[Intel-gfx] [PATCH v2] drm: Fix HDCP failures when SRM fw is missing

[Sean Paul]

On Wed, Apr 29, 2020 at 12:20 PM Ramalingam C <ramalingam.c at intel.com> wrote:
>
> On 2020-04-29 at 10:46:29 -0400, Sean Paul wrote:
> > On Wed, Apr 29, 2020 at 10:22 AM Ramalingam C <ramalingam.c at intel.com> wrote:
> > >
> > > On 2020-04-29 at 09:58:16 -0400, Sean Paul wrote:
> > > > On Wed, Apr 29, 2020 at 9:50 AM Ramalingam C <ramalingam.c at intel.com> wrote:
> > > > >
> > > > > On 2020-04-14 at 15:02:55 -0400, Sean Paul wrote:
> > > > > > From: Sean Paul <seanpaul at chromium.org>
> > > > > >
> > > > > > The SRM cleanup in 79643fddd6eb2 ("drm/hdcp: optimizing the srm
> > > > > > handling") inadvertently altered the behavior of HDCP auth when
> > > > > > the SRM firmware is missing. Before that patch, missing SRM was
> > > > > > interpreted as the device having no revoked keys. With that patch,
> > > > > > if the SRM fw file is missing we reject _all_ keys.
> > > > > >
> > > > > > This patch fixes that regression by returning success if the file
> > > > > > cannot be found. It also checks the return value from request_srm such
> > > > > > that we won't end up trying to parse the ksv list if there is an error
> > > > > > fetching it.
> > > > > >
> > > > > > Fixes: 79643fddd6eb ("drm/hdcp: optimizing the srm handling")
> > > > > > Cc: stable at vger.kernel.org
> > > > > > Cc: Ramalingam C <ramalingam.c at intel.com>
> > > > > > Cc: Sean Paul <sean at poorly.run>
> > > > > > Cc: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
> > > > > > Cc: Maxime Ripard <mripard at kernel.org>
> > > > > > Cc: Thomas Zimmermann <tzimmermann at suse.de>
> > > > > > Cc: David Airlie <airlied at linux.ie>
> > > > > > Cc: Daniel Vetter <daniel at ffwll.ch>
> > > > > > Cc: dri-devel at lists.freedesktop.org
> > > > > > Signed-off-by: Sean Paul <seanpaul at chromium.org>
> > > > > >
> > > > > > Changes in v2:
> > > > > > -Noticed a couple other things to clean up
> > > > > > ---
> > > > > >
> > > > > > Sorry for the quick rev, noticed a couple other loose ends that should
> > > > > > be cleaned up.
> > > > > >
> > > > > >  drivers/gpu/drm/drm_hdcp.c | 8 +++++++-
> > > > > >  1 file changed, 7 insertions(+), 1 deletion(-)
> > > > > >
> > > > > > diff --git a/drivers/gpu/drm/drm_hdcp.c b/drivers/gpu/drm/drm_hdcp.c
> > > > > > index 7f386adcf872..910108ccaae1 100644
> > > > > > --- a/drivers/gpu/drm/drm_hdcp.c
> > > > > > +++ b/drivers/gpu/drm/drm_hdcp.c
> > > > > > @@ -241,8 +241,12 @@ static int drm_hdcp_request_srm(struct drm_device *drm_dev,
> > > > > >
> > > > > >       ret = request_firmware_direct(&fw, (const char *)fw_name,
> > > > > >                                     drm_dev->dev);
> > > > > > -     if (ret < 0)
> > > > > > +     if (ret < 0) {
> > > > > > +             *revoked_ksv_cnt = 0;
> > > > > > +             *revoked_ksv_list = NULL;
> > > > > These two variables are already initialized by the caller.
> > > >
> > > > Right now it is, but that's not guaranteed. In the ret == 0 case, it's
> > > > pretty common for a caller to assume the called function has
> > > > validated/assigned all the function output.
> > > Ok.
> > > >
> > > > > > +             ret = 0;
> > > > > Missing of this should have been caught by CI. May be CI system always
> > > > > having the SRM file from previous execution. Never been removed. IGT
> > > > > need a fix to clean the prior SRM files before execution.
> > > > >
> > > > > CI fix shouldn't block this fix.
> > > > > >               goto exit;
> > > > > > +     }
> > > > > >
> > > > > >       if (fw->size && fw->data)
> > > > > >               ret = drm_hdcp_srm_update(fw->data, fw->size, revoked_ksv_list,
> > > > > > @@ -287,6 +291,8 @@ int drm_hdcp_check_ksvs_revoked(struct drm_device *drm_dev, u8 *ksvs,
> > > > > >
> > > > > >       ret = drm_hdcp_request_srm(drm_dev, &revoked_ksv_list,
> > > > > >                                  &revoked_ksv_cnt);
> > > > > > +     if (ret)
> > > > > > +             return ret;
> > > > > This error code also shouldn't effect the caller(i915)
> > > >
> > > > Why not? I'd assume an invalid SRM revocation list should probably be
> > > > treated as failure?
> > > IMHO invalid SRM revocation need not be treated as HDCP authentication
> > > failure.
> > >
> > > First of all SRM need not supplied by all players. and incase, supplied
> > > SRM is not as per the spec, then we dont have any list of revoked ID.
> > > with this I dont think we need to fail the HDCP authentication. Until we
> > > have valid list of revoked IDs from SRM, and the receiver ID is matching
> > > to one of the revoked IDs, I wouldn't want to fail the HDCP
> > > authentication.
> > >
> >
> > Ok, thanks for the explanation. This all seems reasonable to me.
> >
> > Looks like this can be applied as-is, right?
> Yes.
>

Applied to drm-misc-fixes

Sean

> Thanks,
> Ram
>
> > I'll review the patch you
> > posted so we can ignore the -ve return values.
> >
> > Thanks for the review!
> >
> > Sean
> >
> > > -Ram
> > > >
> > > >
> > > > > hence pushed a
> > > > > change https://patchwork.freedesktop.org/series/76730/
> > > > >
> > > > > With these addresed.
> > > > >
> > > > > LGTM.
> > > > >
> > > > > Reviewed-by: Ramalingam C <ramalingam.c at intel.com>
> > > > > >
> > > > > >       /* revoked_ksv_cnt will be zero when above function failed */
> > > > > >       for (i = 0; i < revoked_ksv_cnt; i++)
> > > > > > --
> > > > > > Sean Paul, Software Engineer, Google / Chromium OS
> > > > > >