Bug 2134397
| Summary: | libtraceevent build is missing symbols | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 9 | Reporter: | Michael Petlan <mpetlan> |
| Component: | libtraceevent | Assignee: | Michael Petlan <mpetlan> |
| Status: | CLOSED ERRATA | QA Contact: | Qiao Zhao <qzhao> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 9.2 | CC: | acme, bhu, ihuguet, jbenc, kernel-general-qe, lilu, mmalik, xuwei, zsun |
| Target Milestone: | rc | Keywords: | Triaged |
| Target Release: | --- | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | libtraceevent-1.5.3-3.el9 | Doc Type: | If docs needed, set a value |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2023-05-09 08:15:04 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
The problem is not in the rebase itself, for me it works on my testing box well:
# rpmquery libtraceevent trace-cmd
libtraceevent-1.5.3-1.el9.x86_64
trace-cmd-2.9.2-9.el9.x86_64
[root@intel-sharkbay-mb-01 ~]# systemctl start trace-cmd
[root@intel-sharkbay-mb-01 ~]# systemctl status trace-cmd
● trace-cmd.service - trace-cmd Flightrecorder
Loaded: loaded (/usr/lib/systemd/system/trace-cmd.service; disabled; vendor preset: disabled)
Active: active (exited) since Fri 2022-10-14 03:21:09 EDT; 16s ago
Process: 125560 ExecStart=/usr/bin/trace-cmd start $OPTS (code=exited, status=0/SUCCESS)
Main PID: 125560 (code=exited, status=0/SUCCESS)
CPU: 54ms
The problem is in brew, since some strange issue has broken the package build. Otherwise, it would normally work.
The example mentioned in this comment works, because the libtraceevent-1.5.3-1.el9 used here is my own local "rpmbuild --rebuild" one.
Oh thanks for the info. It actually reminded me that I also met missing symbols when I update to 1.5.3 in Fedora. I checked the log and see when I disable parallel build the issue actually goes away in Fedora. So maybe you can also check if disable parallel build will fix the issue in RHEL or not https://src.fedoraproject.org/rpms/libtraceevent/c/17c3a0d638abcf502d1311dd8d50f59a7b6112b4?branch=rawhide *** Bug 2135774 has been marked as a duplicate of this bug. *** For all who have been affected and need to use trace-cmd, you can make trace-cmd work again by temporary downgrade to libtraceevent-1.1.1-8.el9 before a permanent solution is applied. Any ETA for this, please? This is blocking us from merging *all* kernel MRs touching networking, since lnst testing depends on brew builds. Posted a new MR with disabling parallel building. If that goes well, it should be solved in a day. *** Bug 2137498 has been marked as a duplicate of this bug. *** Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (libtraceevent bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2023:2465 |
Description of problem: I totally don't know how that happened, but the library built in brew is missing some API symbols. When I build the package myself via `rpmbuild --rebuild`, it is OK. The same applies for the centos-stream build in KOJI -- the build is OK. I have built a scratchbuild in brew and it is OK. So it appears that something must have been broken in brew at the time the build was being built. Version-Release number of selected component (if applicable): libtraceevent-1.5.3-1.el9 from brew How reproducible: 100% -- the library just does not contain some symbols Steps to Reproduce: 1. readelf -Ws /usr/lib64/libtraceevent.so.1.5.3 | grep tep_filter 2. 3. Actual results: no entries are found Expected results: 10 entries are found: 72: 0000000000014c80 33 FUNC GLOBAL DEFAULT 13 tep_filter_free 73: 0000000000015810 492 FUNC GLOBAL DEFAULT 13 tep_filter_copy 95: 0000000000015e40 260 FUNC GLOBAL DEFAULT 13 tep_filter_match 104: 0000000000012790 200 FUNC GLOBAL DEFAULT 13 tep_filter_remove_event 113: 0000000000013f40 59 FUNC GLOBAL DEFAULT 13 tep_filter_alloc 115: 0000000000012980 354 FUNC GLOBAL DEFAULT 13 tep_filter_compare 122: 0000000000013220 86 FUNC GLOBAL DEFAULT 13 tep_filter_strerror 160: 0000000000015160 1706 FUNC GLOBAL DEFAULT 13 tep_filter_add_filter_str 174: 0000000000012920 85 FUNC GLOBAL DEFAULT 13 tep_filter_make_string 197: 0000000000012860 93 FUNC GLOBAL DEFAULT 13 tep_filter_reset Additional info: Link to the broken BREW build: https://brewweb.engineering.redhat.com/brew/buildinfo?buildID=2178971 Link to the correct build in KOJI: https://kojihub.stream.centos.org/koji/buildinfo?buildID=25343 Link to the correct scratchbuild in BREW: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=48278346 It's also worth of noting that there were some outages in Westford, and also that previous build in brew was unsuccessful, because the i686 variant could not be linked. I was investigating it and was totally out of ideas, so I just kicked off the build again with no source changes. It passed and linked i686 libs well. However, now it seems that this build was also broken, but in a different way. Let's just bump NVR and rebuild it, since from the scratchbuild above, and from the koji build for c9s, it is highly probable, that correct library will be build this time.