Hi,
I looked at the modifications to entry_32.S and entry_64.S generated
by perfmon in 2.6.28-rc6
and they are quite minimal:
--- a/arch/x86/kernel/entry_32.S
+++ b/arch/x86/kernel/entry_32.S
@@ -513,7 +513,7 @@ ENDPROC(system_call)
ALIGN
RING0_PTREGS_FRAME # can't unwind into user space anyway
work_pending:
- testb $_TIF_NEED_RESCHED, %cl
+ testw $(_TIF_NEED_RESCHED|_TIF_PERFMON_WORK), %cx
jz work_notifysig
work_resched:
call schedule
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index b86f332..b6f061e 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -869,7 +869,13 @@ END(error_interrupt)
ENTRY(spurious_interrupt)
apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt
END(spurious_interrupt)
-
+
+#ifdef CONFIG_PERFMON
+ENTRY(pmu_interrupt)
+ apicinterrupt LOCAL_PERFMON_VECTOR,smp_pmu_interrupt
+END(pmu_interrupt)
+#endif
+
/*
* Exception entry points.
*/
Looking at Stephen's tree, looks like there were some new macros introduced to
declare the lowest level interrupt handler for entry_64. There is not much I can
do in advance to avoid this kind of problems. The handler is part of perfmon not
some generic x86 modification. Same thing for entry_32.S.
So I am not sure I understand your point about submitting the changes to the x86
tree first.
On Tue, Nov 25, 2008 at 11:22 AM, Ingo Molnar <mingo@xxxxxxx> wrote:
>
> * Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>
>> Hi Stephane,
>>
>> Today's linux-next merge of the perfmon3 tree got a conflict in
>> arch/x86/kernel/entry_64.S between commit
>> 322648d1ba75280d62f114d47048beb0b35f5047 ("x86: include ENTRY/END in
>> entry handlers in entry_64.S") from the x86 tree and commit
>> 3511a04973eaac18ee386f0db8c109e589019511 ("perfmon: x86-64 hooks")
>> from the perfmon3 tree.
>
> very emphatic NAK on carrying such x86 changes in the perfmon tree,
> and putting that into linux-next!
>
> Those perfmon changes should be submitted to the x86 tree. The code it
> touches is sensitive and the design questions matter a lot as well.
>
> Ingo
>
--
To unsubscribe from this list: send the line "unsubscribe linux-next" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
