[PATCH] swsusp: finally solve mysqld problem

[PATCH] swsusp: finally solve mysqld problem

[Rafael J. Wysocki]

Hi,

This patch from Pavel moves userland freeze signals handling into
more logical place.  It now hits even with mysqld running.

Please apply.

Greetings,
Rafael


Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Pavel Machek <pavel@suse.cz>

Index: linux-2.6.16-rc1-mm5/arch/i386/kernel/signal.c
===================================================================
--- linux-2.6.16-rc1-mm5.orig/arch/i386/kernel/signal.c
+++ linux-2.6.16-rc1-mm5/arch/i386/kernel/signal.c
@@ -582,9 +582,6 @@ static void fastcall do_signal(struct pt
 	if (!user_mode(regs))
 		return;
 
-	if (try_to_freeze())
-		goto no_signal;
-
 	if (test_thread_flag(TIF_RESTORE_SIGMASK))
 		oldset = &current->saved_sigmask;
 	else
@@ -613,7 +610,6 @@ static void fastcall do_signal(struct pt
 		return;
 	}
 
-no_signal:
 	/* Did we come from a system call? */
 	if (regs->orig_eax >= 0) {
 		/* Restart the system call - no handlers present */
Index: linux-2.6.16-rc1-mm5/arch/x86_64/kernel/signal.c
===================================================================
--- linux-2.6.16-rc1-mm5.orig/arch/x86_64/kernel/signal.c
+++ linux-2.6.16-rc1-mm5/arch/x86_64/kernel/signal.c
@@ -443,9 +443,6 @@ int do_signal(struct pt_regs *regs, sigs
 	if (!user_mode(regs))
 		return 1;
 
-	if (try_to_freeze())
-		goto no_signal;
-
 	if (!oldset)
 		oldset = &current->blocked;
 
@@ -463,7 +460,6 @@ int do_signal(struct pt_regs *regs, sigs
 		return handle_signal(signr, &info, &ka, oldset, regs);
 	}
 
- no_signal:
 	/* Did we come from a system call? */
 	if ((long)regs->orig_rax >= 0) {
 		/* Restart the system call - no handlers present */
Index: linux-2.6.16-rc1-mm5/kernel/signal.c
===================================================================
--- linux-2.6.16-rc1-mm5.orig/kernel/signal.c
+++ linux-2.6.16-rc1-mm5/kernel/signal.c
@@ -1922,6 +1922,8 @@ int get_signal_to_deliver(siginfo_t *inf
 	sigset_t *mask = &current->blocked;
 	int signr = 0;
 
+	try_to_freeze();
+
 relock:
 	spin_lock_irq(&current->sighand->siglock);
 	for (;;) {
@@ -2307,7 +2309,6 @@ sys_rt_sigtimedwait(const sigset_t __use
 
 			timeout = schedule_timeout_interruptible(timeout);
 
-			try_to_freeze();
 			spin_lock_irq(&current->sighand->siglock);
 			sig = dequeue_signal(current, &these, &info);
 			current->blocked = current->real_blocked;

Re: [PATCH] swsusp: finally solve mysqld problem

[Adrian Bunk]

On Sun, Feb 05, 2006 at 01:21:54PM +0100, Rafael J. Wysocki wrote:

> Hi,

Hi Rafael,

> This patch from Pavel moves userland freeze signals handling into
> more logical place.  It now hits even with mysqld running.
>...

I've seen this patch has been included in Linus' tree.

What exactly was this "mysqld problem" problem, and more specifically, 
is this patch 2.6.16.2 material?

> Greetings,
> Rafael
>...

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [PATCH] swsusp: finally solve mysqld problem

[Rafael J. Wysocki]

Hi,

On Saturday 25 March 2006 19:38, Adrian Bunk wrote:
> On Sun, Feb 05, 2006 at 01:21:54PM +0100, Rafael J. Wysocki wrote:
> 
> > Hi,
> 
> Hi Rafael,
> 
> > This patch from Pavel moves userland freeze signals handling into
> > more logical place.  It now hits even with mysqld running.
> >...
> 
> I've seen this patch has been included in Linus' tree.
> 
> What exactly was this "mysqld problem" problem, and more specifically, 
> is this patch 2.6.16.2 material?

I don't think it's that urgent, but actually Pavel knows the details (test case
etc.).  IIRC there was/is a problem with freezing mysqld due to the
way in which the userland freeze signals are handled without this patch.
Still mysqld is the only known (to me) application affected.

Greetings,
Rafael

Re: [PATCH] swsusp: finally solve mysqld problem

[Pavel Machek]

On So 25-03-06 19:38:17, Adrian Bunk wrote:
> On Sun, Feb 05, 2006 at 01:21:54PM +0100, Rafael J. Wysocki wrote:
> 
> > Hi,
> 
> Hi Rafael,
> 
> > This patch from Pavel moves userland freeze signals handling into
> > more logical place.  It now hits even with mysqld running.
> >...
> 
> I've seen this patch has been included in Linus' tree.
> 
> What exactly was this "mysqld problem" problem, and more specifically, 
> is this patch 2.6.16.2 material?

Suspend failed with mysqld running (for a *long* time). I believe this
patch would be too risky for 2.6.16.2.
								Pavel
-- 
Picture of sleeping (Linux) penguin wanted...