[PATCH 2/6] nptl: Move pthread_kill to libc

[Adhemerval Zanella]

On 04/12/2020 15:24, Florian Weimer wrote:
> * Adhemerval Zanella:
> 
>> On 04/12/2020 15:17, Florian Weimer wrote:
>>> * Adhemerval Zanella via Libc-alpha:
>>>
>>>> diff --git a/nptl/libpthread-compat.c b/nptl/libpthread-compat.c
>>>> index ec02b5dccf..a0856c5586 100644
>>>> --- a/nptl/libpthread-compat.c
>>>> +++ b/nptl/libpthread-compat.c
>>>> @@ -36,6 +36,12 @@ __libpthread_version_placeholder (void)
>>>>     version or later, the placeholder symbol is not needed because
>>>>     there are plenty of other symbols which populate those later
>>>>     versions.  */
>>>> +
>>>> +#if (SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_1_2))
>>>> +compat_symbol_unique (libpthread,
>>>> +		      __libpthread_version_placeholder, GLIBC_2_0);
>>>> +#endif
>>>> +
>>>>  #if (SHLIB_COMPAT (libpthread, GLIBC_2_1_2, GLIBC_2_2))
>>>>  compat_symbol_unique (libpthread,
>>>>  		      __libpthread_version_placeholder, GLIBC_2_1_2);
>>>
>>> I think this isn't necessary yet becase there are many GLIBC_2.0
>>> symbols that keep that version alive.
>>>
>>> I assume you have used the symbol moving script, so those changes
>>> should be fine.
>>>
>>
>> Yes, and I think I have acked on the last review. I think I saw
>> one issue with riscv32, since it is base abi is the current one,
>> that I will check out.
> 
> Okay, should I commit the script as it is today?
> 
> I wanted to add something to auto-generate the new base symbol
> version, but haven't found the time to implement that.
> 

Yes, I will check the possible riscv32 issue once it is upstream.