Re: [PATCH 0/3] sunrpc removal

[Florian Weimer <fweimer@redhat.com>]

* H. J. Lu:

> On Wed, Jul 8, 2020 at 5:06 AM Florian Weimer via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
>>
>> This patch series polishes Petr Vorel's patches.  I think they are ready
>> for inclusion now.
>>
>> I forced the removal for x86-64, and found a bunch of build issues:
>>
>> * bindresvport and the NSS RPC functions need to preserved.  The easiest
>>   way to do that is to move their implementation to the inet
>>   subdirectory, which is what the first patch does.
>>
>> * If sunrpc is gone, __libc_thread_freeres should not call
>>   __rpc_thread_destroy anymore.
>>
>> * The cut-off point is glibc 2.31 for the sunrpc removal, not glibc
>
> Did you mean glibc 2.32?

Sorry, it's after 2.31.  2.31 is the version that goes into the
Makefile.  It's off by one compared to the compat_symbol version
unfortunately, due to internal complications.

>>   2.28.  This affects the csky port (ABI baseline 2.29).
>>
>> The --enable-obsolete-nsl patch deletes more things in this version.
>>
>> I think we should delay merging the --enable-obsolete-rpc removal until
>> the ARC port is merged, along with the required abilist adjustments.
>> But the first two patches do not change the ARC ABI, and could go in
>> now.
>>
>
> I haven't looked at the last 2 patches.  But I don't think they should
> wait for the ARC port if they are good to go in.  Rebase a port shouldn't
> be that difficult.

I don't have a strong opinion on this.

Thanks,
Florian