From: Matthew Brost <matthew.brost@intel.com>
To: intel-xe@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: airlied@gmail.com, christian.koenig@amd.com,
thomas.hellstrom@linux.intel.com, matthew.auld@intel.com,
daniel@ffwll.ch
Subject: [RFC PATCH 02/28] drm/xe: Invalidate media_gt TLBs in PT code
Date: Tue, 27 Aug 2024 19:48:35 -0700 [thread overview]
Message-ID: <20240828024901.2582335-3-matthew.brost@intel.com> (raw)
In-Reply-To: <20240828024901.2582335-1-matthew.brost@intel.com>
Testing on LNL has shown media GT's TLBs need to be invalidated via the
GuC, update PT code appropriately.
v2:
- Do dma_fence_get before first call of invalidation_fence_init (Himal)
- No need to check for valid chain fence (Himal)
v3:
- Use dma-fence-array
Fixes: 3330361543fc ("drm/xe/lnl: Add LNL platform definition")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
drivers/gpu/drm/xe/xe_pt.c | 117 ++++++++++++++++++++++++++++++-------
1 file changed, 96 insertions(+), 21 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c
index 579ed31b46db..d6353e8969f0 100644
--- a/drivers/gpu/drm/xe/xe_pt.c
+++ b/drivers/gpu/drm/xe/xe_pt.c
@@ -3,6 +3,8 @@
* Copyright © 2022 Intel Corporation
*/
+#include <linux/dma-fence-array.h>
+
#include "xe_pt.h"
#include "regs/xe_gtt_defs.h"
@@ -1627,9 +1629,11 @@ xe_pt_update_ops_rfence_interval(struct xe_vm_pgtable_update_ops *pt_update_ops,
static int vma_reserve_fences(struct xe_device *xe, struct xe_vma *vma)
{
+ int shift = xe_device_get_root_tile(xe)->media_gt ? 1 : 0;
+
if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
return dma_resv_reserve_fences(xe_vma_bo(vma)->ttm.base.resv,
- xe->info.tile_count);
+ xe->info.tile_count << shift);
return 0;
}
@@ -1816,6 +1820,7 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
struct xe_vm_pgtable_update_ops *pt_update_ops =
&vops->pt_update_ops[tile->id];
struct xe_vma_op *op;
+ int shift = tile->media_gt ? 1 : 0;
int err;
lockdep_assert_held(&vops->vm->lock);
@@ -1824,7 +1829,7 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
xe_pt_update_ops_init(pt_update_ops);
err = dma_resv_reserve_fences(xe_vm_resv(vops->vm),
- tile_to_xe(tile)->info.tile_count);
+ tile_to_xe(tile)->info.tile_count << shift);
if (err)
return err;
@@ -1849,13 +1854,20 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
static void bind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
struct xe_vm_pgtable_update_ops *pt_update_ops,
- struct xe_vma *vma, struct dma_fence *fence)
+ struct xe_vma *vma, struct dma_fence *fence,
+ struct dma_fence *fence2)
{
- if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+ if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm) {
dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence,
pt_update_ops->wait_vm_bookkeep ?
DMA_RESV_USAGE_KERNEL :
DMA_RESV_USAGE_BOOKKEEP);
+ if (fence2)
+ dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence2,
+ pt_update_ops->wait_vm_bookkeep ?
+ DMA_RESV_USAGE_KERNEL :
+ DMA_RESV_USAGE_BOOKKEEP);
+ }
vma->tile_present |= BIT(tile->id);
vma->tile_staged &= ~BIT(tile->id);
if (xe_vma_is_userptr(vma)) {
@@ -1875,13 +1887,20 @@ static void bind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
static void unbind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
struct xe_vm_pgtable_update_ops *pt_update_ops,
- struct xe_vma *vma, struct dma_fence *fence)
+ struct xe_vma *vma, struct dma_fence *fence,
+ struct dma_fence *fence2)
{
- if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+ if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm) {
dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence,
pt_update_ops->wait_vm_bookkeep ?
DMA_RESV_USAGE_KERNEL :
DMA_RESV_USAGE_BOOKKEEP);
+ if (fence2)
+ dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence2,
+ pt_update_ops->wait_vm_bookkeep ?
+ DMA_RESV_USAGE_KERNEL :
+ DMA_RESV_USAGE_BOOKKEEP);
+ }
vma->tile_present &= ~BIT(tile->id);
if (!vma->tile_present) {
list_del_init(&vma->combined_links.rebind);
@@ -1898,7 +1917,8 @@ static void unbind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
static void op_commit(struct xe_vm *vm,
struct xe_tile *tile,
struct xe_vm_pgtable_update_ops *pt_update_ops,
- struct xe_vma_op *op, struct dma_fence *fence)
+ struct xe_vma_op *op, struct dma_fence *fence,
+ struct dma_fence *fence2)
{
xe_vm_assert_held(vm);
@@ -1907,26 +1927,28 @@ static void op_commit(struct xe_vm *vm,
if (!op->map.immediate && xe_vm_in_fault_mode(vm))
break;
- bind_op_commit(vm, tile, pt_update_ops, op->map.vma, fence);
+ bind_op_commit(vm, tile, pt_update_ops, op->map.vma, fence,
+ fence2);
break;
case DRM_GPUVA_OP_REMAP:
unbind_op_commit(vm, tile, pt_update_ops,
- gpuva_to_vma(op->base.remap.unmap->va), fence);
+ gpuva_to_vma(op->base.remap.unmap->va), fence,
+ fence2);
if (op->remap.prev)
bind_op_commit(vm, tile, pt_update_ops, op->remap.prev,
- fence);
+ fence, fence2);
if (op->remap.next)
bind_op_commit(vm, tile, pt_update_ops, op->remap.next,
- fence);
+ fence, fence2);
break;
case DRM_GPUVA_OP_UNMAP:
unbind_op_commit(vm, tile, pt_update_ops,
- gpuva_to_vma(op->base.unmap.va), fence);
+ gpuva_to_vma(op->base.unmap.va), fence, fence2);
break;
case DRM_GPUVA_OP_PREFETCH:
bind_op_commit(vm, tile, pt_update_ops,
- gpuva_to_vma(op->base.prefetch.va), fence);
+ gpuva_to_vma(op->base.prefetch.va), fence, fence2);
break;
default:
drm_warn(&vm->xe->drm, "NOT POSSIBLE");
@@ -1963,7 +1985,9 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
struct xe_vm_pgtable_update_ops *pt_update_ops =
&vops->pt_update_ops[tile->id];
struct dma_fence *fence;
- struct invalidation_fence *ifence = NULL;
+ struct invalidation_fence *ifence = NULL, *mfence = NULL;
+ struct dma_fence **fences = NULL;
+ struct dma_fence_array *cf = NULL;
struct xe_range_fence *rfence;
struct xe_vma_op *op;
int err = 0, i;
@@ -1996,6 +2020,23 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
err = -ENOMEM;
goto kill_vm_tile1;
}
+ if (tile->media_gt) {
+ mfence = kzalloc(sizeof(*ifence), GFP_KERNEL);
+ if (!mfence) {
+ err = -ENOMEM;
+ goto free_ifence;
+ }
+ fences = kmalloc_array(2, sizeof(*fences), GFP_KERNEL);
+ if (!fences) {
+ err = -ENOMEM;
+ goto free_ifence;
+ }
+ cf = dma_fence_array_alloc(2);
+ if (!cf) {
+ err = -ENOMEM;
+ goto free_ifence;
+ }
+ }
}
rfence = kzalloc(sizeof(*rfence), GFP_KERNEL);
@@ -2027,19 +2068,50 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
/* tlb invalidation must be done before signaling rebind */
if (ifence) {
+ if (mfence)
+ dma_fence_get(fence);
invalidation_fence_init(tile->primary_gt, ifence, fence,
pt_update_ops->start,
pt_update_ops->last, vm->usm.asid);
- fence = &ifence->base.base;
+ if (mfence) {
+ invalidation_fence_init(tile->media_gt, mfence, fence,
+ pt_update_ops->start,
+ pt_update_ops->last, vm->usm.asid);
+ fences[0] = &ifence->base.base;
+ fences[1] = &mfence->base.base;
+ dma_fence_array_init(cf, 2, fences,
+ vm->composite_fence_ctx,
+ vm->composite_fence_seqno++,
+ false);
+ fence = &cf->base;
+ } else {
+ fence = &ifence->base.base;
+ }
}
- dma_resv_add_fence(xe_vm_resv(vm), fence,
- pt_update_ops->wait_vm_bookkeep ?
- DMA_RESV_USAGE_KERNEL :
- DMA_RESV_USAGE_BOOKKEEP);
+ if (!mfence) {
+ dma_resv_add_fence(xe_vm_resv(vm), fence,
+ pt_update_ops->wait_vm_bookkeep ?
+ DMA_RESV_USAGE_KERNEL :
+ DMA_RESV_USAGE_BOOKKEEP);
- list_for_each_entry(op, &vops->list, link)
- op_commit(vops->vm, tile, pt_update_ops, op, fence);
+ list_for_each_entry(op, &vops->list, link)
+ op_commit(vops->vm, tile, pt_update_ops, op, fence, NULL);
+ } else {
+ dma_resv_add_fence(xe_vm_resv(vm), &ifence->base.base,
+ pt_update_ops->wait_vm_bookkeep ?
+ DMA_RESV_USAGE_KERNEL :
+ DMA_RESV_USAGE_BOOKKEEP);
+
+ dma_resv_add_fence(xe_vm_resv(vm), &mfence->base.base,
+ pt_update_ops->wait_vm_bookkeep ?
+ DMA_RESV_USAGE_KERNEL :
+ DMA_RESV_USAGE_BOOKKEEP);
+
+ list_for_each_entry(op, &vops->list, link)
+ op_commit(vops->vm, tile, pt_update_ops, op,
+ &ifence->base.base, &mfence->base.base);
+ }
if (pt_update_ops->needs_userptr_lock)
up_read(&vm->userptr.notifier_lock);
@@ -2049,6 +2121,9 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
free_rfence:
kfree(rfence);
free_ifence:
+ kfree(cf);
+ kfree(fences);
+ kfree(mfence);
kfree(ifence);
kill_vm_tile1:
if (err != -EAGAIN && tile->id)
--
2.34.1
next prev parent reply other threads:[~2024-08-28 2:48 UTC|newest]
Thread overview: 100+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-08-28 2:48 [RFC PATCH 00/28] Introduce GPU SVM and Xe SVM implementation Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 01/28] dma-buf: Split out dma fence array create into alloc and arm functions Matthew Brost
2024-08-28 2:48 ` Matthew Brost [this message]
2024-08-28 2:48 ` [RFC PATCH 03/28] drm/xe: Retry BO allocation Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 04/28] mm/migrate: Add migrate_device_vma_range Matthew Brost
2024-08-29 9:03 ` Daniel Vetter
2024-08-29 15:58 ` Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 05/28] drm/gpusvm: Add support for GPU Shared Virtual Memory Matthew Brost
2024-08-28 14:31 ` Daniel Vetter
2024-08-28 14:46 ` Christian König
2024-08-28 15:43 ` Matthew Brost
2024-08-28 16:06 ` Alex Deucher
2024-08-28 16:25 ` Daniel Vetter
2024-08-29 16:40 ` Matthew Brost
2024-09-02 11:29 ` Daniel Vetter
2024-08-30 5:00 ` Matthew Brost
2024-09-02 11:36 ` Daniel Vetter
2024-08-28 18:50 ` Daniel Vetter
2024-08-29 16:49 ` Matthew Brost
2024-09-02 11:40 ` Daniel Vetter
2024-08-29 9:16 ` Thomas Hellström
2024-08-29 17:45 ` Matthew Brost
2024-08-29 18:13 ` Matthew Brost
2024-08-29 19:18 ` Thomas Hellström
2024-08-29 20:56 ` Matthew Brost
2024-08-30 8:18 ` Thomas Hellström
2024-08-30 13:58 ` Matthew Brost
2024-09-02 9:57 ` Thomas Hellström
2024-08-30 9:57 ` Thomas Hellström
2024-08-30 13:47 ` Matthew Brost
2024-09-02 9:45 ` Thomas Hellström
2024-09-02 12:33 ` Daniel Vetter
2024-09-04 12:27 ` Thomas Hellström
2024-09-24 8:41 ` Simona Vetter
2024-08-30 1:35 ` Matthew Brost
2024-08-29 9:45 ` Daniel Vetter
2024-08-29 17:27 ` Matthew Brost
2024-09-02 11:53 ` Daniel Vetter
2024-09-02 17:03 ` Matthew Brost
2024-09-11 16:06 ` Matthew Brost
2024-08-30 9:16 ` Thomas Hellström
2024-09-02 12:20 ` Daniel Vetter
2024-09-06 18:41 ` Zeng, Oak
2024-09-24 9:25 ` Simona Vetter
2024-09-25 16:34 ` Zeng, Oak
2024-09-24 10:42 ` Thomas Hellström
2024-09-24 16:30 ` Matthew Brost
2024-09-25 21:12 ` Matthew Brost
2024-10-09 10:50 ` Thomas Hellström
2024-10-16 3:18 ` Matthew Brost
2024-10-16 6:27 ` Thomas Hellström
2024-10-16 8:24 ` Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 06/28] drm/xe/uapi: Add DRM_XE_VM_BIND_FLAG_SYSTEM_ALLOCATON flag Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 07/28] drm/xe: Add SVM init / fini to faulting VMs Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 08/28] drm/xe: Add dma_addr res cursor Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 09/28] drm/xe: Add SVM range invalidation Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 10/28] drm/gpuvm: Add DRM_GPUVA_OP_USER Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 11/28] drm/xe: Add (re)bind to SVM page fault handler Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 12/28] drm/xe: Add SVM garbage collector Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 13/28] drm/xe: Add unbind to " Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 14/28] drm/xe: Do not allow system allocator VMA unbind if the GPU has bindings Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 15/28] drm/xe: Enable system allocator uAPI Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 16/28] drm/xe: Add migrate layer functions for SVM support Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 17/28] drm/xe: Add SVM device memory mirroring Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 18/28] drm/xe: Add GPUSVM copy SRAM / VRAM vfunc functions Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 19/28] drm/xe: Update PT layer to understand ranges in VRAM Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 20/28] drm/xe: Add Xe SVM populate_vram_pfn vfunc Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 21/28] drm/xe: Add Xe SVM vram_release vfunc Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 22/28] drm/xe: Add BO flags required for SVM Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 23/28] drm/xe: Add SVM VRAM migration Matthew Brost
2024-08-28 16:06 ` Daniel Vetter
2024-08-28 18:22 ` Daniel Vetter
2024-08-29 9:24 ` Christian König
2024-08-29 9:53 ` Thomas Hellström
2024-08-29 11:02 ` Daniel Vetter
2024-08-29 22:12 ` Matthew Brost
2024-08-29 22:23 ` Matthew Brost
2024-09-02 11:01 ` Christian König
2024-09-02 12:50 ` Daniel Vetter
2024-09-02 12:48 ` Daniel Vetter
2024-09-02 22:20 ` Matthew Brost
2024-09-03 8:07 ` Simona Vetter
2024-08-29 14:30 ` Christian König
2024-08-29 21:53 ` Matthew Brost
2024-08-29 21:48 ` Matthew Brost
2024-09-02 13:02 ` Daniel Vetter
2024-08-28 2:48 ` [RFC PATCH 24/28] drm/xe: Basic SVM BO eviction Matthew Brost
2024-08-29 10:14 ` Daniel Vetter
2024-08-29 15:55 ` Matthew Brost
2024-09-02 13:05 ` Daniel Vetter
2024-08-28 2:48 ` [RFC PATCH 25/28] drm/xe: Add SVM debug Matthew Brost
2024-08-28 2:48 ` [RFC PATCH 26/28] drm/xe: Add modparam for SVM notifier size Matthew Brost
2024-08-28 2:49 ` [RFC PATCH 27/28] drm/xe: Add modparam for SVM prefault Matthew Brost
2024-08-28 2:49 ` [RFC PATCH 28/28] drm/gpusvm: Ensure all pages migrated upon eviction Matthew Brost
2024-08-28 2:55 ` ✓ CI.Patch_applied: success for Introduce GPU SVM and Xe SVM implementation Patchwork
2024-08-28 2:55 ` ✗ CI.checkpatch: warning " Patchwork
2024-08-28 2:56 ` ✗ CI.KUnit: failure " Patchwork
2024-09-24 9:16 ` [RFC PATCH 00/28] " Simona Vetter
2024-09-24 19:36 ` Matthew Brost
2024-09-25 11:41 ` Simona Vetter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240828024901.2582335-3-matthew.brost@intel.com \
--to=matthew.brost@intel.com \
--cc=airlied@gmail.com \
--cc=christian.koenig@amd.com \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.auld@intel.com \
--cc=thomas.hellstrom@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox