Bug 2362863

Summary: glibc: Findings by static analyzers in Fedora 43
Product: [Fedora] Fedora Reporter: Siteshwar Vashisht <svashisht>
Component: glibcAssignee: Arjun Shankar <ashankar>
Status: ASSIGNED --- QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: medium Docs Contact:
Priority: unspecified    
Version: rawhideCC: araghuku, arjun, codonell, dj, fberat, fweimer, jlaw, josmyers, mcermak, mcoufal, mfabian, pfrankli, ralvaro, sipoyare, skolosov, suraj.ghimire7
Target Milestone: ---Flags: codonell: mirror+
Target Release: ---   
Hardware: Unspecified   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: ---
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Siteshwar Vashisht 2025-04-29 09:14:47 UTC
We are using various static analyzers to find bugs in Fedora packages. There are 704 findings[1] reported for glibc in a mass scan[2] performed on Fedora rawhide. Please review this report and either fix these findings or add them to the known-false-positives[3] repository.

[1] https://svashisht.fedorapeople.org/openscanhub/mass-scans/f43-25-Apr-2025/glibc-2.41.9000-10.fc43/scan-results.html
[2] https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/PH5MCW4SPO2D4ITTGDFLSYBSF5FWFYEL/
[3] https://github.com/openscanhub/known-false-positives


Reproducible: Always

Comment 1 Carlos O'Donell 2025-04-29 15:43:28 UTC
Thanks for running the analysis. It will certainly take us a long time to evaluate the results.

Is it feasible to mark them all as a "baseline" and use that going forward to detect changes?

Comment 2 Siteshwar Vashisht 2025-04-29 16:03:42 UTC
(In reply to Carlos O'Donell from comment #1)
> Thanks for running the analysis. It will certainly take us a long time to
> evaluate the results.
> 
> Is it feasible to mark them all as a "baseline" and use that going forward
> to detect changes?

If you only want to review findings introduced in rawhide, they can be seen on the added findings[1] page. See the `+` column in report[2] shared with the Fedora community.

[1] https://svashisht.fedorapeople.org/openscanhub/mass-scans/f43-25-Apr-2025/glibc-2.41.9000-10.fc43/added.html
[2] https://svashisht.fedorapeople.org/openscanhub/mass-scans/f43-25-Apr-2025/

Comment 3 Siteshwar Vashisht 2025-05-05 12:16:34 UTC
A significant amount of findings in the full scan report were for the benchmark tests. I have opened a pull request[1] to exclude benchmark tests path to make the reports more useful.

[1] https://github.com/openscanhub/known-false-positives/pull/17

Comment 4 Carlos O'Donell 2025-05-05 17:08:49 UTC
(In reply to Siteshwar Vashisht from comment #3)
> A significant amount of findings in the full scan report were for the
> benchmark tests. I have opened a pull request[1] to exclude benchmark tests
> path to make the reports more useful.
> 
> [1] https://github.com/openscanhub/known-false-positives/pull/17

From first principles the benchmarks should always be correct. I've commented on the github PR with alternative suggestions.

Comment 7 Carlos O'Donell 2026-02-06 14:46:41 UTC
Siteshwar,

I note that we have updated mass scans here:
https://svashisht.fedorapeople.org/openscanhub/mass-scans/

However, glibc is not in the list for the f44 scans, why might that be?

Comment 8 Siteshwar Vashisht 2026-02-06 17:04:26 UTC
(In reply to Carlos O'Donell from comment #7)
> Siteshwar,
> 
> I note that we have updated mass scans here:
> https://svashisht.fedorapeople.org/openscanhub/mass-scans/
> 
> However, glibc is not in the list for the f44 scans, why might that be?

glibc differential scan failed[1] as the base SRPM for Fedora 43 could not be built[2].

[1] https://openscanhub.fedoraproject.org/task/90933/
[2] https://openscanhub.fedoraproject.org/task/90972/log/stdout.log

Comment 9 Carlos O'Donell 2026-02-06 17:08:15 UTC
(In reply to Siteshwar Vashisht from comment #8)
> (In reply to Carlos O'Donell from comment #7)
> > Siteshwar,
> > 
> > I note that we have updated mass scans here:
> > https://svashisht.fedorapeople.org/openscanhub/mass-scans/
> > 
> > However, glibc is not in the list for the f44 scans, why might that be?
> 
> glibc differential scan failed[1] as the base SRPM for Fedora 43 could not
> be built[2].
> 
> [1] https://openscanhub.fedoraproject.org/task/90933/
> [2] https://openscanhub.fedoraproject.org/task/90972/log/stdout.log

~~~
RPM build errors:
/usr/bin/ld.bfd: cannot find /usr/lib/libatomic.so.1.2.0: No such file or directory <--[gcc]
~~~

This has since been fixed and was part of the GCC 16 transition.

commit 3ca6309421beb8eaa033da5b9c723aecf33f807d
Author: DJ Delorie <dj>
Date:   Tue Jan 6 22:07:58 2026 -0500

    Do not try to link glibc32 with libatomic (#2427390)
    
    GCC now links libatomic on an "as-needed" basis.  However, the build
    environment doesn't have the 32-bit version during our 64-bit builds,
    and the link fails because the file is missing.  Since it isn't needed
    anyway, just remove that as-needed from those builds.
    
    Since the native builds do not need libatomic (which *is* in the build
    environment), as-needing them is harmless as the not-needed-ness means
    there isn't even a DT_NEEDED for them.

Is there any way to redo the differential scan?

Comment 10 Siteshwar Vashisht 2026-02-08 22:17:24 UTC
> Is there any way to redo the differential scan?

It is documented in the Fedora wiki[1]. I have triggered another scan[2] for glibc.

[1] https://fedoraproject.org/wiki/OpenScanHub
[2] https://openscanhub.fedoraproject.org/task/97738/

Comment 11 Siteshwar Vashisht 2026-02-08 22:43:18 UTC
(In reply to Siteshwar Vashisht from comment #10)
> > Is there any way to redo the differential scan?
> 
> It is documented in the Fedora wiki[1]. I have triggered another scan[2] for
> glibc.
> 
> [1] https://fedoraproject.org/wiki/OpenScanHub
> [2] https://openscanhub.fedoraproject.org/task/97738/

The scan has failed again as this commit[1] is not there in the Fedora 43 glibc package. I have triggered another scan[1] by adding custom `-fno-link-libatomic` flag for the Fedora 43 build.

[1] https://src.fedoraproject.org/rpms/glibc/c/3ca6309421beb8eaa033da5b9c723aecf33f807d
[2] https://openscanhub.fedoraproject.org/task/97741/