Re: [PATCH 1/5] drm/i915: don't include CML PCI IDs in CFL

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Wed, May 08, 2024 at 02:45:10PM +0300, Jani Nikula wrote:
> On Wed, 08 May 2024, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote:
> > On Tue, May 07, 2024 at 09:47:16AM -0400, Rodrigo Vivi wrote:
> >> On Tue, May 07, 2024 at 03:56:48PM +0300, Jani Nikula wrote:
> >> > It's confusing for INTEL_CFL_IDS() to include all CML PCI IDs. Even if
> >> > we treat them the same in a lot of places, CML is a platform of its own,
> >> > and the lists of PCI IDs should not conflate them.
> >> > 
> >> > Cc: Bjorn Helgaas <bhelgaas@google.com>
> >> > Cc: linux-pci@vger.kernel.org
> >> > Signed-off-by: Jani Nikula <jani.nikula@intel.com>
> >> > ---
> >> >  arch/x86/kernel/early-quirks.c                      |  1 +
> >> >  drivers/gpu/drm/i915/display/intel_display_device.c |  1 +
> >> >  include/drm/i915_pciids.h                           | 12 +++++++-----
> >> >  3 files changed, 9 insertions(+), 5 deletions(-)
> >> > 
> >> > diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c
> >> > index 59f4aefc6bc1..2e2d15be4025 100644
> >> > --- a/arch/x86/kernel/early-quirks.c
> >> > +++ b/arch/x86/kernel/early-quirks.c
> >> > @@ -547,6 +547,7 @@ static const struct pci_device_id intel_early_ids[] __initconst = {
> >> >  	INTEL_BXT_IDS(&gen9_early_ops),
> >> >  	INTEL_KBL_IDS(&gen9_early_ops),
> >> >  	INTEL_CFL_IDS(&gen9_early_ops),
> >> > +	INTEL_CML_IDS(&gen9_early_ops),
> >> >  	INTEL_GLK_IDS(&gen9_early_ops),
> >> >  	INTEL_CNL_IDS(&gen9_early_ops),
> >> >  	INTEL_ICL_11_IDS(&gen11_early_ops),
> >> > diff --git a/drivers/gpu/drm/i915/display/intel_display_device.c b/drivers/gpu/drm/i915/display/intel_display_device.c
> >> > index 56a2e17d7d9e..3aa7d1cdd228 100644
> >> > --- a/drivers/gpu/drm/i915/display/intel_display_device.c
> >> > +++ b/drivers/gpu/drm/i915/display/intel_display_device.c
> >> > @@ -832,6 +832,7 @@ static const struct {
> >> >  	INTEL_GLK_IDS(&glk_display),
> >> >  	INTEL_KBL_IDS(&skl_display),
> >> >  	INTEL_CFL_IDS(&skl_display),
> >> > +	INTEL_CML_IDS(&skl_display),
> >> >  	INTEL_ICL_11_IDS(&icl_display),
> >> >  	INTEL_EHL_IDS(&jsl_ehl_display),
> >> >  	INTEL_JSL_IDS(&jsl_ehl_display),
> >> > diff --git a/include/drm/i915_pciids.h b/include/drm/i915_pciids.h
> >> > index 85ce33ad6e26..5f52c504ffde 100644
> >> > --- a/include/drm/i915_pciids.h
> >> > +++ b/include/drm/i915_pciids.h
> >> > @@ -472,6 +472,12 @@
> >> >  	INTEL_VGA_DEVICE(0x9BCA, info), \
> >> >  	INTEL_VGA_DEVICE(0x9BCC, info)
> >> >  
> >> > +#define INTEL_CML_IDS(info) \
> >> > +	INTEL_CML_GT1_IDS(info), \
> >> > +	INTEL_CML_GT2_IDS(info), \
> >> > +	INTEL_CML_U_GT1_IDS(info), \
> >> > +	INTEL_CML_U_GT2_IDS(info)
> >> > +
> >> >  #define INTEL_KBL_IDS(info) \
> >> >  	INTEL_KBL_GT1_IDS(info), \
> >> >  	INTEL_KBL_GT2_IDS(info), \
> >> > @@ -535,11 +541,7 @@
> >> >  	INTEL_WHL_U_GT1_IDS(info), \
> >> >  	INTEL_WHL_U_GT2_IDS(info), \
> >> >  	INTEL_WHL_U_GT3_IDS(info), \
> >> > -	INTEL_AML_CFL_GT2_IDS(info), \
> >> > -	INTEL_CML_GT1_IDS(info), \
> >> > -	INTEL_CML_GT2_IDS(info), \
> >> > -	INTEL_CML_U_GT1_IDS(info), \
> >> > -	INTEL_CML_U_GT2_IDS(info)
> >> > +	INTEL_AML_CFL_GT2_IDS(info)
> >> 
> >> Why only CML and not AML and WHL as well?
> >
> > Why do we even have CML as a separate platform? The only difference 
> > I can see is is that we do allow_read_ctx_timestamp() for CML but
> > not for CFL. Does that even make sense?
> 
> git blame tells me:
> 
> 5f4ae2704d59 ("drm/i915: Identify Cometlake platform")
> dbc7e72897a4 ("drm/i915/gt: Make the CTX_TIMESTAMP readable on !rcs")

Right. That explains why we need it on CML+. But is there some reason
we  can't just do it on CFL as well, even if not strictly necessary?
I would assume that setting FORCE_TO_NONPRIV on an already
non-privileged register should be totally fine.

-- 
Ville Syrjälä
Intel