Bug 2052047

Summary:	openblas package's static libraries cause annocheck to never exit
Product:	Red Hat Enterprise Linux 9	Reporter:	David Cantrell <dcantrell>
Component:	annobin	Assignee:	Martin Cermak <mcermak>
annobin sub component:	system-version	QA Contact:
Status:	CLOSED NOTABUG	Docs Contact:
Severity:	unspecified
Priority:	unspecified	CC:	bstinson, fweimer, jwboyer, mcermak, nickc, sipoyare, vkadlcik
Version:	CentOS Stream	Keywords:	Bugfix, Triaged
Target Milestone:	rc	Flags:	nickc: needinfo? (dcantrell)
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	No Doc Update
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2023-07-17 10:12:48 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:

Description David Cantrell 2022-02-08 15:19:39 UTC

The openblas package is unable to run through annocheck.  Using a recent build, such as https://koji.fedoraproject.org/koji/buildinfo?buildID=1890262 (openblas-0.3.19-3.fc36) I cannot get it to complete annocheck unless I exclude the static libraries.

I am invoking this via rpminspect, which is using the defaults for annocheck.  If I tell rpminspect to ignore /usr/lib/*.a and /usr/lib64/*.a for openblas, it completes fine.

It is consistent across architectures.  I at least checked aarch64, ppc64le, s390x, and x86_64 and it behaves the same way.

I am on Fedora 35 and have annobin-annocheck-10.35 installed.

Comment 1 Nick Clifton 2022-02-08 15:55:08 UTC

Hi David,

  Which openblas rpm(s) trigger this problem ?

  I tried testing openblas-0.3.19-3.fc36.x86_64.rpm and had no problems.

  I also tried testing openblas-static-0.3.19-3.fc36.x86_64.rpm.  This
  took a significant amount of time to complete an annocheck run (55 
  seconds according to the wall clock).  But then there are a *lot* of
  components in the libraries and it looks like all of them have
  problems.

  I was using annobin 10.52 so there is also a possibility that this
  is an already fixed problem.  (Although I do not remember fixing an
  infinite loop issue).

Cheers
  Nick

Comment 2 Florian Weimer 2022-02-08 16:12:49 UTC

Is this an actual infinite loop, or is it perhaps due to a large network transfer? Could this be debuginfod?

Comment 4 Nick Clifton 2023-07-17 10:12:48 UTC

Given that we have no response from the reporter, and no other reports of annocheck going into an infinite loop, I am closing this BZ.

I suspect that Florian's guess that the issue might be debuginfod related is spot on.  It should be noted that annocheck has a command line option to run without using debuginfod (-n or --no-use-debuginfod), which might help if similar problems are encountered in the future.