diff options
| author |   Ian Campbell <Ian.Campbell@eu.citrix.com> | 2009-02-06 16:55:58 -0800 |
|---|---|---|
| committer |   Ingo Molnar <mingo@elte.hu> | 2009-02-09 12:18:22 +0100 |
| commit | 90af9514ac99f51e81682c7bec8f9fb88a17a95c (patch) | |
| tree | 479e22492c5821e4386f1a258fa6ea92cde05ad4 /include/linux/kernel.h | |
| parent | xen: make sure that softirqs get handled at the end of event processing (diff) | |
| download | wireguard-linux-90af9514ac99f51e81682c7bec8f9fb88a17a95c.tar.xz wireguard-linux-90af9514ac99f51e81682c7bec8f9fb88a17a95c.zip | |
xen: explicitly initialise the cpu field of irq_info
I was seeing a very odd crash on 64 bit in bind_evtchn_to_cpu because
cpu_from_irq(irq) was coming out as -1. I found this was coming direct
from the mk_ipi_info call.
It's not clear to me that this isn't a compiler bug (implicit
initialisation to zero of unsigned shorts in a struct not handled
correctly?).
On the other hand is it true that all event channels start of bound to
CPU 0? If not then -1 might be correct and the various other functions
should cope with this.
Signed-off-by: Ian Campbell <Ian.Campbell@eu.citrix.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Diffstat (limited to 'include/linux/kernel.h')
0 files changed, 0 insertions, 0 deletions