mirror of
https://github.com/FEX-Emu/linux.git
synced 2024-12-27 11:55:53 +00:00
5a0e3ad6af
percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
1435 lines
36 KiB
C
1435 lines
36 KiB
C
/*********************************************************************
|
|
*
|
|
* Filename: irlap_frame.c
|
|
* Version: 1.0
|
|
* Description: Build and transmit IrLAP frames
|
|
* Status: Stable
|
|
* Author: Dag Brattli <dagb@cs.uit.no>
|
|
* Created at: Tue Aug 19 10:27:26 1997
|
|
* Modified at: Wed Jan 5 08:59:04 2000
|
|
* Modified by: Dag Brattli <dagb@cs.uit.no>
|
|
*
|
|
* Copyright (c) 1998-2000 Dag Brattli <dagb@cs.uit.no>,
|
|
* All Rights Reserved.
|
|
* Copyright (c) 2000-2003 Jean Tourrilhes <jt@hpl.hp.com>
|
|
*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License as
|
|
* published by the Free Software Foundation; either version 2 of
|
|
* the License, or (at your option) any later version.
|
|
*
|
|
* Neither Dag Brattli nor University of Tromsø admit liability nor
|
|
* provide warranty for any of this software. This material is
|
|
* provided "AS-IS" and at no charge.
|
|
*
|
|
********************************************************************/
|
|
|
|
#include <linux/skbuff.h>
|
|
#include <linux/if.h>
|
|
#include <linux/if_ether.h>
|
|
#include <linux/netdevice.h>
|
|
#include <linux/irda.h>
|
|
#include <linux/slab.h>
|
|
|
|
#include <net/pkt_sched.h>
|
|
#include <net/sock.h>
|
|
|
|
#include <asm/byteorder.h>
|
|
|
|
#include <net/irda/irda.h>
|
|
#include <net/irda/irda_device.h>
|
|
#include <net/irda/irlap.h>
|
|
#include <net/irda/wrapper.h>
|
|
#include <net/irda/timer.h>
|
|
#include <net/irda/irlap_frame.h>
|
|
#include <net/irda/qos.h>
|
|
|
|
static void irlap_send_i_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
int command);
|
|
|
|
/*
|
|
* Function irlap_insert_info (self, skb)
|
|
*
|
|
* Insert minimum turnaround time and speed information into the skb. We
|
|
* need to do this since it's per packet relevant information. Safe to
|
|
* have this function inlined since it's only called from one place
|
|
*/
|
|
static inline void irlap_insert_info(struct irlap_cb *self,
|
|
struct sk_buff *skb)
|
|
{
|
|
struct irda_skb_cb *cb = (struct irda_skb_cb *) skb->cb;
|
|
|
|
/*
|
|
* Insert MTT (min. turn time) and speed into skb, so that the
|
|
* device driver knows which settings to use
|
|
*/
|
|
cb->magic = LAP_MAGIC;
|
|
cb->mtt = self->mtt_required;
|
|
cb->next_speed = self->speed;
|
|
|
|
/* Reset */
|
|
self->mtt_required = 0;
|
|
|
|
/*
|
|
* Delay equals negotiated BOFs count, plus the number of BOFs to
|
|
* force the negotiated minimum turnaround time
|
|
*/
|
|
cb->xbofs = self->bofs_count;
|
|
cb->next_xbofs = self->next_bofs;
|
|
cb->xbofs_delay = self->xbofs_delay;
|
|
|
|
/* Reset XBOF's delay (used only for getting min turn time) */
|
|
self->xbofs_delay = 0;
|
|
/* Put the correct xbofs value for the next packet */
|
|
self->bofs_count = self->next_bofs;
|
|
}
|
|
|
|
/*
|
|
* Function irlap_queue_xmit (self, skb)
|
|
*
|
|
* A little wrapper for dev_queue_xmit, so we can insert some common
|
|
* code into it.
|
|
*/
|
|
void irlap_queue_xmit(struct irlap_cb *self, struct sk_buff *skb)
|
|
{
|
|
/* Some common init stuff */
|
|
skb->dev = self->netdev;
|
|
skb_reset_mac_header(skb);
|
|
skb_reset_network_header(skb);
|
|
skb_reset_transport_header(skb);
|
|
skb->protocol = htons(ETH_P_IRDA);
|
|
skb->priority = TC_PRIO_BESTEFFORT;
|
|
|
|
irlap_insert_info(self, skb);
|
|
|
|
if (unlikely(self->mode & IRDA_MODE_MONITOR)) {
|
|
IRDA_DEBUG(3, "%s(): %s is in monitor mode\n", __func__,
|
|
self->netdev->name);
|
|
dev_kfree_skb(skb);
|
|
return;
|
|
}
|
|
|
|
dev_queue_xmit(skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_snrm_cmd (void)
|
|
*
|
|
* Transmits a connect SNRM command frame
|
|
*/
|
|
void irlap_send_snrm_frame(struct irlap_cb *self, struct qos_info *qos)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct snrm_frame *frame;
|
|
int ret;
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
/* Allocate frame */
|
|
tx_skb = alloc_skb(sizeof(struct snrm_frame) +
|
|
IRLAP_NEGOCIATION_PARAMS_LEN,
|
|
GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct snrm_frame *) skb_put(tx_skb, 2);
|
|
|
|
/* Insert connection address field */
|
|
if (qos)
|
|
frame->caddr = CMD_FRAME | CBROADCAST;
|
|
else
|
|
frame->caddr = CMD_FRAME | self->caddr;
|
|
|
|
/* Insert control field */
|
|
frame->control = SNRM_CMD | PF_BIT;
|
|
|
|
/*
|
|
* If we are establishing a connection then insert QoS parameters
|
|
*/
|
|
if (qos) {
|
|
skb_put(tx_skb, 9); /* 25 left */
|
|
frame->saddr = cpu_to_le32(self->saddr);
|
|
frame->daddr = cpu_to_le32(self->daddr);
|
|
|
|
frame->ncaddr = self->caddr;
|
|
|
|
ret = irlap_insert_qos_negotiation_params(self, tx_skb);
|
|
if (ret < 0) {
|
|
dev_kfree_skb(tx_skb);
|
|
return;
|
|
}
|
|
}
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_snrm_cmd (skb, info)
|
|
*
|
|
* Received SNRM (Set Normal Response Mode) command frame
|
|
*
|
|
*/
|
|
static void irlap_recv_snrm_cmd(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
struct snrm_frame *frame;
|
|
|
|
if (pskb_may_pull(skb,sizeof(struct snrm_frame))) {
|
|
frame = (struct snrm_frame *) skb->data;
|
|
|
|
/* Copy the new connection address ignoring the C/R bit */
|
|
info->caddr = frame->ncaddr & 0xFE;
|
|
|
|
/* Check if the new connection address is valid */
|
|
if ((info->caddr == 0x00) || (info->caddr == 0xfe)) {
|
|
IRDA_DEBUG(3, "%s(), invalid connection address!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
|
|
/* Copy peer device address */
|
|
info->daddr = le32_to_cpu(frame->saddr);
|
|
info->saddr = le32_to_cpu(frame->daddr);
|
|
|
|
/* Only accept if addressed directly to us */
|
|
if (info->saddr != self->saddr) {
|
|
IRDA_DEBUG(2, "%s(), not addressed to us!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
irlap_do_event(self, RECV_SNRM_CMD, skb, info);
|
|
} else {
|
|
/* Signal that this SNRM frame does not contain and I-field */
|
|
irlap_do_event(self, RECV_SNRM_CMD, skb, NULL);
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_ua_response_frame (qos)
|
|
*
|
|
* Send UA (Unnumbered Acknowledgement) frame
|
|
*
|
|
*/
|
|
void irlap_send_ua_response_frame(struct irlap_cb *self, struct qos_info *qos)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct ua_frame *frame;
|
|
int ret;
|
|
|
|
IRDA_DEBUG(2, "%s() <%ld>\n", __func__, jiffies);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
/* Allocate frame */
|
|
tx_skb = alloc_skb(sizeof(struct ua_frame) +
|
|
IRLAP_NEGOCIATION_PARAMS_LEN,
|
|
GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct ua_frame *) skb_put(tx_skb, 10);
|
|
|
|
/* Build UA response */
|
|
frame->caddr = self->caddr;
|
|
frame->control = UA_RSP | PF_BIT;
|
|
|
|
frame->saddr = cpu_to_le32(self->saddr);
|
|
frame->daddr = cpu_to_le32(self->daddr);
|
|
|
|
/* Should we send QoS negotiation parameters? */
|
|
if (qos) {
|
|
ret = irlap_insert_qos_negotiation_params(self, tx_skb);
|
|
if (ret < 0) {
|
|
dev_kfree_skb(tx_skb);
|
|
return;
|
|
}
|
|
}
|
|
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
|
|
/*
|
|
* Function irlap_send_dm_frame (void)
|
|
*
|
|
* Send disconnected mode (DM) frame
|
|
*
|
|
*/
|
|
void irlap_send_dm_frame( struct irlap_cb *self)
|
|
{
|
|
struct sk_buff *tx_skb = NULL;
|
|
struct dm_frame *frame;
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
tx_skb = alloc_skb(sizeof(struct dm_frame), GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct dm_frame *)skb_put(tx_skb, 2);
|
|
|
|
if (self->state == LAP_NDM)
|
|
frame->caddr = CBROADCAST;
|
|
else
|
|
frame->caddr = self->caddr;
|
|
|
|
frame->control = DM_RSP | PF_BIT;
|
|
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_disc_frame (void)
|
|
*
|
|
* Send disconnect (DISC) frame
|
|
*
|
|
*/
|
|
void irlap_send_disc_frame(struct irlap_cb *self)
|
|
{
|
|
struct sk_buff *tx_skb = NULL;
|
|
struct disc_frame *frame;
|
|
|
|
IRDA_DEBUG(3, "%s()\n", __func__);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
tx_skb = alloc_skb(sizeof(struct disc_frame), GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct disc_frame *)skb_put(tx_skb, 2);
|
|
|
|
frame->caddr = self->caddr | CMD_FRAME;
|
|
frame->control = DISC_CMD | PF_BIT;
|
|
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_discovery_xid_frame (S, s, command)
|
|
*
|
|
* Build and transmit a XID (eXchange station IDentifier) discovery
|
|
* frame.
|
|
*/
|
|
void irlap_send_discovery_xid_frame(struct irlap_cb *self, int S, __u8 s,
|
|
__u8 command, discovery_t *discovery)
|
|
{
|
|
struct sk_buff *tx_skb = NULL;
|
|
struct xid_frame *frame;
|
|
__u32 bcast = BROADCAST;
|
|
__u8 *info;
|
|
|
|
IRDA_DEBUG(4, "%s(), s=%d, S=%d, command=%d\n", __func__,
|
|
s, S, command);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
IRDA_ASSERT(discovery != NULL, return;);
|
|
|
|
tx_skb = alloc_skb(sizeof(struct xid_frame) + IRLAP_DISCOVERY_INFO_LEN,
|
|
GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
skb_put(tx_skb, 14);
|
|
frame = (struct xid_frame *) tx_skb->data;
|
|
|
|
if (command) {
|
|
frame->caddr = CBROADCAST | CMD_FRAME;
|
|
frame->control = XID_CMD | PF_BIT;
|
|
} else {
|
|
frame->caddr = CBROADCAST;
|
|
frame->control = XID_RSP | PF_BIT;
|
|
}
|
|
frame->ident = XID_FORMAT;
|
|
|
|
frame->saddr = cpu_to_le32(self->saddr);
|
|
|
|
if (command)
|
|
frame->daddr = cpu_to_le32(bcast);
|
|
else
|
|
frame->daddr = cpu_to_le32(discovery->data.daddr);
|
|
|
|
switch (S) {
|
|
case 1:
|
|
frame->flags = 0x00;
|
|
break;
|
|
case 6:
|
|
frame->flags = 0x01;
|
|
break;
|
|
case 8:
|
|
frame->flags = 0x02;
|
|
break;
|
|
case 16:
|
|
frame->flags = 0x03;
|
|
break;
|
|
default:
|
|
frame->flags = 0x02;
|
|
break;
|
|
}
|
|
|
|
frame->slotnr = s;
|
|
frame->version = 0x00;
|
|
|
|
/*
|
|
* Provide info for final slot only in commands, and for all
|
|
* responses. Send the second byte of the hint only if the
|
|
* EXTENSION bit is set in the first byte.
|
|
*/
|
|
if (!command || (frame->slotnr == 0xff)) {
|
|
int len;
|
|
|
|
if (discovery->data.hints[0] & HINT_EXTENSION) {
|
|
info = skb_put(tx_skb, 2);
|
|
info[0] = discovery->data.hints[0];
|
|
info[1] = discovery->data.hints[1];
|
|
} else {
|
|
info = skb_put(tx_skb, 1);
|
|
info[0] = discovery->data.hints[0];
|
|
}
|
|
info = skb_put(tx_skb, 1);
|
|
info[0] = discovery->data.charset;
|
|
|
|
len = IRDA_MIN(discovery->name_len, skb_tailroom(tx_skb));
|
|
info = skb_put(tx_skb, len);
|
|
memcpy(info, discovery->data.info, len);
|
|
}
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_discovery_xid_rsp (skb, info)
|
|
*
|
|
* Received a XID discovery response
|
|
*
|
|
*/
|
|
static void irlap_recv_discovery_xid_rsp(struct irlap_cb *self,
|
|
struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
struct xid_frame *xid;
|
|
discovery_t *discovery = NULL;
|
|
__u8 *discovery_info;
|
|
char *text;
|
|
|
|
IRDA_DEBUG(4, "%s()\n", __func__);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
if (!pskb_may_pull(skb, sizeof(struct xid_frame))) {
|
|
IRDA_ERROR("%s: frame too short!\n", __func__);
|
|
return;
|
|
}
|
|
|
|
xid = (struct xid_frame *) skb->data;
|
|
|
|
info->daddr = le32_to_cpu(xid->saddr);
|
|
info->saddr = le32_to_cpu(xid->daddr);
|
|
|
|
/* Make sure frame is addressed to us */
|
|
if ((info->saddr != self->saddr) && (info->saddr != BROADCAST)) {
|
|
IRDA_DEBUG(0, "%s(), frame is not addressed to us!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
|
|
if ((discovery = kzalloc(sizeof(discovery_t), GFP_ATOMIC)) == NULL) {
|
|
IRDA_WARNING("%s: kmalloc failed!\n", __func__);
|
|
return;
|
|
}
|
|
|
|
discovery->data.daddr = info->daddr;
|
|
discovery->data.saddr = self->saddr;
|
|
discovery->timestamp = jiffies;
|
|
|
|
IRDA_DEBUG(4, "%s(), daddr=%08x\n", __func__,
|
|
discovery->data.daddr);
|
|
|
|
discovery_info = skb_pull(skb, sizeof(struct xid_frame));
|
|
|
|
/* Get info returned from peer */
|
|
discovery->data.hints[0] = discovery_info[0];
|
|
if (discovery_info[0] & HINT_EXTENSION) {
|
|
IRDA_DEBUG(4, "EXTENSION\n");
|
|
discovery->data.hints[1] = discovery_info[1];
|
|
discovery->data.charset = discovery_info[2];
|
|
text = (char *) &discovery_info[3];
|
|
} else {
|
|
discovery->data.hints[1] = 0;
|
|
discovery->data.charset = discovery_info[1];
|
|
text = (char *) &discovery_info[2];
|
|
}
|
|
/*
|
|
* Terminate info string, should be safe since this is where the
|
|
* FCS bytes resides.
|
|
*/
|
|
skb->data[skb->len] = '\0';
|
|
strncpy(discovery->data.info, text, NICKNAME_MAX_LEN);
|
|
discovery->name_len = strlen(discovery->data.info);
|
|
|
|
info->discovery = discovery;
|
|
|
|
irlap_do_event(self, RECV_DISCOVERY_XID_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_discovery_xid_cmd (skb, info)
|
|
*
|
|
* Received a XID discovery command
|
|
*
|
|
*/
|
|
static void irlap_recv_discovery_xid_cmd(struct irlap_cb *self,
|
|
struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
struct xid_frame *xid;
|
|
discovery_t *discovery = NULL;
|
|
__u8 *discovery_info;
|
|
char *text;
|
|
|
|
if (!pskb_may_pull(skb, sizeof(struct xid_frame))) {
|
|
IRDA_ERROR("%s: frame too short!\n", __func__);
|
|
return;
|
|
}
|
|
|
|
xid = (struct xid_frame *) skb->data;
|
|
|
|
info->daddr = le32_to_cpu(xid->saddr);
|
|
info->saddr = le32_to_cpu(xid->daddr);
|
|
|
|
/* Make sure frame is addressed to us */
|
|
if ((info->saddr != self->saddr) && (info->saddr != BROADCAST)) {
|
|
IRDA_DEBUG(0, "%s(), frame is not addressed to us!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
|
|
switch (xid->flags & 0x03) {
|
|
case 0x00:
|
|
info->S = 1;
|
|
break;
|
|
case 0x01:
|
|
info->S = 6;
|
|
break;
|
|
case 0x02:
|
|
info->S = 8;
|
|
break;
|
|
case 0x03:
|
|
info->S = 16;
|
|
break;
|
|
default:
|
|
/* Error!! */
|
|
return;
|
|
}
|
|
info->s = xid->slotnr;
|
|
|
|
discovery_info = skb_pull(skb, sizeof(struct xid_frame));
|
|
|
|
/*
|
|
* Check if last frame
|
|
*/
|
|
if (info->s == 0xff) {
|
|
/* Check if things are sane at this point... */
|
|
if((discovery_info == NULL) ||
|
|
!pskb_may_pull(skb, 3)) {
|
|
IRDA_ERROR("%s: discovery frame too short!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
|
|
/*
|
|
* We now have some discovery info to deliver!
|
|
*/
|
|
discovery = kmalloc(sizeof(discovery_t), GFP_ATOMIC);
|
|
if (!discovery) {
|
|
IRDA_WARNING("%s: unable to malloc!\n", __func__);
|
|
return;
|
|
}
|
|
|
|
discovery->data.daddr = info->daddr;
|
|
discovery->data.saddr = self->saddr;
|
|
discovery->timestamp = jiffies;
|
|
|
|
discovery->data.hints[0] = discovery_info[0];
|
|
if (discovery_info[0] & HINT_EXTENSION) {
|
|
discovery->data.hints[1] = discovery_info[1];
|
|
discovery->data.charset = discovery_info[2];
|
|
text = (char *) &discovery_info[3];
|
|
} else {
|
|
discovery->data.hints[1] = 0;
|
|
discovery->data.charset = discovery_info[1];
|
|
text = (char *) &discovery_info[2];
|
|
}
|
|
/*
|
|
* Terminate string, should be safe since this is where the
|
|
* FCS bytes resides.
|
|
*/
|
|
skb->data[skb->len] = '\0';
|
|
strncpy(discovery->data.info, text, NICKNAME_MAX_LEN);
|
|
discovery->name_len = strlen(discovery->data.info);
|
|
|
|
info->discovery = discovery;
|
|
} else
|
|
info->discovery = NULL;
|
|
|
|
irlap_do_event(self, RECV_DISCOVERY_XID_CMD, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_rr_frame (self, command)
|
|
*
|
|
* Build and transmit RR (Receive Ready) frame. Notice that it is currently
|
|
* only possible to send RR frames with the poll bit set.
|
|
*/
|
|
void irlap_send_rr_frame(struct irlap_cb *self, int command)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct rr_frame *frame;
|
|
|
|
tx_skb = alloc_skb(sizeof(struct rr_frame), GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct rr_frame *)skb_put(tx_skb, 2);
|
|
|
|
frame->caddr = self->caddr;
|
|
frame->caddr |= (command) ? CMD_FRAME : 0;
|
|
|
|
frame->control = RR | PF_BIT | (self->vr << 5);
|
|
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_rd_frame (self)
|
|
*
|
|
* Request disconnect. Used by a secondary station to request the
|
|
* disconnection of the link.
|
|
*/
|
|
void irlap_send_rd_frame(struct irlap_cb *self)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct rd_frame *frame;
|
|
|
|
tx_skb = alloc_skb(sizeof(struct rd_frame), GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
frame = (struct rd_frame *)skb_put(tx_skb, 2);
|
|
|
|
frame->caddr = self->caddr;
|
|
frame->caddr = RD_RSP | PF_BIT;
|
|
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_rr_frame (skb, info)
|
|
*
|
|
* Received RR (Receive Ready) frame from peer station, no harm in
|
|
* making it inline since its called only from one single place
|
|
* (irlap_driver_rcv).
|
|
*/
|
|
static inline void irlap_recv_rr_frame(struct irlap_cb *self,
|
|
struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
info->nr = skb->data[1] >> 5;
|
|
|
|
/* Check if this is a command or a response frame */
|
|
if (command)
|
|
irlap_do_event(self, RECV_RR_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_RR_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_rnr_frame (self, skb, info)
|
|
*
|
|
* Received RNR (Receive Not Ready) frame from peer station
|
|
*
|
|
*/
|
|
static void irlap_recv_rnr_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
info->nr = skb->data[1] >> 5;
|
|
|
|
IRDA_DEBUG(4, "%s(), nr=%d, %ld\n", __func__, info->nr, jiffies);
|
|
|
|
if (command)
|
|
irlap_do_event(self, RECV_RNR_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_RNR_RSP, skb, info);
|
|
}
|
|
|
|
static void irlap_recv_rej_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
IRDA_DEBUG(0, "%s()\n", __func__);
|
|
|
|
info->nr = skb->data[1] >> 5;
|
|
|
|
/* Check if this is a command or a response frame */
|
|
if (command)
|
|
irlap_do_event(self, RECV_REJ_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_REJ_RSP, skb, info);
|
|
}
|
|
|
|
static void irlap_recv_srej_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
IRDA_DEBUG(0, "%s()\n", __func__);
|
|
|
|
info->nr = skb->data[1] >> 5;
|
|
|
|
/* Check if this is a command or a response frame */
|
|
if (command)
|
|
irlap_do_event(self, RECV_SREJ_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_SREJ_RSP, skb, info);
|
|
}
|
|
|
|
static void irlap_recv_disc_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
IRDA_DEBUG(2, "%s()\n", __func__);
|
|
|
|
/* Check if this is a command or a response frame */
|
|
if (command)
|
|
irlap_do_event(self, RECV_DISC_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_RD_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_ua_frame (skb, frame)
|
|
*
|
|
* Received UA (Unnumbered Acknowledgement) frame
|
|
*
|
|
*/
|
|
static inline void irlap_recv_ua_frame(struct irlap_cb *self,
|
|
struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
irlap_do_event(self, RECV_UA_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_data_primary(self, skb)
|
|
*
|
|
* Send I-frames as the primary station but without the poll bit set
|
|
*
|
|
*/
|
|
void irlap_send_data_primary(struct irlap_cb *self, struct sk_buff *skb)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
|
|
if (skb->data[1] == I_FRAME) {
|
|
|
|
/*
|
|
* Insert frame sequence number (Vs) in control field before
|
|
* inserting into transmit window queue.
|
|
*/
|
|
skb->data[1] = I_FRAME | (self->vs << 1);
|
|
|
|
/*
|
|
* Insert frame in store, in case of retransmissions
|
|
* Increase skb reference count, see irlap_do_event()
|
|
*/
|
|
skb_get(skb);
|
|
skb_queue_tail(&self->wx_list, skb);
|
|
|
|
/* Copy buffer */
|
|
tx_skb = skb_clone(skb, GFP_ATOMIC);
|
|
if (tx_skb == NULL) {
|
|
return;
|
|
}
|
|
|
|
self->vs = (self->vs + 1) % 8;
|
|
self->ack_required = FALSE;
|
|
self->window -= 1;
|
|
|
|
irlap_send_i_frame( self, tx_skb, CMD_FRAME);
|
|
} else {
|
|
IRDA_DEBUG(4, "%s(), sending unreliable frame\n", __func__);
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, CMD_FRAME);
|
|
self->window -= 1;
|
|
}
|
|
}
|
|
/*
|
|
* Function irlap_send_data_primary_poll (self, skb)
|
|
*
|
|
* Send I(nformation) frame as primary with poll bit set
|
|
*/
|
|
void irlap_send_data_primary_poll(struct irlap_cb *self, struct sk_buff *skb)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
int transmission_time;
|
|
|
|
/* Stop P timer */
|
|
del_timer(&self->poll_timer);
|
|
|
|
/* Is this reliable or unreliable data? */
|
|
if (skb->data[1] == I_FRAME) {
|
|
|
|
/*
|
|
* Insert frame sequence number (Vs) in control field before
|
|
* inserting into transmit window queue.
|
|
*/
|
|
skb->data[1] = I_FRAME | (self->vs << 1);
|
|
|
|
/*
|
|
* Insert frame in store, in case of retransmissions
|
|
* Increase skb reference count, see irlap_do_event()
|
|
*/
|
|
skb_get(skb);
|
|
skb_queue_tail(&self->wx_list, skb);
|
|
|
|
/* Copy buffer */
|
|
tx_skb = skb_clone(skb, GFP_ATOMIC);
|
|
if (tx_skb == NULL) {
|
|
return;
|
|
}
|
|
|
|
/*
|
|
* Set poll bit if necessary. We do this to the copied
|
|
* skb, since retransmitted need to set or clear the poll
|
|
* bit depending on when they are sent.
|
|
*/
|
|
tx_skb->data[1] |= PF_BIT;
|
|
|
|
self->vs = (self->vs + 1) % 8;
|
|
self->ack_required = FALSE;
|
|
|
|
irlap_next_state(self, LAP_NRM_P);
|
|
irlap_send_i_frame(self, tx_skb, CMD_FRAME);
|
|
} else {
|
|
IRDA_DEBUG(4, "%s(), sending unreliable frame\n", __func__);
|
|
|
|
if (self->ack_required) {
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, CMD_FRAME);
|
|
irlap_next_state(self, LAP_NRM_P);
|
|
irlap_send_rr_frame(self, CMD_FRAME);
|
|
self->ack_required = FALSE;
|
|
} else {
|
|
skb->data[1] |= PF_BIT;
|
|
irlap_next_state(self, LAP_NRM_P);
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, CMD_FRAME);
|
|
}
|
|
}
|
|
|
|
/* How much time we took for transmission of all frames.
|
|
* We don't know, so let assume we used the full window. Jean II */
|
|
transmission_time = self->final_timeout;
|
|
|
|
/* Reset parameter so that we can fill next window */
|
|
self->window = self->window_size;
|
|
|
|
#ifdef CONFIG_IRDA_DYNAMIC_WINDOW
|
|
/* Remove what we have not used. Just do a prorata of the
|
|
* bytes left in window to window capacity.
|
|
* See max_line_capacities[][] in qos.c for details. Jean II */
|
|
transmission_time -= (self->final_timeout * self->bytes_left
|
|
/ self->line_capacity);
|
|
IRDA_DEBUG(4, "%s() adjusting transmission_time : ft=%d, bl=%d, lc=%d -> tt=%d\n", __func__, self->final_timeout, self->bytes_left, self->line_capacity, transmission_time);
|
|
|
|
/* We are allowed to transmit a maximum number of bytes again. */
|
|
self->bytes_left = self->line_capacity;
|
|
#endif /* CONFIG_IRDA_DYNAMIC_WINDOW */
|
|
|
|
/*
|
|
* The network layer has a intermediate buffer between IrLAP
|
|
* and the IrDA driver which can contain 8 frames. So, even
|
|
* though IrLAP is currently sending the *last* frame of the
|
|
* tx-window, the driver most likely has only just started
|
|
* sending the *first* frame of the same tx-window.
|
|
* I.e. we are always at the very begining of or Tx window.
|
|
* Now, we are supposed to set the final timer from the end
|
|
* of our tx-window to let the other peer reply. So, we need
|
|
* to add extra time to compensate for the fact that we
|
|
* are really at the start of tx-window, otherwise the final timer
|
|
* might expire before he can answer...
|
|
* Jean II
|
|
*/
|
|
irlap_start_final_timer(self, self->final_timeout + transmission_time);
|
|
|
|
/*
|
|
* The clever amongst you might ask why we do this adjustement
|
|
* only here, and not in all the other cases in irlap_event.c.
|
|
* In all those other case, we only send a very short management
|
|
* frame (few bytes), so the adjustement would be lost in the
|
|
* noise...
|
|
* The exception of course is irlap_resend_rejected_frame().
|
|
* Jean II */
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_data_secondary_final (self, skb)
|
|
*
|
|
* Send I(nformation) frame as secondary with final bit set
|
|
*
|
|
*/
|
|
void irlap_send_data_secondary_final(struct irlap_cb *self,
|
|
struct sk_buff *skb)
|
|
{
|
|
struct sk_buff *tx_skb = NULL;
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
IRDA_ASSERT(skb != NULL, return;);
|
|
|
|
/* Is this reliable or unreliable data? */
|
|
if (skb->data[1] == I_FRAME) {
|
|
|
|
/*
|
|
* Insert frame sequence number (Vs) in control field before
|
|
* inserting into transmit window queue.
|
|
*/
|
|
skb->data[1] = I_FRAME | (self->vs << 1);
|
|
|
|
/*
|
|
* Insert frame in store, in case of retransmissions
|
|
* Increase skb reference count, see irlap_do_event()
|
|
*/
|
|
skb_get(skb);
|
|
skb_queue_tail(&self->wx_list, skb);
|
|
|
|
tx_skb = skb_clone(skb, GFP_ATOMIC);
|
|
if (tx_skb == NULL) {
|
|
return;
|
|
}
|
|
|
|
tx_skb->data[1] |= PF_BIT;
|
|
|
|
self->vs = (self->vs + 1) % 8;
|
|
self->ack_required = FALSE;
|
|
|
|
irlap_send_i_frame(self, tx_skb, RSP_FRAME);
|
|
} else {
|
|
if (self->ack_required) {
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, RSP_FRAME);
|
|
irlap_send_rr_frame(self, RSP_FRAME);
|
|
self->ack_required = FALSE;
|
|
} else {
|
|
skb->data[1] |= PF_BIT;
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, RSP_FRAME);
|
|
}
|
|
}
|
|
|
|
self->window = self->window_size;
|
|
#ifdef CONFIG_IRDA_DYNAMIC_WINDOW
|
|
/* We are allowed to transmit a maximum number of bytes again. */
|
|
self->bytes_left = self->line_capacity;
|
|
#endif /* CONFIG_IRDA_DYNAMIC_WINDOW */
|
|
|
|
irlap_start_wd_timer(self, self->wd_timeout);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_data_secondary (self, skb)
|
|
*
|
|
* Send I(nformation) frame as secondary without final bit set
|
|
*
|
|
*/
|
|
void irlap_send_data_secondary(struct irlap_cb *self, struct sk_buff *skb)
|
|
{
|
|
struct sk_buff *tx_skb = NULL;
|
|
|
|
/* Is this reliable or unreliable data? */
|
|
if (skb->data[1] == I_FRAME) {
|
|
|
|
/*
|
|
* Insert frame sequence number (Vs) in control field before
|
|
* inserting into transmit window queue.
|
|
*/
|
|
skb->data[1] = I_FRAME | (self->vs << 1);
|
|
|
|
/*
|
|
* Insert frame in store, in case of retransmissions
|
|
* Increase skb reference count, see irlap_do_event()
|
|
*/
|
|
skb_get(skb);
|
|
skb_queue_tail(&self->wx_list, skb);
|
|
|
|
tx_skb = skb_clone(skb, GFP_ATOMIC);
|
|
if (tx_skb == NULL) {
|
|
return;
|
|
}
|
|
|
|
self->vs = (self->vs + 1) % 8;
|
|
self->ack_required = FALSE;
|
|
self->window -= 1;
|
|
|
|
irlap_send_i_frame(self, tx_skb, RSP_FRAME);
|
|
} else {
|
|
irlap_send_ui_frame(self, skb_get(skb), self->caddr, RSP_FRAME);
|
|
self->window -= 1;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Function irlap_resend_rejected_frames (nr)
|
|
*
|
|
* Resend frames which has not been acknowledged. Should be safe to
|
|
* traverse the list without locking it since this function will only be
|
|
* called from interrupt context (BH)
|
|
*/
|
|
void irlap_resend_rejected_frames(struct irlap_cb *self, int command)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct sk_buff *skb;
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
/* Resend unacknowledged frame(s) */
|
|
skb_queue_walk(&self->wx_list, skb) {
|
|
irlap_wait_min_turn_around(self, &self->qos_tx);
|
|
|
|
/* We copy the skb to be retransmitted since we will have to
|
|
* modify it. Cloning will confuse packet sniffers
|
|
*/
|
|
/* tx_skb = skb_clone( skb, GFP_ATOMIC); */
|
|
tx_skb = skb_copy(skb, GFP_ATOMIC);
|
|
if (!tx_skb) {
|
|
IRDA_DEBUG(0, "%s(), unable to copy\n", __func__);
|
|
return;
|
|
}
|
|
|
|
/* Clear old Nr field + poll bit */
|
|
tx_skb->data[1] &= 0x0f;
|
|
|
|
/*
|
|
* Set poll bit on the last frame retransmitted
|
|
*/
|
|
if (skb_queue_is_last(&self->wx_list, skb))
|
|
tx_skb->data[1] |= PF_BIT; /* Set p/f bit */
|
|
else
|
|
tx_skb->data[1] &= ~PF_BIT; /* Clear p/f bit */
|
|
|
|
irlap_send_i_frame(self, tx_skb, command);
|
|
}
|
|
#if 0 /* Not yet */
|
|
/*
|
|
* We can now fill the window with additional data frames
|
|
*/
|
|
while (!skb_queue_empty(&self->txq)) {
|
|
|
|
IRDA_DEBUG(0, "%s(), sending additional frames!\n", __func__);
|
|
if (self->window > 0) {
|
|
skb = skb_dequeue( &self->txq);
|
|
IRDA_ASSERT(skb != NULL, return;);
|
|
|
|
/*
|
|
* If send window > 1 then send frame with pf
|
|
* bit cleared
|
|
*/
|
|
if ((self->window > 1) &&
|
|
!skb_queue_empty(&self->txq)) {
|
|
irlap_send_data_primary(self, skb);
|
|
} else {
|
|
irlap_send_data_primary_poll(self, skb);
|
|
}
|
|
kfree_skb(skb);
|
|
}
|
|
}
|
|
#endif
|
|
}
|
|
|
|
void irlap_resend_rejected_frame(struct irlap_cb *self, int command)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct sk_buff *skb;
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
|
|
/* Resend unacknowledged frame(s) */
|
|
skb = skb_peek(&self->wx_list);
|
|
if (skb != NULL) {
|
|
irlap_wait_min_turn_around(self, &self->qos_tx);
|
|
|
|
/* We copy the skb to be retransmitted since we will have to
|
|
* modify it. Cloning will confuse packet sniffers
|
|
*/
|
|
/* tx_skb = skb_clone( skb, GFP_ATOMIC); */
|
|
tx_skb = skb_copy(skb, GFP_ATOMIC);
|
|
if (!tx_skb) {
|
|
IRDA_DEBUG(0, "%s(), unable to copy\n", __func__);
|
|
return;
|
|
}
|
|
|
|
/* Clear old Nr field + poll bit */
|
|
tx_skb->data[1] &= 0x0f;
|
|
|
|
/* Set poll/final bit */
|
|
tx_skb->data[1] |= PF_BIT; /* Set p/f bit */
|
|
|
|
irlap_send_i_frame(self, tx_skb, command);
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_ui_frame (self, skb, command)
|
|
*
|
|
* Contruct and transmit an Unnumbered Information (UI) frame
|
|
*
|
|
*/
|
|
void irlap_send_ui_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
__u8 caddr, int command)
|
|
{
|
|
IRDA_DEBUG(4, "%s()\n", __func__);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
IRDA_ASSERT(skb != NULL, return;);
|
|
|
|
/* Insert connection address */
|
|
skb->data[0] = caddr | ((command) ? CMD_FRAME : 0);
|
|
|
|
irlap_queue_xmit(self, skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_i_frame (skb)
|
|
*
|
|
* Contruct and transmit Information (I) frame
|
|
*/
|
|
static void irlap_send_i_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
int command)
|
|
{
|
|
/* Insert connection address */
|
|
skb->data[0] = self->caddr;
|
|
skb->data[0] |= (command) ? CMD_FRAME : 0;
|
|
|
|
/* Insert next to receive (Vr) */
|
|
skb->data[1] |= (self->vr << 5); /* insert nr */
|
|
|
|
irlap_queue_xmit(self, skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_i_frame (skb, frame)
|
|
*
|
|
* Receive and parse an I (Information) frame, no harm in making it inline
|
|
* since it's called only from one single place (irlap_driver_rcv).
|
|
*/
|
|
static inline void irlap_recv_i_frame(struct irlap_cb *self,
|
|
struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
info->nr = skb->data[1] >> 5; /* Next to receive */
|
|
info->pf = skb->data[1] & PF_BIT; /* Final bit */
|
|
info->ns = (skb->data[1] >> 1) & 0x07; /* Next to send */
|
|
|
|
/* Check if this is a command or a response frame */
|
|
if (command)
|
|
irlap_do_event(self, RECV_I_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_I_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_ui_frame (self, skb, info)
|
|
*
|
|
* Receive and parse an Unnumbered Information (UI) frame
|
|
*
|
|
*/
|
|
static void irlap_recv_ui_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
IRDA_DEBUG( 4, "%s()\n", __func__);
|
|
|
|
info->pf = skb->data[1] & PF_BIT; /* Final bit */
|
|
|
|
irlap_do_event(self, RECV_UI_FRAME, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_frmr_frame (skb, frame)
|
|
*
|
|
* Received Frame Reject response.
|
|
*
|
|
*/
|
|
static void irlap_recv_frmr_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info)
|
|
{
|
|
__u8 *frame;
|
|
int w, x, y, z;
|
|
|
|
IRDA_DEBUG(0, "%s()\n", __func__);
|
|
|
|
IRDA_ASSERT(self != NULL, return;);
|
|
IRDA_ASSERT(self->magic == LAP_MAGIC, return;);
|
|
IRDA_ASSERT(skb != NULL, return;);
|
|
IRDA_ASSERT(info != NULL, return;);
|
|
|
|
if (!pskb_may_pull(skb, 4)) {
|
|
IRDA_ERROR("%s: frame too short!\n", __func__);
|
|
return;
|
|
}
|
|
|
|
frame = skb->data;
|
|
|
|
info->nr = frame[2] >> 5; /* Next to receive */
|
|
info->pf = frame[2] & PF_BIT; /* Final bit */
|
|
info->ns = (frame[2] >> 1) & 0x07; /* Next to send */
|
|
|
|
w = frame[3] & 0x01;
|
|
x = frame[3] & 0x02;
|
|
y = frame[3] & 0x04;
|
|
z = frame[3] & 0x08;
|
|
|
|
if (w) {
|
|
IRDA_DEBUG(0, "Rejected control field is undefined or not "
|
|
"implemented.\n");
|
|
}
|
|
if (x) {
|
|
IRDA_DEBUG(0, "Rejected control field was invalid because it "
|
|
"contained a non permitted I field.\n");
|
|
}
|
|
if (y) {
|
|
IRDA_DEBUG(0, "Received I field exceeded the maximum negotiated "
|
|
"for the existing connection or exceeded the maximum "
|
|
"this station supports if no connection exists.\n");
|
|
}
|
|
if (z) {
|
|
IRDA_DEBUG(0, "Rejected control field control field contained an "
|
|
"invalid Nr count.\n");
|
|
}
|
|
irlap_do_event(self, RECV_FRMR_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_send_test_frame (self, daddr)
|
|
*
|
|
* Send a test frame response
|
|
*
|
|
*/
|
|
void irlap_send_test_frame(struct irlap_cb *self, __u8 caddr, __u32 daddr,
|
|
struct sk_buff *cmd)
|
|
{
|
|
struct sk_buff *tx_skb;
|
|
struct test_frame *frame;
|
|
__u8 *info;
|
|
|
|
tx_skb = alloc_skb(cmd->len + sizeof(struct test_frame), GFP_ATOMIC);
|
|
if (!tx_skb)
|
|
return;
|
|
|
|
/* Broadcast frames must include saddr and daddr fields */
|
|
if (caddr == CBROADCAST) {
|
|
frame = (struct test_frame *)
|
|
skb_put(tx_skb, sizeof(struct test_frame));
|
|
|
|
/* Insert the swapped addresses */
|
|
frame->saddr = cpu_to_le32(self->saddr);
|
|
frame->daddr = cpu_to_le32(daddr);
|
|
} else
|
|
frame = (struct test_frame *) skb_put(tx_skb, LAP_ADDR_HEADER + LAP_CTRL_HEADER);
|
|
|
|
frame->caddr = caddr;
|
|
frame->control = TEST_RSP | PF_BIT;
|
|
|
|
/* Copy info */
|
|
info = skb_put(tx_skb, cmd->len);
|
|
memcpy(info, cmd->data, cmd->len);
|
|
|
|
/* Return to sender */
|
|
irlap_wait_min_turn_around(self, &self->qos_tx);
|
|
irlap_queue_xmit(self, tx_skb);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_recv_test_frame (self, skb)
|
|
*
|
|
* Receive a test frame
|
|
*
|
|
*/
|
|
static void irlap_recv_test_frame(struct irlap_cb *self, struct sk_buff *skb,
|
|
struct irlap_info *info, int command)
|
|
{
|
|
struct test_frame *frame;
|
|
|
|
IRDA_DEBUG(2, "%s()\n", __func__);
|
|
|
|
if (!pskb_may_pull(skb, sizeof(*frame))) {
|
|
IRDA_ERROR("%s: frame too short!\n", __func__);
|
|
return;
|
|
}
|
|
frame = (struct test_frame *) skb->data;
|
|
|
|
/* Broadcast frames must carry saddr and daddr fields */
|
|
if (info->caddr == CBROADCAST) {
|
|
if (skb->len < sizeof(struct test_frame)) {
|
|
IRDA_DEBUG(0, "%s() test frame too short!\n",
|
|
__func__);
|
|
return;
|
|
}
|
|
|
|
/* Read and swap addresses */
|
|
info->daddr = le32_to_cpu(frame->saddr);
|
|
info->saddr = le32_to_cpu(frame->daddr);
|
|
|
|
/* Make sure frame is addressed to us */
|
|
if ((info->saddr != self->saddr) &&
|
|
(info->saddr != BROADCAST)) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
if (command)
|
|
irlap_do_event(self, RECV_TEST_CMD, skb, info);
|
|
else
|
|
irlap_do_event(self, RECV_TEST_RSP, skb, info);
|
|
}
|
|
|
|
/*
|
|
* Function irlap_driver_rcv (skb, netdev, ptype)
|
|
*
|
|
* Called when a frame is received. Dispatches the right receive function
|
|
* for processing of the frame.
|
|
*
|
|
* Note on skb management :
|
|
* After calling the higher layers of the IrDA stack, we always
|
|
* kfree() the skb, which drop the reference count (and potentially
|
|
* destroy it).
|
|
* If a higher layer of the stack want to keep the skb around (to put
|
|
* in a queue or pass it to the higher layer), it will need to use
|
|
* skb_get() to keep a reference on it. This is usually done at the
|
|
* LMP level in irlmp.c.
|
|
* Jean II
|
|
*/
|
|
int irlap_driver_rcv(struct sk_buff *skb, struct net_device *dev,
|
|
struct packet_type *ptype, struct net_device *orig_dev)
|
|
{
|
|
struct irlap_info info;
|
|
struct irlap_cb *self;
|
|
int command;
|
|
__u8 control;
|
|
int ret = -1;
|
|
|
|
if (!net_eq(dev_net(dev), &init_net))
|
|
goto out;
|
|
|
|
/* FIXME: should we get our own field? */
|
|
self = (struct irlap_cb *) dev->atalk_ptr;
|
|
|
|
/* If the net device is down, then IrLAP is gone! */
|
|
if (!self || self->magic != LAP_MAGIC)
|
|
goto err;
|
|
|
|
/* We are no longer an "old" protocol, so we need to handle
|
|
* share and non linear skbs. This should never happen, so
|
|
* we don't need to be clever about it. Jean II */
|
|
if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL) {
|
|
IRDA_ERROR("%s: can't clone shared skb!\n", __func__);
|
|
goto err;
|
|
}
|
|
|
|
/* Check if frame is large enough for parsing */
|
|
if (!pskb_may_pull(skb, 2)) {
|
|
IRDA_ERROR("%s: frame too short!\n", __func__);
|
|
goto err;
|
|
}
|
|
|
|
command = skb->data[0] & CMD_FRAME;
|
|
info.caddr = skb->data[0] & CBROADCAST;
|
|
|
|
info.pf = skb->data[1] & PF_BIT;
|
|
info.control = skb->data[1] & ~PF_BIT; /* Mask away poll/final bit */
|
|
|
|
control = info.control;
|
|
|
|
/* First we check if this frame has a valid connection address */
|
|
if ((info.caddr != self->caddr) && (info.caddr != CBROADCAST)) {
|
|
IRDA_DEBUG(0, "%s(), wrong connection address!\n",
|
|
__func__);
|
|
goto out;
|
|
}
|
|
/*
|
|
* Optimize for the common case and check if the frame is an
|
|
* I(nformation) frame. Only I-frames have bit 0 set to 0
|
|
*/
|
|
if (~control & 0x01) {
|
|
irlap_recv_i_frame(self, skb, &info, command);
|
|
goto out;
|
|
}
|
|
/*
|
|
* We now check is the frame is an S(upervisory) frame. Only
|
|
* S-frames have bit 0 set to 1 and bit 1 set to 0
|
|
*/
|
|
if (~control & 0x02) {
|
|
/*
|
|
* Received S(upervisory) frame, check which frame type it is
|
|
* only the first nibble is of interest
|
|
*/
|
|
switch (control & 0x0f) {
|
|
case RR:
|
|
irlap_recv_rr_frame(self, skb, &info, command);
|
|
break;
|
|
case RNR:
|
|
irlap_recv_rnr_frame(self, skb, &info, command);
|
|
break;
|
|
case REJ:
|
|
irlap_recv_rej_frame(self, skb, &info, command);
|
|
break;
|
|
case SREJ:
|
|
irlap_recv_srej_frame(self, skb, &info, command);
|
|
break;
|
|
default:
|
|
IRDA_WARNING("%s: Unknown S-frame %02x received!\n",
|
|
__func__, info.control);
|
|
break;
|
|
}
|
|
goto out;
|
|
}
|
|
/*
|
|
* This must be a C(ontrol) frame
|
|
*/
|
|
switch (control) {
|
|
case XID_RSP:
|
|
irlap_recv_discovery_xid_rsp(self, skb, &info);
|
|
break;
|
|
case XID_CMD:
|
|
irlap_recv_discovery_xid_cmd(self, skb, &info);
|
|
break;
|
|
case SNRM_CMD:
|
|
irlap_recv_snrm_cmd(self, skb, &info);
|
|
break;
|
|
case DM_RSP:
|
|
irlap_do_event(self, RECV_DM_RSP, skb, &info);
|
|
break;
|
|
case DISC_CMD: /* And RD_RSP since they have the same value */
|
|
irlap_recv_disc_frame(self, skb, &info, command);
|
|
break;
|
|
case TEST_CMD:
|
|
irlap_recv_test_frame(self, skb, &info, command);
|
|
break;
|
|
case UA_RSP:
|
|
irlap_recv_ua_frame(self, skb, &info);
|
|
break;
|
|
case FRMR_RSP:
|
|
irlap_recv_frmr_frame(self, skb, &info);
|
|
break;
|
|
case UI_FRAME:
|
|
irlap_recv_ui_frame(self, skb, &info);
|
|
break;
|
|
default:
|
|
IRDA_WARNING("%s: Unknown frame %02x received!\n",
|
|
__func__, info.control);
|
|
break;
|
|
}
|
|
out:
|
|
ret = 0;
|
|
err:
|
|
/* Always drop our reference on the skb */
|
|
dev_kfree_skb(skb);
|
|
return ret;
|
|
}
|