winsys/amdgpu: reduce max_alloc_size based on GTT limits

Allocating huge buffers in VRAM is not a problem, but when those buffers
start being migrated, the kernel runs into errors because it cannot split
those buffer up for moving through GTT.

This should fix intermittent failures of
GL45-CTS.texture_buffer.texture_buffer_max_size

Cc: [email protected]
Reviewed-by: Marek Olšák <[email protected]>
(cherry picked from commit 550125e1e73e2441989da11495057a20dd9dad44)
[Emil Velikov: resolve trivial conflicts]
Signed-off-by: Emil Velikov <[email protected]>

Conflicts:
	src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c

1 files changed, 4 insertions, 2 deletions

diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
index e944e62f0aa..573c5acff04 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
@@ -322,8 +322,10 @@ static bool do_winsys_init(struct amdgpu_winsys *ws, int fd)
    /* Set hardware information. */
    ws->info.gart_size = gtt.heap_size;
    ws->info.vram_size = vram.heap_size;
-   /* The kernel can split large buffers, so we can do large allocations. */
-   ws->info.max_alloc_size = MAX2(ws->info.vram_size, ws->info.gart_size) * 0.9;
+   /* The kernel can split large buffers in VRAM but not in GTT, so large
+    * allocations can fail or cause buffer movement failures in the kernel.
+    */
+   ws->info.max_alloc_size = MIN2(ws->info.vram_size * 0.9, ws->info.gart_size * 0.7);
    /* convert the shader clock from KHz to MHz */
    ws->info.max_shader_clock = ws->amdinfo.max_engine_clk / 1000;
    ws->info.max_se = ws->amdinfo.num_shader_engines;