This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: buildbot is green!
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Carlos O'Donell <carlos at redhat dot com>
- Cc: Roland McGrath <roland at hack dot frob dot com>, "GNU C. Library" <libc-alpha at sourceware dot org>
- Date: Wed, 9 Sep 2015 10:30:38 +0000
- Subject: Re: buildbot is green!
- Authentication-results: sourceware.org; auth=none
- References: <20150904224438 dot 4D5382C3B1E at topped-with-meat dot com> <55EFA899 dot 70604 at redhat dot com> <55EFAC58 dot 80204 at redhat dot com>
On Tue, 8 Sep 2015, Carlos O'Donell wrote:
> LC_ALL=C readelf -W -d /mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build/rt/librt.so > /mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build/rt/librt.so.dynT
> test -s /mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build/rt/librt.so.dynT
> mv -f /mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build/rt/librt.so.dynT /mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build/rt/librt.so.dyn
> make[1]: *** [tests] Error 1
> make[1]: Target `check' not remade because of errors.
> make[1]: Leaving directory `/mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/glibc'
> make: *** [check] Error 2
> make: Leaving directory `/mnt/b/glibc-ubuntu-trusty-slave2/glibc-i686-linux/build/build'
>
> Which is elf/Makefile:
>
> 945 $(all-built-dso:=.dyn): %.dyn: %
> 946 @rm -f $@T
> 947 LC_ALL=C $(READELF) -W -d $< > $@T
> 948 test -s $@T
> 949 mv -f $@T $@
>
> It looks like PARALLELMFLAGS is "" so this is -j1 output, my own logs
> show interleaved output (I normally run -j4 or higher).
>
> Therefore the failure was in the `mv`?
If the failure was in the mv I'd expect
make[1]: *** [librt.so.dyn] Error 1
or similar, along with an error message from mv. An error from [tests]
with no previous such message would suggest an error in one of the
commands for "tests". Except that doesn't make sense based on the log
shown (and none of those commands use @ to disable showing the command).
> Which is odd, unless it's the sequence of >, test, and mv that cause an fs-level
> failure. We could run with -vf to get a verbose output?
I'd suggest using make -d (debugging). That would produce huge output
(tens of MB at least, possibly hundreds of MB or more) but hopefully
include enough information to trace what went wrong.
I don't know the right way to get the buildbot to run some tests with make
-d until this error reoccurs.
--
Joseph S. Myers
joseph@codesourcery.com