This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Fix misreported errno on preadv2/pwritev2 (BZ#23579)


On Tue, 23 Oct 2018, Szabolcs Nagy wrote:

> with build-many-glibcs.py in
> logs/glibcs/i686-gnu/010-glibcs-i686-gnu-check-log.txt i see
> 
> In file included from tst-preadvwritev2.c:26:
> tst-preadvwritev2-common.c: In function ‘do_test_with_invalid_iov’:
> tst-preadvwritev2-common.c:89:22: error: ‘IOV_MAX’ undeclared (first use in this function); did you mean ‘INT_MAX’?
>      struct iovec iov[IOV_MAX+1] = { 0 };
>                       ^~~~~~~
>                       INT_MAX
> tst-preadvwritev2-common.c:89:22: note: each undeclared identifier is reported only once for each function it appears in
> tst-preadvwritev2-common.c:89:18: error: unused variable ‘iov’ [-Werror=unused-variable]
>      struct iovec iov[IOV_MAX+1] = { 0 };
>                   ^~~
> cc1: all warnings being treated as errors
> make[3]: *** [../o-iterator.mk:9: /data/glibc-build-j2/build/glibcs/i686-gnu/glibc/misc/tst-preadvwritev2.o] Error 1

I guess this wasn't readily visible to people running build-many-glibcs.py 
tests because the current expected baseline is that the compilation 
testsuite doesn't have clean results on i686-gnu, ever since the C11 
threads support went in (so that this build failure appeared is only 
visible if you actually look into the log files, not if you just look at 
the summary results of build-many-glibcs.py).

It would of course be good to have C11 threads support for Hurd (sharing 
as much as possible of the code to implement C11 threads on top of 
pthreads, and the associated tests for C11 threads, with those used with 
NPTL).  Until we have that, maybe it would be possible to XFAIL the 
relevant conform/ tests in some Hurd sysdeps Makefile (with a comment 
referencing a bug open in Bugzilla about the lack of C11 threads support 
for Hurd)?  Finding regressions with build-many-glibcs.py works much 
better if the normal expected state is no failures at all.

(ia64 has also been broken with GCC mainline for over a month - 
<https://gcc.gnu.org/ml/gcc-patches/2018-09/msg01052.html>.)

-- 
Joseph S. Myers
joseph@codesourcery.com

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]