Bug 1073606 - FTBFS: strace fails to build/testsuite on i586
FTBFS: strace fails to build/testsuite on i586
Status: RESOLVED FIXED
Classification: openSUSE
Product: openSUSE Tumbleweed
Classification: openSUSE
Component: Basesystem
Current
Other Other
: P5 - None : Normal (vote)
: ---
Assigned To: Michael Schröder
E-mail List
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2017-12-20 07:57 UTC by Dominique Leuenberger
Modified: 2017-12-20 17:28 UTC (History)
0 users

See Also:
Found By: ---
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Dominique Leuenberger 2017-12-20 07:57:18 UTC
Link to log:

https://build.opensuse.org/package/live_build_log/openSUSE:Factory:Staging:G/strace/standard/i586

The test suite of strace fails with:

[  703s] FAIL: strace-k
[  703s] ==============
[  703s] 
[  703s] expected: "getpid f3 f2 f1 f0 main "
[  703s] result: "__libc_start_main "
[  703s] getpid()                                = 16424
[  703s]  > [vdso]() [0xcc9]
[  703s]  > /lib/libc-2.26.so(__libc_start_main+0xf3) [0x187b3]
[  703s]  > unexpected_backtracing_error [0x41c000]
[  703s] +++ exited with 0 +++
[  703s] strace-k.test: failed test: ../../strace -e getpid -k ../stack-fcall output mismatch
[  703s] 
[  703s] + exit 1
[  703s] error: Bad exit status from /var/tmp/rpm-tmp.lUii0j (%check)
Comment 1 Dominique Leuenberger 2017-12-20 12:41:52 UTC
I did a bit of digging, taking the recent changes in Tumbleweed into account. The 'root' is glibc, probably paired with some OBS speciality I don't yet understand

The situation seems to be that aggregatepac in combination with multibuild flavors seems not to do what we had before. strace actually builds now with glibc.686 in the Staging projects - but OBS picks glibc.i586 in the non-staging (and Staging A -C where we build glibc directly, not aggregate it)

The setup for Staging:D is like this:

glibc and glibc:686 are aggregated from openSUSE:Factory:Rings:0-Bootstrap to Staging:D/bootstrap-copy (as we don't want to spend the resources of bootstrap cycles in all staging projects)

obs build then picks glibc.i686; before glibc turned into multibuild, we aggregated glibc and glibc.i686 (the top-add patched linked), and obs picked glibc.i586 for building of strace.

I'm assigning this to MLS, who might have much more insight into the scheduler works of OBS; I doubt there is anything wrong with stracce / glibc per se
Comment 2 Dominique Leuenberger 2017-12-20 17:28:55 UTC
Thanks to mls we could confirm that the digging at least was all right - and a fix could be implemented (in short: glibc and glibc:i686 drifted appart with the rebuild counter, which caused this kind of nasty issue).

Added
  <bcntsynctag>glibc</bcntsynctag>

to glib's meta in Ring0 (will also add in oS:F once build is settled there)

Rebased stagings build strace successful again