[Bug default/30394] New: abidiff "execution should not have reached this point"

[Bug default/30394] New: abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

            Bug ID: 30394
           Summary: abidiff "execution should not have reached this point"
           Product: libabigail
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: ben at demerara dot io
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Version:
https://sourceware.org/git/?p=libabigail.git;a=commit;h=6ba26ed6ae74cf54c87cde1df163e5f086bbcfce

Problem pipeline:
https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/4183660486
Problem tarballs:
https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/4183660486/artifacts/browse/libabigail-tars/

There are two separate issues within the same job, I can create a separate
ticket if preferred, though they both seem to reference the same function.

Excerpts from job logs:

    in build_ir_node_from_die at: ../../src/abg-dwarf-reader.cc:15901:
execution should not have reached this point!

    abidiff: ../../src/abg-dwarf-reader.cc:15731:
abigail::ir::type_or_decl_base_sptr
abigail::dwarf::build_ir_node_from_die(reader&, Dwarf_Die*,
abigail::ir::scope_decl*, bool, size_t, bool, bool): Assertion `__abg_cond__'
failed.

I can't reproduce this issue when run from my laptop. Perhaps this isn't an
issue on amd64?

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #1 from Ben Brown <ben at demerara dot io> ---
Hmm, well there goes my theory about amd64:
https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/4186839558

Could something in the headers be tripping it up? I don't currently include
them in the generated tarballs, I'll set that up now. Otherwise there's
possibly something in our build environment at fault.

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #2 from Ben Brown <ben at demerara dot io> ---
(In reply to Ben Brown from comment #1)
> Hmm, well there goes my theory about amd64:
> https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/4186839558

Oops, this is a completely different component with a different error:

    abidiff: ../../src/abg-dwarf-reader.cc:9202: std::string
abigail::dwarf::die_qualified_type_name(const reader&, const Dwarf_Die*,
size_t): Assertion `__abg_cond__' failed.

Sorry for the noise in this ticket.

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #3 from Ben Brown <ben at demerara dot io> ---
Created attachment 14866
  --> https://sourceware.org/bugzilla/attachment.cgi?id=14866&action=edit
gdb backtrace

Frustratingly the error message is different when built locally from the
"execution should not have reached this point" message when built via CI.

However I've managed to extract a backtrace using gdb which will hopefully be
of some use.

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

[Bug default/30394] abidiff "execution should not have reached this point"

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #4 from dodji at redhat dot com ---
Can I get my hands on the directories with the header files please?

I mean, would it be possible to get these in the tarball too?

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #5 from Ben Brown <ben at demerara dot io> ---
(In reply to dodji from comment #4)
> Can I get my hands on the directories with the header files please?
> 
> I mean, would it be possible to get these in the tarball too?

The tarballs available here include the headers:
https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/4271633824/artifacts/browse/libabigail-tars/

They include *all* of the headers in the platform and SDK, as such it's a bit
too large to upload. It's on my to-do list to optimise this somewhat.

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

--- Comment #6 from Ben Brown <ben at demerara dot io> ---
Created attachment 14884
  --> https://sourceware.org/bugzilla/attachment.cgi?id=14884&action=edit
sqlite debug info

As discussed on IRC, copy this to /usr/lib/debug/dwz/components/sqlite.bst/
when comparing the sqlite libraries to reproduce the issue.

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

[Bug default/30394] abidiff "execution should not have reached this point"

[ben at demerara dot io]

https://sourceware.org/bugzilla/show_bug.cgi?id=30394

Ben Brown <ben at demerara dot io> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #7 from Ben Brown <ben at demerara dot io> ---
This has since been resolved, thanks to discussions over IRC.

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