don't ship libgc with -Wl,-Bsymbolic-functions

[Dima Pasechnik]

This option leads to various issues, see the upstream discussion on
https://github.com/ivmai/bdwgc/issues/321
Steps to reproduction are given there.

We ran into it in Sagemath, see https://trac.sagemath.org/ticket/30629
Basically, ecl built with this libgc on WSL is broken.

[Ivan Maidanski (ivmai)]

This is fixed in upstream v8.0.6

[1] https://github.com/ivmai/bdwgc/commit/cf699a0b377494a3ae08cabc067f57485f775c9a
[2] https://github.com/ivmai/bdwgc/releases/tag/v8.0.6

[Nathan Lee (nathan.lee)]

I'm not sure if we want to risk backporting the fix, but attaching a diff that is a cherry-pick from the fix itself (see Ivan's comment).

There was a merge conflict where the HOST_ANDROID logic had not been removed in v7.6.4 yet. Digging the git logs, it was removed in https://github.com/ivmai/bdwgc/commit/7744b5d5dba1bd13abe6ebca258c335ec557f3f6, which points to https://github.com/ivmai/bdwgc/issues/259.

I'm not sure if it is desired to leave in the HOST_ANDROID logic for that particular use case. If this affects ubuntu packages, we can use a patch were this logic is only applied if HOST_ANDROID is not defined (but I suppose this means the patch won't help for users of libgc built with HOST_ANDROID).

-----

I came back to this because we have had a couple reports that the ppa:inkscape.dev/stable experiences the issue outside of WSL (https://gitlab.com/inkscape/inkscape/-/issues/1637#note_1010909501), though I'm not able to replicate it myself.

[Ubuntu Foundations Team Bug Bot (crichton)]

The attachment "__data_start-_end-pair.diff" seems to be a patch. If it isn't, please remove the "patch" flag from the attachment, remove the "patch" tag, and if you are a member of the ~ubuntu-reviewers, unsubscribe the team.

[This is an automated message performed by a Launchpad user owned by ~brian-murray, for any issues please contact him.]