Re: [Xen-devel] [PATCH v6 11/12] x86/microcode: Synchronize late microcode loading

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 13.03.19 at 06:02, <chao.gao@xxxxxxxxx> wrote:
> On Tue, Mar 12, 2019 at 05:07:51PM -0700, Raj, Ashok wrote:
>>On Mon, Mar 11, 2019 at 03:57:35PM +0800, Chao Gao wrote:
>>> +    if ( cpu == cpumask_first(per_cpu(cpu_sibling_mask, cpu)) )
>>> +        ret = microcode_update_cpu();
>>
>>Does ret have any useful things on if the update failed? Doesn't seem 
>>to be used before you overwrite later in collect_cpu_info()?
> 
> It has the reason of failure on error. Actally, there are two reasons:
> one is no patch of newer revision, the other is we tried to update but
> the microcode revision didn't change. I can check this return value and
> print more informative message to admin. And furthermore, for the
> latter, we can remove the ucode patch from caches to address Roger's
> concern expressed in comments to patch 4 & 5.

Btw, I'm not sure removing such ucode from the cache is appropriate:
It may well apply elsewhere, unless there's a clear indication that the
blob is broken. So perhaps there needs to be special casing of -EIO,
which gets returned when the ucode rev reported by the CPU after
the update does not match expectations.

Jan



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel