Message ID | 20191029095856.25431-3-matthew.auld@intel.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | [1/7] drm/i915: define i915_ggtt_has_aperture | expand |
Quoting Matthew Auld (2019-10-29 09:58:52) > From: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > > We can't fence anything without aperture. > > Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > Signed-off-by: Stuart Summers <stuart.summers@intel.com> > Signed-off-by: Matthew Auld <matthew.auld@intel.com> Neat. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> -Chris
On 10/29/19 2:58 AM, Matthew Auld wrote: > From: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > > We can't fence anything without aperture. > > Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > Signed-off-by: Stuart Summers <stuart.summers@intel.com> > Signed-off-by: Matthew Auld <matthew.auld@intel.com> > --- > drivers/gpu/drm/i915/i915_gem_fence_reg.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/drivers/gpu/drm/i915/i915_gem_fence_reg.c b/drivers/gpu/drm/i915/i915_gem_fence_reg.c > index 321189e1b0f2..71efccfde122 100644 > --- a/drivers/gpu/drm/i915/i915_gem_fence_reg.c > +++ b/drivers/gpu/drm/i915/i915_gem_fence_reg.c > @@ -846,8 +846,10 @@ void i915_ggtt_init_fences(struct i915_ggtt *ggtt) > > detect_bit_6_swizzle(ggtt); > > - if (INTEL_GEN(i915) >= 7 && > - !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) > + if (!i915_ggtt_has_aperture(ggtt)) Daniel had mentioned that the fencing HW has been nerfed in gen12 and suggested settings num_fences to zero for all gen12+. Should we go directly with that? Daniele > + num_fences = 0; > + else if (INTEL_GEN(i915) >= 7 && > + !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) > num_fences = 32; > else if (INTEL_GEN(i915) >= 4 || > IS_I945G(i915) || IS_I945GM(i915) || >
Quoting Daniele Ceraolo Spurio (2019-10-29 21:23:16) > > > On 10/29/19 2:58 AM, Matthew Auld wrote: > > From: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > > > > We can't fence anything without aperture. > > > > Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> > > Signed-off-by: Stuart Summers <stuart.summers@intel.com> > > Signed-off-by: Matthew Auld <matthew.auld@intel.com> > > --- > > drivers/gpu/drm/i915/i915_gem_fence_reg.c | 6 ++++-- > > 1 file changed, 4 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/gpu/drm/i915/i915_gem_fence_reg.c b/drivers/gpu/drm/i915/i915_gem_fence_reg.c > > index 321189e1b0f2..71efccfde122 100644 > > --- a/drivers/gpu/drm/i915/i915_gem_fence_reg.c > > +++ b/drivers/gpu/drm/i915/i915_gem_fence_reg.c > > @@ -846,8 +846,10 @@ void i915_ggtt_init_fences(struct i915_ggtt *ggtt) > > > > detect_bit_6_swizzle(ggtt); > > > > - if (INTEL_GEN(i915) >= 7 && > > - !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) > > + if (!i915_ggtt_has_aperture(ggtt)) > > Daniel had mentioned that the fencing HW has been nerfed in gen12 and > suggested settings num_fences to zero for all gen12+. Should we go > directly with that? Do fences exist, yes/no? If there are literally no fences forevermore... tgl begs to differ though. -Chris
On 10/29/19 2:44 PM, Chris Wilson wrote: > Quoting Daniele Ceraolo Spurio (2019-10-29 21:23:16) >> >> >> On 10/29/19 2:58 AM, Matthew Auld wrote: >>> From: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> >>> >>> We can't fence anything without aperture. >>> >>> Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> >>> Signed-off-by: Stuart Summers <stuart.summers@intel.com> >>> Signed-off-by: Matthew Auld <matthew.auld@intel.com> >>> --- >>> drivers/gpu/drm/i915/i915_gem_fence_reg.c | 6 ++++-- >>> 1 file changed, 4 insertions(+), 2 deletions(-) >>> >>> diff --git a/drivers/gpu/drm/i915/i915_gem_fence_reg.c b/drivers/gpu/drm/i915/i915_gem_fence_reg.c >>> index 321189e1b0f2..71efccfde122 100644 >>> --- a/drivers/gpu/drm/i915/i915_gem_fence_reg.c >>> +++ b/drivers/gpu/drm/i915/i915_gem_fence_reg.c >>> @@ -846,8 +846,10 @@ void i915_ggtt_init_fences(struct i915_ggtt *ggtt) >>> >>> detect_bit_6_swizzle(ggtt); >>> >>> - if (INTEL_GEN(i915) >= 7 && >>> - !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) >>> + if (!i915_ggtt_has_aperture(ggtt)) >> >> Daniel had mentioned that the fencing HW has been nerfed in gen12 and >> suggested settings num_fences to zero for all gen12+. Should we go >> directly with that? > > Do fences exist, yes/no? If there are literally no fences forevermore... > tgl begs to differ though. > -Chris > The registers are there in the specs and I don't see any special notes about their usage. I'll leave it to Daniel to comment on the capability reduction he mentioned. Not a blocker for this patch in the meantime. Daniele
diff --git a/drivers/gpu/drm/i915/i915_gem_fence_reg.c b/drivers/gpu/drm/i915/i915_gem_fence_reg.c index 321189e1b0f2..71efccfde122 100644 --- a/drivers/gpu/drm/i915/i915_gem_fence_reg.c +++ b/drivers/gpu/drm/i915/i915_gem_fence_reg.c @@ -846,8 +846,10 @@ void i915_ggtt_init_fences(struct i915_ggtt *ggtt) detect_bit_6_swizzle(ggtt); - if (INTEL_GEN(i915) >= 7 && - !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) + if (!i915_ggtt_has_aperture(ggtt)) + num_fences = 0; + else if (INTEL_GEN(i915) >= 7 && + !(IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915))) num_fences = 32; else if (INTEL_GEN(i915) >= 4 || IS_I945G(i915) || IS_I945GM(i915) ||