RE: [GIT PULL] x86/microcode for v3.9-rc1

From: Yu, Fenghua
Date: Thu Feb 28 2013 - 19:09:33 EST


> -----Original Message-----
> From: Borislav Petkov [mailto:bp@xxxxxxxxx]
> Sent: Thursday, February 28, 2013 3:53 PM
> To: Yu, Fenghua; H. Peter Anvin
> Cc: H. Peter Anvin; Linus Torvalds; Ingo Molnar; Linux Kernel Mailing
> List; Thomas Gleixner; Yinghai Lu
> Subject: Re: [GIT PULL] x86/microcode for v3.9-rc1
>
> On Thu, Feb 28, 2013 at 11:28:06PM +0000, Yu, Fenghua wrote:
> >> gcc is warning that the function is using lots of stack. In the
> >> context that it is running in this is most likely not a problem
> >> given how small the overrun is, but it might be worthwhile to see if
> >> there is anything which can be moved out to static storage or some
> >> other variant.
> >>
> >> Static storage is tricky to use in this context since it runs in
> >> flat linear mode (without paging, and therefore without the +3 GB
> >> offset) on 32 bits.
> >
> > The errors might be related to the arrays defined
> > mc_saved_tmp[MAX_UCODE_COUNT].
> >
> > Could you send your .config to me so that I can reproduce the issue?
> > I don't see the issue in my build environment and in Fengguang's test
> > environment.
>
> Ok, forget it. It was some local .config file corruption which caused
> include/generated/autoconf.h and include/config/auto.conf to have a
> line
>
> CONFIG_FRAME_WARN=1024
>
> which would cause the warnings.
>
> The 1024 ceiling value is also consistent with the warnings complaining
> about something being > 1024 bytes.
>
> Default CONFIG_FRAME_WARN on x86_64 is 2048 which explains why those
> warnings never trigger on 64-bit.
>
> So, we all can relax ourselves, especially I :-)
>
> Thanks.
>
> --
> Regards/Gruss,
> Boris.
>
> Sent from a fat crate under my desk. Formatting is fine.
> --

Ok. Agree, 1024 is too small. Nice to know that:)

Merci.

-Fenghua
èº{.nÇ+‰·Ÿ®‰­†+%ŠËlzwm…ébëæìr¸›zX§»®w¥Š{ayºÊÚë,j­¢f£¢·hš‹àz¹®w¥¢¸ ¢·¦j:+v‰¨ŠwèjØm¶Ÿÿ¾«‘êçzZ+ƒùšŽŠÝj"ú!¶iO•æ¬z·švØ^¶m§ÿðà nÆàþY&—