[Mesa-dev] [PATCH] radeonsi: add a debug flag to zero vram allocations

[Grazvydas Ignotas]

This allows to avoid having to see garbage in Dying Light loading screen
at least, which probably expects Windows/NV behavior of all allocations
being zeroed by default.

Analogous to radv flag with the same name.
---
 src/gallium/drivers/radeonsi/si_pipe.c        | 1 +
 src/gallium/drivers/radeonsi/si_pipe.h        | 1 +
 src/gallium/winsys/amdgpu/drm/amdgpu_bo.c     | 3 +++
 src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c | 1 +
 src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h | 1 +
 5 files changed, 7 insertions(+)

diff --git a/src/gallium/drivers/radeonsi/si_pipe.c b/src/gallium/drivers/radeonsi/si_pipe.c
index cc871b1..1c3405f 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -80,10 +80,11 @@ static const struct debug_named_value debug_options[] = {
 	{ "forcedma", DBG(FORCE_DMA), "Use asynchronous DMA for all operations when possible." },
 	{ "nodma", DBG(NO_ASYNC_DMA), "Disable asynchronous DMA" },
 	{ "nowc", DBG(NO_WC), "Disable GTT write combining" },
 	{ "check_vm", DBG(CHECK_VM), "Check VM faults and dump debug info." },
 	{ "reserve_vmid", DBG(RESERVE_VMID), "Force VMID reservation per context." },
+	{ "zerovram", DBG(ZERO_VRAM), "Clear VRAM allocations." },
 
 	/* 3D engine options: */
 	{ "switch_on_eop", DBG(SWITCH_ON_EOP), "Program WD/IA to switch on end-of-packet." },
 	{ "nooutoforder", DBG(NO_OUT_OF_ORDER), "Disable out-of-order rasterization" },
 	{ "nodpbb", DBG(NO_DPBB), "Disable DPBB." },
diff --git a/src/gallium/drivers/radeonsi/si_pipe.h b/src/gallium/drivers/radeonsi/si_pipe.h
index eb0c226..60dc8b0 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.h
+++ b/src/gallium/drivers/radeonsi/si_pipe.h
@@ -139,10 +139,11 @@ enum {
 	DBG_FORCE_DMA,
 	DBG_NO_ASYNC_DMA,
 	DBG_NO_WC,
 	DBG_CHECK_VM,
 	DBG_RESERVE_VMID,
+	DBG_ZERO_VRAM,
 
 	/* 3D engine options: */
 	DBG_SWITCH_ON_EOP,
 	DBG_NO_OUT_OF_ORDER,
 	DBG_NO_DPBB,
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c b/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
index df8b829..15916ad 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
@@ -426,10 +426,13 @@ static struct amdgpu_winsys_bo *amdgpu_create_bo(struct amdgpu_winsys *ws,
    if (flags & RADEON_FLAG_GTT_WC)
       request.flags |= AMDGPU_GEM_CREATE_CPU_GTT_USWC;
    if (flags & RADEON_FLAG_NO_INTERPROCESS_SHARING &&
        ws->info.has_local_buffers)
       request.flags |= AMDGPU_GEM_CREATE_VM_ALWAYS_VALID;
+   if (ws->zero_all_vram_allocs &&
+       (request.preferred_heap & AMDGPU_GEM_DOMAIN_VRAM))
+      request.flags |= AMDGPU_GEM_CREATE_VRAM_CLEARED;
 
    r = amdgpu_bo_alloc(ws->dev, &request, &buf_handle);
    if (r) {
       fprintf(stderr, "amdgpu: Failed to allocate a buffer:\n");
       fprintf(stderr, "amdgpu:    size      : %"PRIu64" bytes\n", size);
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
index caa7991..44429e2 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
@@ -60,10 +60,11 @@ static bool do_winsys_init(struct amdgpu_winsys *ws, int fd)
    }
 
    ws->check_vm = strstr(debug_get_option("R600_DEBUG", ""), "check_vm") != NULL;
    ws->debug_all_bos = debug_get_option_all_bos();
    ws->reserve_vmid = strstr(debug_get_option("R600_DEBUG", ""), "reserve_vmid") != NULL;
+   ws->zero_all_vram_allocs = strstr(debug_get_option("R600_DEBUG", ""), "zerovram") != NULL;
 
    return true;
 
 fail:
    amdgpu_device_deinitialize(ws->dev);
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
index a6784e8..8079255 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
@@ -77,10 +77,11 @@ struct amdgpu_winsys {
    ADDR_HANDLE addrlib;
 
    bool check_vm;
    bool debug_all_bos;
    bool reserve_vmid;
+   bool zero_all_vram_allocs;
 
    /* List of all allocated buffers */
    simple_mtx_t global_bo_list_lock;
    struct list_head global_bo_list;
    unsigned num_buffers;
-- 
2.7.4