sna: Implement a VMA cache

A VMA cache appears unavoidable thanks to compiz and an excrutiatingly
slow GTT pagefault, though it does look like it will be ineffectual
during everyday usage. Compiz (and presumably other compositing
managers) appears to be undoing all the pagefault minimisation as
demonstrated on gen5 with large XPutImage. It also appears the CPU to
memory bandwidth ratio plays a crucial role in determining whether
going straight to GTT or through the CPU cache is a win - so no trivial
heuristic.

x11perf -putimage10 -putimage500 on i5-2467m:
Before:
  bare:   1150,000   2,410
  compiz:  438,000   2,670
After:
  bare:   1190,000   2,730
  compiz:  437,000   2,690
UXA:
  bare:    658,000   2,670
  compiz:  389,000   2,520

On i3-330m
Before:
  bare:    537,000   1,080
  compiz:  263,000     398
After:
  bare:    606,000   1,360
  compiz:  203,000     985
UXA:
  bare:    294,000   1,070
  compiz:  197,000     821

On pnv:
Before:
  bare:    179,000   213
  compiz:  106,000   123
After:
  bare:    181,000   246
  compiz:  103,000   197
UXA:
  bare:    114,000   312
  compiz:   75,700   191

Reported-by: Michael Larabel <Michael@phoronix.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

1 files changed, 4 insertions, 4 deletions

diff --git a/src/sna/kgem_debug_gen5.c b/src/sna/kgem_debug_gen5.c
index 78ba4432..c4f5df15 100644
--- a/src/sna/kgem_debug_gen5.c
+++ b/src/sna/kgem_debug_gen5.c
@@ -85,7 +85,7 @@ static void gen5_update_vertex_buffer(struct kgem *kgem, const uint32_t *data)
 
 	i = data[0] >> 27;
 	if (state.vb[i].current)
-		munmap(state.vb[i].base, state.vb[i].current->size);
+		kgem_bo_unmap(kgem, state.vb[i].current);
 
 	state.vb[i].handle = reloc->target_handle;
 	state.vb[i].current = bo;
@@ -394,7 +394,7 @@ static void
 put_reloc(struct kgem *kgem, struct reloc *r)
 {
 	if (r->bo != NULL)
-		munmap(r->base, r->bo->size);
+		kgem_bo_umap(kgem, r->bo);
 }
 #endif
 
@@ -673,7 +673,7 @@ static void finish_vertex_buffers(struct kgem *kgem)
 
 	for (i = 0; i < ARRAY_SIZE(state.vb); i++)
 		if (state.vb[i].current)
-			munmap(state.vb[i].base, state.vb[i].current->size);
+			kgem_bo_unmap(kgem, state.vb[i].current);
 }
 
 void kgem_gen5_finish_state(struct kgem *kgem)
@@ -681,7 +681,7 @@ void kgem_gen5_finish_state(struct kgem *kgem)
 	finish_vertex_buffers(kgem);
 
 	if (state.dynamic_state.current)
-		munmap(state.dynamic_state.base, state.dynamic_state.current->size);
+		kgem_bo_unmap(kgem,state. dynamic_state.current);
 
 	memset(&state, 0, sizeof(state));
 }