[PATCH] drm/i915/gtt: drop the page table optimisation

[PATCH] drm/i915/gtt: drop the page table optimisation

[Matthew Auld]

We skip filling out the pt with scratch entries if the va range covers
the entire pt, since we later have to fill it with the PTEs for the
object pages anyway. However this might leave open a small window where
the PTEs don't point to anything valid for the HW to consume.

When for example using 2M GTT pages this fill_px() showed up as being
quite significant in perf measurements, and ends up being completely
wasted since we ignore the pt and just use the pde directly.

Anyway, currently we have our PTE construction split between alloc and
insert, which is probably slightly iffy nowadays, since the alloc
doesn't actually allocate anything anymore, instead it just sets up the
page directories and points the PTEs at the scratch page. Later when we
do the insert step we re-program the PTEs again. Better might be to
squash the alloc and insert into a single step, then bringing back this
optimisation(along with some others) should be possible.

Fixes: 14826673247e ("drm/i915: Only initialize partially filled pagetables")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Jon Bloomfield <jon.bloomfield@intel.com>
Cc: Chris Wilson <chris.p.wilson@intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: <stable@vger.kernel.org> # v4.15+
---
 drivers/gpu/drm/i915/gt/gen8_ppgtt.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
index 3d02c726c746..6e0e52eeb87a 100644
--- a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
+++ b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
@@ -303,10 +303,7 @@ static void __gen8_ppgtt_alloc(struct i915_address_space * const vm,
 			__i915_gem_object_pin_pages(pt->base);
 			i915_gem_object_make_unshrinkable(pt->base);
 
-			if (lvl ||
-			    gen8_pt_count(*start, end) < I915_PDES ||
-			    intel_vgpu_active(vm->i915))
-				fill_px(pt, vm->scratch[lvl]->encode);
+			fill_px(pt, vm->scratch[lvl]->encode);
 
 			spin_lock(&pd->lock);
 			if (likely(!pd->entry[idx])) {
-- 
2.26.3

Re: [PATCH] drm/i915/gtt: drop the page table optimisation

[Daniel Vetter]

On Tue, Jul 13, 2021 at 02:04:31PM +0100, Matthew Auld wrote:
> We skip filling out the pt with scratch entries if the va range covers
> the entire pt, since we later have to fill it with the PTEs for the
> object pages anyway. However this might leave open a small window where
> the PTEs don't point to anything valid for the HW to consume.
> 
> When for example using 2M GTT pages this fill_px() showed up as being
> quite significant in perf measurements, and ends up being completely
> wasted since we ignore the pt and just use the pde directly.
> 
> Anyway, currently we have our PTE construction split between alloc and
> insert, which is probably slightly iffy nowadays, since the alloc
> doesn't actually allocate anything anymore, instead it just sets up the
> page directories and points the PTEs at the scratch page. Later when we
> do the insert step we re-program the PTEs again. Better might be to
> squash the alloc and insert into a single step, then bringing back this
> optimisation(along with some others) should be possible.
> 
> Fixes: 14826673247e ("drm/i915: Only initialize partially filled pagetables")
> Signed-off-by: Matthew Auld <matthew.auld@intel.com>
> Cc: Jon Bloomfield <jon.bloomfield@intel.com>
> Cc: Chris Wilson <chris.p.wilson@intel.com>
> Cc: Daniel Vetter <daniel@ffwll.ch>
> Cc: <stable@vger.kernel.org> # v4.15+

This is some impressively convoluted code, and I'm scared.

But as far as I managed to convince myself, your story here checks out.
Problem will be a bit that this code moved around a _lot_ so we'll need a
lot of dedicated backports :-(

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> ---
>  drivers/gpu/drm/i915/gt/gen8_ppgtt.c | 5 +----
>  1 file changed, 1 insertion(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> index 3d02c726c746..6e0e52eeb87a 100644
> --- a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> +++ b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> @@ -303,10 +303,7 @@ static void __gen8_ppgtt_alloc(struct i915_address_space * const vm,
>  			__i915_gem_object_pin_pages(pt->base);
>  			i915_gem_object_make_unshrinkable(pt->base);
>  
> -			if (lvl ||
> -			    gen8_pt_count(*start, end) < I915_PDES ||
> -			    intel_vgpu_active(vm->i915))
> -				fill_px(pt, vm->scratch[lvl]->encode);
> +			fill_px(pt, vm->scratch[lvl]->encode);
>  
>  			spin_lock(&pd->lock);
>  			if (likely(!pd->entry[idx])) {
> -- 
> 2.26.3
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch