On Fri, Nov 07, 2025 at 10:19:58AM -0600, Lucas De Marchi wrote: > On Fri, Nov 07, 2025 at 10:01:26AM -0600, Govindapillai, Vinod wrote: > > On Fri, 2025-11-07 at 07:45 -0600, Lucas De Marchi wrote: > > > On Fri, Nov 07, 2025 at 01:08:15PM +0200, Jani Nikula wrote: > > > > On Fri, 07 Nov 2025, "Govindapillai, Vinod" > > > > <[email protected]> wrote: > > > > > On Fri, 2025-11-07 at 12:16 +0200, Jani Nikula wrote: > > > > > > On Fri, 07 Nov 2025, Vinod Govindapillai > > > > > > <[email protected]> wrote: > > > > > > > Disable FBC as part for the wa for the bmg variant > > > > > > > > > > > > > > Bspec: 74212 > > > > > > > Signed-off-by: Vinod Govindapillai > > > > > > > <[email protected]> > > > > > > > --- > > > > > > > drivers/gpu/drm/i915/display/intel_display_wa.c | 4 ++++ > > > > > > > drivers/gpu/drm/i915/display/intel_display_wa.h | 1 + > > > > > > > drivers/gpu/drm/i915/display/intel_fbc.c | 10 > > > > > > > ++++++++++ > > > > > > > 3 files changed, 15 insertions(+) > > > > > > > > > > > > > > diff --git a/drivers/gpu/drm/i915/display/intel_display_wa.c > > > > > > > b/drivers/gpu/drm/i915/display/intel_display_wa.c > > > > > > > index c528aaa679ca..ba2272d85a04 100644 > > > > > > > --- a/drivers/gpu/drm/i915/display/intel_display_wa.c > > > > > > > +++ b/drivers/gpu/drm/i915/display/intel_display_wa.c > > > > > > > @@ -10,6 +10,7 @@ > > > > > > > #include "intel_display_core.h" > > > > > > > #include "intel_display_regs.h" > > > > > > > #include "intel_display_wa.h" > > > > > > > +#include "intel_step.h" > > > > > > > > > > > > > > static void gen11_display_wa_apply(struct intel_display > > > > > > > *display) > > > > > > > { > > > > > > > @@ -69,6 +70,9 @@ bool __intel_display_wa(struct > > > > > > > intel_display > > > > > > > *display, enum intel_display_wa wa, > > > > > > > return DISPLAY_VER(display) == 13; > > > > > > > case INTEL_DISPLAY_WA_22014263786: > > > > > > > return IS_DISPLAY_VERx100(display, 1100, > > > > > > > 1400); > > > > > > > + case INTEL_DISPLAY_WA_15018326506: > > > > > > > + return DISPLAY_VERx100(display) == 1401 && > > > > > > > + INTEL_DISPLAY_STEP(display) == > > > > > > > STEP_C0; > > > > > > > > > > > > IS_DISPLAY_VERx100_STEP(). > > > > > > > > > > I was looking into the comments for that macro! This wa apply > > > > > only for > > > > > step c0 (as of now)! I wasn't sure what the "until" part could be > > > > > in > > > > > that macro! So what do you suggest for the "until" STEP_D0 or > > > > > STEP_FOREVER in such cases! > > > > > > > > If it's C0 only, then C1 is the next step, not D0. > > > > > > it would be very weird to be C0 **only**, so I went to check the WA > > > db > > > and I don't really see any stepping restriction > > > > > > Lucas De Marchi > > > > wa_16023588340 disabled FBC in g31. But this wa 15018326506 recommends > > disabling FBC for g31 which has display stepping C0 from bspec 68090. > > right, I see there the DE_ID 0x03804008. For some reason the WA itself > is not updating the stepping correctly, but if it applied to other > steppings it would actually be for g21. Matt / Uma, do you know why the > stepping is not correct?
I'm not sure I follow the discussion here. Wa_16023588340 is the old G21 SoC-level issue related to LMEM BAR access. It's an invasive workaround that requires driver changes on both the GT side and display side, and the display part involves disabling FBC. That workaround is only required on BMG-G21 platforms (of all steppings); the SoC issue is no longer present on BMG-G31 platforms. Wa_15018326506 is an unrelated, display-specific workaround that also requires disabling FBC (for different reasons). The need for Wa_15018326506 was discovered during BMG-G31 testing, although it's likely that the underlying hardware issue existed on BMG-G21 as well and was just masked because FBC was already disabled there. Wa_15018326506 is still going through final approval so the final call on steppings hasn't been made yet, but I suspect it will apply to all steppings (as it's currently set in the workaround database) because there's no reason to believe that earlier steppings wouldn't have been impacted if FBC wasn't already disabled there. Matt > > thanks > Lucas De Marchi -- Matt Roper Graphics Software Engineer Linux GPU Platform Enablement Intel Corporation
