Commit e5e417232e7c9ecc58a77902d2e8dd46792cd092

Authored by Christian Borntraeger
Committed by Linus Torvalds
1 parent 6bffd7b57d

Fix cpu hotplug problem in softirq code

currently cpu hotplug (unplug) seems broken on s390 and likely others. On cpu
unplug the system starts to behave very strange and hangs.

I bisected the problem to the following commit:

commit 48f20a9a9488c432fc86df1ff4b7f4fa895d1183
Author: Olof Johansson <olof@lixom.net>
Date: Tue Mar 4 15:23:25 2008 -0800
	tasklets: execute tasklets in the same order they were queued

Reverting this patch seems to fix the problem.  I looked into takeover_tasklet
and it seems that there is a way to corrupt the tail pointer of the current
cpu.  If the tasklet list of the frozen cpu is empty, the tail pointer of the
current cpu points to the address of the head pointer of the stopped cpu and
not to the next pointer of a tasklet_struct.

This patch avoids the list splice of the list is empty and cpu hotplug seems
to work as the tail pointer is not corrupted.  Olof, can you look into that
patch and ACK/NACK it so Andrew can push this to Linus, if appropriate?
Please note that some lines are longer than 80 chars, but line-wrapping looked
worse that this version.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Olof Johansson <olof@lixom.net>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 1 changed file with 12 additions and 8 deletions Side-by-side Diff

... ... @@ -589,16 +589,20 @@
589 589 local_irq_disable();
590 590  
591 591 /* Find end, append list for that CPU. */
592   - *__get_cpu_var(tasklet_vec).tail = per_cpu(tasklet_vec, cpu).head;
593   - __get_cpu_var(tasklet_vec).tail = per_cpu(tasklet_vec, cpu).tail;
594   - per_cpu(tasklet_vec, cpu).head = NULL;
595   - per_cpu(tasklet_vec, cpu).tail = &per_cpu(tasklet_vec, cpu).head;
  592 + if (&per_cpu(tasklet_vec, cpu).head != per_cpu(tasklet_vec, cpu).tail) {
  593 + *(__get_cpu_var(tasklet_vec).tail) = per_cpu(tasklet_vec, cpu).head;
  594 + __get_cpu_var(tasklet_vec).tail = per_cpu(tasklet_vec, cpu).tail;
  595 + per_cpu(tasklet_vec, cpu).head = NULL;
  596 + per_cpu(tasklet_vec, cpu).tail = &per_cpu(tasklet_vec, cpu).head;
  597 + }
596 598 raise_softirq_irqoff(TASKLET_SOFTIRQ);
597 599  
598   - *__get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).head;
599   - __get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).tail;
600   - per_cpu(tasklet_hi_vec, cpu).head = NULL;
601   - per_cpu(tasklet_hi_vec, cpu).tail = &per_cpu(tasklet_hi_vec, cpu).head;
  600 + if (&per_cpu(tasklet_hi_vec, cpu).head != per_cpu(tasklet_hi_vec, cpu).tail) {
  601 + *__get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).head;
  602 + __get_cpu_var(tasklet_hi_vec).tail = per_cpu(tasklet_hi_vec, cpu).tail;
  603 + per_cpu(tasklet_hi_vec, cpu).head = NULL;
  604 + per_cpu(tasklet_hi_vec, cpu).tail = &per_cpu(tasklet_hi_vec, cpu).head;
  605 + }
602 606 raise_softirq_irqoff(HI_SOFTIRQ);
603 607  
604 608 local_irq_enable();