[Intel-gfx] [RFC 1/2] drm/i915: Add RPL-U sub platform
Matt Roper
matthew.d.roper at intel.com
Fri Jan 27 18:04:43 UTC 2023
On Fri, Jan 27, 2023 at 01:34:31AM -0800, Borah, Chaitanya Kumar wrote:
> Hello Jani and Matt,
>
> > -----Original Message-----
> > From: Jani Nikula <jani.nikula at linux.intel.com>
> > Sent: Tuesday, January 24, 2023 8:02 PM
> > To: Borah, Chaitanya Kumar <chaitanya.kumar.borah at intel.com>; intel-
> > gfx at lists.freedesktop.org
> > Cc: Shankar, Uma <uma.shankar at intel.com>; Syrjala, Ville
> > <ville.syrjala at intel.com>; Srivatsa, Anusha <anusha.srivatsa at intel.com>;
> > Roper, Matthew D <matthew.d.roper at intel.com>; Atwood, Matthew S
> > <matthew.s.atwood at intel.com>; Borah, Chaitanya Kumar
> > <chaitanya.kumar.borah at intel.com>
> > Subject: Re: [RFC 1/2] drm/i915: Add RPL-U sub platform
> >
> > On Tue, 17 Jan 2023, Chaitanya Kumar Borah
> > <chaitanya.kumar.borah at intel.com> wrote:
> > > Separate out RPLU device ids and add them to both RPL and newly
> > > created RPL-U subplatforms.
> > >
> > > v2: (Matt)
> > > - Sort PCI-IDs numerically
> > > - Name the sub-platform to accurately depict what it is for
> > > - Make RPL-U part of RPL subplatform
> > >
> > > v3: revert to RPL-U subplatform (Jani)
> > >
> > > Signed-off-by: Chaitanya Kumar Borah
> > <chaitanya.kumar.borah at intel.com>
> > > ---
> > > drivers/gpu/drm/i915/i915_drv.h | 2 ++
> > > drivers/gpu/drm/i915/i915_pci.c | 1 +
> > > drivers/gpu/drm/i915/intel_device_info.c | 8 ++++++++
> > > drivers/gpu/drm/i915/intel_device_info.h | 2 ++
> > > include/drm/i915_pciids.h | 11 +++++++----
> > > 5 files changed, 20 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/i915_drv.h
> > > b/drivers/gpu/drm/i915/i915_drv.h index 48fd82722f12..c88e514728a0
> > > 100644
> > > --- a/drivers/gpu/drm/i915/i915_drv.h
> > > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > > @@ -619,6 +619,8 @@ IS_SUBPLATFORM(const struct drm_i915_private
> > *i915,
> > > IS_SUBPLATFORM(dev_priv, INTEL_ALDERLAKE_P,
> > INTEL_SUBPLATFORM_N)
> > > #define IS_ADLP_RPLP(dev_priv) \
> > > IS_SUBPLATFORM(dev_priv, INTEL_ALDERLAKE_P,
> > INTEL_SUBPLATFORM_RPL)
> > > +#define IS_ADLP_RPLU(dev_priv) \
> > > + IS_SUBPLATFORM(dev_priv, INTEL_ALDERLAKE_P,
> > INTEL_SUBPLATFORM_RPLU)
> > > #define IS_HSW_EARLY_SDV(dev_priv) (IS_HASWELL(dev_priv) && \
> > > (INTEL_DEVID(dev_priv) & 0xFF00) ==
> > 0x0C00) #define
> > > IS_BDW_ULT(dev_priv) \ diff --git a/drivers/gpu/drm/i915/i915_pci.c
> > > b/drivers/gpu/drm/i915/i915_pci.c index 6cc65079b18d..e9f3b99b3e00
> > > 100644
> > > --- a/drivers/gpu/drm/i915/i915_pci.c
> > > +++ b/drivers/gpu/drm/i915/i915_pci.c
> > > @@ -1234,6 +1234,7 @@ static const struct pci_device_id pciidlist[] = {
> > > INTEL_DG1_IDS(&dg1_info),
> > > INTEL_RPLS_IDS(&adl_s_info),
> > > INTEL_RPLP_IDS(&adl_p_info),
> > > + INTEL_RPLU_IDS(&adl_p_info),
> >
> > You may want to drop this change, see later comment on how and why.
> >
> > > INTEL_DG2_IDS(&dg2_info),
> > > INTEL_ATS_M_IDS(&ats_m_info),
> > > INTEL_MTL_IDS(&mtl_info),
> > > diff --git a/drivers/gpu/drm/i915/intel_device_info.c
> > > b/drivers/gpu/drm/i915/intel_device_info.c
> > > index 849baf6c3b3c..fec8bd116436 100644
> > > --- a/drivers/gpu/drm/i915/intel_device_info.c
> > > +++ b/drivers/gpu/drm/i915/intel_device_info.c
> > > @@ -199,6 +199,11 @@ static const u16 subplatform_n_ids[] = { static
> > > const u16 subplatform_rpl_ids[] = {
> > > INTEL_RPLS_IDS(0),
> > > INTEL_RPLP_IDS(0),
> > > + INTEL_RPLU_IDS(0)
> >
> > Please always include the trailing , at the end to make future changes easier.
> > (However, you may want to drop this change altogether, see later
> > comment.)
> >
> > > +};
> > > +
> > > +static const u16 subplatform_rplu_ids[] = {
> > > + INTEL_RPLU_IDS(0),
> > > };
> > >
> > > static const u16 subplatform_g10_ids[] = { @@ -268,6 +273,9 @@ static
> > > void intel_device_info_subplatform_init(struct drm_i915_private *i915)
> > > } else if (find_devid(devid, subplatform_rpl_ids,
> > > ARRAY_SIZE(subplatform_rpl_ids))) {
> > > mask = BIT(INTEL_SUBPLATFORM_RPL);
> > > + if (find_devid(devid, subplatform_rplu_ids,
> > > + ARRAY_SIZE(subplatform_rplu_ids)))
> > > + mask |= BIT(INTEL_SUBPLATFORM_RPLU);
> > > } else if (find_devid(devid, subplatform_g10_ids,
> > > ARRAY_SIZE(subplatform_g10_ids))) {
> > > mask = BIT(INTEL_SUBPLATFORM_G10);
> > > diff --git a/drivers/gpu/drm/i915/intel_device_info.h
> > > b/drivers/gpu/drm/i915/intel_device_info.h
> > > index d588e5fd2eea..4a5cd337e4b5 100644
> > > --- a/drivers/gpu/drm/i915/intel_device_info.h
> > > +++ b/drivers/gpu/drm/i915/intel_device_info.h
> > > @@ -127,6 +127,8 @@ enum intel_platform {
> > > * bit set
> > > */
> > > #define INTEL_SUBPLATFORM_N 1
> > > +/* Sub Platform for RPL-U */
> >
> > This comment really adds nothing, it's exactly the same as the macro name.
> >
>
> Ack.
>
> > > +#define INTEL_SUBPLATFORM_RPLU 2
> > >
> > > /* MTL */
> > > #define INTEL_SUBPLATFORM_M 0
> > > diff --git a/include/drm/i915_pciids.h b/include/drm/i915_pciids.h
> > > index 4a4c190f7698..758be5fb09a2 100644
> > > --- a/include/drm/i915_pciids.h
> > > +++ b/include/drm/i915_pciids.h
> > > @@ -684,14 +684,17 @@
> > > INTEL_VGA_DEVICE(0xA78A, info), \
> > > INTEL_VGA_DEVICE(0xA78B, info)
> > >
> > > +/* RPL-U */
> > > +#define INTEL_RPLU_IDS(info) \
> > > + INTEL_VGA_DEVICE(0xA721, info), \
> > > + INTEL_VGA_DEVICE(0xA7A1, info), \
> > > + INTEL_VGA_DEVICE(0xA7A9, info)
> > > +
> > > /* RPL-P */
> > > #define INTEL_RPLP_IDS(info) \
> > > INTEL_VGA_DEVICE(0xA720, info), \
> > > - INTEL_VGA_DEVICE(0xA721, info), \
> > > INTEL_VGA_DEVICE(0xA7A0, info), \
> > > - INTEL_VGA_DEVICE(0xA7A1, info), \
> > > - INTEL_VGA_DEVICE(0xA7A8, info), \
> > > - INTEL_VGA_DEVICE(0xA7A9, info)
> > > + INTEL_VGA_DEVICE(0xA7A8, info)
> >
> > Changing the INTEL_RPLP_IDS impacts arch/x86/kernel/early-quirks.c too. As
> > is, this drops the early quirks from RPL-U.
> >
>
> Yes I missed the early quirks part in this revision. Thank you Jani for pointing out.
>
> > Your options are 1) modify early-quirks.c too, or 2) include RPL-U ids in RPL-P:
> >
> > #define INTEL_RPLP_IDS(info) \
> > + INTEL_RPLU_IDS(info), \
> > - INTEL_VGA_DEVICE(0xA721, info), \
> > INTEL_VGA_DEVICE(0xA7A0, info), \
> > - INTEL_VGA_DEVICE(0xA7A1, info), \
> > - INTEL_VGA_DEVICE(0xA7A8, info), \
> > - INTEL_VGA_DEVICE(0xA7A9, info)
> > + INTEL_VGA_DEVICE(0xA7A8, info)
> >
> > With option 2, you also don't need to add INTEL_RPLU_IDS separately to
> > subplatform_rpl_ids[] or pciidlist[].
> >
> > I might lean towards option 2, but no strong opinions.
> >
>
> Thank you Jani for your suggestion. I am also inclined to the solution of adding RPLU IDs to RPLP platform.
>
> To summarize, we would add the INTEL_RPLU_IDS(info) into INTEL_RPLP_IDS(info). As Jani pointed out, with this, there is no need to add INTEL_RPLU_IDS separately to
> subplatform_rpl_ids[] or pciidlist[]. This also aligns with Bspec(55376) as RPL-U ids are listed under "RPL-P SKUs and Device IDs Info"
>
> @Matt: I remember, we had discussed a similar option in our communications. Do you have any strong opinion against it?
That sounds fine to me.
Thanks.
Matt
>
> If not I would go ahead with this implementation.
>
> Regards
>
> Chaitanya
>
> >
> > BR,
> > Jani.
> >
> >
> > >
> > > /* DG2 */
> > > #define INTEL_DG2_G10_IDS(info) \
> >
> > --
> > Jani Nikula, Intel Open Source Graphics Center
--
Matt Roper
Graphics Software Engineer
Linux GPU Platform Enablement
Intel Corporation
More information about the Intel-gfx
mailing list