2010-12-14 00:34:40 +00:00
|
|
|
/*
|
|
|
|
* QEMU AHCI Emulation
|
|
|
|
*
|
|
|
|
* Copyright (c) 2010 qiaochong@loongson.cn
|
|
|
|
* Copyright (c) 2010 Roland Elek <elek.roland@gmail.com>
|
|
|
|
* Copyright (c) 2010 Sebastian Herbszt <herbszt@gmx.de>
|
|
|
|
* Copyright (c) 2010 Alexander Graf <agraf@suse.de>
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library 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
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with this library; if not, see <http://www.gnu.org/licenses/>.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <hw/hw.h>
|
2012-12-12 12:24:50 +00:00
|
|
|
#include <hw/pci/msi.h>
|
2013-02-05 16:06:20 +00:00
|
|
|
#include <hw/i386/pc.h>
|
2012-12-12 12:24:50 +00:00
|
|
|
#include <hw/pci/pci.h>
|
2012-01-26 11:43:47 +00:00
|
|
|
#include <hw/sysbus.h>
|
2010-12-14 00:34:40 +00:00
|
|
|
|
2012-12-17 17:19:49 +00:00
|
|
|
#include "monitor/monitor.h"
|
2014-10-07 11:59:18 +00:00
|
|
|
#include "sysemu/block-backend.h"
|
2012-12-17 17:20:04 +00:00
|
|
|
#include "sysemu/dma.h"
|
2010-12-14 00:34:40 +00:00
|
|
|
#include "internal.h"
|
|
|
|
#include <hw/ide/pci.h>
|
2011-02-01 14:51:26 +00:00
|
|
|
#include <hw/ide/ahci.h>
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
/* #define DEBUG_AHCI */
|
|
|
|
|
|
|
|
#ifdef DEBUG_AHCI
|
|
|
|
#define DPRINTF(port, fmt, ...) \
|
|
|
|
do { fprintf(stderr, "ahci: %s: [%d] ", __FUNCTION__, port); \
|
|
|
|
fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
|
|
|
|
#else
|
|
|
|
#define DPRINTF(port, fmt, ...) do {} while(0)
|
|
|
|
#endif
|
|
|
|
|
|
|
|
static void check_cmd(AHCIState *s, int port);
|
|
|
|
static int handle_cmd(AHCIState *s,int port,int slot);
|
|
|
|
static void ahci_reset_port(AHCIState *s, int port);
|
|
|
|
static void ahci_write_fis_d2h(AHCIDevice *ad, uint8_t *cmd_fis);
|
2011-02-01 14:51:29 +00:00
|
|
|
static void ahci_init_d2h(AHCIDevice *ad);
|
2014-10-01 22:55:47 +00:00
|
|
|
static int ahci_dma_prepare_buf(IDEDMA *dma, int is_write);
|
|
|
|
static void ahci_commit_buf(IDEDMA *dma, uint32_t tx_bytes);
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
static uint32_t ahci_port_read(AHCIState *s, int port, int offset)
|
|
|
|
{
|
|
|
|
uint32_t val;
|
|
|
|
AHCIPortRegs *pr;
|
|
|
|
pr = &s->dev[port].port_regs;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
case PORT_LST_ADDR:
|
|
|
|
val = pr->lst_addr;
|
|
|
|
break;
|
|
|
|
case PORT_LST_ADDR_HI:
|
|
|
|
val = pr->lst_addr_hi;
|
|
|
|
break;
|
|
|
|
case PORT_FIS_ADDR:
|
|
|
|
val = pr->fis_addr;
|
|
|
|
break;
|
|
|
|
case PORT_FIS_ADDR_HI:
|
|
|
|
val = pr->fis_addr_hi;
|
|
|
|
break;
|
|
|
|
case PORT_IRQ_STAT:
|
|
|
|
val = pr->irq_stat;
|
|
|
|
break;
|
|
|
|
case PORT_IRQ_MASK:
|
|
|
|
val = pr->irq_mask;
|
|
|
|
break;
|
|
|
|
case PORT_CMD:
|
|
|
|
val = pr->cmd;
|
|
|
|
break;
|
|
|
|
case PORT_TFDATA:
|
2014-08-21 17:44:36 +00:00
|
|
|
val = pr->tfdata;
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case PORT_SIG:
|
|
|
|
val = pr->sig;
|
|
|
|
break;
|
|
|
|
case PORT_SCR_STAT:
|
2014-10-07 11:59:18 +00:00
|
|
|
if (s->dev[port].port.ifs[0].blk) {
|
2010-12-14 00:34:40 +00:00
|
|
|
val = SATA_SCR_SSTATUS_DET_DEV_PRESENT_PHY_UP |
|
|
|
|
SATA_SCR_SSTATUS_SPD_GEN1 | SATA_SCR_SSTATUS_IPM_ACTIVE;
|
|
|
|
} else {
|
|
|
|
val = SATA_SCR_SSTATUS_DET_NODEV;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case PORT_SCR_CTL:
|
|
|
|
val = pr->scr_ctl;
|
|
|
|
break;
|
|
|
|
case PORT_SCR_ERR:
|
|
|
|
val = pr->scr_err;
|
|
|
|
break;
|
|
|
|
case PORT_SCR_ACT:
|
|
|
|
pr->scr_act &= ~s->dev[port].finished;
|
|
|
|
s->dev[port].finished = 0;
|
|
|
|
val = pr->scr_act;
|
|
|
|
break;
|
|
|
|
case PORT_CMD_ISSUE:
|
|
|
|
val = pr->cmd_issue;
|
|
|
|
break;
|
|
|
|
case PORT_RESERVED:
|
|
|
|
default:
|
|
|
|
val = 0;
|
|
|
|
}
|
|
|
|
DPRINTF(port, "offset: 0x%x val: 0x%x\n", offset, val);
|
|
|
|
return val;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_irq_raise(AHCIState *s, AHCIDevice *dev)
|
|
|
|
{
|
2013-06-30 12:19:24 +00:00
|
|
|
AHCIPCIState *d = container_of(s, AHCIPCIState, ahci);
|
2014-03-18 19:36:13 +00:00
|
|
|
PCIDevice *pci_dev =
|
|
|
|
(PCIDevice *)object_dynamic_cast(OBJECT(d), TYPE_PCI_DEVICE);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
DPRINTF(0, "raise irq\n");
|
|
|
|
|
2014-03-18 19:36:13 +00:00
|
|
|
if (pci_dev && msi_enabled(pci_dev)) {
|
2013-06-30 12:19:24 +00:00
|
|
|
msi_notify(pci_dev, 0);
|
2010-12-14 00:34:40 +00:00
|
|
|
} else {
|
|
|
|
qemu_irq_raise(s->irq);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_irq_lower(AHCIState *s, AHCIDevice *dev)
|
|
|
|
{
|
2013-06-30 12:19:24 +00:00
|
|
|
AHCIPCIState *d = container_of(s, AHCIPCIState, ahci);
|
2014-03-18 19:36:13 +00:00
|
|
|
PCIDevice *pci_dev =
|
|
|
|
(PCIDevice *)object_dynamic_cast(OBJECT(d), TYPE_PCI_DEVICE);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
DPRINTF(0, "lower irq\n");
|
|
|
|
|
2014-03-18 19:36:13 +00:00
|
|
|
if (!pci_dev || !msi_enabled(pci_dev)) {
|
2010-12-14 00:34:40 +00:00
|
|
|
qemu_irq_lower(s->irq);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_check_irq(AHCIState *s)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
DPRINTF(-1, "check irq %#x\n", s->control_regs.irqstatus);
|
|
|
|
|
2012-01-30 22:29:48 +00:00
|
|
|
s->control_regs.irqstatus = 0;
|
2011-02-01 14:51:31 +00:00
|
|
|
for (i = 0; i < s->ports; i++) {
|
2010-12-14 00:34:40 +00:00
|
|
|
AHCIPortRegs *pr = &s->dev[i].port_regs;
|
|
|
|
if (pr->irq_stat & pr->irq_mask) {
|
|
|
|
s->control_regs.irqstatus |= (1 << i);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (s->control_regs.irqstatus &&
|
|
|
|
(s->control_regs.ghc & HOST_CTL_IRQ_EN)) {
|
|
|
|
ahci_irq_raise(s, NULL);
|
|
|
|
} else {
|
|
|
|
ahci_irq_lower(s, NULL);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_trigger_irq(AHCIState *s, AHCIDevice *d,
|
|
|
|
int irq_type)
|
|
|
|
{
|
|
|
|
DPRINTF(d->port_no, "trigger irq %#x -> %x\n",
|
|
|
|
irq_type, d->port_regs.irq_mask & irq_type);
|
|
|
|
|
|
|
|
d->port_regs.irq_stat |= irq_type;
|
|
|
|
ahci_check_irq(s);
|
|
|
|
}
|
|
|
|
|
2014-07-03 08:26:27 +00:00
|
|
|
static void map_page(AddressSpace *as, uint8_t **ptr, uint64_t addr,
|
|
|
|
uint32_t wanted)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
2012-10-23 10:30:10 +00:00
|
|
|
hwaddr len = wanted;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
if (*ptr) {
|
2014-07-03 08:26:27 +00:00
|
|
|
dma_memory_unmap(as, *ptr, len, DMA_DIRECTION_FROM_DEVICE, len);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
2014-07-03 08:26:27 +00:00
|
|
|
*ptr = dma_memory_map(as, addr, &len, DMA_DIRECTION_FROM_DEVICE);
|
2010-12-14 00:34:40 +00:00
|
|
|
if (len < wanted) {
|
2014-07-03 08:26:27 +00:00
|
|
|
dma_memory_unmap(as, *ptr, len, DMA_DIRECTION_FROM_DEVICE, len);
|
2010-12-14 00:34:40 +00:00
|
|
|
*ptr = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_port_write(AHCIState *s, int port, int offset, uint32_t val)
|
|
|
|
{
|
|
|
|
AHCIPortRegs *pr = &s->dev[port].port_regs;
|
|
|
|
|
|
|
|
DPRINTF(port, "offset: 0x%x val: 0x%x\n", offset, val);
|
|
|
|
switch (offset) {
|
|
|
|
case PORT_LST_ADDR:
|
|
|
|
pr->lst_addr = val;
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &s->dev[port].lst,
|
2010-12-14 00:34:40 +00:00
|
|
|
((uint64_t)pr->lst_addr_hi << 32) | pr->lst_addr, 1024);
|
|
|
|
s->dev[port].cur_cmd = NULL;
|
|
|
|
break;
|
|
|
|
case PORT_LST_ADDR_HI:
|
|
|
|
pr->lst_addr_hi = val;
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &s->dev[port].lst,
|
2010-12-14 00:34:40 +00:00
|
|
|
((uint64_t)pr->lst_addr_hi << 32) | pr->lst_addr, 1024);
|
|
|
|
s->dev[port].cur_cmd = NULL;
|
|
|
|
break;
|
|
|
|
case PORT_FIS_ADDR:
|
|
|
|
pr->fis_addr = val;
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &s->dev[port].res_fis,
|
2010-12-14 00:34:40 +00:00
|
|
|
((uint64_t)pr->fis_addr_hi << 32) | pr->fis_addr, 256);
|
|
|
|
break;
|
|
|
|
case PORT_FIS_ADDR_HI:
|
|
|
|
pr->fis_addr_hi = val;
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &s->dev[port].res_fis,
|
2010-12-14 00:34:40 +00:00
|
|
|
((uint64_t)pr->fis_addr_hi << 32) | pr->fis_addr, 256);
|
|
|
|
break;
|
|
|
|
case PORT_IRQ_STAT:
|
|
|
|
pr->irq_stat &= ~val;
|
2012-01-30 22:29:48 +00:00
|
|
|
ahci_check_irq(s);
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case PORT_IRQ_MASK:
|
|
|
|
pr->irq_mask = val & 0xfdc000ff;
|
|
|
|
ahci_check_irq(s);
|
|
|
|
break;
|
|
|
|
case PORT_CMD:
|
|
|
|
pr->cmd = val & ~(PORT_CMD_LIST_ON | PORT_CMD_FIS_ON);
|
|
|
|
|
|
|
|
if (pr->cmd & PORT_CMD_START) {
|
|
|
|
pr->cmd |= PORT_CMD_LIST_ON;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (pr->cmd & PORT_CMD_FIS_RX) {
|
|
|
|
pr->cmd |= PORT_CMD_FIS_ON;
|
|
|
|
}
|
|
|
|
|
2011-02-01 14:51:29 +00:00
|
|
|
/* XXX usually the FIS would be pending on the bus here and
|
|
|
|
issuing deferred until the OS enables FIS receival.
|
|
|
|
Instead, we only submit it once - which works in most
|
|
|
|
cases, but is a hack. */
|
|
|
|
if ((pr->cmd & PORT_CMD_FIS_ON) &&
|
|
|
|
!s->dev[port].init_d2h_sent) {
|
|
|
|
ahci_init_d2h(&s->dev[port]);
|
2013-01-15 15:12:09 +00:00
|
|
|
s->dev[port].init_d2h_sent = true;
|
2011-02-01 14:51:29 +00:00
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
check_cmd(s, port);
|
|
|
|
break;
|
|
|
|
case PORT_TFDATA:
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Read Only. */
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case PORT_SIG:
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Read Only */
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case PORT_SCR_STAT:
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Read Only */
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case PORT_SCR_CTL:
|
|
|
|
if (((pr->scr_ctl & AHCI_SCR_SCTL_DET) == 1) &&
|
|
|
|
((val & AHCI_SCR_SCTL_DET) == 0)) {
|
|
|
|
ahci_reset_port(s, port);
|
|
|
|
}
|
|
|
|
pr->scr_ctl = val;
|
|
|
|
break;
|
|
|
|
case PORT_SCR_ERR:
|
|
|
|
pr->scr_err &= ~val;
|
|
|
|
break;
|
|
|
|
case PORT_SCR_ACT:
|
|
|
|
/* RW1 */
|
|
|
|
pr->scr_act |= val;
|
|
|
|
break;
|
|
|
|
case PORT_CMD_ISSUE:
|
|
|
|
pr->cmd_issue |= val;
|
|
|
|
check_cmd(s, port);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-10-23 10:30:10 +00:00
|
|
|
static uint64_t ahci_mem_read(void *opaque, hwaddr addr,
|
2011-08-08 13:09:14 +00:00
|
|
|
unsigned size)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
2011-08-08 13:09:14 +00:00
|
|
|
AHCIState *s = opaque;
|
2010-12-14 00:34:40 +00:00
|
|
|
uint32_t val = 0;
|
|
|
|
|
|
|
|
if (addr < AHCI_GENERIC_HOST_CONTROL_REGS_MAX_ADDR) {
|
|
|
|
switch (addr) {
|
|
|
|
case HOST_CAP:
|
|
|
|
val = s->control_regs.cap;
|
|
|
|
break;
|
|
|
|
case HOST_CTL:
|
|
|
|
val = s->control_regs.ghc;
|
|
|
|
break;
|
|
|
|
case HOST_IRQ_STAT:
|
|
|
|
val = s->control_regs.irqstatus;
|
|
|
|
break;
|
|
|
|
case HOST_PORTS_IMPL:
|
|
|
|
val = s->control_regs.impl;
|
|
|
|
break;
|
|
|
|
case HOST_VERSION:
|
|
|
|
val = s->control_regs.version;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
DPRINTF(-1, "(addr 0x%08X), val 0x%08X\n", (unsigned) addr, val);
|
|
|
|
} else if ((addr >= AHCI_PORT_REGS_START_ADDR) &&
|
2011-02-01 14:51:31 +00:00
|
|
|
(addr < (AHCI_PORT_REGS_START_ADDR +
|
|
|
|
(s->ports * AHCI_PORT_ADDR_OFFSET_LEN)))) {
|
2010-12-14 00:34:40 +00:00
|
|
|
val = ahci_port_read(s, (addr - AHCI_PORT_REGS_START_ADDR) >> 7,
|
|
|
|
addr & AHCI_PORT_ADDR_OFFSET_MASK);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2012-10-23 10:30:10 +00:00
|
|
|
static void ahci_mem_write(void *opaque, hwaddr addr,
|
2011-08-08 13:09:14 +00:00
|
|
|
uint64_t val, unsigned size)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
2011-08-08 13:09:14 +00:00
|
|
|
AHCIState *s = opaque;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
/* Only aligned reads are allowed on AHCI */
|
|
|
|
if (addr & 3) {
|
|
|
|
fprintf(stderr, "ahci: Mis-aligned write to addr 0x"
|
|
|
|
TARGET_FMT_plx "\n", addr);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (addr < AHCI_GENERIC_HOST_CONTROL_REGS_MAX_ADDR) {
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(-1, "(addr 0x%08X), val 0x%08"PRIX64"\n", (unsigned) addr, val);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
switch (addr) {
|
|
|
|
case HOST_CAP: /* R/WO, RO */
|
|
|
|
/* FIXME handle R/WO */
|
|
|
|
break;
|
|
|
|
case HOST_CTL: /* R/W */
|
|
|
|
if (val & HOST_CTL_RESET) {
|
|
|
|
DPRINTF(-1, "HBA Reset\n");
|
2012-05-11 14:42:36 +00:00
|
|
|
ahci_reset(s);
|
2010-12-14 00:34:40 +00:00
|
|
|
} else {
|
|
|
|
s->control_regs.ghc = (val & 0x3) | HOST_CTL_AHCI_EN;
|
|
|
|
ahci_check_irq(s);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case HOST_IRQ_STAT: /* R/WC, RO */
|
|
|
|
s->control_regs.irqstatus &= ~val;
|
|
|
|
ahci_check_irq(s);
|
|
|
|
break;
|
|
|
|
case HOST_PORTS_IMPL: /* R/WO, RO */
|
|
|
|
/* FIXME handle R/WO */
|
|
|
|
break;
|
|
|
|
case HOST_VERSION: /* RO */
|
|
|
|
/* FIXME report write? */
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
DPRINTF(-1, "write to unknown register 0x%x\n", (unsigned)addr);
|
|
|
|
}
|
|
|
|
} else if ((addr >= AHCI_PORT_REGS_START_ADDR) &&
|
2011-02-01 14:51:31 +00:00
|
|
|
(addr < (AHCI_PORT_REGS_START_ADDR +
|
|
|
|
(s->ports * AHCI_PORT_ADDR_OFFSET_LEN)))) {
|
2010-12-14 00:34:40 +00:00
|
|
|
ahci_port_write(s, (addr - AHCI_PORT_REGS_START_ADDR) >> 7,
|
|
|
|
addr & AHCI_PORT_ADDR_OFFSET_MASK, val);
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2012-02-05 10:19:07 +00:00
|
|
|
static const MemoryRegionOps ahci_mem_ops = {
|
2011-08-08 13:09:14 +00:00
|
|
|
.read = ahci_mem_read,
|
|
|
|
.write = ahci_mem_write,
|
|
|
|
.endianness = DEVICE_LITTLE_ENDIAN,
|
2010-12-14 00:34:40 +00:00
|
|
|
};
|
|
|
|
|
2012-10-23 10:30:10 +00:00
|
|
|
static uint64_t ahci_idp_read(void *opaque, hwaddr addr,
|
2011-08-27 09:12:28 +00:00
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
AHCIState *s = opaque;
|
|
|
|
|
|
|
|
if (addr == s->idp_offset) {
|
|
|
|
/* index register */
|
|
|
|
return s->idp_index;
|
|
|
|
} else if (addr == s->idp_offset + 4) {
|
|
|
|
/* data register - do memory read at location selected by index */
|
|
|
|
return ahci_mem_read(opaque, s->idp_index, size);
|
|
|
|
} else {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-10-23 10:30:10 +00:00
|
|
|
static void ahci_idp_write(void *opaque, hwaddr addr,
|
2011-08-27 09:12:28 +00:00
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
AHCIState *s = opaque;
|
|
|
|
|
|
|
|
if (addr == s->idp_offset) {
|
|
|
|
/* index register - mask off reserved bits */
|
|
|
|
s->idp_index = (uint32_t)val & ((AHCI_MEM_BAR_SIZE - 1) & ~3);
|
|
|
|
} else if (addr == s->idp_offset + 4) {
|
|
|
|
/* data register - do memory write at location selected by index */
|
|
|
|
ahci_mem_write(opaque, s->idp_index, val, size);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-02-05 10:19:07 +00:00
|
|
|
static const MemoryRegionOps ahci_idp_ops = {
|
2011-08-27 09:12:28 +00:00
|
|
|
.read = ahci_idp_read,
|
|
|
|
.write = ahci_idp_write,
|
|
|
|
.endianness = DEVICE_LITTLE_ENDIAN,
|
|
|
|
};
|
|
|
|
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
static void ahci_reg_init(AHCIState *s)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
s->control_regs.cap = (s->ports - 1) |
|
2010-12-14 00:34:40 +00:00
|
|
|
(AHCI_NUM_COMMAND_SLOTS << 8) |
|
|
|
|
(AHCI_SUPPORTED_SPEED_GEN1 << AHCI_SUPPORTED_SPEED) |
|
|
|
|
HOST_CAP_NCQ | HOST_CAP_AHCI;
|
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
s->control_regs.impl = (1 << s->ports) - 1;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
s->control_regs.version = AHCI_VERSION_1_0;
|
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
for (i = 0; i < s->ports; i++) {
|
2010-12-14 00:34:40 +00:00
|
|
|
s->dev[i].port_state = STATE_RUN;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void check_cmd(AHCIState *s, int port)
|
|
|
|
{
|
|
|
|
AHCIPortRegs *pr = &s->dev[port].port_regs;
|
|
|
|
int slot;
|
|
|
|
|
|
|
|
if ((pr->cmd & PORT_CMD_START) && pr->cmd_issue) {
|
|
|
|
for (slot = 0; (slot < 32) && pr->cmd_issue; slot++) {
|
2014-03-28 15:12:55 +00:00
|
|
|
if ((pr->cmd_issue & (1U << slot)) &&
|
2010-12-14 00:34:40 +00:00
|
|
|
!handle_cmd(s, port, slot)) {
|
2014-03-28 15:12:55 +00:00
|
|
|
pr->cmd_issue &= ~(1U << slot);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_check_cmd_bh(void *opaque)
|
|
|
|
{
|
|
|
|
AHCIDevice *ad = opaque;
|
|
|
|
|
|
|
|
qemu_bh_delete(ad->check_bh);
|
|
|
|
ad->check_bh = NULL;
|
|
|
|
|
|
|
|
if ((ad->busy_slot != -1) &&
|
|
|
|
!(ad->port.ifs[0].status & (BUSY_STAT|DRQ_STAT))) {
|
|
|
|
/* no longer busy */
|
|
|
|
ad->port_regs.cmd_issue &= ~(1 << ad->busy_slot);
|
|
|
|
ad->busy_slot = -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
check_cmd(ad->hba, ad->port_no);
|
|
|
|
}
|
|
|
|
|
2011-02-01 14:51:29 +00:00
|
|
|
static void ahci_init_d2h(AHCIDevice *ad)
|
|
|
|
{
|
2012-05-22 23:26:42 +00:00
|
|
|
uint8_t init_fis[20];
|
2011-02-01 14:51:29 +00:00
|
|
|
IDEState *ide_state = &ad->port.ifs[0];
|
|
|
|
|
|
|
|
memset(init_fis, 0, sizeof(init_fis));
|
|
|
|
|
|
|
|
init_fis[4] = 1;
|
|
|
|
init_fis[12] = 1;
|
|
|
|
|
|
|
|
if (ide_state->drive_kind == IDE_CD) {
|
|
|
|
init_fis[5] = ide_state->lcyl;
|
|
|
|
init_fis[6] = ide_state->hcyl;
|
|
|
|
}
|
|
|
|
|
|
|
|
ahci_write_fis_d2h(ad, init_fis);
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
static void ahci_reset_port(AHCIState *s, int port)
|
|
|
|
{
|
|
|
|
AHCIDevice *d = &s->dev[port];
|
|
|
|
AHCIPortRegs *pr = &d->port_regs;
|
|
|
|
IDEState *ide_state = &d->port.ifs[0];
|
|
|
|
int i;
|
|
|
|
|
|
|
|
DPRINTF(port, "reset port\n");
|
|
|
|
|
|
|
|
ide_bus_reset(&d->port);
|
|
|
|
ide_state->ncq_queues = AHCI_MAX_CMDS;
|
|
|
|
|
|
|
|
pr->scr_stat = 0;
|
|
|
|
pr->scr_err = 0;
|
|
|
|
pr->scr_act = 0;
|
2014-08-21 17:44:36 +00:00
|
|
|
pr->tfdata = 0x7F;
|
|
|
|
pr->sig = 0xFFFFFFFF;
|
2010-12-14 00:34:40 +00:00
|
|
|
d->busy_slot = -1;
|
2013-01-15 15:12:09 +00:00
|
|
|
d->init_d2h_sent = false;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
ide_state = &s->dev[port].port.ifs[0];
|
2014-10-07 11:59:18 +00:00
|
|
|
if (!ide_state->blk) {
|
2010-12-14 00:34:40 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* reset ncq queue */
|
|
|
|
for (i = 0; i < AHCI_MAX_CMDS; i++) {
|
|
|
|
NCQTransferState *ncq_tfs = &s->dev[port].ncq_tfs[i];
|
|
|
|
if (!ncq_tfs->used) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ncq_tfs->aiocb) {
|
2014-10-07 11:59:18 +00:00
|
|
|
blk_aio_cancel(ncq_tfs->aiocb);
|
2010-12-14 00:34:40 +00:00
|
|
|
ncq_tfs->aiocb = NULL;
|
|
|
|
}
|
|
|
|
|
2014-10-07 11:59:18 +00:00
|
|
|
/* Maybe we just finished the request thanks to blk_aio_cancel() */
|
2012-01-30 22:29:47 +00:00
|
|
|
if (!ncq_tfs->used) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
qemu_sglist_destroy(&ncq_tfs->sglist);
|
|
|
|
ncq_tfs->used = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
s->dev[port].port_state = STATE_RUN;
|
2014-10-07 11:59:18 +00:00
|
|
|
if (!ide_state->blk) {
|
2014-08-21 17:44:36 +00:00
|
|
|
pr->sig = 0;
|
2010-12-19 10:43:09 +00:00
|
|
|
ide_state->status = SEEK_STAT | WRERR_STAT;
|
2010-12-14 00:34:40 +00:00
|
|
|
} else if (ide_state->drive_kind == IDE_CD) {
|
2014-08-21 17:44:36 +00:00
|
|
|
pr->sig = SATA_SIGNATURE_CDROM;
|
2010-12-14 00:34:40 +00:00
|
|
|
ide_state->lcyl = 0x14;
|
|
|
|
ide_state->hcyl = 0xeb;
|
|
|
|
DPRINTF(port, "set lcyl = %d\n", ide_state->lcyl);
|
|
|
|
ide_state->status = SEEK_STAT | WRERR_STAT | READY_STAT;
|
|
|
|
} else {
|
2014-08-21 17:44:36 +00:00
|
|
|
pr->sig = SATA_SIGNATURE_DISK;
|
2010-12-14 00:34:40 +00:00
|
|
|
ide_state->status = SEEK_STAT | WRERR_STAT;
|
|
|
|
}
|
|
|
|
|
|
|
|
ide_state->error = 1;
|
2011-02-01 14:51:29 +00:00
|
|
|
ahci_init_d2h(d);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void debug_print_fis(uint8_t *fis, int cmd_len)
|
|
|
|
{
|
|
|
|
#ifdef DEBUG_AHCI
|
|
|
|
int i;
|
|
|
|
|
|
|
|
fprintf(stderr, "fis:");
|
|
|
|
for (i = 0; i < cmd_len; i++) {
|
|
|
|
if ((i & 0xf) == 0) {
|
|
|
|
fprintf(stderr, "\n%02x:",i);
|
|
|
|
}
|
|
|
|
fprintf(stderr, "%02x ",fis[i]);
|
|
|
|
}
|
|
|
|
fprintf(stderr, "\n");
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_write_fis_sdb(AHCIState *s, int port, uint32_t finished)
|
|
|
|
{
|
2014-08-21 17:44:36 +00:00
|
|
|
AHCIDevice *ad = &s->dev[port];
|
|
|
|
AHCIPortRegs *pr = &ad->port_regs;
|
2010-12-14 00:34:40 +00:00
|
|
|
IDEState *ide_state;
|
2014-10-01 22:55:51 +00:00
|
|
|
SDBFIS *sdb_fis;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
if (!s->dev[port].res_fis ||
|
|
|
|
!(pr->cmd & PORT_CMD_FIS_RX)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2014-10-01 22:55:51 +00:00
|
|
|
sdb_fis = (SDBFIS *)&ad->res_fis[RES_FIS_SDBFIS];
|
2014-08-21 17:44:36 +00:00
|
|
|
ide_state = &ad->port.ifs[0];
|
2010-12-14 00:34:40 +00:00
|
|
|
|
2014-10-01 22:55:51 +00:00
|
|
|
sdb_fis->type = 0xA1;
|
|
|
|
/* Interrupt pending & Notification bit */
|
|
|
|
sdb_fis->flags = (ad->hba->control_regs.irqstatus ? (1 << 6) : 0);
|
|
|
|
sdb_fis->status = ide_state->status & 0x77;
|
|
|
|
sdb_fis->error = ide_state->error;
|
|
|
|
/* update SAct field in SDB_FIS */
|
2010-12-14 00:34:40 +00:00
|
|
|
s->dev[port].finished |= finished;
|
2014-10-01 22:55:51 +00:00
|
|
|
sdb_fis->payload = cpu_to_le32(ad->finished);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Update shadow registers (except BSY 0x80 and DRQ 0x08) */
|
|
|
|
pr->tfdata = (ad->port.ifs[0].error << 8) |
|
|
|
|
(ad->port.ifs[0].status & 0x77) |
|
|
|
|
(pr->tfdata & 0x88);
|
|
|
|
|
|
|
|
ahci_trigger_irq(s, ad, PORT_IRQ_SDB_FIS);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
2014-08-04 21:11:18 +00:00
|
|
|
static void ahci_write_fis_pio(AHCIDevice *ad, uint16_t len)
|
|
|
|
{
|
|
|
|
AHCIPortRegs *pr = &ad->port_regs;
|
|
|
|
uint8_t *pio_fis, *cmd_fis;
|
|
|
|
uint64_t tbl_addr;
|
|
|
|
dma_addr_t cmd_len = 0x80;
|
2014-10-01 22:55:46 +00:00
|
|
|
IDEState *s = &ad->port.ifs[0];
|
2014-08-04 21:11:18 +00:00
|
|
|
|
|
|
|
if (!ad->res_fis || !(pr->cmd & PORT_CMD_FIS_RX)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* map cmd_fis */
|
|
|
|
tbl_addr = le64_to_cpu(ad->cur_cmd->tbl_addr);
|
|
|
|
cmd_fis = dma_memory_map(ad->hba->as, tbl_addr, &cmd_len,
|
|
|
|
DMA_DIRECTION_TO_DEVICE);
|
|
|
|
|
|
|
|
if (cmd_fis == NULL) {
|
|
|
|
DPRINTF(ad->port_no, "dma_memory_map failed in ahci_write_fis_pio");
|
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_HBUS_ERR);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cmd_len != 0x80) {
|
|
|
|
DPRINTF(ad->port_no,
|
|
|
|
"dma_memory_map mapped too few bytes in ahci_write_fis_pio");
|
|
|
|
dma_memory_unmap(ad->hba->as, cmd_fis, cmd_len,
|
|
|
|
DMA_DIRECTION_TO_DEVICE, cmd_len);
|
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_HBUS_ERR);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
pio_fis = &ad->res_fis[RES_FIS_PSFIS];
|
|
|
|
|
|
|
|
pio_fis[0] = 0x5f;
|
|
|
|
pio_fis[1] = (ad->hba->control_regs.irqstatus ? (1 << 6) : 0);
|
2014-10-01 22:55:46 +00:00
|
|
|
pio_fis[2] = s->status;
|
|
|
|
pio_fis[3] = s->error;
|
|
|
|
|
|
|
|
pio_fis[4] = s->sector;
|
|
|
|
pio_fis[5] = s->lcyl;
|
|
|
|
pio_fis[6] = s->hcyl;
|
|
|
|
pio_fis[7] = s->select;
|
|
|
|
pio_fis[8] = s->hob_sector;
|
|
|
|
pio_fis[9] = s->hob_lcyl;
|
|
|
|
pio_fis[10] = s->hob_hcyl;
|
|
|
|
pio_fis[11] = 0;
|
2014-08-04 21:11:18 +00:00
|
|
|
pio_fis[12] = cmd_fis[12];
|
|
|
|
pio_fis[13] = cmd_fis[13];
|
|
|
|
pio_fis[14] = 0;
|
2014-10-01 22:55:46 +00:00
|
|
|
pio_fis[15] = s->status;
|
2014-08-04 21:11:18 +00:00
|
|
|
pio_fis[16] = len & 255;
|
|
|
|
pio_fis[17] = len >> 8;
|
|
|
|
pio_fis[18] = 0;
|
|
|
|
pio_fis[19] = 0;
|
|
|
|
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Update shadow registers: */
|
|
|
|
pr->tfdata = (ad->port.ifs[0].error << 8) |
|
|
|
|
ad->port.ifs[0].status;
|
|
|
|
|
2014-08-04 21:11:18 +00:00
|
|
|
if (pio_fis[2] & ERR_STAT) {
|
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_TF_ERR);
|
|
|
|
}
|
|
|
|
|
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_PIOS_FIS);
|
|
|
|
|
|
|
|
dma_memory_unmap(ad->hba->as, cmd_fis, cmd_len,
|
|
|
|
DMA_DIRECTION_TO_DEVICE, cmd_len);
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
static void ahci_write_fis_d2h(AHCIDevice *ad, uint8_t *cmd_fis)
|
|
|
|
{
|
|
|
|
AHCIPortRegs *pr = &ad->port_regs;
|
|
|
|
uint8_t *d2h_fis;
|
|
|
|
int i;
|
2012-06-27 04:50:41 +00:00
|
|
|
dma_addr_t cmd_len = 0x80;
|
2010-12-14 00:34:40 +00:00
|
|
|
int cmd_mapped = 0;
|
2014-10-01 22:55:46 +00:00
|
|
|
IDEState *s = &ad->port.ifs[0];
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
if (!ad->res_fis || !(pr->cmd & PORT_CMD_FIS_RX)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!cmd_fis) {
|
|
|
|
/* map cmd_fis */
|
|
|
|
uint64_t tbl_addr = le64_to_cpu(ad->cur_cmd->tbl_addr);
|
2013-04-10 16:15:49 +00:00
|
|
|
cmd_fis = dma_memory_map(ad->hba->as, tbl_addr, &cmd_len,
|
2012-06-27 04:50:41 +00:00
|
|
|
DMA_DIRECTION_TO_DEVICE);
|
2010-12-14 00:34:40 +00:00
|
|
|
cmd_mapped = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
d2h_fis = &ad->res_fis[RES_FIS_RFIS];
|
|
|
|
|
|
|
|
d2h_fis[0] = 0x34;
|
|
|
|
d2h_fis[1] = (ad->hba->control_regs.irqstatus ? (1 << 6) : 0);
|
2014-10-01 22:55:46 +00:00
|
|
|
d2h_fis[2] = s->status;
|
|
|
|
d2h_fis[3] = s->error;
|
|
|
|
|
|
|
|
d2h_fis[4] = s->sector;
|
|
|
|
d2h_fis[5] = s->lcyl;
|
|
|
|
d2h_fis[6] = s->hcyl;
|
|
|
|
d2h_fis[7] = s->select;
|
|
|
|
d2h_fis[8] = s->hob_sector;
|
|
|
|
d2h_fis[9] = s->hob_lcyl;
|
|
|
|
d2h_fis[10] = s->hob_hcyl;
|
|
|
|
d2h_fis[11] = 0;
|
2010-12-14 00:34:40 +00:00
|
|
|
d2h_fis[12] = cmd_fis[12];
|
|
|
|
d2h_fis[13] = cmd_fis[13];
|
2012-05-22 23:26:42 +00:00
|
|
|
for (i = 14; i < 20; i++) {
|
2010-12-14 00:34:40 +00:00
|
|
|
d2h_fis[i] = 0;
|
|
|
|
}
|
|
|
|
|
2014-08-21 17:44:36 +00:00
|
|
|
/* Update shadow registers: */
|
|
|
|
pr->tfdata = (ad->port.ifs[0].error << 8) |
|
|
|
|
ad->port.ifs[0].status;
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
if (d2h_fis[2] & ERR_STAT) {
|
2014-08-04 21:11:15 +00:00
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_TF_ERR);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
ahci_trigger_irq(ad->hba, ad, PORT_IRQ_D2H_REG_FIS);
|
|
|
|
|
|
|
|
if (cmd_mapped) {
|
2013-04-10 16:15:49 +00:00
|
|
|
dma_memory_unmap(ad->hba->as, cmd_fis, cmd_len,
|
2012-06-27 04:50:41 +00:00
|
|
|
DMA_DIRECTION_TO_DEVICE, cmd_len);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-07-01 11:13:27 +00:00
|
|
|
static int prdt_tbl_entry_size(const AHCI_SG *tbl)
|
|
|
|
{
|
|
|
|
return (le32_to_cpu(tbl->flags_size) & AHCI_PRDT_SIZE_MASK) + 1;
|
|
|
|
}
|
|
|
|
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
static int ahci_populate_sglist(AHCIDevice *ad, QEMUSGList *sglist,
|
|
|
|
int32_t offset)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
AHCICmdHdr *cmd = ad->cur_cmd;
|
|
|
|
uint32_t opts = le32_to_cpu(cmd->opts);
|
|
|
|
uint64_t prdt_addr = le64_to_cpu(cmd->tbl_addr) + 0x80;
|
|
|
|
int sglist_alloc_hint = opts >> AHCI_CMD_HDR_PRDT_LEN;
|
2012-06-27 04:50:41 +00:00
|
|
|
dma_addr_t prdt_len = (sglist_alloc_hint * sizeof(AHCI_SG));
|
|
|
|
dma_addr_t real_prdt_len = prdt_len;
|
2010-12-14 00:34:40 +00:00
|
|
|
uint8_t *prdt;
|
|
|
|
int i;
|
|
|
|
int r = 0;
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
uint64_t sum = 0;
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
int off_idx = -1;
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
int64_t off_pos = -1;
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
int tbl_entry_size;
|
2013-06-03 12:17:19 +00:00
|
|
|
IDEBus *bus = &ad->port;
|
|
|
|
BusState *qbus = BUS(bus);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
/*
|
|
|
|
* Note: AHCI PRDT can describe up to 256GiB. SATA/ATA only support
|
|
|
|
* transactions of up to 32MiB as of ATA8-ACS3 rev 1b, assuming a
|
|
|
|
* 512 byte sector size. We limit the PRDT in this implementation to
|
|
|
|
* a reasonably large 2GiB, which can accommodate the maximum transfer
|
|
|
|
* request for sector sizes up to 32K.
|
|
|
|
*/
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
if (!sglist_alloc_hint) {
|
|
|
|
DPRINTF(ad->port_no, "no sg list given by guest: 0x%08x\n", opts);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* map PRDT */
|
2013-04-10 16:15:49 +00:00
|
|
|
if (!(prdt = dma_memory_map(ad->hba->as, prdt_addr, &prdt_len,
|
2012-06-27 04:50:41 +00:00
|
|
|
DMA_DIRECTION_TO_DEVICE))){
|
2010-12-14 00:34:40 +00:00
|
|
|
DPRINTF(ad->port_no, "map failed\n");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (prdt_len < real_prdt_len) {
|
|
|
|
DPRINTF(ad->port_no, "mapped less than expected\n");
|
|
|
|
r = -1;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Get entries in the PRDT, init a qemu sglist accordingly */
|
|
|
|
if (sglist_alloc_hint > 0) {
|
|
|
|
AHCI_SG *tbl = (AHCI_SG *)prdt;
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
sum = 0;
|
2010-12-14 00:34:40 +00:00
|
|
|
for (i = 0; i < sglist_alloc_hint; i++) {
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
/* flags_size is zero-based */
|
2014-07-01 11:13:27 +00:00
|
|
|
tbl_entry_size = prdt_tbl_entry_size(&tbl[i]);
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
if (offset <= (sum + tbl_entry_size)) {
|
|
|
|
off_idx = i;
|
|
|
|
off_pos = offset - sum;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
sum += tbl_entry_size;
|
|
|
|
}
|
|
|
|
if ((off_idx == -1) || (off_pos < 0) || (off_pos > tbl_entry_size)) {
|
|
|
|
DPRINTF(ad->port_no, "%s: Incorrect offset! "
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
"off_idx: %d, off_pos: %"PRId64"\n",
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
__func__, off_idx, off_pos);
|
|
|
|
r = -1;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2013-06-03 12:17:19 +00:00
|
|
|
qemu_sglist_init(sglist, qbus->parent, (sglist_alloc_hint - off_idx),
|
|
|
|
ad->hba->as);
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
qemu_sglist_add(sglist, le64_to_cpu(tbl[off_idx].addr + off_pos),
|
2014-07-01 11:13:27 +00:00
|
|
|
prdt_tbl_entry_size(&tbl[off_idx]) - off_pos);
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
|
|
|
|
for (i = off_idx + 1; i < sglist_alloc_hint; i++) {
|
2010-12-14 00:34:40 +00:00
|
|
|
/* flags_size is zero-based */
|
|
|
|
qemu_sglist_add(sglist, le64_to_cpu(tbl[i].addr),
|
2014-07-01 11:13:27 +00:00
|
|
|
prdt_tbl_entry_size(&tbl[i]));
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
if (sglist->size > INT32_MAX) {
|
|
|
|
error_report("AHCI Physical Region Descriptor Table describes "
|
|
|
|
"more than 2 GiB.\n");
|
|
|
|
qemu_sglist_destroy(sglist);
|
|
|
|
r = -1;
|
|
|
|
goto out;
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2013-04-10 16:15:49 +00:00
|
|
|
dma_memory_unmap(ad->hba->as, prdt, prdt_len,
|
2012-06-27 04:50:41 +00:00
|
|
|
DMA_DIRECTION_TO_DEVICE, prdt_len);
|
2010-12-14 00:34:40 +00:00
|
|
|
return r;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ncq_cb(void *opaque, int ret)
|
|
|
|
{
|
|
|
|
NCQTransferState *ncq_tfs = (NCQTransferState *)opaque;
|
|
|
|
IDEState *ide_state = &ncq_tfs->drive->port.ifs[0];
|
|
|
|
|
2014-09-11 05:41:07 +00:00
|
|
|
if (ret == -ECANCELED) {
|
|
|
|
return;
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
/* Clear bit for this tag in SActive */
|
|
|
|
ncq_tfs->drive->port_regs.scr_act &= ~(1 << ncq_tfs->tag);
|
|
|
|
|
|
|
|
if (ret < 0) {
|
|
|
|
/* error */
|
|
|
|
ide_state->error = ABRT_ERR;
|
|
|
|
ide_state->status = READY_STAT | ERR_STAT;
|
|
|
|
ncq_tfs->drive->port_regs.scr_err |= (1 << ncq_tfs->tag);
|
|
|
|
} else {
|
|
|
|
ide_state->status = READY_STAT | SEEK_STAT;
|
|
|
|
}
|
|
|
|
|
|
|
|
ahci_write_fis_sdb(ncq_tfs->drive->hba, ncq_tfs->drive->port_no,
|
|
|
|
(1 << ncq_tfs->tag));
|
|
|
|
|
|
|
|
DPRINTF(ncq_tfs->drive->port_no, "NCQ transfer tag %d finished\n",
|
|
|
|
ncq_tfs->tag);
|
|
|
|
|
2014-10-07 11:59:18 +00:00
|
|
|
block_acct_done(blk_get_stats(ncq_tfs->drive->port.ifs[0].blk),
|
2014-09-05 13:46:18 +00:00
|
|
|
&ncq_tfs->acct);
|
2010-12-14 00:34:40 +00:00
|
|
|
qemu_sglist_destroy(&ncq_tfs->sglist);
|
|
|
|
ncq_tfs->used = 0;
|
|
|
|
}
|
|
|
|
|
2014-11-03 23:56:15 +00:00
|
|
|
static int is_ncq(uint8_t ata_cmd)
|
|
|
|
{
|
|
|
|
/* Based on SATA 3.2 section 13.6.3.2 */
|
|
|
|
switch (ata_cmd) {
|
|
|
|
case READ_FPDMA_QUEUED:
|
|
|
|
case WRITE_FPDMA_QUEUED:
|
|
|
|
case NCQ_NON_DATA:
|
|
|
|
case RECEIVE_FPDMA_QUEUED:
|
|
|
|
case SEND_FPDMA_QUEUED:
|
|
|
|
return 1;
|
|
|
|
default:
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
static void process_ncq_command(AHCIState *s, int port, uint8_t *cmd_fis,
|
|
|
|
int slot)
|
|
|
|
{
|
|
|
|
NCQFrame *ncq_fis = (NCQFrame*)cmd_fis;
|
|
|
|
uint8_t tag = ncq_fis->tag >> 3;
|
|
|
|
NCQTransferState *ncq_tfs = &s->dev[port].ncq_tfs[tag];
|
|
|
|
|
|
|
|
if (ncq_tfs->used) {
|
|
|
|
/* error - already in use */
|
|
|
|
fprintf(stderr, "%s: tag %d already used\n", __FUNCTION__, tag);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
ncq_tfs->used = 1;
|
|
|
|
ncq_tfs->drive = &s->dev[port];
|
|
|
|
ncq_tfs->slot = slot;
|
|
|
|
ncq_tfs->lba = ((uint64_t)ncq_fis->lba5 << 40) |
|
|
|
|
((uint64_t)ncq_fis->lba4 << 32) |
|
|
|
|
((uint64_t)ncq_fis->lba3 << 24) |
|
|
|
|
((uint64_t)ncq_fis->lba2 << 16) |
|
|
|
|
((uint64_t)ncq_fis->lba1 << 8) |
|
|
|
|
(uint64_t)ncq_fis->lba0;
|
|
|
|
|
|
|
|
/* Note: We calculate the sector count, but don't currently rely on it.
|
|
|
|
* The total size of the DMA buffer tells us the transfer size instead. */
|
|
|
|
ncq_tfs->sector_count = ((uint16_t)ncq_fis->sector_count_high << 8) |
|
|
|
|
ncq_fis->sector_count_low;
|
|
|
|
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(port, "NCQ transfer LBA from %"PRId64" to %"PRId64", "
|
|
|
|
"drive max %"PRId64"\n",
|
2010-12-14 00:34:40 +00:00
|
|
|
ncq_tfs->lba, ncq_tfs->lba + ncq_tfs->sector_count - 2,
|
|
|
|
s->dev[port].port.ifs[0].nb_sectors - 1);
|
|
|
|
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
ahci_populate_sglist(&s->dev[port], &ncq_tfs->sglist, 0);
|
2010-12-14 00:34:40 +00:00
|
|
|
ncq_tfs->tag = tag;
|
|
|
|
|
|
|
|
switch(ncq_fis->command) {
|
|
|
|
case READ_FPDMA_QUEUED:
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(port, "NCQ reading %d sectors from LBA %"PRId64", "
|
|
|
|
"tag %d\n",
|
2010-12-14 00:34:40 +00:00
|
|
|
ncq_tfs->sector_count-1, ncq_tfs->lba, ncq_tfs->tag);
|
|
|
|
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(port, "tag %d aio read %"PRId64"\n",
|
|
|
|
ncq_tfs->tag, ncq_tfs->lba);
|
2011-08-25 06:26:01 +00:00
|
|
|
|
2014-10-07 11:59:18 +00:00
|
|
|
dma_acct_start(ncq_tfs->drive->port.ifs[0].blk, &ncq_tfs->acct,
|
2014-09-05 13:46:17 +00:00
|
|
|
&ncq_tfs->sglist, BLOCK_ACCT_READ);
|
2014-10-07 11:59:18 +00:00
|
|
|
ncq_tfs->aiocb = dma_blk_read(ncq_tfs->drive->port.ifs[0].blk,
|
|
|
|
&ncq_tfs->sglist, ncq_tfs->lba,
|
|
|
|
ncq_cb, ncq_tfs);
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
case WRITE_FPDMA_QUEUED:
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(port, "NCQ writing %d sectors to LBA %"PRId64", tag %d\n",
|
2010-12-14 00:34:40 +00:00
|
|
|
ncq_tfs->sector_count-1, ncq_tfs->lba, ncq_tfs->tag);
|
|
|
|
|
2011-10-26 11:06:14 +00:00
|
|
|
DPRINTF(port, "tag %d aio write %"PRId64"\n",
|
|
|
|
ncq_tfs->tag, ncq_tfs->lba);
|
2011-08-25 06:26:01 +00:00
|
|
|
|
2014-10-07 11:59:18 +00:00
|
|
|
dma_acct_start(ncq_tfs->drive->port.ifs[0].blk, &ncq_tfs->acct,
|
2014-09-05 13:46:17 +00:00
|
|
|
&ncq_tfs->sglist, BLOCK_ACCT_WRITE);
|
2014-10-07 11:59:18 +00:00
|
|
|
ncq_tfs->aiocb = dma_blk_write(ncq_tfs->drive->port.ifs[0].blk,
|
|
|
|
&ncq_tfs->sglist, ncq_tfs->lba,
|
|
|
|
ncq_cb, ncq_tfs);
|
2010-12-14 00:34:40 +00:00
|
|
|
break;
|
|
|
|
default:
|
2014-11-03 23:56:15 +00:00
|
|
|
if (is_ncq(cmd_fis[2])) {
|
|
|
|
DPRINTF(port,
|
|
|
|
"error: unsupported NCQ command (0x%02x) received\n",
|
|
|
|
cmd_fis[2]);
|
|
|
|
} else {
|
|
|
|
DPRINTF(port,
|
|
|
|
"error: tried to process non-NCQ command as NCQ\n");
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
qemu_sglist_destroy(&ncq_tfs->sglist);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int handle_cmd(AHCIState *s, int port, int slot)
|
|
|
|
{
|
|
|
|
IDEState *ide_state;
|
|
|
|
uint32_t opts;
|
|
|
|
uint64_t tbl_addr;
|
|
|
|
AHCICmdHdr *cmd;
|
|
|
|
uint8_t *cmd_fis;
|
2012-06-27 04:50:41 +00:00
|
|
|
dma_addr_t cmd_len;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
if (s->dev[port].port.ifs[0].status & (BUSY_STAT|DRQ_STAT)) {
|
|
|
|
/* Engine currently busy, try again later */
|
|
|
|
DPRINTF(port, "engine busy\n");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!s->dev[port].lst) {
|
|
|
|
DPRINTF(port, "error: lst not given but cmd handled");
|
|
|
|
return -1;
|
|
|
|
}
|
2014-11-03 23:56:17 +00:00
|
|
|
cmd = &((AHCICmdHdr *)s->dev[port].lst)[slot];
|
2010-12-14 00:34:40 +00:00
|
|
|
/* remember current slot handle for later */
|
|
|
|
s->dev[port].cur_cmd = cmd;
|
|
|
|
|
2014-11-03 23:56:17 +00:00
|
|
|
/* The device we are working for */
|
|
|
|
ide_state = &s->dev[port].port.ifs[0];
|
|
|
|
if (!ide_state->blk) {
|
|
|
|
DPRINTF(port, "error: guest accessed unused port");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
opts = le32_to_cpu(cmd->opts);
|
|
|
|
tbl_addr = le64_to_cpu(cmd->tbl_addr);
|
|
|
|
cmd_len = 0x80;
|
2013-04-10 16:15:49 +00:00
|
|
|
cmd_fis = dma_memory_map(s->as, tbl_addr, &cmd_len,
|
2012-06-27 04:50:41 +00:00
|
|
|
DMA_DIRECTION_FROM_DEVICE);
|
2010-12-14 00:34:40 +00:00
|
|
|
if (!cmd_fis) {
|
|
|
|
DPRINTF(port, "error: guest passed us an invalid cmd fis\n");
|
|
|
|
return -1;
|
2014-11-03 23:56:17 +00:00
|
|
|
} else if (cmd_len != 0x80) {
|
|
|
|
ahci_trigger_irq(s, &s->dev[port], PORT_IRQ_HBUS_ERR);
|
|
|
|
DPRINTF(port, "error: dma_memory_map failed: "
|
|
|
|
"(len(%02"PRIx64") != 0x80)\n",
|
|
|
|
cmd_len);
|
2010-12-14 00:34:40 +00:00
|
|
|
goto out;
|
|
|
|
}
|
2014-11-03 23:56:17 +00:00
|
|
|
debug_print_fis(cmd_fis, 0x80);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
switch (cmd_fis[0]) {
|
|
|
|
case SATA_FIS_TYPE_REGISTER_H2D:
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
DPRINTF(port, "unknown command cmd_fis[0]=%02x cmd_fis[1]=%02x "
|
|
|
|
"cmd_fis[2]=%02x\n", cmd_fis[0], cmd_fis[1],
|
|
|
|
cmd_fis[2]);
|
|
|
|
goto out;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2014-11-03 23:56:18 +00:00
|
|
|
if (cmd_fis[1] & 0x0F) {
|
|
|
|
DPRINTF(port, "Port Multiplier not supported."
|
|
|
|
" cmd_fis[0]=%02x cmd_fis[1]=%02x cmd_fis[2]=%02x\n",
|
|
|
|
cmd_fis[0], cmd_fis[1], cmd_fis[2]);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cmd_fis[1] & 0x70) {
|
|
|
|
DPRINTF(port, "Reserved flags set in H2D Register FIS."
|
|
|
|
" cmd_fis[0]=%02x cmd_fis[1]=%02x cmd_fis[2]=%02x\n",
|
|
|
|
cmd_fis[0], cmd_fis[1], cmd_fis[2]);
|
|
|
|
goto out;
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
ahci: Fix FIS decomposition
This patch introduces a few changes to how FIS packets are
deciphered in the AHCI virtual device. The summary of
changes can be grouped into two pieces:
[A] Changes to how we apply a preliminary sieve to FISes,
[B] Changes in how we internalize a decomposed FIS.
== Changes to how we apply a preliminary sieve to FISes ==
(1) Packets may now either update the Control register or
the Command register, but not both. This is according
to the SATA 3.2 specification which states:
"...the device either initiates processing of the command
indicated in the Command register or initiates processing
of the control request indicated [...] depending on the
state of the C bit in the FIS."
See SATA 3.2 section 10.5.5.4, "Reception" in the 10.5.5
"Register Host to Device FIS" section.
This change accounts for the first two regions of change
within the diff. All other changes belong to the following
changes.
== Changes in how we internalize a decomposed FIS ==
(2) Instead of trying to extract the sector number out of the
FIS from bytes 4-10 and setting it with ide_set_sector,
we set the appropriate IDEState registers and trust that
ide_get_sector can retrieve the correct sector later.
By "constructing" the sector for use with ide_set_sector,
we are duplicating the mechanisms of ide_get_sector.
This change makes the FIS decomposition more obvious.
SATA 3.2 as a specification does not make the legacy
register mapping with respect to the D2H FIS obvious.
However, SATA 3.2 section 10.5.5.1 "Register Host to
Device FIS layout" describes all of the "cmd_fis"
bytes:
0 - FIS Type (0x27)
1 - Port Multiplier Port and Command Update flag
2 - ATA Command
3 - Features_Low
4 - LBA 7:0
5 - LBA 15:8
6 - LBA 23:16
7 - Device, AKA "Drive Select."
8 - LBA 31:24
9 - LBA 39:32
10 - LBA 47:40
11 - Features_High
12 - Count Low
13 - Count High
14 - ICC
15 - Control
16-19 - Auxiliary (for NCQ, defined per-command)
Most of these registers map to existing IDEState registers
in obvious ways, especially features, select, hob_features,
and nsector (count). ICC is reserved in older specifications
but is not supported in our implementation, and remains
unused here. The Control register is not valid for a command
that is trying to update the command register and is to be
considered reserved at this point.
What is not obvious is the LBA register mappings, but SATA 1.0
can help inform of us legacy device support, see SATA 1.0 section
8.5.2 "Register - Host to Device."
LBA 7:0 - Sector Number (sector)
LBA 15:8 - Cyl Low (lcyl)
LBA 23:16 - Cyl High (hcyl)
LBA 31:24 - Sector Num Exp. (hob_sector)
LBA 39:32 - Cyl Low Exp. (hob_lcyl)
LBA 47:40 - Cyl High Exp. (hob_hcyl)
These mappings help guide which registers the FIS should be decomposed
into/towards for CHS, LBA28 and LBA48 commands.
As a note: The prior confusion that can be seen in the documentation
arises from the fact that CHS and LBA28 commands use the low nybble
of the drive select register to store LBA 27:24, whereas LNA48 commands
use the hob_sector, hob_lcyl and hob_hcyl registers as explained above.
The decomposition as it stands now will correctly decompose CHS, LBA28
and LBA48 commands into their appropriate registers where the core
IDE/ATAPI layers can deal with them correctly.
See the below point for more information.
(3) We save cmd_fis[7] as ide_state->select, which informs
decisions about if we are using LBA or CHS.
This corrects a bug in AHCI wherein we attempt to set and/or
retrieve the sector number by using ide_set_sector and
ide_get_sector, which depend on the select register to
determine if we are using LBA or CHS.
Without this adjustment, LBA48 read/writes are currently
broken. Thanks to Eniac Zheng @ HP for pointing this out.
(4) Save cmd_fis[11] as ide_state->hob_feature, as defined in SATA 3.2.
(5) For several ATA commands, the sector count register set to 0
is a magic number that means 256 sectors. For LBA48 commands,
this means 65,536 sectors. We drop the magic sector correction
here, and trust the ide core layer to handle the conversion
appropriately, in ide_cmd_lba48_transform(). As it stands,
the current AHCI code is only compliant with LBA28 commands.
By simply removing the magic, it will work with LBA28 and LBA48.
(6) We expand FIS decomposition to include both ATAPI and IDE devices.
We leave the logic of determining if the fields are valid or not
to the respective layers.
This change intends to make it clearer that AHCI is only a
composition mechanism for the FIS packets: the meanings of
the registers is best left to the implementation layers for
those devices.
(7) Forcefully setting the feature, hcyl and lcyl registers for ATAPI
commands is removed.
- The hcyl and lcyl magic present here is valid at boot only,
and should not be overridden for every PACKET command.
- The feature register is defined as valid for the PACKET command,
so we should not suppress it. The ATAPI layer does not even
currently depend on or require 0x01 as mandatory.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1415058979-16604-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-11-03 23:56:16 +00:00
|
|
|
if (!(cmd_fis[1] & SATA_FIS_REG_H2D_UPDATE_COMMAND_REGISTER)) {
|
|
|
|
switch (s->dev[port].port_state) {
|
2010-12-14 00:34:40 +00:00
|
|
|
case STATE_RUN:
|
|
|
|
if (cmd_fis[15] & ATA_SRST) {
|
|
|
|
s->dev[port].port_state = STATE_RESET;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case STATE_RESET:
|
|
|
|
if (!(cmd_fis[15] & ATA_SRST)) {
|
|
|
|
ahci_reset_port(s, port);
|
|
|
|
}
|
|
|
|
break;
|
ahci: Fix FIS decomposition
This patch introduces a few changes to how FIS packets are
deciphered in the AHCI virtual device. The summary of
changes can be grouped into two pieces:
[A] Changes to how we apply a preliminary sieve to FISes,
[B] Changes in how we internalize a decomposed FIS.
== Changes to how we apply a preliminary sieve to FISes ==
(1) Packets may now either update the Control register or
the Command register, but not both. This is according
to the SATA 3.2 specification which states:
"...the device either initiates processing of the command
indicated in the Command register or initiates processing
of the control request indicated [...] depending on the
state of the C bit in the FIS."
See SATA 3.2 section 10.5.5.4, "Reception" in the 10.5.5
"Register Host to Device FIS" section.
This change accounts for the first two regions of change
within the diff. All other changes belong to the following
changes.
== Changes in how we internalize a decomposed FIS ==
(2) Instead of trying to extract the sector number out of the
FIS from bytes 4-10 and setting it with ide_set_sector,
we set the appropriate IDEState registers and trust that
ide_get_sector can retrieve the correct sector later.
By "constructing" the sector for use with ide_set_sector,
we are duplicating the mechanisms of ide_get_sector.
This change makes the FIS decomposition more obvious.
SATA 3.2 as a specification does not make the legacy
register mapping with respect to the D2H FIS obvious.
However, SATA 3.2 section 10.5.5.1 "Register Host to
Device FIS layout" describes all of the "cmd_fis"
bytes:
0 - FIS Type (0x27)
1 - Port Multiplier Port and Command Update flag
2 - ATA Command
3 - Features_Low
4 - LBA 7:0
5 - LBA 15:8
6 - LBA 23:16
7 - Device, AKA "Drive Select."
8 - LBA 31:24
9 - LBA 39:32
10 - LBA 47:40
11 - Features_High
12 - Count Low
13 - Count High
14 - ICC
15 - Control
16-19 - Auxiliary (for NCQ, defined per-command)
Most of these registers map to existing IDEState registers
in obvious ways, especially features, select, hob_features,
and nsector (count). ICC is reserved in older specifications
but is not supported in our implementation, and remains
unused here. The Control register is not valid for a command
that is trying to update the command register and is to be
considered reserved at this point.
What is not obvious is the LBA register mappings, but SATA 1.0
can help inform of us legacy device support, see SATA 1.0 section
8.5.2 "Register - Host to Device."
LBA 7:0 - Sector Number (sector)
LBA 15:8 - Cyl Low (lcyl)
LBA 23:16 - Cyl High (hcyl)
LBA 31:24 - Sector Num Exp. (hob_sector)
LBA 39:32 - Cyl Low Exp. (hob_lcyl)
LBA 47:40 - Cyl High Exp. (hob_hcyl)
These mappings help guide which registers the FIS should be decomposed
into/towards for CHS, LBA28 and LBA48 commands.
As a note: The prior confusion that can be seen in the documentation
arises from the fact that CHS and LBA28 commands use the low nybble
of the drive select register to store LBA 27:24, whereas LNA48 commands
use the hob_sector, hob_lcyl and hob_hcyl registers as explained above.
The decomposition as it stands now will correctly decompose CHS, LBA28
and LBA48 commands into their appropriate registers where the core
IDE/ATAPI layers can deal with them correctly.
See the below point for more information.
(3) We save cmd_fis[7] as ide_state->select, which informs
decisions about if we are using LBA or CHS.
This corrects a bug in AHCI wherein we attempt to set and/or
retrieve the sector number by using ide_set_sector and
ide_get_sector, which depend on the select register to
determine if we are using LBA or CHS.
Without this adjustment, LBA48 read/writes are currently
broken. Thanks to Eniac Zheng @ HP for pointing this out.
(4) Save cmd_fis[11] as ide_state->hob_feature, as defined in SATA 3.2.
(5) For several ATA commands, the sector count register set to 0
is a magic number that means 256 sectors. For LBA48 commands,
this means 65,536 sectors. We drop the magic sector correction
here, and trust the ide core layer to handle the conversion
appropriately, in ide_cmd_lba48_transform(). As it stands,
the current AHCI code is only compliant with LBA28 commands.
By simply removing the magic, it will work with LBA28 and LBA48.
(6) We expand FIS decomposition to include both ATAPI and IDE devices.
We leave the logic of determining if the fields are valid or not
to the respective layers.
This change intends to make it clearer that AHCI is only a
composition mechanism for the FIS packets: the meanings of
the registers is best left to the implementation layers for
those devices.
(7) Forcefully setting the feature, hcyl and lcyl registers for ATAPI
commands is removed.
- The hcyl and lcyl magic present here is valid at boot only,
and should not be overridden for every PACKET command.
- The feature register is defined as valid for the PACKET command,
so we should not suppress it. The ATAPI layer does not even
currently depend on or require 0x01 as mandatory.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1415058979-16604-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-11-03 23:56:16 +00:00
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
ahci: Fix FIS decomposition
This patch introduces a few changes to how FIS packets are
deciphered in the AHCI virtual device. The summary of
changes can be grouped into two pieces:
[A] Changes to how we apply a preliminary sieve to FISes,
[B] Changes in how we internalize a decomposed FIS.
== Changes to how we apply a preliminary sieve to FISes ==
(1) Packets may now either update the Control register or
the Command register, but not both. This is according
to the SATA 3.2 specification which states:
"...the device either initiates processing of the command
indicated in the Command register or initiates processing
of the control request indicated [...] depending on the
state of the C bit in the FIS."
See SATA 3.2 section 10.5.5.4, "Reception" in the 10.5.5
"Register Host to Device FIS" section.
This change accounts for the first two regions of change
within the diff. All other changes belong to the following
changes.
== Changes in how we internalize a decomposed FIS ==
(2) Instead of trying to extract the sector number out of the
FIS from bytes 4-10 and setting it with ide_set_sector,
we set the appropriate IDEState registers and trust that
ide_get_sector can retrieve the correct sector later.
By "constructing" the sector for use with ide_set_sector,
we are duplicating the mechanisms of ide_get_sector.
This change makes the FIS decomposition more obvious.
SATA 3.2 as a specification does not make the legacy
register mapping with respect to the D2H FIS obvious.
However, SATA 3.2 section 10.5.5.1 "Register Host to
Device FIS layout" describes all of the "cmd_fis"
bytes:
0 - FIS Type (0x27)
1 - Port Multiplier Port and Command Update flag
2 - ATA Command
3 - Features_Low
4 - LBA 7:0
5 - LBA 15:8
6 - LBA 23:16
7 - Device, AKA "Drive Select."
8 - LBA 31:24
9 - LBA 39:32
10 - LBA 47:40
11 - Features_High
12 - Count Low
13 - Count High
14 - ICC
15 - Control
16-19 - Auxiliary (for NCQ, defined per-command)
Most of these registers map to existing IDEState registers
in obvious ways, especially features, select, hob_features,
and nsector (count). ICC is reserved in older specifications
but is not supported in our implementation, and remains
unused here. The Control register is not valid for a command
that is trying to update the command register and is to be
considered reserved at this point.
What is not obvious is the LBA register mappings, but SATA 1.0
can help inform of us legacy device support, see SATA 1.0 section
8.5.2 "Register - Host to Device."
LBA 7:0 - Sector Number (sector)
LBA 15:8 - Cyl Low (lcyl)
LBA 23:16 - Cyl High (hcyl)
LBA 31:24 - Sector Num Exp. (hob_sector)
LBA 39:32 - Cyl Low Exp. (hob_lcyl)
LBA 47:40 - Cyl High Exp. (hob_hcyl)
These mappings help guide which registers the FIS should be decomposed
into/towards for CHS, LBA28 and LBA48 commands.
As a note: The prior confusion that can be seen in the documentation
arises from the fact that CHS and LBA28 commands use the low nybble
of the drive select register to store LBA 27:24, whereas LNA48 commands
use the hob_sector, hob_lcyl and hob_hcyl registers as explained above.
The decomposition as it stands now will correctly decompose CHS, LBA28
and LBA48 commands into their appropriate registers where the core
IDE/ATAPI layers can deal with them correctly.
See the below point for more information.
(3) We save cmd_fis[7] as ide_state->select, which informs
decisions about if we are using LBA or CHS.
This corrects a bug in AHCI wherein we attempt to set and/or
retrieve the sector number by using ide_set_sector and
ide_get_sector, which depend on the select register to
determine if we are using LBA or CHS.
Without this adjustment, LBA48 read/writes are currently
broken. Thanks to Eniac Zheng @ HP for pointing this out.
(4) Save cmd_fis[11] as ide_state->hob_feature, as defined in SATA 3.2.
(5) For several ATA commands, the sector count register set to 0
is a magic number that means 256 sectors. For LBA48 commands,
this means 65,536 sectors. We drop the magic sector correction
here, and trust the ide core layer to handle the conversion
appropriately, in ide_cmd_lba48_transform(). As it stands,
the current AHCI code is only compliant with LBA28 commands.
By simply removing the magic, it will work with LBA28 and LBA48.
(6) We expand FIS decomposition to include both ATAPI and IDE devices.
We leave the logic of determining if the fields are valid or not
to the respective layers.
This change intends to make it clearer that AHCI is only a
composition mechanism for the FIS packets: the meanings of
the registers is best left to the implementation layers for
those devices.
(7) Forcefully setting the feature, hcyl and lcyl registers for ATAPI
commands is removed.
- The hcyl and lcyl magic present here is valid at boot only,
and should not be overridden for every PACKET command.
- The feature register is defined as valid for the PACKET command,
so we should not suppress it. The ATAPI layer does not even
currently depend on or require 0x01 as mandatory.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1415058979-16604-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-11-03 23:56:16 +00:00
|
|
|
else if (cmd_fis[1] & SATA_FIS_REG_H2D_UPDATE_COMMAND_REGISTER) {
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
/* Check for NCQ command */
|
2014-11-03 23:56:15 +00:00
|
|
|
if (is_ncq(cmd_fis[2])) {
|
2010-12-14 00:34:40 +00:00
|
|
|
process_ncq_command(s, port, cmd_fis, slot);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
ahci: Fix FIS decomposition
This patch introduces a few changes to how FIS packets are
deciphered in the AHCI virtual device. The summary of
changes can be grouped into two pieces:
[A] Changes to how we apply a preliminary sieve to FISes,
[B] Changes in how we internalize a decomposed FIS.
== Changes to how we apply a preliminary sieve to FISes ==
(1) Packets may now either update the Control register or
the Command register, but not both. This is according
to the SATA 3.2 specification which states:
"...the device either initiates processing of the command
indicated in the Command register or initiates processing
of the control request indicated [...] depending on the
state of the C bit in the FIS."
See SATA 3.2 section 10.5.5.4, "Reception" in the 10.5.5
"Register Host to Device FIS" section.
This change accounts for the first two regions of change
within the diff. All other changes belong to the following
changes.
== Changes in how we internalize a decomposed FIS ==
(2) Instead of trying to extract the sector number out of the
FIS from bytes 4-10 and setting it with ide_set_sector,
we set the appropriate IDEState registers and trust that
ide_get_sector can retrieve the correct sector later.
By "constructing" the sector for use with ide_set_sector,
we are duplicating the mechanisms of ide_get_sector.
This change makes the FIS decomposition more obvious.
SATA 3.2 as a specification does not make the legacy
register mapping with respect to the D2H FIS obvious.
However, SATA 3.2 section 10.5.5.1 "Register Host to
Device FIS layout" describes all of the "cmd_fis"
bytes:
0 - FIS Type (0x27)
1 - Port Multiplier Port and Command Update flag
2 - ATA Command
3 - Features_Low
4 - LBA 7:0
5 - LBA 15:8
6 - LBA 23:16
7 - Device, AKA "Drive Select."
8 - LBA 31:24
9 - LBA 39:32
10 - LBA 47:40
11 - Features_High
12 - Count Low
13 - Count High
14 - ICC
15 - Control
16-19 - Auxiliary (for NCQ, defined per-command)
Most of these registers map to existing IDEState registers
in obvious ways, especially features, select, hob_features,
and nsector (count). ICC is reserved in older specifications
but is not supported in our implementation, and remains
unused here. The Control register is not valid for a command
that is trying to update the command register and is to be
considered reserved at this point.
What is not obvious is the LBA register mappings, but SATA 1.0
can help inform of us legacy device support, see SATA 1.0 section
8.5.2 "Register - Host to Device."
LBA 7:0 - Sector Number (sector)
LBA 15:8 - Cyl Low (lcyl)
LBA 23:16 - Cyl High (hcyl)
LBA 31:24 - Sector Num Exp. (hob_sector)
LBA 39:32 - Cyl Low Exp. (hob_lcyl)
LBA 47:40 - Cyl High Exp. (hob_hcyl)
These mappings help guide which registers the FIS should be decomposed
into/towards for CHS, LBA28 and LBA48 commands.
As a note: The prior confusion that can be seen in the documentation
arises from the fact that CHS and LBA28 commands use the low nybble
of the drive select register to store LBA 27:24, whereas LNA48 commands
use the hob_sector, hob_lcyl and hob_hcyl registers as explained above.
The decomposition as it stands now will correctly decompose CHS, LBA28
and LBA48 commands into their appropriate registers where the core
IDE/ATAPI layers can deal with them correctly.
See the below point for more information.
(3) We save cmd_fis[7] as ide_state->select, which informs
decisions about if we are using LBA or CHS.
This corrects a bug in AHCI wherein we attempt to set and/or
retrieve the sector number by using ide_set_sector and
ide_get_sector, which depend on the select register to
determine if we are using LBA or CHS.
Without this adjustment, LBA48 read/writes are currently
broken. Thanks to Eniac Zheng @ HP for pointing this out.
(4) Save cmd_fis[11] as ide_state->hob_feature, as defined in SATA 3.2.
(5) For several ATA commands, the sector count register set to 0
is a magic number that means 256 sectors. For LBA48 commands,
this means 65,536 sectors. We drop the magic sector correction
here, and trust the ide core layer to handle the conversion
appropriately, in ide_cmd_lba48_transform(). As it stands,
the current AHCI code is only compliant with LBA28 commands.
By simply removing the magic, it will work with LBA28 and LBA48.
(6) We expand FIS decomposition to include both ATAPI and IDE devices.
We leave the logic of determining if the fields are valid or not
to the respective layers.
This change intends to make it clearer that AHCI is only a
composition mechanism for the FIS packets: the meanings of
the registers is best left to the implementation layers for
those devices.
(7) Forcefully setting the feature, hcyl and lcyl registers for ATAPI
commands is removed.
- The hcyl and lcyl magic present here is valid at boot only,
and should not be overridden for every PACKET command.
- The feature register is defined as valid for the PACKET command,
so we should not suppress it. The ATAPI layer does not even
currently depend on or require 0x01 as mandatory.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1415058979-16604-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-11-03 23:56:16 +00:00
|
|
|
/* Decompose the FIS:
|
|
|
|
* AHCI does not interpret FIS packets, it only forwards them.
|
|
|
|
* SATA 1.0 describes how to decode LBA28 and CHS FIS packets.
|
|
|
|
* Later specifications, e.g, SATA 3.2, describe LBA48 FIS packets.
|
|
|
|
*
|
|
|
|
* ATA4 describes sector number for LBA28/CHS commands.
|
|
|
|
* ATA6 describes sector number for LBA48 commands.
|
|
|
|
* ATA8 deprecates CHS fully, describing only LBA28/48.
|
|
|
|
*
|
|
|
|
* We dutifully convert the FIS into IDE registers, and allow the
|
|
|
|
* core layer to interpret them as needed. */
|
2010-12-14 00:34:40 +00:00
|
|
|
ide_state->feature = cmd_fis[3];
|
ahci: Fix FIS decomposition
This patch introduces a few changes to how FIS packets are
deciphered in the AHCI virtual device. The summary of
changes can be grouped into two pieces:
[A] Changes to how we apply a preliminary sieve to FISes,
[B] Changes in how we internalize a decomposed FIS.
== Changes to how we apply a preliminary sieve to FISes ==
(1) Packets may now either update the Control register or
the Command register, but not both. This is according
to the SATA 3.2 specification which states:
"...the device either initiates processing of the command
indicated in the Command register or initiates processing
of the control request indicated [...] depending on the
state of the C bit in the FIS."
See SATA 3.2 section 10.5.5.4, "Reception" in the 10.5.5
"Register Host to Device FIS" section.
This change accounts for the first two regions of change
within the diff. All other changes belong to the following
changes.
== Changes in how we internalize a decomposed FIS ==
(2) Instead of trying to extract the sector number out of the
FIS from bytes 4-10 and setting it with ide_set_sector,
we set the appropriate IDEState registers and trust that
ide_get_sector can retrieve the correct sector later.
By "constructing" the sector for use with ide_set_sector,
we are duplicating the mechanisms of ide_get_sector.
This change makes the FIS decomposition more obvious.
SATA 3.2 as a specification does not make the legacy
register mapping with respect to the D2H FIS obvious.
However, SATA 3.2 section 10.5.5.1 "Register Host to
Device FIS layout" describes all of the "cmd_fis"
bytes:
0 - FIS Type (0x27)
1 - Port Multiplier Port and Command Update flag
2 - ATA Command
3 - Features_Low
4 - LBA 7:0
5 - LBA 15:8
6 - LBA 23:16
7 - Device, AKA "Drive Select."
8 - LBA 31:24
9 - LBA 39:32
10 - LBA 47:40
11 - Features_High
12 - Count Low
13 - Count High
14 - ICC
15 - Control
16-19 - Auxiliary (for NCQ, defined per-command)
Most of these registers map to existing IDEState registers
in obvious ways, especially features, select, hob_features,
and nsector (count). ICC is reserved in older specifications
but is not supported in our implementation, and remains
unused here. The Control register is not valid for a command
that is trying to update the command register and is to be
considered reserved at this point.
What is not obvious is the LBA register mappings, but SATA 1.0
can help inform of us legacy device support, see SATA 1.0 section
8.5.2 "Register - Host to Device."
LBA 7:0 - Sector Number (sector)
LBA 15:8 - Cyl Low (lcyl)
LBA 23:16 - Cyl High (hcyl)
LBA 31:24 - Sector Num Exp. (hob_sector)
LBA 39:32 - Cyl Low Exp. (hob_lcyl)
LBA 47:40 - Cyl High Exp. (hob_hcyl)
These mappings help guide which registers the FIS should be decomposed
into/towards for CHS, LBA28 and LBA48 commands.
As a note: The prior confusion that can be seen in the documentation
arises from the fact that CHS and LBA28 commands use the low nybble
of the drive select register to store LBA 27:24, whereas LNA48 commands
use the hob_sector, hob_lcyl and hob_hcyl registers as explained above.
The decomposition as it stands now will correctly decompose CHS, LBA28
and LBA48 commands into their appropriate registers where the core
IDE/ATAPI layers can deal with them correctly.
See the below point for more information.
(3) We save cmd_fis[7] as ide_state->select, which informs
decisions about if we are using LBA or CHS.
This corrects a bug in AHCI wherein we attempt to set and/or
retrieve the sector number by using ide_set_sector and
ide_get_sector, which depend on the select register to
determine if we are using LBA or CHS.
Without this adjustment, LBA48 read/writes are currently
broken. Thanks to Eniac Zheng @ HP for pointing this out.
(4) Save cmd_fis[11] as ide_state->hob_feature, as defined in SATA 3.2.
(5) For several ATA commands, the sector count register set to 0
is a magic number that means 256 sectors. For LBA48 commands,
this means 65,536 sectors. We drop the magic sector correction
here, and trust the ide core layer to handle the conversion
appropriately, in ide_cmd_lba48_transform(). As it stands,
the current AHCI code is only compliant with LBA28 commands.
By simply removing the magic, it will work with LBA28 and LBA48.
(6) We expand FIS decomposition to include both ATAPI and IDE devices.
We leave the logic of determining if the fields are valid or not
to the respective layers.
This change intends to make it clearer that AHCI is only a
composition mechanism for the FIS packets: the meanings of
the registers is best left to the implementation layers for
those devices.
(7) Forcefully setting the feature, hcyl and lcyl registers for ATAPI
commands is removed.
- The hcyl and lcyl magic present here is valid at boot only,
and should not be overridden for every PACKET command.
- The feature register is defined as valid for the PACKET command,
so we should not suppress it. The ATAPI layer does not even
currently depend on or require 0x01 as mandatory.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1415058979-16604-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-11-03 23:56:16 +00:00
|
|
|
ide_state->sector = cmd_fis[4]; /* LBA 7:0 */
|
|
|
|
ide_state->lcyl = cmd_fis[5]; /* LBA 15:8 */
|
|
|
|
ide_state->hcyl = cmd_fis[6]; /* LBA 23:16 */
|
|
|
|
ide_state->select = cmd_fis[7]; /* LBA 27:24 (LBA28) */
|
|
|
|
ide_state->hob_sector = cmd_fis[8]; /* LBA 31:24 */
|
|
|
|
ide_state->hob_lcyl = cmd_fis[9]; /* LBA 39:32 */
|
|
|
|
ide_state->hob_hcyl = cmd_fis[10]; /* LBA 47:40 */
|
|
|
|
ide_state->hob_feature = cmd_fis[11];
|
|
|
|
ide_state->nsector = (int64_t)((cmd_fis[13] << 8) | cmd_fis[12]);
|
|
|
|
/* 14, 16, 17, 18, 19: Reserved (SATA 1.0) */
|
|
|
|
/* 15: Only valid when UPDATE_COMMAND not set. */
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
/* Copy the ACMD field (ATAPI packet, if any) from the AHCI command
|
|
|
|
* table to ide_state->io_buffer
|
|
|
|
*/
|
|
|
|
if (opts & AHCI_CMD_ATAPI) {
|
|
|
|
memcpy(ide_state->io_buffer, &cmd_fis[AHCI_COMMAND_TABLE_ACMD], 0x10);
|
|
|
|
debug_print_fis(ide_state->io_buffer, 0x10);
|
2013-01-15 15:12:09 +00:00
|
|
|
s->dev[port].done_atapi_packet = false;
|
2010-12-14 00:34:40 +00:00
|
|
|
/* XXX send PIO setup FIS */
|
|
|
|
}
|
|
|
|
|
|
|
|
ide_state->error = 0;
|
|
|
|
|
|
|
|
/* Reset transferred byte counter */
|
|
|
|
cmd->status = 0;
|
|
|
|
|
|
|
|
/* We're ready to process the command in FIS byte 2. */
|
|
|
|
ide_exec_cmd(&s->dev[port].port, cmd_fis[2]);
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2013-04-10 16:15:49 +00:00
|
|
|
dma_memory_unmap(s->as, cmd_fis, cmd_len, DMA_DIRECTION_FROM_DEVICE,
|
2012-06-27 04:50:41 +00:00
|
|
|
cmd_len);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
if (s->dev[port].port.ifs[0].status & (BUSY_STAT|DRQ_STAT)) {
|
|
|
|
/* async command, complete later */
|
|
|
|
s->dev[port].busy_slot = slot;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* done handling the command */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* DMA dev <-> ram */
|
2014-08-04 21:11:09 +00:00
|
|
|
static void ahci_start_transfer(IDEDMA *dma)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
|
|
|
IDEState *s = &ad->port.ifs[0];
|
|
|
|
uint32_t size = (uint32_t)(s->data_end - s->data_ptr);
|
|
|
|
/* write == ram -> device */
|
|
|
|
uint32_t opts = le32_to_cpu(ad->cur_cmd->opts);
|
|
|
|
int is_write = opts & AHCI_CMD_WRITE;
|
|
|
|
int is_atapi = opts & AHCI_CMD_ATAPI;
|
|
|
|
int has_sglist = 0;
|
|
|
|
|
|
|
|
if (is_atapi && !ad->done_atapi_packet) {
|
|
|
|
/* already prepopulated iobuffer */
|
2013-01-15 15:12:09 +00:00
|
|
|
ad->done_atapi_packet = true;
|
2014-11-11 00:41:40 +00:00
|
|
|
size = 0;
|
2010-12-14 00:34:40 +00:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
ahci: unify sglist preparation
The intent of this patch is to further unify the creation and
deletion of the sglist used for all AHCI transfers, including
emulated PIO, ATAPI R/W, and native DMA R/W.
By replacing ahci_start_transfer's call to ahci_populate_sglist
with ahci_dma_prepare_buf, we reduce the number of direct calls
where we manipulate the scatter-gather list in the AHCI code.
To make this switch, the constant "0" passed as an offset
in ahci_dma_prepare_buf is adjusted to use io_buffer_offset.
For DMA pathways, this has no effect: io_buffer_offset is always
updated to 0 at the beginning of a DMA transfer loop regardless.
DMA pathways through ide_dma_cb() update the io_buffer_offset
accordingly, and for circumstances where we might make several
trips through this loop, this may actually correct a design flaw.
For PIO pathways, the newly updated ahci_dma_prepare_buf will
now prepare the sglist at the correct offset. It will also set
io_buffer_size, but this is not used in the cmd_read_pio or
cmd_write_pio pathways.
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-3-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:38 +00:00
|
|
|
if (ahci_dma_prepare_buf(dma, is_write)) {
|
2010-12-14 00:34:40 +00:00
|
|
|
has_sglist = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
DPRINTF(ad->port_no, "%sing %d bytes on %s w/%s sglist\n",
|
|
|
|
is_write ? "writ" : "read", size, is_atapi ? "atapi" : "ata",
|
|
|
|
has_sglist ? "" : "o");
|
|
|
|
|
2011-12-15 13:32:04 +00:00
|
|
|
if (has_sglist && size) {
|
|
|
|
if (is_write) {
|
|
|
|
dma_buf_write(s->data_ptr, size, &s->sg);
|
|
|
|
} else {
|
|
|
|
dma_buf_read(s->data_ptr, size, &s->sg);
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
|
|
|
/* declare that we processed everything */
|
|
|
|
s->data_ptr = s->data_end;
|
|
|
|
|
2014-10-01 22:55:47 +00:00
|
|
|
/* Update number of transferred bytes, destroy sglist */
|
|
|
|
ahci_commit_buf(dma, size);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
s->end_transfer_func(s);
|
2014-08-04 21:11:18 +00:00
|
|
|
|
|
|
|
if (!(s->status & DRQ_STAT)) {
|
|
|
|
/* done with PIO send/receive */
|
|
|
|
ahci_write_fis_pio(ad, le32_to_cpu(ad->cur_cmd->status));
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_start_dma(IDEDMA *dma, IDEState *s,
|
2014-10-07 11:59:15 +00:00
|
|
|
BlockCompletionFunc *dma_cb)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
2013-01-04 19:44:41 +00:00
|
|
|
#ifdef DEBUG_AHCI
|
2010-12-14 00:34:40 +00:00
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
2013-01-04 19:44:41 +00:00
|
|
|
#endif
|
2010-12-14 00:34:40 +00:00
|
|
|
DPRINTF(ad->port_no, "\n");
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
s->io_buffer_offset = 0;
|
2010-12-14 00:34:40 +00:00
|
|
|
dma_cb(s, 0);
|
|
|
|
}
|
|
|
|
|
2014-10-01 22:55:47 +00:00
|
|
|
/**
|
|
|
|
* Called in DMA R/W chains to read the PRDT, utilizing ahci_populate_sglist.
|
|
|
|
* Not currently invoked by PIO R/W chains,
|
|
|
|
* which invoke ahci_populate_sglist via ahci_start_transfer.
|
|
|
|
*/
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
static int32_t ahci_dma_prepare_buf(IDEDMA *dma, int is_write)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
|
|
|
IDEState *s = &ad->port.ifs[0];
|
|
|
|
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
if (ahci_populate_sglist(ad, &s->sg, s->io_buffer_offset) == -1) {
|
|
|
|
DPRINTF(ad->port_no, "ahci_dma_prepare_buf failed.\n");
|
|
|
|
return -1;
|
|
|
|
}
|
2011-12-15 13:32:04 +00:00
|
|
|
s->io_buffer_size = s->sg.size;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
DPRINTF(ad->port_no, "len=%#x\n", s->io_buffer_size);
|
ide: Correct handling of malformed/short PRDTs
This impacts both BMDMA and AHCI HBA interfaces for IDE.
Currently, we confuse the difference between a PRDT having
"0 bytes" and a PRDT having "0 complete sectors."
When we receive an incomplete sector, inconsistent error checking
leads to an infinite loop wherein the call succeeds, but it
didn't give us enough bytes -- leading us to re-call the
DMA chain over and over again. This leads to, in the BMDMA case,
leaked memory for short PRDTs, and infinite loops and resource
usage in the AHCI case.
The .prepare_buf() callback is reworked to return the number of
bytes that it successfully prepared. 0 is a valid, non-error
answer that means the table was empty and described no bytes.
-1 indicates an error.
Our current implementation uses the io_buffer in IDEState to
ultimately describe the size of a prepared scatter-gather list.
Even though the AHCI PRDT/SGList can be as large as 256GiB, the
AHCI command header limits transactions to just 4GiB. ATA8-ACS3,
however, defines the largest transaction to be an LBA48 command
that transfers 65,536 sectors. With a 512 byte sector size, this
is just 32MiB.
Since our current state structures use the int type to describe
the size of the buffer, and this state is migrated as int32, we
are limited to describing 2GiB buffer sizes unless we change the
migration protocol.
For this reason, this patch begins to unify the assertions in the
IDE pathways that the scatter-gather list provided by either the
AHCI PRDT or the PCI BMDMA PRDs can only describe, at a maximum,
2GiB. This should be resilient enough unless we need a sector
size that exceeds 32KiB.
Further, the likelihood of any guest operating system actually
attempting to transfer this much data in a single operation is
very slim.
To this end, the IDEState variables have been updated to more
explicitly clarify our maximum supported size. Callers to the
prepare_buf callback have been reworked to understand the new
return code, and all versions of the prepare_buf callback have
been adjusted accordingly.
Lastly, the ahci_populate_sglist helper, relied upon by the
AHCI implementation of .prepare_buf() as well as the PCI
implementation of the callback have had overflow assertions
added to help make clear the reasonings behind the various
type changes.
[Added %d -> %"PRId64" fix John sent because off_pos changed from int to
int64_t.
--Stefan]
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1414785819-26209-4-git-send-email-jsnow@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2014-10-31 20:03:39 +00:00
|
|
|
return s->io_buffer_size;
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
2014-10-01 22:55:47 +00:00
|
|
|
/**
|
|
|
|
* Destroys the scatter-gather list,
|
|
|
|
* and updates the command header with a bytes-read value.
|
|
|
|
* called explicitly via ahci_dma_rw_buf (ATAPI DMA),
|
|
|
|
* and ahci_start_transfer (PIO R/W),
|
|
|
|
* and called via callback from ide_dma_cb for DMA R/W paths.
|
|
|
|
*/
|
|
|
|
static void ahci_commit_buf(IDEDMA *dma, uint32_t tx_bytes)
|
|
|
|
{
|
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
|
|
|
IDEState *s = &ad->port.ifs[0];
|
|
|
|
|
|
|
|
tx_bytes += le32_to_cpu(ad->cur_cmd->status);
|
|
|
|
ad->cur_cmd->status = cpu_to_le32(tx_bytes);
|
|
|
|
|
|
|
|
qemu_sglist_destroy(&s->sg);
|
|
|
|
}
|
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
static int ahci_dma_rw_buf(IDEDMA *dma, int is_write)
|
|
|
|
{
|
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
|
|
|
IDEState *s = &ad->port.ifs[0];
|
|
|
|
uint8_t *p = s->io_buffer + s->io_buffer_index;
|
|
|
|
int l = s->io_buffer_size - s->io_buffer_index;
|
|
|
|
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
if (ahci_populate_sglist(ad, &s->sg, s->io_buffer_offset)) {
|
2010-12-14 00:34:40 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (is_write) {
|
2011-12-15 13:32:04 +00:00
|
|
|
dma_buf_read(p, l, &s->sg);
|
2010-12-14 00:34:40 +00:00
|
|
|
} else {
|
2011-12-15 13:32:04 +00:00
|
|
|
dma_buf_write(p, l, &s->sg);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
2014-10-01 22:55:47 +00:00
|
|
|
/* free sglist, update byte count */
|
|
|
|
ahci_commit_buf(dma, l);
|
2012-08-03 19:57:10 +00:00
|
|
|
|
2010-12-14 00:34:40 +00:00
|
|
|
s->io_buffer_index += l;
|
ahci: Fix ahci cdrom read corruptions for reads > 128k
While testing q35, which has its cdrom attached to the ahci controller, I found
that the Fedora 17 install would panic on boot. The panic occurs while
squashfs is trying to read from the cdrom. The errors are:
[ 8.622711] SQUASHFS error: xz_dec_run error, data probably corrupt
[ 8.625180] SQUASHFS error: squashfs_read_data failed to read block
0x20be48a
I was also able to produce corrupt data reads using an installed piix based
qemu machine, using 'dd'. I found that the corruptions were only occuring when
then read size was greater than 128k. For example, the following command
results in corrupted reads:
dd if=/dev/sr0 of=/tmp/blah bs=256k iflag=direct
The > 128k size reads exercise a different code path than 128k and below. In
ide_atapi_cmd_read_dma_cb() s->io_buffer_size is capped at 128k. Thus,
ide_atapi_cmd_read_dma_cb() is called a second time when the read is > 128k.
However, ahci_dma_rw_buf() restart the read from offset 0, instead of at 128k.
Thus, resulting in a corrupted read.
To fix this, I've introduced 'io_buffer_offset' field in IDEState to keep
track of the offset. I've also modified ahci_populate_sglist() to take a new
3rd offset argument, so that the sglist is property initialized.
I've tested this patch using 'dd' testing, and Fedora 17 now correctly boots
and installs on q35 with the cdrom ahci controller.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-08-03 19:57:06 +00:00
|
|
|
s->io_buffer_offset += l;
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
DPRINTF(ad->port_no, "len=%#x\n", l);
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int ahci_dma_set_unit(IDEDMA *dma, int unit)
|
|
|
|
{
|
|
|
|
/* only a single unit per link */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-08-04 21:11:17 +00:00
|
|
|
static void ahci_cmd_done(IDEDMA *dma)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
AHCIDevice *ad = DO_UPCAST(AHCIDevice, dma, dma);
|
|
|
|
|
2014-08-04 21:11:17 +00:00
|
|
|
DPRINTF(ad->port_no, "cmd done\n");
|
2010-12-14 00:34:40 +00:00
|
|
|
|
|
|
|
/* update d2h status */
|
|
|
|
ahci_write_fis_d2h(ad, NULL);
|
|
|
|
|
2011-05-09 15:48:19 +00:00
|
|
|
if (!ad->check_bh) {
|
|
|
|
/* maybe we still have something to process, check later */
|
|
|
|
ad->check_bh = qemu_bh_new(ahci_check_cmd_bh, ad);
|
|
|
|
qemu_bh_schedule(ad->check_bh);
|
|
|
|
}
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void ahci_irq_set(void *opaque, int n, int level)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2011-07-29 17:26:33 +00:00
|
|
|
static void ahci_dma_restart_cb(void *opaque, int running, RunState state)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static const IDEDMAOps ahci_dma_ops = {
|
|
|
|
.start_dma = ahci_start_dma,
|
|
|
|
.start_transfer = ahci_start_transfer,
|
|
|
|
.prepare_buf = ahci_dma_prepare_buf,
|
2014-10-01 22:55:47 +00:00
|
|
|
.commit_buf = ahci_commit_buf,
|
2010-12-14 00:34:40 +00:00
|
|
|
.rw_buf = ahci_dma_rw_buf,
|
|
|
|
.set_unit = ahci_dma_set_unit,
|
2014-08-04 21:11:17 +00:00
|
|
|
.cmd_done = ahci_cmd_done,
|
2010-12-14 00:34:40 +00:00
|
|
|
.restart_cb = ahci_dma_restart_cb,
|
|
|
|
};
|
|
|
|
|
2013-04-10 16:15:49 +00:00
|
|
|
void ahci_init(AHCIState *s, DeviceState *qdev, AddressSpace *as, int ports)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
|
|
|
qemu_irq *irqs;
|
|
|
|
int i;
|
|
|
|
|
2013-04-10 16:15:49 +00:00
|
|
|
s->as = as;
|
2011-02-01 14:51:31 +00:00
|
|
|
s->ports = ports;
|
block: Use g_new() & friends where that makes obvious sense
g_new(T, n) is neater than g_malloc(sizeof(T) * n). It's also safer,
for two reasons. One, it catches multiplication overflowing size_t.
Two, it returns T * rather than void *, which lets the compiler catch
more type errors.
Patch created with Coccinelle, with two manual changes on top:
* Add const to bdrv_iterate_format() to keep the types straight
* Convert the allocation in bdrv_drop_intermediate(), which Coccinelle
inexplicably misses
Coccinelle semantic patch:
@@
type T;
@@
-g_malloc(sizeof(T))
+g_new(T, 1)
@@
type T;
@@
-g_try_malloc(sizeof(T))
+g_try_new(T, 1)
@@
type T;
@@
-g_malloc0(sizeof(T))
+g_new0(T, 1)
@@
type T;
@@
-g_try_malloc0(sizeof(T))
+g_try_new0(T, 1)
@@
type T;
expression n;
@@
-g_malloc(sizeof(T) * (n))
+g_new(T, n)
@@
type T;
expression n;
@@
-g_try_malloc(sizeof(T) * (n))
+g_try_new(T, n)
@@
type T;
expression n;
@@
-g_malloc0(sizeof(T) * (n))
+g_new0(T, n)
@@
type T;
expression n;
@@
-g_try_malloc0(sizeof(T) * (n))
+g_try_new0(T, n)
@@
type T;
expression p, n;
@@
-g_realloc(p, sizeof(T) * (n))
+g_renew(T, p, n)
@@
type T;
expression p, n;
@@
-g_try_realloc(p, sizeof(T) * (n))
+g_try_renew(T, p, n)
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2014-08-19 08:31:08 +00:00
|
|
|
s->dev = g_new0(AHCIDevice, ports);
|
2010-12-14 00:34:40 +00:00
|
|
|
ahci_reg_init(s);
|
2011-08-08 13:09:14 +00:00
|
|
|
/* XXX BAR size should be 1k, but that breaks, so bump it to 4k for now */
|
2013-06-07 01:25:08 +00:00
|
|
|
memory_region_init_io(&s->mem, OBJECT(qdev), &ahci_mem_ops, s,
|
|
|
|
"ahci", AHCI_MEM_BAR_SIZE);
|
|
|
|
memory_region_init_io(&s->idp, OBJECT(qdev), &ahci_idp_ops, s,
|
|
|
|
"ahci-idp", 32);
|
2011-08-27 09:12:28 +00:00
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
irqs = qemu_allocate_irqs(ahci_irq_set, s, s->ports);
|
2010-12-14 00:34:40 +00:00
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
for (i = 0; i < s->ports; i++) {
|
2010-12-14 00:34:40 +00:00
|
|
|
AHCIDevice *ad = &s->dev[i];
|
|
|
|
|
2013-08-23 18:18:50 +00:00
|
|
|
ide_bus_new(&ad->port, sizeof(ad->port), qdev, i, 1);
|
2010-12-14 00:34:40 +00:00
|
|
|
ide_init2(&ad->port, irqs[i]);
|
|
|
|
|
|
|
|
ad->hba = s;
|
|
|
|
ad->port_no = i;
|
|
|
|
ad->port.dma = &ad->dma;
|
|
|
|
ad->port.dma->ops = &ahci_dma_ops;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-02-01 14:51:31 +00:00
|
|
|
void ahci_uninit(AHCIState *s)
|
|
|
|
{
|
2011-08-21 03:09:37 +00:00
|
|
|
g_free(s->dev);
|
2011-02-01 14:51:31 +00:00
|
|
|
}
|
|
|
|
|
2012-05-11 14:42:36 +00:00
|
|
|
void ahci_reset(AHCIState *s)
|
2010-12-14 00:34:40 +00:00
|
|
|
{
|
2011-09-12 08:19:25 +00:00
|
|
|
AHCIPortRegs *pr;
|
2010-12-14 00:34:40 +00:00
|
|
|
int i;
|
|
|
|
|
2012-05-11 14:42:36 +00:00
|
|
|
s->control_regs.irqstatus = 0;
|
2013-09-28 21:09:35 +00:00
|
|
|
/* AHCI Enable (AE)
|
|
|
|
* The implementation of this bit is dependent upon the value of the
|
|
|
|
* CAP.SAM bit. If CAP.SAM is '0', then GHC.AE shall be read-write and
|
|
|
|
* shall have a reset value of '0'. If CAP.SAM is '1', then AE shall be
|
|
|
|
* read-only and shall have a reset value of '1'.
|
|
|
|
*
|
|
|
|
* We set HOST_CAP_AHCI so we must enable AHCI at reset.
|
|
|
|
*/
|
|
|
|
s->control_regs.ghc = HOST_CTL_AHCI_EN;
|
2011-02-01 14:51:30 +00:00
|
|
|
|
2012-05-11 14:42:36 +00:00
|
|
|
for (i = 0; i < s->ports; i++) {
|
|
|
|
pr = &s->dev[i].port_regs;
|
2011-09-12 08:19:25 +00:00
|
|
|
pr->irq_stat = 0;
|
|
|
|
pr->irq_mask = 0;
|
|
|
|
pr->scr_ctl = 0;
|
ahci: properly reset PxCMD on HBA reset
While testing q35, I found that windows 7 (specifically, windows 7 ultimate
with sp1 x64), wouldn't install because it can't find the cdrom or disk drive.
The failure message is: 'A required cd/dvd device driver is missing. If you
have a driver floppy disk, CD, DVD, or USB flash drive, please insert it now.'
This can also be reproduced on piix by adding an ahci controller, and
observing that windows 7 does not see any devices behind it.
The problem is that when windows issues a HBA reset, qemu does not reset the
individual ports' PxCMD register. Windows 7 then reads back the PxCMD register
and presumably assumes that the ahci controller has already been initialized.
Windows then never sets up the PxIE register to enable interrupts, and thus it
never gets irqs back when it sends ata device inquiry commands.
This change brings qemu into ahci 1.3 specification compliance.
Section 10.4.3 HBA Reset:
"
When GHC.HR is set to '1', GHC.AE, GHC.IE, the IS register, and all port
register fields (except PxFB/PxFBU/PxCLB/PxCLBU) that are not HwInit in the
HBA's register memory space are reset.
"
I've also re-tested Fedora 16 and 17 to verify that they continue to work with
this change.
Signed-off-by: Jason Baron <jbaron@redhat.com>
Acked-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2012-09-04 20:08:08 +00:00
|
|
|
pr->cmd = PORT_CMD_SPIN_UP | PORT_CMD_POWER_ON;
|
2012-05-11 14:42:36 +00:00
|
|
|
ahci_reset_port(s, i);
|
2010-12-14 00:34:40 +00:00
|
|
|
}
|
|
|
|
}
|
2012-01-26 11:43:47 +00:00
|
|
|
|
2013-01-04 19:44:42 +00:00
|
|
|
static const VMStateDescription vmstate_ahci_device = {
|
|
|
|
.name = "ahci port",
|
|
|
|
.version_id = 1,
|
2014-04-16 13:32:32 +00:00
|
|
|
.fields = (VMStateField[]) {
|
2013-01-04 19:44:42 +00:00
|
|
|
VMSTATE_IDE_BUS(port, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_state, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(finished, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.lst_addr, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.lst_addr_hi, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.fis_addr, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.fis_addr_hi, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.irq_stat, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.irq_mask, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.cmd, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.tfdata, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.sig, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.scr_stat, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.scr_ctl, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.scr_err, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.scr_act, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(port_regs.cmd_issue, AHCIDevice),
|
|
|
|
VMSTATE_BOOL(done_atapi_packet, AHCIDevice),
|
|
|
|
VMSTATE_INT32(busy_slot, AHCIDevice),
|
|
|
|
VMSTATE_BOOL(init_d2h_sent, AHCIDevice),
|
|
|
|
VMSTATE_END_OF_LIST()
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static int ahci_state_post_load(void *opaque, int version_id)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct AHCIDevice *ad;
|
|
|
|
AHCIState *s = opaque;
|
|
|
|
|
|
|
|
for (i = 0; i < s->ports; i++) {
|
|
|
|
ad = &s->dev[i];
|
|
|
|
AHCIPortRegs *pr = &ad->port_regs;
|
|
|
|
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &ad->lst,
|
2013-01-04 19:44:42 +00:00
|
|
|
((uint64_t)pr->lst_addr_hi << 32) | pr->lst_addr, 1024);
|
2014-07-03 08:26:27 +00:00
|
|
|
map_page(s->as, &ad->res_fis,
|
2013-01-04 19:44:42 +00:00
|
|
|
((uint64_t)pr->fis_addr_hi << 32) | pr->fis_addr, 256);
|
|
|
|
/*
|
|
|
|
* All pending i/o should be flushed out on a migrate. However,
|
|
|
|
* we might not have cleared the busy_slot since this is done
|
|
|
|
* in a bh. Also, issue i/o against any slots that are pending.
|
|
|
|
*/
|
|
|
|
if ((ad->busy_slot != -1) &&
|
|
|
|
!(ad->port.ifs[0].status & (BUSY_STAT|DRQ_STAT))) {
|
|
|
|
pr->cmd_issue &= ~(1 << ad->busy_slot);
|
|
|
|
ad->busy_slot = -1;
|
|
|
|
}
|
|
|
|
check_cmd(s, i);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
const VMStateDescription vmstate_ahci = {
|
|
|
|
.name = "ahci",
|
|
|
|
.version_id = 1,
|
|
|
|
.post_load = ahci_state_post_load,
|
2014-04-16 13:32:32 +00:00
|
|
|
.fields = (VMStateField[]) {
|
2013-01-04 19:44:42 +00:00
|
|
|
VMSTATE_STRUCT_VARRAY_POINTER_INT32(dev, AHCIState, ports,
|
|
|
|
vmstate_ahci_device, AHCIDevice),
|
|
|
|
VMSTATE_UINT32(control_regs.cap, AHCIState),
|
|
|
|
VMSTATE_UINT32(control_regs.ghc, AHCIState),
|
|
|
|
VMSTATE_UINT32(control_regs.irqstatus, AHCIState),
|
|
|
|
VMSTATE_UINT32(control_regs.impl, AHCIState),
|
|
|
|
VMSTATE_UINT32(control_regs.version, AHCIState),
|
|
|
|
VMSTATE_UINT32(idp_index, AHCIState),
|
2014-04-03 16:51:18 +00:00
|
|
|
VMSTATE_INT32_EQUAL(ports, AHCIState),
|
2013-01-04 19:44:42 +00:00
|
|
|
VMSTATE_END_OF_LIST()
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
2013-07-01 10:18:30 +00:00
|
|
|
#define TYPE_SYSBUS_AHCI "sysbus-ahci"
|
|
|
|
#define SYSBUS_AHCI(obj) OBJECT_CHECK(SysbusAHCIState, (obj), TYPE_SYSBUS_AHCI)
|
|
|
|
|
2012-01-26 11:43:47 +00:00
|
|
|
typedef struct SysbusAHCIState {
|
2013-07-01 10:18:30 +00:00
|
|
|
/*< private >*/
|
|
|
|
SysBusDevice parent_obj;
|
|
|
|
/*< public >*/
|
|
|
|
|
2012-01-26 11:43:47 +00:00
|
|
|
AHCIState ahci;
|
|
|
|
uint32_t num_ports;
|
|
|
|
} SysbusAHCIState;
|
|
|
|
|
|
|
|
static const VMStateDescription vmstate_sysbus_ahci = {
|
|
|
|
.name = "sysbus-ahci",
|
2013-01-04 19:44:42 +00:00
|
|
|
.unmigratable = 1, /* Still buggy under I/O load */
|
2014-04-16 13:32:32 +00:00
|
|
|
.fields = (VMStateField[]) {
|
2014-03-18 19:36:13 +00:00
|
|
|
VMSTATE_AHCI(ahci, SysbusAHCIState),
|
2013-01-04 19:44:42 +00:00
|
|
|
VMSTATE_END_OF_LIST()
|
|
|
|
},
|
2012-01-26 11:43:47 +00:00
|
|
|
};
|
|
|
|
|
2012-05-11 14:42:36 +00:00
|
|
|
static void sysbus_ahci_reset(DeviceState *dev)
|
|
|
|
{
|
2013-07-01 10:18:30 +00:00
|
|
|
SysbusAHCIState *s = SYSBUS_AHCI(dev);
|
2012-05-11 14:42:36 +00:00
|
|
|
|
|
|
|
ahci_reset(&s->ahci);
|
|
|
|
}
|
|
|
|
|
2013-07-01 10:18:31 +00:00
|
|
|
static void sysbus_ahci_realize(DeviceState *dev, Error **errp)
|
2012-01-26 11:43:47 +00:00
|
|
|
{
|
2013-07-01 10:18:31 +00:00
|
|
|
SysBusDevice *sbd = SYS_BUS_DEVICE(dev);
|
2013-07-01 10:18:30 +00:00
|
|
|
SysbusAHCIState *s = SYSBUS_AHCI(dev);
|
2012-01-26 11:43:47 +00:00
|
|
|
|
2014-03-18 19:36:13 +00:00
|
|
|
ahci_init(&s->ahci, dev, &address_space_memory, s->num_ports);
|
2013-07-01 10:18:31 +00:00
|
|
|
|
|
|
|
sysbus_init_mmio(sbd, &s->ahci.mem);
|
|
|
|
sysbus_init_irq(sbd, &s->ahci.irq);
|
2012-01-26 11:43:47 +00:00
|
|
|
}
|
|
|
|
|
2011-12-08 03:34:16 +00:00
|
|
|
static Property sysbus_ahci_properties[] = {
|
|
|
|
DEFINE_PROP_UINT32("num-ports", SysbusAHCIState, num_ports, 1),
|
|
|
|
DEFINE_PROP_END_OF_LIST(),
|
|
|
|
};
|
|
|
|
|
2012-01-24 19:12:29 +00:00
|
|
|
static void sysbus_ahci_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
2011-12-08 03:34:16 +00:00
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
2012-01-24 19:12:29 +00:00
|
|
|
|
2013-07-01 10:18:31 +00:00
|
|
|
dc->realize = sysbus_ahci_realize;
|
2011-12-08 03:34:16 +00:00
|
|
|
dc->vmsd = &vmstate_sysbus_ahci;
|
|
|
|
dc->props = sysbus_ahci_properties;
|
2012-05-11 14:42:36 +00:00
|
|
|
dc->reset = sysbus_ahci_reset;
|
2013-07-29 14:17:45 +00:00
|
|
|
set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
|
2012-01-24 19:12:29 +00:00
|
|
|
}
|
|
|
|
|
2013-01-10 15:19:07 +00:00
|
|
|
static const TypeInfo sysbus_ahci_info = {
|
2013-07-01 10:18:30 +00:00
|
|
|
.name = TYPE_SYSBUS_AHCI,
|
2011-12-08 03:34:16 +00:00
|
|
|
.parent = TYPE_SYS_BUS_DEVICE,
|
|
|
|
.instance_size = sizeof(SysbusAHCIState),
|
|
|
|
.class_init = sysbus_ahci_class_init,
|
2012-01-26 11:43:47 +00:00
|
|
|
};
|
|
|
|
|
2012-02-09 14:20:55 +00:00
|
|
|
static void sysbus_ahci_register_types(void)
|
2012-01-26 11:43:47 +00:00
|
|
|
{
|
2011-12-08 03:34:16 +00:00
|
|
|
type_register_static(&sysbus_ahci_info);
|
2012-01-26 11:43:47 +00:00
|
|
|
}
|
|
|
|
|
2012-02-09 14:20:55 +00:00
|
|
|
type_init(sysbus_ahci_register_types)
|
2014-10-01 18:19:29 +00:00
|
|
|
|
|
|
|
void ahci_ide_create_devs(PCIDevice *dev, DriveInfo **hd)
|
|
|
|
{
|
|
|
|
AHCIPCIState *d = ICH_AHCI(dev);
|
|
|
|
AHCIState *ahci = &d->ahci;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < ahci->ports; i++) {
|
|
|
|
if (hd[i] == NULL) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
ide_create_drive(&ahci->dev[i].port, 0, hd[i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|