Re: [PATCH v2 0/4] livepatch: Add garbage collection for shadow variables

From: Marcos Paulo de Souza <mpdesouza@suse.de>
To: Petr Mladek <pmladek@suse.com>
Cc: Marcos Paulo de Souza <mpdesouza@suse.com>,
	linux-kernel@vger.kernel.org, live-patching@vger.kernel.org,
	jpoimboe@redhat.com, joe.lawrence@redhat.com
Subject: Re: [PATCH v2 0/4] livepatch: Add garbage collection for shadow variables
Date: Mon, 23 Jan 2023 14:33:31 -0300	[thread overview]
Message-ID: <20230123173331.2rvelrrbkaitw56r@daedalus> (raw)
In-Reply-To: <Y2D4ZgWqB0E9viPy@alley>

On Tue, Nov 01, 2022 at 11:43:50AM +0100, Petr Mladek wrote:
> On Wed 2022-10-26 16:41:18, Marcos Paulo de Souza wrote:
> > Hello,
> > 
> > This is the v2 of the livepatch shadow GC patches. The changes are minor since
> > nobody asked for for big code changes.
> > 
> > Changes from v1:
> > * Reworked commit messages (Petr)
> > * Added my SoB which was missing in some patches, or the ordering was wrong. (Josh)
> > * Change __klp_shadow_get_or_use to __klp_shadow_get_or_add_locked and add a comment (Petr)
> > * Add lockdep_assert_held on __klp_shadow_get_or_add_locked (Petr)
> >   about it's meaning (Petr)
> > * CCing LKML (Josh)
> > 
> > Some observations:
> > * Petr has reviewed some of the patches that we created. I kept the Reviewed-by
> >   tags since he wrote the patches some time ago and now he reviewed them again
> >   on the ML.
> > * There were questions about possible problems about using klp_shadow_types
> >   instead of using ids, but Petr already explained that internally it still uses
> >   the id to find the correct livepatch.
> > * Regarding the possibility of multiple patches use the same ID, the problem
> >   already existed before. Petr suggested using a "stringified" version using
> >   name and id, but nobody has commented yet. I can implement such feature in a
> >   v3 if necessary.
> > 
> > Marcos Paulo de Souza (2):
> >   livepatch/shadow: Introduce klp_shadow_type structure
> >   livepatch/shadow: Add garbage collection of shadow variables
> > 
> > Petr Mladek (2):
> >   livepatch/shadow: Separate code to get or use pre-allocated shadow
> >     variable
> >   livepatch/shadow: Separate code removing all shadow variables for a
> >     given id
> 
> From my POV, the patchset is ready for pushing upstream.

Petr, what do you think about merging the first two patches, since they just
cleanups and simplifications?

> 
> Well, we need to get approval from kpatch-build users. Joe described
> possible problems in replay for v3, see
> https://lore.kernel.org/r/b5fc2891-2fb0-4aa7-01dd-861da22bb7ea@redhat.com
> 
> Best Regards,
> Petr