On Wed, 5 Jun 2024 at 01:30, Peter Robinson <pbrobinson@xxxxxxxxx> wrote:
>
> On Tue, 4 Jun 2024 at 23:18, Dmitry Baryshkov
> <dmitry.baryshkov@xxxxxxxxxx> wrote:
> >
> > On Wed, 5 Jun 2024 at 00:40, Peter Robinson <pbrobinson@xxxxxxxxx> wrote:
> > >
> > > Add missing MODULE_FIRMWARE for firmware in linux-firmware,
> > > this is needed for automatically adding firmware to things
> > > like initrds when the drivers are built as modules. This is
> > > useful for devices like the X13s and the RBx devices on
> > > general distros.
> > >
> > > Fixes: 5e7665b5e484b ("drm/msm/adreno: Add Adreno A690 support")
> > > Fixes: 18397519cb622 ("drm/msm/adreno: Add A702 support")
> > > Fixes: 3e7042ba87da ("drm/msm/adreno: Add ZAP firmware name to A635")
> > > Signed-off-by: Peter Robinson <pbrobinson@xxxxxxxxx>
> > > ---
> > > drivers/gpu/drm/msm/adreno/adreno_device.c | 3 +++
> > > 1 file changed, 3 insertions(+)
> > >
> > > diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c
> > > index c3703a51287b4..fede5159e7f5b 100644
> > > --- a/drivers/gpu/drm/msm/adreno/adreno_device.c
> > > +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c
> > > @@ -609,8 +609,11 @@ MODULE_FIRMWARE("qcom/a650_gmu.bin");
> > > MODULE_FIRMWARE("qcom/a650_sqe.fw");
> > > MODULE_FIRMWARE("qcom/a660_gmu.bin");
> > > MODULE_FIRMWARE("qcom/a660_sqe.fw");
> > > +MODULE_FIRMWARE("qcom/a660_zap.mbn");
> >
> > -ENOSUCHFILE. It should qcom/particular-SoC/a660_zap.mbn
>
> > > +MODULE_FIRMWARE("qcom/a702_sqe.fw");
> > > MODULE_FIRMWARE("qcom/leia_pfp_470.fw");
> > > MODULE_FIRMWARE("qcom/leia_pm4_470.fw");
> > > +MODULE_FIRMWARE("qcom/sc8280xp/LENOVO/21BX/qcdxkmsuc8280.mbn");
> >
> > I'm a bit sceptical here. Each device has its own zap MBN file (this
> > one is also a fancy named a690_zap.mbn). Do we want to list all such
> > files? Consider all the vendors, which are open-source / Linux
> > friendly, like FairPhone, OnePlus, etc.
>
> That's what's in linux-firmware, I would have thought they would be
> SoC specific as opposed to device specific, but that's not what we
> currently have, are we supposed to have devices that selectively work
> based on what firmware they have?
The firmware is signed with the manufacturer's key. End-user devices
verify the signature while loading the firmware and decline firmware
without vendor's signature.
>
> > > MODULE_FIRMWARE("qcom/yamato_pfp.fw");
> > > MODULE_FIRMWARE("qcom/yamato_pm4.fw");
--
With best wishes
Dmitry
