Re: [PATCH][BUG] Badness in smp_call_function at arch/i386/kernel/smp.c:552

From: Zwane Mwaikambo
Date: Fri Dec 03 2004 - 02:37:13 EST


On Thu, 2 Dec 2004, Andrew Morton wrote:

> Well, sort-of.
>
> If __handle_sysrq was really a normal IRQ handler then the correct thing to
> do here is to replace spin_lock_irqsave() with spin_lock(). But
> __handle_sysrq() can also be called via /proc/sysrq-trigger and via the
> handlers of multiple interrupt sources. So we're stuck with using
> spin_lock_irqsave().
>
> However enabling interrupts as you've done menas that theoretically we
> could deadlock on sysrq_key_table_lock if another sysrq happens at the
> wrong time.
>
> Which deadlock opportunity would you prefer? ;)

Agreed, there is actually a higher chance of the smp_call_function
deadlock occuring since the __handle_sysrq one relies on another sysrq
event occuring via a different IRQ line interrupt handler, so
we would have to do sysrq via serial and then sysrq via keyboard to cause
the deadlock. Perhaps just make it a spin_trylock?

Index: linux-2.6.10-rc2-mm4/drivers/char/sysrq.c
===================================================================
RCS file: /home/cvsroot/linux-2.6.10-rc2-mm4/drivers/char/sysrq.c,v
retrieving revision 1.1.1.1
diff -u -p -B -r1.1.1.1 sysrq.c
--- linux-2.6.10-rc2-mm4/drivers/char/sysrq.c 30 Nov 2004 18:52:31 -0000 1.1.1.1
+++ linux-2.6.10-rc2-mm4/drivers/char/sysrq.c 3 Dec 2004 07:21:04 -0000
@@ -370,7 +370,10 @@ void __handle_sysrq(int key, struct pt_r
int i, j;
unsigned long flags;

- spin_lock_irqsave(&sysrq_key_table_lock, flags);
+ local_irq_save(flags);
+ if (!spin_trylock(&sysrq_key_table_lock))
+ goto bail;
+
orig_log_level = console_loglevel;
console_loglevel = 7;
printk(KERN_INFO "SysRq : ");
@@ -396,7 +399,9 @@ void __handle_sysrq(int key, struct pt_r
printk ("\n");
console_loglevel = orig_log_level;
}
- spin_unlock_irqrestore(&sysrq_key_table_lock, flags);
+ spin_unlock(&sysrq_key_table_lock);
+bail:
+ local_irq_restore(flags);
}

/*
Index: linux-2.6.10-rc2-mm4/arch/i386/kernel/reboot.c
===================================================================
RCS file: /home/cvsroot/linux-2.6.10-rc2-mm4/arch/i386/kernel/reboot.c,v
retrieving revision 1.1.1.1
diff -u -p -B -r1.1.1.1 reboot.c
--- linux-2.6.10-rc2-mm4/arch/i386/kernel/reboot.c 30 Nov 2004 18:52:19 -0000 1.1.1.1
+++ linux-2.6.10-rc2-mm4/arch/i386/kernel/reboot.c 3 Dec 2004 07:23:44 -0000
@@ -274,6 +274,8 @@ void machine_shutdown(void)
#ifdef CONFIG_SMP
int reboot_cpu_id;

+ local_irq_enable();
+
/* The boot cpu is always logical cpu 0 */
reboot_cpu_id = 0;

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/