[PATCH 2/2] drm/i915: disable fbc due to Wa_16023588340

Fri Jun 28 05:30:57 UTC 2024

On Wed, Jun 26, 2024 at 10:42:24AM GMT, Matt Roper wrote:
>On Wed, Jun 26, 2024 at 01:26:00PM -0400, Rodrigo Vivi wrote:
>> On Wed, Jun 26, 2024 at 05:17:41PM +0100, Matthew Auld wrote:
>> > On 26/06/2024 16:53, Rodrigo Vivi wrote:
>> > > On Wed, Jun 19, 2024 at 03:31:27PM +0100, Matthew Auld wrote:
>> > > > On BMG-G21 we need to disable fbc due to complications around the WA.
>> > > >
>> > > > Signed-off-by: Matthew Auld <matthew.auld at intel.com>
>> > > > Cc: Jonathan Cavitt <jonathan.cavitt at intel.com>
>> > > > Cc: Matt Roper <matthew.d.roper at intel.com>
>> > > > Cc: Lucas De Marchi <lucas.demarchi at intel.com>
>> > > > Cc: Vinod Govindapillai <vinod.govindapillai at intel.com>
>> > > > Cc: intel-gfx at lists.freedesktop.org
>> > > > ---
>> > > >   drivers/gpu/drm/i915/display/intel_display_wa.h |  8 ++++++++
>> > > >   drivers/gpu/drm/i915/display/intel_fbc.c        |  6 ++++++
>> > > >   drivers/gpu/drm/xe/Makefile                     |  4 +++-
>> > > >   drivers/gpu/drm/xe/display/xe_display_wa.c      | 16 ++++++++++++++++
>> > > >   4 files changed, 33 insertions(+), 1 deletion(-)
>> > > >   create mode 100644 drivers/gpu/drm/xe/display/xe_display_wa.c
>> > > >
>> > > > diff --git a/drivers/gpu/drm/i915/display/intel_display_wa.h b/drivers/gpu/drm/i915/display/intel_display_wa.h
>> > > > index 63201d09852c..be644ab6ae00 100644
>> > > > --- a/drivers/gpu/drm/i915/display/intel_display_wa.h
>> > > > +++ b/drivers/gpu/drm/i915/display/intel_display_wa.h
>> > > > @@ -6,8 +6,16 @@
>> > > >   #ifndef __INTEL_DISPLAY_WA_H__
>> > > >   #define __INTEL_DISPLAY_WA_H__
>> > > > +#include <linux/types.h>
>> > > > +
>> > > >   struct drm_i915_private;
>> > > >   void intel_display_wa_apply(struct drm_i915_private *i915);
>> > > > +#ifdef I915
>> > > > +static inline bool intel_display_needs_wa_16023588340(struct drm_i915_private *i915) { return false; }
>> > > > +#else
>> > > > +bool intel_display_needs_wa_16023588340(struct drm_i915_private *i915);
>> > > > +#endif
>> > >
>> > > please avoid the ifdef I915 in new patches as we are trying to get away from that
>> > > in favor of a clean separation.
>> >
>> > Can you please share an example for the best way to do that here, with clean
>> > separation?
>>
>> hmmm... looking more to the patch now...
>> I don't believe that the WA/RTP rule from Xe should leak into i915 to be honest.
>>
>> It looks like we are trending to a separate intel-display.ko that shouldn't depend
>> on driver's declarations like this.
>>
>> Ideally I would also say that wa in the display code should relly on the 'D'
>> (display-id) of the GMD-ID. But I see that this 16023588340 is for the 'G' ip.
>> So, perhaps the display code should inspect the 'G' id from the device inside
>> display code?
>
>This is one of those rare cases where a GT-based workaround also has a
>side effect of "oh, and you also need to disable FBC in the display
>driver."  So as you said, the need to disable FBC is entirely tied to
>details on the graphics side of the IP, not the display version.  :-(
>
>So there are two options --- either you duplicate the logic to decide
>whether the workaround applies in both the display driver and the core
>(i915/Xe) driver, or you make the core driver the authoritative decision
>maker for GT-based workarounds and give the display driver some way to
>query the core driver.  The patch here is following the latter approach,
>and for the short term future while display code just gets re-compiled
>into each core driver, this seems to be an accurate implementation
>(always false on i915 builds, and querying RTP for Xe builds).  As we
>proceed with moving intel_display into its own standalone driver, we'll
>need a more formal display<->core driver interface and it will probably
>make sense to have a formal "query a GT workaround" entrypoint as part
>of that interface so that we don't need to keep adding more one-off
>"needs_XXXXX" for future workarounds that wind up in the same boat.

agreed. Let's not leak the decision on other places: it belongs in the
core driver.

When there's the proper separation, then I believe we can just copy the
root_gt->wa_active.oob over to the display struct. And then implement a
macro on the display side to do the same check. Or we may have a set of
function pointers and one of them would be to query if a WA should be
enabled.

Lucas De Marchi

>
>
>Matt
>
>>
>> Jani, thoughts on this?
>>
>> >
>> > I can add a new .c just for intel_display_needs_wa_16023588340 and move it
>> > there, which then avoids the ifdef I think, but that then adds an entirely
>> > new file just for this tiny stub. Unless I can dump it somewhere else?
>>
>> One temporary workaround that I see without the ifdef I915 would be to
>> declare this function in i915_drv.h so in xe you add to the compat-i915-headers
>> instead of creating a new file there.
>>
>> >
>> > >
>> > > > +
>> > > >   #endif
>> > > > diff --git a/drivers/gpu/drm/i915/display/intel_fbc.c b/drivers/gpu/drm/i915/display/intel_fbc.c
>> > > > index 67116c9f1464..8488f82143a4 100644
>> > > > --- a/drivers/gpu/drm/i915/display/intel_fbc.c
>> > > > +++ b/drivers/gpu/drm/i915/display/intel_fbc.c
>> > > > @@ -56,6 +56,7 @@
>> > > >   #include "intel_display_device.h"
>> > > >   #include "intel_display_trace.h"
>> > > >   #include "intel_display_types.h"
>> > > > +#include "intel_display_wa.h"
>> > > >   #include "intel_fbc.h"
>> > > >   #include "intel_fbc_regs.h"
>> > > >   #include "intel_frontbuffer.h"
>> > > > @@ -1237,6 +1238,11 @@ static int intel_fbc_check_plane(struct intel_atomic_state *state,
>> > > >   		return 0;
>> > > >   	}
>> > > > +	if (intel_display_needs_wa_16023588340(i915)) {
>> > > > +		plane_state->no_fbc_reason = "Wa_16023588340";
>> > > > +		return 0;
>> > > > +	}
>> > > > +
>> > > >   	/* WaFbcTurnOffFbcWhenHyperVisorIsUsed:skl,bxt */
>> > > >   	if (i915_vtd_active(i915) && (IS_SKYLAKE(i915) || IS_BROXTON(i915))) {
>> > > >   		plane_state->no_fbc_reason = "VT-d enabled";
>> > > > diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile
>> > > > index 0e16e5029081..f7521fd5db4c 100644
>> > > > --- a/drivers/gpu/drm/xe/Makefile
>> > > > +++ b/drivers/gpu/drm/xe/Makefile
>> > > > @@ -34,7 +34,8 @@ uses_generated_oob := \
>> > > >   	$(obj)/xe_ring_ops.o \
>> > > >   	$(obj)/xe_vm.o \
>> > > >   	$(obj)/xe_wa.o \
>> > > > -	$(obj)/xe_ttm_stolen_mgr.o
>> > > > +	$(obj)/xe_ttm_stolen_mgr.o \
>> > > > +	$(obj)/display/xe_display_wa.o \
>> > > >   $(uses_generated_oob): $(generated_oob)
>> > > > @@ -192,6 +193,7 @@ xe-$(CONFIG_DRM_XE_DISPLAY) += \
>> > > >   	display/xe_display.o \
>> > > >   	display/xe_display_misc.o \
>> > > >   	display/xe_display_rps.o \
>> > > > +	display/xe_display_wa.o \
>> > > >   	display/xe_dsb_buffer.o \
>> > > >   	display/xe_fb_pin.o \
>> > > >   	display/xe_hdcp_gsc.o \
>> > > > diff --git a/drivers/gpu/drm/xe/display/xe_display_wa.c b/drivers/gpu/drm/xe/display/xe_display_wa.c
>> > > > new file mode 100644
>> > > > index 000000000000..68e3d1959ad6
>> > > > --- /dev/null
>> > > > +++ b/drivers/gpu/drm/xe/display/xe_display_wa.c
>> > > > @@ -0,0 +1,16 @@
>> > > > +// SPDX-License-Identifier: MIT
>> > > > +/*
>> > > > + * Copyright © 2024 Intel Corporation
>> > > > + */
>> > > > +
>> > > > +#include "intel_display_wa.h"
>> > > > +
>> > > > +#include "xe_device.h"
>> > > > +#include "xe_wa.h"
>> > > > +
>> > > > +#include <generated/xe_wa_oob.h>
>> > > > +
>> > > > +bool intel_display_needs_wa_16023588340(struct drm_i915_private *i915)
>> > > > +{
>> > > > +	return XE_WA(xe_root_mmio_gt(i915), 16023588340);
>> > > > +}
>> > > > --
>> > > > 2.45.1
>> > > >
>
>-- 
>Matt Roper
>Graphics Software Engineer
>Linux GPU Platform Enablement
>Intel Corporation