Skip to content

Commit

Permalink
[IA64] Clear IRQ affinity when unregistered
Browse files Browse the repository at this point in the history
When we offline a CPU, migrate_irqs() tries to determine whether the
affinity bits of the IRQ descriptor match any of the remaining online
CPUs.  If not, it fixes up the interrupt to point somewhere else.
Unfortunately, if an IRQ is unregistered the IRQ descriptor may still
have affinity to the CPU being offlined, but the no_irq_chip handler
doesn't provide a set_affinity function.  This causes us to hit the
WARN_ON in migrate_irqs().

The easiest solution seems to be setting all the bits in the affinity
mask when the last interrupt is removed from the vector.  I hit this on
an older kernel with Xen/ia64 using driver domains (so it probably needs
more testing on upstream).  Xen essentially uses the bind/unbind
interface in sysfs to unregister a device from a driver and thus
unregister the interrupt.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
  • Loading branch information
Alex Williamson authored and Tony Luck committed Feb 5, 2007
1 parent 06f87ad commit 451fe00
Showing 1 changed file with 5 additions and 0 deletions.
5 changes: 5 additions & 0 deletions arch/ia64/kernel/iosapic.c
Original file line number Diff line number Diff line change
Expand Up @@ -925,6 +925,11 @@ iosapic_unregister_intr (unsigned int gsi)
/* Clear the interrupt controller descriptor */
idesc->chip = &no_irq_type;

#ifdef CONFIG_SMP
/* Clear affinity */
cpus_setall(idesc->affinity);
#endif

/* Clear the interrupt information */
memset(&iosapic_intr_info[vector], 0,
sizeof(struct iosapic_intr_info));
Expand Down

0 comments on commit 451fe00

Please sign in to comment.