git/linux-cryptodev-2.6
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/herbert/cryptodev-2.6.git synced 2026-04-18 03:23:53 -04:00
Code Issues Packages Projects Releases Wiki Activity

drm/amdgpu: use hmm_pfns instead of array of pages

Browse Source 
we dont need to allocate local array of pages to hold
the pages returned by the hmm, instead we could use
the hmm_range structure itself to get to hmm_pfn
and get the required pages directly.

This avoids call to alloc/free quite a lot.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Acked-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Sunil Khatri
2025-09-17 20:12:43 +05:30
committed by Alex Deucher
parent ae4d627e43
commit c5b3cc417b
9 changed files with 25 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
View File

@@ -29,6 +29,7 @@
#include <linux/pagemap.h>
#include <linux/sync_file.h>
#include <linux/dma-buf.h>
#include <linux/hmm.h>
#include <drm/amdgpu_drm.h>
#include <drm/drm_syncobj.h>
@@ -885,24 +886,12 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
struct amdgpu_bo *bo = e->bo;
int i;
e->user_pages = kvcalloc(bo->tbo.ttm->num_pages,
sizeof(struct page *),
GFP_KERNEL);
if (!e->user_pages) {
drm_err(adev_to_drm(p->adev), "kvmalloc_array failure\n");
r = -ENOMEM;
r = amdgpu_ttm_tt_get_user_pages(bo, &e->range);
if (r)
goto out_free_user_pages;
}
r = amdgpu_ttm_tt_get_user_pages(bo, e->user_pages, &e->range);
if (r) {
kvfree(e->user_pages);
e->user_pages = NULL;
goto out_free_user_pages;
}
for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
if (bo->tbo.ttm->pages[i] != e->user_pages[i]) {
if (bo->tbo.ttm->pages[i] != hmm_pfn_to_page(e->range->hmm_pfns[i])) {
userpage_invalidated = true;
break;
}
@@ -946,7 +935,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
}
if (amdgpu_ttm_tt_is_userptr(e->bo->tbo.ttm) &&
e->user_invalidated && e->user_pages) {
e->user_invalidated) {
amdgpu_bo_placement_from_domain(e->bo,
AMDGPU_GEM_DOMAIN_CPU);
r = ttm_bo_validate(&e->bo->tbo, &e->bo->placement,
@@ -955,11 +944,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
goto out_free_user_pages;
amdgpu_ttm_tt_set_user_pages(e->bo->tbo.ttm,
e->user_pages);
e->range);
}
kvfree(e->user_pages);
e->user_pages = NULL;
}
amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
@@ -1001,11 +987,7 @@ out_free_user_pages:
amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
struct amdgpu_bo *bo = e->bo;
if (!e->user_pages)
continue;
amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range);
kvfree(e->user_pages);
e->user_pages = NULL;
e->range = NULL;
}
mutex_unlock(&p->bo_list->bo_list_mutex);