Re: GDB 17.2 release update -- 2026-04-25

[Andrew Burgess <aburgess@redhat.com>]

Joel Brobecker <brobecker@adacore.com> writes:

> Hi everyone,
>
> Quick update on the 17.2 release, where the target date is next weekend:
>
>     Sat-Sun May 02-03
>
> As of now, we have 2 PRs left to fix in our TODO list, normally
> in good hands. Please let us know if there are some other last-minute
> issues you think we should include in the 17.2 corrective release.
>
> Otherwise, as soon as the two PRs are fixed, I plan to publish
> the 17.2 release the first weekend that follows the fix.
>
> Here is a quick recap of the changes since last week:
>
> Fixed Since the Previous Update:
> --------------------------------
>
>  * [Thiago] build/33747
>    Incompatible with MUSL libc: no member named 'c_ospeed' in 'termios'
>    https://sourceware.org/bugzilla/show_bug.cgi?id=33747
>
> Added Since the Last Update:
> ----------------------------
>
>  * [TomDV] tui/31449
>    TUI crash with debuginfod support
>    https://sourceware.org/bugzilla/show_bug.cgi?id=31449
>
>         https://sourceware.org/pipermail/gdb-patches/2026-April/226551.html
>
>  * [TomDV] tui/33794
>    [gdb/tui] Recursive readline use
>    https://sourceware.org/bugzilla/show_bug.cgi?id=33794

On these last two issues, I took a look at Tom's patches, and ended up
posting an alternative proposal (to both issues) here:

  https://inbox.sourceware.org/gdb-patches/cover.1777645161.git.aburgess@redhat.com

I let Tom know, and he seems generally happy with this proposal, see:

  https://inbox.sourceware.org/gdb-patches/4300deb1-765a-48fe-9d3f-69122748547d@suse.de

though this wasn't a full approval for the patches.

My take is that these two bugs should NOT hold up the 17.2 release.
They have both existed ever since the debuginfod prompt was added back
in GDB 12, so holding up this point release seems pretty arbitrary.

In addition, triggering this bug is (IMHO) pretty unlikely (though not
impossible), but more importantly, it can only trigger early in a debug
session.  Once the user has done something to trigger the debuginfod
prompt they are "safe" from this bug, which means the pain of living
with this is 17.2 (as with all earlier releases) is pretty low; yes GDB
might crash, but the user will not have done much in their session by
that point, so they'll not have lost much.

Anyway, these are just my thoughts, in case this makes the go / no-go
choice any easier.

Thanks,
Andrew