This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 1/2] arm: Implement memcpy ifunc selection in C
On 09/10/2017 09:00, Joseph Myers wrote:
> On Mon, 9 Oct 2017, Adhemerval Zanella wrote:
>
>> There are two different issues here:
>>
>> 1. My understanding if for 1. is for gcc mainline we are aiming to *not*
>> support glibc builds support for gnu indirection function
>> (--enable-gnu-indirect-function). I think this is too restrictive and
>> since it is possible to support glibc build whether gcc has it supports
>> should not prevent this specific support on glibc.
>
> I believe the goal should be to phase out support for multiarch glibc
> built by compilers without ifunc attribute support for that architecture.
> (Meaning the configure test for multiarch support would end up requiring
> the attribute support, say once we require GCC >= 8 to build glibc.)
> Using the attribute gives better debug info, and we want to avoid having
> too many arbitrarily different ways to build glibc.
>
I agree with you, but currently glibc build just fail with gcc mainline
without ifunc attribute support for such architectures. What about to
add a configure test for glibc master to check for such failure on
configure (mismatch alias for ifunc resolvers)?