Re: Regression with patch "kbuild: rpm-pkg: Generate debuginfo package manually"

[Nathan Chancellor <nathan@kernel.org>]

On Mon, Feb 09, 2026 at 03:25:54PM -0700, Nathan Chancellor wrote:
> On Thu, Feb 05, 2026 at 05:24:16PM -0700, Nathan Chancellor wrote:
> > On Thu, Feb 05, 2026 at 05:56:20PM -0600, Steve French wrote:
> > > I noticed with the recent patch below:
> > > 
> > >     commit 62089b804895e845f82e132ea9d46a1fc53ed5a7
> > >     Author: Nathan Chancellor <nathan@kernel.org>
> > >     Date:   Wed Jan 21 15:29:15 2026 -0700
> > > 
> > >        kbuild: rpm-pkg: Generate debuginfo package manually
> > > 
> > > that all of our test builds were failing to build kernel RPMs for
> > > 6.19-rc8 and later (see e.g.
> > > http://osmb311-linux-testing.southcentralus.cloudapp.azure.com/#/builders/1/builds/99/steps/9/logs/stdio).
> > 
> > Sorry for the breakage :(
> > 
> > Can you provide me with the .config? I want to see if I can reproduce
> > this locally.
> ...
> > I wonder if we are relying on something in a newer rpm version that is
> > not available on an older version that is present in RHEL.
> 
> Ooookay, after much gnashing of teeth investigating different
> environments and such, it looks like my change implicitly relies on some
> improvements to debuginfo package generation that are in rpm 4.20 or
> newer (Fedora 41 or newer), which is why I have not seen any issues up
> until this point. I am not entirely sure why yet though but I wanted to
> provide a small update that I am still looking into this. I plan to send
> the main 7.0 Kbuild pull request then I should be able to focus more
> intently on this.

Just to close the loop for future travelers:

  https://lore.kernel.org/20260210-kbuild-fix-debuginfo-rpm-v1-0-0730b92b14bc@kernel.org/

Cheers,
Nathan