Re: [Xen-devel] [PATCH] xen/x86: Move microcode into its own directory

[Jan Beulich <jbeulich@xxxxxxxx>]

On 19.03.2020 11:41, Andrew Cooper wrote:
> On 19/03/2020 09:59, Jan Beulich wrote:
>> On 19.03.2020 10:52, Andrew Cooper wrote:
>>> On 19/03/2020 09:21, Jan Beulich wrote:
>>>> On 18.03.2020 22:05, Andrew Cooper wrote:
>>>>> Split the existing asm/microcode.h in half, keeping the per-cpu cpu_sig
>>>>> available to external users, and moving everything else into private.h
>>>>>
>>>>> Take the opportunity to trim and clean up the include lists for all 3 
>>>>> source
>>>>> files, all of which include rather more than necessary.
>>>>>
>>>>> No functional change.
>>>>>
>>>>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
>>>> Acked-by: Jan Beulich <jbeulich@xxxxxxxx>
>>>> albeit preferably with ...
>>>>
>>>>> ---
>>>>>  xen/arch/x86/Makefile                              |  4 +--
>>>>>  xen/arch/x86/microcode/Makefile                    |  3 ++
>>>>>  xen/arch/x86/{microcode_amd.c => microcode/amd.c}  | 12 ++++----
>>>>>  xen/arch/x86/{microcode.c => microcode/core.c}     | 15 +++-------
>>>>>  .../x86/{microcode_intel.c => microcode/intel.c}   |  9 ++----
>>>>>  .../microcode.h => arch/x86/microcode/private.h}   | 19 ++++---------
>>>> ... these going into xen/arch/x86/cpu/microcode/. Thoughts?
>>> TBH, I've always found the cpu/ directory redundant.  Everything in
>>> arch/x86 is part of the CPU, and these days, even drivers/passthrough is
>>> part of the CPU.
>> I'm surprised of you saying so - certainly e.g. memory management
>> stuff also interfaces with the CPU, but is imo still helpful to be
>> separated.
> 
> I can see an argument for things like domain.c not living in cpu/, but
> where do we draw the line?
> 
> Should traps.c be considered cpu/ or not?

Perhaps partly here and there.

>  What about FPU handling?

Yes, this would belong under cpu/ imo.

>> Likewise while IOMMU stuff may today be part of the
>> CPU package, it's still not core CPU functionality imo.
> 
> Sure, for small systems, but considering it is effectively mandatory for
> a >255 cpu system, I'd no longer agree.

That still doesn't make the IOMMU part of the core CPU. Nor
is it technically impossible to run >255 CPU systems without
IOMMU, it's just not very efficient interrupt distribution
wise.

> After all, we know its not safe running an Intel system until you've
> turned on every thread's CR4.MCE, even if you don't actually want to use
> the thread.

Well, CR4.MCE and in fact all MCA handling is CPU stuff,
and hence imo validly lives under cpu/mcheck/.

Jan

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