This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v8 6/8] nptl: Add abilist symbols for C11 threads
- From: Florian Weimer <fweimer at redhat dot com>
- To: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>, libc-alpha at sourceware dot org
- Date: Fri, 13 Jul 2018 19:49:43 +0200
- Subject: Re: [PATCH v8 6/8] nptl: Add abilist symbols for C11 threads
- References: <1517591084-11347-1-git-send-email-adhemerval.zanella@linaro.org> <1517591084-11347-7-git-send-email-adhemerval.zanella@linaro.org> <97b42cb3-e76a-ac71-65cd-c1f6d8c404aa@redhat.com> <90dc04ab-81df-b0d7-a086-d00ea5cc979d@linaro.org>
On 07/13/2018 05:49 PM, Adhemerval Zanella wrote:
Thanks for all the reviews, I have updated my personal branch [1] with
all you required changes:
Thank you for your patience. This is looking really good now.
- ABI fixes for mips-gnu-linux-soft.
I can confirm that these are fixed.
- pthread_create / pthread_join cast inconsistency.
- sysdeps/nptl/threads.h moved to nptl/threads.h.
- thrd_sleep is now a cancellable entrypoint.
- once_flag and ONCE_FLAG_INIT and __ONCE_ALIGNMENT definition.
Should the type be a struct?
I think ONCE_FLAG_INIT needs to be a compound literal, not an
initializer, at least that's how I read the standard (“which expands to
a value that can be used to initialize an object”).
- thread_local guards for C++.
Missing indentation of the #define. The manual could mention that for
C++, C++11 or later needs to be used to get the thread_local keyword
(not the macro).
- TSS_DTOR_ITERATIONS assert check.
I am not sure about the thread error code mapping to errno, it will
required tying POSIX error code to C11 use some internal mechanism
It will most like require generate a platform-specific file
auto-generated from errno definitions (more build complexity), and
I am not sure if this micro-optimization will really yields any
measurable gain.
I'm fine with leaving it as is.
Thanks,
Florian