Re: [PATCH] x86: Fix bug about glibc.cpu.hwcaps.

[caiyinyu <caiyinyu@loongson.cn>]

在 2023/3/2 上午12:42, Noah Goldstein 写道:
> On Wed, Mar 1, 2023 at 3:11 AM caiyinyu <caiyinyu@loongson.cn> wrote:
>>
>> 在 2023/3/1 下午12:19, Noah Goldstein 写道:
>>> On Tue, Feb 28, 2023 at 9:22 PM caiyinyu <caiyinyu@loongson.cn> wrote:
>>>> 1. export GLIBC_TUNABLES=glibc.cpu.hwcaps=-AVX512
>>>> 2. Add  _dl_printf("p -- %s\n", p); just before switch(nl) in
>>>>      sysdeps/x86/cpu-tunables.c
>>>> 3. compiled and run ./testrun.sh /usr/bin/ls
>>>> you will get:
>>>>
>>>> p -- -AVX512
>>>> p -- LC_ADDRESS=en_US.UTF-8
>>>> p -- LC_NUMERIC=C
>>>> ...
>>>>
>>>> The function, TUNABLE_CALLBACK (set_hwcaps)
>>>> (tunable_val_t *valp), checks far more than it should and it
>>>> should stop at end of "-AVX512".
>>>> ---
>>>>    sysdeps/x86/cpu-tunables.c | 6 +++---
>>>>    1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/sysdeps/x86/cpu-tunables.c b/sysdeps/x86/cpu-tunables.c
>>>> index d3e1367bda..772fb0c4c2 100644
>>>> --- a/sysdeps/x86/cpu-tunables.c
>>>> +++ b/sysdeps/x86/cpu-tunables.c
>>>> @@ -107,13 +107,13 @@ TUNABLE_CALLBACK (set_hwcaps) (tunable_val_t *valp)
>>>>         NOTE: the IFUNC selection may change over time.  Please check all
>>>>         multiarch implementations when experimenting.  */
>>>>
>>>> -  const char *p = valp->strval;
>>>> +  const char *p = valp->strval, *c;
>>>>      struct cpu_features *cpu_features = &GLRO(dl_x86_cpu_features);
>>>>      size_t len;
>>>>
>>>>      do
>>>>        {
>>>> -      const char *c, *n;
>>>> +      const char *n;
>>>>          bool disable;
>>>>          size_t nl;
>>>>
>>>> @@ -283,7 +283,7 @@ TUNABLE_CALLBACK (set_hwcaps) (tunable_val_t *valp)
>>>>           }
>>>>          p += len + 1;
>>>>        }
>>>> -  while (*p != '\0');
>>>> +  while (*c != '\0');
>>>>    }
>>>>
>>>>    # if CET_ENABLED
>>>> --
>>>> 2.31.1
>>>>
>>> Can you make a bugzilla for this?
>> bugzilla:
>>
>> https://sourceware.org/bugzilla/show_bug.cgi?id=30183
>>
>>
> Can you append '[BZ #30183]' to the commit message.
> Otherwise looks good.


OK.

https://sourceware.org/pipermail/libc-alpha/2023-March/145996.html