This is the mail archive of the cygwin mailing list for the Cygwin 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: failure notice


On 2018-04-13 23:00, R0b0t1 wrote:
> On Fri, Apr 13, 2018 at 10:51 PM, Brian Inglis
> <Brian.Inglis@systematicsw.ab.ca> wrote:
>> On 2018-04-13 21:10, Jeffrey Walton wrote:
>>>> On Fri, Apr 13, 2018 at 8:36 PM, Jeffrey Walton <noloader@gmail.com> wrote:
>>>>> On Fri, Apr 13, 2018 at 7:51 AM, Corinna Vinschen
>>>>> <corinna-cygwin@cygwin.com> wrote:
>>>>>> On Apr 12 23:01, Jeffrey Walton wrote:
>>>>>>> I'm working on an AMD A6-9220 and seeing unusual results from
>>>>>>> /proc/cpuinfo. I think this may be an issue with the latest Cygwin. It
>>>>>>> may be present in earlier versions, too.
>>>>>>> Russinovich's coreinfo is shown below
>>>>>>> (https://docs.microsoft.com/en-us/sysinternals/downloads/coreinfo).
>>>>>>> Notice /proc/cpuinfo is missing aesni, pclmul, rdrand, SSE4.1, SSE4.2,
>>>>>>> AVX, etc.
>>>>>> Note that, in theory, cpuinfo has to be extended for each new CPU
>>>>>> generation.  That's a lot of work for marginal gain (Cygwin's not a real
>>>>>> kernel) so I'm doing this only very seldomly.
>>>>>> Patches welcome, of course!

I noticed that AMD cpuinfo lost cache size info at some point in the past.

>>>>> Thanks Corinna. I think I found the file of interest at fhandler_proc.cc.
>>>>> Whitespace is a bit off. It is a mix of tabs and space:
>>>>>       if (features1 & (1 << 0))
>>>>>     print (" fpu");
>>>>>       if (features1 & (1 << 1))
>>>>>     print (" vme");
>>>>> Should I perform a whitespace check-in before things begin? Or can you
>>>>> knock it out?
>>>> The attachment is pp 572-74 from AMD's Programmer's Guide, Volume 3,
>>>> General-Purpose and System Instructions (December 2017)
>>>> (https://support.amd.com/TechDocs/24594.pdf). I believe it has the all
>>>> the necessary information.
>>>> Are there any objections to using it?

How else could AMD cpuid function results be decoded?
See also Appendix E Obtaining Processor Information Via the CPUID Instruction.

>>> Any ideas how to get this through? It is a three page PDF extracted from the
>>> AMD manual. It has the necessary information for the CPUID updates.
>> Just include the link to Appendix D section 2 from the ToC:
>> https://support.amd.com/TechDocs/24594.pdf#G14.931059

PDF readers e.g. {mu,q,x}pdf, gv, gsview, normally have ways to extract pages in
other formats; otherwise copy and paste can be used, preferably to a spreadsheet
program to handle tables properly, before saving in another format.

> He may be concerned about the longterm availability of the referenced passages.

That document has snapshots on the Wayback machine at archive.org; but see
Appendix E Obtaining Processor Information Via the CPUID Instruction ibid.
https://support.amd.com/TechDocs/24594.pdf#G15.998445
"The information in this appendix supersedes the contents of the CPUID
Specification, order #25481, which is now obsolete."

One of the nice features of the amd64 cpuid instruction is that the architecture
is augmented to add meaning to formerly undefined or reserved functions, fields,
or bits, that can be used efficiently as specified, unlike the adhoc approaches
and lookups required for some i64 info.

-- 
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple


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