[PATCH 1/2] Revert "drm/xe/lnl: Offload system clear page activity to GPU"
Nirmoy Das
nirmoy.das at intel.com
Wed Aug 28 09:14:08 UTC 2024
On 8/28/2024 11:11 AM, Thomas Hellström wrote:
> Hi,
>
> On Wed, 2024-08-28 at 10:36 +0200, Nirmoy Das wrote:
>> This optimization relied on having to clear CCS on allocations.
>> If there is no need to clear CCS on allocations then this would
>> mostly
>> help in reducing CPU utilization.
>>
>> Revert this patch at this moment because of:
>> 1 Currently Xe can't do clear on free and using a invalid ttm flag,
>> TTM_TT_FLAG_CLEARED_ON_FREE which could poison global ttm pool on
>> multi-device setup.
>>
>> 2 Also for LNL CPU:WB doesn't require clearing CCS as such BO will
>> not be allowed to bind with compression PTE. Subsequent patch will
>> disable clearing CCS for CPU:WB BOs for LNL.
>>
>> This reverts commit 23683061805be368c8d1c7e7ff52abc470cac275.
>>
>> Cc: Christian König <christian.koenig at amd.com>
>> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray at intel.com>
>> Cc: Lucas De Marchi <lucas.demarchi at intel.com>
>> Cc: Matthew Auld <matthew.auld at intel.com>
>> Cc: Matthew Brost <matthew.brost at intel.com>
>> Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
Signed-off-by: Nirmoy Das <nirmoy.das at intel.com>
> You'd need an Signed-off-by:
Sorry, missed "-s" while committing
>
> With that,
> Reviewed-by: Thomas Hellström <thomas.hellstrom at linux.intel.com>
Thanks,
Nirmoy
>
>
>> ---
>> drivers/gpu/drm/xe/xe_bo.c | 26 ++------------------------
>> drivers/gpu/drm/xe/xe_device_types.h | 2 --
>> drivers/gpu/drm/xe/xe_ttm_sys_mgr.c | 12 ------------
>> 3 files changed, 2 insertions(+), 38 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
>> index 9d6632f92fa9..25d0c939ba31 100644
>> --- a/drivers/gpu/drm/xe/xe_bo.c
>> +++ b/drivers/gpu/drm/xe/xe_bo.c
>> @@ -396,14 +396,6 @@ static struct ttm_tt *xe_ttm_tt_create(struct
>> ttm_buffer_object *ttm_bo,
>> caching = ttm_uncached;
>> }
>>
>> - /*
>> - * If the device can support gpu clear system pages then set
>> proper ttm
>> - * flag. Zeroed pages are only required for
>> ttm_bo_type_device so
>> - * unwanted data is not leaked to userspace.
>> - */
>> - if (ttm_bo->type == ttm_bo_type_device && xe-
>>> mem.gpu_page_clear_sys)
>> - page_flags |= TTM_TT_FLAG_CLEARED_ON_FREE;
>> -
>> err = ttm_tt_init(&tt->ttm, &bo->ttm, page_flags, caching,
>> extra_pages);
>> if (err) {
>> kfree(tt);
>> @@ -425,10 +417,6 @@ static int xe_ttm_tt_populate(struct ttm_device
>> *ttm_dev, struct ttm_tt *tt,
>> if (tt->page_flags & TTM_TT_FLAG_EXTERNAL)
>> return 0;
>>
>> - /* Clear TTM_TT_FLAG_ZERO_ALLOC when GPU is set to clear
>> system pages */
>> - if (tt->page_flags & TTM_TT_FLAG_CLEARED_ON_FREE)
>> - tt->page_flags &= ~TTM_TT_FLAG_ZERO_ALLOC;
>> -
>> err = ttm_pool_alloc(&ttm_dev->pool, tt, ctx);
>> if (err)
>> return err;
>> @@ -671,16 +659,8 @@ static int xe_bo_move(struct ttm_buffer_object
>> *ttm_bo, bool evict,
>> bool needs_clear;
>> bool handle_system_ccs = (!IS_DGFX(xe) &&
>> xe_bo_needs_ccs_pages(bo) &&
>> ttm && ttm_tt_is_populated(ttm)) ?
>> true : false;
>> - bool clear_system_pages;
>> int ret = 0;
>>
>> - /*
>> - * Clear TTM_TT_FLAG_CLEARED_ON_FREE on bo creation path
>> when
>> - * moving to system as the bo doesn't have dma_mapping.
>> - */
>> - if (!old_mem && ttm && !ttm_tt_is_populated(ttm))
>> - ttm->page_flags &= ~TTM_TT_FLAG_CLEARED_ON_FREE;
>> -
>> /* Bo creation path, moving to system or TT. */
>> if ((!old_mem && ttm) && !handle_system_ccs) {
>> if (new_mem->mem_type == XE_PL_TT)
>> @@ -703,10 +683,8 @@ static int xe_bo_move(struct ttm_buffer_object
>> *ttm_bo, bool evict,
>> move_lacks_source = handle_system_ccs ? (!bo->ccs_cleared)
>> :
>> (!mem_type_is_vram(o
>> ld_mem_type) && !tt_has_data);
>>
>> - clear_system_pages = ttm && (ttm->page_flags &
>> TTM_TT_FLAG_CLEARED_ON_FREE);
>> needs_clear = (ttm && ttm->page_flags &
>> TTM_TT_FLAG_ZERO_ALLOC) ||
>> - (!ttm && ttm_bo->type == ttm_bo_type_device) ||
>> - clear_system_pages;
>> + (!ttm && ttm_bo->type == ttm_bo_type_device);
>>
>> if (new_mem->mem_type == XE_PL_TT) {
>> ret = xe_tt_map_sg(ttm);
>> @@ -818,7 +796,7 @@ static int xe_bo_move(struct ttm_buffer_object
>> *ttm_bo, bool evict,
>> if (move_lacks_source) {
>> u32 flags = 0;
>>
>> - if (mem_type_is_vram(new_mem->mem_type) ||
>> clear_system_pages)
>> + if (mem_type_is_vram(new_mem->mem_type))
>> flags |= XE_MIGRATE_CLEAR_FLAG_FULL;
>> else if (handle_system_ccs)
>> flags |=
>> XE_MIGRATE_CLEAR_FLAG_CCS_DATA;
>> diff --git a/drivers/gpu/drm/xe/xe_device_types.h
>> b/drivers/gpu/drm/xe/xe_device_types.h
>> index 4ecd620921a3..e73fb0c23932 100644
>> --- a/drivers/gpu/drm/xe/xe_device_types.h
>> +++ b/drivers/gpu/drm/xe/xe_device_types.h
>> @@ -333,8 +333,6 @@ struct xe_device {
>> struct xe_mem_region vram;
>> /** @mem.sys_mgr: system TTM manager */
>> struct ttm_resource_manager sys_mgr;
>> - /** @mem.gpu_page_clear_sys: clear system pages
>> offloaded to GPU */
>> - bool gpu_page_clear_sys;
>> } mem;
>>
>> /** @sriov: device level virtualization data */
>> diff --git a/drivers/gpu/drm/xe/xe_ttm_sys_mgr.c
>> b/drivers/gpu/drm/xe/xe_ttm_sys_mgr.c
>> index e0ac20f20758..9844a8edbfe1 100644
>> --- a/drivers/gpu/drm/xe/xe_ttm_sys_mgr.c
>> +++ b/drivers/gpu/drm/xe/xe_ttm_sys_mgr.c
>> @@ -117,17 +117,5 @@ int xe_ttm_sys_mgr_init(struct xe_device *xe)
>> ttm_resource_manager_init(man, &xe->ttm, gtt_size >>
>> PAGE_SHIFT);
>> ttm_set_driver_manager(&xe->ttm, XE_PL_TT, man);
>> ttm_resource_manager_set_used(man, true);
>> -
>> - /*
>> - * On iGFX device with flat CCS, we clear CCS metadata,
>> let's extend that
>> - * and use GPU to clear pages as well.
>> - *
>> - * Disable this when init_on_free and/or init_on_alloc is on
>> to avoid double
>> - * zeroing pages with CPU and GPU.
>> - */
>> - if (xe_device_has_flat_ccs(xe) && !IS_DGFX(xe) &&
>> - !want_init_on_alloc(GFP_USER) && !want_init_on_free())
>> - xe->mem.gpu_page_clear_sys = true;
>> -
>> return drmm_add_action_or_reset(&xe->drm, ttm_sys_mgr_fini,
>> xe);
>> }
More information about the dri-devel
mailing list