Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220819210112.7924-1-rdunlap@infradead.org> <87o7wdkkt4.fsf@mpe.ellerman.id.au>
 <YwNovfuf3pDBh2Zk@infradead.org>
In-Reply-To: <YwNovfuf3pDBh2Zk@infradead.org>
From:   Alex Deucher <alexdeucher@gmail.com>
Date:   Mon, 22 Aug 2022 09:31:30 -0400
Message-ID: <CADnq5_NOedr2TGH0W9wVAPkrJyyJ_+eAkEUORbYvShY4UW90sw@mail.gmail.com>
Subject: Re: [PATCH] powerpc: export cpu_smallcore_map for modules
To:     Christoph Hellwig <hch@infradead.org>
Cc:     Michael Ellerman <mpe@ellerman.id.au>,
        "Gautham R . Shenoy" <ego@linux.vnet.ibm.com>,
        amd-gfx@lists.freedesktop.org, "Pan, Xinhui" <Xinhui.Pan@amd.com>,
        Felix Kuehling <Felix.Kuehling@amd.com>,
        Randy Dunlap <rdunlap@infradead.org>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        linux-kernel@vger.kernel.org,
        Christophe Leroy <christophe.leroy@csgroup.eu>,
        Tejun Heo <tj@kernel.org>, Nicholas Piggin <npiggin@gmail.com>,
        Alex Deucher <alexander.deucher@amd.com>,
        linuxppc-dev@lists.ozlabs.org,
        =?UTF-8?Q?Christian_K=C3=B6nig?= <christian.koenig@amd.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Aug 22, 2022 at 9:16 AM Christoph Hellwig <hch@infradead.org> wrote:
>
> On Mon, Aug 22, 2022 at 01:40:23PM +1000, Michael Ellerman wrote:
> > Randy Dunlap <rdunlap@infradead.org> writes:
> > > drivers/gpu/drm/amd/amdkfd/kfd_device.c calls cpu_smt_mask().
> > > This is an inline function on powerpc which references
> > > cpu_smallcore_map.
> > >
> > > Fixes: 425752c63b6f ("powerpc: Detect the presence of big-cores via "ibm, thread-groups"")
> > > Fixes: 7bc913085765 ("drm/amdkfd: Try to schedule bottom half on same core")
> >
> > That 2nd commit is not in mainline, only linux-next.
> >
> > I don't mind merging this fix preemptively, but is that SHA stable?
>
> I really do not think this has any business being exported at all.
>
> kfd_queue_work is not something that should be done in a driver.
> Something like this belongs into the workqueue core, not in an
> underdocumented helper in a random driver.
>
> Drm guys:  once again, please please work with the maintainers instead
> of just making up random stuff in the drivers.

Discussions are already ongoing with the workqueue folks.  I'll drop
this for now.

Alex