summaryrefslogtreecommitdiffstats
path: root/arch/ia64/include/asm
diff options
context:
space:
mode:
authorTony Luck <tony.luck@intel.com>2013-03-20 10:30:15 -0700
committerTony Luck <tony.luck@intel.com>2013-04-02 09:37:06 -0700
commitd303e9e98fce56cdb3c6f2ac92f626fc2bd51c77 (patch)
tree80421d00fe3c7c7502124e641c8918f129e90876 /arch/ia64/include/asm
parent96edc754aa714e51d2044af91b96cc7420c5cb01 (diff)
Fix initialization of CMCI/CMCP interrupts
Back 2010 during a revamp of the irq code some initializations were moved from ia64_mca_init() to ia64_mca_late_init() in commit c75f2aa13f5b268aba369b5dc566088b5194377c Cannot use register_percpu_irq() from ia64_mca_init() But this was hideously wrong. First of all these initializations are now down far too late. Specifically after all the other cpus have been brought up and initialized their own CMC vectors from smp_callin(). Also ia64_mca_late_init() may be called from any cpu so the line: ia64_mca_cmc_vector_setup(); /* Setup vector on BSP */ is generally not executed on the BSP, and so the CMC vector isn't setup at all on that processor. Make use of the arch_early_irq_init() hook to get this code executed at just the right moment: not too early, not too late. Reported-by: Fred Hartnett <fred.hartnett@hp.com> Tested-by: Fred Hartnett <fred.hartnett@hp.com> Cc: stable@kernel.org # v2.6.37+ Signed-off-by: Tony Luck <tony.luck@intel.com>
Diffstat (limited to 'arch/ia64/include/asm')
-rw-r--r--arch/ia64/include/asm/mca.h1
1 files changed, 1 insertions, 0 deletions
diff --git a/arch/ia64/include/asm/mca.h b/arch/ia64/include/asm/mca.h
index 43f96ab18fa..8c709616871 100644
--- a/arch/ia64/include/asm/mca.h
+++ b/arch/ia64/include/asm/mca.h
@@ -143,6 +143,7 @@ extern unsigned long __per_cpu_mca[NR_CPUS];
extern int cpe_vector;
extern int ia64_cpe_irq;
extern void ia64_mca_init(void);
+extern void ia64_mca_irq_init(void);
extern void ia64_mca_cpu_init(void *);
extern void ia64_os_mca_dispatch(void);
extern void ia64_os_mca_dispatch_end(void);