Quoting Matthew Auld (2021-02-12 13:43:46)
> On Fri, 12 Feb 2021 at 10:22, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > Introduce the concept of padding the i915_vma with guard pages before
> > and aft. The major consequence is that all ordinary uses of i915_vma
> > must use i915_vma_offset/i915_vma_size and not i915_vma.node.start/size
> > directly, as the drm_mm_node will include the guard pages that surround
> > our object.
> >
> > So in this patch, we look for all uses of i915_vma->node.start that
> > instead need to include the guard offset and switch them to
> > i915_vma_offset(), and in a few cases to i915_ggtt_offset(). Notable
> > exceptions are the selftests, which expect exact behaviour.
> >
> > The biggest connundrum is how exactly to mix request a fixed address
> > with guard pages, particular through the existing uABI. The user does
> > not know about guard pages, so such must be transparent to the user, and
> > so the execobj.offset must be that of the object itself excluding the
> > guard. So a PIN_OFFSET_FIXED must then be exclusive of the guard pages.
> >
> > In the next patch, we start using guard pages for scanout objects. While
> > these are limited to GGTT vma, on a few platforms these vma (or at least
> > an alias of the vma) is shared with userspace, so we may leak the
> > existence of such guards if we are not careful to ensure that the
> > execobj.offset is transparent and excludes the guards. (On such platforms,
> > without full-ppgtt, userspace has to use relocations so the presence of
> > more untouchable regions within its GTT such be of no further issue.)
> >
> > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > Cc: Matthew Auld <matthew.auld@xxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/gt/intel_ggtt.c | 12 ++++++++++--
> > drivers/gpu/drm/i915/i915_vma.c | 10 +++++++---
> > drivers/gpu/drm/i915/i915_vma.h | 8 ++++----
> > drivers/gpu/drm/i915/i915_vma_types.h | 3 ++-
> > 4 files changed, 23 insertions(+), 10 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c
> > index c5803c434d33..6b326138e765 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c
> > @@ -238,8 +238,12 @@ static void gen8_ggtt_insert_entries(struct i915_address_space *vm,
> >
> > gte = (gen8_pte_t __iomem *)ggtt->gsm;
> > gte += vma->node.start / I915_GTT_PAGE_SIZE;
> > - end = gte + vma->node.size / I915_GTT_PAGE_SIZE;
> >
> > + end = gte + vma->guard / I915_GTT_PAGE_SIZE;
> > + while (gte < end)
> > + gen8_set_pte(gte++, vm->scratch[0]->encode);
> > +
> > + end += (vma->node.size - vma->guard) / I915_GTT_PAGE_SIZE;
> > for_each_sgt_daddr(addr, iter, vma->pages)
> > gen8_set_pte(gte++, pte_encode | addr);
> > GEM_BUG_ON(gte > end);
> > @@ -289,8 +293,12 @@ static void gen6_ggtt_insert_entries(struct i915_address_space *vm,
> >
> > gte = (gen6_pte_t __iomem *)ggtt->gsm;
> > gte += vma->node.start / I915_GTT_PAGE_SIZE;
> > - end = gte + vma->node.size / I915_GTT_PAGE_SIZE;
> >
> > + end = gte + vma->guard / I915_GTT_PAGE_SIZE;
> > + while (gte < end)
> > + gen8_set_pte(gte++, vm->scratch[0]->encode);
> > +
> > + end += (vma->node.size - vma->guard) / I915_GTT_PAGE_SIZE;
> > for_each_sgt_daddr(addr, iter, vma->pages)
> > iowrite32(vm->pte_encode(addr, level, flags), gte++);
> > GEM_BUG_ON(gte > end);
> > diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
> > index 17fe455bd770..155f510b4cc6 100644
> > --- a/drivers/gpu/drm/i915/i915_vma.c
> > +++ b/drivers/gpu/drm/i915/i915_vma.c
> > @@ -623,7 +623,7 @@ bool i915_gem_valid_gtt_space(struct i915_vma *vma, unsigned long color)
> > static int
> > i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
> > {
> > - unsigned long color;
> > + unsigned long color, guard;
> > u64 start, end;
> > int ret;
> >
> > @@ -631,13 +631,16 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
> > GEM_BUG_ON(drm_mm_node_allocated(&vma->node));
> >
> > size = max(size, vma->size);
> > - alignment = max(alignment, vma->display_alignment);
> > + alignment = max_t(typeof(alignment), alignment, vma->display_alignment);
> > if (flags & PIN_MAPPABLE) {
> > size = max_t(typeof(size), size, vma->fence_size);
> > alignment = max_t(typeof(alignment),
> > alignment, vma->fence_alignment);
> > }
> >
> > + guard = 0;
> > + size += 2 * guard;
> > +
> > GEM_BUG_ON(!IS_ALIGNED(size, I915_GTT_PAGE_SIZE));
> > GEM_BUG_ON(!IS_ALIGNED(alignment, I915_GTT_MIN_ALIGNMENT));
> > GEM_BUG_ON(!is_power_of_2(alignment));
> > @@ -674,7 +677,7 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
> > return -EINVAL;
>
> Should we bother adjusting the overflows checking to account for the
> padding? Also maybe check that offset >= guard?
offset >= guard only for pinned.
And yes, it is sensible that we check we don't overflow with adding a guard.
>
> if (flags & PIN_OFFSET_FIXED) {
> u64 offset = flags & PIN_OFFSET_MASK;
> +
> + if (offset < guard)
> + return -ENOSPC;
> +
> if (!IS_ALIGNED(offset, alignment) ||
> - range_overflows(offset, size, end))
> + range_overflows(offset, size - 2 * guard, end))
> return -EINVAL;
>
> If the padding spills over the end, maybe it's ok, so long as the
> "real" range doesn't spill over, like if it needs to be placed within
> the mappable range. Otherwise maybe something like
> range_overflows(offset - guard, size - guard, end)?
My thought was that the user has no knowledge of the guard pages, nor
direct access to them [in theory], so they should not be penalised for
their placement, as far as is possible.
That we have to prevent wrap around and so restrict total placement to
[guard, end - guard] unfortunately belies our attempt to be
undetectable.
So the -EINVAL check should be against the full range, but we should
also do
if (offset < guard || offset > gtt_end - guard)
return -ENOSPC;
> Also bumping the alignment constraint for the OFFSET_FIXED case does
> seem a little scary? Is it not possible to just add the required
> padding around both ends of the provided offset?
Hmm. We don't need to bump the alignment, we just need to make sure the
guard is a multiple of the alignment (so that start + guard still meets
the alignment constraints).
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
