On 04/04/2025 8:48 am, Jan Beulich wrote:
> On 03.04.2025 19:57, Andrew Cooper wrote:
>> Bloat-o-meter reports:
>>
>> add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-29 (-29)
>> Function old new delta
>> _probe_mask_msr 99 94 -5
>> init_amd 2418 2394 -24
>>
>> but that's because .fixup doesn't contain sized/typed symbols. This also
>> drops two "mov -EFAULT, %reg; jmp ...;" sequences too. The net saving is
>> -50.
>>
>> wrmsr_amd_safe()'s return value is only checked against 0 (if at all), and
>> because of this, the compiler can now avoid manifesting 0/-EFAULT entirely,
>> and the %[fault] label simply lands on the right basic block.
>>
>> Convert to Xen style while rewriting.
>>
>> No functional change.
>>
>> Signed-off-by: Andrew Cooper <andrew.coop...@citrix.com>
>> ---
>> CC: Jan Beulich <jbeul...@suse.com>
>> CC: Roger Pau Monné <roger....@citrix.com>
>>
>> Slightly RFC. We use -EIO elsewhere for this purpose, and nothing in this
>> logic cares.
> {rd,wr}msr_safe() both use -EFAULT. What's "elsewhere" here?
Oh, that would be Linux. Sorry, too much time spent staring at the same
logic in different codebases.
>
>> I was pleasently surprised by the manifestation of -EFAULT going away
>> entirely.
> I fear I don't understand this, given the -EFAULT is still there in the
> new code.
I meant about what the optimiser can do. It hadn't occurred to me that
that was a valid transformation.
>
> Irrespective of these remarks:
> Reviewed-by: Jan Beulich <jbeul...@suse.com>
Thanks.
~Andrew
