From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Tue, 12 Jun 2018 14:28:20 -0400
Subject: drm/amdgpu: Consolidate visible vs. real vram check v2.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Git-commit: c8c5e569c5b0c9ad523c35497af2ae5788a29581
Patch-mainline: v4.19-rc1
References: FATE#326289 FATE#326079 FATE#326049 FATE#322398 FATE#326166
Move all instnaces of this check into a function in amdgpu_gmc.h
Rename the original function to a more proper name.
v2:
Add more places to cleanup.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Petr Tesarik <ptesarik@suse.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 11 ++++++-----
drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h | 15 +++++++++++++++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 ++++--------
5 files changed, 27 insertions(+), 15 deletions(-)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -31,6 +31,7 @@
#include <drm/drm_syncobj.h>
#include "amdgpu.h"
#include "amdgpu_trace.h"
+#include "amdgpu_gmc.h"
static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
struct drm_amdgpu_cs_chunk_fence *data,
@@ -302,7 +303,7 @@ static void amdgpu_cs_get_threshold_for_
*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
/* Do the same for visible VRAM if half of it is free */
- if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size) {
+ if (!amdgpu_gmc_vram_full_visible(&adev->gmc)) {
u64 total_vis_vram = adev->gmc.visible_vram_size;
u64 used_vis_vram =
amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
@@ -359,7 +360,7 @@ static int amdgpu_cs_bo_validate(struct
* to move it. Don't move anything if the threshold is zero.
*/
if (p->bytes_moved < p->bytes_moved_threshold) {
- if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+ if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
(bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
/* And don't move a CPU_ACCESS_REQUIRED BO to limited
* visible VRAM if we've depleted our allowance to do
@@ -381,7 +382,7 @@ retry:
r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
p->bytes_moved += ctx.bytes_moved;
- if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+ if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
amdgpu_bo_in_cpu_visible_vram(bo))
p->bytes_moved_vis += ctx.bytes_moved;
@@ -434,8 +435,8 @@ static bool amdgpu_cs_try_evict(struct a
/* Good we can try to move this BO somewhere else */
update_bytes_moved_vis =
- adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
- amdgpu_bo_in_cpu_visible_vram(bo);
+ !amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+ amdgpu_bo_in_cpu_visible_vram(bo);
amdgpu_ttm_placement_from_domain(bo, other);
r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
p->bytes_moved += ctx.bytes_moved;
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -109,4 +109,19 @@ struct amdgpu_gmc {
const struct amdgpu_gmc_funcs *gmc_funcs;
};
+/**
+ * amdgpu_gmc_vram_full_visible - Check if full VRAM is visible through the BAR
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * True if full VRAM is visible through the BAR
+ */
+static inline bool amdgpu_gmc_vram_full_visible(struct amdgpu_gmc *gmc)
+{
+ WARN_ON(gmc->real_vram_size < gmc->visible_vram_size);
+
+ return (gmc->real_vram_size == gmc->visible_vram_size);
+}
+
#endif
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -463,7 +463,7 @@ static int amdgpu_bo_do_create(struct am
if (unlikely(r != 0))
return r;
- if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+ if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
bo->tbo.mem.mem_type == TTM_PL_VRAM &&
bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT)
amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved,
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -277,7 +277,7 @@ static void amdgpu_evict_flags(struct tt
if (!adev->mman.buffer_funcs_enabled) {
/* Move to system memory */
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
- } else if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+ } else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
!(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
amdgpu_bo_in_cpu_visible_vram(abo)) {
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -33,6 +33,7 @@
#include "amdgpu.h"
#include "amdgpu_trace.h"
#include "amdgpu_amdkfd.h"
+#include "amdgpu_gmc.h"
/*
* GPUVM
@@ -588,11 +589,6 @@ bool amdgpu_vm_need_pipeline_sync(struct
return vm_flush_needed || gds_switch_needed;
}
-static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev)
-{
- return (adev->gmc.real_vram_size == adev->gmc.visible_vram_size);
-}
-
/**
* amdgpu_vm_flush - hardware flush the vm
*
@@ -2422,7 +2418,7 @@ int amdgpu_vm_init(struct amdgpu_device
}
DRM_DEBUG_DRIVER("VM update mode is %s\n",
vm->use_cpu_for_update ? "CPU" : "SDMA");
- WARN_ONCE((vm->use_cpu_for_update & !amdgpu_vm_is_large_bar(adev)),
+ WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
"CPU update of VM recommended only for large BAR system\n");
vm->last_update = NULL;
@@ -2538,7 +2534,7 @@ int amdgpu_vm_make_compute(struct amdgpu
vm->pte_support_ats = pte_support_ats;
DRM_DEBUG_DRIVER("VM update mode is %s\n",
vm->use_cpu_for_update ? "CPU" : "SDMA");
- WARN_ONCE((vm->use_cpu_for_update & !amdgpu_vm_is_large_bar(adev)),
+ WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
"CPU update of VM recommended only for large BAR system\n");
if (vm->pasid) {
@@ -2713,7 +2709,7 @@ void amdgpu_vm_manager_init(struct amdgp
*/
#ifdef CONFIG_X86_64
if (amdgpu_vm_update_mode == -1) {
- if (amdgpu_vm_is_large_bar(adev))
+ if (amdgpu_gmc_vram_full_visible(&adev->gmc))
adev->vm_manager.vm_update_mode =
AMDGPU_VM_USE_CPU_FOR_COMPUTE;
else