dipankar@in.ibm.com: Fw: [PATCH 2.5.7] smp_call_function trivial change:
Hi Rusty,
I discussed this with Andrea long ago, but never got around
to send the patch. Better late than never.
Going by the documentation and use of _bh version of spin_lock(),
smp_call_function() is allowed to be called from BH context,
We can run into a deadlock with some locks if we do so.
This because reader-writer locks can sometimes be used optimally
by not disabling irqs while taking the reader side if only the
reader side of the lock is taken from irq context.
CPU #0 CPU #1
read_lock(&tasklist_lock)
write_lock_irq(&tasklist_lock)
[spins with interrupt disabled]
[Interrupted by BH]
smp_call_function() for BH
handler
[ doesn't take the IPI]
So, cpu #1 doesn't take the IPI and cpu #0 spinwaits
for the IPI handler to start, resulting in a deadlock.
The last time I looked, I couldn't see smp_call_function() being
called from BH context anywhere. So, there is no immediate problem.
However it seems right to correct the documentation and also not
disable BH while taking the call lock since it isn't necessary.
Patch for 2.5.7 included.
--- trivial-2.4.19-pre7/arch/i386/kernel/smp.c.orig Mon Apr 22 13:21:41 2002
+++ trivial-2.4.19-pre7/arch/i386/kernel/smp.c Mon Apr 22 13:21:41 2002
@@ -528,7 +528,7 @@
* remote CPUs are nearly ready to execute <<func>> or are or have executed.
*
* You must not call this function with disabled interrupts or from a
- * hardware interrupt handler, you may call it from a bottom half handler.
+ * hardware interrupt handler or from a bottom half handler.
*/
{
struct call_data_struct data;
@@ -544,7 +544,7 @@
if (wait)
atomic_set(&data.finished, 0);
- spin_lock_bh(&call_lock);
+ spin_lock(&call_lock);
call_data = &data;
wmb();
/* Send a message to all other CPUs and wait for them to respond */
@@ -557,7 +557,7 @@
if (wait)
while (atomic_read(&data.finished) != cpus)
barrier();
- spin_unlock_bh(&call_lock);
+ spin_unlock(&call_lock);
return 0;
}
-- Anyone who quotes me in their sig is an idiot. -- Rusty Russell. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Apr 23 2002 - 22:00:30 EST