On Sat, Apr 20, 2019 at 2:13 AM Borislav Petkov <b...@alien8.de> wrote:
>
> On Fri, Apr 19, 2019 at 10:43:03PM -0700, Cong Wang wrote:
> > With this change, although not even compiled, mcelog should still
> > receive correctable memory errors like before, even when we have
> > CONFIG_RAS_CEC=y.
> >
> > Does this make any sense to you?
>
> Yes, the answer is in the mail you snipped. Did you read it?
I read it, all of your response is based on your speculation that I
don't have CONFIG_X86_MCELOG_LEGACY=y, which is clearly
a misunderstanding.
You didn't answer my question here, because I asked you whether
the following change (PoC only) makes sense:
@ -567,12 +567,12 @@ static int mce_first_notifier(struct
notifier_block *nb, unsigned long val,
void *data)
{
struct mce *m = (struct mce *)data;
+ bool consumed;
if (!m)
return NOTIFY_DONE;
- if (cec_add_mce(m))
- return NOTIFY_STOP;
+ consumed = cec_add_mce(m);
/* Emit the trace record: */
trace_mce_record(m);
@@ -581,7 +581,7 @@ static int mce_first_notifier(struct
notifier_block *nb, unsigned long val,
mce_notify_irq();
- return NOTIFY_DONE;
+ return consumed ? NOTIFY_STOP : NOTIFY_DONE;
}
>
> Hint: disable CONFIG_RAS_CEC.
I knew disabling it could cure the problem from the beginning, please
save your own time by not repeating things we both already knew. :)
Once again, I still don't think it is the right answer, which is also why I
keep finding different solutions. I know you disagree, but you never explain
why you disagree, you speculated CONFIG_X86_MCELOG_LEGACY,
which is completely a misunderstanding.
I brought up CONFIG_X86_MCELOG_LEGACY simply to show you
how we could break mcelog _LOUDLY_ if we really decide to break it,
currently it just breaks silently. You misinterpret it as if I understand
CONFIG_RAS as a replacement for CONFIG_X86_MCELOG_LEGACY,
which is a very sad misunderstanding.
Thanks.
