Re: [PATCH v1] drm/xe/vf: Avoid reading media version when media GT is disabled

["Piotr Piórkowski" <piotr.piorkowski@intel.com>]

Lin, Shuicheng <shuicheng.lin@intel.com> wrote on pią [2026-sty-30 19:20:08 +0100]:
> On Fri, Jan 30, 2026 10:09 AM Shuicheng Lin wrote:
> > -----Original Message-----
> > From: Intel-xe <intel-xe-bounces@lists.freedesktop.org> On Behalf Of Lin,
> > Shuicheng
> > Sent: Friday, January 30, 2026 10:09 AM
> > To: Piorkowski, Piotr <piotr.piorkowski@intel.com>; intel-
> > xe@lists.freedesktop.org
> > Cc: Piorkowski, Piotr <piotr.piorkowski@intel.com>; Roper, Matthew D
> > <matthew.d.roper@intel.com>; Wajdeczko, Michal
> > <Michal.Wajdeczko@intel.com>
> > Subject: RE: [PATCH v1] drm/xe/vf: Avoid reading media version when media
> > GT is disabled
> > 
> > On Fri, Jan 30, 2026 7:55 AM Piotr Piórkowski wrote:
> > > Subject: [PATCH v1] drm/xe/vf: Avoid reading media version when media
> > > GT is disabled
> > 
> > The change also affects pf, maybe drop vf in the title?
> > 

You're right. My change also affects PF, which wasn't my goal.
I'll move these changes directly to the VF path.
These changes do not make sense on PF, because reading the registers
does not hurt us. The problem arises when we try to communicate with
Media GuC on VF, which has not even been loaded by PF.
> > >
> > > From: Piotr Piórkowski <piotr.piorkowski@intel.com>
> > >
> > > When the media GT is not allowed, a VF must not attempt to read the
> > > media version from the GuC. The GuC may not be loaded, and any attempt
> > > to communicate with it would result in a timeout and a VF probe failure:
> > >
> > > (...)
> > > [ 1912.406046] xe 0000:01:00.1: [drm] *ERROR* Tile0: GT1: GuC mmio
> > > request 0x5507: no reply 0x5507 [ 1912.407277] xe 0000:01:00.1: [drm]
> > > *ERROR* Tile0: GT1: [GUC COMMUNICATION] MMIO send failed (-
> > > ETIMEDOUT) [ 1912.408689] xe 0000:01:00.1: [drm] *ERROR* VF: Tile0:
> > > GT1: Failed to reset GuC state (-ETIMEDOUT) [ 1912.413986] xe
> > > 0000:01:00.1: probe with driver xe failed with error -110
> > >
> > > Let's skip reading the media version for VFs when the media GT is not
> > allowed.
> > >
> > > Fixes: 7abd69278bb5 ("drm/xe/configfs: Add attribute to disable GT
> > > types")
> > > Signed-off-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
> > > Cc: Matt Roper <matthew.d.roper@intel.com>
> > > Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
> > > ---
> > >  drivers/gpu/drm/xe/xe_pci.c | 4 ++++
> > >  1 file changed, 4 insertions(+)
> > >
> > > diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
> > > index b5e8935fff1d..6ac1baabab4e 100644
> > > --- a/drivers/gpu/drm/xe/xe_pci.c
> > > +++ b/drivers/gpu/drm/xe/xe_pci.c
> > > @@ -644,6 +644,10 @@ static int handle_gmdid(struct xe_device *xe,
> > >  			ver / 100, ver % 100);
> > >  	}
> > >
> > > +	/* Don't try read media ver if media GT is not allowed */
> > 
> > Nit: "try read" should be "try to read".

ok

> > 
> > > +	if (!xe_configfs_media_gt_allowed(to_pci_dev(xe->drm.dev)))
> > > +		return 0;
> > > +
> > 
> > I am curious if primary gt is disabled, will it also have similar issue?
> > If not, what cause the difference?
> > Thanks.
> 
> I should try it first.
> Primary gt cannot be disabled currently.
> "
> [49365.389692] xe 0000:03:00.0: [drm] Primary GT disabled via configfs
> [49365.389695] xe 0000:03:00.0: [drm] *ERROR* Cannot probe device with without a primary GT
> "

In the case of VF, the primary GT would have the same problem,
but we are not considering disabling primary GT.


Thanks
Piotr

> 
> Shuicheng
> 
> > 
> > Shuicheng
> > 
> > >  	ret = read_gmdid(xe, GMDID_MEDIA, &ver, media_revid);
> > >  	if (ret)
> > >  		return ret;
> > > --
> > > 2.34.1
> 

--