Bug 2139495

Summary: Both dynamic-tags and not-dynamic-tags tested
Product: [Fedora] Fedora Reporter: Mark Wielaard <mjw>
Component: annobinAssignee: Nick Clifton <nickc>
Status: MODIFIED --- QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 38CC: dcantrell, fweimer, jakub, nickc, sipoyare, yahmad
Target Milestone: ---Keywords: Bugfix
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: No Doc Update
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 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:

Description Mark Wielaard 2022-11-02 17:20:23 UTC
On Fedora rawhide aarch64 both the dynamic-tags test and not-dynamic-tags test get tested.

The dynamic-tags test PASSes, the not-dynamic-tags test FAILs (as expected).
Only the first test should be run for fedora rawhide.

Example test farm run:
https://artifacts.dev.testing-farm.io/f352b4ba-0df3-4ce5-92f2-4d5ba5e71a31/

This is with the following bohdi update:
https://bodhi.fedoraproject.org/updates/FEDORA-2022-e7e405733d

For the aarch binaries from this build:
https://koji.fedoraproject.org/koji/buildinfo?buildID=2083327

Comment 1 Nick Clifton 2022-11-02 17:46:50 UTC
David - could this be an rpminspect bug ?  Or a libannocheck bug ?

Annocheck is not supposed to enable both the dynamic-tags and the not-dynamic-tags test at the same time.  But if rpminspect is using libannocheck then maybe this is why both tests are being enabled.

Comment 2 Nick Clifton 2022-11-03 09:31:34 UTC
(In reply to Nick Clifton from comment #1)
 
> Annocheck is not supposed to enable both the dynamic-tags and the
> not-dynamic-tags test at the same time.  But if rpminspect is using
> libannocheck then maybe this is why both tests are being enabled.

I have just checked.  Libannocheck's enable_all_tests() function does
enable both the dynamic-tags and not-dynamic-tags test, so if rpminspect
is using this function then that would explain the bug.  I have a local
fix for this issue, but I am little bit concerned because the test run
appears to show annocheck being run as a separate process.  If that is
the case, then fixing libannocheck is not going to help.

Comment 3 David Cantrell 2022-11-07 14:31:49 UTC
I have not merged my rpminspect branch that uses libannocheck yet, so what runs in gating for the rpminspect's annocheck inspection is the fork and exec of annocheck(1).

The above information is useful though.  Nick and I discussed annocheck profiles and what I have put together is rather than calling enable_all_tests(), I first enable the profile if one maps to the product release string.  That is configured on the rpminspect side via new config file settings.  Basically all of the Fedora product releases will enable the "rawhide" annocheck profile while RHEL releases (and CentOS) will get the corresponding elX profile enabled.  After that, rpminspect will process any annocheck command line arguments from the rpminspect config file.  This is an unfortunate side effect I have to deal with on the rpminspect side because I've already established this as "API".  rpminspect will translate the command line arguments it finds in the rpminspect config file to enable or disable calls for the corresponding test.

I plan to merge the branch this week.  I lost some time due to rpmdiff concerns from another team with regards to migration over to rpminspect as well as a conference last week that I was attending.

Nick, in my libannocheck branch for rpminspect, if I enable the "rawhide" profile for Fedora builds, will that do the right thing with regards to enabling/disabling the correct checks?

Comment 4 Nick Clifton 2022-11-07 17:06:39 UTC
(In reply to David Cantrell from comment #3)
> I have not merged my rpminspect branch that uses libannocheck yet, so what
> runs in gating for the rpminspect's annocheck inspection is the fork and
> exec of annocheck(1).

Darn - that means that there is a real problem with annocheck.


> Nick, in my libannocheck branch for rpminspect, if I enable the "rawhide"
> profile for Fedora builds, will that do the right thing with regards to
> enabling/disabling the correct checks?

Yes, it should.

Comment 5 Nick Clifton 2022-11-07 18:13:58 UTC
(Might be) Fixed in: annobin-10.92-1.fc38

Comment 6 Ben Cotton 2023-02-07 15:13:33 UTC
This bug appears to have been reported against 'rawhide' during the Fedora Linux 38 development cycle.
Changing version to 38.