Comment 26 for bug 2042902

Revision history for this message
Andreas Hasenack (ahasenack) wrote :

@Artur,

a) you had a previous debdiff for mantic which used version 1.2.9-1ubuntu3.1, which is what we expect for an SRU to mantic, but now in the mantic unapproved queue I see that you changed the version to 1.2.9-1ubuntu4 again.

The upgrade path between ubuntu releases still works:
alsa-ucm-conf | 1.2.9-1ubuntu3 | mantic | source
alsa-ucm-conf | 1.2.9-1ubuntu4 | mantic/unapproved/9e381c3 | source
alsa-ucm-conf | 1.2.10-1ubuntu1 | noble | source

But since you were asked to change this once in comment #8, and did so in comment #10, I would ask that we stick to the correct version like you did before already :)

I could fix that for you, but I have other questions:

b) In the test plan:
> 1. $ alsactl init
> alsa-lib main.c:779:(execute_sequence) exec '/bin/rm -rf /var/lib/alsa/card0.conf.d' failed (exit code
> 1)

> (note: assuming that failed to remove the empty directory is not a critical problem)

First of all, thanks a lot for noticing this!

I checked the alsa-lib main.c:779 code in jammy, and it does cause that function to exit with an error. What implications that has to the rest of the initialization I'm not sure:

            if (ignore_error == false && err != 0) {
                uc_error("exec '%s' failed (exit code %d)", s->data.exec, err);
                goto __fail;
            }
...
      __fail:
    free(cdev);
    return err;

Did you get this error in all ubuntu releases with the patched package? What about on systems which do NOT have this particular audio card, but another one? I.e., would we regress those?

Whether it's a big problem or not, depends on who is calling alsactl init, and what they do when this command fails.

I briefly searched in alsa-lib where it would be asking for that rm -rf, but couldn't find it. I also checked git history, and didn't see relevant mentions of that.

Do you have a better idea of what's going on with that rm -rf, why it's being attempted on an empty directory, and if that's the case, why not ignore the error?