Skip to contentSkip to conversation

swsusp: fix suspending with mysqld

PATCH Pavel Machek 3 participants 5 messages Oct 4, 2004 – Oct 5, 2004

lore
Share & tools
lore
#0 lore raw

Hi!

mysqld does signal calls in pretty tight loop, and swsusp is not able to stop processes in such case. This should fix it. Please apply,
Pavel

kernel/signal.c +3 -2
Index: linux/kernel/signal.c
===================================================================
@@ -1483,8 +1483,7 @@
 	unsigned long flags;
 	struct sighand_struct *psig;
 
-	if (sig == -1)
-		BUG();
+	BUG_ON(sig == -1);
 
  	/* do_notify_parent_cldstop should have been called instead.  */
  	BUG_ON(tsk->state & (TASK_STOPPED|TASK_TRACED));
@@ -2260,6 +2259,8 @@
 			ret = -EINTR;
 	}
 
+	if (current->flags & PF_FREEZE)
+		refrigerator(1);
 	return ret;
 }
 
Signature · 2 lines
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!
#1 lore raw
Quoted 5 lines · Pavel Machek

Hi!

mysqld does signal calls in pretty tight loop, and swsusp is not able to stop processes in such case. This should fix it. Please apply,
Pavel

Pavel,

I applied your patch to 2.6.9-rc3. Unfortunately, now the system doesn't suspend anymore, it comes back almost immediately:

Stopping tasks: ====================================================| Freeing memory: ...................................|
radeonfb: suspending to state: 3...
agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0.
agpgart: Putting AGP V2 device at 0000:00:00.0 into 0x mode
agpgart: Putting AGP V2 device at 0000:01:00.0 into 0x mode
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1d.1 to 64
PCI: Setting latency timer of device 0000:00:1d.1 to 64
PCI: Setting latency timer of device 0000:00:1d.2 to 64
PCI: Setting latency timer of device 0000:00:1d.2 to 64
PCI: Setting latency timer of device 0000:00:1d.7 to 64
ACPI: PCI interrupt 0000:00:1f.1[A] -> GSI 10 (level, low) -> IRQ 10
ACPI: PCI interrupt 0000:00:1f.5[B] -> GSI 10 (level, low) -> IRQ 10
PCI: Setting latency timer of device 0000:00:1f.5 to 64
ACPI: PCI interrupt 0000:00:1f.6[B] -> GSI 10 (level, low) -> IRQ 10
PCI: Setting latency timer of device 0000:00:1f.6 to 64
radeonfb: resumed !
Restarting tasks... done

The system never reaches suspend.

Jan

Signature · 3 lines
BOFH excuse #323:

Your processor has processed too many instructions.  Turn it off immediately, do not type any commands!!
#2 lore raw

Hi!

Quoted 5 lines
Quoted 2 lines

mysqld does signal calls in pretty tight loop, and swsusp is not able to stop processes in such case. This should fix it. Please apply,

I applied your patch to 2.6.9-rc3. Unfortunately, now the system doesn't suspend anymore, it comes back almost immediately:

And it did work before that patch? I fail to see how this patch could have broken anything.
Pavel

Signature · 1 line
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         
#3 lore raw
Quoted 10 lines · Pavel Machek

Hi!

Quoted 5 lines
Quoted 2 lines

mysqld does signal calls in pretty tight loop, and swsusp is not able to stop processes in such case. This should fix it. Please apply,

I applied your patch to 2.6.9-rc3. Unfortunately, now the system doesn't suspend anymore, it comes back almost immediately:

And it did work before that patch? I fail to see how this patch could have broken anything.

Suspending worked (besides mysql, which I had to kill manually). With this patch it doesn't.

Jan

Signature · 3 lines
BOFH excuse #418:

Sysadmins busy fighting SPAM.
#4 lore raw
In reply to Pavel Machek

kernel/signal.c line 2265

Quoted 13 lines · Pavel Machek

mysqld does signal calls in pretty tight loop, and swsusp is not able to stop processes in such case. This should fix it. Please apply,

...

kernel/signal.c +2 -0
@@ -2260,6 +2259,8 @@
 			ret = -EINTR;
 	}
 
+	if (current->flags & PF_FREEZE)
+		refrigerator(1);
 	return ret;
 }

This seems hacky, and arbitrary. I mean, why add this in sys_rt_sigtimedwait()? Because that is the syscall which mysql appears to use? What if a different application were to be using a different syscall? Do we add a refrigerator() call there too?

Or am I missing something?