This bug was fixed in the package libvirt - 3.6.0-1ubuntu5~cloud0
---
libvirt (3.6.0-1ubuntu5~cloud0) xenial-pike; urgency=medium
.
* New update for the Ubuntu Cloud Archive.
.
libvirt (3.6.0-1ubuntu5) artful; urgency=medium
.
* d/p/u/gnulib-getopt-posix-Fix-build-failure-wh
OpenStack regression tests have passed successfully.
pike-proposed with next charms:
==
Totals
==
Ran: 102 tests in 1904.7099 sec.
- Passed: 93
- Skipped: 9
- Expected Fail: 0
- Un
Hello ChristianEhrhardt, or anyone else affected,
Accepted libvirt into pike-proposed. The package will build now and be
available in the Ubuntu Cloud Archive in a few hours, and then in the
-proposed repository.
Please help us by testing this new package. To enable the -proposed
repository:
s
This bug was fixed in the package libvirt - 3.6.0-1ubuntu5
---
libvirt (3.6.0-1ubuntu5) artful; urgency=medium
* d/p/u/gnulib-getopt-posix-Fix-build-failure-when-using-ac_cv_head.patch:
fix FTBFS with glibc 2.26 (LP: #1718668)
-- Christian Ehrhardt Thu, 28 Sep
2017 08:18:10
Being a pure bugfix upload (and required for any latter service of
libvirt in Artful and UCA-Pike) it is ok to upload this fix into Artful
despite the current Freezes in place.
Regression checks were good, so uploading 3.6.0-1ubuntu5 now.
Seeing it in unapproved [1] so upload was ok.
[1]:
https:/
Discussion ended in a suggested fix in [1].
Builds are fine on all arches now [2], given the changes I'm running a shorter
set of regression checks to be sure (no test would be too few, but no need for
ALL on all-arch given an this is an optarg change).
Will update here and upload after tests ar
We are closing in on the upstream discussion and this is now
reproducible with glibc/gnulib along without libvirt. Holding back an
ugly fix for now until we know the right way to do it.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
h
Worked on all architectures, since the upstream discussion is going on
I'm giving it a day if there is a better fix, otherwise I'll submit the
interim fix to fix it for the time being.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
ht
Upstream discussion at [1] and a preliminary fix building and testing
atm.
[1]: https://www.redhat.com/archives/libvir-
list/2017-September/msg01039.html
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/
So summarizing, while I fail to see the way to a fix I can now quite well
summarize the issue and provide steps to reproduce.
I'll go to the devel mailing list with that:
Hi,
there seems to be an incompatibility to the last glibc due to [1].
Eventually this breaks gnulib unittests (and maybe mor
On zesty:
.. /usr/include/unistd.h
... /usr/include/features.h
/usr/include/x86_64-linux-gnu/sys/cdefs.h
. /usr/include/x86_64-linux-gnu/bits/wordsize.h
/usr/include/x86_64-linux-gnu/gnu/stubs.h
. /usr/include/x86_64-linux-gnu/gnu/stubs-64.h
... /usr/include/x86_64-linux-gnu/bits/
I simplified the code to be somewhat readable.
One can on zesty remove the path to reach ../../gnulib/lib/getopt.h and co and
get the bad behavior to get to the systems unistd.h and run with that.
On Artful it is that behavior no matter if the path is added or not.
Since the source is the same th
Unfortunately gnulib as-from-git builds fine even from the hash it was in
libvirt.
/sigh
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1718668
Title:
Libvirt FTBFS in Artful on x86
To manage notif
./gnulib-tool --test getopt-* all three working
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1718668
Title:
Libvirt FTBFS in Artful on x86
To manage notifications about this bug go to:
https://bug
Well that sounded too good, but it is already applied in what I merged for
artful.
It seems there is something new for glibc 2.26 we have now.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1718668
Ti
commit e3461d1c21a99bcef1b8826f710434e0ffb5adea
Author: Paul Eggert
Date: Sun Jun 11 15:53:09 2017 -0700
getopt-posix: port to glibc 2.25.90
Problem reported by Daniel P. Berrange in:
http://lists.gnu.org/archive/html/bug-gnulib/2017-06/msg3.html
* lib/getopt-pfx-core.h
As always when you know what to look for one can trace further.
config.h and config.log has no massive difference that would indicate the issue.
But before the linking already on the preprocessor step it is broken.
So we can ignore the linker step.
The good case has:
# 91 "../../../../gnulib/tests
Artful:
symbol=getopt; lookup in file=./debian/build/gnulib/tests/test-getopt-posix
[0]
symbol=getopt; lookup in file=/lib/x86_64-linux-gnu/libpthread.so.0 [0]
symbol=getopt; lookup in file=/lib/x86_64-linux-gnu/libc.so.6 [0]
binding file ./debian/build/gnulib/tests/test-getopt-posix [0
Position slightly changed getopt.c, line 1188 -> getopt.c, line 735
But way more important the "good case" does not hit the getopt at all.
The backtrace finds the true reason which is that it stopped using its gnulib
implementation.
The configure still detects to use the local gnulib thing and se
Build with -O0 for debugging and -E to check precompiler
gcc -DHAVE_CONFIG_H -DEXEEXT=\"\" -I. -I../../../../gnulib/tests -I../..
-DIN_LIBVIRT_GNULIB_TESTS=1 -I. -I../../../../gnulib/tests -I../..
-I../../../../gnulib/tests/../.. -I../../gnulib/lib
-I../../../../gnulib/tests/../../gnulib/lib -g
grep __GETOPT_PREFIX debian/build/config.log | uniq
| #define __GETOPT_PREFIX rpl_
#define __GETOPT_PREFIX rpl_
The actual config defines the symbol in both cases.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad
Building the same libvirt source on zetsy works as it did befor eon artful.
So likely one of the later lib updates.
Will use the zesty build as a comparison where in configure and co it
might be different.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is su
Generated debian/build/config.h has:
debian/build/config.h:2942:#define __GETOPT_PREFIX rpl_
Didn't it before?
Unfortunately not in build logs e.g. config.log has that.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.laun
gcc -DHAVE_CONFIG_H -DEXEEXT=\"\" -I. -I../../../../gnulib/tests -I../..
-DIN_LIBVIRT_GNULIB_TESTS=1 -I. -I../../../../gnulib/tests -I../..
-I../../../../gnulib/tests/../.. -I../../gnulib/lib
-I../../../../gnulib/tests/../../gnulib/lib -g -O2 -c -o test-getline.o
../../../../gnulib/tests/tes
So it checks for something not meant to be the way it is [1]
I'd assume the action to actually check is wrong.
That depends on the var "posixly", maybe that changed.
It is defined as:
bool posixly = !!getenv ("POSIXLY_CORRECT");
/* See comment in getopt.c:
glibc gets a LSB-compliant getop
The tests checks that argv isn't touched by getopt.
There are multiple calls which work before the breaking one:
The calls set different options:
setenv ("POSIXLY_CORRECT", "1", 1);
Works
then
unsetenv ("POSIXLY_CORRECT");
fails
In this env getopt_loop moves the args around:
(gdb) p argv
$14 = {
That will do to recreate:
autopkgtest -o test-libvirt-build --shell-fail --apt-upgrade
libvirt_3.6.0-1ubuntu5~ppa01.dsc -- qemu ~/work/autopkgtest-artful-amd64.img
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad
** Tags added: artful ftbfs
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1718668
Title:
Libvirt FTBFS in Artful on x86
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubun
28 matches
Mail list logo
