Re: [committed] libitm: Disable TSX on processors on which it may be broken.

[Uros Bizjak <ubizjak at gmail dot com>]

On Wed, Jan 18, 2017 at 11:08 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
> On Wed, Jan 18, 2017 at 10:48 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
>> Hello!
>>
>>> This fix follows the same approach that glibc uses to disable TSX on
>>> processors on which it is broken.  TSX can also be disabled through a
>>> microcode update on these processors, but glibc consensus is that it
>>> cannot be detected reliably whether the microcode update has been
>>> applied.  Thus, we just look for affected models/steppings.
>>>
>>> Tested on x86_64-linux (but I don't have a machine with broken TSX
>>> available).
>>>
>>>        libitm/ChangeLog
>>>
>>>        * config/x86/target.h (htm_available): Add check for some processors
>>>        on which TSX is broken.
>>
>> +      __cpuid (0, a, b, c, d);
>> +      if (b == 0x756e6547 && c == 0x6c65746e && d == 0x49656e69)
>>
>> You can use:
>>
>> #define signature_INTEL_ebx    0x756e6547
>> #define signature_INTEL_ecx    0x6c65746e
>> #define signature_INTEL_edx    0x49656e69
>>
>> defines from cpuid.h here.
>
> Actually, just provide a non-NULL second argument to __get_cpuid_max.
> It will return %ebx from cpuid, which should be enough to detect Intel
> processor.


Attached is the patch I have committed to mainline SVN after a short
off-line discussion with Torvald.

2017-01-19  Uros Bizjak  <ubizjak@gmail.com>

    * config/x86/target.h (htm_available): Determine vendor from
    __get_cpuid_max return.  Use signature_INTEL_ebx.  Cleanup.

Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.

Committed to mainline SVN.

Uros.

Attachment: m.diff.txt
Description: Text document