Re: [PATCH] x86/mce: Pay no attention to 'F' bit in MCACOD whenparsing 'UC' errors.

[Naveen N. Rao]

On 07/24/2013 11:46 AM, Chen Gong wrote:
> On Tue, Jul 23, 2013 at 03:51:14PM -0700, Tony Luck wrote:
>> Date: Tue, 23 Jul 2013 15:51:14 -0700
>> From: Tony Luck <tony.luck@xxxxxxxxx>
>> To: Linux Kernel Mailing List <linux-kernel@xxxxxxxxxxxxxxx>
>> Cc: Borislav Petkov <bp@xxxxxxx>, Chen Gong <gong.chen@xxxxxxxxxxxxxxx>,
>>   "Naveen N. Rao" <naveen.n.rao@xxxxxxxxxxxxxxxxxx>
>> Subject: Re: [PATCH] x86/mce: Pay no attention to 'F' bit in MCACOD when
>>   parsing 'UC' errors.
>>
>> Gah ... there is another bug in that unaffected thread entry.  The check
>> for
>> MCG_STATUS should be for RIPV=1 *and* EIPV=0
>>
>
> I set "MCGMASK(MCG_STATUS_RIPV, MCG_STATUS_RIPV)" becase
> I want it to cover Non-Affected Logical Processors (1,0)
> and Affected Logical Processor/Recoverable continuable (1,1).
>
> I think both of them are continuable so they should be as
> *KEEP*.

For affected logical processors, we won't be able to continue if we were in
kernel-space. Right? So, it looks like we should panic and I think this gets
covered by "Action required: unknown MCACOD" entry later on, though a more
explicit entry might help. For user-space, the next two entries cover AR.

Does the below help or am I reading this wrong?

Thanks,
Naveen

--
We have three categories under MCA Action Required (AR):
1. Unaffected threads/cpu (RIPV=1,EIPV=0): always continuable
2. Affected threads (RIPV=EIPV=1): continuable
3. Affected threads (RIPV=0): not continuable

The consolidated entry (Tony's new patch) should only cover (1).

(2) and (3) are covered for user-space by the two entries following the entry
for (1) for data load and instruction fetch errors.

(3) is covered for kernel-space by the earlier entry for "In kernel and no
restart IP" where we panic. The below patch is to make (2) explicit for
kernel-space.


Signed-off-by: Naveen N. Rao <naveen.n.rao@xxxxxxxxxxxxxxxxxx>
---
 arch/x86/kernel/cpu/mcheck/mce-severity.c |    6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/arch/x86/kernel/cpu/mcheck/mce-severity.c b/arch/x86/kernel/cpu/mcheck/mce-severity.c
index e2703520..585ddbb 100644
--- a/arch/x86/kernel/cpu/mcheck/mce-severity.c
+++ b/arch/x86/kernel/cpu/mcheck/mce-severity.c
@@ -115,6 +115,12 @@ static struct severity {
 		MCGMASK(MCG_STATUS_RIPV, MCG_STATUS_RIPV)
 		),
 	MCESEV(
+		PANIC, "Action required but kernel thread is not continuable",
+		SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR|MCI_ADDR, MCI_UC_SAR|MCI_ADDR),
+		MCGMASK(MCG_STATUS_RIPV|MCG_STATUS_EIPV, MCG_STATUS_RIPV|MCG_STATUS_EIPV),
+		KERNEL
+		),
+	MCESEV(
 		AR, "Action required: data load error in a user process",
 		SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR|MCI_ADDR|MCACOD, MCI_UC_SAR|MCI_ADDR|MCACOD_DATA),
 		USER

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/