[Bug default/32794] New: [Release 2.6] abg-ir.cc:8794: abigail::interned_string abigail::ir::get_type_name(const abigail::ir::type_base*, bool, bool): Assertion `__abg_cond__' failed.

[romain.geissler at amadeus dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=32794

            Bug ID: 32794
           Summary: [Release 2.6] abg-ir.cc:8794: abigail::interned_string
                    abigail::ir::get_type_name(const
                    abigail::ir::type_base*, bool, bool): Assertion
                    `__abg_cond__' failed.
           Product: libabigail
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: default
          Assignee: unassigned at sourceware dot org
          Reporter: romain.geissler at amadeus dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Salut Dodji,

One of my user reported me today this failure with libabigail 2.6:

04:50:06  Comparing the ABIs of the current component with version 31-0-2-0
located at /remote/tmp/rnd-css/bms_workflow_replica/ra/raf/db/31-0-2-0
04:50:06  abidiff cmd:
/opt/1A/toolchain/x86_64-v21/build-pack/default/bin/abidiff
--no-unreferenced-symbols --headers-dir1
/remote/tmp/rnd-css/bms_workflow_replica/ra/raf/db/31-0-2-0/include
--headers-dir2 /home/jenkins/workspace/REVOBE_raf_PR-252/db/include
/remote/tmp/rnd-css/bms_workflow_replica/ra/raf/db/31-0-2-0/lib/x86_64-v21/Release/librafKitDb.so
/home/jenkins/workspace/REVOBE_raf_PR-252/db/lib/x86_64-v21/Release/librafKitDb.so
04:50:06  
04:50:07  abidiff:
/workdir/BUILD_OS_PACK/src/libabigail-2.6/src/abg-ir.cc:8794:
abigail::interned_string abigail::ir::get_type_name(const
abigail::ir::type_base*, bool, bool): Assertion `__abg_cond__' failed.

As you know now what we call "toolchain & Middleware Pack" inside Amadeus, this
is  happening with toolchains I have delivered already 2 or 3 weeks ago, and
it's the first time since then a user reports me a problem, so it seems rather
rare. This toolchain is based on gcc 11.

For now I have just these Jenkins logs, I don't have the actual binaries, so
this bug report is quite useless. If I can get my hands on the two binaries and
can reproduce, I fear we will hit the issue that these binaries are both too
big to be added in this bugzilla, and are actually proprietary code that I
can't publish on a public bugzilla. How would you like to proceed then ? I open
a Red Hat case where I can upload big binaries and send you the link via mail ?

Cheers,
Romain

-- 
You are receiving this mail because:
You are on the CC list for the bug.