This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] sysconf: Use conservative default for _SC_NPROCESSORS_ONLN [BZ #21542]


On Fri, Jun 9, 2017 at 5:35 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 06/09/2017 02:32 PM, H.J. Lu wrote:
>> On Fri, Jun 9, 2017 at 5:28 AM, Florian Weimer <fweimer@redhat.com> wrote:
>>> On 06/09/2017 02:26 PM, H.J. Lu wrote:
>>>> On x86, we compute the number of threads which share L3
>>>> cache in a package via CPUID.  We should use this info as
>>>> default on x86.
>>>
>>> That would only allow us to return a larger value than 2 in this case,
>>> wouldn't it?
>>>
>>> I don't know if the additional complexity is worth it.  The 1/2
>>
>> The information is there.  We just need to store the result in
>> cpu_features and return it in __default_nprocs ().
>
> It still needs another sysdeps override for __default_nprocs ().  This
> is what I meant.
>
>>> distinction is what patterns, I don't think it's possible to get this
>>> data reliably from CPUID.
>>
>> No.  Reliable data needs ACPI table.
>
> Exactly.  A single-core, single-thread processor might still be running
> in a multi-socket configuration.  And with virtualization, it's not safe
> to list CPU names, either.
>
> Florian

Just a suggestion.  I have no strong opinion.

-- 
H.J.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]