* [PATCH][ATM] svc's possibly race with sigd
@ 2003-05-05 22:53 chas williams
2003-05-06 6:43 ` David S. Miller
0 siblings, 1 reply; 2+ messages in thread
From: chas williams @ 2003-05-05 22:53 UTC (permalink / raw)
To: davem; +Cc: linux-kernel
pretty sure the author wanted the wait queues in place before
talking to sigd which might cause a wakeup on the vcc in question.
this race is VERY unlikely, since going down to user space and back
is quite slow.
--- linux-2.5.68/net/atm/svc.c.000 Mon May 5 18:31:34 2003
+++ linux-2.5.68/net/atm/svc.c Mon May 5 18:33:04 2003
@@ -64,8 +64,8 @@
DPRINTK("svc_disconnect %p\n",vcc);
if (test_bit(ATM_VF_REGIS,&vcc->flags)) {
- sigd_enq(vcc,as_close,NULL,NULL,NULL);
add_wait_queue(&vcc->sleep,&wait);
+ sigd_enq(vcc,as_close,NULL,NULL,NULL);
while (!test_bit(ATM_VF_RELEASED,&vcc->flags) && sigd) {
set_current_state(TASK_UNINTERRUPTIBLE);
schedule();
@@ -124,8 +124,8 @@
if (!test_bit(ATM_VF_HASQOS,&vcc->flags)) return -EBADFD;
vcc->local = *addr;
vcc->reply = WAITING;
- sigd_enq(vcc,as_bind,NULL,NULL,&vcc->local);
add_wait_queue(&vcc->sleep,&wait);
+ sigd_enq(vcc,as_bind,NULL,NULL,&vcc->local);
while (vcc->reply == WAITING && sigd) {
set_current_state(TASK_UNINTERRUPTIBLE);
schedule();
@@ -169,12 +169,13 @@
!vcc->qos.rxtp.traffic_class) return -EINVAL;
vcc->remote = *addr;
vcc->reply = WAITING;
+ add_wait_queue(&vcc->sleep,&wait);
sigd_enq(vcc,as_connect,NULL,NULL,&vcc->remote);
if (flags & O_NONBLOCK) {
+ remove_wait_queue(&vcc->sleep,&wait);
sock->state = SS_CONNECTING;
return -EINPROGRESS;
}
- add_wait_queue(&vcc->sleep,&wait);
error = 0;
while (vcc->reply == WAITING && sigd) {
set_current_state(TASK_INTERRUPTIBLE);
@@ -243,8 +244,8 @@
/* let server handle listen on unbound sockets */
if (test_bit(ATM_VF_SESSION,&vcc->flags)) return -EINVAL;
vcc->reply = WAITING;
- sigd_enq(vcc,as_listen,NULL,NULL,&vcc->local);
add_wait_queue(&vcc->sleep,&wait);
+ sigd_enq(vcc,as_listen,NULL,NULL,&vcc->local);
while (vcc->reply == WAITING && sigd) {
set_current_state(TASK_UNINTERRUPTIBLE);
schedule();
@@ -313,8 +314,8 @@
}
/* wait should be short, so we ignore the non-blocking flag */
new_vcc->reply = WAITING;
- sigd_enq(new_vcc,as_accept,old_vcc,NULL,NULL);
add_wait_queue(&new_vcc->sleep,&wait);
+ sigd_enq(new_vcc,as_accept,old_vcc,NULL,NULL);
while (new_vcc->reply == WAITING && sigd) {
set_current_state(TASK_UNINTERRUPTIBLE);
schedule();
@@ -347,8 +348,8 @@
DECLARE_WAITQUEUE(wait,current);
vcc->reply = WAITING;
- sigd_enq2(vcc,as_modify,NULL,NULL,&vcc->local,qos,0);
add_wait_queue(&vcc->sleep,&wait);
+ sigd_enq2(vcc,as_modify,NULL,NULL,&vcc->local,qos,0);
while (vcc->reply == WAITING && !test_bit(ATM_VF_RELEASED,&vcc->flags)
&& sigd) {
set_current_state(TASK_UNINTERRUPTIBLE);
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH][ATM] svc's possibly race with sigd
2003-05-05 22:53 [PATCH][ATM] svc's possibly race with sigd chas williams
@ 2003-05-06 6:43 ` David S. Miller
0 siblings, 0 replies; 2+ messages in thread
From: David S. Miller @ 2003-05-06 6:43 UTC (permalink / raw)
To: chas; +Cc: linux-kernel
From: chas williams <chas@locutus.cmf.nrl.navy.mil>
Date: Mon, 5 May 2003 18:53:30 -0400
pretty sure the author wanted the wait queues in place before
talking to sigd which might cause a wakeup on the vcc in question.
this race is VERY unlikely, since going down to user space and back
is quite slow.
Applied, thanks chas.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2003-05-06 7:38 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-05-05 22:53 [PATCH][ATM] svc's possibly race with sigd chas williams
2003-05-06 6:43 ` David S. Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).