[PATCH v2 2/3] drm/xe/xe2_hpg: Add set of workarounds

[Bhadane, Dnyaneshwar]

> -----Original Message-----
> From: De Marchi, Lucas <lucas.demarchi at intel.com>
> Sent: Thursday, June 5, 2025 9:16 PM
> To: Roper, Matthew D <matthew.d.roper at intel.com>
> Cc: Bhadane, Dnyaneshwar <dnyaneshwar.bhadane at intel.com>; intel-
> xe at lists.freedesktop.org; Chauhan, Shekhar <shekhar.chauhan at intel.com>
> Subject: Re: [PATCH v2 2/3] drm/xe/xe2_hpg: Add set of workarounds
> 
> On Wed, Jun 04, 2025 at 11:54:19AM -0700, Matt Roper wrote:
> >On Wed, Jun 04, 2025 at 08:22:35PM +0530, Dnyaneshwar Bhadane wrote:
> >> From: Shekhar Chauhan <shekhar.chauhan at intel.com>
> >>
> >> Add set of workarounds for xe2_hpg.
> >>
> >> -v2: Fix xe2_hpg GMD version for some workarounds.
> >>
> >> Signed-off-by: Shekhar Chauhan <shekhar.chauhan at intel.com>
> >> Signed-off-by: Dnyaneshwar Bhadane <dnyaneshwar.bhadane at intel.com>
> >> ---
> >>  drivers/gpu/drm/xe/xe_wa.c         | 55 +++++++++++++++++++++---------
> >>  drivers/gpu/drm/xe/xe_wa_oob.rules |  4 +--
> >>  2 files changed, 40 insertions(+), 19 deletions(-)
> >>
> >> diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c
> >> index 67196baa4249..2c0e0adc5904 100644
> >> --- a/drivers/gpu/drm/xe/xe_wa.c
> >> +++ b/drivers/gpu/drm/xe/xe_wa.c
> >> @@ -489,8 +489,8 @@ static const struct xe_rtp_entry_sr engine_was[] = {
> >>  	  XE_RTP_RULES(GRAPHICS_VERSION(2004),
> FUNC(xe_rtp_match_first_render_or_compute)),
> >>  	  XE_RTP_ACTIONS(SET(ROW_CHICKEN3,
> XE2_EUPEND_CHK_FLUSH_DIS))
> >>  	},
> >> -	{ XE_RTP_NAME("18034896535, 16021540221"), /* 16021540221:
> GRAPHICS_STEP(A0, B0) */
> >> -	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, 2004),
> >> +	{ XE_RTP_NAME("16021540221"),
> >> +	  XE_RTP_RULES(GRAPHICS_VERSION(2004), GRAPHICS_STEP(A0, B0),
> >
> >Breaking out 20.01/20.02 into a separate entry farther down is good
> >since it keeps the LPG and HPG handling separate, and also avoids
> >blindly applying this 20.03 which doesn't exist today, but could
> >potentially show up in the future.
> >
> >However is there actually a benefit to splitting 16021540221 (a-step
> >only) and 18034896535 (all steppings) into separate entries given that
> >the two workarounds have the same actions?  I think breaking it out
> >like this will actually cause a warning if run on an A-step platform
> >since we'll have two entries trying to update the same bits of the same
> >register (see compatible_entries() in xe_reg_sr.c).
> 
> humn... it looks like our kunit in drivers/gpu/drm/xe/tests/xe_wa_test.c
> is only covering the gt table, not engine. We should add entries for
> 
> 	KUNIT_CASE_PARAM(xe_wa_engine, platform_gen_params),
> 
> and
> 
> 	KUNIT_CASE_PARAM(xe_wa_lrc, platform_gen_params),
> 
> to cover this in kunit and avoid introducing these bugs.
> Shekhar/Dnyaneshwar, would you be able to do that? It can (should?) be
> separate from these patches to avoid unnecessary delays in applying these
> workarounds.
Thanks Lucas, I will take care of this, and I will send the separate patch for this. 

Dnyaneshwar, 

> 
> thanks
> Lucas De Marchi