[Bug nptl/25847] pthread_cond_signal failed to wake up pthread_cond_wait due to a bug in undoing stealing

[carlos at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25847

--- Comment #80 from Carlos O'Donell <carlos at redhat dot com> ---
(In reply to Florian Weimer from comment #79)
> (In reply to Ismail Donmez from comment #78)
> > Hi,
> > 
> > (In reply to Florian Weimer from comment #77)
> > > Debian could also make sure that they
> > > integrate the re-initialization patch into their version of Berkeley DB.
> > 
> > Not with Debian, but could you share a link/commit to this patch? Thanks!
> 
> I believe it's this one (but there are other downstream patches related to
> locking):
> https://gitlab.com/redhat/centos-stream/rpms/libdb/-/blob/c9s/db-5.3.28-
> condition_variable.patch?ref_type=heads
> 
> This does not apply to the AGPL versions of Berkeley DB, I think. They no
> longer use process-shared condition variables.

Adding my notes here for clarity.

The following branches have the fix:

release/2.41/master
release/2.40/master
release/2.39/master
release/2.38/master

The fix has not been backported to release/2.37/master or earlier.

There are 2 outstanding patch series from Wind River to backport them to 2.36
and 2.35, which we haven't committed:

https://inbox.sourceware.org/libc-stable/20250723101827.930871-1-sunilkumar.dora@windriver.com/
https://inbox.sourceware.org/libc-stable/20250722150321.2251739-1-sunilkumar.dora@windriver.com/

Both contain the initializer fix for increased compatibility:
https://inbox.sourceware.org/libc-stable/20250723101827.930871-11-sunilkumar.dora@windriver.com/

Even with that you might still see issues with serialized condition variables
(which is not permitted by the standard).

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