2005-04-16 22:20:36 +00:00
/*
2005-08-29 00:18:39 +00:00
* libata - core . c - helper library for ATA
*
* Maintained by : Jeff Garzik < jgarzik @ pobox . com >
* Please ALWAYS copy linux - ide @ vger . kernel . org
* on emails .
*
* Copyright 2003 - 2004 Red Hat , Inc . All rights reserved .
* Copyright 2003 - 2004 Jeff Garzik
*
*
* This program is free software ; you can redistribute it and / or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation ; either version 2 , or ( at your option )
* any later version .
*
* This program is distributed in the hope that it will be useful ,
* but WITHOUT ANY WARRANTY ; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE . See the
* GNU General Public License for more details .
*
* You should have received a copy of the GNU General Public License
* along with this program ; see the file COPYING . If not , write to
* the Free Software Foundation , 675 Mass Ave , Cambridge , MA 0213 9 , USA .
*
*
* libata documentation is available via ' make { ps | pdf } docs ' ,
* as Documentation / DocBook / libata . *
*
* Hardware documentation available from http : //www.t13.org/ and
* http : //www.sata-io.org/
*
2007-11-19 14:30:16 +00:00
* Standards documents from :
* http : //www.t13.org (ATA standards, PCI DMA IDE spec)
* http : //www.t10.org (SCSI MMC - for ATAPI MMC)
* http : //www.sata-io.org (SATA)
* http : //www.compactflash.org (CF)
* http : //www.qic.org (QIC157 - Tape and DSC)
* http : //www.ce-ata.org (CE-ATA: not supported)
*
2005-04-16 22:20:36 +00:00
*/
# include <linux/kernel.h>
# include <linux/module.h>
# include <linux/pci.h>
# include <linux/init.h>
# include <linux/list.h>
# include <linux/mm.h>
# include <linux/spinlock.h>
# include <linux/blkdev.h>
# include <linux/delay.h>
# include <linux/timer.h>
# include <linux/interrupt.h>
# include <linux/completion.h>
# include <linux/suspend.h>
# include <linux/workqueue.h>
2005-09-17 07:55:31 +00:00
# include <linux/scatterlist.h>
2007-10-19 10:42:56 +00:00
# include <linux/io.h>
2009-01-04 13:32:28 +00:00
# include <linux/async.h>
2009-03-19 17:32:21 +00:00
# include <linux/log2.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
# include <linux/slab.h>
2005-04-16 22:20:36 +00:00
# include <scsi/scsi.h>
2005-11-07 05:59:37 +00:00
# include <scsi/scsi_cmnd.h>
2005-04-16 22:20:36 +00:00
# include <scsi/scsi_host.h>
# include <linux/libata.h>
# include <asm/byteorder.h>
2007-12-12 03:21:52 +00:00
# include <linux/cdrom.h>
2010-04-12 12:11:41 +00:00
# include <linux/ratelimit.h>
2005-04-16 22:20:36 +00:00
# include "libata.h"
2010-05-25 19:31:38 +00:00
# include "libata-transport.h"
2007-01-31 12:43:15 +00:00
2006-05-31 09:27:46 +00:00
/* debounce timing parameters in msecs { interval, duration, timeout } */
2006-07-03 07:07:26 +00:00
const unsigned long sata_deb_timing_normal [ ] = { 5 , 100 , 2000 } ;
const unsigned long sata_deb_timing_hotplug [ ] = { 25 , 500 , 2000 } ;
const unsigned long sata_deb_timing_long [ ] = { 100 , 2000 , 5000 } ;
2006-05-31 09:27:46 +00:00
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
const struct ata_port_operations ata_base_port_ops = {
2008-04-07 13:47:18 +00:00
. prereset = ata_std_prereset ,
2008-04-07 13:47:18 +00:00
. postreset = ata_std_postreset ,
libata: make reset related methods proper port operations
Currently reset methods are not specified directly in the
ata_port_operations table. If a LLD wants to use custom reset
methods, it should construct and use a error_handler which uses those
reset methods. It's done this way for two reasons.
First, the ops table already contained too many methods and adding
four more of them would noticeably increase the amount of necessary
boilerplate code all over low level drivers.
Second, as ->error_handler uses those reset methods, it can get
confusing. ie. By overriding ->error_handler, those reset ops can be
made useless making layering a bit hazy.
Now that ops table uses inheritance, the first problem doesn't exist
anymore. The second isn't completely solved but is relieved by
providing default values - most drivers can just override what it has
implemented and don't have to concern itself about higher level
callbacks. In fact, there currently is no driver which actually
modifies error handling behavior. Drivers which override
->error_handler just wraps the standard error handler only to prepare
the controller for EH. I don't think making ops layering strict has
any noticeable benefit.
This patch makes ->prereset, ->softreset, ->hardreset, ->postreset and
their PMP counterparts propoer ops. Default ops are provided in the
base ops tables and drivers are converted to override individual reset
methods instead of creating custom error_handler.
* ata_std_error_handler() doesn't use sata_std_hardreset() if SCRs
aren't accessible. sata_promise doesn't need to use separate
error_handlers for PATA and SATA anymore.
* softreset is broken for sata_inic162x and sata_sx4. As libata now
always prefers hardreset, this doesn't really matter but the ops are
forced to NULL using ATA_OP_NULL for documentation purpose.
* pata_hpt374 needs to use different prereset for the first and second
PCI functions. This used to be done by branching from
hpt374_error_handler(). The proper way to do this is to use
separate ops and port_info tables for each function. Converted.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:50 +00:00
. error_handler = ata_std_error_handler ,
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
} ;
const struct ata_port_operations sata_port_ops = {
. inherits = & ata_base_port_ops ,
. qc_defer = ata_std_qc_defer ,
2008-04-07 13:47:19 +00:00
. hardreset = sata_std_hardreset ,
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
} ;
2006-05-15 11:57:53 +00:00
static unsigned int ata_dev_init_params ( struct ata_device * dev ,
u16 heads , u16 sectors ) ;
static unsigned int ata_dev_set_xfermode ( struct ata_device * dev ) ;
static void ata_dev_xfermask ( struct ata_device * dev ) ;
2007-07-05 04:31:27 +00:00
static unsigned long ata_dev_blacklisted ( const struct ata_device * dev ) ;
2005-04-16 22:20:36 +00:00
2007-04-17 14:44:07 +00:00
unsigned int ata_print_id = 1 ;
2005-04-16 22:20:36 +00:00
2008-02-13 00:15:09 +00:00
struct ata_force_param {
const char * name ;
unsigned int cbl ;
int spd_limit ;
unsigned long xfer_mask ;
unsigned int horkage_on ;
unsigned int horkage_off ;
2008-08-13 11:19:09 +00:00
unsigned int lflags ;
2008-02-13 00:15:09 +00:00
} ;
struct ata_force_ent {
int port ;
int device ;
struct ata_force_param param ;
} ;
static struct ata_force_ent * ata_force_tbl ;
static int ata_force_tbl_size ;
static char ata_force_param_buf [ PAGE_SIZE ] __initdata ;
2008-03-09 11:21:53 +00:00
/* param_buf is thrown away after initialization, disallow read */
module_param_string ( force , ata_force_param_buf , sizeof ( ata_force_param_buf ) , 0 ) ;
2008-02-13 00:15:09 +00:00
MODULE_PARM_DESC ( force , " Force ATA configurations including cable type, link speed and transfer mode (see Documentation/kernel-parameters.txt for details) " ) ;
2008-07-30 22:52:40 +00:00
static int atapi_enabled = 1 ;
2005-08-30 07:37:42 +00:00
module_param ( atapi_enabled , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( atapi_enabled , " Enable discovery of ATAPI devices (0=off, 1=on [default]) " ) ;
2005-08-30 07:37:42 +00:00
2008-02-25 00:07:25 +00:00
static int atapi_dmadir = 0 ;
2006-04-04 02:57:18 +00:00
module_param ( atapi_dmadir , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( atapi_dmadir , " Enable ATAPI DMADIR bridge support (0=off [default], 1=on) " ) ;
2006-04-04 02:57:18 +00:00
2007-08-07 16:08:45 +00:00
int atapi_passthru16 = 1 ;
module_param ( atapi_passthru16 , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( atapi_passthru16 , " Enable ATA_16 passthru for ATAPI devices (0=off, 1=on [default]) " ) ;
2007-08-07 16:08:45 +00:00
2006-02-28 03:31:19 +00:00
int libata_fua = 0 ;
module_param_named ( fua , libata_fua , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( fua , " FUA support (0=off [default], 1=on) " ) ;
2006-02-28 03:31:19 +00:00
2007-10-19 10:42:56 +00:00
static int ata_ignore_hpa ;
2007-04-10 23:23:13 +00:00
module_param_named ( ignore_hpa , ata_ignore_hpa , int , 0644 ) ;
MODULE_PARM_DESC ( ignore_hpa , " Ignore HPA limit (0=keep BIOS limits, 1=ignore limits, using full disk) " ) ;
2007-10-02 16:38:26 +00:00
static int libata_dma_mask = ATA_DMA_MASK_ATA | ATA_DMA_MASK_ATAPI | ATA_DMA_MASK_CFA ;
module_param_named ( dma , libata_dma_mask , int , 0444 ) ;
MODULE_PARM_DESC ( dma , " DMA enable/disable (0x1==ATA, 0x2==ATAPI, 0x4==CF) " ) ;
2008-05-19 17:17:54 +00:00
static int ata_probe_timeout ;
2006-06-25 08:36:52 +00:00
module_param ( ata_probe_timeout , int , 0444 ) ;
MODULE_PARM_DESC ( ata_probe_timeout , " Set ATA probing timeout (seconds) " ) ;
2007-10-05 20:28:36 +00:00
int libata_noacpi = 0 ;
2007-03-28 05:57:37 +00:00
module_param_named ( noacpi , libata_noacpi , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( noacpi , " Disable the use of ACPI in probe/suspend/resume (0=off [default], 1=on) " ) ;
2006-09-28 18:29:01 +00:00
2007-11-05 03:05:49 +00:00
int libata_allow_tpm = 0 ;
module_param_named ( allow_tpm , libata_allow_tpm , int , 0444 ) ;
2009-06-22 07:46:20 +00:00
MODULE_PARM_DESC ( allow_tpm , " Permit the use of TPM commands (0=off [default], 1=on) " ) ;
2007-11-05 03:05:49 +00:00
2010-05-19 13:38:58 +00:00
static int atapi_an ;
module_param ( atapi_an , int , 0444 ) ;
MODULE_PARM_DESC ( atapi_an , " Enable ATAPI AN media presence notification (0=0ff [default], 1=on) " ) ;
2005-04-16 22:20:36 +00:00
MODULE_AUTHOR ( " Jeff Garzik " ) ;
MODULE_DESCRIPTION ( " Library module for ATA devices " ) ;
MODULE_LICENSE ( " GPL " ) ;
MODULE_VERSION ( DRV_VERSION ) ;
2005-06-02 22:17:13 +00:00
2009-01-29 11:31:31 +00:00
static bool ata_sstatus_online ( u32 sstatus )
{
return ( sstatus & 0xf ) = = 0x3 ;
}
2008-11-03 11:03:17 +00:00
/**
* ata_link_next - link iteration helper
* @ link : the previous link , NULL to start
* @ ap : ATA port containing links to iterate
* @ mode : iteration mode , one of ATA_LITER_ *
*
* LOCKING :
* Host lock or EH context .
2008-07-31 08:02:41 +00:00
*
2008-11-03 11:03:17 +00:00
* RETURNS :
* Pointer to the next link .
2008-07-31 08:02:41 +00:00
*/
2008-11-03 11:03:17 +00:00
struct ata_link * ata_link_next ( struct ata_link * link , struct ata_port * ap ,
enum ata_link_iter_mode mode )
2008-07-31 08:02:41 +00:00
{
2008-11-03 11:03:17 +00:00
BUG_ON ( mode ! = ATA_LITER_EDGE & &
mode ! = ATA_LITER_PMP_FIRST & & mode ! = ATA_LITER_HOST_FIRST ) ;
2008-07-31 08:02:41 +00:00
/* NULL link indicates start of iteration */
2008-11-03 11:03:17 +00:00
if ( ! link )
switch ( mode ) {
case ATA_LITER_EDGE :
case ATA_LITER_PMP_FIRST :
if ( sata_pmp_attached ( ap ) )
return ap - > pmp_link ;
/* fall through */
case ATA_LITER_HOST_FIRST :
return & ap - > link ;
}
2008-07-31 08:02:41 +00:00
2008-11-03 11:03:17 +00:00
/* we just iterated over the host link, what's next? */
if ( link = = & ap - > link )
switch ( mode ) {
case ATA_LITER_HOST_FIRST :
if ( sata_pmp_attached ( ap ) )
return ap - > pmp_link ;
/* fall through */
case ATA_LITER_PMP_FIRST :
if ( unlikely ( ap - > slave_link ) )
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
return ap - > slave_link ;
2008-11-03 11:03:17 +00:00
/* fall through */
case ATA_LITER_EDGE :
2008-07-31 08:02:41 +00:00
return NULL ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
}
2008-07-31 08:02:41 +00:00
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/* slave_link excludes PMP */
if ( unlikely ( link = = ap - > slave_link ) )
return NULL ;
2008-11-03 11:03:17 +00:00
/* we were over a PMP link */
2008-07-31 08:02:41 +00:00
if ( + + link < ap - > pmp_link + ap - > nr_pmp_links )
return link ;
2008-11-03 11:03:17 +00:00
if ( mode = = ATA_LITER_PMP_FIRST )
return & ap - > link ;
2008-07-31 08:02:41 +00:00
return NULL ;
}
2008-11-03 11:03:17 +00:00
/**
* ata_dev_next - device iteration helper
* @ dev : the previous device , NULL to start
* @ link : ATA link containing devices to iterate
* @ mode : iteration mode , one of ATA_DITER_ *
*
* LOCKING :
* Host lock or EH context .
*
* RETURNS :
* Pointer to the next device .
*/
struct ata_device * ata_dev_next ( struct ata_device * dev , struct ata_link * link ,
enum ata_dev_iter_mode mode )
{
BUG_ON ( mode ! = ATA_DITER_ENABLED & & mode ! = ATA_DITER_ENABLED_REVERSE & &
mode ! = ATA_DITER_ALL & & mode ! = ATA_DITER_ALL_REVERSE ) ;
/* NULL dev indicates start of iteration */
if ( ! dev )
switch ( mode ) {
case ATA_DITER_ENABLED :
case ATA_DITER_ALL :
dev = link - > device ;
goto check ;
case ATA_DITER_ENABLED_REVERSE :
case ATA_DITER_ALL_REVERSE :
dev = link - > device + ata_link_max_devices ( link ) - 1 ;
goto check ;
}
next :
/* move to the next one */
switch ( mode ) {
case ATA_DITER_ENABLED :
case ATA_DITER_ALL :
if ( + + dev < link - > device + ata_link_max_devices ( link ) )
goto check ;
return NULL ;
case ATA_DITER_ENABLED_REVERSE :
case ATA_DITER_ALL_REVERSE :
if ( - - dev > = link - > device )
goto check ;
return NULL ;
}
check :
if ( ( mode = = ATA_DITER_ENABLED | | mode = = ATA_DITER_ENABLED_REVERSE ) & &
! ata_dev_enabled ( dev ) )
goto next ;
return dev ;
}
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/**
* ata_dev_phys_link - find physical link for a device
* @ dev : ATA device to look up physical link for
*
* Look up physical link which @ dev is attached to . Note that
* this is different from @ dev - > link only when @ dev is on slave
* link . For all other cases , it ' s the same as @ dev - > link .
*
* LOCKING :
* Don ' t care .
*
* RETURNS :
* Pointer to the found physical link .
*/
struct ata_link * ata_dev_phys_link ( struct ata_device * dev )
{
struct ata_port * ap = dev - > link - > ap ;
if ( ! ap - > slave_link )
return dev - > link ;
if ( ! dev - > devno )
return & ap - > link ;
return ap - > slave_link ;
}
2008-02-13 00:15:09 +00:00
/**
* ata_force_cbl - force cable type according to libata . force
2008-02-22 20:21:37 +00:00
* @ ap : ATA port of interest
2008-02-13 00:15:09 +00:00
*
* Force cable type according to libata . force and whine about it .
* The last entry which has matching port number is used , so it
* can be specified as part of device force parameters . For
* example , both " a:40c,1.00:udma4 " and " 1.00:40c,udma4 " have the
* same effect .
*
* LOCKING :
* EH context .
*/
void ata_force_cbl ( struct ata_port * ap )
{
int i ;
for ( i = ata_force_tbl_size - 1 ; i > = 0 ; i - - ) {
const struct ata_force_ent * fe = & ata_force_tbl [ i ] ;
if ( fe - > port ! = - 1 & & fe - > port ! = ap - > print_id )
continue ;
if ( fe - > param . cbl = = ATA_CBL_NONE )
continue ;
ap - > cbl = fe - > param . cbl ;
ata_port_printk ( ap , KERN_NOTICE ,
" FORCE: cable set to %s \n " , fe - > param . name ) ;
return ;
}
}
/**
2008-08-13 11:19:09 +00:00
* ata_force_link_limits - force link limits according to libata . force
2008-02-13 00:15:09 +00:00
* @ link : ATA link of interest
*
2008-08-13 11:19:09 +00:00
* Force link flags and SATA spd limit according to libata . force
* and whine about it . When only the port part is specified
* ( e . g . 1 : ) , the limit applies to all links connected to both
* the host link and all fan - out ports connected via PMP . If the
* device part is specified as 0 ( e . g . 1.00 : ) , it specifies the
* first fan - out link not the host link . Device number 15 always
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
* points to the host link whether PMP is attached or not . If the
* controller has slave link , device number 16 points to it .
2008-02-13 00:15:09 +00:00
*
* LOCKING :
* EH context .
*/
2008-08-13 11:19:09 +00:00
static void ata_force_link_limits ( struct ata_link * link )
2008-02-13 00:15:09 +00:00
{
2008-08-13 11:19:09 +00:00
bool did_spd = false ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
int linkno = link - > pmp ;
int i ;
2008-02-13 00:15:09 +00:00
if ( ata_is_host_link ( link ) )
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
linkno + = 15 ;
2008-02-13 00:15:09 +00:00
for ( i = ata_force_tbl_size - 1 ; i > = 0 ; i - - ) {
const struct ata_force_ent * fe = & ata_force_tbl [ i ] ;
if ( fe - > port ! = - 1 & & fe - > port ! = link - > ap - > print_id )
continue ;
if ( fe - > device ! = - 1 & & fe - > device ! = linkno )
continue ;
2008-08-13 11:19:09 +00:00
/* only honor the first spd limit */
if ( ! did_spd & & fe - > param . spd_limit ) {
link - > hw_sata_spd_limit = ( 1 < < fe - > param . spd_limit ) - 1 ;
ata_link_printk ( link , KERN_NOTICE ,
" FORCE: PHY spd limit set to %s \n " ,
fe - > param . name ) ;
did_spd = true ;
}
2008-02-13 00:15:09 +00:00
2008-08-13 11:19:09 +00:00
/* let lflags stack */
if ( fe - > param . lflags ) {
link - > flags | = fe - > param . lflags ;
ata_link_printk ( link , KERN_NOTICE ,
" FORCE: link flag 0x%x forced -> 0x%x \n " ,
fe - > param . lflags , link - > flags ) ;
}
2008-02-13 00:15:09 +00:00
}
}
/**
* ata_force_xfermask - force xfermask according to libata . force
* @ dev : ATA device of interest
*
* Force xfer_mask according to libata . force and whine about it .
* For consistency with link selection , device number 15 selects
* the first device connected to the host link .
*
* LOCKING :
* EH context .
*/
static void ata_force_xfermask ( struct ata_device * dev )
{
int devno = dev - > link - > pmp + dev - > devno ;
int alt_devno = devno ;
int i ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/* allow n.15/16 for devices attached to host port */
if ( ata_is_host_link ( dev - > link ) )
alt_devno + = 15 ;
2008-02-13 00:15:09 +00:00
for ( i = ata_force_tbl_size - 1 ; i > = 0 ; i - - ) {
const struct ata_force_ent * fe = & ata_force_tbl [ i ] ;
unsigned long pio_mask , mwdma_mask , udma_mask ;
if ( fe - > port ! = - 1 & & fe - > port ! = dev - > link - > ap - > print_id )
continue ;
if ( fe - > device ! = - 1 & & fe - > device ! = devno & &
fe - > device ! = alt_devno )
continue ;
if ( ! fe - > param . xfer_mask )
continue ;
ata_unpack_xfermask ( fe - > param . xfer_mask ,
& pio_mask , & mwdma_mask , & udma_mask ) ;
if ( udma_mask )
dev - > udma_mask = udma_mask ;
else if ( mwdma_mask ) {
dev - > udma_mask = 0 ;
dev - > mwdma_mask = mwdma_mask ;
} else {
dev - > udma_mask = 0 ;
dev - > mwdma_mask = 0 ;
dev - > pio_mask = pio_mask ;
}
ata_dev_printk ( dev , KERN_NOTICE ,
" FORCE: xfer_mask set to %s \n " , fe - > param . name ) ;
return ;
}
}
/**
* ata_force_horkage - force horkage according to libata . force
* @ dev : ATA device of interest
*
* Force horkage according to libata . force and whine about it .
* For consistency with link selection , device number 15 selects
* the first device connected to the host link .
*
* LOCKING :
* EH context .
*/
static void ata_force_horkage ( struct ata_device * dev )
{
int devno = dev - > link - > pmp + dev - > devno ;
int alt_devno = devno ;
int i ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/* allow n.15/16 for devices attached to host port */
if ( ata_is_host_link ( dev - > link ) )
alt_devno + = 15 ;
2008-02-13 00:15:09 +00:00
for ( i = 0 ; i < ata_force_tbl_size ; i + + ) {
const struct ata_force_ent * fe = & ata_force_tbl [ i ] ;
if ( fe - > port ! = - 1 & & fe - > port ! = dev - > link - > ap - > print_id )
continue ;
if ( fe - > device ! = - 1 & & fe - > device ! = devno & &
fe - > device ! = alt_devno )
continue ;
if ( ! ( ~ dev - > horkage & fe - > param . horkage_on ) & &
! ( dev - > horkage & fe - > param . horkage_off ) )
continue ;
dev - > horkage | = fe - > param . horkage_on ;
dev - > horkage & = ~ fe - > param . horkage_off ;
ata_dev_printk ( dev , KERN_NOTICE ,
" FORCE: horkage modified (%s) \n " , fe - > param . name ) ;
}
}
2008-04-02 08:28:46 +00:00
/**
* atapi_cmd_type - Determine ATAPI command type from SCSI opcode
* @ opcode : SCSI opcode
*
* Determine ATAPI command type from @ opcode .
*
* LOCKING :
* None .
*
* RETURNS :
* ATAPI_ { READ | WRITE | READ_CD | PASS_THRU | MISC }
*/
int atapi_cmd_type ( u8 opcode )
{
switch ( opcode ) {
case GPCMD_READ_10 :
case GPCMD_READ_12 :
return ATAPI_READ ;
case GPCMD_WRITE_10 :
case GPCMD_WRITE_12 :
case GPCMD_WRITE_AND_VERIFY_10 :
return ATAPI_WRITE ;
case GPCMD_READ_CD :
case GPCMD_READ_CD_MSF :
return ATAPI_READ_CD ;
2008-04-02 08:35:19 +00:00
case ATA_16 :
case ATA_12 :
if ( atapi_passthru16 )
return ATAPI_PASS_THRU ;
/* fall thru */
2008-04-02 08:28:46 +00:00
default :
return ATAPI_MISC ;
}
}
2005-04-16 22:20:36 +00:00
/**
* ata_tf_to_fis - Convert ATA taskfile to SATA FIS structure
* @ tf : Taskfile to convert
* @ pmp : Port multiplier port
2007-07-16 05:29:38 +00:00
* @ is_cmd : This FIS is for command
* @ fis : Buffer into which data will output
2005-04-16 22:20:36 +00:00
*
* Converts a standard ATA taskfile to a Serial ATA
* FIS structure ( Register - Host to Device ) .
*
* LOCKING :
* Inherited from caller .
*/
2007-07-16 05:29:38 +00:00
void ata_tf_to_fis ( const struct ata_taskfile * tf , u8 pmp , int is_cmd , u8 * fis )
2005-04-16 22:20:36 +00:00
{
2007-07-16 05:29:38 +00:00
fis [ 0 ] = 0x27 ; /* Register - Host to Device FIS */
fis [ 1 ] = pmp & 0xf ; /* Port multiplier number*/
if ( is_cmd )
fis [ 1 ] | = ( 1 < < 7 ) ; /* bit 7 indicates Command FIS */
2005-04-16 22:20:36 +00:00
fis [ 2 ] = tf - > command ;
fis [ 3 ] = tf - > feature ;
fis [ 4 ] = tf - > lbal ;
fis [ 5 ] = tf - > lbam ;
fis [ 6 ] = tf - > lbah ;
fis [ 7 ] = tf - > device ;
fis [ 8 ] = tf - > hob_lbal ;
fis [ 9 ] = tf - > hob_lbam ;
fis [ 10 ] = tf - > hob_lbah ;
fis [ 11 ] = tf - > hob_feature ;
fis [ 12 ] = tf - > nsect ;
fis [ 13 ] = tf - > hob_nsect ;
fis [ 14 ] = 0 ;
fis [ 15 ] = tf - > ctl ;
fis [ 16 ] = 0 ;
fis [ 17 ] = 0 ;
fis [ 18 ] = 0 ;
fis [ 19 ] = 0 ;
}
/**
* ata_tf_from_fis - Convert SATA FIS to ATA taskfile
* @ fis : Buffer from which data will be input
* @ tf : Taskfile to output
*
2005-11-12 23:55:45 +00:00
* Converts a serial ATA FIS structure to a standard ATA taskfile .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
* Inherited from caller .
*/
2005-10-22 18:27:05 +00:00
void ata_tf_from_fis ( const u8 * fis , struct ata_taskfile * tf )
2005-04-16 22:20:36 +00:00
{
tf - > command = fis [ 2 ] ; /* status */
tf - > feature = fis [ 3 ] ; /* error */
tf - > lbal = fis [ 4 ] ;
tf - > lbam = fis [ 5 ] ;
tf - > lbah = fis [ 6 ] ;
tf - > device = fis [ 7 ] ;
tf - > hob_lbal = fis [ 8 ] ;
tf - > hob_lbam = fis [ 9 ] ;
tf - > hob_lbah = fis [ 10 ] ;
tf - > nsect = fis [ 12 ] ;
tf - > hob_nsect = fis [ 13 ] ;
}
2005-10-12 07:06:27 +00:00
static const u8 ata_rw_cmds [ ] = {
/* pio multi */
ATA_CMD_READ_MULTI ,
ATA_CMD_WRITE_MULTI ,
ATA_CMD_READ_MULTI_EXT ,
ATA_CMD_WRITE_MULTI_EXT ,
2006-01-06 08:56:18 +00:00
0 ,
0 ,
0 ,
ATA_CMD_WRITE_MULTI_FUA_EXT ,
2005-10-12 07:06:27 +00:00
/* pio */
ATA_CMD_PIO_READ ,
ATA_CMD_PIO_WRITE ,
ATA_CMD_PIO_READ_EXT ,
ATA_CMD_PIO_WRITE_EXT ,
2006-01-06 08:56:18 +00:00
0 ,
0 ,
0 ,
0 ,
2005-10-12 07:06:27 +00:00
/* dma */
ATA_CMD_READ ,
ATA_CMD_WRITE ,
ATA_CMD_READ_EXT ,
2006-01-06 08:56:18 +00:00
ATA_CMD_WRITE_EXT ,
0 ,
0 ,
0 ,
ATA_CMD_WRITE_FUA_EXT
2005-10-12 07:06:27 +00:00
} ;
2005-04-16 22:20:36 +00:00
/**
2005-10-12 07:06:27 +00:00
* ata_rwcmd_protocol - set taskfile r / w commands and protocol
2006-11-14 13:47:10 +00:00
* @ tf : command to examine and configure
* @ dev : device tf belongs to
2005-04-16 22:20:36 +00:00
*
2006-03-24 14:56:57 +00:00
* Examine the device configuration and tf - > flags to calculate
2005-10-12 07:06:27 +00:00
* the proper read / write commands and protocol to use .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
* caller .
*/
2006-11-14 13:47:10 +00:00
static int ata_rwcmd_protocol ( struct ata_taskfile * tf , struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2006-01-06 08:56:18 +00:00
u8 cmd ;
2005-04-16 22:20:36 +00:00
2006-01-06 08:56:18 +00:00
int index , fua , lba48 , write ;
2006-03-24 14:56:57 +00:00
2006-01-06 08:56:18 +00:00
fua = ( tf - > flags & ATA_TFLAG_FUA ) ? 4 : 0 ;
2005-10-12 07:06:27 +00:00
lba48 = ( tf - > flags & ATA_TFLAG_LBA48 ) ? 2 : 0 ;
write = ( tf - > flags & ATA_TFLAG_WRITE ) ? 1 : 0 ;
2005-04-16 22:20:36 +00:00
2005-10-12 07:06:27 +00:00
if ( dev - > flags & ATA_DFLAG_PIO ) {
tf - > protocol = ATA_PROT_PIO ;
2006-01-06 08:56:18 +00:00
index = dev - > multi_count ? 0 : 8 ;
2007-08-06 09:36:22 +00:00
} else if ( lba48 & & ( dev - > link - > ap - > flags & ATA_FLAG_PIO_LBA48 ) ) {
2006-01-17 20:50:31 +00:00
/* Unable to use DMA due to host limitation */
tf - > protocol = ATA_PROT_PIO ;
2006-02-13 10:55:25 +00:00
index = dev - > multi_count ? 0 : 8 ;
2005-10-12 07:06:27 +00:00
} else {
tf - > protocol = ATA_PROT_DMA ;
2006-01-06 08:56:18 +00:00
index = 16 ;
2005-10-12 07:06:27 +00:00
}
2005-04-16 22:20:36 +00:00
2006-01-06 08:56:18 +00:00
cmd = ata_rw_cmds [ index + fua + lba48 + write ] ;
if ( cmd ) {
tf - > command = cmd ;
return 0 ;
}
return - 1 ;
2005-04-16 22:20:36 +00:00
}
2006-11-14 13:37:35 +00:00
/**
* ata_tf_read_block - Read block address from ATA taskfile
* @ tf : ATA taskfile of interest
* @ dev : ATA device @ tf belongs to
*
* LOCKING :
* None .
*
* Read block address from @ tf . This function can handle all
* three address formats - LBA , LBA48 and CHS . tf - > protocol and
* flags select the address format to use .
*
* RETURNS :
* Block address read from @ tf .
*/
u64 ata_tf_read_block ( struct ata_taskfile * tf , struct ata_device * dev )
{
u64 block = 0 ;
if ( tf - > flags & ATA_TFLAG_LBA ) {
if ( tf - > flags & ATA_TFLAG_LBA48 ) {
block | = ( u64 ) tf - > hob_lbah < < 40 ;
block | = ( u64 ) tf - > hob_lbam < < 32 ;
2008-11-04 18:34:48 +00:00
block | = ( u64 ) tf - > hob_lbal < < 24 ;
2006-11-14 13:37:35 +00:00
} else
block | = ( tf - > device & 0xf ) < < 24 ;
block | = tf - > lbah < < 16 ;
block | = tf - > lbam < < 8 ;
block | = tf - > lbal ;
} else {
u32 cyl , head , sect ;
cyl = tf - > lbam | ( tf - > lbah < < 8 ) ;
head = tf - > device & 0xf ;
sect = tf - > lbal ;
2009-08-16 12:21:21 +00:00
if ( ! sect ) {
ata_dev_printk ( dev , KERN_WARNING , " device reported "
" invalid CHS sector 0 \n " ) ;
sect = 1 ; /* oh well */
}
block = ( cyl * dev - > heads + head ) * dev - > sectors + sect - 1 ;
2006-11-14 13:37:35 +00:00
}
return block ;
}
2006-11-14 13:47:10 +00:00
/**
* ata_build_rw_tf - Build ATA taskfile for given read / write request
* @ tf : Target ATA taskfile
* @ dev : ATA device @ tf belongs to
* @ block : Block address
* @ n_block : Number of blocks
* @ tf_flags : RW / FUA etc . . .
* @ tag : tag
*
* LOCKING :
* None .
*
* Build ATA taskfile @ tf for read / write request described by
* @ block , @ n_block , @ tf_flags and @ tag on @ dev .
*
* RETURNS :
*
* 0 on success , - ERANGE if the request is too large for @ dev ,
* - EINVAL if the request is invalid .
*/
int ata_build_rw_tf ( struct ata_taskfile * tf , struct ata_device * dev ,
u64 block , u32 n_block , unsigned int tf_flags ,
unsigned int tag )
{
tf - > flags | = ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE ;
tf - > flags | = tf_flags ;
2007-02-20 14:20:27 +00:00
if ( ata_ncq_enabled ( dev ) & & likely ( tag ! = ATA_TAG_INTERNAL ) ) {
2006-11-14 13:47:10 +00:00
/* yay, NCQ */
if ( ! lba_48_ok ( block , n_block ) )
return - ERANGE ;
tf - > protocol = ATA_PROT_NCQ ;
tf - > flags | = ATA_TFLAG_LBA | ATA_TFLAG_LBA48 ;
if ( tf - > flags & ATA_TFLAG_WRITE )
tf - > command = ATA_CMD_FPDMA_WRITE ;
else
tf - > command = ATA_CMD_FPDMA_READ ;
tf - > nsect = tag < < 3 ;
tf - > hob_feature = ( n_block > > 8 ) & 0xff ;
tf - > feature = n_block & 0xff ;
tf - > hob_lbah = ( block > > 40 ) & 0xff ;
tf - > hob_lbam = ( block > > 32 ) & 0xff ;
tf - > hob_lbal = ( block > > 24 ) & 0xff ;
tf - > lbah = ( block > > 16 ) & 0xff ;
tf - > lbam = ( block > > 8 ) & 0xff ;
tf - > lbal = block & 0xff ;
tf - > device = 1 < < 6 ;
if ( tf - > flags & ATA_TFLAG_FUA )
tf - > device | = 1 < < 7 ;
} else if ( dev - > flags & ATA_DFLAG_LBA ) {
tf - > flags | = ATA_TFLAG_LBA ;
if ( lba_28_ok ( block , n_block ) ) {
/* use LBA28 */
tf - > device | = ( block > > 24 ) & 0xf ;
} else if ( lba_48_ok ( block , n_block ) ) {
if ( ! ( dev - > flags & ATA_DFLAG_LBA48 ) )
return - ERANGE ;
/* use LBA48 */
tf - > flags | = ATA_TFLAG_LBA48 ;
tf - > hob_nsect = ( n_block > > 8 ) & 0xff ;
tf - > hob_lbah = ( block > > 40 ) & 0xff ;
tf - > hob_lbam = ( block > > 32 ) & 0xff ;
tf - > hob_lbal = ( block > > 24 ) & 0xff ;
} else
/* request too large even for LBA48 */
return - ERANGE ;
if ( unlikely ( ata_rwcmd_protocol ( tf , dev ) < 0 ) )
return - EINVAL ;
tf - > nsect = n_block & 0xff ;
tf - > lbah = ( block > > 16 ) & 0xff ;
tf - > lbam = ( block > > 8 ) & 0xff ;
tf - > lbal = block & 0xff ;
tf - > device | = ATA_LBA ;
} else {
/* CHS */
u32 sect , head , cyl , track ;
/* The request -may- be too large for CHS addressing. */
if ( ! lba_28_ok ( block , n_block ) )
return - ERANGE ;
if ( unlikely ( ata_rwcmd_protocol ( tf , dev ) < 0 ) )
return - EINVAL ;
/* Convert LBA to CHS */
track = ( u32 ) block / dev - > sectors ;
cyl = track / dev - > heads ;
head = track % dev - > heads ;
sect = ( u32 ) block % dev - > sectors + 1 ;
DPRINTK ( " block %u track %u cyl %u head %u sect %u \n " ,
( u32 ) block , track , cyl , head , sect ) ;
/* Check whether the converted CHS can fit.
Cylinder : 0 - 65535
Head : 0 - 15
Sector : 1 - 255 */
if ( ( cyl > > 16 ) | | ( head > > 4 ) | | ( sect > > 8 ) | | ( ! sect ) )
return - ERANGE ;
tf - > nsect = n_block & 0xff ; /* Sector count 0 means 256 sectors */
tf - > lbal = sect ;
tf - > lbam = cyl ;
tf - > lbah = cyl > > 8 ;
tf - > device | = head ;
}
return 0 ;
}
2006-03-05 19:31:56 +00:00
/**
* ata_pack_xfermask - Pack pio , mwdma and udma masks into xfer_mask
* @ pio_mask : pio_mask
* @ mwdma_mask : mwdma_mask
* @ udma_mask : udma_mask
*
* Pack @ pio_mask , @ mwdma_mask and @ udma_mask into a single
* unsigned int xfer_mask .
*
* LOCKING :
* None .
*
* RETURNS :
* Packed xfer_mask .
*/
2007-11-27 10:43:42 +00:00
unsigned long ata_pack_xfermask ( unsigned long pio_mask ,
unsigned long mwdma_mask ,
unsigned long udma_mask )
2006-03-05 19:31:56 +00:00
{
return ( ( pio_mask < < ATA_SHIFT_PIO ) & ATA_MASK_PIO ) |
( ( mwdma_mask < < ATA_SHIFT_MWDMA ) & ATA_MASK_MWDMA ) |
( ( udma_mask < < ATA_SHIFT_UDMA ) & ATA_MASK_UDMA ) ;
}
2006-03-24 05:07:49 +00:00
/**
* ata_unpack_xfermask - Unpack xfer_mask into pio , mwdma and udma masks
* @ xfer_mask : xfer_mask to unpack
* @ pio_mask : resulting pio_mask
* @ mwdma_mask : resulting mwdma_mask
* @ udma_mask : resulting udma_mask
*
* Unpack @ xfer_mask into @ pio_mask , @ mwdma_mask and @ udma_mask .
* Any NULL distination masks will be ignored .
*/
2007-11-27 10:43:42 +00:00
void ata_unpack_xfermask ( unsigned long xfer_mask , unsigned long * pio_mask ,
unsigned long * mwdma_mask , unsigned long * udma_mask )
2006-03-24 05:07:49 +00:00
{
if ( pio_mask )
* pio_mask = ( xfer_mask & ATA_MASK_PIO ) > > ATA_SHIFT_PIO ;
if ( mwdma_mask )
* mwdma_mask = ( xfer_mask & ATA_MASK_MWDMA ) > > ATA_SHIFT_MWDMA ;
if ( udma_mask )
* udma_mask = ( xfer_mask & ATA_MASK_UDMA ) > > ATA_SHIFT_UDMA ;
}
2006-03-05 19:31:56 +00:00
static const struct ata_xfer_ent {
2006-03-31 13:48:52 +00:00
int shift , bits ;
2006-03-05 19:31:56 +00:00
u8 base ;
} ata_xfer_tbl [ ] = {
2007-11-27 10:43:40 +00:00
{ ATA_SHIFT_PIO , ATA_NR_PIO_MODES , XFER_PIO_0 } ,
{ ATA_SHIFT_MWDMA , ATA_NR_MWDMA_MODES , XFER_MW_DMA_0 } ,
{ ATA_SHIFT_UDMA , ATA_NR_UDMA_MODES , XFER_UDMA_0 } ,
2006-03-05 19:31:56 +00:00
{ - 1 , } ,
} ;
/**
* ata_xfer_mask2mode - Find matching XFER_ * for the given xfer_mask
* @ xfer_mask : xfer_mask of interest
*
* Return matching XFER_ * value for @ xfer_mask . Only the highest
* bit of @ xfer_mask is considered .
*
* LOCKING :
* None .
*
* RETURNS :
2007-11-27 10:43:40 +00:00
* Matching XFER_ * value , 0xff if no match found .
2006-03-05 19:31:56 +00:00
*/
2007-11-27 10:43:42 +00:00
u8 ata_xfer_mask2mode ( unsigned long xfer_mask )
2006-03-05 19:31:56 +00:00
{
int highbit = fls ( xfer_mask ) - 1 ;
const struct ata_xfer_ent * ent ;
for ( ent = ata_xfer_tbl ; ent - > shift > = 0 ; ent + + )
if ( highbit > = ent - > shift & & highbit < ent - > shift + ent - > bits )
return ent - > base + highbit - ent - > shift ;
2007-11-27 10:43:40 +00:00
return 0xff ;
2006-03-05 19:31:56 +00:00
}
/**
* ata_xfer_mode2mask - Find matching xfer_mask for XFER_ *
* @ xfer_mode : XFER_ * of interest
*
* Return matching xfer_mask for @ xfer_mode .
*
* LOCKING :
* None .
*
* RETURNS :
* Matching xfer_mask , 0 if no match found .
*/
2007-11-27 10:43:42 +00:00
unsigned long ata_xfer_mode2mask ( u8 xfer_mode )
2006-03-05 19:31:56 +00:00
{
const struct ata_xfer_ent * ent ;
for ( ent = ata_xfer_tbl ; ent - > shift > = 0 ; ent + + )
if ( xfer_mode > = ent - > base & & xfer_mode < ent - > base + ent - > bits )
2007-11-27 10:43:40 +00:00
return ( ( 2 < < ( ent - > shift + xfer_mode - ent - > base ) ) - 1 )
& ~ ( ( 1 < < ent - > shift ) - 1 ) ;
2006-03-05 19:31:56 +00:00
return 0 ;
}
/**
* ata_xfer_mode2shift - Find matching xfer_shift for XFER_ *
* @ xfer_mode : XFER_ * of interest
*
* Return matching xfer_shift for @ xfer_mode .
*
* LOCKING :
* None .
*
* RETURNS :
* Matching xfer_shift , - 1 if no match found .
*/
2007-11-27 10:43:42 +00:00
int ata_xfer_mode2shift ( unsigned long xfer_mode )
2006-03-05 19:31:56 +00:00
{
const struct ata_xfer_ent * ent ;
for ( ent = ata_xfer_tbl ; ent - > shift > = 0 ; ent + + )
if ( xfer_mode > = ent - > base & & xfer_mode < ent - > base + ent - > bits )
return ent - > shift ;
return - 1 ;
}
2005-04-16 22:20:36 +00:00
/**
2006-03-05 19:31:56 +00:00
* ata_mode_string - convert xfer_mask to string
* @ xfer_mask : mask of bits supported ; only highest bit counts .
2005-04-16 22:20:36 +00:00
*
* Determine string which represents the highest speed
2006-03-05 19:31:56 +00:00
* ( highest bit in @ modemask ) .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
* None .
*
* RETURNS :
* Constant C string representing highest speed listed in
2006-03-05 19:31:56 +00:00
* @ mode_mask , or the constant C string " <n/a> " .
2005-04-16 22:20:36 +00:00
*/
2007-11-27 10:43:42 +00:00
const char * ata_mode_string ( unsigned long xfer_mask )
2005-04-16 22:20:36 +00:00
{
2006-03-05 19:31:57 +00:00
static const char * const xfer_mode_str [ ] = {
" PIO0 " ,
" PIO1 " ,
" PIO2 " ,
" PIO3 " ,
" PIO4 " ,
2006-08-10 17:52:12 +00:00
" PIO5 " ,
" PIO6 " ,
2006-03-05 19:31:57 +00:00
" MWDMA0 " ,
" MWDMA1 " ,
" MWDMA2 " ,
2006-08-10 17:52:12 +00:00
" MWDMA3 " ,
" MWDMA4 " ,
2006-03-05 19:31:57 +00:00
" UDMA/16 " ,
" UDMA/25 " ,
" UDMA/33 " ,
" UDMA/44 " ,
" UDMA/66 " ,
" UDMA/100 " ,
" UDMA/133 " ,
" UDMA7 " ,
} ;
2006-03-05 19:31:56 +00:00
int highbit ;
2005-04-16 22:20:36 +00:00
2006-03-05 19:31:56 +00:00
highbit = fls ( xfer_mask ) - 1 ;
if ( highbit > = 0 & & highbit < ARRAY_SIZE ( xfer_mode_str ) )
return xfer_mode_str [ highbit ] ;
2005-04-16 22:20:36 +00:00
return " <n/a> " ;
}
2010-05-25 19:31:38 +00:00
const char * sata_spd_string ( unsigned int spd )
2006-03-31 16:38:17 +00:00
{
static const char * const spd_str [ ] = {
" 1.5 Gbps " ,
" 3.0 Gbps " ,
2008-12-30 03:00:37 +00:00
" 6.0 Gbps " ,
2006-03-31 16:38:17 +00:00
} ;
if ( spd = = 0 | | ( spd - 1 ) > = ARRAY_SIZE ( spd_str ) )
return " <unknown> " ;
return spd_str [ spd - 1 ] ;
}
2005-04-16 22:20:36 +00:00
/**
* ata_dev_classify - determine device type based on ATA - spec signature
* @ tf : ATA taskfile register set for device to be identified
*
* Determine from taskfile register contents whether a device is
* ATA or ATAPI , as per " Signature and persistence " section
* of ATA / PI spec ( volume 1 , sect 5.14 ) .
*
* LOCKING :
* None .
*
* RETURNS :
2007-09-23 04:19:54 +00:00
* Device type , % ATA_DEV_ATA , % ATA_DEV_ATAPI , % ATA_DEV_PMP or
* % ATA_DEV_UNKNOWN the event of failure .
2005-04-16 22:20:36 +00:00
*/
2005-10-22 18:27:05 +00:00
unsigned int ata_dev_classify ( const struct ata_taskfile * tf )
2005-04-16 22:20:36 +00:00
{
/* Apple's open source Darwin code hints that some devices only
* put a proper signature into the LBA mid / high registers ,
* So , we only check those . It ' s sufficient for uniqueness .
2007-09-23 04:19:54 +00:00
*
* ATA / ATAPI - 7 ( d1532v1r1 : Feb . 19 , 2003 ) specified separate
* signatures for ATA and ATAPI devices attached on SerialATA ,
* 0x3c / 0xc3 and 0x69 / 0x96 respectively . However , SerialATA
* spec has never mentioned about using different signatures
* for ATA / ATAPI devices . Then , Serial ATA II : Port
* Multiplier specification began to use 0x69 / 0x96 to identify
* port multpliers and 0x3c / 0xc3 to identify SEMB device .
* ATA / ATAPI - 7 dropped descriptions about 0x3c / 0xc3 and
* 0x69 / 0x96 shortly and described them as reserved for
* SerialATA .
*
* We follow the current spec and consider that 0x69 / 0x96
* identifies a port multiplier and 0x3c / 0xc3 a SEMB device .
2009-04-14 21:21:10 +00:00
* Unfortunately , WDC WD1600JS - 62 MHB5 ( a hard drive ) reports
* SEMB signature . This is worked around in
* ata_dev_read_id ( ) .
2005-04-16 22:20:36 +00:00
*/
2007-09-23 04:19:54 +00:00
if ( ( tf - > lbam = = 0 ) & & ( tf - > lbah = = 0 ) ) {
2005-04-16 22:20:36 +00:00
DPRINTK ( " found ATA device by sig \n " ) ;
return ATA_DEV_ATA ;
}
2007-09-23 04:19:54 +00:00
if ( ( tf - > lbam = = 0x14 ) & & ( tf - > lbah = = 0xeb ) ) {
2005-04-16 22:20:36 +00:00
DPRINTK ( " found ATAPI device by sig \n " ) ;
return ATA_DEV_ATAPI ;
}
2007-09-23 04:19:54 +00:00
if ( ( tf - > lbam = = 0x69 ) & & ( tf - > lbah = = 0x96 ) ) {
DPRINTK ( " found PMP device by sig \n " ) ;
return ATA_DEV_PMP ;
}
if ( ( tf - > lbam = = 0x3c ) & & ( tf - > lbah = = 0xc3 ) ) {
2009-04-14 21:21:10 +00:00
DPRINTK ( " found SEMB device by sig (could be ATA device) \n " ) ;
return ATA_DEV_SEMB ;
2007-09-23 04:19:54 +00:00
}
2005-04-16 22:20:36 +00:00
DPRINTK ( " unknown device \n " ) ;
return ATA_DEV_UNKNOWN ;
}
/**
2006-02-13 01:02:46 +00:00
* ata_id_string - Convert IDENTIFY DEVICE page into string
2005-04-16 22:20:36 +00:00
* @ id : IDENTIFY DEVICE results we will examine
* @ s : string into which data is output
* @ ofs : offset into identify device page
* @ len : length of string to return . must be an even number .
*
* The strings in the IDENTIFY DEVICE page are broken up into
* 16 - bit chunks . Run through the string , and output each
* 8 - bit chunk linearly , regardless of platform .
*
* LOCKING :
* caller .
*/
2006-02-13 01:02:46 +00:00
void ata_id_string ( const u16 * id , unsigned char * s ,
unsigned int ofs , unsigned int len )
2005-04-16 22:20:36 +00:00
{
unsigned int c ;
2008-07-24 16:16:06 +00:00
BUG_ON ( len & 1 ) ;
2005-04-16 22:20:36 +00:00
while ( len > 0 ) {
c = id [ ofs ] > > 8 ;
* s = c ;
s + + ;
c = id [ ofs ] & 0xff ;
* s = c ;
s + + ;
ofs + + ;
len - = 2 ;
}
}
2006-02-12 13:47:04 +00:00
/**
2006-02-13 01:02:46 +00:00
* ata_id_c_string - Convert IDENTIFY DEVICE page into C string
2006-02-12 13:47:04 +00:00
* @ id : IDENTIFY DEVICE results we will examine
* @ s : string into which data is output
* @ ofs : offset into identify device page
* @ len : length of string to return . must be an odd number .
*
2006-02-13 01:02:46 +00:00
* This function is identical to ata_id_string except that it
2006-02-12 13:47:04 +00:00
* trims trailing spaces and terminates the resulting string with
* null . @ len must be actual maximum length ( even number ) + 1.
*
* LOCKING :
* caller .
*/
2006-02-13 01:02:46 +00:00
void ata_id_c_string ( const u16 * id , unsigned char * s ,
unsigned int ofs , unsigned int len )
2006-02-12 13:47:04 +00:00
{
unsigned char * p ;
2006-02-13 01:02:46 +00:00
ata_id_string ( id , s , ofs , len - 1 ) ;
2006-02-12 13:47:04 +00:00
p = s + strnlen ( s , len - 1 ) ;
while ( p > s & & p [ - 1 ] = = ' ' )
p - - ;
* p = ' \0 ' ;
}
2005-06-02 22:17:13 +00:00
2007-09-03 03:31:58 +00:00
static u64 ata_id_n_sectors ( const u16 * id )
{
if ( ata_id_has_lba ( id ) ) {
if ( ata_id_has_lba48 ( id ) )
2009-02-17 02:15:08 +00:00
return ata_id_u64 ( id , ATA_ID_LBA_CAPACITY_2 ) ;
2007-09-03 03:31:58 +00:00
else
2009-02-17 02:15:08 +00:00
return ata_id_u32 ( id , ATA_ID_LBA_CAPACITY ) ;
2007-09-03 03:31:58 +00:00
} else {
if ( ata_id_current_chs_valid ( id ) )
2009-02-17 02:15:08 +00:00
return id [ ATA_ID_CUR_CYLS ] * id [ ATA_ID_CUR_HEADS ] *
id [ ATA_ID_CUR_SECTORS ] ;
2007-09-03 03:31:58 +00:00
else
2009-02-17 02:15:08 +00:00
return id [ ATA_ID_CYLS ] * id [ ATA_ID_HEADS ] *
id [ ATA_ID_SECTORS ] ;
2007-09-03 03:31:58 +00:00
}
}
2008-03-27 10:14:23 +00:00
u64 ata_tf_to_lba48 ( const struct ata_taskfile * tf )
2007-04-10 23:23:13 +00:00
{
u64 sectors = 0 ;
sectors | = ( ( u64 ) ( tf - > hob_lbah & 0xff ) ) < < 40 ;
sectors | = ( ( u64 ) ( tf - > hob_lbam & 0xff ) ) < < 32 ;
2008-10-28 23:52:20 +00:00
sectors | = ( ( u64 ) ( tf - > hob_lbal & 0xff ) ) < < 24 ;
2007-04-10 23:23:13 +00:00
sectors | = ( tf - > lbah & 0xff ) < < 16 ;
sectors | = ( tf - > lbam & 0xff ) < < 8 ;
sectors | = ( tf - > lbal & 0xff ) ;
2008-03-27 10:14:23 +00:00
return sectors ;
2007-04-10 23:23:13 +00:00
}
2008-03-27 10:14:23 +00:00
u64 ata_tf_to_lba ( const struct ata_taskfile * tf )
2007-04-10 23:23:13 +00:00
{
u64 sectors = 0 ;
sectors | = ( tf - > device & 0x0f ) < < 24 ;
sectors | = ( tf - > lbah & 0xff ) < < 16 ;
sectors | = ( tf - > lbam & 0xff ) < < 8 ;
sectors | = ( tf - > lbal & 0xff ) ;
2008-03-27 10:14:23 +00:00
return sectors ;
2007-04-10 23:23:13 +00:00
}
/**
2007-09-03 03:32:30 +00:00
* ata_read_native_max_address - Read native max address
* @ dev : target device
* @ max_sectors : out parameter for the result native max address
2007-04-10 23:23:13 +00:00
*
2007-09-03 03:32:30 +00:00
* Perform an LBA48 or LBA28 native size query upon the device in
* question .
2007-04-10 23:23:13 +00:00
*
2007-09-03 03:32:30 +00:00
* RETURNS :
* 0 on success , - EACCES if command is aborted by the drive .
* - EIO on other errors .
2007-04-10 23:23:13 +00:00
*/
2007-09-03 03:32:30 +00:00
static int ata_read_native_max_address ( struct ata_device * dev , u64 * max_sectors )
2007-04-10 23:23:13 +00:00
{
2007-09-03 03:32:30 +00:00
unsigned int err_mask ;
2007-04-10 23:23:13 +00:00
struct ata_taskfile tf ;
2007-09-03 03:32:30 +00:00
int lba48 = ata_id_has_lba48 ( dev - > id ) ;
2007-04-10 23:23:13 +00:00
ata_tf_init ( dev , & tf ) ;
2007-09-03 03:32:30 +00:00
/* always clear all address registers */
2007-04-10 23:23:13 +00:00
tf . flags | = ATA_TFLAG_DEVICE | ATA_TFLAG_ISADDR ;
2007-09-03 03:32:30 +00:00
if ( lba48 ) {
tf . command = ATA_CMD_READ_NATIVE_MAX_EXT ;
tf . flags | = ATA_TFLAG_LBA48 ;
} else
tf . command = ATA_CMD_READ_NATIVE_MAX ;
2007-04-10 23:23:13 +00:00
tf . protocol | = ATA_PROT_NODATA ;
2007-09-03 03:32:30 +00:00
tf . device | = ATA_LBA ;
2007-10-09 06:05:44 +00:00
err_mask = ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2007-09-03 03:32:30 +00:00
if ( err_mask ) {
ata_dev_printk ( dev , KERN_WARNING , " failed to read native "
" max address (err_mask=0x%x) \n " , err_mask ) ;
if ( err_mask = = AC_ERR_DEV & & ( tf . feature & ATA_ABORTED ) )
return - EACCES ;
return - EIO ;
}
2007-04-10 23:23:13 +00:00
2007-09-03 03:32:30 +00:00
if ( lba48 )
2008-03-27 10:14:23 +00:00
* max_sectors = ata_tf_to_lba48 ( & tf ) + 1 ;
2007-09-03 03:32:30 +00:00
else
2008-03-27 10:14:23 +00:00
* max_sectors = ata_tf_to_lba ( & tf ) + 1 ;
2007-10-19 10:42:56 +00:00
if ( dev - > horkage & ATA_HORKAGE_HPA_SIZE )
2007-09-29 08:06:48 +00:00
( * max_sectors ) - - ;
2007-09-03 03:32:30 +00:00
return 0 ;
2007-04-10 23:23:13 +00:00
}
/**
2007-09-03 03:32:30 +00:00
* ata_set_max_sectors - Set max sectors
* @ dev : target device
2007-05-02 00:35:55 +00:00
* @ new_sectors : new max sectors value to set for the device
2007-04-10 23:23:13 +00:00
*
2007-09-03 03:32:30 +00:00
* Set max sectors of @ dev to @ new_sectors .
*
* RETURNS :
* 0 on success , - EACCES if command is aborted or denied ( due to
* previous non - volatile SET_MAX ) by the drive . - EIO on other
* errors .
2007-04-10 23:23:13 +00:00
*/
2007-09-03 03:32:57 +00:00
static int ata_set_max_sectors ( struct ata_device * dev , u64 new_sectors )
2007-04-10 23:23:13 +00:00
{
2007-09-03 03:32:30 +00:00
unsigned int err_mask ;
2007-04-10 23:23:13 +00:00
struct ata_taskfile tf ;
2007-09-03 03:32:30 +00:00
int lba48 = ata_id_has_lba48 ( dev - > id ) ;
2007-04-10 23:23:13 +00:00
new_sectors - - ;
ata_tf_init ( dev , & tf ) ;
tf . flags | = ATA_TFLAG_DEVICE | ATA_TFLAG_ISADDR ;
2007-09-03 03:32:30 +00:00
if ( lba48 ) {
tf . command = ATA_CMD_SET_MAX_EXT ;
tf . flags | = ATA_TFLAG_LBA48 ;
tf . hob_lbal = ( new_sectors > > 24 ) & 0xff ;
tf . hob_lbam = ( new_sectors > > 32 ) & 0xff ;
tf . hob_lbah = ( new_sectors > > 40 ) & 0xff ;
2007-09-21 11:07:14 +00:00
} else {
2007-09-03 03:32:30 +00:00
tf . command = ATA_CMD_SET_MAX ;
2007-09-21 11:07:14 +00:00
tf . device | = ( new_sectors > > 24 ) & 0xf ;
}
2007-04-10 23:23:13 +00:00
tf . protocol | = ATA_PROT_NODATA ;
2007-09-03 03:32:30 +00:00
tf . device | = ATA_LBA ;
2007-04-10 23:23:13 +00:00
tf . lbal = ( new_sectors > > 0 ) & 0xff ;
tf . lbam = ( new_sectors > > 8 ) & 0xff ;
tf . lbah = ( new_sectors > > 16 ) & 0xff ;
2007-10-09 06:05:44 +00:00
err_mask = ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2007-09-03 03:32:30 +00:00
if ( err_mask ) {
ata_dev_printk ( dev , KERN_WARNING , " failed to set "
" max address (err_mask=0x%x) \n " , err_mask ) ;
if ( err_mask = = AC_ERR_DEV & &
( tf . feature & ( ATA_ABORTED | ATA_IDNF ) ) )
return - EACCES ;
return - EIO ;
}
return 0 ;
2007-04-10 23:23:13 +00:00
}
/**
* ata_hpa_resize - Resize a device with an HPA set
* @ dev : Device to resize
*
* Read the size of an LBA28 or LBA48 disk with HPA features and resize
* it if required to the full size of the media . The caller must check
* the drive has the HPA feature set enabled .
2007-09-03 03:32:57 +00:00
*
* RETURNS :
* 0 on success , - errno on failure .
2007-04-10 23:23:13 +00:00
*/
2007-09-03 03:32:57 +00:00
static int ata_hpa_resize ( struct ata_device * dev )
2007-04-10 23:23:13 +00:00
{
2007-09-03 03:32:57 +00:00
struct ata_eh_context * ehc = & dev - > link - > eh_context ;
int print_info = ehc - > i . flags & ATA_EHI_PRINTINFO ;
2010-04-05 01:33:13 +00:00
bool unlock_hpa = ata_ignore_hpa | | dev - > flags & ATA_DFLAG_UNLOCK_HPA ;
2007-09-03 03:32:57 +00:00
u64 sectors = ata_id_n_sectors ( dev - > id ) ;
u64 native_sectors ;
2007-09-03 03:32:30 +00:00
int rc ;
2007-05-22 00:14:23 +00:00
2007-09-03 03:32:57 +00:00
/* do we need to do it? */
if ( dev - > class ! = ATA_DEV_ATA | |
! ata_id_has_lba ( dev - > id ) | | ! ata_id_hpa_enabled ( dev - > id ) | |
( dev - > horkage & ATA_HORKAGE_BROKEN_HPA ) )
2007-09-03 03:32:30 +00:00
return 0 ;
2007-04-10 23:23:13 +00:00
2007-09-03 03:32:57 +00:00
/* read native max address */
rc = ata_read_native_max_address ( dev , & native_sectors ) ;
if ( rc ) {
2008-03-23 12:05:15 +00:00
/* If device aborted the command or HPA isn't going to
* be unlocked , skip HPA resizing .
2007-09-03 03:32:57 +00:00
*/
2010-04-05 01:33:13 +00:00
if ( rc = = - EACCES | | ! unlock_hpa ) {
2007-09-03 03:32:57 +00:00
ata_dev_printk ( dev , KERN_WARNING , " HPA support seems "
2008-03-23 12:05:15 +00:00
" broken, skipping HPA handling \n " ) ;
2007-09-03 03:32:57 +00:00
dev - > horkage | = ATA_HORKAGE_BROKEN_HPA ;
/* we can continue if device aborted the command */
if ( rc = = - EACCES )
rc = 0 ;
2007-04-10 23:23:13 +00:00
}
2007-06-25 11:45:54 +00:00
2007-09-03 03:32:57 +00:00
return rc ;
}
2009-07-15 08:11:41 +00:00
dev - > n_native_sectors = native_sectors ;
2007-09-03 03:32:57 +00:00
/* nothing to do? */
2010-04-05 01:33:13 +00:00
if ( native_sectors < = sectors | | ! unlock_hpa ) {
2007-09-03 03:32:57 +00:00
if ( ! print_info | | native_sectors = = sectors )
return 0 ;
if ( native_sectors > sectors )
ata_dev_printk ( dev , KERN_INFO ,
" HPA detected: current %llu, native %llu \n " ,
( unsigned long long ) sectors ,
( unsigned long long ) native_sectors ) ;
else if ( native_sectors < sectors )
ata_dev_printk ( dev , KERN_WARNING ,
" native sectors (%llu) is smaller than "
" sectors (%llu) \n " ,
( unsigned long long ) native_sectors ,
( unsigned long long ) sectors ) ;
return 0 ;
}
/* let's unlock HPA */
rc = ata_set_max_sectors ( dev , native_sectors ) ;
if ( rc = = - EACCES ) {
/* if device aborted the command, skip HPA resizing */
ata_dev_printk ( dev , KERN_WARNING , " device aborted resize "
" (%llu -> %llu), skipping HPA handling \n " ,
( unsigned long long ) sectors ,
( unsigned long long ) native_sectors ) ;
dev - > horkage | = ATA_HORKAGE_BROKEN_HPA ;
return 0 ;
} else if ( rc )
return rc ;
/* re-read IDENTIFY data */
rc = ata_dev_reread_id ( dev , 0 ) ;
if ( rc ) {
ata_dev_printk ( dev , KERN_ERR , " failed to re-read IDENTIFY "
" data after HPA resizing \n " ) ;
return rc ;
}
if ( print_info ) {
u64 new_sectors = ata_id_n_sectors ( dev - > id ) ;
ata_dev_printk ( dev , KERN_INFO ,
" HPA unlocked: %llu -> %llu, native %llu \n " ,
( unsigned long long ) sectors ,
( unsigned long long ) new_sectors ,
( unsigned long long ) native_sectors ) ;
}
return 0 ;
2007-04-10 23:23:13 +00:00
}
2005-04-16 22:20:36 +00:00
/**
* ata_dump_id - IDENTIFY DEVICE info debugging output
2006-02-12 13:47:05 +00:00
* @ id : IDENTIFY DEVICE page to dump
2005-04-16 22:20:36 +00:00
*
2006-02-12 13:47:05 +00:00
* Dump selected 16 - bit words from the given IDENTIFY DEVICE
* page .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
* caller .
*/
2006-02-12 13:47:05 +00:00
static inline void ata_dump_id ( const u16 * id )
2005-04-16 22:20:36 +00:00
{
DPRINTK ( " 49==0x%04x "
" 53==0x%04x "
" 63==0x%04x "
" 64==0x%04x "
" 75==0x%04x \n " ,
2006-02-12 13:47:05 +00:00
id [ 49 ] ,
id [ 53 ] ,
id [ 63 ] ,
id [ 64 ] ,
id [ 75 ] ) ;
2005-04-16 22:20:36 +00:00
DPRINTK ( " 80==0x%04x "
" 81==0x%04x "
" 82==0x%04x "
" 83==0x%04x "
" 84==0x%04x \n " ,
2006-02-12 13:47:05 +00:00
id [ 80 ] ,
id [ 81 ] ,
id [ 82 ] ,
id [ 83 ] ,
id [ 84 ] ) ;
2005-04-16 22:20:36 +00:00
DPRINTK ( " 88==0x%04x "
" 93==0x%04x \n " ,
2006-02-12 13:47:05 +00:00
id [ 88 ] ,
id [ 93 ] ) ;
2005-04-16 22:20:36 +00:00
}
2006-03-05 19:31:56 +00:00
/**
* ata_id_xfermask - Compute xfermask from the given IDENTIFY data
* @ id : IDENTIFY data to compute xfer mask from
*
* Compute the xfermask for this device . This is not as trivial
* as it seems if we must consider early devices correctly .
*
* FIXME : pre IDE drive timing ( do we care ? ) .
*
* LOCKING :
* None .
*
* RETURNS :
* Computed xfermask
*/
2007-11-27 10:43:42 +00:00
unsigned long ata_id_xfermask ( const u16 * id )
2006-03-05 19:31:56 +00:00
{
2007-11-27 10:43:42 +00:00
unsigned long pio_mask , mwdma_mask , udma_mask ;
2006-03-05 19:31:56 +00:00
/* Usual case. Word 53 indicates word 64 is valid */
if ( id [ ATA_ID_FIELD_VALID ] & ( 1 < < 1 ) ) {
pio_mask = id [ ATA_ID_PIO_MODES ] & 0x03 ;
pio_mask < < = 3 ;
pio_mask | = 0x7 ;
} else {
/* If word 64 isn't valid then Word 51 high byte holds
* the PIO timing number for the maximum . Turn it into
* a mask .
*/
2007-01-29 12:28:47 +00:00
u8 mode = ( id [ ATA_ID_OLD_PIO_MODES ] > > 8 ) & 0xFF ;
2006-09-29 17:26:47 +00:00
if ( mode < 5 ) /* Valid PIO range */
2007-10-19 10:42:56 +00:00
pio_mask = ( 2 < < mode ) - 1 ;
2006-09-29 17:26:47 +00:00
else
pio_mask = 1 ;
2006-03-05 19:31:56 +00:00
/* But wait.. there's more. Design your standards by
* committee and you too can get a free iordy field to
* process . However its the speeds not the modes that
* are supported . . . Note drivers using the timing API
* will get this right anyway
*/
}
mwdma_mask = id [ ATA_ID_MWDMA_MODES ] & 0x07 ;
2006-03-12 03:34:35 +00:00
2006-08-10 17:52:12 +00:00
if ( ata_id_is_cfa ( id ) ) {
/*
* Process compact flash extended modes
*/
2009-04-13 16:50:00 +00:00
int pio = ( id [ ATA_ID_CFA_MODES ] > > 0 ) & 0x7 ;
int dma = ( id [ ATA_ID_CFA_MODES ] > > 3 ) & 0x7 ;
2006-08-10 17:52:12 +00:00
if ( pio )
pio_mask | = ( 1 < < 5 ) ;
if ( pio > 1 )
pio_mask | = ( 1 < < 6 ) ;
if ( dma )
mwdma_mask | = ( 1 < < 3 ) ;
if ( dma > 1 )
mwdma_mask | = ( 1 < < 4 ) ;
}
2006-03-12 03:34:35 +00:00
udma_mask = 0 ;
if ( id [ ATA_ID_FIELD_VALID ] & ( 1 < < 2 ) )
udma_mask = id [ ATA_ID_UDMA_MODES ] & 0xff ;
2006-03-05 19:31:56 +00:00
return ata_pack_xfermask ( pio_mask , mwdma_mask , udma_mask ) ;
}
2007-01-03 23:09:36 +00:00
static void ata_qc_complete_internal ( struct ata_queued_cmd * qc )
2005-12-13 05:48:31 +00:00
{
2006-01-23 04:09:36 +00:00
struct completion * waiting = qc - > private_data ;
2005-12-13 05:48:31 +00:00
complete ( waiting ) ;
}
/**
2006-11-14 13:47:09 +00:00
* ata_exec_internal_sg - execute libata internal command
2005-12-13 05:48:31 +00:00
* @ dev : Device to which the command is sent
* @ tf : Taskfile registers for the command and the result
2006-04-02 09:51:53 +00:00
* @ cdb : CDB for packet command
2005-12-13 05:48:31 +00:00
* @ dma_dir : Data tranfer direction of the command
2007-10-18 21:12:26 +00:00
* @ sgl : sg list for the data buffer of the command
2006-11-14 13:47:09 +00:00
* @ n_elem : Number of sg entries
2007-10-09 06:05:44 +00:00
* @ timeout : Timeout in msecs ( 0 for default )
2005-12-13 05:48:31 +00:00
*
* Executes libata internal command with timeout . @ tf contains
* command on entry and result on return . Timeout and error
* conditions are reported via return value . No recovery action
* is taken after a command times out . It ' s caller ' s duty to
* clean up after timeout .
*
* LOCKING :
* None . Should be called with kernel context , might sleep .
2006-06-12 05:09:49 +00:00
*
* RETURNS :
* Zero on success , AC_ERR_ * mask on failure
2005-12-13 05:48:31 +00:00
*/
2006-11-14 13:47:09 +00:00
unsigned ata_exec_internal_sg ( struct ata_device * dev ,
struct ata_taskfile * tf , const u8 * cdb ,
2007-10-16 09:14:12 +00:00
int dma_dir , struct scatterlist * sgl ,
2007-10-09 06:05:44 +00:00
unsigned int n_elem , unsigned long timeout )
2005-12-13 05:48:31 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_link * link = dev - > link ;
struct ata_port * ap = link - > ap ;
2005-12-13 05:48:31 +00:00
u8 command = tf - > command ;
2008-05-19 17:17:54 +00:00
int auto_timeout = 0 ;
2005-12-13 05:48:31 +00:00
struct ata_queued_cmd * qc ;
2006-05-15 11:58:02 +00:00
unsigned int tag , preempted_tag ;
2006-05-15 12:03:43 +00:00
u32 preempted_sactive , preempted_qc_active ;
2007-09-23 04:14:12 +00:00
int preempted_nr_active_links ;
2006-07-03 07:25:26 +00:00
DECLARE_COMPLETION_ONSTACK ( wait ) ;
2005-12-13 05:48:31 +00:00
unsigned long flags ;
2006-01-23 04:09:36 +00:00
unsigned int err_mask ;
2006-05-15 11:58:14 +00:00
int rc ;
2005-12-13 05:48:31 +00:00
2006-06-23 03:46:10 +00:00
spin_lock_irqsave ( ap - > lock , flags ) ;
2005-12-13 05:48:31 +00:00
2006-05-15 11:58:09 +00:00
/* no internal command while frozen */
2006-06-28 16:29:30 +00:00
if ( ap - > pflags & ATA_PFLAG_FROZEN ) {
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2006-05-15 11:58:09 +00:00
return AC_ERR_SYSTEM ;
}
2006-05-15 11:58:02 +00:00
/* initialize internal qc */
2005-12-13 05:48:31 +00:00
2006-05-15 11:58:02 +00:00
/* XXX: Tag 0 is used for drivers with legacy EH as some
* drivers choke if any other tag is given . This breaks
* ata_tag_internal ( ) test for those drivers . Don ' t use new
* EH stuff without converting to it .
*/
if ( ap - > ops - > error_handler )
tag = ATA_TAG_INTERNAL ;
else
tag = 0 ;
2008-11-10 05:48:21 +00:00
if ( test_and_set_bit ( tag , & ap - > qc_allocated ) )
BUG ( ) ;
2006-05-15 11:58:03 +00:00
qc = __ata_qc_from_tag ( ap , tag ) ;
2006-05-15 11:58:02 +00:00
qc - > tag = tag ;
qc - > scsicmd = NULL ;
qc - > ap = ap ;
qc - > dev = dev ;
ata_qc_reinit ( qc ) ;
2007-08-06 09:36:22 +00:00
preempted_tag = link - > active_tag ;
preempted_sactive = link - > sactive ;
2006-05-15 12:03:43 +00:00
preempted_qc_active = ap - > qc_active ;
2007-09-23 04:14:12 +00:00
preempted_nr_active_links = ap - > nr_active_links ;
2007-08-06 09:36:22 +00:00
link - > active_tag = ATA_TAG_POISON ;
link - > sactive = 0 ;
2006-05-15 12:03:43 +00:00
ap - > qc_active = 0 ;
2007-09-23 04:14:12 +00:00
ap - > nr_active_links = 0 ;
2006-05-15 11:58:02 +00:00
/* prepare & issue qc */
2005-12-13 05:48:31 +00:00
qc - > tf = * tf ;
2006-04-02 09:51:53 +00:00
if ( cdb )
memcpy ( qc - > cdb , cdb , ATAPI_CDB_LEN ) ;
2006-05-15 11:57:40 +00:00
qc - > flags | = ATA_QCFLAG_RESULT_TF ;
2005-12-13 05:48:31 +00:00
qc - > dma_dir = dma_dir ;
if ( dma_dir ! = DMA_NONE ) {
2006-11-14 13:47:09 +00:00
unsigned int i , buflen = 0 ;
2007-10-16 09:14:12 +00:00
struct scatterlist * sg ;
2006-11-14 13:47:09 +00:00
2007-10-16 09:14:12 +00:00
for_each_sg ( sgl , sg , n_elem , i )
buflen + = sg - > length ;
2006-11-14 13:47:09 +00:00
2007-10-16 09:14:12 +00:00
ata_sg_init ( qc , sgl , n_elem ) ;
2007-01-30 17:32:26 +00:00
qc - > nbytes = buflen ;
2005-12-13 05:48:31 +00:00
}
2006-01-23 04:09:36 +00:00
qc - > private_data = & wait ;
2005-12-13 05:48:31 +00:00
qc - > complete_fn = ata_qc_complete_internal ;
2006-03-31 11:41:11 +00:00
ata_qc_issue ( qc ) ;
2005-12-13 05:48:31 +00:00
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2005-12-13 05:48:31 +00:00
2008-05-19 17:17:54 +00:00
if ( ! timeout ) {
if ( ata_probe_timeout )
timeout = ata_probe_timeout * 1000 ;
else {
timeout = ata_internal_cmd_timeout ( dev , command ) ;
auto_timeout = 1 ;
}
}
2007-10-09 06:05:44 +00:00
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
if ( ap - > ops - > error_handler )
ata_eh_release ( ap ) ;
2007-10-09 06:05:44 +00:00
rc = wait_for_completion_timeout ( & wait , msecs_to_jiffies ( timeout ) ) ;
2006-05-15 11:58:14 +00:00
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
if ( ap - > ops - > error_handler )
ata_eh_acquire ( ap ) ;
2010-05-10 19:41:38 +00:00
ata_sff_flush_pio_task ( ap ) ;
2006-03-14 03:19:04 +00:00
2006-05-15 11:58:14 +00:00
if ( ! rc ) {
2006-06-23 03:46:10 +00:00
spin_lock_irqsave ( ap - > lock , flags ) ;
2005-12-13 05:48:31 +00:00
/* We're racing with irq here. If we lose, the
* following test prevents us from completing the qc
2006-05-15 11:58:14 +00:00
* twice . If we win , the port is frozen and will be
* cleaned up by - > post_internal_cmd ( ) .
2005-12-13 05:48:31 +00:00
*/
2006-01-23 04:09:36 +00:00
if ( qc - > flags & ATA_QCFLAG_ACTIVE ) {
2006-05-15 11:58:14 +00:00
qc - > err_mask | = AC_ERR_TIMEOUT ;
if ( ap - > ops - > error_handler )
ata_port_freeze ( ap ) ;
else
ata_qc_complete ( qc ) ;
2006-05-15 11:57:56 +00:00
2006-06-23 06:29:08 +00:00
if ( ata_msg_warn ( ap ) )
ata_dev_printk ( dev , KERN_WARNING ,
2006-06-25 11:00:35 +00:00
" qc timeout (cmd 0x%x) \n " , command ) ;
2005-12-13 05:48:31 +00:00
}
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2005-12-13 05:48:31 +00:00
}
2006-05-15 11:58:14 +00:00
/* do post_internal_cmd */
if ( ap - > ops - > post_internal_cmd )
ap - > ops - > post_internal_cmd ( qc ) ;
2007-03-20 06:24:11 +00:00
/* perform minimal error analysis */
if ( qc - > flags & ATA_QCFLAG_FAILED ) {
if ( qc - > result_tf . command & ( ATA_ERR | ATA_DF ) )
qc - > err_mask | = AC_ERR_DEV ;
if ( ! qc - > err_mask )
qc - > err_mask | = AC_ERR_OTHER ;
if ( qc - > err_mask & ~ AC_ERR_OTHER )
qc - > err_mask & = ~ AC_ERR_OTHER ;
2006-05-15 11:58:14 +00:00
}
2006-05-15 11:57:33 +00:00
/* finish up */
2006-06-23 03:46:10 +00:00
spin_lock_irqsave ( ap - > lock , flags ) ;
2006-05-15 11:57:33 +00:00
2006-05-15 11:57:40 +00:00
* tf = qc - > result_tf ;
2006-01-23 04:09:36 +00:00
err_mask = qc - > err_mask ;
ata_qc_free ( qc ) ;
2007-08-06 09:36:22 +00:00
link - > active_tag = preempted_tag ;
link - > sactive = preempted_sactive ;
2006-05-15 12:03:43 +00:00
ap - > qc_active = preempted_qc_active ;
2007-09-23 04:14:12 +00:00
ap - > nr_active_links = preempted_nr_active_links ;
2006-01-23 04:09:36 +00:00
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2006-05-15 11:57:33 +00:00
2008-05-19 17:17:54 +00:00
if ( ( err_mask & AC_ERR_TIMEOUT ) & & auto_timeout )
ata_internal_cmd_timed_out ( dev , command ) ;
2006-01-23 04:09:36 +00:00
return err_mask ;
2005-12-13 05:48:31 +00:00
}
2006-11-14 13:47:09 +00:00
/**
2006-12-11 17:15:31 +00:00
* ata_exec_internal - execute libata internal command
2006-11-14 13:47:09 +00:00
* @ dev : Device to which the command is sent
* @ tf : Taskfile registers for the command and the result
* @ cdb : CDB for packet command
* @ dma_dir : Data tranfer direction of the command
* @ buf : Data buffer of the command
* @ buflen : Length of data buffer
2007-10-09 06:05:44 +00:00
* @ timeout : Timeout in msecs ( 0 for default )
2006-11-14 13:47:09 +00:00
*
* Wrapper around ata_exec_internal_sg ( ) which takes simple
* buffer instead of sg list .
*
* LOCKING :
* None . Should be called with kernel context , might sleep .
*
* RETURNS :
* Zero on success , AC_ERR_ * mask on failure
*/
unsigned ata_exec_internal ( struct ata_device * dev ,
struct ata_taskfile * tf , const u8 * cdb ,
2007-10-09 06:05:44 +00:00
int dma_dir , void * buf , unsigned int buflen ,
unsigned long timeout )
2006-11-14 13:47:09 +00:00
{
2006-12-11 17:15:31 +00:00
struct scatterlist * psg = NULL , sg ;
unsigned int n_elem = 0 ;
2006-11-14 13:47:09 +00:00
2006-12-11 17:15:31 +00:00
if ( dma_dir ! = DMA_NONE ) {
WARN_ON ( ! buf ) ;
sg_init_one ( & sg , buf , buflen ) ;
psg = & sg ;
n_elem + + ;
}
2006-11-14 13:47:09 +00:00
2007-10-09 06:05:44 +00:00
return ata_exec_internal_sg ( dev , tf , cdb , dma_dir , psg , n_elem ,
timeout ) ;
2006-11-14 13:47:09 +00:00
}
2006-06-24 11:30:19 +00:00
/**
* ata_do_simple_cmd - execute simple internal command
* @ dev : Device to which the command is sent
* @ cmd : Opcode to execute
*
* Execute a ' simple ' command , that only consists of the opcode
* ' cmd ' itself , without filling any other registers
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*
* RETURNS :
* Zero on success , AC_ERR_ * mask on failure
2006-06-24 11:30:19 +00:00
*/
2006-06-24 11:30:19 +00:00
unsigned int ata_do_simple_cmd ( struct ata_device * dev , u8 cmd )
2006-06-24 11:30:19 +00:00
{
struct ata_taskfile tf ;
ata_tf_init ( dev , & tf ) ;
tf . command = cmd ;
tf . flags | = ATA_TFLAG_DEVICE ;
tf . protocol = ATA_PROT_NODATA ;
2007-10-09 06:05:44 +00:00
return ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2006-06-24 11:30:19 +00:00
}
2006-01-09 17:18:14 +00:00
/**
* ata_pio_need_iordy - check if iordy needed
* @ adev : ATA device
*
* Check if the current speed of the device requires IORDY . Used
* by various controllers for chip configuration .
*/
unsigned int ata_pio_need_iordy ( const struct ata_device * adev )
{
2009-06-11 02:04:45 +00:00
/* Don't set IORDY if we're preparing for reset. IORDY may
* lead to controller lock up on certain controllers if the
* port is not occupied . See bko # 11703 for details .
*/
if ( adev - > link - > ap - > pflags & ATA_PFLAG_RESETTING )
return 0 ;
/* Controller doesn't support IORDY. Probably a pointless
* check as the caller should know this .
*/
2007-08-06 09:36:22 +00:00
if ( adev - > link - > ap - > flags & ATA_FLAG_NO_IORDY )
2006-01-09 17:18:14 +00:00
return 0 ;
2008-12-10 23:39:12 +00:00
/* CF spec. r4.1 Table 22 says no iordy on PIO5 and PIO6. */
if ( ata_id_is_cfa ( adev - > id )
& & ( adev - > pio_mode = = XFER_PIO_5 | | adev - > pio_mode = = XFER_PIO_6 ) )
return 0 ;
2007-03-08 23:22:59 +00:00
/* PIO3 and higher it is mandatory */
if ( adev - > pio_mode > XFER_PIO_2 )
return 1 ;
/* We turn it on when possible */
if ( ata_id_has_iordy ( adev - > id ) )
2006-01-09 17:18:14 +00:00
return 1 ;
2007-03-08 23:22:59 +00:00
return 0 ;
}
2006-03-24 14:56:57 +00:00
2007-03-08 23:22:59 +00:00
/**
* ata_pio_mask_no_iordy - Return the non IORDY mask
* @ adev : ATA device
*
* Compute the highest mode possible if we are not using iordy . Return
* - 1 if no iordy mode is available .
*/
static u32 ata_pio_mask_no_iordy ( const struct ata_device * adev )
{
2006-01-09 17:18:14 +00:00
/* If we have no drive specific rule, then PIO 2 is non IORDY */
if ( adev - > id [ ATA_ID_FIELD_VALID ] & 2 ) { /* EIDE */
2007-03-08 23:22:59 +00:00
u16 pio = adev - > id [ ATA_ID_EIDE_PIO ] ;
2006-01-09 17:18:14 +00:00
/* Is the speed faster than the drive allows non IORDY ? */
if ( pio ) {
/* This is cycle times not frequency - watch the logic! */
if ( pio > 240 ) /* PIO2 is 240nS per cycle */
2007-03-08 23:22:59 +00:00
return 3 < < ATA_SHIFT_PIO ;
return 7 < < ATA_SHIFT_PIO ;
2006-01-09 17:18:14 +00:00
}
}
2007-03-08 23:22:59 +00:00
return 3 < < ATA_SHIFT_PIO ;
2006-01-09 17:18:14 +00:00
}
2008-07-24 16:16:06 +00:00
/**
* ata_do_dev_read_id - default ID read method
* @ dev : device
* @ tf : proposed taskfile
* @ id : data buffer
*
* Issue the identify taskfile and hand back the buffer containing
* identify data . For some RAID controllers and for pre ATA devices
* this function is wrapped or replaced by the driver
*/
unsigned int ata_do_dev_read_id ( struct ata_device * dev ,
struct ata_taskfile * tf , u16 * id )
{
return ata_exec_internal ( dev , tf , NULL , DMA_FROM_DEVICE ,
id , sizeof ( id [ 0 ] ) * ATA_ID_WORDS , 0 ) ;
}
2005-04-16 22:20:36 +00:00
/**
2006-02-20 17:12:11 +00:00
* ata_dev_read_id - Read ID data from the specified device
* @ dev : target device
* @ p_class : pointer to class of the target device ( may be changed )
2006-11-10 09:08:10 +00:00
* @ flags : ATA_READID_ * flags
2006-05-15 11:57:35 +00:00
* @ id : buffer to read IDENTIFY data into
2005-04-16 22:20:36 +00:00
*
2006-02-20 17:12:11 +00:00
* Read ID data from the specified device . ATA_CMD_ID_ATA is
* performed on ATA devices and ATA_CMD_ID_ATAPI on ATAPI
2006-03-24 16:33:34 +00:00
* devices . This function also issues ATA_CMD_INIT_DEV_PARAMS
* for pre - ATA4 drives .
2005-04-16 22:20:36 +00:00
*
2007-08-08 13:27:00 +00:00
* FIXME : ATA_CMD_ID_ATA is optional for early drives and right
2007-10-19 10:42:56 +00:00
* now we abort if we hit that case .
2007-08-08 13:27:00 +00:00
*
2005-04-16 22:20:36 +00:00
* LOCKING :
2006-02-20 17:12:11 +00:00
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno otherwise .
2005-04-16 22:20:36 +00:00
*/
2006-05-31 09:27:44 +00:00
int ata_dev_read_id ( struct ata_device * dev , unsigned int * p_class ,
2006-11-10 09:08:10 +00:00
unsigned int flags , u16 * id )
2005-04-16 22:20:36 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_port * ap = dev - > link - > ap ;
2006-02-20 17:12:11 +00:00
unsigned int class = * p_class ;
2005-12-13 05:49:31 +00:00
struct ata_taskfile tf ;
2006-02-20 17:12:11 +00:00
unsigned int err_mask = 0 ;
const char * reason ;
2009-04-14 21:21:10 +00:00
bool is_semb = class = = ATA_DEV_SEMB ;
2007-05-11 12:35:29 +00:00
int may_fallback = 1 , tried_spinup = 0 ;
2006-02-20 17:12:11 +00:00
int rc ;
2005-04-16 22:20:36 +00:00
2006-06-23 06:29:08 +00:00
if ( ata_msg_ctl ( ap ) )
2008-03-06 02:24:52 +00:00
ata_dev_printk ( dev , KERN_DEBUG , " %s: ENTER \n " , __func__ ) ;
2005-04-16 22:20:36 +00:00
2008-07-24 16:16:06 +00:00
retry :
2006-05-15 11:57:53 +00:00
ata_tf_init ( dev , & tf ) ;
2005-12-13 05:49:31 +00:00
2006-02-20 17:12:11 +00:00
switch ( class ) {
2009-04-14 21:21:10 +00:00
case ATA_DEV_SEMB :
class = ATA_DEV_ATA ; /* some hard drives report SEMB sig */
2006-02-20 17:12:11 +00:00
case ATA_DEV_ATA :
2005-12-13 05:49:31 +00:00
tf . command = ATA_CMD_ID_ATA ;
2006-02-20 17:12:11 +00:00
break ;
case ATA_DEV_ATAPI :
2005-12-13 05:49:31 +00:00
tf . command = ATA_CMD_ID_ATAPI ;
2006-02-20 17:12:11 +00:00
break ;
default :
rc = - ENODEV ;
reason = " unsupported class " ;
goto err_out ;
2005-04-16 22:20:36 +00:00
}
2005-12-13 05:49:31 +00:00
tf . protocol = ATA_PROT_PIO ;
2007-02-07 20:37:41 +00:00
/* Some devices choke if TF registers contain garbage. Make
* sure those are properly initialized .
*/
tf . flags | = ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE ;
/* Device presence detection is unreliable on some
* controllers . Always poll IDENTIFY if available .
*/
tf . flags | = ATA_TFLAG_POLLING ;
2005-04-16 22:20:36 +00:00
2008-07-24 16:16:06 +00:00
if ( ap - > ops - > read_id )
err_mask = ap - > ops - > read_id ( dev , & tf , id ) ;
else
err_mask = ata_do_dev_read_id ( dev , & tf , id ) ;
2005-12-13 05:49:31 +00:00
if ( err_mask ) {
2006-12-03 12:34:13 +00:00
if ( err_mask & AC_ERR_NODEV_HINT ) {
2008-03-23 06:16:53 +00:00
ata_dev_printk ( dev , KERN_DEBUG ,
" NODEV after polling detection \n " ) ;
2006-11-10 09:08:10 +00:00
return - ENOENT ;
}
2009-04-14 21:21:10 +00:00
if ( is_semb ) {
ata_dev_printk ( dev , KERN_INFO , " IDENTIFY failed on "
" device w/ SEMB sig, disabled \n " ) ;
/* SEMB is not supported yet */
* p_class = ATA_DEV_SEMB_UNSUP ;
return 0 ;
}
2008-03-23 06:16:53 +00:00
if ( ( err_mask = = AC_ERR_DEV ) & & ( tf . feature & ATA_ABORTED ) ) {
/* Device or controller might have reported
* the wrong device class . Give a shot at the
* other IDENTIFY if the current one is
* aborted by the device .
*/
if ( may_fallback ) {
may_fallback = 0 ;
if ( class = = ATA_DEV_ATA )
class = ATA_DEV_ATAPI ;
else
class = ATA_DEV_ATA ;
goto retry ;
}
/* Control reaches here iff the device aborted
* both flavors of IDENTIFYs which happens
* sometimes with phantom devices .
*/
ata_dev_printk ( dev , KERN_DEBUG ,
" both IDENTIFYs aborted, assuming NODEV \n " ) ;
return - ENOENT ;
2007-05-11 12:35:29 +00:00
}
2006-02-20 17:12:11 +00:00
rc = - EIO ;
reason = " I/O error " ;
2005-04-16 22:20:36 +00:00
goto err_out ;
}
2010-05-23 10:59:11 +00:00
if ( dev - > horkage & ATA_HORKAGE_DUMP_ID ) {
ata_dev_printk ( dev , KERN_DEBUG , " dumping IDENTIFY data, "
" class=%d may_fallback=%d tried_spinup=%d \n " ,
class , may_fallback , tried_spinup ) ;
print_hex_dump ( KERN_DEBUG , " " , DUMP_PREFIX_OFFSET ,
16 , 2 , id , ATA_ID_WORDS * sizeof ( * id ) , true ) ;
}
2007-05-11 12:35:29 +00:00
/* Falling back doesn't make sense if ID data was read
* successfully at least once .
*/
may_fallback = 0 ;
2006-02-20 17:12:11 +00:00
swap_buf_le16 ( id , ATA_ID_WORDS ) ;
2005-04-16 22:20:36 +00:00
2006-02-20 17:12:11 +00:00
/* sanity check */
2006-09-13 03:35:49 +00:00
rc = - EINVAL ;
2007-06-07 15:13:55 +00:00
reason = " device reports invalid type " ;
2006-09-13 03:35:49 +00:00
if ( class = = ATA_DEV_ATA ) {
if ( ! ata_id_is_ata ( id ) & & ! ata_id_is_cfa ( id ) )
goto err_out ;
} else {
if ( ata_id_is_ata ( id ) )
goto err_out ;
2006-02-20 17:12:11 +00:00
}
libata: Handle drives that require a spin-up command before first access
(S)ATA drives can be configured for "power-up in standby",
a mode whereby a specific "spin up now!" command is required
before the first media access.
Currently, a drive with this feature enabled can not be used at all
with libata, and once in this mode, the drive becomes a doorstop.
The older drivers/ide subsystem at least enumerates the drive,
so that it can be woken up after the fact from a userspace HDIO_*
command, but not libata.
This patch adds support to libata for the "power-up in standby"
mode where a "spin up now!" command (SET_FEATURES) is needed.
With this, libata will recognize such drives, spin them up,
and then re-IDENTIFY them if necessary to get a full/complete
set of drive features data.
Drives in this state are determined by looking for
special values in id[2], as documented in the current ATA specs.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-04-17 22:26:07 +00:00
if ( ! tried_spinup & & ( id [ 2 ] = = 0x37c8 | | id [ 2 ] = = 0x738c ) ) {
tried_spinup = 1 ;
/*
* Drive powered - up in standby mode , and requires a specific
* SET_FEATURES spin - up subcommand before it will accept
* anything other than the original IDENTIFY command .
*/
2007-10-25 04:33:27 +00:00
err_mask = ata_dev_set_feature ( dev , SETFEATURES_SPINUP , 0 ) ;
2007-08-10 20:59:35 +00:00
if ( err_mask & & id [ 2 ] ! = 0x738c ) {
libata: Handle drives that require a spin-up command before first access
(S)ATA drives can be configured for "power-up in standby",
a mode whereby a specific "spin up now!" command is required
before the first media access.
Currently, a drive with this feature enabled can not be used at all
with libata, and once in this mode, the drive becomes a doorstop.
The older drivers/ide subsystem at least enumerates the drive,
so that it can be woken up after the fact from a userspace HDIO_*
command, but not libata.
This patch adds support to libata for the "power-up in standby"
mode where a "spin up now!" command (SET_FEATURES) is needed.
With this, libata will recognize such drives, spin them up,
and then re-IDENTIFY them if necessary to get a full/complete
set of drive features data.
Drives in this state are determined by looking for
special values in id[2], as documented in the current ATA specs.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-04-17 22:26:07 +00:00
rc = - EIO ;
reason = " SPINUP failed " ;
goto err_out ;
}
/*
* If the drive initially returned incomplete IDENTIFY info ,
* we now must reissue the IDENTIFY command .
*/
if ( id [ 2 ] = = 0x37c8 )
goto retry ;
}
2006-11-10 09:08:10 +00:00
if ( ( flags & ATA_READID_POSTRESET ) & & class = = ATA_DEV_ATA ) {
2006-02-20 17:12:11 +00:00
/*
* The exact sequence expected by certain pre - ATA4 drives is :
* SRST RESET
2007-08-08 13:27:00 +00:00
* IDENTIFY ( optional in early ATA )
* INITIALIZE DEVICE PARAMETERS ( later IDE and ATA )
2006-02-20 17:12:11 +00:00
* anything else . .
* Some drives were very specific about that exact sequence .
2007-08-08 13:27:00 +00:00
*
* Note that ATA4 says lba is mandatory so the second check
2009-12-18 20:40:42 +00:00
* should never trigger .
2006-02-20 17:12:11 +00:00
*/
if ( ata_id_major_version ( id ) < 4 | | ! ata_id_has_lba ( id ) ) {
2006-05-15 11:57:53 +00:00
err_mask = ata_dev_init_params ( dev , id [ 3 ] , id [ 6 ] ) ;
2006-02-20 17:12:11 +00:00
if ( err_mask ) {
rc = - EIO ;
reason = " INIT_DEV_PARAMS failed " ;
goto err_out ;
}
/* current CHS translation info (id[53-58]) might be
* changed . reread the identify device info .
*/
2006-11-10 09:08:10 +00:00
flags & = ~ ATA_READID_POSTRESET ;
2006-02-20 17:12:11 +00:00
goto retry ;
}
}
* p_class = class ;
2006-05-15 11:57:35 +00:00
2006-02-20 17:12:11 +00:00
return 0 ;
err_out :
2006-06-25 11:00:35 +00:00
if ( ata_msg_warn ( ap ) )
2006-06-23 06:29:08 +00:00
ata_dev_printk ( dev , KERN_WARNING , " failed to IDENTIFY "
2006-06-25 11:00:35 +00:00
" (%s, err_mask=0x%x) \n " , reason , err_mask ) ;
2006-02-20 17:12:11 +00:00
return rc ;
}
2009-01-29 11:31:36 +00:00
static int ata_do_link_spd_horkage ( struct ata_device * dev )
{
struct ata_link * plink = ata_dev_phys_link ( dev ) ;
u32 target , target_limit ;
if ( ! sata_scr_valid ( plink ) )
return 0 ;
if ( dev - > horkage & ATA_HORKAGE_1_5_GBPS )
target = 1 ;
else
return 0 ;
target_limit = ( 1 < < target ) - 1 ;
/* if already on stricter limit, no need to push further */
if ( plink - > sata_spd_limit < = target_limit )
return 0 ;
plink - > sata_spd_limit = target_limit ;
/* Request another EH round by returning -EAGAIN if link is
* going faster than the target speed . Forward progress is
* guaranteed by setting sata_spd_limit to target_limit above .
*/
if ( plink - > sata_spd > target ) {
ata_dev_printk ( dev , KERN_INFO ,
" applying link speed limit horkage to %s \n " ,
sata_spd_string ( target ) ) ;
return - EAGAIN ;
}
return 0 ;
}
2006-05-15 11:57:53 +00:00
static inline u8 ata_dev_knobble ( struct ata_device * dev )
2006-03-01 07:09:36 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_port * ap = dev - > link - > ap ;
2008-08-27 13:23:18 +00:00
if ( ata_dev_blacklisted ( dev ) & ATA_HORKAGE_BRIDGE_OK )
return 0 ;
2007-08-06 09:36:22 +00:00
return ( ( ap - > cbl = = ATA_CBL_SATA ) & & ( ! ata_id_is_sata ( dev - > id ) ) ) ;
2006-03-01 07:09:36 +00:00
}
2009-07-27 01:24:35 +00:00
static int ata_dev_config_ncq ( struct ata_device * dev ,
2006-05-15 12:03:48 +00:00
char * desc , size_t desc_sz )
{
2007-08-06 09:36:22 +00:00
struct ata_port * ap = dev - > link - > ap ;
2006-05-15 12:03:48 +00:00
int hdepth = 0 , ddepth = ata_id_queue_depth ( dev - > id ) ;
2009-07-27 01:24:35 +00:00
unsigned int err_mask ;
char * aa_desc = " " ;
2006-05-15 12:03:48 +00:00
if ( ! ata_id_has_ncq ( dev - > id ) ) {
desc [ 0 ] = ' \0 ' ;
2009-07-27 01:24:35 +00:00
return 0 ;
2006-05-15 12:03:48 +00:00
}
2007-07-05 04:31:27 +00:00
if ( dev - > horkage & ATA_HORKAGE_NONCQ ) {
2006-10-28 02:08:46 +00:00
snprintf ( desc , desc_sz , " NCQ (not used) " ) ;
2009-07-27 01:24:35 +00:00
return 0 ;
2006-10-28 02:08:46 +00:00
}
2006-05-15 12:03:48 +00:00
if ( ap - > flags & ATA_FLAG_NCQ ) {
2006-08-24 07:19:22 +00:00
hdepth = min ( ap - > scsi_host - > can_queue , ATA_MAX_QUEUE - 1 ) ;
2006-05-15 12:03:48 +00:00
dev - > flags | = ATA_DFLAG_NCQ ;
}
2009-07-27 01:24:35 +00:00
if ( ! ( dev - > horkage & ATA_HORKAGE_BROKEN_FPDMA_AA ) & &
( ap - > flags & ATA_FLAG_FPDMA_AA ) & &
ata_id_has_fpdma_aa ( dev - > id ) ) {
err_mask = ata_dev_set_feature ( dev , SETFEATURES_SATA_ENABLE ,
SATA_FPDMA_AA ) ;
if ( err_mask ) {
ata_dev_printk ( dev , KERN_ERR , " failed to enable AA "
" (error_mask=0x%x) \n " , err_mask ) ;
if ( err_mask ! = AC_ERR_DEV ) {
dev - > horkage | = ATA_HORKAGE_BROKEN_FPDMA_AA ;
return - EIO ;
}
} else
aa_desc = " , AA " ;
}
2006-05-15 12:03:48 +00:00
if ( hdepth > = ddepth )
2009-07-27 01:24:35 +00:00
snprintf ( desc , desc_sz , " NCQ (depth %d)%s " , ddepth , aa_desc ) ;
2006-05-15 12:03:48 +00:00
else
2009-07-27 01:24:35 +00:00
snprintf ( desc , desc_sz , " NCQ (depth %d/%d)%s " , hdepth ,
ddepth , aa_desc ) ;
return 0 ;
2006-05-15 12:03:48 +00:00
}
2006-02-20 17:12:11 +00:00
/**
2006-03-01 07:09:35 +00:00
* ata_dev_configure - Configure the specified ATA / ATAPI device
* @ dev : Target device to configure
*
* Configure @ dev according to @ dev - > id . Generic and low - level
* driver specific fixups are also applied .
2006-02-20 17:12:11 +00:00
*
* LOCKING :
2006-03-01 07:09:35 +00:00
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno otherwise
2006-02-20 17:12:11 +00:00
*/
2006-11-01 09:38:52 +00:00
int ata_dev_configure ( struct ata_device * dev )
2006-02-20 17:12:11 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_port * ap = dev - > link - > ap ;
struct ata_eh_context * ehc = & dev - > link - > eh_context ;
libata: reimplement ACPI invocation
This patch reimplements ACPI invocation such that, instead of
exporting ACPI details to the rest of libata, ACPI event handlers -
ata_acpi_on_resume() and ata_acpi_on_devcfg() - are used. These two
functions are responsible for determining whether specific ACPI method
is used and when.
On resume, _GTF is scheduled by setting ATA_DFLAG_ACPI_PENDING device
flag. This is done this way to avoid performing the action on wrong
device device (device swapping while suspended).
On every ata_dev_configure(), ata_acpi_on_devcfg() is called, which
performs _SDD and _GTF. _GTF is performed only after resuming and, if
SATA, hardreset as the ACPI spec specifies. As _GTF may contain
arbitrary commands, IDENTIFY page is re-read after _GTF taskfiles are
executed.
If one of ACPI methods fails, ata_acpi_on_devcfg() retries on the
first failure. If it fails again on the second try, ACPI is disabled
on the device. Note that successful configuration clears ACPI failed
status.
With all feature checks moved to the above two functions,
do_drive_set_taskfiles() is trivial and thus collapsed into
ata_acpi_exec_tfs(), which is now static and converted to return the
number of executed taskfiles to be used by ata_acpi_on_resume(). As
failures are handled properly, ata_acpi_push_id() now returns -errno
on errors instead of unconditional zero.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-14 18:28:16 +00:00
int print_info = ehc - > i . flags & ATA_EHI_PRINTINFO ;
2006-03-13 10:48:04 +00:00
const u16 * id = dev - > id ;
2007-11-27 10:43:42 +00:00
unsigned long xfer_mask ;
2006-08-10 17:52:12 +00:00
char revbuf [ 7 ] ; /* XYZ-99\0 */
2007-01-31 06:00:40 +00:00
char fwrevbuf [ ATA_ID_FW_REV_LEN + 1 ] ;
char modelbuf [ ATA_ID_PROD_LEN + 1 ] ;
2006-06-28 13:30:31 +00:00
int rc ;
2006-02-20 17:12:11 +00:00
2006-06-23 06:29:08 +00:00
if ( ! ata_dev_enabled ( dev ) & & ata_msg_info ( ap ) ) {
2007-02-20 16:06:51 +00:00
ata_dev_printk ( dev , KERN_INFO , " %s: ENTER/EXIT -- nodev \n " ,
2008-03-06 02:24:52 +00:00
__func__ ) ;
2006-03-01 07:09:35 +00:00
return 0 ;
2006-02-20 17:12:11 +00:00
}
2006-06-23 06:29:08 +00:00
if ( ata_msg_probe ( ap ) )
2008-03-06 02:24:52 +00:00
ata_dev_printk ( dev , KERN_DEBUG , " %s: ENTER \n " , __func__ ) ;
2005-04-16 22:20:36 +00:00
2007-07-05 04:31:27 +00:00
/* set horkage */
dev - > horkage | = ata_dev_blacklisted ( dev ) ;
2008-02-13 00:15:09 +00:00
ata_force_horkage ( dev ) ;
2007-07-05 04:31:27 +00:00
2008-05-18 16:15:14 +00:00
if ( dev - > horkage & ATA_HORKAGE_DISABLE ) {
ata_dev_printk ( dev , KERN_INFO ,
" unsupported device, disabling \n " ) ;
ata_dev_disable ( dev ) ;
return 0 ;
}
2008-07-30 22:52:40 +00:00
if ( ( ! atapi_enabled | | ( ap - > flags & ATA_FLAG_NO_ATAPI ) ) & &
dev - > class = = ATA_DEV_ATAPI ) {
ata_dev_printk ( dev , KERN_WARNING ,
" WARNING: ATAPI is %s, device ignored. \n " ,
atapi_enabled ? " not supported with this driver "
: " disabled " ) ;
ata_dev_disable ( dev ) ;
return 0 ;
}
2009-01-29 11:31:36 +00:00
rc = ata_do_link_spd_horkage ( dev ) ;
if ( rc )
return rc ;
libata: reimplement ACPI invocation
This patch reimplements ACPI invocation such that, instead of
exporting ACPI details to the rest of libata, ACPI event handlers -
ata_acpi_on_resume() and ata_acpi_on_devcfg() - are used. These two
functions are responsible for determining whether specific ACPI method
is used and when.
On resume, _GTF is scheduled by setting ATA_DFLAG_ACPI_PENDING device
flag. This is done this way to avoid performing the action on wrong
device device (device swapping while suspended).
On every ata_dev_configure(), ata_acpi_on_devcfg() is called, which
performs _SDD and _GTF. _GTF is performed only after resuming and, if
SATA, hardreset as the ACPI spec specifies. As _GTF may contain
arbitrary commands, IDENTIFY page is re-read after _GTF taskfiles are
executed.
If one of ACPI methods fails, ata_acpi_on_devcfg() retries on the
first failure. If it fails again on the second try, ACPI is disabled
on the device. Note that successful configuration clears ACPI failed
status.
With all feature checks moved to the above two functions,
do_drive_set_taskfiles() is trivial and thus collapsed into
ata_acpi_exec_tfs(), which is now static and converted to return the
number of executed taskfiles to be used by ata_acpi_on_resume(). As
failures are handled properly, ata_acpi_push_id() now returns -errno
on errors instead of unconditional zero.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-14 18:28:16 +00:00
/* let ACPI work its magic */
rc = ata_acpi_on_devcfg ( dev ) ;
if ( rc )
return rc ;
2006-11-11 00:14:47 +00:00
2007-09-03 03:32:57 +00:00
/* massage HPA, do it early as it might change IDENTIFY data */
rc = ata_hpa_resize ( dev ) ;
if ( rc )
return rc ;
2006-03-13 10:51:19 +00:00
/* print device capabilities */
2006-06-23 06:29:08 +00:00
if ( ata_msg_probe ( ap ) )
2006-06-25 11:00:35 +00:00
ata_dev_printk ( dev , KERN_DEBUG ,
" %s: cfg 49:%04x 82:%04x 83:%04x 84:%04x "
" 85:%04x 86:%04x 87:%04x 88:%04x \n " ,
2008-03-06 02:24:52 +00:00
__func__ ,
2006-05-15 11:57:56 +00:00
id [ 49 ] , id [ 82 ] , id [ 83 ] , id [ 84 ] ,
id [ 85 ] , id [ 86 ] , id [ 87 ] , id [ 88 ] ) ;
2006-03-13 10:51:19 +00:00
2006-03-05 08:55:58 +00:00
/* initialize to-be-configured parameters */
2006-04-02 09:51:53 +00:00
dev - > flags & = ~ ATA_DFLAG_CFG_MASK ;
2006-03-05 08:55:58 +00:00
dev - > max_sectors = 0 ;
dev - > cdb_len = 0 ;
dev - > n_sectors = 0 ;
dev - > cylinders = 0 ;
dev - > heads = 0 ;
dev - > sectors = 0 ;
2009-03-19 17:32:21 +00:00
dev - > multi_count = 0 ;
2006-03-05 08:55:58 +00:00
2005-04-16 22:20:36 +00:00
/*
* common ATA , ATAPI feature tests
*/
2006-03-05 19:31:56 +00:00
/* find max transfer mode; for printk only */
2006-03-13 10:48:04 +00:00
xfer_mask = ata_id_xfermask ( id ) ;
2005-04-16 22:20:36 +00:00
2006-06-23 06:29:08 +00:00
if ( ata_msg_probe ( ap ) )
ata_dump_id ( id ) ;
2005-04-16 22:20:36 +00:00
2007-06-05 05:01:33 +00:00
/* SCSI only uses 4-char revisions, dump full 8 chars from ATA */
ata_id_c_string ( dev - > id , fwrevbuf , ATA_ID_FW_REV ,
sizeof ( fwrevbuf ) ) ;
ata_id_c_string ( dev - > id , modelbuf , ATA_ID_PROD ,
sizeof ( modelbuf ) ) ;
2005-04-16 22:20:36 +00:00
/* ATA-specific feature tests */
if ( dev - > class = = ATA_DEV_ATA ) {
2006-08-10 17:52:12 +00:00
if ( ata_id_is_cfa ( id ) ) {
2009-04-13 16:50:00 +00:00
/* CPRM may make this media unusable */
if ( id [ ATA_ID_CFA_KEY_MGMT ] & 1 )
2007-02-20 16:06:51 +00:00
ata_dev_printk ( dev , KERN_WARNING ,
" supports DRM functions and may "
tree-wide: fix comment/printk typos
"gadget", "through", "command", "maintain", "maintain", "controller", "address",
"between", "initiali[zs]e", "instead", "function", "select", "already",
"equal", "access", "management", "hierarchy", "registration", "interest",
"relative", "memory", "offset", "already",
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2010-11-01 19:38:34 +00:00
" not be fully accessible. \n " ) ;
2006-08-10 17:52:12 +00:00
snprintf ( revbuf , 7 , " CFA " ) ;
2007-11-05 03:05:49 +00:00
} else {
2007-10-19 10:42:56 +00:00
snprintf ( revbuf , 7 , " ATA-%d " , ata_id_major_version ( id ) ) ;
2007-11-05 03:05:49 +00:00
/* Warn the user if the device has TPM extensions */
if ( ata_id_has_tpm ( id ) )
ata_dev_printk ( dev , KERN_WARNING ,
" supports DRM functions and may "
tree-wide: fix comment/printk typos
"gadget", "through", "command", "maintain", "maintain", "controller", "address",
"between", "initiali[zs]e", "instead", "function", "select", "already",
"equal", "access", "management", "hierarchy", "registration", "interest",
"relative", "memory", "offset", "already",
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2010-11-01 19:38:34 +00:00
" not be fully accessible. \n " ) ;
2007-11-05 03:05:49 +00:00
}
2006-08-10 17:52:12 +00:00
2006-03-13 10:48:04 +00:00
dev - > n_sectors = ata_id_n_sectors ( id ) ;
2006-02-12 13:47:04 +00:00
2009-03-19 17:32:21 +00:00
/* get current R/W Multiple count setting */
if ( ( dev - > id [ 47 ] > > 8 ) = = 0x80 & & ( dev - > id [ 59 ] & 0x100 ) ) {
unsigned int max = dev - > id [ 47 ] & 0xff ;
unsigned int cnt = dev - > id [ 59 ] & 0xff ;
/* only recognize/allow powers of two here */
if ( is_power_of_2 ( max ) & & is_power_of_2 ( cnt ) )
if ( cnt < = max )
dev - > multi_count = cnt ;
}
2007-01-31 06:00:40 +00:00
2006-03-13 10:48:04 +00:00
if ( ata_id_has_lba ( id ) ) {
2006-03-05 08:55:58 +00:00
const char * lba_desc ;
2009-07-27 01:24:35 +00:00
char ncq_desc [ 24 ] ;
2005-05-12 19:29:42 +00:00
2006-03-05 08:55:58 +00:00
lba_desc = " LBA " ;
dev - > flags | = ATA_DFLAG_LBA ;
2006-03-13 10:48:04 +00:00
if ( ata_id_has_lba48 ( id ) ) {
2005-05-12 19:29:42 +00:00
dev - > flags | = ATA_DFLAG_LBA48 ;
2006-03-05 08:55:58 +00:00
lba_desc = " LBA48 " ;
2006-11-11 11:10:45 +00:00
if ( dev - > n_sectors > = ( 1UL < < 28 ) & &
ata_id_has_flush_ext ( id ) )
dev - > flags | = ATA_DFLAG_FLUSH_EXT ;
2006-03-05 08:55:58 +00:00
}
2005-05-12 19:29:42 +00:00
2006-05-15 12:03:48 +00:00
/* config NCQ */
2009-07-27 01:24:35 +00:00
rc = ata_dev_config_ncq ( dev , ncq_desc , sizeof ( ncq_desc ) ) ;
if ( rc )
return rc ;
2006-05-15 12:03:48 +00:00
2005-05-12 19:29:42 +00:00
/* print device info to dmesg */
2007-01-31 06:00:40 +00:00
if ( ata_msg_drv ( ap ) & & print_info ) {
ata_dev_printk ( dev , KERN_INFO ,
" %s: %s, %s, max %s \n " ,
revbuf , modelbuf , fwrevbuf ,
ata_mode_string ( xfer_mask ) ) ;
ata_dev_printk ( dev , KERN_INFO ,
" %Lu sectors, multi %u: %s %s \n " ,
2006-05-15 11:57:56 +00:00
( unsigned long long ) dev - > n_sectors ,
2007-01-31 06:00:40 +00:00
dev - > multi_count , lba_desc , ncq_desc ) ;
}
2006-03-01 07:09:35 +00:00
} else {
2005-05-12 19:29:42 +00:00
/* CHS */
/* Default translation */
2006-03-13 10:48:04 +00:00
dev - > cylinders = id [ 1 ] ;
dev - > heads = id [ 3 ] ;
dev - > sectors = id [ 6 ] ;
2005-05-12 19:29:42 +00:00
2006-03-13 10:48:04 +00:00
if ( ata_id_current_chs_valid ( id ) ) {
2005-05-12 19:29:42 +00:00
/* Current CHS translation is valid. */
2006-03-13 10:48:04 +00:00
dev - > cylinders = id [ 54 ] ;
dev - > heads = id [ 55 ] ;
dev - > sectors = id [ 56 ] ;
2005-05-12 19:29:42 +00:00
}
/* print device info to dmesg */
2007-01-31 06:00:40 +00:00
if ( ata_msg_drv ( ap ) & & print_info ) {
2006-06-25 11:00:35 +00:00
ata_dev_printk ( dev , KERN_INFO ,
2007-01-31 06:00:40 +00:00
" %s: %s, %s, max %s \n " ,
revbuf , modelbuf , fwrevbuf ,
ata_mode_string ( xfer_mask ) ) ;
2007-02-26 10:51:33 +00:00
ata_dev_printk ( dev , KERN_INFO ,
2007-01-31 06:00:40 +00:00
" %Lu sectors, multi %u, CHS %u/%u/%u \n " ,
( unsigned long long ) dev - > n_sectors ,
dev - > multi_count , dev - > cylinders ,
dev - > heads , dev - > sectors ) ;
}
2005-11-01 11:33:20 +00:00
}
2006-02-12 14:32:58 +00:00
dev - > cdb_len = 16 ;
2005-04-16 22:20:36 +00:00
}
/* ATAPI-specific feature tests */
2005-11-14 19:14:16 +00:00
else if ( dev - > class = = ATA_DEV_ATAPI ) {
2007-09-23 04:14:11 +00:00
const char * cdb_intr_string = " " ;
const char * atapi_an_string = " " ;
2008-02-21 04:25:50 +00:00
const char * dma_dir_string = " " ;
2007-09-23 04:14:13 +00:00
u32 sntf ;
2006-03-31 05:29:04 +00:00
2006-03-13 10:48:04 +00:00
rc = atapi_cdb_len ( id ) ;
2005-04-16 22:20:36 +00:00
if ( ( rc < 12 ) | | ( rc > ATAPI_CDB_LEN ) ) {
2006-06-23 06:29:08 +00:00
if ( ata_msg_warn ( ap ) )
2006-06-25 11:00:35 +00:00
ata_dev_printk ( dev , KERN_WARNING ,
" unsupported CDB len \n " ) ;
2006-03-01 07:09:35 +00:00
rc = - EINVAL ;
2005-04-16 22:20:36 +00:00
goto err_out_nosup ;
}
2006-02-12 14:32:58 +00:00
dev - > cdb_len = ( unsigned int ) rc ;
2005-04-16 22:20:36 +00:00
2007-09-23 04:14:13 +00:00
/* Enable ATAPI AN if both the host and device have
* the support . If PMP is attached , SNTF is required
* to enable ATAPI AN to discern between PHY status
* changed notifications and ATAPI ANs .
2007-08-15 07:57:11 +00:00
*/
2010-05-19 13:38:58 +00:00
if ( atapi_an & &
( ap - > flags & ATA_FLAG_AN ) & & ata_id_has_atapi_AN ( id ) & &
2008-04-07 13:47:22 +00:00
( ! sata_pmp_attached ( ap ) | |
2007-09-23 04:14:13 +00:00
sata_scr_read ( & ap - > link , SCR_NOTIFICATION , & sntf ) = = 0 ) ) {
2007-09-23 04:14:11 +00:00
unsigned int err_mask ;
2007-08-15 07:57:11 +00:00
/* issue SET feature command to turn this on */
2007-10-25 04:33:27 +00:00
err_mask = ata_dev_set_feature ( dev ,
SETFEATURES_SATA_ENABLE , SATA_AN ) ;
2007-09-23 04:14:11 +00:00
if ( err_mask )
2007-08-15 07:57:11 +00:00
ata_dev_printk ( dev , KERN_ERR ,
2007-09-23 04:14:11 +00:00
" failed to enable ATAPI AN "
" (err_mask=0x%x) \n " , err_mask ) ;
else {
2007-08-15 07:57:11 +00:00
dev - > flags | = ATA_DFLAG_AN ;
2007-09-23 04:14:11 +00:00
atapi_an_string = " , ATAPI AN " ;
}
2007-08-15 07:57:11 +00:00
}
2006-03-31 05:29:04 +00:00
if ( ata_id_cdb_intr ( dev - > id ) ) {
2005-09-27 09:38:03 +00:00
dev - > flags | = ATA_DFLAG_CDB_INTR ;
2006-03-31 05:29:04 +00:00
cdb_intr_string = " , CDB intr " ;
}
2005-09-27 09:38:03 +00:00
2008-02-21 04:25:50 +00:00
if ( atapi_dmadir | | atapi_id_dmadir ( dev - > id ) ) {
dev - > flags | = ATA_DFLAG_DMADIR ;
dma_dir_string = " , DMADIR " ;
}
2005-04-16 22:20:36 +00:00
/* print device info to dmesg */
2006-06-27 12:51:25 +00:00
if ( ata_msg_drv ( ap ) & & print_info )
2007-06-05 05:01:33 +00:00
ata_dev_printk ( dev , KERN_INFO ,
2008-02-21 04:25:50 +00:00
" ATAPI: %s, %s, max %s%s%s%s \n " ,
2007-06-05 05:01:33 +00:00
modelbuf , fwrevbuf ,
2006-05-15 11:59:15 +00:00
ata_mode_string ( xfer_mask ) ,
2008-02-21 04:25:50 +00:00
cdb_intr_string , atapi_an_string ,
dma_dir_string ) ;
2005-04-16 22:20:36 +00:00
}
2006-11-01 09:39:55 +00:00
/* determine max_sectors */
dev - > max_sectors = ATA_MAX_SECTORS ;
if ( dev - > flags & ATA_DFLAG_LBA48 )
dev - > max_sectors = ATA_MAX_SECTORS_LBA48 ;
2007-10-25 13:21:16 +00:00
/* Limit PATA drive on SATA cable bridge transfers to udma5,
200 sectors */
2006-05-15 11:57:53 +00:00
if ( ata_dev_knobble ( dev ) ) {
2006-06-27 12:51:25 +00:00
if ( ata_msg_drv ( ap ) & & print_info )
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_INFO ,
" applying bridge limits \n " ) ;
2006-03-24 05:07:50 +00:00
dev - > udma_mask & = ATA_UDMA5 ;
2006-03-01 07:09:36 +00:00
dev - > max_sectors = ATA_MAX_SECTORS ;
}
2007-10-30 15:44:35 +00:00
if ( ( dev - > class = = ATA_DEV_ATAPI ) & &
2007-11-15 01:35:47 +00:00
( atapi_command_packet_set ( id ) = = TYPE_TAPE ) ) {
2007-10-30 15:44:35 +00:00
dev - > max_sectors = ATA_MAX_SECTORS_TAPE ;
2007-11-15 01:35:47 +00:00
dev - > horkage | = ATA_HORKAGE_STUCK_ERR ;
}
2007-10-30 15:44:35 +00:00
2007-07-05 04:31:27 +00:00
if ( dev - > horkage & ATA_HORKAGE_MAX_SEC_128 )
2007-04-12 04:38:11 +00:00
dev - > max_sectors = min_t ( unsigned int , ATA_MAX_SECTORS_128 ,
dev - > max_sectors ) ;
2007-04-02 03:34:15 +00:00
2006-03-01 07:09:36 +00:00
if ( ap - > ops - > dev_config )
2007-03-02 00:56:15 +00:00
ap - > ops - > dev_config ( dev ) ;
2006-03-01 07:09:36 +00:00
2007-10-25 13:21:16 +00:00
if ( dev - > horkage & ATA_HORKAGE_DIAGNOSTIC ) {
/* Let the user know. We don't want to disallow opens for
rescue purposes , or in case the vendor is just a blithering
idiot . Do this after the dev_config call as some controllers
with buggy firmware may want to avoid reporting false device
bugs */
if ( print_info ) {
ata_dev_printk ( dev , KERN_WARNING ,
" Drive reports diagnostics failure. This may indicate a drive \n " ) ;
ata_dev_printk ( dev , KERN_WARNING ,
" fault or invalid emulation. Contact drive vendor for information. \n " ) ;
}
}
2008-11-27 04:36:48 +00:00
if ( ( dev - > horkage & ATA_HORKAGE_FIRMWARE_WARN ) & & print_info ) {
ata_dev_printk ( dev , KERN_WARNING , " WARNING: device requires "
" firmware update to be fully functional. \n " ) ;
ata_dev_printk ( dev , KERN_WARNING , " contact the vendor "
" or visit http://ata.wiki.kernel.org. \n " ) ;
}
2006-03-01 07:09:35 +00:00
return 0 ;
2005-04-16 22:20:36 +00:00
err_out_nosup :
2006-06-23 06:29:08 +00:00
if ( ata_msg_probe ( ap ) )
2006-06-25 11:00:35 +00:00
ata_dev_printk ( dev , KERN_DEBUG ,
2008-03-06 02:24:52 +00:00
" %s: EXIT, err \n " , __func__ ) ;
2006-03-01 07:09:35 +00:00
return rc ;
2005-04-16 22:20:36 +00:00
}
2007-03-06 10:37:56 +00:00
/**
2007-03-08 23:19:19 +00:00
* ata_cable_40wire - return 40 wire cable type
2007-03-06 10:37:56 +00:00
* @ ap : port
*
2007-03-08 23:19:19 +00:00
* Helper method for drivers which want to hardwire 40 wire cable
2007-03-06 10:37:56 +00:00
* detection .
*/
int ata_cable_40wire ( struct ata_port * ap )
{
return ATA_CBL_PATA40 ;
}
/**
2007-03-08 23:19:19 +00:00
* ata_cable_80wire - return 80 wire cable type
2007-03-06 10:37:56 +00:00
* @ ap : port
*
2007-03-08 23:19:19 +00:00
* Helper method for drivers which want to hardwire 80 wire cable
2007-03-06 10:37:56 +00:00
* detection .
*/
int ata_cable_80wire ( struct ata_port * ap )
{
return ATA_CBL_PATA80 ;
}
/**
* ata_cable_unknown - return unknown PATA cable .
* @ ap : port
*
* Helper method for drivers which have no PATA cable detection .
*/
int ata_cable_unknown ( struct ata_port * ap )
{
return ATA_CBL_PATA_UNK ;
}
2007-11-27 10:43:48 +00:00
/**
* ata_cable_ignore - return ignored PATA cable .
* @ ap : port
*
* Helper method for drivers which don ' t use cable type to limit
* transfer mode .
*/
int ata_cable_ignore ( struct ata_port * ap )
{
return ATA_CBL_PATA_IGN ;
}
2007-03-06 10:37:56 +00:00
/**
* ata_cable_sata - return SATA cable type
* @ ap : port
*
* Helper method for drivers which have SATA cables
*/
int ata_cable_sata ( struct ata_port * ap )
{
return ATA_CBL_SATA ;
}
2005-04-16 22:20:36 +00:00
/**
* ata_bus_probe - Reset and probe ATA bus
* @ ap : Bus to probe
*
2005-05-30 23:49:12 +00:00
* Master ATA bus probing function . Initiates a hardware - dependent
* bus reset , then attempts to identify any devices found on
* the bus .
*
2005-04-16 22:20:36 +00:00
* LOCKING :
2005-05-30 23:49:12 +00:00
* PCI / etc . bus probe sem .
2005-04-16 22:20:36 +00:00
*
* RETURNS :
2006-03-31 16:38:17 +00:00
* Zero on success , negative errno otherwise .
2005-04-16 22:20:36 +00:00
*/
2006-08-07 19:27:31 +00:00
int ata_bus_probe ( struct ata_port * ap )
2005-04-16 22:20:36 +00:00
{
2006-03-01 07:09:36 +00:00
unsigned int classes [ ATA_MAX_DEVICES ] ;
2006-04-02 08:54:46 +00:00
int tries [ ATA_MAX_DEVICES ] ;
2007-08-06 09:36:23 +00:00
int rc ;
2006-03-31 16:38:18 +00:00
struct ata_device * dev ;
2005-04-16 22:20:36 +00:00
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ALL )
2007-08-06 09:36:23 +00:00
tries [ dev - > devno ] = ATA_PROBE_MAX_TRIES ;
2006-04-02 08:54:46 +00:00
retry :
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ALL ) {
2007-10-29 07:41:09 +00:00
/* If we issue an SRST then an ATA drive (not ATAPI)
* may change configuration and be in PIO0 timing . If
* we do a hard reset ( or are coming from power on )
* this is true for ATA or ATAPI . Until we ' ve set a
* suitable controller mode we should not touch the
* bus as we may be talking too fast .
*/
dev - > pio_mode = XFER_PIO_0 ;
/* If the controller has a pio mode setup function
* then use it to set the chipset to rights . Don ' t
* touch the DMA setup as that will be dealt with when
* configuring devices .
*/
if ( ap - > ops - > set_piomode )
ap - > ops - > set_piomode ( ap , dev ) ;
}
2006-03-12 16:57:01 +00:00
/* reset and determine device classes */
2006-05-31 09:28:22 +00:00
ap - > ops - > phy_reset ( ap ) ;
2006-03-11 15:57:39 +00:00
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ALL ) {
2010-05-10 19:41:30 +00:00
if ( dev - > class ! = ATA_DEV_UNKNOWN )
2006-05-31 09:28:22 +00:00
classes [ dev - > devno ] = dev - > class ;
else
classes [ dev - > devno ] = ATA_DEV_NONE ;
2006-03-12 16:57:01 +00:00
2006-05-31 09:28:22 +00:00
dev - > class = ATA_DEV_UNKNOWN ;
2006-03-01 07:09:36 +00:00
}
2005-04-16 22:20:36 +00:00
2007-03-02 22:47:28 +00:00
/* read IDENTIFY page and configure devices. We have to do the identify
specific sequence bass - ackwards so that PDIAG - is released by
the slave device */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ALL_REVERSE ) {
2007-08-06 09:36:23 +00:00
if ( tries [ dev - > devno ] )
dev - > class = classes [ dev - > devno ] ;
2006-03-01 07:09:35 +00:00
2006-04-02 08:54:46 +00:00
if ( ! ata_dev_enabled ( dev ) )
2006-03-01 07:09:35 +00:00
continue ;
2006-11-10 09:08:10 +00:00
rc = ata_dev_read_id ( dev , & dev - > class , ATA_READID_POSTRESET ,
dev - > id ) ;
2006-04-02 08:54:46 +00:00
if ( rc )
goto fail ;
2007-03-02 22:47:28 +00:00
}
2007-03-06 10:37:56 +00:00
/* Now ask for the cable type as PDIAG- should have been released */
if ( ap - > ops - > cable_detect )
ap - > cbl = ap - > ops - > cable_detect ( ap ) ;
2008-11-03 11:03:17 +00:00
/* We may have SATA bridge glue hiding here irrespective of
* the reported cable types and sensed types . When SATA
* drives indicate we have a bridge , we don ' t know which end
* of the link the bridge is which is a problem .
*/
ata_for_each_dev ( dev , & ap - > link , ENABLED )
2007-08-22 22:22:45 +00:00
if ( ata_id_is_sata ( dev - > id ) )
ap - > cbl = ATA_CBL_SATA ;
2007-03-02 22:47:28 +00:00
/* After the identify sequence we can now set up the devices. We do
this in the normal order so that the user doesn ' t get confused */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ENABLED ) {
2007-08-06 09:36:22 +00:00
ap - > link . eh_context . i . flags | = ATA_EHI_PRINTINFO ;
2006-11-01 09:38:52 +00:00
rc = ata_dev_configure ( dev ) ;
2007-08-06 09:36:22 +00:00
ap - > link . eh_context . i . flags & = ~ ATA_EHI_PRINTINFO ;
2006-04-02 08:54:46 +00:00
if ( rc )
goto fail ;
2005-04-16 22:20:36 +00:00
}
2006-03-31 16:38:18 +00:00
/* configure transfer mode */
2007-08-06 09:36:23 +00:00
rc = ata_set_mode ( & ap - > link , & dev ) ;
2007-02-02 07:22:30 +00:00
if ( rc )
2006-04-11 13:26:29 +00:00
goto fail ;
2005-04-16 22:20:36 +00:00
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , & ap - > link , ENABLED )
return 0 ;
2005-04-16 22:20:36 +00:00
2006-03-31 16:38:17 +00:00
return - ENODEV ;
2006-04-02 08:54:46 +00:00
fail :
2007-02-02 07:22:30 +00:00
tries [ dev - > devno ] - - ;
2006-04-02 08:54:46 +00:00
switch ( rc ) {
case - EINVAL :
2007-02-02 07:22:30 +00:00
/* eeek, something went very wrong, give up */
2006-04-02 08:54:46 +00:00
tries [ dev - > devno ] = 0 ;
break ;
2007-02-02 07:22:30 +00:00
case - ENODEV :
/* give it just one more chance */
tries [ dev - > devno ] = min ( tries [ dev - > devno ] , 1 ) ;
2006-04-02 08:54:46 +00:00
case - EIO :
2007-02-02 07:22:30 +00:00
if ( tries [ dev - > devno ] = = 1 ) {
/* This is the last chance, better to slow
* down than lose it .
*/
2009-01-29 11:31:33 +00:00
sata_down_spd_limit ( & ap - > link , 0 ) ;
2007-02-02 07:22:30 +00:00
ata_down_xfermask_limit ( dev , ATA_DNXFER_PIO ) ;
}
2006-04-02 08:54:46 +00:00
}
2007-02-02 07:22:30 +00:00
if ( ! tries [ dev - > devno ] )
2006-05-15 11:57:53 +00:00
ata_dev_disable ( dev ) ;
2006-04-11 13:26:29 +00:00
2006-04-02 08:54:46 +00:00
goto retry ;
2005-04-16 22:20:36 +00:00
}
2005-12-19 13:35:02 +00:00
/**
* sata_print_link_status - Print SATA link status
2007-08-06 09:36:23 +00:00
* @ link : SATA link to printk link status about
2005-12-19 13:35:02 +00:00
*
* This function prints link speed and status of a SATA link .
*
* LOCKING :
* None .
*/
2008-04-21 08:51:11 +00:00
static void sata_print_link_status ( struct ata_link * link )
2005-12-19 13:35:02 +00:00
{
2006-04-02 15:09:41 +00:00
u32 sstatus , scontrol , tmp ;
2005-12-19 13:35:02 +00:00
2007-08-06 09:36:23 +00:00
if ( sata_scr_read ( link , SCR_STATUS , & sstatus ) )
2005-12-19 13:35:02 +00:00
return ;
2007-08-06 09:36:23 +00:00
sata_scr_read ( link , SCR_CONTROL , & scontrol ) ;
2005-12-19 13:35:02 +00:00
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
if ( ata_phys_link_online ( link ) ) {
2005-12-19 13:35:02 +00:00
tmp = ( sstatus > > 4 ) & 0xf ;
2007-08-06 09:36:23 +00:00
ata_link_printk ( link , KERN_INFO ,
2006-05-15 11:57:56 +00:00
" SATA link up %s (SStatus %X SControl %X) \n " ,
sata_spd_string ( tmp ) , sstatus , scontrol ) ;
2005-12-19 13:35:02 +00:00
} else {
2007-08-06 09:36:23 +00:00
ata_link_printk ( link , KERN_INFO ,
2006-05-15 11:57:56 +00:00
" SATA link down (SStatus %X SControl %X) \n " ,
sstatus , scontrol ) ;
2005-12-19 13:35:02 +00:00
}
}
2006-03-23 15:38:34 +00:00
/**
* ata_dev_pair - return other device on cable
* @ adev : device
*
* Obtain the other device on the same cable , or if none is
* present NULL is returned
*/
2006-03-24 14:56:57 +00:00
2006-05-15 11:57:53 +00:00
struct ata_device * ata_dev_pair ( struct ata_device * adev )
2006-03-23 15:38:34 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_link * link = adev - > link ;
struct ata_device * pair = & link - > device [ 1 - adev - > devno ] ;
2006-03-31 16:38:18 +00:00
if ( ! ata_dev_enabled ( pair ) )
2006-03-23 15:38:34 +00:00
return NULL ;
return pair ;
}
2006-04-02 11:53:28 +00:00
/**
2006-05-15 11:57:23 +00:00
* sata_down_spd_limit - adjust SATA spd limit downward
2007-08-06 09:36:23 +00:00
* @ link : Link to adjust SATA spd limit for
2009-01-29 11:31:33 +00:00
* @ spd_limit : Additional limit
2006-04-02 11:53:28 +00:00
*
2007-08-06 09:36:23 +00:00
* Adjust SATA spd limit of @ link downward . Note that this
2006-04-02 11:53:28 +00:00
* function only adjusts the limit . The change must be applied
2006-05-15 11:57:23 +00:00
* using sata_set_spd ( ) .
2006-04-02 11:53:28 +00:00
*
2009-01-29 11:31:33 +00:00
* If @ spd_limit is non - zero , the speed is limited to equal to or
* lower than @ spd_limit if such speed is supported . If
* @ spd_limit is slower than any supported speed , only the lowest
* supported speed is allowed .
*
2006-04-02 11:53:28 +00:00
* LOCKING :
* Inherited from caller .
*
* RETURNS :
* 0 on success , negative errno on failure
*/
2009-01-29 11:31:33 +00:00
int sata_down_spd_limit ( struct ata_link * link , u32 spd_limit )
2006-04-02 11:53:28 +00:00
{
2006-05-15 11:57:47 +00:00
u32 sstatus , spd , mask ;
2009-01-29 11:31:33 +00:00
int rc , bit ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
if ( ! sata_scr_valid ( link ) )
2007-07-16 05:29:40 +00:00
return - EOPNOTSUPP ;
/* If SCR can be read, use it to determine the current SPD.
2007-08-06 09:36:23 +00:00
* If not , use cached value in link - > sata_spd .
2007-07-16 05:29:40 +00:00
*/
2007-08-06 09:36:23 +00:00
rc = sata_scr_read ( link , SCR_STATUS , & sstatus ) ;
2009-01-29 11:31:31 +00:00
if ( rc = = 0 & & ata_sstatus_online ( sstatus ) )
2007-07-16 05:29:40 +00:00
spd = ( sstatus > > 4 ) & 0xf ;
else
2007-08-06 09:36:23 +00:00
spd = link - > sata_spd ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
mask = link - > sata_spd_limit ;
2006-04-02 11:53:28 +00:00
if ( mask < = 1 )
return - EINVAL ;
2007-07-16 05:29:40 +00:00
/* unconditionally mask off the highest bit */
2009-01-29 11:31:33 +00:00
bit = fls ( mask ) - 1 ;
mask & = ~ ( 1 < < bit ) ;
2006-04-02 11:53:28 +00:00
2007-07-16 05:29:40 +00:00
/* Mask off all speeds higher than or equal to the current
* one . Force 1.5 Gbps if current SPD is not available .
*/
if ( spd > 1 )
mask & = ( 1 < < ( spd - 1 ) ) - 1 ;
else
mask & = 1 ;
/* were we already at the bottom? */
2006-04-02 11:53:28 +00:00
if ( ! mask )
return - EINVAL ;
2009-01-29 11:31:33 +00:00
if ( spd_limit ) {
if ( mask & ( ( 1 < < spd_limit ) - 1 ) )
mask & = ( 1 < < spd_limit ) - 1 ;
else {
bit = ffs ( mask ) - 1 ;
mask = 1 < < bit ;
}
}
2007-08-06 09:36:23 +00:00
link - > sata_spd_limit = mask ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
ata_link_printk ( link , KERN_WARNING , " limiting SATA link speed to %s \n " ,
2006-05-15 11:57:56 +00:00
sata_spd_string ( fls ( mask ) ) ) ;
2006-04-02 11:53:28 +00:00
return 0 ;
}
2007-08-06 09:36:23 +00:00
static int __sata_set_spd_needed ( struct ata_link * link , u32 * scontrol )
2006-04-02 11:53:28 +00:00
{
2007-10-31 01:17:07 +00:00
struct ata_link * host_link = & link - > ap - > link ;
u32 limit , target , spd ;
2006-04-02 11:53:28 +00:00
2007-10-31 01:17:07 +00:00
limit = link - > sata_spd_limit ;
/* Don't configure downstream link faster than upstream link.
* It doesn ' t speed up anything and some PMPs choke on such
* configuration .
*/
if ( ! ata_is_host_link ( link ) & & host_link - > sata_spd )
limit & = ( 1 < < host_link - > sata_spd ) - 1 ;
if ( limit = = UINT_MAX )
target = 0 ;
2006-04-02 11:53:28 +00:00
else
2007-10-31 01:17:07 +00:00
target = fls ( limit ) ;
2006-04-02 11:53:28 +00:00
spd = ( * scontrol > > 4 ) & 0xf ;
2007-10-31 01:17:07 +00:00
* scontrol = ( * scontrol & ~ 0xf0 ) | ( ( target & 0xf ) < < 4 ) ;
2006-04-02 11:53:28 +00:00
2007-10-31 01:17:07 +00:00
return spd ! = target ;
2006-04-02 11:53:28 +00:00
}
/**
2006-05-15 11:57:23 +00:00
* sata_set_spd_needed - is SATA spd configuration needed
2007-08-06 09:36:23 +00:00
* @ link : Link in question
2006-04-02 11:53:28 +00:00
*
* Test whether the spd limit in SControl matches
2007-08-06 09:36:23 +00:00
* @ link - > sata_spd_limit . This function is used to determine
2006-04-02 11:53:28 +00:00
* whether hardreset is necessary to apply SATA spd
* configuration .
*
* LOCKING :
* Inherited from caller .
*
* RETURNS :
* 1 if SATA spd configuration is needed , 0 otherwise .
*/
2008-04-21 08:51:17 +00:00
static int sata_set_spd_needed ( struct ata_link * link )
2006-04-02 11:53:28 +00:00
{
u32 scontrol ;
2007-08-06 09:36:23 +00:00
if ( sata_scr_read ( link , SCR_CONTROL , & scontrol ) )
2007-10-31 01:17:06 +00:00
return 1 ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
return __sata_set_spd_needed ( link , & scontrol ) ;
2006-04-02 11:53:28 +00:00
}
/**
2006-05-15 11:57:23 +00:00
* sata_set_spd - set SATA spd according to spd limit
2007-08-06 09:36:23 +00:00
* @ link : Link to set SATA spd for
2006-04-02 11:53:28 +00:00
*
2007-08-06 09:36:23 +00:00
* Set SATA spd of @ link according to sata_spd_limit .
2006-04-02 11:53:28 +00:00
*
* LOCKING :
* Inherited from caller .
*
* RETURNS :
* 0 if spd doesn ' t need to be changed , 1 if spd has been
2006-05-15 11:57:47 +00:00
* changed . Negative errno if SCR registers are inaccessible .
2006-04-02 11:53:28 +00:00
*/
2007-08-06 09:36:23 +00:00
int sata_set_spd ( struct ata_link * link )
2006-04-02 11:53:28 +00:00
{
u32 scontrol ;
2006-05-15 11:57:47 +00:00
int rc ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ) )
2006-05-15 11:57:47 +00:00
return rc ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
if ( ! __sata_set_spd_needed ( link , & scontrol ) )
2006-04-02 11:53:28 +00:00
return 0 ;
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_write ( link , SCR_CONTROL , scontrol ) ) )
2006-05-15 11:57:47 +00:00
return rc ;
2006-04-02 11:53:28 +00:00
return 1 ;
}
2005-10-21 23:01:32 +00:00
/*
* This mode timing computation functionality is ported over from
* drivers / ide / ide - timing . h and was originally written by Vojtech Pavlik
*/
/*
2006-08-10 17:52:12 +00:00
* PIO 0 - 4 , MWDMA 0 - 2 and UDMA 0 - 6 timings ( in nanoseconds ) .
2005-10-21 23:01:32 +00:00
* These were taken from ATA / ATAPI - 6 standard , rev 0 a , except
2006-08-10 17:52:12 +00:00
* for UDMA6 , which is currently supported only by Maxtor drives .
*
* For PIO 5 / 6 MWDMA 3 / 4 see the CFA specification 3.0 .
2005-10-21 23:01:32 +00:00
*/
static const struct ata_timing ata_timing [ ] = {
2009-01-16 01:45:31 +00:00
/* { XFER_PIO_SLOW, 120, 290, 240, 960, 290, 240, 0, 960, 0 }, */
{ XFER_PIO_0 , 70 , 290 , 240 , 600 , 165 , 150 , 0 , 600 , 0 } ,
{ XFER_PIO_1 , 50 , 290 , 93 , 383 , 125 , 100 , 0 , 383 , 0 } ,
{ XFER_PIO_2 , 30 , 290 , 40 , 330 , 100 , 90 , 0 , 240 , 0 } ,
{ XFER_PIO_3 , 30 , 80 , 70 , 180 , 80 , 70 , 0 , 180 , 0 } ,
{ XFER_PIO_4 , 25 , 70 , 25 , 120 , 70 , 25 , 0 , 120 , 0 } ,
{ XFER_PIO_5 , 15 , 65 , 25 , 100 , 65 , 25 , 0 , 100 , 0 } ,
{ XFER_PIO_6 , 10 , 55 , 20 , 80 , 55 , 20 , 0 , 80 , 0 } ,
{ XFER_SW_DMA_0 , 120 , 0 , 0 , 0 , 480 , 480 , 50 , 960 , 0 } ,
{ XFER_SW_DMA_1 , 90 , 0 , 0 , 0 , 240 , 240 , 30 , 480 , 0 } ,
{ XFER_SW_DMA_2 , 60 , 0 , 0 , 0 , 120 , 120 , 20 , 240 , 0 } ,
{ XFER_MW_DMA_0 , 60 , 0 , 0 , 0 , 215 , 215 , 20 , 480 , 0 } ,
{ XFER_MW_DMA_1 , 45 , 0 , 0 , 0 , 80 , 50 , 5 , 150 , 0 } ,
{ XFER_MW_DMA_2 , 25 , 0 , 0 , 0 , 70 , 25 , 5 , 120 , 0 } ,
{ XFER_MW_DMA_3 , 25 , 0 , 0 , 0 , 65 , 25 , 5 , 100 , 0 } ,
{ XFER_MW_DMA_4 , 25 , 0 , 0 , 0 , 55 , 20 , 5 , 80 , 0 } ,
/* { XFER_UDMA_SLOW, 0, 0, 0, 0, 0, 0, 0, 0, 150 }, */
{ XFER_UDMA_0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 120 } ,
{ XFER_UDMA_1 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 80 } ,
{ XFER_UDMA_2 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 60 } ,
{ XFER_UDMA_3 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 45 } ,
{ XFER_UDMA_4 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 30 } ,
{ XFER_UDMA_5 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 20 } ,
{ XFER_UDMA_6 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 15 } ,
2005-10-21 23:01:32 +00:00
{ 0xFF }
} ;
2007-10-19 10:42:56 +00:00
# define ENOUGH(v, unit) (((v)-1) / (unit)+1)
# define EZ(v, unit) ((v)?ENOUGH(v, unit):0)
2005-10-21 23:01:32 +00:00
static void ata_timing_quantize ( const struct ata_timing * t , struct ata_timing * q , int T , int UT )
{
2009-01-16 01:45:31 +00:00
q - > setup = EZ ( t - > setup * 1000 , T ) ;
q - > act8b = EZ ( t - > act8b * 1000 , T ) ;
q - > rec8b = EZ ( t - > rec8b * 1000 , T ) ;
q - > cyc8b = EZ ( t - > cyc8b * 1000 , T ) ;
q - > active = EZ ( t - > active * 1000 , T ) ;
q - > recover = EZ ( t - > recover * 1000 , T ) ;
q - > dmack_hold = EZ ( t - > dmack_hold * 1000 , T ) ;
q - > cycle = EZ ( t - > cycle * 1000 , T ) ;
q - > udma = EZ ( t - > udma * 1000 , UT ) ;
2005-10-21 23:01:32 +00:00
}
void ata_timing_merge ( const struct ata_timing * a , const struct ata_timing * b ,
struct ata_timing * m , unsigned int what )
{
if ( what & ATA_TIMING_SETUP ) m - > setup = max ( a - > setup , b - > setup ) ;
if ( what & ATA_TIMING_ACT8B ) m - > act8b = max ( a - > act8b , b - > act8b ) ;
if ( what & ATA_TIMING_REC8B ) m - > rec8b = max ( a - > rec8b , b - > rec8b ) ;
if ( what & ATA_TIMING_CYC8B ) m - > cyc8b = max ( a - > cyc8b , b - > cyc8b ) ;
if ( what & ATA_TIMING_ACTIVE ) m - > active = max ( a - > active , b - > active ) ;
if ( what & ATA_TIMING_RECOVER ) m - > recover = max ( a - > recover , b - > recover ) ;
2009-01-16 01:45:31 +00:00
if ( what & ATA_TIMING_DMACK_HOLD ) m - > dmack_hold = max ( a - > dmack_hold , b - > dmack_hold ) ;
2005-10-21 23:01:32 +00:00
if ( what & ATA_TIMING_CYCLE ) m - > cycle = max ( a - > cycle , b - > cycle ) ;
if ( what & ATA_TIMING_UDMA ) m - > udma = max ( a - > udma , b - > udma ) ;
}
2007-11-27 10:43:39 +00:00
const struct ata_timing * ata_timing_find_mode ( u8 xfer_mode )
2005-10-21 23:01:32 +00:00
{
2007-11-27 10:43:40 +00:00
const struct ata_timing * t = ata_timing ;
while ( xfer_mode > t - > mode )
t + + ;
2005-10-21 23:01:32 +00:00
2007-11-27 10:43:40 +00:00
if ( xfer_mode = = t - > mode )
return t ;
return NULL ;
2005-10-21 23:01:32 +00:00
}
int ata_timing_compute ( struct ata_device * adev , unsigned short speed ,
struct ata_timing * t , int T , int UT )
{
2010-01-18 17:13:57 +00:00
const u16 * id = adev - > id ;
2005-10-21 23:01:32 +00:00
const struct ata_timing * s ;
struct ata_timing p ;
/*
2006-03-24 14:56:57 +00:00
* Find the mode .
2005-11-16 09:06:18 +00:00
*/
2005-10-21 23:01:32 +00:00
if ( ! ( s = ata_timing_find_mode ( speed ) ) )
return - EINVAL ;
2005-11-16 09:06:18 +00:00
memcpy ( t , s , sizeof ( * s ) ) ;
2005-10-21 23:01:32 +00:00
/*
* If the drive is an EIDE drive , it can tell us it needs extended
* PIO / MW_DMA cycle timing .
*/
2010-01-18 17:13:57 +00:00
if ( id [ ATA_ID_FIELD_VALID ] & 2 ) { /* EIDE drive */
2005-10-21 23:01:32 +00:00
memset ( & p , 0 , sizeof ( p ) ) ;
2010-01-18 17:13:57 +00:00
2007-10-19 10:42:56 +00:00
if ( speed > = XFER_PIO_0 & & speed < = XFER_SW_DMA_0 ) {
2010-01-18 17:13:57 +00:00
if ( speed < = XFER_PIO_2 )
p . cycle = p . cyc8b = id [ ATA_ID_EIDE_PIO ] ;
else if ( ( speed < = XFER_PIO_4 ) | |
( speed = = XFER_PIO_5 & & ! ata_id_is_cfa ( id ) ) )
p . cycle = p . cyc8b = id [ ATA_ID_EIDE_PIO_IORDY ] ;
} else if ( speed > = XFER_MW_DMA_0 & & speed < = XFER_MW_DMA_2 )
p . cycle = id [ ATA_ID_EIDE_DMA_MIN ] ;
2005-10-21 23:01:32 +00:00
ata_timing_merge ( & p , t , t , ATA_TIMING_CYCLE | ATA_TIMING_CYC8B ) ;
}
/*
* Convert the timing to bus clock counts .
*/
2005-11-16 09:06:18 +00:00
ata_timing_quantize ( t , t , T , UT ) ;
2005-10-21 23:01:32 +00:00
/*
2006-01-28 18:15:32 +00:00
* Even in DMA / UDMA modes we still use PIO access for IDENTIFY ,
* S . M . A . R . T * and some other commands . We have to ensure that the
* DMA cycle timing is slower / equal than the fastest PIO timing .
2005-10-21 23:01:32 +00:00
*/
2006-12-07 12:41:18 +00:00
if ( speed > XFER_PIO_6 ) {
2005-10-21 23:01:32 +00:00
ata_timing_compute ( adev , adev - > pio_mode , & p , T , UT ) ;
ata_timing_merge ( & p , t , t , ATA_TIMING_ALL ) ;
}
/*
2006-01-28 18:15:32 +00:00
* Lengthen active & recovery time so that cycle time is correct .
2005-10-21 23:01:32 +00:00
*/
if ( t - > act8b + t - > rec8b < t - > cyc8b ) {
t - > act8b + = ( t - > cyc8b - ( t - > act8b + t - > rec8b ) ) / 2 ;
t - > rec8b = t - > cyc8b - t - > act8b ;
}
if ( t - > active + t - > recover < t - > cycle ) {
t - > active + = ( t - > cycle - ( t - > active + t - > recover ) ) / 2 ;
t - > recover = t - > cycle - t - > active ;
}
2007-05-22 00:14:23 +00:00
2007-04-23 10:55:36 +00:00
/* In a few cases quantisation may produce enough errors to
leave t - > cycle too low for the sum of active and recovery
if so we must correct this */
if ( t - > active + t - > recover > t - > cycle )
t - > cycle = t - > active + t - > recover ;
2005-10-21 23:01:32 +00:00
return 0 ;
}
2007-12-18 07:33:05 +00:00
/**
* ata_timing_cycle2mode - find xfer mode for the specified cycle duration
* @ xfer_shift : ATA_SHIFT_ * value for transfer type to examine .
* @ cycle : cycle duration in ns
*
* Return matching xfer mode for @ cycle . The returned mode is of
* the transfer type specified by @ xfer_shift . If @ cycle is too
* slow for @ xfer_shift , 0xff is returned . If @ cycle is faster
* than the fastest known mode , the fasted mode is returned .
*
* LOCKING :
* None .
*
* RETURNS :
* Matching xfer_mode , 0xff if no match found .
*/
u8 ata_timing_cycle2mode ( unsigned int xfer_shift , int cycle )
{
u8 base_mode = 0xff , last_mode = 0xff ;
const struct ata_xfer_ent * ent ;
const struct ata_timing * t ;
for ( ent = ata_xfer_tbl ; ent - > shift > = 0 ; ent + + )
if ( ent - > shift = = xfer_shift )
base_mode = ent - > base ;
for ( t = ata_timing_find_mode ( base_mode ) ;
t & & ata_xfer_mode2shift ( t - > mode ) = = xfer_shift ; t + + ) {
unsigned short this_cycle ;
switch ( xfer_shift ) {
case ATA_SHIFT_PIO :
case ATA_SHIFT_MWDMA :
this_cycle = t - > cycle ;
break ;
case ATA_SHIFT_UDMA :
this_cycle = t - > udma ;
break ;
default :
return 0xff ;
}
if ( cycle > this_cycle )
break ;
last_mode = t - > mode ;
}
return last_mode ;
}
2006-04-02 08:54:46 +00:00
/**
* ata_down_xfermask_limit - adjust dev xfer masks downward
* @ dev : Device to adjust xfer masks
2007-02-02 07:22:30 +00:00
* @ sel : ATA_DNXFER_ * selector
2006-04-02 08:54:46 +00:00
*
* Adjust xfer masks of @ dev downward . Note that this function
* does not apply the change . Invoking ata_set_mode ( ) afterwards
* will apply the limit .
*
* LOCKING :
* Inherited from caller .
*
* RETURNS :
* 0 on success , negative errno on failure
*/
2007-02-02 07:22:30 +00:00
int ata_down_xfermask_limit ( struct ata_device * dev , unsigned int sel )
2006-04-02 08:54:46 +00:00
{
2007-02-02 07:22:30 +00:00
char buf [ 32 ] ;
2007-11-27 10:43:42 +00:00
unsigned long orig_mask , xfer_mask ;
unsigned long pio_mask , mwdma_mask , udma_mask ;
2007-02-02 07:22:30 +00:00
int quiet , highbit ;
2006-04-02 08:54:46 +00:00
2007-02-02 07:22:30 +00:00
quiet = ! ! ( sel & ATA_DNXFER_QUIET ) ;
sel & = ~ ATA_DNXFER_QUIET ;
2006-04-02 08:54:46 +00:00
2007-02-02 07:22:30 +00:00
xfer_mask = orig_mask = ata_pack_xfermask ( dev - > pio_mask ,
dev - > mwdma_mask ,
dev - > udma_mask ) ;
ata_unpack_xfermask ( xfer_mask , & pio_mask , & mwdma_mask , & udma_mask ) ;
2006-04-02 08:54:46 +00:00
2007-02-02 07:22:30 +00:00
switch ( sel ) {
case ATA_DNXFER_PIO :
highbit = fls ( pio_mask ) - 1 ;
pio_mask & = ~ ( 1 < < highbit ) ;
break ;
case ATA_DNXFER_DMA :
if ( udma_mask ) {
highbit = fls ( udma_mask ) - 1 ;
udma_mask & = ~ ( 1 < < highbit ) ;
if ( ! udma_mask )
return - ENOENT ;
} else if ( mwdma_mask ) {
highbit = fls ( mwdma_mask ) - 1 ;
mwdma_mask & = ~ ( 1 < < highbit ) ;
if ( ! mwdma_mask )
return - ENOENT ;
}
break ;
case ATA_DNXFER_40C :
udma_mask & = ATA_UDMA_MASK_40C ;
break ;
case ATA_DNXFER_FORCE_PIO0 :
pio_mask & = 1 ;
case ATA_DNXFER_FORCE_PIO :
mwdma_mask = 0 ;
udma_mask = 0 ;
break ;
default :
BUG ( ) ;
}
xfer_mask & = ata_pack_xfermask ( pio_mask , mwdma_mask , udma_mask ) ;
if ( ! ( xfer_mask & ATA_MASK_PIO ) | | xfer_mask = = orig_mask )
return - ENOENT ;
if ( ! quiet ) {
if ( xfer_mask & ( ATA_MASK_MWDMA | ATA_MASK_UDMA ) )
snprintf ( buf , sizeof ( buf ) , " %s:%s " ,
ata_mode_string ( xfer_mask ) ,
ata_mode_string ( xfer_mask & ATA_MASK_PIO ) ) ;
else
snprintf ( buf , sizeof ( buf ) , " %s " ,
ata_mode_string ( xfer_mask ) ) ;
ata_dev_printk ( dev , KERN_WARNING ,
" limiting speed to %s \n " , buf ) ;
}
2006-04-02 08:54:46 +00:00
ata_unpack_xfermask ( xfer_mask , & dev - > pio_mask , & dev - > mwdma_mask ,
& dev - > udma_mask ) ;
return 0 ;
}
2006-05-15 11:57:53 +00:00
static int ata_dev_set_mode ( struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2009-07-09 00:27:50 +00:00
struct ata_port * ap = dev - > link - > ap ;
2007-08-06 09:36:22 +00:00
struct ata_eh_context * ehc = & dev - > link - > eh_context ;
2009-07-09 00:27:50 +00:00
const bool nosetxfer = dev - > horkage & ATA_HORKAGE_NOSETXFER ;
2008-02-07 01:34:08 +00:00
const char * dev_err_whine = " " ;
int ign_dev_err = 0 ;
2009-07-09 00:27:50 +00:00
unsigned int err_mask = 0 ;
2006-03-24 06:25:31 +00:00
int rc ;
2005-04-16 22:20:36 +00:00
2006-04-02 09:51:53 +00:00
dev - > flags & = ~ ATA_DFLAG_PIO ;
2005-04-16 22:20:36 +00:00
if ( dev - > xfer_shift = = ATA_SHIFT_PIO )
dev - > flags | = ATA_DFLAG_PIO ;
2009-07-09 00:27:50 +00:00
if ( nosetxfer & & ap - > flags & ATA_FLAG_SATA & & ata_id_is_sata ( dev - > id ) )
dev_err_whine = " (SET_XFERMODE skipped) " ;
else {
if ( nosetxfer )
ata_dev_printk ( dev , KERN_WARNING ,
" NOSETXFER but PATA detected - can't "
" skip SETXFER, might malfunction \n " ) ;
err_mask = ata_dev_set_xfermode ( dev ) ;
}
2007-10-19 10:42:56 +00:00
2008-02-07 01:34:08 +00:00
if ( err_mask & ~ AC_ERR_DEV )
goto fail ;
/* revalidate */
ehc - > i . flags | = ATA_EHI_POST_SETMODE ;
rc = ata_dev_revalidate ( dev , ATA_DEV_UNKNOWN , 0 ) ;
ehc - > i . flags & = ~ ATA_EHI_POST_SETMODE ;
if ( rc )
return rc ;
2008-04-08 15:36:44 +00:00
if ( dev - > xfer_shift = = ATA_SHIFT_PIO ) {
/* Old CFA may refuse this command, which is just fine */
if ( ata_id_is_cfa ( dev - > id ) )
ign_dev_err = 1 ;
/* Catch several broken garbage emulations plus some pre
ATA devices */
if ( ata_id_major_version ( dev - > id ) = = 0 & &
dev - > pio_mode < = XFER_PIO_2 )
ign_dev_err = 1 ;
/* Some very old devices and some bad newer ones fail
any kind of SET_XFERMODE request but support PIO0 - 2
timings and no IORDY */
if ( ! ata_id_has_iordy ( dev - > id ) & & dev - > pio_mode < = XFER_PIO_2 )
ign_dev_err = 1 ;
}
2007-10-15 18:18:27 +00:00
/* Early MWDMA devices do DMA but don't allow DMA mode setting.
Don ' t fail an MWDMA0 set IFF the device indicates it is in MWDMA0 */
2007-10-25 13:21:16 +00:00
if ( dev - > xfer_shift = = ATA_SHIFT_MWDMA & &
2007-10-15 18:18:27 +00:00
dev - > dma_mode = = XFER_MW_DMA_0 & &
( dev - > id [ 63 ] > > 8 ) & 1 )
2008-02-07 01:34:08 +00:00
ign_dev_err = 1 ;
2007-10-15 18:18:27 +00:00
2008-02-07 01:34:08 +00:00
/* if the device is actually configured correctly, ignore dev err */
if ( dev - > xfer_mode = = ata_xfer_mask2mode ( ata_id_xfermask ( dev - > id ) ) )
ign_dev_err = 1 ;
2005-04-16 22:20:36 +00:00
2008-02-07 01:34:08 +00:00
if ( err_mask & AC_ERR_DEV ) {
if ( ! ign_dev_err )
goto fail ;
else
dev_err_whine = " (device error ignored) " ;
}
2006-03-05 08:55:58 +00:00
2006-03-05 19:31:57 +00:00
DPRINTK ( " xfer_shift=%u, xfer_mode=0x%x \n " ,
dev - > xfer_shift , ( int ) dev - > xfer_mode ) ;
2005-04-16 22:20:36 +00:00
2008-02-07 01:34:08 +00:00
ata_dev_printk ( dev , KERN_INFO , " configured for %s%s \n " ,
ata_mode_string ( ata_xfer_mode2mask ( dev - > xfer_mode ) ) ,
dev_err_whine ) ;
2006-03-24 06:25:31 +00:00
return 0 ;
2008-02-07 01:34:08 +00:00
fail :
ata_dev_printk ( dev , KERN_ERR , " failed to set xfermode "
" (err_mask=0x%x) \n " , err_mask ) ;
return - EIO ;
2005-04-16 22:20:36 +00:00
}
/**
2007-03-06 10:37:52 +00:00
* ata_do_set_mode - Program timings and issue SET FEATURES - XFER
2007-08-06 09:36:23 +00:00
* @ link : link on which timings will be programmed
2008-02-03 15:08:11 +00:00
* @ r_failed_dev : out parameter for failed device
2005-04-16 22:20:36 +00:00
*
2007-03-06 10:37:52 +00:00
* Standard implementation of the function used to tune and set
* ATA device disk transfer mode ( PIO3 , UDMA6 , etc . ) . If
* ata_dev_set_mode ( ) fails , pointer to the failing device is
2006-03-31 16:38:18 +00:00
* returned in @ r_failed_dev .
2005-05-30 19:41:05 +00:00
*
2005-04-16 22:20:36 +00:00
* LOCKING :
2005-05-30 23:49:12 +00:00
* PCI / etc . bus probe sem .
2006-03-31 16:38:18 +00:00
*
* RETURNS :
* 0 on success , negative errno otherwise
2005-04-16 22:20:36 +00:00
*/
2007-03-06 10:37:52 +00:00
2007-08-06 09:36:23 +00:00
int ata_do_set_mode ( struct ata_link * link , struct ata_device * * r_failed_dev )
2005-04-16 22:20:36 +00:00
{
2007-08-06 09:36:23 +00:00
struct ata_port * ap = link - > ap ;
2006-03-31 16:38:18 +00:00
struct ata_device * dev ;
2007-08-06 09:36:23 +00:00
int rc = 0 , used_dma = 0 , found = 0 ;
2006-05-15 11:57:37 +00:00
2006-03-05 19:31:57 +00:00
/* step 1: calculate xfer_mask */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , link , ENABLED ) {
2007-11-27 10:43:42 +00:00
unsigned long pio_mask , dma_mask ;
2007-10-02 16:38:26 +00:00
unsigned int mode_mask ;
2006-03-05 19:31:57 +00:00
2007-10-02 16:38:26 +00:00
mode_mask = ATA_DMA_MASK_ATA ;
if ( dev - > class = = ATA_DEV_ATAPI )
mode_mask = ATA_DMA_MASK_ATAPI ;
else if ( ata_id_is_cfa ( dev - > id ) )
mode_mask = ATA_DMA_MASK_CFA ;
2006-05-15 11:57:53 +00:00
ata_dev_xfermask ( dev ) ;
2008-02-13 00:15:09 +00:00
ata_force_xfermask ( dev ) ;
2005-04-16 22:20:36 +00:00
2006-03-24 05:07:50 +00:00
pio_mask = ata_pack_xfermask ( dev - > pio_mask , 0 , 0 ) ;
dma_mask = ata_pack_xfermask ( 0 , dev - > mwdma_mask , dev - > udma_mask ) ;
2007-10-02 16:38:26 +00:00
if ( libata_dma_mask & mode_mask )
dma_mask = ata_pack_xfermask ( 0 , dev - > mwdma_mask , dev - > udma_mask ) ;
else
dma_mask = 0 ;
2006-03-24 05:07:50 +00:00
dev - > pio_mode = ata_xfer_mask2mode ( pio_mask ) ;
dev - > dma_mode = ata_xfer_mask2mode ( dma_mask ) ;
2006-03-27 17:58:20 +00:00
2006-03-31 16:38:18 +00:00
found = 1 ;
2008-08-01 08:18:34 +00:00
if ( ata_dma_enabled ( dev ) )
2006-03-27 17:58:20 +00:00
used_dma = 1 ;
2006-03-05 19:31:57 +00:00
}
2006-03-31 16:38:18 +00:00
if ( ! found )
2006-03-31 16:38:18 +00:00
goto out ;
2006-03-05 19:31:57 +00:00
/* step 2: always set host PIO timings */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , link , ENABLED ) {
2007-11-27 10:43:40 +00:00
if ( dev - > pio_mode = = 0xff ) {
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_WARNING , " no PIO support \n " ) ;
2006-03-31 16:38:18 +00:00
rc = - EINVAL ;
2006-03-31 16:38:18 +00:00
goto out ;
2006-03-31 16:38:18 +00:00
}
dev - > xfer_mode = dev - > pio_mode ;
dev - > xfer_shift = ATA_SHIFT_PIO ;
if ( ap - > ops - > set_piomode )
ap - > ops - > set_piomode ( ap , dev ) ;
}
2005-04-16 22:20:36 +00:00
2006-03-05 19:31:57 +00:00
/* step 3: set host DMA timings */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , link , ENABLED ) {
if ( ! ata_dma_enabled ( dev ) )
2006-03-31 16:38:18 +00:00
continue ;
dev - > xfer_mode = dev - > dma_mode ;
dev - > xfer_shift = ata_xfer_mode2shift ( dev - > dma_mode ) ;
if ( ap - > ops - > set_dmamode )
ap - > ops - > set_dmamode ( ap , dev ) ;
}
2005-04-16 22:20:36 +00:00
/* step 4: update devices' xfer mode */
2008-11-03 11:03:17 +00:00
ata_for_each_dev ( dev , link , ENABLED ) {
2006-05-15 11:57:53 +00:00
rc = ata_dev_set_mode ( dev ) ;
2006-03-31 16:38:17 +00:00
if ( rc )
2006-03-31 16:38:18 +00:00
goto out ;
2006-03-24 06:25:31 +00:00
}
2005-04-16 22:20:36 +00:00
2006-03-31 16:38:18 +00:00
/* Record simplex status. If we selected DMA then the other
* host channels are not permitted to do so .
2006-03-27 17:58:20 +00:00
*/
2006-08-24 07:19:22 +00:00
if ( used_dma & & ( ap - > host - > flags & ATA_HOST_SIMPLEX ) )
2007-03-01 17:36:46 +00:00
ap - > host - > simplex_claimed = ap ;
2006-03-27 17:58:20 +00:00
2006-03-31 16:38:18 +00:00
out :
if ( rc )
* r_failed_dev = dev ;
return rc ;
2005-04-16 22:20:36 +00:00
}
2008-04-07 13:47:19 +00:00
/**
* ata_wait_ready - wait for link to become ready
* @ link : link to be waited on
* @ deadline : deadline jiffies for the operation
* @ check_ready : callback to check link readiness
*
* Wait for @ link to become ready . @ check_ready should return
* positive number if @ link is ready , 0 if it isn ' t , - ENODEV if
* link doesn ' t seem to be occupied , other errno for other error
* conditions .
*
* Transient - ENODEV conditions are allowed for
* ATA_TMOUT_FF_WAIT .
*
* LOCKING :
* EH context .
*
* RETURNS :
* 0 if @ linke is ready before @ deadline ; otherwise , - errno .
*/
int ata_wait_ready ( struct ata_link * link , unsigned long deadline ,
int ( * check_ready ) ( struct ata_link * link ) )
{
unsigned long start = jiffies ;
2010-04-09 10:46:38 +00:00
unsigned long nodev_deadline ;
2008-04-07 13:47:19 +00:00
int warned = 0 ;
2010-04-09 10:46:38 +00:00
/* choose which 0xff timeout to use, read comment in libata.h */
if ( link - > ap - > host - > flags & ATA_HOST_PARALLEL_SCAN )
nodev_deadline = ata_deadline ( start , ATA_TMOUT_FF_WAIT_LONG ) ;
else
nodev_deadline = ata_deadline ( start , ATA_TMOUT_FF_WAIT ) ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/* Slave readiness can't be tested separately from master. On
* M / S emulation configuration , this function should be called
* only on the master and it will handle both master and slave .
*/
WARN_ON ( link = = link - > ap - > slave_link ) ;
2008-04-07 13:47:19 +00:00
if ( time_after ( nodev_deadline , deadline ) )
nodev_deadline = deadline ;
while ( 1 ) {
unsigned long now = jiffies ;
int ready , tmp ;
ready = tmp = check_ready ( link ) ;
if ( ready > 0 )
return 0 ;
2010-04-09 10:46:38 +00:00
/*
* - ENODEV could be transient . Ignore - ENODEV if link
2008-04-07 13:47:19 +00:00
* is online . Also , some SATA devices take a long
2010-04-09 10:46:38 +00:00
* time to clear 0xff after reset . Wait for
* ATA_TMOUT_FF_WAIT [ _LONG ] on - ENODEV if link isn ' t
* offline .
2008-04-07 13:47:19 +00:00
*
* Note that some PATA controllers ( pata_ali ) explode
* if status register is read more than once when
* there ' s no device attached .
*/
if ( ready = = - ENODEV ) {
if ( ata_link_online ( link ) )
ready = 0 ;
else if ( ( link - > ap - > flags & ATA_FLAG_SATA ) & &
! ata_link_offline ( link ) & &
time_before ( now , nodev_deadline ) )
ready = 0 ;
}
if ( ready )
return ready ;
if ( time_after ( now , deadline ) )
return - EBUSY ;
if ( ! warned & & time_after ( now , start + 5 * HZ ) & &
( deadline - now > 3 * HZ ) ) {
ata_link_printk ( link , KERN_WARNING ,
" link is slow to respond, please be patient "
" (ready=%d) \n " , tmp ) ;
warned = 1 ;
}
2010-09-06 15:56:29 +00:00
ata_msleep ( link - > ap , 50 ) ;
2008-04-07 13:47:19 +00:00
}
}
/**
* ata_wait_after_reset - wait for link to become ready after reset
* @ link : link to be waited on
* @ deadline : deadline jiffies for the operation
* @ check_ready : callback to check link readiness
*
* Wait for @ link to become ready after reset .
*
* LOCKING :
* EH context .
*
* RETURNS :
* 0 if @ linke is ready before @ deadline ; otherwise , - errno .
*/
2008-04-25 01:37:34 +00:00
int ata_wait_after_reset ( struct ata_link * link , unsigned long deadline ,
2008-04-07 13:47:19 +00:00
int ( * check_ready ) ( struct ata_link * link ) )
{
2010-09-06 15:56:29 +00:00
ata_msleep ( link - > ap , ATA_WAIT_AFTER_RESET ) ;
2008-04-07 13:47:19 +00:00
return ata_wait_ready ( link , deadline , check_ready ) ;
}
2006-05-31 09:27:46 +00:00
/**
2007-08-06 09:36:23 +00:00
* sata_link_debounce - debounce SATA phy status
* @ link : ATA link to debounce SATA phy status for
2006-05-31 09:27:46 +00:00
* @ params : timing parameters { interval , duratinon , timeout } in msec
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* @ deadline : deadline jiffies for the operation
2006-05-31 09:27:46 +00:00
*
2010-09-01 15:50:05 +00:00
* Make sure SStatus of @ link reaches stable state , determined by
2006-05-31 09:27:46 +00:00
* holding the same value where DET is not 1 for @ duration polled
* every @ interval , before @ timeout . Timeout constraints the
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* beginning of the stable state . Because DET gets stuck at 1 on
* some controllers after hot unplugging , this functions waits
2006-05-31 09:27:46 +00:00
* until timeout then returns 0 if DET is stable at 1.
*
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* @ timeout is further limited by @ deadline . The sooner of the
* two is used .
*
2006-05-31 09:27:46 +00:00
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno on failure .
*/
2007-08-06 09:36:23 +00:00
int sata_link_debounce ( struct ata_link * link , const unsigned long * params ,
unsigned long deadline )
2006-02-02 09:20:00 +00:00
{
2008-05-19 17:17:51 +00:00
unsigned long interval = params [ 0 ] ;
unsigned long duration = params [ 1 ] ;
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
unsigned long last_jiffies , t ;
2006-05-31 09:27:46 +00:00
u32 last , cur ;
int rc ;
2008-05-19 17:17:51 +00:00
t = ata_deadline ( jiffies , params [ 2 ] ) ;
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
if ( time_before ( t , deadline ) )
deadline = t ;
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_STATUS , & cur ) ) )
2006-05-31 09:27:46 +00:00
return rc ;
cur & = 0xf ;
last = cur ;
last_jiffies = jiffies ;
while ( 1 ) {
2010-09-06 15:56:29 +00:00
ata_msleep ( link - > ap , interval ) ;
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_STATUS , & cur ) ) )
2006-05-31 09:27:46 +00:00
return rc ;
cur & = 0xf ;
/* DET stable? */
if ( cur = = last ) {
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
if ( cur = = 1 & & time_before ( jiffies , deadline ) )
2006-05-31 09:27:46 +00:00
continue ;
2008-05-19 17:17:51 +00:00
if ( time_after ( jiffies ,
ata_deadline ( last_jiffies , duration ) ) )
2006-05-31 09:27:46 +00:00
return 0 ;
continue ;
}
/* unstable, start over */
last = cur ;
last_jiffies = jiffies ;
2007-07-16 05:29:40 +00:00
/* Check deadline. If debouncing failed, return
* - EPIPE to tell upper layer to lower link speed .
*/
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
if ( time_after ( jiffies , deadline ) )
2007-07-16 05:29:40 +00:00
return - EPIPE ;
2006-05-31 09:27:46 +00:00
}
}
/**
2007-08-06 09:36:23 +00:00
* sata_link_resume - resume SATA link
* @ link : ATA link to resume SATA
2006-05-31 09:27:46 +00:00
* @ params : timing parameters { interval , duratinon , timeout } in msec
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* @ deadline : deadline jiffies for the operation
2006-05-31 09:27:46 +00:00
*
2007-08-06 09:36:23 +00:00
* Resume SATA phy @ link and debounce it .
2006-05-31 09:27:46 +00:00
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno on failure .
*/
2007-08-06 09:36:23 +00:00
int sata_link_resume ( struct ata_link * link , const unsigned long * params ,
unsigned long deadline )
2006-05-31 09:27:46 +00:00
{
2010-01-11 02:14:44 +00:00
int tries = ATA_LINK_RESUME_TRIES ;
2008-04-07 13:47:19 +00:00
u32 scontrol , serror ;
2006-05-15 11:57:47 +00:00
int rc ;
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ) )
2006-05-15 11:57:47 +00:00
return rc ;
2006-02-02 09:20:00 +00:00
2010-01-11 02:14:44 +00:00
/*
* Writes to SControl sometimes get ignored under certain
* controllers ( ata_piix SIDPR ) . Make sure DET actually is
* cleared .
*/
do {
scontrol = ( scontrol & 0x0f0 ) | 0x300 ;
if ( ( rc = sata_scr_write ( link , SCR_CONTROL , scontrol ) ) )
return rc ;
/*
* Some PHYs react badly if SStatus is pounded
* immediately after resuming . Delay 200 ms before
* debouncing .
*/
2010-09-06 15:56:29 +00:00
ata_msleep ( link - > ap , 200 ) ;
2006-05-15 11:57:47 +00:00
2010-01-11 02:14:44 +00:00
/* is SControl restored correctly? */
if ( ( rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ) )
return rc ;
} while ( ( scontrol & 0xf0f ) ! = 0x300 & & - - tries ) ;
2006-02-02 09:20:00 +00:00
2010-01-11 02:14:44 +00:00
if ( ( scontrol & 0xf0f ) ! = 0x300 ) {
ata_link_printk ( link , KERN_ERR ,
" failed to resume link (SControl %X) \n " ,
scontrol ) ;
return 0 ;
}
if ( tries < ATA_LINK_RESUME_TRIES )
ata_link_printk ( link , KERN_WARNING ,
" link resume succeeded after %d retries \n " ,
ATA_LINK_RESUME_TRIES - tries ) ;
2006-02-02 09:20:00 +00:00
2008-04-07 13:47:19 +00:00
if ( ( rc = sata_link_debounce ( link , params , deadline ) ) )
return rc ;
2008-05-18 16:15:08 +00:00
/* clear SError, some PHYs require this even for SRST to work */
2008-04-07 13:47:19 +00:00
if ( ! ( rc = sata_scr_read ( link , SCR_ERROR , & serror ) ) )
rc = sata_scr_write ( link , SCR_ERROR , serror ) ;
2008-05-18 16:15:08 +00:00
return rc ! = - EINVAL ? rc : 0 ;
2006-02-02 09:20:00 +00:00
}
2010-09-01 15:50:05 +00:00
/**
* sata_link_scr_lpm - manipulate SControl IPM and SPM fields
* @ link : ATA link to manipulate SControl for
* @ policy : LPM policy to configure
* @ spm_wakeup : initiate LPM transition to active state
*
* Manipulate the IPM field of the SControl register of @ link
* according to @ policy . If @ policy is ATA_LPM_MAX_POWER and
* @ spm_wakeup is % true , the SPM field is manipulated to wake up
* the link . This function also clears PHYRDY_CHG before
* returning .
*
* LOCKING :
* EH context .
*
* RETURNS :
* 0 on succes , - errno otherwise .
*/
int sata_link_scr_lpm ( struct ata_link * link , enum ata_lpm_policy policy ,
bool spm_wakeup )
{
struct ata_eh_context * ehc = & link - > eh_context ;
bool woken_up = false ;
u32 scontrol ;
int rc ;
rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ;
if ( rc )
return rc ;
switch ( policy ) {
case ATA_LPM_MAX_POWER :
/* disable all LPM transitions */
scontrol | = ( 0x3 < < 8 ) ;
/* initiate transition to active state */
if ( spm_wakeup ) {
scontrol | = ( 0x4 < < 12 ) ;
woken_up = true ;
}
break ;
case ATA_LPM_MED_POWER :
/* allow LPM to PARTIAL */
scontrol & = ~ ( 0x1 < < 8 ) ;
scontrol | = ( 0x2 < < 8 ) ;
break ;
case ATA_LPM_MIN_POWER :
2011-03-04 18:24:11 +00:00
if ( ata_link_nr_enabled ( link ) > 0 )
/* no restrictions on LPM transitions */
scontrol & = ~ ( 0x3 < < 8 ) ;
else {
/* empty port, power off */
scontrol & = ~ 0xf ;
scontrol | = ( 0x1 < < 2 ) ;
}
2010-09-01 15:50:05 +00:00
break ;
default :
WARN_ON ( 1 ) ;
}
rc = sata_scr_write ( link , SCR_CONTROL , scontrol ) ;
if ( rc )
return rc ;
/* give the link time to transit out of LPM state */
if ( woken_up )
msleep ( 10 ) ;
/* clear PHYRDY_CHG from SError */
ehc - > i . serror & = ~ SERR_PHYRDY_CHG ;
return sata_scr_write ( link , SCR_ERROR , SERR_PHYRDY_CHG ) ;
}
2006-05-31 09:27:48 +00:00
/**
2008-04-07 13:47:18 +00:00
* ata_std_prereset - prepare for reset
2007-08-06 09:36:23 +00:00
* @ link : ATA link to be reset
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* @ deadline : deadline jiffies for the operation
2006-05-31 09:27:48 +00:00
*
2007-08-06 09:36:23 +00:00
* @ link is about to be reset . Initialize it . Failure from
2007-02-02 07:50:52 +00:00
* prereset makes libata abort whole reset sequence and give up
* that port , so prereset should be best - effort . It does its
* best to prepare for reset sequence but if things go wrong , it
* should just whine , not fail .
2006-05-31 09:27:48 +00:00
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno otherwise .
*/
2008-04-07 13:47:18 +00:00
int ata_std_prereset ( struct ata_link * link , unsigned long deadline )
2006-05-31 09:27:48 +00:00
{
2007-08-06 09:36:23 +00:00
struct ata_port * ap = link - > ap ;
2007-08-06 09:36:23 +00:00
struct ata_eh_context * ehc = & link - > eh_context ;
2006-07-03 07:07:26 +00:00
const unsigned long * timing = sata_ehc_deb_timing ( ehc ) ;
2006-05-31 09:27:48 +00:00
int rc ;
/* if we're about to do hardreset, nothing more to do */
if ( ehc - > i . action & ATA_EH_HARDRESET )
return 0 ;
2007-08-06 09:36:23 +00:00
/* if SATA, resume link */
2007-05-21 16:33:47 +00:00
if ( ap - > flags & ATA_FLAG_SATA ) {
2007-08-06 09:36:23 +00:00
rc = sata_link_resume ( link , timing , deadline ) ;
2007-02-02 07:50:52 +00:00
/* whine about phy resume failure but proceed */
if ( rc & & rc ! = - EOPNOTSUPP )
2007-08-06 09:36:23 +00:00
ata_link_printk ( link , KERN_WARNING , " failed to resume "
2006-05-31 09:27:48 +00:00
" link for reset (errno=%d) \n " , rc ) ;
}
2008-04-07 16:46:56 +00:00
/* no point in trying softreset on offline link */
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
if ( ata_phys_link_offline ( link ) )
2008-04-07 16:46:56 +00:00
ehc - > i . action & = ~ ATA_EH_SOFTRESET ;
2006-05-31 09:27:48 +00:00
return 0 ;
}
2006-01-24 08:05:22 +00:00
/**
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
* sata_link_hardreset - reset link via SATA phy reset
* @ link : link to reset
* @ timing : timing parameters { interval , duratinon , timeout } in msec
libata: add deadline support to prereset and reset methods
Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function. This patch is in preparation for EH timing
improvements.
* ata_wait_ready() never does busy sleep. It's only used from EH and
no wait in EH is that urgent. This function also prints 'be
patient' message automatically after 5 secs of waiting if more than
3 secs is remaining till deadline.
* ata_bus_post_reset() now fails with error code if any of its wait
fails. This is important because earlier reset tries will have
shorter timeout than the spec requires. If a device fails to
respond before the short timeout, reset should be retried with
longer timeout rather than silently ignoring the device.
There are three behavior differences.
1. Timeout is applied to both devices at once, not separately. This
is more consistent with what the spec says.
2. When a device passes devchk but fails to become ready before
deadline. Previouly, post_reset would just succeed and let
device classification remove the device. New code fails the
reset thus causing reset retry. After a few times, EH will give
up disabling the port.
3. When slave device passes devchk but fails to become accessible
(TF-wise) after reset. Original code disables dev1 after 30s
timeout and continues as if the device doesn't exist, while the
patched code fails reset. When this happens, new code fails
reset on whole port rather than proceeding with only the primary
device.
If the failing device is suffering transient problems, new code
retries reset which is a better behavior. If the failing device is
actually broken, the net effect is identical to it, but not to the
other device sharing the channel. In the previous code, reset would
have succeeded after 30s thus detecting the working one. In the new
code, reset fails and whole port gets disabled. IMO, it's a
pathological case anyway (broken device sharing bus with working
one) and doesn't really matter.
* ata_bus_softreset() is changed to return error code from
ata_bus_post_reset(). It used to return 0 unconditionally.
* Spin up waiting is to be removed and not converted to honor
deadline.
* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-02-02 07:50:52 +00:00
* @ deadline : deadline jiffies for the operation
2008-04-07 13:47:19 +00:00
* @ online : optional out parameter indicating link onlineness
* @ check_ready : optional callback to check link readiness
2006-01-24 08:05:22 +00:00
*
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
* SATA phy - reset @ link using DET bits of SControl register .
2008-04-07 13:47:19 +00:00
* After hardreset , link readiness is waited upon using
* ata_wait_ready ( ) if @ check_ready is specified . LLDs are
* allowed to not specify @ check_ready and wait itself after this
* function returns . Device classification is LLD ' s
* responsibility .
*
* * @ online is set to one iff reset succeeded and @ link is online
* after reset .
2006-01-24 08:05:22 +00:00
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , - errno otherwise .
*/
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
int sata_link_hardreset ( struct ata_link * link , const unsigned long * timing ,
2008-04-07 13:47:19 +00:00
unsigned long deadline ,
bool * online , int ( * check_ready ) ( struct ata_link * ) )
2006-01-24 08:05:22 +00:00
{
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
u32 scontrol ;
2006-05-15 11:57:47 +00:00
int rc ;
2006-03-31 16:38:18 +00:00
2006-01-24 08:05:22 +00:00
DPRINTK ( " ENTER \n " ) ;
2008-04-07 13:47:19 +00:00
if ( online )
* online = false ;
2007-08-06 09:36:23 +00:00
if ( sata_set_spd_needed ( link ) ) {
2006-04-02 11:53:28 +00:00
/* SATA spec says nothing about how to reconfigure
* spd . To be on the safe side , turn off phy during
* reconfiguration . This works for at least ICH7 AHCI
* and Sil3124 .
*/
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ) )
2006-11-01 08:59:53 +00:00
goto out ;
2006-05-15 11:57:47 +00:00
2006-07-05 19:06:13 +00:00
scontrol = ( scontrol & 0x0f0 ) | 0x304 ;
2006-05-15 11:57:47 +00:00
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_write ( link , SCR_CONTROL , scontrol ) ) )
2006-11-01 08:59:53 +00:00
goto out ;
2006-04-02 11:53:28 +00:00
2007-08-06 09:36:23 +00:00
sata_set_spd ( link ) ;
2006-04-02 11:53:28 +00:00
}
/* issue phy wake/reset */
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_read ( link , SCR_CONTROL , & scontrol ) ) )
2006-11-01 08:59:53 +00:00
goto out ;
2006-05-15 11:57:47 +00:00
2006-03-31 16:38:18 +00:00
scontrol = ( scontrol & 0x0f0 ) | 0x301 ;
2006-05-15 11:57:47 +00:00
2007-08-06 09:36:23 +00:00
if ( ( rc = sata_scr_write_flush ( link , SCR_CONTROL , scontrol ) ) )
2006-11-01 08:59:53 +00:00
goto out ;
2006-01-24 08:05:22 +00:00
2006-04-02 11:53:28 +00:00
/* Couldn't find anything in SATA I/II specs, but AHCI-1.1
2006-01-24 08:05:22 +00:00
* 10.4 .2 says at least 1 ms .
*/
2010-09-06 15:56:29 +00:00
ata_msleep ( link - > ap , 1 ) ;
2006-01-24 08:05:22 +00:00
2007-08-06 09:36:23 +00:00
/* bring link back */
rc = sata_link_resume ( link , timing , deadline ) ;
2008-04-07 13:47:19 +00:00
if ( rc )
goto out ;
/* if link is offline nothing more to do */
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
if ( ata_phys_link_offline ( link ) )
2008-04-07 13:47:19 +00:00
goto out ;
/* Link is online. From this point, -ENODEV too is an error. */
if ( online )
* online = true ;
2008-04-07 13:47:22 +00:00
if ( sata_pmp_supported ( link - > ap ) & & ata_is_host_link ( link ) ) {
2008-04-07 13:47:19 +00:00
/* If PMP is supported, we have to do follow-up SRST.
* Some PMPs don ' t send D2H Reg FIS after hardreset if
* the first port is empty . Wait only for
* ATA_TMOUT_PMP_SRST_WAIT .
*/
if ( check_ready ) {
unsigned long pmp_deadline ;
2008-05-19 17:17:51 +00:00
pmp_deadline = ata_deadline ( jiffies ,
ATA_TMOUT_PMP_SRST_WAIT ) ;
2008-04-07 13:47:19 +00:00
if ( time_after ( pmp_deadline , deadline ) )
pmp_deadline = deadline ;
ata_wait_ready ( link , pmp_deadline , check_ready ) ;
}
rc = - EAGAIN ;
goto out ;
}
rc = 0 ;
if ( check_ready )
rc = ata_wait_ready ( link , deadline , check_ready ) ;
2006-11-01 08:59:53 +00:00
out :
2008-05-18 16:15:05 +00:00
if ( rc & & rc ! = - EAGAIN ) {
/* online is set iff link is online && reset succeeded */
if ( online )
* online = false ;
2008-04-07 13:47:19 +00:00
ata_link_printk ( link , KERN_ERR ,
" COMRESET failed (errno=%d) \n " , rc ) ;
2008-05-18 16:15:05 +00:00
}
2006-11-01 08:59:53 +00:00
DPRINTK ( " EXIT, rc=%d \n " , rc ) ;
return rc ;
}
2008-04-07 13:47:19 +00:00
/**
* sata_std_hardreset - COMRESET w / o waiting or classification
* @ link : link to reset
* @ class : resulting class of attached device
* @ deadline : deadline jiffies for the operation
*
* Standard SATA COMRESET w / o waiting or classification .
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 if link offline , - EAGAIN if link online , - errno on errors .
*/
int sata_std_hardreset ( struct ata_link * link , unsigned int * class ,
unsigned long deadline )
{
const unsigned long * timing = sata_ehc_deb_timing ( & link - > eh_context ) ;
bool online ;
int rc ;
/* do hardreset */
rc = sata_link_hardreset ( link , timing , deadline , & online , NULL ) ;
return online ? - EAGAIN : rc ;
}
2006-01-24 08:05:22 +00:00
/**
2008-04-07 13:47:18 +00:00
* ata_std_postreset - standard postreset callback
2007-08-06 09:36:23 +00:00
* @ link : the target ata_link
2006-01-24 08:05:22 +00:00
* @ classes : classes of attached devices
*
* This function is invoked after a successful reset . Note that
* the device might have been reset more than once using
* different reset methods before postreset is invoked .
*
* LOCKING :
* Kernel thread context ( may sleep )
*/
2008-04-07 13:47:18 +00:00
void ata_std_postreset ( struct ata_link * link , unsigned int * classes )
2006-01-24 08:05:22 +00:00
{
2008-05-18 16:15:08 +00:00
u32 serror ;
2006-01-24 08:05:22 +00:00
DPRINTK ( " ENTER \n " ) ;
2008-05-18 16:15:08 +00:00
/* reset complete, clear SError */
if ( ! sata_scr_read ( link , SCR_ERROR , & serror ) )
sata_scr_write ( link , SCR_ERROR , serror ) ;
2006-01-24 08:05:22 +00:00
/* print link status */
2007-08-06 09:36:23 +00:00
sata_print_link_status ( link ) ;
2006-01-24 08:05:22 +00:00
DPRINTK ( " EXIT \n " ) ;
}
2006-03-05 08:55:58 +00:00
/**
* ata_dev_same_device - Determine whether new ID matches configured device
* @ dev : device to compare against
* @ new_class : class of the new device
* @ new_id : IDENTIFY page of the new device
*
* Compare @ new_class and @ new_id against @ dev and determine
* whether @ dev is the device indicated by @ new_class and
* @ new_id .
*
* LOCKING :
* None .
*
* RETURNS :
* 1 if @ dev matches @ new_class and @ new_id , 0 otherwise .
*/
2006-05-15 11:57:53 +00:00
static int ata_dev_same_device ( struct ata_device * dev , unsigned int new_class ,
const u16 * new_id )
2006-03-05 08:55:58 +00:00
{
const u16 * old_id = dev - > id ;
2007-01-02 11:18:49 +00:00
unsigned char model [ 2 ] [ ATA_ID_PROD_LEN + 1 ] ;
unsigned char serial [ 2 ] [ ATA_ID_SERNO_LEN + 1 ] ;
2006-03-05 08:55:58 +00:00
if ( dev - > class ! = new_class ) {
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_INFO , " class mismatch %d != %d \n " ,
dev - > class , new_class ) ;
2006-03-05 08:55:58 +00:00
return 0 ;
}
2007-01-02 11:18:49 +00:00
ata_id_c_string ( old_id , model [ 0 ] , ATA_ID_PROD , sizeof ( model [ 0 ] ) ) ;
ata_id_c_string ( new_id , model [ 1 ] , ATA_ID_PROD , sizeof ( model [ 1 ] ) ) ;
ata_id_c_string ( old_id , serial [ 0 ] , ATA_ID_SERNO , sizeof ( serial [ 0 ] ) ) ;
ata_id_c_string ( new_id , serial [ 1 ] , ATA_ID_SERNO , sizeof ( serial [ 1 ] ) ) ;
2006-03-05 08:55:58 +00:00
if ( strcmp ( model [ 0 ] , model [ 1 ] ) ) {
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_INFO , " model number mismatch "
" '%s' != '%s' \n " , model [ 0 ] , model [ 1 ] ) ;
2006-03-05 08:55:58 +00:00
return 0 ;
}
if ( strcmp ( serial [ 0 ] , serial [ 1 ] ) ) {
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_INFO , " serial number mismatch "
" '%s' != '%s' \n " , serial [ 0 ] , serial [ 1 ] ) ;
2006-03-05 08:55:58 +00:00
return 0 ;
}
return 1 ;
}
/**
2007-05-14 18:28:15 +00:00
* ata_dev_reread_id - Re - read IDENTIFY data
2007-06-19 08:10:50 +00:00
* @ dev : target ATA device
2006-11-10 09:08:10 +00:00
* @ readid_flags : read ID flags
2006-03-05 08:55:58 +00:00
*
* Re - read IDENTIFY page and make sure @ dev is still attached to
* the port .
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , negative errno otherwise
*/
2007-05-14 18:28:15 +00:00
int ata_dev_reread_id ( struct ata_device * dev , unsigned int readid_flags )
2006-03-05 08:55:58 +00:00
{
2006-04-02 09:51:52 +00:00
unsigned int class = dev - > class ;
2007-08-06 09:36:22 +00:00
u16 * id = ( void * ) dev - > link - > ap - > sector_buf ;
2006-03-05 08:55:58 +00:00
int rc ;
2006-05-15 11:57:35 +00:00
/* read ID data */
2006-11-10 09:08:10 +00:00
rc = ata_dev_read_id ( dev , & class , readid_flags , id ) ;
2006-03-05 08:55:58 +00:00
if ( rc )
2007-05-14 18:28:15 +00:00
return rc ;
2006-03-05 08:55:58 +00:00
/* is the device still there? */
2007-05-14 18:28:15 +00:00
if ( ! ata_dev_same_device ( dev , class , id ) )
return - ENODEV ;
2006-03-05 08:55:58 +00:00
2006-05-15 11:57:35 +00:00
memcpy ( dev - > id , id , sizeof ( id [ 0 ] ) * ATA_ID_WORDS ) ;
2007-05-14 18:28:15 +00:00
return 0 ;
}
/**
* ata_dev_revalidate - Revalidate ATA device
* @ dev : device to revalidate
2007-09-23 04:14:12 +00:00
* @ new_class : new class code
2007-05-14 18:28:15 +00:00
* @ readid_flags : read ID flags
*
* Re - read IDENTIFY page , make sure @ dev is still attached to the
* port and reconfigure it according to the new IDENTIFY page .
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , negative errno otherwise
*/
2007-09-23 04:14:12 +00:00
int ata_dev_revalidate ( struct ata_device * dev , unsigned int new_class ,
unsigned int readid_flags )
2007-05-14 18:28:15 +00:00
{
2007-05-14 18:28:15 +00:00
u64 n_sectors = dev - > n_sectors ;
2009-07-15 08:11:41 +00:00
u64 n_native_sectors = dev - > n_native_sectors ;
2007-05-14 18:28:15 +00:00
int rc ;
if ( ! ata_dev_enabled ( dev ) )
return - ENODEV ;
2007-09-23 04:14:12 +00:00
/* fail early if !ATA && !ATAPI to avoid issuing [P]IDENTIFY to PMP */
if ( ata_class_enabled ( new_class ) & &
2009-04-26 14:33:34 +00:00
new_class ! = ATA_DEV_ATA & &
new_class ! = ATA_DEV_ATAPI & &
new_class ! = ATA_DEV_SEMB ) {
2007-09-23 04:14:12 +00:00
ata_dev_printk ( dev , KERN_INFO , " class mismatch %u != %u \n " ,
dev - > class , new_class ) ;
rc = - ENODEV ;
goto fail ;
}
2007-05-14 18:28:15 +00:00
/* re-read ID */
rc = ata_dev_reread_id ( dev , readid_flags ) ;
if ( rc )
goto fail ;
2006-03-05 08:55:58 +00:00
/* configure device according to the new ID */
2006-11-01 09:38:52 +00:00
rc = ata_dev_configure ( dev ) ;
2007-05-14 18:28:15 +00:00
if ( rc )
goto fail ;
/* verify n_sectors hasn't changed */
2010-04-05 01:33:13 +00:00
if ( dev - > class ! = ATA_DEV_ATA | | ! n_sectors | |
dev - > n_sectors = = n_sectors )
return 0 ;
/* n_sectors has changed */
ata_dev_printk ( dev , KERN_WARNING , " n_sectors mismatch %llu != %llu \n " ,
( unsigned long long ) n_sectors ,
( unsigned long long ) dev - > n_sectors ) ;
/*
* Something could have caused HPA to be unlocked
* involuntarily . If n_native_sectors hasn ' t changed and the
* new size matches it , keep the device .
*/
if ( dev - > n_native_sectors = = n_native_sectors & &
dev - > n_sectors > n_sectors & & dev - > n_sectors = = n_native_sectors ) {
ata_dev_printk ( dev , KERN_WARNING ,
" new n_sectors matches native, probably "
2010-05-15 18:09:33 +00:00
" late HPA unlock, n_sectors updated \n " ) ;
/* use the larger n_sectors */
2010-04-05 01:33:13 +00:00
return 0 ;
2007-05-14 18:28:15 +00:00
}
2010-04-05 01:33:13 +00:00
/*
* Some BIOSes boot w / o HPA but resume w / HPA locked . Try
* unlocking HPA in those cases .
*
* https : //bugzilla.kernel.org/show_bug.cgi?id=15396
*/
if ( dev - > n_native_sectors = = n_native_sectors & &
dev - > n_sectors < n_sectors & & n_sectors = = n_native_sectors & &
! ( dev - > horkage & ATA_HORKAGE_BROKEN_HPA ) ) {
ata_dev_printk ( dev , KERN_WARNING ,
" old n_sectors matches native, probably "
" late HPA lock, will try to unlock HPA \n " ) ;
/* try unlocking HPA */
dev - > flags | = ATA_DFLAG_UNLOCK_HPA ;
rc = - EIO ;
} else
rc = - ENODEV ;
2006-03-05 08:55:58 +00:00
2010-04-05 01:33:13 +00:00
/* restore original n_[native_]sectors and fail */
dev - > n_native_sectors = n_native_sectors ;
dev - > n_sectors = n_sectors ;
2006-03-05 08:55:58 +00:00
fail :
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_ERR , " revalidation failed (errno=%d) \n " , rc ) ;
2006-03-05 08:55:58 +00:00
return rc ;
}
2006-10-28 02:08:46 +00:00
struct ata_blacklist_entry {
const char * model_num ;
const char * model_rev ;
unsigned long horkage ;
} ;
static const struct ata_blacklist_entry ata_device_blacklist [ ] = {
/* Devices with DMA related problems under Linux */
{ " WDC AC11000H " , NULL , ATA_HORKAGE_NODMA } ,
{ " WDC AC22100H " , NULL , ATA_HORKAGE_NODMA } ,
{ " WDC AC32500H " , NULL , ATA_HORKAGE_NODMA } ,
{ " WDC AC33100H " , NULL , ATA_HORKAGE_NODMA } ,
{ " WDC AC31600H " , NULL , ATA_HORKAGE_NODMA } ,
{ " WDC AC32100H " , " 24.09P07 " , ATA_HORKAGE_NODMA } ,
{ " WDC AC23200L " , " 21.10N21 " , ATA_HORKAGE_NODMA } ,
{ " Compaq CRD-8241B " , NULL , ATA_HORKAGE_NODMA } ,
{ " CRD-8400B " , NULL , ATA_HORKAGE_NODMA } ,
2010-07-01 22:18:12 +00:00
{ " CRD-848[02]B " , NULL , ATA_HORKAGE_NODMA } ,
2006-10-28 02:08:46 +00:00
{ " CRD-84 " , NULL , ATA_HORKAGE_NODMA } ,
{ " SanDisk SDP3B " , NULL , ATA_HORKAGE_NODMA } ,
{ " SanDisk SDP3B-64 " , NULL , ATA_HORKAGE_NODMA } ,
{ " SANYO CD-ROM CRD " , NULL , ATA_HORKAGE_NODMA } ,
{ " HITACHI CDR-8 " , NULL , ATA_HORKAGE_NODMA } ,
2010-07-01 22:18:12 +00:00
{ " HITACHI CDR-8[34]35 " , NULL , ATA_HORKAGE_NODMA } ,
2006-10-28 02:08:46 +00:00
{ " Toshiba CD-ROM XM-6202B " , NULL , ATA_HORKAGE_NODMA } ,
{ " TOSHIBA CD-ROM XM-1702BC " , NULL , ATA_HORKAGE_NODMA } ,
{ " CD-532E-A " , NULL , ATA_HORKAGE_NODMA } ,
{ " E-IDE CD-ROM CR-840 " , NULL , ATA_HORKAGE_NODMA } ,
{ " CD-ROM Drive/F5A " , NULL , ATA_HORKAGE_NODMA } ,
{ " WPI CDD-820 " , NULL , ATA_HORKAGE_NODMA } ,
{ " SAMSUNG CD-ROM SC-148C " , NULL , ATA_HORKAGE_NODMA } ,
{ " SAMSUNG CD-ROM SC " , NULL , ATA_HORKAGE_NODMA } ,
{ " ATAPI CD-ROM DRIVE 40X MAXIMUM " , NULL , ATA_HORKAGE_NODMA } ,
{ " _NEC DV5800A " , NULL , ATA_HORKAGE_NODMA } ,
2007-10-19 10:42:56 +00:00
{ " SAMSUNG CD-ROM SN-124 " , " N001 " , ATA_HORKAGE_NODMA } ,
2007-05-21 18:31:03 +00:00
{ " Seagate STT20000A " , NULL , ATA_HORKAGE_NODMA } ,
2007-09-23 04:19:54 +00:00
/* Odd clown on sil3726/4726 PMPs */
2008-05-18 16:15:14 +00:00
{ " Config Disk " , NULL , ATA_HORKAGE_DISABLE } ,
2006-10-28 02:08:46 +00:00
2007-04-02 03:34:15 +00:00
/* Weird ATAPI devices */
2007-06-26 17:49:38 +00:00
{ " TORiSAN DVD-ROM DRD-N216 " , NULL , ATA_HORKAGE_MAX_SEC_128 } ,
2008-11-03 10:01:09 +00:00
{ " QUANTUM DAT DAT72-000 " , NULL , ATA_HORKAGE_ATAPI_MOD16_DMA } ,
2007-04-02 03:34:15 +00:00
2006-10-28 02:08:46 +00:00
/* Devices we expect to fail diagnostics */
/* Devices where NCQ should be avoided */
/* NCQ is slow */
2007-10-19 10:42:56 +00:00
{ " WDC WD740ADFD-00 " , NULL , ATA_HORKAGE_NONCQ } ,
2007-12-07 03:46:23 +00:00
{ " WDC WD740ADFD-00NLR1 " , NULL , ATA_HORKAGE_NONCQ , } ,
2007-02-28 06:21:23 +00:00
/* http://thread.gmane.org/gmane.linux.ide/14907 */
{ " FUJITSU MHT2060BH " , NULL , ATA_HORKAGE_NONCQ } ,
2007-03-27 05:43:44 +00:00
/* NCQ is broken */
2007-09-20 20:31:47 +00:00
{ " Maxtor * " , " BANC* " , ATA_HORKAGE_NONCQ } ,
2007-09-20 14:22:47 +00:00
{ " Maxtor 7V300F0 " , " VA111630 " , ATA_HORKAGE_NONCQ } ,
2007-10-04 02:06:56 +00:00
{ " ST380817AS " , " 3.42 " , ATA_HORKAGE_NONCQ } ,
2007-12-09 10:45:39 +00:00
{ " ST3160023AS " , " 3.42 " , ATA_HORKAGE_NONCQ } ,
2008-12-22 10:35:22 +00:00
{ " OCZ CORE_SSD " , " 02.10104 " , ATA_HORKAGE_NONCQ } ,
2007-09-20 20:31:47 +00:00
2008-11-27 04:36:48 +00:00
/* Seagate NCQ + FLUSH CACHE firmware bug */
2010-07-05 22:53:37 +00:00
{ " ST31500341AS " , " SD1[5-9] " , ATA_HORKAGE_NONCQ |
2008-11-27 04:36:48 +00:00
ATA_HORKAGE_FIRMWARE_WARN } ,
2008-12-11 04:42:42 +00:00
2010-07-05 22:53:37 +00:00
{ " ST31000333AS " , " SD1[5-9] " , ATA_HORKAGE_NONCQ |
2008-12-11 04:42:42 +00:00
ATA_HORKAGE_FIRMWARE_WARN } ,
2010-07-05 22:53:37 +00:00
{ " ST3640[36]23AS " , " SD1[5-9] " , ATA_HORKAGE_NONCQ |
2008-12-11 04:42:42 +00:00
ATA_HORKAGE_FIRMWARE_WARN } ,
2010-07-05 22:53:37 +00:00
{ " ST3320[68]13AS " , " SD1[5-9] " , ATA_HORKAGE_NONCQ |
2008-11-27 04:36:48 +00:00
ATA_HORKAGE_FIRMWARE_WARN } ,
2007-04-03 04:05:29 +00:00
/* Blacklist entries taken from Silicon Image 3124/3132
Windows driver . inf file - also several Linux problem reports */
{ " HTS541060G9SA00 " , " MB3OC60D " , ATA_HORKAGE_NONCQ , } ,
{ " HTS541080G9SA00 " , " MB4OC60D " , ATA_HORKAGE_NONCQ , } ,
{ " HTS541010G9SA00 " , " MBZOC60D " , ATA_HORKAGE_NONCQ , } ,
2006-10-28 02:08:46 +00:00
2010-04-05 01:51:26 +00:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=15573 */
{ " C300-CTFDDAC128MAG " , " 0001 " , ATA_HORKAGE_NONCQ , } ,
2007-08-29 02:58:33 +00:00
/* devices which puke on READ_NATIVE_MAX */
{ " HDS724040KLSA80 " , " KFAOA20N " , ATA_HORKAGE_BROKEN_HPA , } ,
{ " WDC WD3200JD-00KLB0 " , " WD-WCAMR1130137 " , ATA_HORKAGE_BROKEN_HPA } ,
{ " WDC WD2500JD-00HBB0 " , " WD-WMAL71490727 " , ATA_HORKAGE_BROKEN_HPA } ,
{ " MAXTOR 6L080L4 " , " A93.0500 " , ATA_HORKAGE_BROKEN_HPA } ,
2006-10-28 02:08:46 +00:00
2009-08-06 16:59:15 +00:00
/* this one allows HPA unlocking but fails IOs on the area */
{ " OCZ-VERTEX " , " 1.30 " , ATA_HORKAGE_BROKEN_HPA } ,
2007-09-29 08:06:48 +00:00
/* Devices which report 1 sector over size HPA */
{ " ST340823A " , NULL , ATA_HORKAGE_HPA_SIZE , } ,
{ " ST320413A " , NULL , ATA_HORKAGE_HPA_SIZE , } ,
2008-02-19 00:41:25 +00:00
{ " ST310211A " , NULL , ATA_HORKAGE_HPA_SIZE , } ,
2007-09-29 08:06:48 +00:00
2007-11-05 22:58:58 +00:00
/* Devices which get the IVB wrong */
{ " QUANTUM FIREBALLlct10 05 " , " A03.0900 " , ATA_HORKAGE_IVB , } ,
2008-04-29 13:08:36 +00:00
/* Maybe we should just blacklist TSSTcorp... */
2010-07-01 22:18:12 +00:00
{ " TSSTcorp CDDVDW SH-S202[HJN] " , " SB0[01] " , ATA_HORKAGE_IVB , } ,
2007-11-05 22:58:58 +00:00
2008-08-27 13:23:18 +00:00
/* Devices that do not need bridging limits applied */
{ " MTRON MSP-SATA* " , NULL , ATA_HORKAGE_BRIDGE_OK , } ,
2009-01-29 11:31:36 +00:00
/* Devices which aren't very happy with higher link speeds */
{ " WD My Book " , NULL , ATA_HORKAGE_1_5_GBPS , } ,
2009-07-09 00:27:50 +00:00
/*
* Devices which choke on SETXFER . Applies only if both the
* device and controller are SATA .
*/
2011-06-22 10:13:10 +00:00
{ " PIONEER DVD-RW DVRTD08 " , NULL , ATA_HORKAGE_NOSETXFER } ,
{ " PIONEER DVD-RW DVR-212D " , NULL , ATA_HORKAGE_NOSETXFER } ,
{ " PIONEER DVD-RW DVR-216D " , NULL , ATA_HORKAGE_NOSETXFER } ,
2009-07-09 00:27:50 +00:00
2006-10-28 02:08:46 +00:00
/* End Marker */
{ }
2005-04-16 22:20:36 +00:00
} ;
2006-03-24 14:56:57 +00:00
2010-07-01 22:16:14 +00:00
/**
* glob_match - match a text string against a glob - style pattern
* @ text : the string to be examined
* @ pattern : the glob - style pattern to be matched against
*
* Either / both of text and pattern can be empty strings .
*
* Match text against a glob - style pattern , with wildcards and simple sets :
*
* ? matches any single character .
* * matches any run of characters .
* [ xyz ] matches a single character from the set : x , y , or z .
2010-07-05 14:25:45 +00:00
* [ a - d ] matches a single character from the range : a , b , c , or d .
* [ a - d0 - 9 ] matches a single character from either range .
2010-07-01 22:16:14 +00:00
*
2010-07-05 14:25:45 +00:00
* The special characters ? , [ , - , or * , can be matched using a set , eg . [ * ]
* Behaviour with malformed patterns is undefined , though generally reasonable .
2010-07-01 22:16:14 +00:00
*
2010-08-09 23:28:30 +00:00
* Sample patterns : " SD1? " , " SD1[0-5] " , " *R0 " , " SD*1?[012]*xx "
2010-07-01 22:16:14 +00:00
*
* This function uses one level of recursion per ' * ' in pattern .
* Since it calls _nothing_ else , and has _no_ explicit local variables ,
* this will not cause stack problems for any reasonable use here .
*
* RETURNS :
* 0 on match , 1 otherwise .
*/
static int glob_match ( const char * text , const char * pattern )
2007-09-20 20:31:47 +00:00
{
2010-07-01 22:16:14 +00:00
do {
/* Match single character or a '?' wildcard */
if ( * text = = * pattern | | * pattern = = ' ? ' ) {
if ( ! * pattern + + )
return 0 ; /* End of both strings: match */
} else {
/* Match single char against a '[' bracketed ']' pattern set */
if ( ! * text | | * pattern ! = ' [ ' )
break ; /* Not a pattern set */
2010-07-05 14:25:45 +00:00
while ( * + + pattern & & * pattern ! = ' ] ' & & * text ! = * pattern ) {
if ( * pattern = = ' - ' & & * ( pattern - 1 ) ! = ' [ ' )
if ( * text > * ( pattern - 1 ) & & * text < * ( pattern + 1 ) ) {
+ + pattern ;
break ;
}
}
2010-07-01 22:16:14 +00:00
if ( ! * pattern | | * pattern = = ' ] ' )
return 1 ; /* No match */
while ( * pattern & & * pattern + + ! = ' ] ' ) ;
}
} while ( * + + text & & * pattern ) ;
/* Match any run of chars against a '*' wildcard */
if ( * pattern = = ' * ' ) {
if ( ! * + + pattern )
return 0 ; /* Match: avoid recursion at end of pattern */
/* Loop to handle additional pattern chars after the wildcard */
while ( * text ) {
if ( glob_match ( text , pattern ) = = 0 )
return 0 ; /* Remainder matched */
+ + text ; /* Absorb (match) this char and try again */
2007-10-15 19:43:12 +00:00
}
}
2010-07-01 22:16:14 +00:00
if ( ! * text & & ! * pattern )
return 0 ; /* End of both strings: match */
return 1 ; /* No match */
2007-09-20 20:31:47 +00:00
}
2011-02-15 06:13:24 +00:00
2007-07-05 04:31:27 +00:00
static unsigned long ata_dev_blacklisted ( const struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2007-01-02 11:19:40 +00:00
unsigned char model_num [ ATA_ID_PROD_LEN + 1 ] ;
unsigned char model_rev [ ATA_ID_FW_REV_LEN + 1 ] ;
2006-10-28 02:08:46 +00:00
const struct ata_blacklist_entry * ad = ata_device_blacklist ;
2006-06-22 05:00:25 +00:00
2007-01-02 11:19:40 +00:00
ata_id_c_string ( dev - > id , model_num , ATA_ID_PROD , sizeof ( model_num ) ) ;
ata_id_c_string ( dev - > id , model_rev , ATA_ID_FW_REV , sizeof ( model_rev ) ) ;
2005-04-16 22:20:36 +00:00
2006-10-28 02:08:46 +00:00
while ( ad - > model_num ) {
2010-07-01 22:16:14 +00:00
if ( ! glob_match ( model_num , ad - > model_num ) ) {
2006-10-28 02:08:46 +00:00
if ( ad - > model_rev = = NULL )
return ad - > horkage ;
2010-07-01 22:16:14 +00:00
if ( ! glob_match ( model_rev , ad - > model_rev ) )
2006-10-28 02:08:46 +00:00
return ad - > horkage ;
2006-03-22 15:54:04 +00:00
}
2006-10-28 02:08:46 +00:00
ad + + ;
2006-03-22 15:54:04 +00:00
}
2005-04-16 22:20:36 +00:00
return 0 ;
}
2006-10-28 02:08:46 +00:00
static int ata_dma_blacklisted ( const struct ata_device * dev )
{
/* We don't support polling DMA.
* DMA blacklist those ATAPI devices with CDB - intr ( and use PIO )
* if the LLDD handles only interrupts in the HSM_ST_LAST state .
*/
2007-08-06 09:36:22 +00:00
if ( ( dev - > link - > ap - > flags & ATA_FLAG_PIO_POLLING ) & &
2006-10-28 02:08:46 +00:00
( dev - > flags & ATA_DFLAG_CDB_INTR ) )
return 1 ;
2007-07-05 04:31:27 +00:00
return ( dev - > horkage & ATA_HORKAGE_NODMA ) ? 1 : 0 ;
2006-10-28 02:08:46 +00:00
}
2007-11-05 22:58:58 +00:00
/**
* ata_is_40wire - check drive side detection
* @ dev : device
*
* Perform drive side detection decoding , allowing for device vendors
* who can ' t follow the documentation .
*/
static int ata_is_40wire ( struct ata_device * dev )
{
if ( dev - > horkage & ATA_HORKAGE_IVB )
return ata_drive_40wire_relaxed ( dev - > id ) ;
return ata_drive_40wire ( dev - > id ) ;
}
2008-03-28 21:33:46 +00:00
/**
* cable_is_40wire - 40 / 80 / SATA decider
* @ ap : port to consider
*
* This function encapsulates the policy for speed management
* in one place . At the moment we don ' t cache the result but
* there is a good case for setting ap - > cbl to the result when
* we are called with unknown cables ( and figuring out if it
* impacts hotplug at all ) .
*
* Return 1 if the cable appears to be 40 wire .
*/
static int cable_is_40wire ( struct ata_port * ap )
{
struct ata_link * link ;
struct ata_device * dev ;
2008-10-27 10:59:23 +00:00
/* If the controller thinks we are 40 wire, we are. */
2008-03-28 21:33:46 +00:00
if ( ap - > cbl = = ATA_CBL_PATA40 )
return 1 ;
2008-10-27 10:59:23 +00:00
/* If the controller thinks we are 80 wire, we are. */
2008-03-28 21:33:46 +00:00
if ( ap - > cbl = = ATA_CBL_PATA80 | | ap - > cbl = = ATA_CBL_SATA )
return 0 ;
2008-10-27 10:59:23 +00:00
/* If the system is known to be 40 wire short cable (eg
* laptop ) , then we allow 80 wire modes even if the drive
* isn ' t sure .
*/
2008-04-08 15:37:50 +00:00
if ( ap - > cbl = = ATA_CBL_PATA40_SHORT )
return 0 ;
2008-10-27 10:59:23 +00:00
/* If the controller doesn't know, we scan.
*
* Note : We look for all 40 wire detects at this point . Any
* 80 wire detect is taken to be 80 wire cable because
* - in many setups only the one drive ( slave if present ) will
* give a valid detect
* - if you have a non detect capable drive you don ' t want it
* to colour the choice
*/
2008-11-03 11:03:17 +00:00
ata_for_each_link ( link , ap , EDGE ) {
ata_for_each_dev ( dev , link , ENABLED ) {
if ( ! ata_is_40wire ( dev ) )
2008-03-28 21:33:46 +00:00
return 0 ;
}
}
return 1 ;
}
2006-03-05 19:31:57 +00:00
/**
* ata_dev_xfermask - Compute supported xfermask of the given device
* @ dev : Device to compute xfermask for
*
2006-03-24 05:07:50 +00:00
* Compute supported xfermask of @ dev and store it in
* dev - > * _mask . This function is responsible for applying all
* known limits including host controller limits , device
* blacklist , etc . . .
2006-03-05 19:31:57 +00:00
*
* LOCKING :
* None .
*/
2006-05-15 11:57:53 +00:00
static void ata_dev_xfermask ( struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_link * link = dev - > link ;
struct ata_port * ap = link - > ap ;
2006-08-24 07:19:22 +00:00
struct ata_host * host = ap - > host ;
2006-03-05 19:31:57 +00:00
unsigned long xfer_mask ;
2005-04-16 22:20:36 +00:00
2006-08-10 07:59:07 +00:00
/* controller modes available */
2006-04-02 08:54:47 +00:00
xfer_mask = ata_pack_xfermask ( ap - > pio_mask ,
ap - > mwdma_mask , ap - > udma_mask ) ;
2007-03-06 10:37:51 +00:00
/* drive modes available */
2006-08-10 07:59:07 +00:00
xfer_mask & = ata_pack_xfermask ( dev - > pio_mask ,
dev - > mwdma_mask , dev - > udma_mask ) ;
xfer_mask & = ata_id_xfermask ( dev - > id ) ;
2006-04-02 08:54:47 +00:00
2006-08-10 17:52:12 +00:00
/*
* CFA Advanced TrueIDE timings are not allowed on a shared
* cable
*/
if ( ata_dev_pair ( dev ) ) {
/* No PIO5 or PIO6 */
xfer_mask & = ~ ( 0x03 < < ( ATA_SHIFT_PIO + 5 ) ) ;
/* No MWDMA3 or MWDMA 4 */
xfer_mask & = ~ ( 0x03 < < ( ATA_SHIFT_MWDMA + 3 ) ) ;
}
2006-08-10 07:59:07 +00:00
if ( ata_dma_blacklisted ( dev ) ) {
xfer_mask & = ~ ( ATA_MASK_MWDMA | ATA_MASK_UDMA ) ;
2006-05-15 11:57:56 +00:00
ata_dev_printk ( dev , KERN_WARNING ,
" device is on DMA blacklist, disabling DMA \n " ) ;
2006-08-10 07:59:07 +00:00
}
2006-03-05 19:31:57 +00:00
Fix simplex adapters with libata
Recently I got my hands on nVidia's MCP61 PM-AM board, and
it contains IDE chip configured by BIOS with only primary
channel enabled. This confuses code which probes for
device DMA capabilities - it gets 0x60 (happy duplex
device) from primary channel BMDMA, but 0xFF (nobody here)
from secondary channel BMDMA. Due to this code then believes
that chip is simplex. I do not address this problem in
my patch, as I'm not sure how to handle this. Probably
ata_pci_init_one should have bitmap of enabled/possible
interfaces instead of their count, but it looks like
quite intrusive change, and maybe we do not care - for device
with only one channel simplex and regular DMA engines are
same.
But making device simplex pointed out that support for
DMA on simplex devices is currently broken - ata_dev_xfermask
tests whether device is simplex and if it is whether DMA
engine was assigned to this port. If not then it strips
out DMA bits from device. Problem is that code which assigns
DMA engine to port in ata_set_mode first detect device
mode and assigns DMA engine to channel only if some DMA
capable device was found.
And as xfermask stripped out DMA bits, host->simplex_claimed
is always NULL with current implementation.
By allowing DMA either if simplex_claimed is NULL or if it
points to current port DMA can be finally used - it gets
assigned to first port which contains any DMA capable
device.
Before:
pata_amd 0000:00:06.0: version 0.2.8
PCI: Setting latency timer of device 0000:00:06.0 to 64
ata5: PATA max UDMA/133 cmd 0x000101f0 ctl 0x000103f6 bmdma 0x0001f000 irq 14
ata6: PATA max UDMA/133 cmd 0x00010170 ctl 0x00010376 bmdma 0x0001f008 irq 15
scsi4 : pata_amd
ata5.00: ATAPI, max UDMA/66
ata5.00: simplex DMA is claimed by other device, disabling DMA
ata5.00: configured for PIO4
scsi5 : pata_amd
ata6: port disabled. ignoring.
ata6: reset failed, giving up
scsi 4:0:0:0: CD-ROM ATAPI DVD W DH16W1P LG12 PQ: 0 ANSI: 5
After:
pata_amd 0000:00:06.0: version 0.2.8
PCI: Setting latency timer of device 0000:00:06.0 to 64
ata5: PATA max UDMA/133 cmd 0x000101f0 ctl 0x000103f6 bmdma 0x0001f000 irq 14
ata6: PATA max UDMA/133 cmd 0x00010170 ctl 0x00010376 bmdma 0x0001f008 irq 15
scsi4 : pata_amd
ata5.00: ATAPI, max UDMA/66
ata5.00: configured for UDMA/33
scsi5 : pata_amd
ata6: port disabled. ignoring.
ata6: reset failed, giving up
scsi 4:0:0:0: CD-ROM ATAPI DVD W DH16W1P LG12 PQ: 0 ANSI: 5
Signed-off-by: Petr Vandrovec <petr@vandrovec.name>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-03-08 09:12:12 +00:00
if ( ( host - > flags & ATA_HOST_SIMPLEX ) & &
2007-10-19 10:42:56 +00:00
host - > simplex_claimed & & host - > simplex_claimed ! = ap ) {
2006-08-10 07:59:07 +00:00
xfer_mask & = ~ ( ATA_MASK_MWDMA | ATA_MASK_UDMA ) ;
ata_dev_printk ( dev , KERN_WARNING , " simplex DMA is claimed by "
" other device, disabling DMA \n " ) ;
2006-03-27 17:58:20 +00:00
}
2006-04-02 08:54:47 +00:00
2007-03-09 14:56:46 +00:00
if ( ap - > flags & ATA_FLAG_NO_IORDY )
xfer_mask & = ata_pio_mask_no_iordy ( dev ) ;
2006-03-27 17:58:20 +00:00
if ( ap - > ops - > mode_filter )
2007-03-09 14:34:07 +00:00
xfer_mask = ap - > ops - > mode_filter ( dev , xfer_mask ) ;
2006-03-27 17:58:20 +00:00
2007-03-06 10:37:51 +00:00
/* Apply cable rule here. Don't apply it early because when
* we handle hot plug the cable type can itself change .
* Check this last so that we know if the transfer rate was
* solely limited by the cable .
* Unknown or 80 wire cables reported host side are checked
* drive side as well . Cases where we know a 40 wire cable
* is used safely for 80 are not checked here .
*/
if ( xfer_mask & ( 0xF8 < < ATA_SHIFT_UDMA ) )
/* UDMA/44 or higher would be available */
2008-03-28 21:33:46 +00:00
if ( cable_is_40wire ( ap ) ) {
2007-10-19 10:42:56 +00:00
ata_dev_printk ( dev , KERN_WARNING ,
2007-03-06 10:37:51 +00:00
" limited to UDMA/33 due to 40-wire cable \n " ) ;
xfer_mask & = ~ ( 0xF8 < < ATA_SHIFT_UDMA ) ;
}
2006-04-02 08:54:47 +00:00
ata_unpack_xfermask ( xfer_mask , & dev - > pio_mask ,
& dev - > mwdma_mask , & dev - > udma_mask ) ;
2005-04-16 22:20:36 +00:00
}
/**
* ata_dev_set_xfermode - Issue SET FEATURES - XFER MODE command
* @ dev : Device to which command will be sent
*
2005-05-30 19:41:05 +00:00
* Issue SET FEATURES - XFER MODE command to device @ dev
* on port @ ap .
*
2005-04-16 22:20:36 +00:00
* LOCKING :
2005-05-30 23:49:12 +00:00
* PCI / etc . bus probe sem .
2006-03-24 06:25:31 +00:00
*
* RETURNS :
* 0 on success , AC_ERR_ * mask otherwise .
2005-04-16 22:20:36 +00:00
*/
2006-05-15 11:57:53 +00:00
static unsigned int ata_dev_set_xfermode ( struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2005-12-13 05:49:31 +00:00
struct ata_taskfile tf ;
2006-03-24 06:25:31 +00:00
unsigned int err_mask ;
2005-04-16 22:20:36 +00:00
/* set up set-features taskfile */
DPRINTK ( " set features - xfer mode \n " ) ;
2007-05-27 13:10:40 +00:00
/* Some controllers and ATAPI devices show flaky interrupt
* behavior after setting xfer mode . Use polling instead .
*/
2006-05-15 11:57:53 +00:00
ata_tf_init ( dev , & tf ) ;
2005-12-13 05:49:31 +00:00
tf . command = ATA_CMD_SET_FEATURES ;
tf . feature = SETFEATURES_XFER ;
2007-05-27 13:10:40 +00:00
tf . flags | = ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE | ATA_TFLAG_POLLING ;
2005-12-13 05:49:31 +00:00
tf . protocol = ATA_PROT_NODATA ;
2007-11-19 14:33:11 +00:00
/* If we are using IORDY we must send the mode setting command */
2007-11-24 02:12:14 +00:00
if ( ata_pio_need_iordy ( dev ) )
tf . nsect = dev - > xfer_mode ;
2007-11-19 14:33:11 +00:00
/* If the device has IORDY and the controller does not - turn it off */
else if ( ata_id_has_iordy ( dev - > id ) )
2007-11-24 02:12:14 +00:00
tf . nsect = 0x01 ;
2007-11-19 14:33:11 +00:00
else /* In the ancient relic department - skip all of this */
return 0 ;
2005-04-16 22:20:36 +00:00
2007-10-09 06:05:44 +00:00
err_mask = ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2007-08-15 07:57:11 +00:00
DPRINTK ( " EXIT, err_mask=%x \n " , err_mask ) ;
return err_mask ;
}
2010-09-01 15:50:05 +00:00
2007-08-15 07:57:11 +00:00
/**
2007-10-25 04:33:27 +00:00
* ata_dev_set_feature - Issue SET FEATURES - SATA FEATURES
2007-08-15 07:57:11 +00:00
* @ dev : Device to which command will be sent
* @ enable : Whether to enable or disable the feature
2007-10-25 04:33:27 +00:00
* @ feature : The sector count represents the feature to set
2007-08-15 07:57:11 +00:00
*
* Issue SET FEATURES - SATA FEATURES command to device @ dev
2007-10-25 04:33:27 +00:00
* on port @ ap with sector count
2007-08-15 07:57:11 +00:00
*
* LOCKING :
* PCI / etc . bus probe sem .
*
* RETURNS :
* 0 on success , AC_ERR_ * mask otherwise .
*/
2010-09-01 15:50:05 +00:00
unsigned int ata_dev_set_feature ( struct ata_device * dev , u8 enable , u8 feature )
2007-08-15 07:57:11 +00:00
{
struct ata_taskfile tf ;
unsigned int err_mask ;
/* set up set-features taskfile */
DPRINTK ( " set features - SATA features \n " ) ;
ata_tf_init ( dev , & tf ) ;
tf . command = ATA_CMD_SET_FEATURES ;
tf . feature = enable ;
tf . flags | = ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE ;
tf . protocol = ATA_PROT_NODATA ;
2007-10-25 04:33:27 +00:00
tf . nsect = feature ;
2007-08-15 07:57:11 +00:00
2007-10-09 06:05:44 +00:00
err_mask = ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2005-04-16 22:20:36 +00:00
2006-03-24 06:25:31 +00:00
DPRINTK ( " EXIT, err_mask=%x \n " , err_mask ) ;
return err_mask ;
2005-04-16 22:20:36 +00:00
}
2005-05-12 19:29:42 +00:00
/**
* ata_dev_init_params - Issue INIT DEV PARAMS command
* @ dev : Device to which command will be sent
2006-05-18 17:50:18 +00:00
* @ heads : Number of heads ( taskfile parameter )
* @ sectors : Number of sectors ( taskfile parameter )
2005-05-12 19:29:42 +00:00
*
* LOCKING :
2006-02-15 09:24:09 +00:00
* Kernel thread context ( may sleep )
*
* RETURNS :
* 0 on success , AC_ERR_ * mask otherwise .
2005-05-12 19:29:42 +00:00
*/
2006-05-15 11:57:53 +00:00
static unsigned int ata_dev_init_params ( struct ata_device * dev ,
u16 heads , u16 sectors )
2005-05-12 19:29:42 +00:00
{
2005-12-13 05:49:31 +00:00
struct ata_taskfile tf ;
2006-02-15 09:24:09 +00:00
unsigned int err_mask ;
2005-05-12 19:29:42 +00:00
/* Number of sectors per track 1-255. Number of heads 1-16 */
if ( sectors < 1 | | sectors > 255 | | heads < 1 | | heads > 16 )
2006-03-27 08:39:18 +00:00
return AC_ERR_INVALID ;
2005-05-12 19:29:42 +00:00
/* set up init dev params taskfile */
DPRINTK ( " init dev params \n " ) ;
2006-05-15 11:57:53 +00:00
ata_tf_init ( dev , & tf ) ;
2005-12-13 05:49:31 +00:00
tf . command = ATA_CMD_INIT_DEV_PARAMS ;
tf . flags | = ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE ;
tf . protocol = ATA_PROT_NODATA ;
tf . nsect = sectors ;
tf . device | = ( heads - 1 ) & 0x0f ; /* max head = num. of heads - 1 */
2005-05-12 19:29:42 +00:00
2007-10-09 06:05:44 +00:00
err_mask = ata_exec_internal ( dev , & tf , NULL , DMA_NONE , NULL , 0 , 0 ) ;
2007-08-08 13:28:49 +00:00
/* A clean abort indicates an original or just out of spec drive
and we should continue as we issue the setup based on the
drive reported working geometry */
if ( err_mask = = AC_ERR_DEV & & ( tf . feature & ATA_ABORTED ) )
err_mask = 0 ;
2005-05-12 19:29:42 +00:00
2006-02-15 09:24:09 +00:00
DPRINTK ( " EXIT, err_mask=%x \n " , err_mask ) ;
return err_mask ;
2005-05-12 19:29:42 +00:00
}
2005-04-16 22:20:36 +00:00
/**
2005-05-30 23:49:12 +00:00
* ata_sg_clean - Unmap DMA memory associated with command
* @ qc : Command containing DMA memory to be released
*
* Unmap all mapped DMA memory associated with this command .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
2006-08-24 07:19:22 +00:00
* spin_lock_irqsave ( host lock )
2005-04-16 22:20:36 +00:00
*/
2006-11-14 13:47:10 +00:00
void ata_sg_clean ( struct ata_queued_cmd * qc )
2005-04-16 22:20:36 +00:00
{
struct ata_port * ap = qc - > ap ;
2007-12-05 07:43:11 +00:00
struct scatterlist * sg = qc - > sg ;
2005-04-16 22:20:36 +00:00
int dir = qc - > dma_dir ;
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( sg = = NULL ) ;
2005-04-16 22:20:36 +00:00
2008-02-19 10:36:56 +00:00
VPRINTK ( " unmapping %u sg elements \n " , qc - > n_elem ) ;
2005-04-16 22:20:36 +00:00
2008-02-19 10:36:56 +00:00
if ( qc - > n_elem )
2009-02-27 08:35:43 +00:00
dma_unmap_sg ( ap - > dev , sg , qc - > orig_n_elem , dir ) ;
2005-04-16 22:20:36 +00:00
qc - > flags & = ~ ATA_QCFLAG_DMAMAP ;
2007-12-05 07:43:11 +00:00
qc - > sg = NULL ;
2005-04-16 22:20:36 +00:00
}
/**
2008-06-17 03:36:26 +00:00
* atapi_check_dma - Check whether ATAPI DMA can be supported
2005-04-16 22:20:36 +00:00
* @ qc : Metadata associated with taskfile to check
*
2005-05-30 19:41:05 +00:00
* Allow low - level driver to filter ATA PACKET commands , returning
* a status indicating whether or not it is OK to use DMA for the
* supplied PACKET command .
*
2005-04-16 22:20:36 +00:00
* LOCKING :
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
* spin_lock_irqsave ( host lock )
*
* RETURNS : 0 when ATAPI DMA can be used
* nonzero otherwise
*/
2008-06-17 03:36:26 +00:00
int atapi_check_dma ( struct ata_queued_cmd * qc )
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
{
struct ata_port * ap = qc - > ap ;
2006-03-25 10:11:12 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
/* Don't allow DMA if it isn't multiple of 16 bytes. Quite a
* few ATAPI devices choke on such DMA requests .
*/
2008-11-03 10:01:09 +00:00
if ( ! ( qc - > dev - > horkage & ATA_HORKAGE_ATAPI_MOD16_DMA ) & &
unlikely ( qc - > nbytes & 15 ) )
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
return 1 ;
2006-03-25 09:43:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
if ( ap - > ops - > check_atapi_dma )
return ap - > ops - > check_atapi_dma ( qc ) ;
2006-03-25 09:43:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
return 0 ;
}
2005-04-16 22:20:36 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
/**
* ata_std_qc_defer - Check whether a qc needs to be deferred
* @ qc : ATA command in question
*
* Non - NCQ commands cannot run with any other command , NCQ or
* not . As upper layer only knows the queue depth , we are
* responsible for maintaining exclusion . This function checks
* whether a new command @ qc can be issued .
*
* LOCKING :
* spin_lock_irqsave ( host lock )
*
* RETURNS :
* ATA_DEFER_ * if deferring is needed , 0 otherwise .
*/
int ata_std_qc_defer ( struct ata_queued_cmd * qc )
{
struct ata_link * link = qc - > dev - > link ;
2006-03-25 09:43:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
if ( qc - > tf . protocol = = ATA_PROT_NCQ ) {
if ( ! ata_tag_valid ( link - > active_tag ) )
return 0 ;
} else {
if ( ! ata_tag_valid ( link - > active_tag ) & & ! link - > sactive )
return 0 ;
}
2006-03-25 09:43:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
return ATA_DEFER_LINK ;
}
2006-03-25 09:45:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
void ata_noop_qc_prep ( struct ata_queued_cmd * qc ) { }
2005-04-16 22:20:36 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
/**
* ata_sg_init - Associate command with scatter - gather table .
* @ qc : Command to be associated
* @ sg : Scatter - gather table .
* @ n_elem : Number of elements in s / g table .
*
* Initialize the data - related elements of queued_cmd @ qc
* to point to a scatter - gather table @ sg , containing @ n_elem
* elements .
*
* LOCKING :
* spin_lock_irqsave ( host lock )
*/
void ata_sg_init ( struct ata_queued_cmd * qc , struct scatterlist * sg ,
unsigned int n_elem )
{
qc - > sg = sg ;
qc - > n_elem = n_elem ;
qc - > cursg = qc - > sg ;
}
2006-03-25 09:48:02 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
/**
* ata_sg_setup - DMA - map the scatter - gather table associated with a command .
* @ qc : Command with scatter - gather table to be mapped .
*
* DMA - map the scatter - gather table associated with queued_cmd @ qc .
*
* LOCKING :
* spin_lock_irqsave ( host lock )
*
* RETURNS :
* Zero on success , negative on error .
*
*/
static int ata_sg_setup ( struct ata_queued_cmd * qc )
{
struct ata_port * ap = qc - > ap ;
unsigned int n_elem ;
2005-04-16 22:20:36 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
VPRINTK ( " ENTER, ata%u \n " , ap - > print_id ) ;
2006-03-25 09:43:49 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
n_elem = dma_map_sg ( ap - > dev , qc - > sg , qc - > n_elem , qc - > dma_dir ) ;
if ( n_elem < 1 )
return - 1 ;
2006-03-25 09:48:02 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
DPRINTK ( " %d sg elements mapped \n " , n_elem ) ;
2009-02-27 08:35:43 +00:00
qc - > orig_n_elem = qc - > n_elem ;
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
qc - > n_elem = n_elem ;
qc - > flags | = ATA_QCFLAG_DMAMAP ;
2005-04-16 22:20:36 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
return 0 ;
2005-04-16 22:20:36 +00:00
}
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
/**
* swap_buf_le16 - swap halves of 16 - bit words in place
* @ buf : Buffer to swap
* @ buf_words : Number of 16 - bit words in buffer .
*
* Swap halves of 16 - bit words if needed to convert from
* little - endian byte order to native cpu byte order , or
* vice - versa .
*
* LOCKING :
* Inherited from caller .
*/
void swap_buf_le16 ( u16 * buf , unsigned int buf_words )
2006-03-05 06:29:09 +00:00
{
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
# ifdef __BIG_ENDIAN
unsigned int i ;
2006-03-05 06:29:09 +00:00
libata: reorganize SFF related stuff
* Move SFF related functions from libata-core.c to libata-sff.c.
ata_[bmdma_]sff_port_ops, ata_devchk(), ata_dev_try_classify(),
ata_std_dev_select(), ata_tf_to_host(), ata_busy_sleep(),
ata_wait_after_reset(), ata_wait_ready(), ata_bus_post_reset(),
ata_bus_softreset(), ata_bus_reset(), ata_std_softreset(),
sata_std_hardreset(), ata_fill_sg(), ata_fill_sg_dumb(),
ata_qc_prep(), ata_dump_qc_prep(), ata_data_xfer(),
ata_data_xfer_noirq(), ata_pio_sector(), ata_pio_sectors(),
atapi_send_cdb(), __atapi_pio_bytes(), atapi_pio_bytes(),
ata_hsm_ok_in_wq(), ata_hsm_qc_complete(), ata_hsm_move(),
ata_pio_task(), ata_qc_issue_prot(), ata_host_intr(),
ata_interrupt(), ata_std_ports()
* Make ata_pio_queue_task() global as it's now called from
libata-sff.c.
* Move SFF related stuff in include/linux/libata.h and
drivers/ata/libata.h into one place. While at it, move timing
constants into the global enum definition and fortify comments a
bit.
This patch strictly moves stuff around and as such doesn't cause any
functional difference.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
2008-03-25 13:16:41 +00:00
for ( i = 0 ; i < buf_words ; i + + )
buf [ i ] = le16_to_cpu ( buf [ i ] ) ;
# endif /* __BIG_ENDIAN */
2006-03-05 06:29:09 +00:00
}
2008-11-10 05:48:21 +00:00
/**
* ata_qc_new - Request an available ATA command , for queueing
2009-01-21 00:28:59 +00:00
* @ ap : target port
2008-11-10 05:48:21 +00:00
*
* LOCKING :
* None .
*/
static struct ata_queued_cmd * ata_qc_new ( struct ata_port * ap )
{
struct ata_queued_cmd * qc = NULL ;
unsigned int i ;
/* no command while frozen */
if ( unlikely ( ap - > pflags & ATA_PFLAG_FROZEN ) )
return NULL ;
/* the last tag is reserved for internal command. */
for ( i = 0 ; i < ATA_MAX_QUEUE - 1 ; i + + )
if ( ! test_and_set_bit ( i , & ap - > qc_allocated ) ) {
qc = __ata_qc_from_tag ( ap , i ) ;
break ;
}
if ( qc )
qc - > tag = i ;
return qc ;
}
2005-04-16 22:20:36 +00:00
/**
* ata_qc_new_init - Request an available ATA command , and initialize it
* @ dev : Device from whom we request an available command structure
*
* LOCKING :
2005-05-30 23:49:12 +00:00
* None .
2005-04-16 22:20:36 +00:00
*/
2008-11-10 05:48:21 +00:00
struct ata_queued_cmd * ata_qc_new_init ( struct ata_device * dev )
2005-04-16 22:20:36 +00:00
{
2007-08-06 09:36:22 +00:00
struct ata_port * ap = dev - > link - > ap ;
2005-04-16 22:20:36 +00:00
struct ata_queued_cmd * qc ;
2008-11-10 05:48:21 +00:00
qc = ata_qc_new ( ap ) ;
2005-04-16 22:20:36 +00:00
if ( qc ) {
qc - > scsicmd = NULL ;
qc - > ap = ap ;
qc - > dev = dev ;
2005-11-14 19:14:16 +00:00
ata_qc_reinit ( qc ) ;
2005-04-16 22:20:36 +00:00
}
return qc ;
}
2008-11-10 05:48:21 +00:00
/**
* ata_qc_free - free unused ata_queued_cmd
* @ qc : Command to complete
*
* Designed to free unused ata_queued_cmd object
* in case something prevents using it .
*
* LOCKING :
* spin_lock_irqsave ( host lock )
*/
void ata_qc_free ( struct ata_queued_cmd * qc )
{
2009-10-17 06:41:47 +00:00
struct ata_port * ap ;
2008-11-10 05:48:21 +00:00
unsigned int tag ;
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( qc = = NULL ) ; /* ata_qc_from_tag _might_ return NULL */
2009-10-17 06:41:47 +00:00
ap = qc - > ap ;
2008-11-10 05:48:21 +00:00
qc - > flags = 0 ;
tag = qc - > tag ;
if ( likely ( ata_tag_valid ( tag ) ) ) {
qc - > tag = ATA_TAG_POISON ;
clear_bit ( tag , & ap - > qc_allocated ) ;
}
}
2006-02-11 06:13:49 +00:00
void __ata_qc_complete ( struct ata_queued_cmd * qc )
2005-04-16 22:20:36 +00:00
{
2009-10-17 06:41:47 +00:00
struct ata_port * ap ;
struct ata_link * link ;
2006-05-15 12:03:43 +00:00
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( qc = = NULL ) ; /* ata_qc_from_tag _might_ return NULL */
WARN_ON_ONCE ( ! ( qc - > flags & ATA_QCFLAG_ACTIVE ) ) ;
2009-10-17 06:41:47 +00:00
ap = qc - > ap ;
link = qc - > dev - > link ;
2005-04-16 22:20:36 +00:00
if ( likely ( qc - > flags & ATA_QCFLAG_DMAMAP ) )
ata_sg_clean ( qc ) ;
2006-05-15 11:57:32 +00:00
/* command should be marked inactive atomically with qc completion */
2007-09-23 04:14:12 +00:00
if ( qc - > tf . protocol = = ATA_PROT_NCQ ) {
2007-08-06 09:36:22 +00:00
link - > sactive & = ~ ( 1 < < qc - > tag ) ;
2007-09-23 04:14:12 +00:00
if ( ! link - > sactive )
ap - > nr_active_links - - ;
} else {
2007-08-06 09:36:22 +00:00
link - > active_tag = ATA_TAG_POISON ;
2007-09-23 04:14:12 +00:00
ap - > nr_active_links - - ;
}
/* clear exclusive status */
if ( unlikely ( qc - > flags & ATA_QCFLAG_CLEAR_EXCL & &
ap - > excl_link = = link ) )
ap - > excl_link = NULL ;
2006-05-15 11:57:32 +00:00
2005-08-16 06:25:38 +00:00
/* atapi: mark qc as inactive to prevent the interrupt handler
* from completing the command twice later , before the error handler
* is called . ( when rc ! = 0 and atapi request sense is needed )
*/
qc - > flags & = ~ ATA_QCFLAG_ACTIVE ;
2006-05-15 12:03:43 +00:00
ap - > qc_active & = ~ ( 1 < < qc - > tag ) ;
2005-08-16 06:25:38 +00:00
2005-04-16 22:20:36 +00:00
/* call completion callback */
2006-01-23 04:09:36 +00:00
qc - > complete_fn ( qc ) ;
2005-04-16 22:20:36 +00:00
}
2006-11-14 13:37:35 +00:00
static void fill_result_tf ( struct ata_queued_cmd * qc )
{
struct ata_port * ap = qc - > ap ;
qc - > result_tf . flags = qc - > tf . flags ;
2008-04-07 13:47:20 +00:00
ap - > ops - > qc_fill_rtf ( qc ) ;
2006-11-14 13:37:35 +00:00
}
2007-11-27 10:28:58 +00:00
static void ata_verify_xfer ( struct ata_queued_cmd * qc )
{
struct ata_device * dev = qc - > dev ;
if ( ata_is_nodata ( qc - > tf . protocol ) )
return ;
if ( ( dev - > mwdma_mask | | dev - > udma_mask ) & & ata_is_pio ( qc - > tf . protocol ) )
return ;
dev - > flags & = ~ ATA_DFLAG_DUBIOUS_XFER ;
}
2006-05-15 11:58:05 +00:00
/**
* ata_qc_complete - Complete an active ATA command
* @ qc : Command to complete
*
2010-06-25 13:03:34 +00:00
* Indicate to the mid and upper layers that an ATA command has
* completed , with either an ok or not - ok status .
*
* Refrain from calling this function multiple times when
* successfully completing multiple NCQ commands .
* ata_qc_complete_multiple ( ) should be used instead , which will
* properly update IRQ expect state .
2006-05-15 11:58:05 +00:00
*
* LOCKING :
2006-08-24 07:19:22 +00:00
* spin_lock_irqsave ( host lock )
2006-05-15 11:58:05 +00:00
*/
void ata_qc_complete ( struct ata_queued_cmd * qc )
{
struct ata_port * ap = qc - > ap ;
/* XXX: New EH and old EH use different mechanisms to
* synchronize EH with regular execution path .
*
* In new EH , a failed qc is marked with ATA_QCFLAG_FAILED .
* Normal execution path is responsible for not accessing a
* failed qc . libata core enforces the rule by returning NULL
* from ata_qc_from_tag ( ) for failed qcs .
*
* Old EH depends on ata_qc_complete ( ) nullifying completion
* requests if ATA_QCFLAG_EH_SCHEDULED is set . Old EH does
* not synchronize with interrupt handler . Only PIO task is
* taken care of .
*/
if ( ap - > ops - > error_handler ) {
2007-10-25 09:22:44 +00:00
struct ata_device * dev = qc - > dev ;
struct ata_eh_info * ehi = & dev - > link - > eh_info ;
2006-05-15 11:58:05 +00:00
if ( unlikely ( qc - > err_mask ) )
qc - > flags | = ATA_QCFLAG_FAILED ;
2010-12-09 14:59:32 +00:00
/*
* Finish internal commands without any further processing
* and always with the result TF filled .
*/
if ( unlikely ( ata_tag_internal ( qc - > tag ) ) ) {
2009-10-16 04:00:51 +00:00
fill_result_tf ( qc ) ;
2010-12-09 14:59:32 +00:00
__ata_qc_complete ( qc ) ;
return ;
}
2009-10-16 04:00:51 +00:00
2010-12-09 14:59:32 +00:00
/*
* Non - internal qc has failed . Fill the result TF and
* summon EH .
*/
if ( unlikely ( qc - > flags & ATA_QCFLAG_FAILED ) ) {
fill_result_tf ( qc ) ;
ata_qc_schedule_eh ( qc ) ;
2009-10-16 04:00:51 +00:00
return ;
2006-05-15 11:58:05 +00:00
}
2009-09-17 09:45:27 +00:00
WARN_ON_ONCE ( ap - > pflags & ATA_PFLAG_FROZEN ) ;
2006-05-15 11:58:05 +00:00
/* read result TF if requested */
if ( qc - > flags & ATA_QCFLAG_RESULT_TF )
2006-11-14 13:37:35 +00:00
fill_result_tf ( qc ) ;
2006-05-15 11:58:05 +00:00
2007-10-25 09:22:44 +00:00
/* Some commands need post-processing after successful
* completion .
*/
switch ( qc - > tf . command ) {
case ATA_CMD_SET_FEATURES :
if ( qc - > tf . feature ! = SETFEATURES_WC_ON & &
qc - > tf . feature ! = SETFEATURES_WC_OFF )
break ;
/* fall through */
case ATA_CMD_INIT_DEV_PARAMS : /* CHS translation changed */
case ATA_CMD_SET_MULTI : /* multi_count changed */
/* revalidate device */
ehi - > dev_action [ dev - > devno ] | = ATA_EH_REVALIDATE ;
ata_port_schedule_eh ( ap ) ;
break ;
2007-10-25 09:30:36 +00:00
case ATA_CMD_SLEEP :
dev - > flags | = ATA_DFLAG_SLEEPING ;
break ;
2007-10-25 09:22:44 +00:00
}
2007-11-27 10:28:58 +00:00
if ( unlikely ( dev - > flags & ATA_DFLAG_DUBIOUS_XFER ) )
ata_verify_xfer ( qc ) ;
2006-05-15 11:58:05 +00:00
__ata_qc_complete ( qc ) ;
} else {
if ( qc - > flags & ATA_QCFLAG_EH_SCHEDULED )
return ;
/* read result TF if failed or requested */
if ( qc - > err_mask | | qc - > flags & ATA_QCFLAG_RESULT_TF )
2006-11-14 13:37:35 +00:00
fill_result_tf ( qc ) ;
2006-05-15 11:58:05 +00:00
__ata_qc_complete ( qc ) ;
}
}
2006-05-15 12:03:43 +00:00
/**
* ata_qc_complete_multiple - Complete multiple qcs successfully
* @ ap : port in question
* @ qc_active : new qc_active mask
*
* Complete in - flight commands . This functions is meant to be
* called from low - level driver ' s interrupt routine to complete
* requests normally . ap - > qc_active and @ qc_active is compared
* and commands are completed accordingly .
*
2010-06-25 13:03:34 +00:00
* Always use this function when completing multiple NCQ commands
* from IRQ handlers instead of calling ata_qc_complete ( )
* multiple times to keep IRQ expect status properly in sync .
*
2006-05-15 12:03:43 +00:00
* LOCKING :
2006-08-24 07:19:22 +00:00
* spin_lock_irqsave ( host lock )
2006-05-15 12:03:43 +00:00
*
* RETURNS :
* Number of completed commands on success , - errno otherwise .
*/
2008-04-07 13:47:20 +00:00
int ata_qc_complete_multiple ( struct ata_port * ap , u32 qc_active )
2006-05-15 12:03:43 +00:00
{
int nr_done = 0 ;
u32 done_mask ;
done_mask = ap - > qc_active ^ qc_active ;
if ( unlikely ( done_mask & qc_active ) ) {
ata_port_printk ( ap , KERN_ERR , " illegal qc_active transition "
" (%08x->%08x) \n " , ap - > qc_active , qc_active ) ;
return - EINVAL ;
}
2009-05-20 07:44:39 +00:00
while ( done_mask ) {
2006-05-15 12:03:43 +00:00
struct ata_queued_cmd * qc ;
2009-05-20 07:44:39 +00:00
unsigned int tag = __ffs ( done_mask ) ;
2006-05-15 12:03:43 +00:00
2009-05-20 07:44:39 +00:00
qc = ata_qc_from_tag ( ap , tag ) ;
if ( qc ) {
2006-05-15 12:03:43 +00:00
ata_qc_complete ( qc ) ;
nr_done + + ;
}
2009-05-20 07:44:39 +00:00
done_mask & = ~ ( 1 < < tag ) ;
2006-05-15 12:03:43 +00:00
}
return nr_done ;
}
2005-04-16 22:20:36 +00:00
/**
* ata_qc_issue - issue taskfile to device
* @ qc : command to issue to device
*
* Prepare an ATA command to submission to device .
* This includes mapping the data into a DMA - able
* area , filling in the S / G table , and finally
* writing the taskfile to hardware , starting the command .
*
* LOCKING :
2006-08-24 07:19:22 +00:00
* spin_lock_irqsave ( host lock )
2005-04-16 22:20:36 +00:00
*/
2006-03-31 11:41:11 +00:00
void ata_qc_issue ( struct ata_queued_cmd * qc )
2005-04-16 22:20:36 +00:00
{
struct ata_port * ap = qc - > ap ;
2007-08-06 09:36:22 +00:00
struct ata_link * link = qc - > dev - > link ;
2007-11-27 10:28:53 +00:00
u8 prot = qc - > tf . protocol ;
2005-04-16 22:20:36 +00:00
2006-05-15 12:03:43 +00:00
/* Make sure only one non-NCQ command is outstanding. The
* check is skipped for old EH because it reuses active qc to
* request ATAPI sense .
*/
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( ap - > ops - > error_handler & & ata_tag_valid ( link - > active_tag ) ) ;
2006-05-15 12:03:43 +00:00
2007-12-05 01:36:13 +00:00
if ( ata_is_ncq ( prot ) ) {
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( link - > sactive & ( 1 < < qc - > tag ) ) ;
2007-09-23 04:14:12 +00:00
if ( ! link - > sactive )
ap - > nr_active_links + + ;
2007-08-06 09:36:22 +00:00
link - > sactive | = 1 < < qc - > tag ;
2006-05-15 12:03:43 +00:00
} else {
2009-01-09 10:19:14 +00:00
WARN_ON_ONCE ( link - > sactive ) ;
2007-09-23 04:14:12 +00:00
ap - > nr_active_links + + ;
2007-08-06 09:36:22 +00:00
link - > active_tag = qc - > tag ;
2006-05-15 12:03:43 +00:00
}
2006-03-31 11:36:47 +00:00
qc - > flags | = ATA_QCFLAG_ACTIVE ;
2006-05-15 12:03:43 +00:00
ap - > qc_active | = 1 < < qc - > tag ;
2006-03-31 11:36:47 +00:00
2010-08-23 09:27:27 +00:00
/*
* We guarantee to LLDs that they will have at least one
2007-12-05 07:43:10 +00:00
* non - zero sg if the command is a data command .
*/
2010-08-23 09:27:27 +00:00
if ( WARN_ON_ONCE ( ata_is_data ( prot ) & &
( ! qc - > sg | | ! qc - > n_elem | | ! qc - > nbytes ) ) )
goto sys_err ;
2007-12-05 07:43:10 +00:00
2007-11-27 10:28:53 +00:00
if ( ata_is_dma ( prot ) | | ( ata_is_pio ( prot ) & &
2007-12-05 07:43:10 +00:00
( ap - > flags & ATA_FLAG_PIO_DMA ) ) )
2007-12-05 07:43:09 +00:00
if ( ata_sg_setup ( qc ) )
2010-08-23 09:27:27 +00:00
goto sys_err ;
2005-04-16 22:20:36 +00:00
libata: prefer hardreset
When both soft and hard resets are available, libata preferred
softreset till now. The logic behind it was to be softer to devices;
however, this doesn't really help much. Rationales for the change:
* BIOS may freeze lock certain things during boot and softreset can't
unlock those. This by itself is okay but during operation PHY event
or other error conditions can trigger hardreset and the device may
end up with different configuration.
For example, after a hardreset, previously unlockable HPA can be
unlocked resulting in different device size and thus revalidation
failure. Similar condition can occur during or after resume.
* Certain ATAPI devices require hardreset to recover after certain
error conditions. On PATA, this is done by issuing the DEVICE RESET
command. On SATA, COMRESET has equivalent effect. The problem is
that DEVICE RESET needs its own execution protocol.
For SFF controllers with bare TF access, it can be easily
implemented but more advanced controllers (e.g. ahci and sata_sil24)
require specialized implementations. Simply using hardreset solves
the problem nicely.
* COMRESET initialization sequence is the norm in SATA land and many
SATA devices don't work properly if only SRST is used. For example,
some PMPs behave this way and libata works around by always issuing
hardreset if the host supports PMP.
Like the above example, libata has developed a number of mechanisms
aiming to promote softreset to hardreset if softreset is not going
to work. This approach is time consuming and error prone.
Also, note that, dependingon how you read the specs, it could be
argued that PMP fan-out ports require COMRESET to start operation.
In fact, all the PMPs on the market except one don't work properly
if COMRESET is not issued to fan-out ports after PMP reset.
* COMRESET is an integral part of SATA connection and any working
device should be able to handle COMRESET properly. After all, it's
the way to signal hardreset during reboot. This is the most used
and recommended (at least by the ahci spec) method of resetting
devices.
So, this patch makes libata prefer hardreset over softreset by making
the following changes.
* Rename ATA_EH_RESET_MASK to ATA_EH_RESET and use it whereever
ATA_EH_{SOFT|HARD}RESET used to be used. ATA_EH_{SOFT|HARD}RESET is
now only used to tell prereset whether soft or hard reset will be
issued.
* Strip out now unneeded promote-to-hardreset logics from
ata_eh_reset(), ata_std_prereset(), sata_pmp_std_prereset() and
other places.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-01-23 15:05:14 +00:00
/* if device is sleeping, schedule reset and abort the link */
2007-10-25 09:30:36 +00:00
if ( unlikely ( qc - > dev - > flags & ATA_DFLAG_SLEEPING ) ) {
libata: prefer hardreset
When both soft and hard resets are available, libata preferred
softreset till now. The logic behind it was to be softer to devices;
however, this doesn't really help much. Rationales for the change:
* BIOS may freeze lock certain things during boot and softreset can't
unlock those. This by itself is okay but during operation PHY event
or other error conditions can trigger hardreset and the device may
end up with different configuration.
For example, after a hardreset, previously unlockable HPA can be
unlocked resulting in different device size and thus revalidation
failure. Similar condition can occur during or after resume.
* Certain ATAPI devices require hardreset to recover after certain
error conditions. On PATA, this is done by issuing the DEVICE RESET
command. On SATA, COMRESET has equivalent effect. The problem is
that DEVICE RESET needs its own execution protocol.
For SFF controllers with bare TF access, it can be easily
implemented but more advanced controllers (e.g. ahci and sata_sil24)
require specialized implementations. Simply using hardreset solves
the problem nicely.
* COMRESET initialization sequence is the norm in SATA land and many
SATA devices don't work properly if only SRST is used. For example,
some PMPs behave this way and libata works around by always issuing
hardreset if the host supports PMP.
Like the above example, libata has developed a number of mechanisms
aiming to promote softreset to hardreset if softreset is not going
to work. This approach is time consuming and error prone.
Also, note that, dependingon how you read the specs, it could be
argued that PMP fan-out ports require COMRESET to start operation.
In fact, all the PMPs on the market except one don't work properly
if COMRESET is not issued to fan-out ports after PMP reset.
* COMRESET is an integral part of SATA connection and any working
device should be able to handle COMRESET properly. After all, it's
the way to signal hardreset during reboot. This is the most used
and recommended (at least by the ahci spec) method of resetting
devices.
So, this patch makes libata prefer hardreset over softreset by making
the following changes.
* Rename ATA_EH_RESET_MASK to ATA_EH_RESET and use it whereever
ATA_EH_{SOFT|HARD}RESET used to be used. ATA_EH_{SOFT|HARD}RESET is
now only used to tell prereset whether soft or hard reset will be
issued.
* Strip out now unneeded promote-to-hardreset logics from
ata_eh_reset(), ata_std_prereset(), sata_pmp_std_prereset() and
other places.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-01-23 15:05:14 +00:00
link - > eh_info . action | = ATA_EH_RESET ;
2007-10-25 09:30:36 +00:00
ata_ehi_push_desc ( & link - > eh_info , " waking up from sleep " ) ;
ata_link_abort ( link ) ;
return ;
}
2005-04-16 22:20:36 +00:00
ap - > ops - > qc_prep ( qc ) ;
2006-03-31 11:41:11 +00:00
qc - > err_mask | = ap - > ops - > qc_issue ( qc ) ;
if ( unlikely ( qc - > err_mask ) )
goto err ;
return ;
2005-04-16 22:20:36 +00:00
2010-08-23 09:27:27 +00:00
sys_err :
2006-03-31 11:41:11 +00:00
qc - > err_mask | = AC_ERR_SYSTEM ;
err :
ata_qc_complete ( qc ) ;
2005-04-16 22:20:36 +00:00
}
2006-05-15 11:57:46 +00:00
/**
* sata_scr_valid - test whether SCRs are accessible
2007-08-06 09:36:23 +00:00
* @ link : ATA link to test SCR accessibility for
2006-05-15 11:57:46 +00:00
*
2007-08-06 09:36:23 +00:00
* Test whether SCRs are accessible for @ link .
2006-05-15 11:57:46 +00:00
*
* LOCKING :
* None .
*
* RETURNS :
* 1 if SCRs are accessible , 0 otherwise .
*/
2007-08-06 09:36:23 +00:00
int sata_scr_valid ( struct ata_link * link )
2006-05-15 11:57:46 +00:00
{
2007-08-06 09:36:23 +00:00
struct ata_port * ap = link - > ap ;
2007-05-21 16:33:47 +00:00
return ( ap - > flags & ATA_FLAG_SATA ) & & ap - > ops - > scr_read ;
2006-05-15 11:57:46 +00:00
}
/**
* sata_scr_read - read SCR register of the specified port
2007-08-06 09:36:23 +00:00
* @ link : ATA link to read SCR for
2006-05-15 11:57:46 +00:00
* @ reg : SCR to read
* @ val : Place to store read value
*
2007-08-06 09:36:23 +00:00
* Read SCR register @ reg of @ link into * @ val . This function is
2007-09-23 04:19:54 +00:00
* guaranteed to succeed if @ link is ap - > link , the cable type of
* the port is SATA and the port implements - > scr_read .
2006-05-15 11:57:46 +00:00
*
* LOCKING :
2007-09-23 04:19:54 +00:00
* None if @ link is ap - > link . Kernel thread context otherwise .
2006-05-15 11:57:46 +00:00
*
* RETURNS :
* 0 on success , negative errno on failure .
*/
2007-08-06 09:36:23 +00:00
int sata_scr_read ( struct ata_link * link , int reg , u32 * val )
2006-05-15 11:57:46 +00:00
{
2007-09-23 04:19:54 +00:00
if ( ata_is_host_link ( link ) ) {
if ( sata_scr_valid ( link ) )
2008-07-31 08:02:40 +00:00
return link - > ap - > ops - > scr_read ( link , reg , val ) ;
2007-09-23 04:19:54 +00:00
return - EOPNOTSUPP ;
}
return sata_pmp_scr_read ( link , reg , val ) ;
2006-05-15 11:57:46 +00:00
}
/**
* sata_scr_write - write SCR register of the specified port
2007-08-06 09:36:23 +00:00
* @ link : ATA link to write SCR for
2006-05-15 11:57:46 +00:00
* @ reg : SCR to write
* @ val : value to write
*
2007-08-06 09:36:23 +00:00
* Write @ val to SCR register @ reg of @ link . This function is
2007-09-23 04:19:54 +00:00
* guaranteed to succeed if @ link is ap - > link , the cable type of
* the port is SATA and the port implements - > scr_read .
2006-05-15 11:57:46 +00:00
*
* LOCKING :
2007-09-23 04:19:54 +00:00
* None if @ link is ap - > link . Kernel thread context otherwise .
2006-05-15 11:57:46 +00:00
*
* RETURNS :
* 0 on success , negative errno on failure .
*/
2007-08-06 09:36:23 +00:00
int sata_scr_write ( struct ata_link * link , int reg , u32 val )
2006-05-15 11:57:46 +00:00
{
2007-09-23 04:19:54 +00:00
if ( ata_is_host_link ( link ) ) {
if ( sata_scr_valid ( link ) )
2008-07-31 08:02:40 +00:00
return link - > ap - > ops - > scr_write ( link , reg , val ) ;
2007-09-23 04:19:54 +00:00
return - EOPNOTSUPP ;
}
2007-08-06 09:36:23 +00:00
2007-09-23 04:19:54 +00:00
return sata_pmp_scr_write ( link , reg , val ) ;
2006-05-15 11:57:46 +00:00
}
/**
* sata_scr_write_flush - write SCR register of the specified port and flush
2007-08-06 09:36:23 +00:00
* @ link : ATA link to write SCR for
2006-05-15 11:57:46 +00:00
* @ reg : SCR to write
* @ val : value to write
*
* This function is identical to sata_scr_write ( ) except that this
* function performs flush after writing to the register .
*
* LOCKING :
2007-09-23 04:19:54 +00:00
* None if @ link is ap - > link . Kernel thread context otherwise .
2006-05-15 11:57:46 +00:00
*
* RETURNS :
* 0 on success , negative errno on failure .
*/
2007-08-06 09:36:23 +00:00
int sata_scr_write_flush ( struct ata_link * link , int reg , u32 val )
2006-05-15 11:57:46 +00:00
{
2007-09-23 04:19:54 +00:00
if ( ata_is_host_link ( link ) ) {
int rc ;
2007-07-16 05:29:40 +00:00
2007-09-23 04:19:54 +00:00
if ( sata_scr_valid ( link ) ) {
2008-07-31 08:02:40 +00:00
rc = link - > ap - > ops - > scr_write ( link , reg , val ) ;
2007-09-23 04:19:54 +00:00
if ( rc = = 0 )
2008-07-31 08:02:40 +00:00
rc = link - > ap - > ops - > scr_read ( link , reg , & val ) ;
2007-09-23 04:19:54 +00:00
return rc ;
}
return - EOPNOTSUPP ;
2006-05-15 11:57:46 +00:00
}
2007-09-23 04:19:54 +00:00
return sata_pmp_scr_write ( link , reg , val ) ;
2006-05-15 11:57:46 +00:00
}
/**
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
* ata_phys_link_online - test whether the given link is online
2007-08-06 09:36:23 +00:00
* @ link : ATA link to test
2006-05-15 11:57:46 +00:00
*
2007-08-06 09:36:23 +00:00
* Test whether @ link is online . Note that this function returns
* 0 if online status of @ link cannot be obtained , so
* ata_link_online ( link ) ! = ! ata_link_offline ( link ) .
2006-05-15 11:57:46 +00:00
*
* LOCKING :
* None .
*
* RETURNS :
2008-07-31 08:02:42 +00:00
* True if the port online status is available and online .
2006-05-15 11:57:46 +00:00
*/
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
bool ata_phys_link_online ( struct ata_link * link )
2006-05-15 11:57:46 +00:00
{
u32 sstatus ;
2007-08-06 09:36:23 +00:00
if ( sata_scr_read ( link , SCR_STATUS , & sstatus ) = = 0 & &
2009-01-29 11:31:31 +00:00
ata_sstatus_online ( sstatus ) )
2008-07-31 08:02:42 +00:00
return true ;
return false ;
2006-05-15 11:57:46 +00:00
}
/**
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
* ata_phys_link_offline - test whether the given link is offline
2007-08-06 09:36:23 +00:00
* @ link : ATA link to test
2006-05-15 11:57:46 +00:00
*
2007-08-06 09:36:23 +00:00
* Test whether @ link is offline . Note that this function
* returns 0 if offline status of @ link cannot be obtained , so
* ata_link_online ( link ) ! = ! ata_link_offline ( link ) .
2006-05-15 11:57:46 +00:00
*
* LOCKING :
* None .
*
* RETURNS :
2008-07-31 08:02:42 +00:00
* True if the port offline status is available and offline .
2006-05-15 11:57:46 +00:00
*/
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
bool ata_phys_link_offline ( struct ata_link * link )
2006-05-15 11:57:46 +00:00
{
u32 sstatus ;
2007-08-06 09:36:23 +00:00
if ( sata_scr_read ( link , SCR_STATUS , & sstatus ) = = 0 & &
2009-01-29 11:31:31 +00:00
! ata_sstatus_online ( sstatus ) )
2008-07-31 08:02:42 +00:00
return true ;
return false ;
2006-05-15 11:57:46 +00:00
}
2005-06-02 22:17:13 +00:00
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/**
* ata_link_online - test whether the given link is online
* @ link : ATA link to test
*
* Test whether @ link is online . This is identical to
* ata_phys_link_online ( ) when there ' s no slave link . When
* there ' s a slave link , this function should only be called on
* the master link and will return true if any of M / S links is
* online .
*
* LOCKING :
* None .
*
* RETURNS :
* True if the port online status is available and online .
*/
bool ata_link_online ( struct ata_link * link )
{
struct ata_link * slave = link - > ap - > slave_link ;
WARN_ON ( link = = slave ) ; /* shouldn't be called on slave link */
return ata_phys_link_online ( link ) | |
( slave & & ata_phys_link_online ( slave ) ) ;
}
/**
* ata_link_offline - test whether the given link is offline
* @ link : ATA link to test
*
* Test whether @ link is offline . This is identical to
* ata_phys_link_offline ( ) when there ' s no slave link . When
* there ' s a slave link , this function should only be called on
* the master link and will return true if both M / S links are
* offline .
*
* LOCKING :
* None .
*
* RETURNS :
* True if the port offline status is available and offline .
*/
bool ata_link_offline ( struct ata_link * link )
{
struct ata_link * slave = link - > ap - > slave_link ;
WARN_ON ( link = = slave ) ; /* shouldn't be called on slave link */
return ata_phys_link_offline ( link ) & &
( ! slave | | ata_phys_link_offline ( slave ) ) ;
}
2007-03-02 08:32:47 +00:00
# ifdef CONFIG_PM
2006-08-24 07:19:22 +00:00
static int ata_host_request_pm ( struct ata_host * host , pm_message_t mesg ,
unsigned int action , unsigned int ehi_flags ,
int wait )
2006-07-03 07:07:27 +00:00
{
unsigned long flags ;
int i , rc ;
2006-08-24 07:19:22 +00:00
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
2007-08-06 09:36:24 +00:00
struct ata_link * link ;
2006-07-03 07:07:27 +00:00
/* Previous resume operation might still be in
* progress . Wait for PM_PENDING to clear .
*/
if ( ap - > pflags & ATA_PFLAG_PM_PENDING ) {
ata_port_wait_eh ( ap ) ;
WARN_ON ( ap - > pflags & ATA_PFLAG_PM_PENDING ) ;
}
/* request PM ops to EH */
spin_lock_irqsave ( ap - > lock , flags ) ;
ap - > pm_mesg = mesg ;
if ( wait ) {
rc = 0 ;
ap - > pm_result = & rc ;
}
ap - > pflags | = ATA_PFLAG_PM_PENDING ;
2008-11-03 11:03:17 +00:00
ata_for_each_link ( link , ap , HOST_FIRST ) {
2007-08-06 09:36:24 +00:00
link - > eh_info . action | = action ;
link - > eh_info . flags | = ehi_flags ;
}
2006-07-03 07:07:27 +00:00
ata_port_schedule_eh ( ap ) ;
spin_unlock_irqrestore ( ap - > lock , flags ) ;
/* wait and check result */
if ( wait ) {
ata_port_wait_eh ( ap ) ;
WARN_ON ( ap - > pflags & ATA_PFLAG_PM_PENDING ) ;
if ( rc )
return rc ;
}
}
return 0 ;
}
/**
2006-08-24 07:19:22 +00:00
* ata_host_suspend - suspend host
* @ host : host to suspend
2006-07-03 07:07:27 +00:00
* @ mesg : PM message
*
2006-08-24 07:19:22 +00:00
* Suspend @ host . Actual operation is performed by EH . This
2006-07-03 07:07:27 +00:00
* function requests EH to perform PM operations and waits for EH
* to finish .
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*
* RETURNS :
* 0 on success , - errno on failure .
*/
2006-08-24 07:19:22 +00:00
int ata_host_suspend ( struct ata_host * host , pm_message_t mesg )
2006-07-03 07:07:27 +00:00
{
2010-09-07 12:05:31 +00:00
unsigned int ehi_flags = ATA_EHI_QUIET ;
2007-05-04 19:27:47 +00:00
int rc ;
2006-07-03 07:07:27 +00:00
2010-09-07 12:05:31 +00:00
/*
* On some hardware , device fails to respond after spun down
* for suspend . As the device won ' t be used before being
* resumed , we don ' t need to touch the device . Ask EH to skip
* the usual stuff and proceed directly to suspend .
*
* http : //thread.gmane.org/gmane.linux.ide/46764
*/
if ( mesg . event = = PM_EVENT_SUSPEND )
ehi_flags | = ATA_EHI_NO_AUTOPSY | ATA_EHI_NO_RECOVERY ;
rc = ata_host_request_pm ( host , mesg , 0 , ehi_flags , 1 ) ;
2008-02-25 22:31:10 +00:00
if ( rc = = 0 )
host - > dev - > power . power_state = mesg ;
2006-07-03 07:07:27 +00:00
return rc ;
}
/**
2006-08-24 07:19:22 +00:00
* ata_host_resume - resume host
* @ host : host to resume
2006-07-03 07:07:27 +00:00
*
2006-08-24 07:19:22 +00:00
* Resume @ host . Actual operation is performed by EH . This
2006-07-03 07:07:27 +00:00
* function requests EH to perform PM operations and returns .
2011-03-31 01:57:33 +00:00
* Note that all resume operations are performed parallelly .
2006-07-03 07:07:27 +00:00
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*/
2006-08-24 07:19:22 +00:00
void ata_host_resume ( struct ata_host * host )
2006-07-03 07:07:27 +00:00
{
libata: prefer hardreset
When both soft and hard resets are available, libata preferred
softreset till now. The logic behind it was to be softer to devices;
however, this doesn't really help much. Rationales for the change:
* BIOS may freeze lock certain things during boot and softreset can't
unlock those. This by itself is okay but during operation PHY event
or other error conditions can trigger hardreset and the device may
end up with different configuration.
For example, after a hardreset, previously unlockable HPA can be
unlocked resulting in different device size and thus revalidation
failure. Similar condition can occur during or after resume.
* Certain ATAPI devices require hardreset to recover after certain
error conditions. On PATA, this is done by issuing the DEVICE RESET
command. On SATA, COMRESET has equivalent effect. The problem is
that DEVICE RESET needs its own execution protocol.
For SFF controllers with bare TF access, it can be easily
implemented but more advanced controllers (e.g. ahci and sata_sil24)
require specialized implementations. Simply using hardreset solves
the problem nicely.
* COMRESET initialization sequence is the norm in SATA land and many
SATA devices don't work properly if only SRST is used. For example,
some PMPs behave this way and libata works around by always issuing
hardreset if the host supports PMP.
Like the above example, libata has developed a number of mechanisms
aiming to promote softreset to hardreset if softreset is not going
to work. This approach is time consuming and error prone.
Also, note that, dependingon how you read the specs, it could be
argued that PMP fan-out ports require COMRESET to start operation.
In fact, all the PMPs on the market except one don't work properly
if COMRESET is not issued to fan-out ports after PMP reset.
* COMRESET is an integral part of SATA connection and any working
device should be able to handle COMRESET properly. After all, it's
the way to signal hardreset during reboot. This is the most used
and recommended (at least by the ahci spec) method of resetting
devices.
So, this patch makes libata prefer hardreset over softreset by making
the following changes.
* Rename ATA_EH_RESET_MASK to ATA_EH_RESET and use it whereever
ATA_EH_{SOFT|HARD}RESET used to be used. ATA_EH_{SOFT|HARD}RESET is
now only used to tell prereset whether soft or hard reset will be
issued.
* Strip out now unneeded promote-to-hardreset logics from
ata_eh_reset(), ata_std_prereset(), sata_pmp_std_prereset() and
other places.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-01-23 15:05:14 +00:00
ata_host_request_pm ( host , PMSG_ON , ATA_EH_RESET ,
2006-08-24 07:19:22 +00:00
ATA_EHI_NO_AUTOPSY | ATA_EHI_QUIET , 0 ) ;
2008-02-25 22:31:10 +00:00
host - > dev - > power . power_state = PMSG_ON ;
2006-07-03 07:07:27 +00:00
}
2007-03-02 08:32:47 +00:00
# endif
2006-07-03 07:07:27 +00:00
2006-05-31 09:27:30 +00:00
/**
* ata_dev_init - Initialize an ata_device structure
* @ dev : Device structure to initialize
*
* Initialize @ dev in preparation for probing .
*
* LOCKING :
* Inherited from caller .
*/
void ata_dev_init ( struct ata_device * dev )
{
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
struct ata_link * link = ata_dev_phys_link ( dev ) ;
2007-08-06 09:36:22 +00:00
struct ata_port * ap = link - > ap ;
2006-05-31 09:27:32 +00:00
unsigned long flags ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/* SATA spd limit is bound to the attached device, reset together */
2007-08-06 09:36:22 +00:00
link - > sata_spd_limit = link - > hw_sata_spd_limit ;
link - > sata_spd = 0 ;
2006-05-31 09:27:38 +00:00
2006-05-31 09:27:32 +00:00
/* High bits of dev->flags are used to record warm plug
* requests which occur asynchronously . Synchronize using
2006-08-24 07:19:22 +00:00
* host lock .
2006-05-31 09:27:32 +00:00
*/
2006-06-23 03:46:10 +00:00
spin_lock_irqsave ( ap - > lock , flags ) ;
2006-05-31 09:27:32 +00:00
dev - > flags & = ~ ATA_DFLAG_INIT_MASK ;
2007-09-03 03:20:11 +00:00
dev - > horkage = 0 ;
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2006-05-31 09:27:30 +00:00
2009-01-29 11:31:32 +00:00
memset ( ( void * ) dev + ATA_DEVICE_CLEAR_BEGIN , 0 ,
ATA_DEVICE_CLEAR_END - ATA_DEVICE_CLEAR_BEGIN ) ;
2006-05-31 09:27:30 +00:00
dev - > pio_mask = UINT_MAX ;
dev - > mwdma_mask = UINT_MAX ;
dev - > udma_mask = UINT_MAX ;
}
2007-08-06 09:36:23 +00:00
/**
* ata_link_init - Initialize an ata_link structure
* @ ap : ATA port link is attached to
* @ link : Link structure to initialize
2007-08-06 09:36:23 +00:00
* @ pmp : Port multiplier port number
2007-08-06 09:36:23 +00:00
*
* Initialize @ link .
*
* LOCKING :
* Kernel thread context ( may sleep )
*/
2007-09-23 04:14:12 +00:00
void ata_link_init ( struct ata_port * ap , struct ata_link * link , int pmp )
2007-08-06 09:36:23 +00:00
{
int i ;
/* clear everything except for devices */
2010-05-25 19:31:38 +00:00
memset ( ( void * ) link + ATA_LINK_CLEAR_BEGIN , 0 ,
ATA_LINK_CLEAR_END - ATA_LINK_CLEAR_BEGIN ) ;
2007-08-06 09:36:23 +00:00
link - > ap = ap ;
2007-08-06 09:36:23 +00:00
link - > pmp = pmp ;
2007-08-06 09:36:23 +00:00
link - > active_tag = ATA_TAG_POISON ;
link - > hw_sata_spd_limit = UINT_MAX ;
/* can't use iterator, ap isn't initialized yet */
for ( i = 0 ; i < ATA_MAX_DEVICES ; i + + ) {
struct ata_device * dev = & link - > device [ i ] ;
dev - > link = link ;
dev - > devno = dev - link - > device ;
2009-09-15 19:17:28 +00:00
# ifdef CONFIG_ATA_ACPI
dev - > gtf_filter = ata_acpi_gtf_filter ;
# endif
2007-08-06 09:36:23 +00:00
ata_dev_init ( dev ) ;
}
}
/**
* sata_link_init_spd - Initialize link - > sata_spd_limit
* @ link : Link to configure sata_spd_limit for
*
* Initialize @ link - > [ hw_ ] sata_spd_limit to the currently
* configured value .
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*
* RETURNS :
* 0 on success , - errno on failure .
*/
2007-09-23 04:14:12 +00:00
int sata_link_init_spd ( struct ata_link * link )
2007-08-06 09:36:23 +00:00
{
2008-02-13 00:15:09 +00:00
u8 spd ;
2007-08-06 09:36:23 +00:00
int rc ;
2008-07-31 07:09:34 +00:00
rc = sata_scr_read ( link , SCR_CONTROL , & link - > saved_scontrol ) ;
2007-08-06 09:36:23 +00:00
if ( rc )
return rc ;
2008-07-31 07:09:34 +00:00
spd = ( link - > saved_scontrol > > 4 ) & 0xf ;
2007-08-06 09:36:23 +00:00
if ( spd )
link - > hw_sata_spd_limit & = ( 1 < < spd ) - 1 ;
2008-08-13 11:19:09 +00:00
ata_force_link_limits ( link ) ;
2008-02-13 00:15:09 +00:00
2007-08-06 09:36:23 +00:00
link - > sata_spd_limit = link - > hw_sata_spd_limit ;
return 0 ;
}
2005-04-16 22:20:36 +00:00
/**
2007-04-17 14:44:07 +00:00
* ata_port_alloc - allocate and initialize basic ATA port resources
* @ host : ATA host this allocated port belongs to
2005-04-16 22:20:36 +00:00
*
2007-04-17 14:44:07 +00:00
* Allocate and initialize basic ATA port resources .
*
* RETURNS :
* Allocate ATA port on success , NULL on failure .
2005-05-30 23:49:12 +00:00
*
2005-04-16 22:20:36 +00:00
* LOCKING :
2007-04-17 14:44:07 +00:00
* Inherited from calling layer ( may sleep ) .
2005-04-16 22:20:36 +00:00
*/
2007-04-17 14:44:07 +00:00
struct ata_port * ata_port_alloc ( struct ata_host * host )
2005-04-16 22:20:36 +00:00
{
2007-04-17 14:44:07 +00:00
struct ata_port * ap ;
2005-04-16 22:20:36 +00:00
2007-04-17 14:44:07 +00:00
DPRINTK ( " ENTER \n " ) ;
ap = kzalloc ( sizeof ( * ap ) , GFP_KERNEL ) ;
if ( ! ap )
return NULL ;
2011-02-15 06:13:24 +00:00
2011-03-16 13:58:32 +00:00
ap - > pflags | = ATA_PFLAG_INITIALIZING | ATA_PFLAG_FROZEN ;
2006-08-24 07:19:22 +00:00
ap - > lock = & host - > lock ;
2007-04-17 14:44:07 +00:00
ap - > print_id = - 1 ;
2006-08-24 07:19:22 +00:00
ap - > host = host ;
2007-04-17 14:44:07 +00:00
ap - > dev = host - > dev ;
2006-06-12 03:17:01 +00:00
# if defined(ATA_VERBOSE_DEBUG)
/* turn on all debugging levels */
ap - > msg_enable = 0x00FF ;
# elif defined(ATA_DEBUG)
ap - > msg_enable = ATA_MSG_DRV | ATA_MSG_INFO | ATA_MSG_CTL | ATA_MSG_WARN | ATA_MSG_ERR ;
2006-06-25 11:00:35 +00:00
# else
2006-06-23 06:29:08 +00:00
ap - > msg_enable = ATA_MSG_DRV | ATA_MSG_ERR | ATA_MSG_WARN ;
2006-06-12 03:17:01 +00:00
# endif
2005-04-16 22:20:36 +00:00
2010-07-02 08:03:52 +00:00
mutex_init ( & ap - > scsi_scan_mutex ) ;
2006-11-22 14:55:48 +00:00
INIT_DELAYED_WORK ( & ap - > hotplug_task , ata_scsi_hotplug ) ;
INIT_WORK ( & ap - > scsi_rescan_task , ata_scsi_dev_rescan ) ;
2006-01-23 04:09:37 +00:00
INIT_LIST_HEAD ( & ap - > eh_done_q ) ;
2006-05-31 09:27:27 +00:00
init_waitqueue_head ( & ap - > eh_wait_q ) ;
2008-09-21 09:54:08 +00:00
init_completion ( & ap - > park_req_pending ) ;
2007-07-16 05:29:41 +00:00
init_timer_deferrable ( & ap - > fastdrain_timer ) ;
ap - > fastdrain_timer . function = ata_eh_fastdrain_timerfn ;
ap - > fastdrain_timer . data = ( unsigned long ) ap ;
2005-04-16 22:20:36 +00:00
2006-05-15 11:57:44 +00:00
ap - > cbl = ATA_CBL_NONE ;
2007-08-06 09:36:23 +00:00
ata_link_init ( ap , & ap - > link , 0 ) ;
2005-04-16 22:20:36 +00:00
# ifdef ATA_IRQ_TRAP
ap - > stats . unhandled_irq = 1 ;
ap - > stats . idle_irq = 1 ;
# endif
2010-05-10 19:41:35 +00:00
ata_sff_port_init ( ap ) ;
2005-04-16 22:20:36 +00:00
return ap ;
}
2007-01-20 07:00:28 +00:00
static void ata_host_release ( struct device * gendev , void * res )
{
struct ata_host * host = dev_get_drvdata ( gendev ) ;
int i ;
2007-03-09 10:36:12 +00:00
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
2007-04-17 14:44:06 +00:00
if ( ! ap )
continue ;
if ( ap - > scsi_host )
2007-03-09 10:36:12 +00:00
scsi_host_put ( ap - > scsi_host ) ;
2007-09-23 04:19:54 +00:00
kfree ( ap - > pmp_link ) ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
kfree ( ap - > slave_link ) ;
2007-04-17 14:44:06 +00:00
kfree ( ap ) ;
2007-03-09 10:36:12 +00:00
host - > ports [ i ] = NULL ;
}
2007-02-27 13:33:21 +00:00
dev_set_drvdata ( gendev , NULL ) ;
2007-01-20 07:00:28 +00:00
}
2007-04-17 14:44:07 +00:00
/**
* ata_host_alloc - allocate and init basic ATA host resources
* @ dev : generic device this host is associated with
* @ max_ports : maximum number of ATA ports associated with this host
*
* Allocate and initialize basic ATA host resources . LLD calls
* this function to allocate a host , initializes it fully and
* attaches it using ata_host_register ( ) .
*
* @ max_ports ports are allocated and host - > n_ports is
* initialized to @ max_ports . The caller is allowed to decrease
* host - > n_ports before calling ata_host_register ( ) . The unused
* ports will be automatically freed on registration .
*
* RETURNS :
* Allocate ATA host on success , NULL on failure .
*
* LOCKING :
* Inherited from calling layer ( may sleep ) .
*/
struct ata_host * ata_host_alloc ( struct device * dev , int max_ports )
{
struct ata_host * host ;
size_t sz ;
int i ;
DPRINTK ( " ENTER \n " ) ;
if ( ! devres_open_group ( dev , NULL , GFP_KERNEL ) )
return NULL ;
/* alloc a container for our list of ATA ports (buses) */
sz = sizeof ( struct ata_host ) + ( max_ports + 1 ) * sizeof ( void * ) ;
/* alloc a container for our list of ATA ports (buses) */
host = devres_alloc ( ata_host_release , sz , GFP_KERNEL ) ;
if ( ! host )
goto err_out ;
devres_add ( dev , host ) ;
dev_set_drvdata ( dev , host ) ;
spin_lock_init ( & host - > lock ) ;
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
mutex_init ( & host - > eh_mutex ) ;
2007-04-17 14:44:07 +00:00
host - > dev = dev ;
host - > n_ports = max_ports ;
/* allocate ports bound to this host */
for ( i = 0 ; i < max_ports ; i + + ) {
struct ata_port * ap ;
ap = ata_port_alloc ( host ) ;
if ( ! ap )
goto err_out ;
ap - > port_no = i ;
host - > ports [ i ] = ap ;
}
devres_remove_group ( dev , NULL ) ;
return host ;
err_out :
devres_release_group ( dev , NULL ) ;
return NULL ;
}
2007-04-17 14:44:07 +00:00
/**
* ata_host_alloc_pinfo - alloc host and init with port_info array
* @ dev : generic device this host is associated with
* @ ppi : array of ATA port_info to initialize host with
* @ n_ports : number of ATA ports attached to this host
*
* Allocate ATA host and initialize with info from @ ppi . If NULL
* terminated , @ ppi may contain fewer entries than @ n_ports . The
* last entry will be used for the remaining ports .
*
* RETURNS :
* Allocate ATA host on success , NULL on failure .
*
* LOCKING :
* Inherited from calling layer ( may sleep ) .
*/
struct ata_host * ata_host_alloc_pinfo ( struct device * dev ,
const struct ata_port_info * const * ppi ,
int n_ports )
{
const struct ata_port_info * pi ;
struct ata_host * host ;
int i , j ;
host = ata_host_alloc ( dev , n_ports ) ;
if ( ! host )
return NULL ;
for ( i = 0 , j = 0 , pi = NULL ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
if ( ppi [ j ] )
pi = ppi [ j + + ] ;
ap - > pio_mask = pi - > pio_mask ;
ap - > mwdma_mask = pi - > mwdma_mask ;
ap - > udma_mask = pi - > udma_mask ;
ap - > flags | = pi - > flags ;
2007-08-06 09:36:23 +00:00
ap - > link . flags | = pi - > link_flags ;
2007-04-17 14:44:07 +00:00
ap - > ops = pi - > port_ops ;
if ( ! host - > ops & & ( pi - > port_ops ! = & ata_dummy_port_ops ) )
host - > ops = pi - > port_ops ;
}
return host ;
}
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
/**
* ata_slave_link_init - initialize slave link
* @ ap : port to initialize slave link for
*
* Create and initialize slave link for @ ap . This enables slave
* link handling on the port .
*
* In libata , a port contains links and a link contains devices .
* There is single host link but if a PMP is attached to it ,
* there can be multiple fan - out links . On SATA , there ' s usually
* a single device connected to a link but PATA and SATA
* controllers emulating TF based interface can have two - master
* and slave .
*
* However , there are a few controllers which don ' t fit into this
* abstraction too well - SATA controllers which emulate TF
* interface with both master and slave devices but also have
* separate SCR register sets for each device . These controllers
* need separate links for physical link handling
* ( e . g . onlineness , link speed ) but should be treated like a
* traditional M / S controller for everything else ( e . g . command
* issue , softreset ) .
*
* slave_link is libata ' s way of handling this class of
* controllers without impacting core layer too much . For
* anything other than physical link handling , the default host
* link is used for both master and slave . For physical link
* handling , separate @ ap - > slave_link is used . All dirty details
* are implemented inside libata core layer . From LLD ' s POV , the
* only difference is that prereset , hardreset and postreset are
* called once more for the slave link , so the reset sequence
* looks like the following .
*
* prereset ( M ) - > prereset ( S ) - > hardreset ( M ) - > hardreset ( S ) - >
* softreset ( M ) - > postreset ( M ) - > postreset ( S )
*
* Note that softreset is called only for the master . Softreset
* resets both M / S by definition , so SRST on master should handle
* both ( the standard method will work just fine ) .
*
* LOCKING :
* Should be called before host is registered .
*
* RETURNS :
* 0 on success , - errno on failure .
*/
int ata_slave_link_init ( struct ata_port * ap )
{
struct ata_link * link ;
WARN_ON ( ap - > slave_link ) ;
WARN_ON ( ap - > flags & ATA_FLAG_PMP ) ;
link = kzalloc ( sizeof ( * link ) , GFP_KERNEL ) ;
if ( ! link )
return - ENOMEM ;
ata_link_init ( ap , link , 1 ) ;
ap - > slave_link = link ;
return 0 ;
}
2007-11-08 04:09:00 +00:00
static void ata_host_stop ( struct device * gendev , void * res )
{
struct ata_host * host = dev_get_drvdata ( gendev ) ;
int i ;
WARN_ON ( ! ( host - > flags & ATA_HOST_STARTED ) ) ;
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
if ( ap - > ops - > port_stop )
ap - > ops - > port_stop ( ap ) ;
}
if ( host - > ops - > host_stop )
host - > ops - > host_stop ( host ) ;
}
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
/**
* ata_finalize_port_ops - finalize ata_port_operations
* @ ops : ata_port_operations to finalize
*
* An ata_port_operations can inherit from another ops and that
* ops can again inherit from another . This can go on as many
* times as necessary as long as there is no loop in the
* inheritance chain .
*
* Ops tables are finalized when the host is started . NULL or
* unspecified entries are inherited from the closet ancestor
* which has the method and the entry is populated with it .
* After finalization , the ops table directly points to all the
* methods and - > inherits is no longer necessary and cleared .
*
* Using ATA_OP_NULL , inheriting ops can force a method to NULL .
*
* LOCKING :
* None .
*/
static void ata_finalize_port_ops ( struct ata_port_operations * ops )
{
2008-05-29 17:58:14 +00:00
static DEFINE_SPINLOCK ( lock ) ;
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
const struct ata_port_operations * cur ;
void * * begin = ( void * * ) ops ;
void * * end = ( void * * ) & ops - > inherits ;
void * * pp ;
if ( ! ops | | ! ops - > inherits )
return ;
spin_lock ( & lock ) ;
for ( cur = ops - > inherits ; cur ; cur = cur - > inherits ) {
void * * inherit = ( void * * ) cur ;
for ( pp = begin ; pp < end ; pp + + , inherit + + )
if ( ! * pp )
* pp = * inherit ;
}
for ( pp = begin ; pp < end ; pp + + )
if ( IS_ERR ( * pp ) )
* pp = NULL ;
ops - > inherits = NULL ;
spin_unlock ( & lock ) ;
}
2007-04-17 14:44:06 +00:00
/**
* ata_host_start - start and freeze ports of an ATA host
* @ host : ATA host to start ports for
*
* Start and then freeze ports of @ host . Started status is
* recorded in host - > flags , so this function can be called
* multiple times . Ports are guaranteed to get started only
2007-04-17 14:44:07 +00:00
* once . If host - > ops isn ' t initialized yet , its set to the
* first non - dummy port ops .
2007-04-17 14:44:06 +00:00
*
* LOCKING :
* Inherited from calling layer ( may sleep ) .
*
* RETURNS :
* 0 if all ports are started successfully , - errno otherwise .
*/
int ata_host_start ( struct ata_host * host )
{
2007-11-08 04:09:00 +00:00
int have_stop = 0 ;
void * start_dr = NULL ;
2007-04-17 14:44:06 +00:00
int i , rc ;
if ( host - > flags & ATA_HOST_STARTED )
return 0 ;
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
ata_finalize_port_ops ( host - > ops ) ;
2007-04-17 14:44:06 +00:00
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
ata_finalize_port_ops ( ap - > ops ) ;
2007-04-17 14:44:07 +00:00
if ( ! host - > ops & & ! ata_port_is_dummy ( ap ) )
host - > ops = ap - > ops ;
2007-11-08 04:09:00 +00:00
if ( ap - > ops - > port_stop )
have_stop = 1 ;
}
if ( host - > ops - > host_stop )
have_stop = 1 ;
if ( have_stop ) {
start_dr = devres_alloc ( ata_host_stop , 0 , GFP_KERNEL ) ;
if ( ! start_dr )
return - ENOMEM ;
}
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
2007-04-17 14:44:06 +00:00
if ( ap - > ops - > port_start ) {
rc = ap - > ops - > port_start ( ap ) ;
if ( rc ) {
2007-11-30 15:23:16 +00:00
if ( rc ! = - ENODEV )
2008-01-10 22:33:09 +00:00
dev_printk ( KERN_ERR , host - > dev ,
" failed to start port %d "
" (errno=%d) \n " , i , rc ) ;
2007-04-17 14:44:06 +00:00
goto err_out ;
}
}
ata_eh_freeze_port ( ap ) ;
}
2007-11-08 04:09:00 +00:00
if ( start_dr )
devres_add ( host - > dev , start_dr ) ;
2007-04-17 14:44:06 +00:00
host - > flags | = ATA_HOST_STARTED ;
return 0 ;
err_out :
while ( - - i > = 0 ) {
struct ata_port * ap = host - > ports [ i ] ;
if ( ap - > ops - > port_stop )
ap - > ops - > port_stop ( ap ) ;
}
2007-11-08 04:09:00 +00:00
devres_free ( start_dr ) ;
2007-04-17 14:44:06 +00:00
return rc ;
}
2006-08-07 19:27:10 +00:00
/**
2006-08-24 07:19:22 +00:00
* ata_sas_host_init - Initialize a host struct
* @ host : host to initialize
* @ dev : device host is attached to
* @ flags : host flags
* @ ops : port_ops
2006-08-07 19:27:10 +00:00
*
* LOCKING :
* PCI / etc . bus probe sem .
*
*/
2007-04-17 14:44:07 +00:00
/* KILLME - the only user left is ipr */
2006-08-24 07:19:22 +00:00
void ata_host_init ( struct ata_host * host , struct device * dev ,
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
unsigned long flags , struct ata_port_operations * ops )
2006-08-07 19:27:10 +00:00
{
2006-08-24 07:19:22 +00:00
spin_lock_init ( & host - > lock ) ;
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
mutex_init ( & host - > eh_mutex ) ;
2006-08-24 07:19:22 +00:00
host - > dev = dev ;
host - > flags = flags ;
host - > ops = ops ;
2006-08-07 19:27:10 +00:00
}
2011-01-23 14:28:33 +00:00
int ata_port_probe ( struct ata_port * ap )
2009-01-04 13:32:28 +00:00
{
2011-01-23 14:28:33 +00:00
int rc = 0 ;
2009-01-09 23:54:07 +00:00
2009-01-04 13:32:28 +00:00
/* probe */
if ( ap - > ops - > error_handler ) {
struct ata_eh_info * ehi = & ap - > link . eh_info ;
unsigned long flags ;
/* kick EH for boot probing */
spin_lock_irqsave ( ap - > lock , flags ) ;
ehi - > probe_mask | = ATA_ALL_DEVICES ;
2010-09-01 15:50:06 +00:00
ehi - > action | = ATA_EH_RESET ;
2009-01-04 13:32:28 +00:00
ehi - > flags | = ATA_EHI_NO_AUTOPSY | ATA_EHI_QUIET ;
ap - > pflags & = ~ ATA_PFLAG_INITIALIZING ;
ap - > pflags | = ATA_PFLAG_LOADING ;
ata_port_schedule_eh ( ap ) ;
spin_unlock_irqrestore ( ap - > lock , flags ) ;
/* wait for EH to finish */
ata_port_wait_eh ( ap ) ;
} else {
DPRINTK ( " ata%u: bus probe begin \n " , ap - > print_id ) ;
rc = ata_bus_probe ( ap ) ;
DPRINTK ( " ata%u: bus probe end \n " , ap - > print_id ) ;
}
2011-01-23 14:28:33 +00:00
return rc ;
}
static void async_port_probe ( void * data , async_cookie_t cookie )
{
struct ata_port * ap = data ;
2011-02-15 06:13:24 +00:00
2011-01-23 14:28:33 +00:00
/*
* If we ' re not allowed to scan this host in parallel ,
* we need to wait until all previous scans have completed
* before going further .
* Jeff Garzik says this is only within a controller , so we
* don ' t need to wait for port 0 , only for later ports .
*/
if ( ! ( ap - > host - > flags & ATA_HOST_PARALLEL_SCAN ) & & ap - > port_no ! = 0 )
async_synchronize_cookie ( cookie ) ;
( void ) ata_port_probe ( ap ) ;
2009-01-05 23:07:07 +00:00
/* in order to keep device order, we need to synchronize at this point */
async_synchronize_cookie ( cookie ) ;
ata_scsi_scan_host ( ap , 1 ) ;
2009-01-04 13:32:28 +00:00
}
2011-01-23 14:28:33 +00:00
2007-04-17 14:44:07 +00:00
/**
* ata_host_register - register initialized ATA host
* @ host : ATA host to register
* @ sht : template for SCSI host
*
* Register initialized ATA host . @ host is allocated using
* ata_host_alloc ( ) and fully initialized by LLD . This function
* starts ports , registers @ host with ATA and SCSI layers and
* probe registered devices .
*
* LOCKING :
* Inherited from calling layer ( may sleep ) .
*
* RETURNS :
* 0 on success , - errno otherwise .
*/
int ata_host_register ( struct ata_host * host , struct scsi_host_template * sht )
{
int i , rc ;
/* host must have been started */
if ( ! ( host - > flags & ATA_HOST_STARTED ) ) {
dev_printk ( KERN_ERR , host - > dev ,
" BUG: trying to register unstarted host \n " ) ;
WARN_ON ( 1 ) ;
return - EINVAL ;
}
/* Blow away unused ports. This happens when LLD can't
* determine the exact number of ports to allocate at
* allocation time .
*/
for ( i = host - > n_ports ; host - > ports [ i ] ; i + + )
kfree ( host - > ports [ i ] ) ;
/* give ports names and add SCSI hosts */
for ( i = 0 ; i < host - > n_ports ; i + + )
host - > ports [ i ] - > print_id = ata_print_id + + ;
2011-02-15 06:13:24 +00:00
2010-05-25 19:31:38 +00:00
/* Create associated sysfs transport objects */
for ( i = 0 ; i < host - > n_ports ; i + + ) {
rc = ata_tport_add ( host - > dev , host - > ports [ i ] ) ;
if ( rc ) {
goto err_tadd ;
}
}
2007-04-17 14:44:07 +00:00
rc = ata_scsi_add_hosts ( host , sht ) ;
if ( rc )
2010-05-25 19:31:38 +00:00
goto err_tadd ;
2007-04-17 14:44:07 +00:00
2007-05-14 18:28:16 +00:00
/* associate with ACPI nodes */
ata_acpi_associate ( host ) ;
2007-04-17 14:44:07 +00:00
/* set cable, sata_spd_limit and report */
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
unsigned long xfer_mask ;
/* set SATA cable type if still unset */
if ( ap - > cbl = = ATA_CBL_NONE & & ( ap - > flags & ATA_FLAG_SATA ) )
ap - > cbl = ATA_CBL_SATA ;
/* init sata_spd_limit to the current value */
2007-08-06 09:36:23 +00:00
sata_link_init_spd ( & ap - > link ) ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
if ( ap - > slave_link )
sata_link_init_spd ( ap - > slave_link ) ;
2007-04-17 14:44:07 +00:00
2007-08-18 04:14:55 +00:00
/* print per-port info to dmesg */
2007-04-17 14:44:07 +00:00
xfer_mask = ata_pack_xfermask ( ap - > pio_mask , ap - > mwdma_mask ,
ap - > udma_mask ) ;
2007-10-09 05:57:25 +00:00
if ( ! ata_port_is_dummy ( ap ) ) {
2007-08-18 04:14:55 +00:00
ata_port_printk ( ap , KERN_INFO ,
" %cATA max %s %s \n " ,
2007-05-21 16:33:47 +00:00
( ap - > flags & ATA_FLAG_SATA ) ? ' S ' : ' P ' ,
2007-04-17 14:44:07 +00:00
ata_mode_string ( xfer_mask ) ,
2007-08-18 04:14:55 +00:00
ap - > link . eh_info . desc ) ;
2007-10-09 05:57:25 +00:00
ata_ehi_clear_desc ( & ap - > link . eh_info ) ;
} else
2007-04-17 14:44:07 +00:00
ata_port_printk ( ap , KERN_INFO , " DUMMY \n " ) ;
}
2009-04-08 16:19:39 +00:00
/* perform each probe asynchronously */
2007-04-17 14:44:07 +00:00
for ( i = 0 ; i < host - > n_ports ; i + + ) {
struct ata_port * ap = host - > ports [ i ] ;
2009-01-04 13:32:28 +00:00
async_schedule ( async_port_probe , ap ) ;
2007-04-17 14:44:07 +00:00
}
return 0 ;
2010-05-25 19:31:38 +00:00
err_tadd :
while ( - - i > = 0 ) {
ata_tport_delete ( host - > ports [ i ] ) ;
}
return rc ;
2007-04-17 14:44:07 +00:00
}
2007-04-17 14:44:07 +00:00
/**
* ata_host_activate - start host , request IRQ and register it
* @ host : target ATA host
* @ irq : IRQ to request
* @ irq_handler : irq_handler used when requesting IRQ
* @ irq_flags : irq_flags used when requesting IRQ
* @ sht : scsi_host_template to use when registering the host
*
* After allocating an ATA host and initializing it , most libata
* LLDs perform three steps to activate the host - start host ,
* request IRQ and register it . This helper takes necessasry
* arguments and performs the three steps in one go .
*
2007-11-08 02:14:56 +00:00
* An invalid IRQ skips the IRQ registration and expects the host to
* have set polling mode on the port . In this case , @ irq_handler
* should be NULL .
*
2007-04-17 14:44:07 +00:00
* LOCKING :
* Inherited from calling layer ( may sleep ) .
*
* RETURNS :
* 0 on success , - errno otherwise .
*/
int ata_host_activate ( struct ata_host * host , int irq ,
irq_handler_t irq_handler , unsigned long irq_flags ,
struct scsi_host_template * sht )
{
2007-08-18 04:14:55 +00:00
int i , rc ;
2007-04-17 14:44:07 +00:00
rc = ata_host_start ( host ) ;
if ( rc )
return rc ;
2007-11-08 02:14:56 +00:00
/* Special case for polling mode */
if ( ! irq ) {
WARN_ON ( irq_handler ) ;
return ata_host_register ( host , sht ) ;
}
2007-04-17 14:44:07 +00:00
rc = devm_request_irq ( host - > dev , irq , irq_handler , irq_flags ,
dev_driver_string ( host - > dev ) , host ) ;
if ( rc )
return rc ;
2007-08-18 04:14:55 +00:00
for ( i = 0 ; i < host - > n_ports ; i + + )
ata_port_desc ( host - > ports [ i ] , " irq %d " , irq ) ;
2007-07-02 16:38:47 +00:00
2007-04-17 14:44:07 +00:00
rc = ata_host_register ( host , sht ) ;
/* if failed, just free the IRQ and leave ports alone */
if ( rc )
devm_free_irq ( host - > dev , irq , host ) ;
return rc ;
}
2006-05-31 09:28:13 +00:00
/**
* ata_port_detach - Detach ATA port in prepration of device removal
* @ ap : ATA port to be detached
*
* Detach all ATA devices and the associated SCSI devices of @ ap ;
* then , remove the associated SCSI host . @ ap is guaranteed to
* be quiescent on return from this function .
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*/
2007-10-24 16:23:06 +00:00
static void ata_port_detach ( struct ata_port * ap )
2006-05-31 09:28:13 +00:00
{
unsigned long flags ;
if ( ! ap - > ops - > error_handler )
2006-08-04 18:59:11 +00:00
goto skip_eh ;
2006-05-31 09:28:13 +00:00
/* tell EH we're leaving & flush EH */
2006-06-23 03:46:10 +00:00
spin_lock_irqsave ( ap - > lock , flags ) ;
2006-06-28 16:29:30 +00:00
ap - > pflags | = ATA_PFLAG_UNLOADING ;
2008-11-03 11:04:37 +00:00
ata_port_schedule_eh ( ap ) ;
2006-06-23 03:46:10 +00:00
spin_unlock_irqrestore ( ap - > lock , flags ) ;
2006-05-31 09:28:13 +00:00
2008-11-03 11:04:37 +00:00
/* wait till EH commits suicide */
2006-05-31 09:28:13 +00:00
ata_port_wait_eh ( ap ) ;
2008-11-03 11:04:37 +00:00
/* it better be dead now */
WARN_ON ( ! ( ap - > pflags & ATA_PFLAG_UNLOADED ) ) ;
2006-05-31 09:28:13 +00:00
2010-12-14 15:21:17 +00:00
cancel_delayed_work_sync ( & ap - > hotplug_task ) ;
2006-05-31 09:28:13 +00:00
2006-08-04 18:59:11 +00:00
skip_eh :
2010-05-25 19:31:38 +00:00
if ( ap - > pmp_link ) {
int i ;
for ( i = 0 ; i < SATA_PMP_MAX_PORTS ; i + + )
ata_tlink_delete ( & ap - > pmp_link [ i ] ) ;
}
ata_tport_delete ( ap ) ;
2006-05-31 09:28:13 +00:00
/* remove the associated SCSI host */
2006-08-24 07:19:22 +00:00
scsi_remove_host ( ap - > scsi_host ) ;
2006-05-31 09:28:13 +00:00
}
2007-01-20 07:00:26 +00:00
/**
* ata_host_detach - Detach all ports of an ATA host
* @ host : Host to detach
*
* Detach all ports of @ host .
*
* LOCKING :
* Kernel thread context ( may sleep ) .
*/
void ata_host_detach ( struct ata_host * host )
{
int i ;
for ( i = 0 ; i < host - > n_ports ; i + + )
ata_port_detach ( host - > ports [ i ] ) ;
2007-12-15 06:05:01 +00:00
/* the host is dead now, dissociate ACPI */
ata_acpi_dissociate ( host ) ;
2007-01-20 07:00:26 +00:00
}
2005-08-30 09:42:52 +00:00
# ifdef CONFIG_PCI
2005-04-16 22:20:36 +00:00
/**
* ata_pci_remove_one - PCI layer callback for device removal
* @ pdev : PCI device that was removed
*
2007-01-20 07:00:28 +00:00
* PCI layer indicates to libata via this hook that hot - unplug or
* module unload event has occurred . Detach all ports . Resource
* release is handled via devres .
2005-04-16 22:20:36 +00:00
*
* LOCKING :
* Inherited from PCI layer ( may sleep ) .
*/
2007-01-20 07:00:28 +00:00
void ata_pci_remove_one ( struct pci_dev * pdev )
2005-04-16 22:20:36 +00:00
{
2007-10-11 21:12:35 +00:00
struct device * dev = & pdev - > dev ;
2006-08-24 07:19:22 +00:00
struct ata_host * host = dev_get_drvdata ( dev ) ;
2005-04-16 22:20:36 +00:00
2007-01-20 07:00:28 +00:00
ata_host_detach ( host ) ;
2005-04-16 22:20:36 +00:00
}
/* move to PCI subsystem */
2005-10-22 18:27:05 +00:00
int pci_test_config_bits ( struct pci_dev * pdev , const struct pci_bits * bits )
2005-04-16 22:20:36 +00:00
{
unsigned long tmp = 0 ;
switch ( bits - > width ) {
case 1 : {
u8 tmp8 = 0 ;
pci_read_config_byte ( pdev , bits - > reg , & tmp8 ) ;
tmp = tmp8 ;
break ;
}
case 2 : {
u16 tmp16 = 0 ;
pci_read_config_word ( pdev , bits - > reg , & tmp16 ) ;
tmp = tmp16 ;
break ;
}
case 4 : {
u32 tmp32 = 0 ;
pci_read_config_dword ( pdev , bits - > reg , & tmp32 ) ;
tmp = tmp32 ;
break ;
}
default :
return - EINVAL ;
}
tmp & = bits - > mask ;
return ( tmp = = bits - > val ) ? 1 : 0 ;
}
2006-01-06 08:28:07 +00:00
2007-03-02 08:32:47 +00:00
# ifdef CONFIG_PM
2006-07-26 07:58:33 +00:00
void ata_pci_device_do_suspend ( struct pci_dev * pdev , pm_message_t mesg )
2006-01-06 08:28:07 +00:00
{
pci_save_state ( pdev ) ;
2007-02-20 09:14:48 +00:00
pci_disable_device ( pdev ) ;
2006-07-03 07:07:27 +00:00
2008-02-23 18:13:25 +00:00
if ( mesg . event & PM_EVENT_SLEEP )
2006-07-03 07:07:27 +00:00
pci_set_power_state ( pdev , PCI_D3hot ) ;
2006-01-06 08:28:07 +00:00
}
2006-12-26 10:39:50 +00:00
int ata_pci_device_do_resume ( struct pci_dev * pdev )
2006-01-06 08:28:07 +00:00
{
2006-12-26 10:39:50 +00:00
int rc ;
2006-01-06 08:28:07 +00:00
pci_set_power_state ( pdev , PCI_D0 ) ;
pci_restore_state ( pdev ) ;
2006-12-26 10:39:50 +00:00
2007-01-20 07:00:28 +00:00
rc = pcim_enable_device ( pdev ) ;
2006-12-26 10:39:50 +00:00
if ( rc ) {
dev_printk ( KERN_ERR , & pdev - > dev ,
" failed to enable device after resume (%d) \n " , rc ) ;
return rc ;
}
2006-01-06 08:28:07 +00:00
pci_set_master ( pdev ) ;
2006-12-26 10:39:50 +00:00
return 0 ;
2006-07-03 07:07:27 +00:00
}
2006-07-26 07:58:33 +00:00
int ata_pci_device_suspend ( struct pci_dev * pdev , pm_message_t mesg )
2006-07-03 07:07:27 +00:00
{
2006-08-24 07:19:22 +00:00
struct ata_host * host = dev_get_drvdata ( & pdev - > dev ) ;
2006-07-03 07:07:27 +00:00
int rc = 0 ;
2006-08-24 07:19:22 +00:00
rc = ata_host_suspend ( host , mesg ) ;
2006-07-03 07:07:27 +00:00
if ( rc )
return rc ;
2006-07-26 07:58:33 +00:00
ata_pci_device_do_suspend ( pdev , mesg ) ;
2006-07-03 07:07:27 +00:00
return 0 ;
}
int ata_pci_device_resume ( struct pci_dev * pdev )
{
2006-08-24 07:19:22 +00:00
struct ata_host * host = dev_get_drvdata ( & pdev - > dev ) ;
2006-12-26 10:39:50 +00:00
int rc ;
2006-07-03 07:07:27 +00:00
2006-12-26 10:39:50 +00:00
rc = ata_pci_device_do_resume ( pdev ) ;
if ( rc = = 0 )
ata_host_resume ( host ) ;
return rc ;
2006-01-06 08:28:07 +00:00
}
2007-03-02 08:32:47 +00:00
# endif /* CONFIG_PM */
2005-04-16 22:20:36 +00:00
# endif /* CONFIG_PCI */
2008-02-13 00:15:09 +00:00
static int __init ata_parse_force_one ( char * * cur ,
struct ata_force_ent * force_ent ,
const char * * reason )
{
/* FIXME: Currently, there's no way to tag init const data and
* using __initdata causes build failure on some versions of
* gcc . Once __initdataconst is implemented , add const to the
* following structure .
*/
static struct ata_force_param force_tbl [ ] __initdata = {
{ " 40c " , . cbl = ATA_CBL_PATA40 } ,
{ " 80c " , . cbl = ATA_CBL_PATA80 } ,
{ " short40c " , . cbl = ATA_CBL_PATA40_SHORT } ,
{ " unk " , . cbl = ATA_CBL_PATA_UNK } ,
{ " ign " , . cbl = ATA_CBL_PATA_IGN } ,
{ " sata " , . cbl = ATA_CBL_SATA } ,
{ " 1.5Gbps " , . spd_limit = 1 } ,
{ " 3.0Gbps " , . spd_limit = 2 } ,
{ " noncq " , . horkage_on = ATA_HORKAGE_NONCQ } ,
{ " ncq " , . horkage_off = ATA_HORKAGE_NONCQ } ,
2010-05-23 10:59:11 +00:00
{ " dump_id " , . horkage_on = ATA_HORKAGE_DUMP_ID } ,
2008-02-13 00:15:09 +00:00
{ " pio0 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 0 ) } ,
{ " pio1 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 1 ) } ,
{ " pio2 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 2 ) } ,
{ " pio3 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 3 ) } ,
{ " pio4 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 4 ) } ,
{ " pio5 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 5 ) } ,
{ " pio6 " , . xfer_mask = 1 < < ( ATA_SHIFT_PIO + 6 ) } ,
{ " mwdma0 " , . xfer_mask = 1 < < ( ATA_SHIFT_MWDMA + 0 ) } ,
{ " mwdma1 " , . xfer_mask = 1 < < ( ATA_SHIFT_MWDMA + 1 ) } ,
{ " mwdma2 " , . xfer_mask = 1 < < ( ATA_SHIFT_MWDMA + 2 ) } ,
{ " mwdma3 " , . xfer_mask = 1 < < ( ATA_SHIFT_MWDMA + 3 ) } ,
{ " mwdma4 " , . xfer_mask = 1 < < ( ATA_SHIFT_MWDMA + 4 ) } ,
{ " udma0 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 0 ) } ,
{ " udma16 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 0 ) } ,
{ " udma/16 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 0 ) } ,
{ " udma1 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 1 ) } ,
{ " udma25 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 1 ) } ,
{ " udma/25 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 1 ) } ,
{ " udma2 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 2 ) } ,
{ " udma33 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 2 ) } ,
{ " udma/33 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 2 ) } ,
{ " udma3 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 3 ) } ,
{ " udma44 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 3 ) } ,
{ " udma/44 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 3 ) } ,
{ " udma4 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 4 ) } ,
{ " udma66 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 4 ) } ,
{ " udma/66 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 4 ) } ,
{ " udma5 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 5 ) } ,
{ " udma100 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 5 ) } ,
{ " udma/100 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 5 ) } ,
{ " udma6 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 6 ) } ,
{ " udma133 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 6 ) } ,
{ " udma/133 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 6 ) } ,
{ " udma7 " , . xfer_mask = 1 < < ( ATA_SHIFT_UDMA + 7 ) } ,
2008-08-13 11:19:09 +00:00
{ " nohrst " , . lflags = ATA_LFLAG_NO_HRST } ,
{ " nosrst " , . lflags = ATA_LFLAG_NO_SRST } ,
{ " norst " , . lflags = ATA_LFLAG_NO_HRST | ATA_LFLAG_NO_SRST } ,
2008-02-13 00:15:09 +00:00
} ;
char * start = * cur , * p = * cur ;
char * id , * val , * endp ;
const struct ata_force_param * match_fp = NULL ;
int nr_matches = 0 , i ;
/* find where this param ends and update *cur */
while ( * p ! = ' \0 ' & & * p ! = ' , ' )
p + + ;
if ( * p = = ' \0 ' )
* cur = p ;
else
* cur = p + 1 ;
* p = ' \0 ' ;
/* parse */
p = strchr ( start , ' : ' ) ;
if ( ! p ) {
val = strstrip ( start ) ;
goto parse_val ;
}
* p = ' \0 ' ;
id = strstrip ( start ) ;
val = strstrip ( p + 1 ) ;
/* parse id */
p = strchr ( id , ' . ' ) ;
if ( p ) {
* p + + = ' \0 ' ;
force_ent - > device = simple_strtoul ( p , & endp , 10 ) ;
if ( p = = endp | | * endp ! = ' \0 ' ) {
* reason = " invalid device " ;
return - EINVAL ;
}
}
force_ent - > port = simple_strtoul ( id , & endp , 10 ) ;
if ( p = = endp | | * endp ! = ' \0 ' ) {
* reason = " invalid port/link " ;
return - EINVAL ;
}
parse_val :
/* parse val, allow shortcuts so that both 1.5 and 1.5Gbps work */
for ( i = 0 ; i < ARRAY_SIZE ( force_tbl ) ; i + + ) {
const struct ata_force_param * fp = & force_tbl [ i ] ;
if ( strncasecmp ( val , fp - > name , strlen ( val ) ) )
continue ;
nr_matches + + ;
match_fp = fp ;
if ( strcasecmp ( val , fp - > name ) = = 0 ) {
nr_matches = 1 ;
break ;
}
}
if ( ! nr_matches ) {
* reason = " unknown value " ;
return - EINVAL ;
}
if ( nr_matches > 1 ) {
* reason = " ambigious value " ;
return - EINVAL ;
}
force_ent - > param = * match_fp ;
return 0 ;
}
static void __init ata_parse_force_param ( void )
{
int idx = 0 , size = 1 ;
int last_port = - 1 , last_device = - 1 ;
char * p , * cur , * next ;
/* calculate maximum number of params and allocate force_tbl */
for ( p = ata_force_param_buf ; * p ; p + + )
if ( * p = = ' , ' )
size + + ;
ata_force_tbl = kzalloc ( sizeof ( ata_force_tbl [ 0 ] ) * size , GFP_KERNEL ) ;
if ( ! ata_force_tbl ) {
printk ( KERN_WARNING " ata: failed to extend force table, "
" libata.force ignored \n " ) ;
return ;
}
/* parse and populate the table */
for ( cur = ata_force_param_buf ; * cur ! = ' \0 ' ; cur = next ) {
const char * reason = " " ;
struct ata_force_ent te = { . port = - 1 , . device = - 1 } ;
next = cur ;
if ( ata_parse_force_one ( & next , & te , & reason ) ) {
printk ( KERN_WARNING " ata: failed to parse force "
" parameter \" %s \" (%s) \n " ,
cur , reason ) ;
continue ;
}
if ( te . port = = - 1 ) {
te . port = last_port ;
te . device = last_device ;
}
ata_force_tbl [ idx + + ] = te ;
last_port = te . port ;
last_device = te . device ;
}
ata_force_tbl_size = idx ;
}
2005-04-16 22:20:36 +00:00
static int __init ata_init ( void )
{
2010-05-25 19:31:38 +00:00
int rc ;
2010-05-10 19:41:35 +00:00
2008-02-13 00:15:09 +00:00
ata_parse_force_param ( ) ;
2010-05-10 19:41:35 +00:00
rc = ata_sff_init ( ) ;
2010-07-02 08:03:52 +00:00
if ( rc ) {
kfree ( ata_force_tbl ) ;
return rc ;
}
2006-05-31 09:27:42 +00:00
2010-05-25 19:31:38 +00:00
libata_transport_init ( ) ;
ata_scsi_transport_template = ata_attach_transport ( ) ;
if ( ! ata_scsi_transport_template ) {
ata_sff_exit ( ) ;
rc = - ENOMEM ;
goto err_out ;
2011-02-15 06:13:24 +00:00
}
2010-05-25 19:31:38 +00:00
2005-04-16 22:20:36 +00:00
printk ( KERN_DEBUG " libata version " DRV_VERSION " loaded. \n " ) ;
return 0 ;
2010-05-25 19:31:38 +00:00
err_out :
return rc ;
2005-04-16 22:20:36 +00:00
}
static void __exit ata_exit ( void )
{
2010-05-25 19:31:38 +00:00
ata_release_transport ( ata_scsi_transport_template ) ;
libata_transport_exit ( ) ;
2010-05-10 19:41:35 +00:00
ata_sff_exit ( ) ;
2008-02-13 00:15:09 +00:00
kfree ( ata_force_tbl ) ;
2005-04-16 22:20:36 +00:00
}
2006-11-13 22:32:36 +00:00
subsys_initcall ( ata_init ) ;
2005-04-16 22:20:36 +00:00
module_exit ( ata_exit ) ;
2010-04-12 12:11:41 +00:00
static DEFINE_RATELIMIT_STATE ( ratelimit , HZ / 5 , 1 ) ;
2005-10-05 06:58:32 +00:00
int ata_ratelimit ( void )
{
2010-04-12 12:11:41 +00:00
return __ratelimit ( & ratelimit ) ;
2005-10-05 06:58:32 +00:00
}
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
/**
* ata_msleep - ATA EH owner aware msleep
* @ ap : ATA port to attribute the sleep to
* @ msecs : duration to sleep in milliseconds
*
* Sleeps @ msecs . If the current task is owner of @ ap ' s EH , the
* ownership is released before going to sleep and reacquired
* after the sleep is complete . IOW , other ports sharing the
* @ ap - > host will be allowed to own the EH while this task is
* sleeping .
*
* LOCKING :
* Might sleep .
*/
2010-09-06 15:56:29 +00:00
void ata_msleep ( struct ata_port * ap , unsigned int msecs )
{
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
bool owns_eh = ap & & ap - > host - > eh_owner = = current ;
if ( owns_eh )
ata_eh_release ( ap ) ;
2010-09-06 15:56:29 +00:00
msleep ( msecs ) ;
libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release
ap->lock explicitly when accessing hardware or shared data structures.
However, once EH is active, it's assumed that the port is owned by EH
and EH methods don't explicitly take ap->lock unless race from irq
handler or other code paths are expected. However, libata EH didn't
guarantee exclusion among EHs for ports of the same host. IOW,
multiple EHs may execute in parallel on multiple ports of the same
controller.
In many cases, especially in SATA, the ports are completely
independent of each other and this doesn't cause problems; however,
there are cases where different ports share the same resource, which
lead to obscure timing related bugs such as the one fixed by commit
213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH
begins, it acquires per-host EH ownership by calling ata_eh_acquire().
When EH finishes, the ownership is released by calling
ata_eh_release(). EH ownership is also released whenever the EH
thread goes to sleep from ata_msleep() or explicitly and reacquired
after waking up.
This ensures that while EH is actively accessing the hardware, it has
exclusive access to it while allowing EHs to interleave and progress
in parallel as they hit waiting stages, which dominate the time spent
in EH. This achieves cross-port EH exclusion without pervasive and
fragile changes while still allowing parallel EH for the most part.
This was first reported by yuanding02@gmail.com more than three years
ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reported-by: yuanding02@gmail.com
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2010-09-06 15:57:14 +00:00
if ( owns_eh )
ata_eh_acquire ( ap ) ;
2010-09-06 15:56:29 +00:00
}
2006-04-11 13:22:29 +00:00
/**
* ata_wait_register - wait until register value changes
2010-09-06 15:56:29 +00:00
* @ ap : ATA port to wait register for , can be NULL
2006-04-11 13:22:29 +00:00
* @ reg : IO - mapped register
* @ mask : Mask to apply to read register value
* @ val : Wait condition
2008-05-19 17:17:51 +00:00
* @ interval : polling interval in milliseconds
* @ timeout : timeout in milliseconds
2006-04-11 13:22:29 +00:00
*
* Waiting for some bits of register to change is a common
* operation for ATA controllers . This function reads 32 bit LE
* IO - mapped register @ reg and tests for the following condition .
*
* ( * @ reg & mask ) ! = val
*
* If the condition is met , it returns ; otherwise , the process is
* repeated after @ interval_msec until timeout .
*
* LOCKING :
* Kernel thread context ( may sleep )
*
* RETURNS :
* The final register value .
*/
2010-09-06 15:56:29 +00:00
u32 ata_wait_register ( struct ata_port * ap , void __iomem * reg , u32 mask , u32 val ,
2008-05-19 17:17:51 +00:00
unsigned long interval , unsigned long timeout )
2006-04-11 13:22:29 +00:00
{
2008-05-19 17:17:51 +00:00
unsigned long deadline ;
2006-04-11 13:22:29 +00:00
u32 tmp ;
tmp = ioread32 ( reg ) ;
/* Calculate timeout _after_ the first read to make sure
* preceding writes reach the controller before starting to
* eat away the timeout .
*/
2008-05-19 17:17:51 +00:00
deadline = ata_deadline ( jiffies , timeout ) ;
2006-04-11 13:22:29 +00:00
2008-05-19 17:17:51 +00:00
while ( ( tmp & mask ) = = val & & time_before ( jiffies , deadline ) ) {
2010-09-06 15:56:29 +00:00
ata_msleep ( ap , interval ) ;
2006-04-11 13:22:29 +00:00
tmp = ioread32 ( reg ) ;
}
return tmp ;
}
2006-08-10 07:59:12 +00:00
/*
* Dummy port_ops
*/
2008-04-07 13:47:21 +00:00
static unsigned int ata_dummy_qc_issue ( struct ata_queued_cmd * qc )
2006-08-10 07:59:12 +00:00
{
2008-04-07 13:47:21 +00:00
return AC_ERR_SYSTEM ;
2006-08-10 07:59:12 +00:00
}
2008-04-07 13:47:21 +00:00
static void ata_dummy_error_handler ( struct ata_port * ap )
2006-08-10 07:59:12 +00:00
{
2008-04-07 13:47:21 +00:00
/* truly dummy */
2006-08-10 07:59:12 +00:00
}
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
struct ata_port_operations ata_dummy_port_ops = {
2006-08-10 07:59:12 +00:00
. qc_prep = ata_noop_qc_prep ,
. qc_issue = ata_dummy_qc_issue ,
2008-04-07 13:47:21 +00:00
. error_handler = ata_dummy_error_handler ,
2006-08-10 07:59:12 +00:00
} ;
2007-04-17 14:44:07 +00:00
const struct ata_port_info ata_dummy_port_info = {
. port_ops = & ata_dummy_port_ops ,
} ;
2005-04-16 22:20:36 +00:00
/*
* libata is essentially a library of internal helper functions for
* low - level ATA host controller drivers . As such , the API / ABI is
* likely to change as new drivers are added and updated .
* Do not depend on ABI / API stability .
*/
2006-07-03 07:07:26 +00:00
EXPORT_SYMBOL_GPL ( sata_deb_timing_normal ) ;
EXPORT_SYMBOL_GPL ( sata_deb_timing_hotplug ) ;
EXPORT_SYMBOL_GPL ( sata_deb_timing_long ) ;
libata: implement and use ops inheritance
libata lets low level drivers build ata_port_operations table and
register it with libata core layer. This allows low level drivers
high level of flexibility but also burdens them with lots of
boilerplate entries.
This becomes worse for drivers which support related similar
controllers which differ slightly. They share most of the operations
except for a few. However, the driver still needs to list all
operations for each variant. This results in large number of
duplicate entries, which is not only inefficient but also error-prone
as it becomes very difficult to tell what the actual differences are.
This duplicate boilerplates all over the low level drivers also make
updating the core layer exteremely difficult and error-prone. When
compounded with multi-branched development model, it ends up
accumulating inconsistencies over time. Some of those inconsistencies
cause immediate problems and fixed. Others just remain there dormant
making maintenance increasingly difficult.
To rectify the problem, this patch implements ata_port_operations
inheritance. To allow LLDs to easily re-use their own ops tables
overriding only specific methods, this patch implements poor man's
class inheritance. An ops table has ->inherits field which can be set
to any ops table as long as it doesn't create a loop. When the host
is started, the inheritance chain is followed and any operation which
isn't specified is taken from the nearest ancestor which has it
specified. This operation is called finalization and done only once
per an ops table and the LLD doesn't have to do anything special about
it other than making the ops table non-const such that libata can
update it.
libata provides four base ops tables lower drivers can inherit from -
base, sata, pmp, sff and bmdma. To avoid overriding these ops
accidentaly, these ops are declared const and LLDs should always
inherit these instead of using them directly.
After finalization, all the ops table are identical before and after
the patch except for setting .irq_handler to ata_interrupt in drivers
which didn't use to. The .irq_handler doesn't have any actual effect
and the field will soon be removed by later patch.
* sata_sx4 is still using old style EH and currently doesn't take
advantage of ops inheritance.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:49 +00:00
EXPORT_SYMBOL_GPL ( ata_base_port_ops ) ;
EXPORT_SYMBOL_GPL ( sata_port_ops ) ;
2006-08-10 07:59:12 +00:00
EXPORT_SYMBOL_GPL ( ata_dummy_port_ops ) ;
2007-04-17 14:44:07 +00:00
EXPORT_SYMBOL_GPL ( ata_dummy_port_info ) ;
2008-11-03 11:03:17 +00:00
EXPORT_SYMBOL_GPL ( ata_link_next ) ;
EXPORT_SYMBOL_GPL ( ata_dev_next ) ;
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL_GPL ( ata_std_bios_param ) ;
2010-05-15 18:09:34 +00:00
EXPORT_SYMBOL_GPL ( ata_scsi_unlock_native_capacity ) ;
2006-08-24 07:19:22 +00:00
EXPORT_SYMBOL_GPL ( ata_host_init ) ;
2007-04-17 14:44:07 +00:00
EXPORT_SYMBOL_GPL ( ata_host_alloc ) ;
2007-04-17 14:44:07 +00:00
EXPORT_SYMBOL_GPL ( ata_host_alloc_pinfo ) ;
libata: implement slave_link
Explanation taken from the comment of ata_slave_link_init().
In libata, a port contains links and a link contains devices. There
is single host link but if a PMP is attached to it, there can be
multiple fan-out links. On SATA, there's usually a single device
connected to a link but PATA and SATA controllers emulating TF based
interface can have two - master and slave.
However, there are a few controllers which don't fit into this
abstraction too well - SATA controllers which emulate TF interface
with both master and slave devices but also have separate SCR
register sets for each device. These controllers need separate links
for physical link handling (e.g. onlineness, link speed) but should
be treated like a traditional M/S controller for everything else
(e.g. command issue, softreset).
slave_link is libata's way of handling this class of controllers
without impacting core layer too much. For anything other than
physical link handling, the default host link is used for both master
and slave. For physical link handling, separate @ap->slave_link is
used. All dirty details are implemented inside libata core layer.
From LLD's POV, the only difference is that prereset, hardreset and
postreset are called once more for the slave link, so the reset
sequence looks like the following.
prereset(M) -> prereset(S) -> hardreset(M) -> hardreset(S) ->
softreset(M) -> postreset(M) -> postreset(S)
Note that softreset is called only for the master. Softreset resets
both M/S by definition, so SRST on master should handle both (the
standard method will work just fine).
As slave_link excludes PMP support and only code paths which deal with
the attributes of physical link are affected, all the changes are
localized to libata.h, libata-core.c and libata-eh.c.
* ata_is_host_link() updated so that slave_link is considered as host
link too.
* iterator extended to iterate over the slave_link when using the
underbarred version.
* force param handling updated such that devno 16 is mapped to the
slave link/device.
* ata_link_on/offline() updated to return the combined result from
master and slave link. ata_phys_link_on/offline() are the direct
versions.
* EH autopsy and report are performed separately for master slave
links. Reset is udpated to implement the above described reset
sequence.
Except for reset update, most changes are minor, many of them just
modifying dev->link to ata_dev_phys_link(dev) or using phys online
test instead.
After this update, LLDs can take full advantage of per-dev SCR
registers by simply turning on slave link.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
2008-07-31 08:02:43 +00:00
EXPORT_SYMBOL_GPL ( ata_slave_link_init ) ;
2007-04-17 14:44:06 +00:00
EXPORT_SYMBOL_GPL ( ata_host_start ) ;
2007-04-17 14:44:07 +00:00
EXPORT_SYMBOL_GPL ( ata_host_register ) ;
2007-04-17 14:44:07 +00:00
EXPORT_SYMBOL_GPL ( ata_host_activate ) ;
2007-01-20 07:00:26 +00:00
EXPORT_SYMBOL_GPL ( ata_host_detach ) ;
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL_GPL ( ata_sg_init ) ;
2006-05-15 11:58:05 +00:00
EXPORT_SYMBOL_GPL ( ata_qc_complete ) ;
2006-05-15 12:03:43 +00:00
EXPORT_SYMBOL_GPL ( ata_qc_complete_multiple ) ;
2008-04-02 08:28:46 +00:00
EXPORT_SYMBOL_GPL ( atapi_cmd_type ) ;
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL_GPL ( ata_tf_to_fis ) ;
EXPORT_SYMBOL_GPL ( ata_tf_from_fis ) ;
2007-11-27 10:43:39 +00:00
EXPORT_SYMBOL_GPL ( ata_pack_xfermask ) ;
EXPORT_SYMBOL_GPL ( ata_unpack_xfermask ) ;
EXPORT_SYMBOL_GPL ( ata_xfer_mask2mode ) ;
EXPORT_SYMBOL_GPL ( ata_xfer_mode2mask ) ;
EXPORT_SYMBOL_GPL ( ata_xfer_mode2shift ) ;
EXPORT_SYMBOL_GPL ( ata_mode_string ) ;
EXPORT_SYMBOL_GPL ( ata_id_xfermask ) ;
2007-03-06 10:37:52 +00:00
EXPORT_SYMBOL_GPL ( ata_do_set_mode ) ;
2007-09-23 04:14:12 +00:00
EXPORT_SYMBOL_GPL ( ata_std_qc_defer ) ;
2006-03-17 23:04:03 +00:00
EXPORT_SYMBOL_GPL ( ata_noop_qc_prep ) ;
2007-02-20 18:01:59 +00:00
EXPORT_SYMBOL_GPL ( ata_dev_disable ) ;
2006-05-15 11:57:23 +00:00
EXPORT_SYMBOL_GPL ( sata_set_spd ) ;
2008-04-07 13:47:19 +00:00
EXPORT_SYMBOL_GPL ( ata_wait_after_reset ) ;
2007-08-06 09:36:23 +00:00
EXPORT_SYMBOL_GPL ( sata_link_debounce ) ;
EXPORT_SYMBOL_GPL ( sata_link_resume ) ;
2010-09-01 15:50:05 +00:00
EXPORT_SYMBOL_GPL ( sata_link_scr_lpm ) ;
2008-04-07 13:47:18 +00:00
EXPORT_SYMBOL_GPL ( ata_std_prereset ) ;
2007-08-06 09:36:23 +00:00
EXPORT_SYMBOL_GPL ( sata_link_hardreset ) ;
2008-04-07 13:47:19 +00:00
EXPORT_SYMBOL_GPL ( sata_std_hardreset ) ;
2008-04-07 13:47:18 +00:00
EXPORT_SYMBOL_GPL ( ata_std_postreset ) ;
2006-03-24 14:56:57 +00:00
EXPORT_SYMBOL_GPL ( ata_dev_classify ) ;
EXPORT_SYMBOL_GPL ( ata_dev_pair ) ;
2005-10-05 06:58:32 +00:00
EXPORT_SYMBOL_GPL ( ata_ratelimit ) ;
2010-09-06 15:56:29 +00:00
EXPORT_SYMBOL_GPL ( ata_msleep ) ;
2006-04-11 13:22:29 +00:00
EXPORT_SYMBOL_GPL ( ata_wait_register ) ;
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL_GPL ( ata_scsi_queuecmd ) ;
EXPORT_SYMBOL_GPL ( ata_scsi_slave_config ) ;
2006-05-31 09:28:07 +00:00
EXPORT_SYMBOL_GPL ( ata_scsi_slave_destroy ) ;
2006-05-15 12:03:48 +00:00
EXPORT_SYMBOL_GPL ( ata_scsi_change_queue_depth ) ;
2006-05-15 11:57:46 +00:00
EXPORT_SYMBOL_GPL ( sata_scr_valid ) ;
EXPORT_SYMBOL_GPL ( sata_scr_read ) ;
EXPORT_SYMBOL_GPL ( sata_scr_write ) ;
EXPORT_SYMBOL_GPL ( sata_scr_write_flush ) ;
2007-08-06 09:36:23 +00:00
EXPORT_SYMBOL_GPL ( ata_link_online ) ;
EXPORT_SYMBOL_GPL ( ata_link_offline ) ;
2007-03-02 08:32:47 +00:00
# ifdef CONFIG_PM
2006-08-24 07:19:22 +00:00
EXPORT_SYMBOL_GPL ( ata_host_suspend ) ;
EXPORT_SYMBOL_GPL ( ata_host_resume ) ;
2007-03-02 08:32:47 +00:00
# endif /* CONFIG_PM */
2006-02-13 01:02:46 +00:00
EXPORT_SYMBOL_GPL ( ata_id_string ) ;
EXPORT_SYMBOL_GPL ( ata_id_c_string ) ;
2008-07-24 16:16:06 +00:00
EXPORT_SYMBOL_GPL ( ata_do_dev_read_id ) ;
2005-04-16 22:20:36 +00:00
EXPORT_SYMBOL_GPL ( ata_scsi_simulate ) ;
2006-01-09 17:18:14 +00:00
EXPORT_SYMBOL_GPL ( ata_pio_need_iordy ) ;
2007-11-27 10:43:39 +00:00
EXPORT_SYMBOL_GPL ( ata_timing_find_mode ) ;
2005-10-21 23:01:32 +00:00
EXPORT_SYMBOL_GPL ( ata_timing_compute ) ;
EXPORT_SYMBOL_GPL ( ata_timing_merge ) ;
2007-12-18 07:33:05 +00:00
EXPORT_SYMBOL_GPL ( ata_timing_cycle2mode ) ;
2005-10-21 23:01:32 +00:00
2005-04-16 22:20:36 +00:00
# ifdef CONFIG_PCI
EXPORT_SYMBOL_GPL ( pci_test_config_bits ) ;
EXPORT_SYMBOL_GPL ( ata_pci_remove_one ) ;
2007-03-02 08:32:47 +00:00
# ifdef CONFIG_PM
2006-07-03 07:07:27 +00:00
EXPORT_SYMBOL_GPL ( ata_pci_device_do_suspend ) ;
EXPORT_SYMBOL_GPL ( ata_pci_device_do_resume ) ;
2006-01-06 08:28:07 +00:00
EXPORT_SYMBOL_GPL ( ata_pci_device_suspend ) ;
EXPORT_SYMBOL_GPL ( ata_pci_device_resume ) ;
2007-03-02 08:32:47 +00:00
# endif /* CONFIG_PM */
2005-04-16 22:20:36 +00:00
# endif /* CONFIG_PCI */
2006-01-06 08:28:07 +00:00
2007-07-16 05:29:39 +00:00
EXPORT_SYMBOL_GPL ( __ata_ehi_push_desc ) ;
EXPORT_SYMBOL_GPL ( ata_ehi_push_desc ) ;
EXPORT_SYMBOL_GPL ( ata_ehi_clear_desc ) ;
2007-08-18 04:14:55 +00:00
EXPORT_SYMBOL_GPL ( ata_port_desc ) ;
# ifdef CONFIG_PCI
EXPORT_SYMBOL_GPL ( ata_port_pbar_desc ) ;
# endif /* CONFIG_PCI */
2006-05-15 11:58:07 +00:00
EXPORT_SYMBOL_GPL ( ata_port_schedule_eh ) ;
2007-08-06 09:36:23 +00:00
EXPORT_SYMBOL_GPL ( ata_link_abort ) ;
2006-05-15 11:58:07 +00:00
EXPORT_SYMBOL_GPL ( ata_port_abort ) ;
2006-05-15 11:58:09 +00:00
EXPORT_SYMBOL_GPL ( ata_port_freeze ) ;
2007-09-23 04:14:13 +00:00
EXPORT_SYMBOL_GPL ( sata_async_notification ) ;
2006-05-15 11:58:09 +00:00
EXPORT_SYMBOL_GPL ( ata_eh_freeze_port ) ;
EXPORT_SYMBOL_GPL ( ata_eh_thaw_port ) ;
2006-04-02 09:51:53 +00:00
EXPORT_SYMBOL_GPL ( ata_eh_qc_complete ) ;
EXPORT_SYMBOL_GPL ( ata_eh_qc_retry ) ;
2008-05-02 06:14:53 +00:00
EXPORT_SYMBOL_GPL ( ata_eh_analyze_ncq_error ) ;
2006-05-15 11:58:22 +00:00
EXPORT_SYMBOL_GPL ( ata_do_eh ) ;
libata: make reset related methods proper port operations
Currently reset methods are not specified directly in the
ata_port_operations table. If a LLD wants to use custom reset
methods, it should construct and use a error_handler which uses those
reset methods. It's done this way for two reasons.
First, the ops table already contained too many methods and adding
four more of them would noticeably increase the amount of necessary
boilerplate code all over low level drivers.
Second, as ->error_handler uses those reset methods, it can get
confusing. ie. By overriding ->error_handler, those reset ops can be
made useless making layering a bit hazy.
Now that ops table uses inheritance, the first problem doesn't exist
anymore. The second isn't completely solved but is relieved by
providing default values - most drivers can just override what it has
implemented and don't have to concern itself about higher level
callbacks. In fact, there currently is no driver which actually
modifies error handling behavior. Drivers which override
->error_handler just wraps the standard error handler only to prepare
the controller for EH. I don't think making ops layering strict has
any noticeable benefit.
This patch makes ->prereset, ->softreset, ->hardreset, ->postreset and
their PMP counterparts propoer ops. Default ops are provided in the
base ops tables and drivers are converted to override individual reset
methods instead of creating custom error_handler.
* ata_std_error_handler() doesn't use sata_std_hardreset() if SCRs
aren't accessible. sata_promise doesn't need to use separate
error_handlers for PATA and SATA anymore.
* softreset is broken for sata_inic162x and sata_sx4. As libata now
always prefers hardreset, this doesn't really matter but the ops are
forced to NULL using ATA_OP_NULL for documentation purpose.
* pata_hpt374 needs to use different prereset for the first and second
PCI functions. This used to be done by branching from
hpt374_error_handler(). The proper way to do this is to use
separate ops and port_info tables for each function. Converted.
Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 03:22:50 +00:00
EXPORT_SYMBOL_GPL ( ata_std_error_handler ) ;
2007-03-06 10:37:56 +00:00
EXPORT_SYMBOL_GPL ( ata_cable_40wire ) ;
EXPORT_SYMBOL_GPL ( ata_cable_80wire ) ;
EXPORT_SYMBOL_GPL ( ata_cable_unknown ) ;
2007-11-27 10:43:48 +00:00
EXPORT_SYMBOL_GPL ( ata_cable_ignore ) ;
2007-03-06 10:37:56 +00:00
EXPORT_SYMBOL_GPL ( ata_cable_sata ) ;
