From: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
To: Matthew Brost <matthew.brost@intel.com>
Cc: <intel-xe@lists.freedesktop.org>, <maarten@lankhorst.se>,
<stuart.summers@intel.com>
Subject: Re: [PATCH] drm/xe: Create LRC BO without VM
Date: Wed, 28 May 2025 22:25:15 -0700 [thread overview]
Message-ID: <aDfvu99Y17J1Xa8L@nvishwa1-desk1> (raw)
In-Reply-To: <aDfgDYLLBipJgA6f@lstrano-desk.jf.intel.com>
On Wed, May 28, 2025 at 09:18:21PM -0700, Matthew Brost wrote:
>On Wed, May 28, 2025 at 09:04:49PM -0700, Niranjana Vishwanathapura wrote:
>> Specifying VM during lrc->bo creation requires VM's reference
>> to be held for the lifetime of lrc->bo as it will use VM's dma
>> reservation object. Using VM's dma reservation object for
>> lrc->bo doesn't provide any advantage. Hence do not pass VM
>> while creating lrc->bo.
>>
>> Fixes: 264eecdba211 ("drm/xe: Decouple xe_exec_queue and xe_lrc")
>> Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
>
>Good cleanup. I thought this through and can't reason why this would
>work or any advantages of the LRC using VM's dma-resv.
>
>One nit, xe_lrc_finish can use xe_bo_unpin_map_no_vm on lrc->bo now.
>
>Asssuming CI passes and the above nit fixed:
>Reviewed-by: Matthew Brost <matthew.brost@intel.com>
>
Thanks. I have uploaded v2 to use xe_bo_unpin_map_no_vm.
Niranjana
>> ---
>> drivers/gpu/drm/xe/xe_exec_queue.c | 9 ---------
>> drivers/gpu/drm/xe/xe_lrc.c | 18 +++---------------
>> 2 files changed, 3 insertions(+), 24 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_exec_queue.c b/drivers/gpu/drm/xe/xe_exec_queue.c
>> index ce78cee5dec6..32c6923df7c3 100644
>> --- a/drivers/gpu/drm/xe/xe_exec_queue.c
>> +++ b/drivers/gpu/drm/xe/xe_exec_queue.c
>> @@ -132,12 +132,6 @@ static int __xe_exec_queue_init(struct xe_exec_queue *q)
>> flags |= XE_LRC_CREATE_RUNALONE;
>> }
>>
>> - if (vm) {
>> - err = xe_vm_lock(vm, true);
>> - if (err)
>> - return err;
>> - }
>> -
>> for (i = 0; i < q->width; ++i) {
>> q->lrc[i] = xe_lrc_create(q->hwe, q->vm, SZ_16K, q->msix_vec, flags);
>> if (IS_ERR(q->lrc[i])) {
>> @@ -146,9 +140,6 @@ static int __xe_exec_queue_init(struct xe_exec_queue *q)
>> }
>> }
>>
>> - if (vm)
>> - xe_vm_unlock(vm);
>> -
>> err = q->ops->init(q);
>> if (err)
>> goto err_lrc;
>> diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c
>> index 61a2e87990a9..974fe509df0e 100644
>> --- a/drivers/gpu/drm/xe/xe_lrc.c
>> +++ b/drivers/gpu/drm/xe/xe_lrc.c
>> @@ -1007,7 +1007,7 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe,
>> * FIXME: Perma-pinning LRC as we don't yet support moving GGTT address
>> * via VM bind calls.
>> */
>> - lrc->bo = xe_bo_create_pin_map(xe, tile, vm, lrc_size,
>> + lrc->bo = xe_bo_create_pin_map(xe, tile, NULL, lrc_size,
>> ttm_bo_type_kernel,
>> bo_flags);
>> if (IS_ERR(lrc->bo))
>> @@ -1795,9 +1795,6 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc)
>> if (!snapshot)
>> return NULL;
>>
>> - if (lrc->bo->vm)
>> - xe_vm_get(lrc->bo->vm);
>> -
>> snapshot->context_desc = xe_lrc_ggtt_addr(lrc);
>> snapshot->ring_addr = __xe_lrc_ring_ggtt_addr(lrc);
>> snapshot->indirect_context_desc = xe_lrc_indirect_ring_ggtt_addr(lrc);
>> @@ -1819,14 +1816,12 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc)
>> void xe_lrc_snapshot_capture_delayed(struct xe_lrc_snapshot *snapshot)
>> {
>> struct xe_bo *bo;
>> - struct xe_vm *vm;
>> struct iosys_map src;
>>
>> if (!snapshot)
>> return;
>>
>> bo = snapshot->lrc_bo;
>> - vm = bo->vm;
>> snapshot->lrc_bo = NULL;
>>
>> snapshot->lrc_snapshot = kvmalloc(snapshot->lrc_size, GFP_KERNEL);
>> @@ -1846,8 +1841,6 @@ void xe_lrc_snapshot_capture_delayed(struct xe_lrc_snapshot *snapshot)
>> xe_bo_unlock(bo);
>> put_bo:
>> xe_bo_put(bo);
>> - if (vm)
>> - xe_vm_put(vm);
>> }
>>
>> void xe_lrc_snapshot_print(struct xe_lrc_snapshot *snapshot, struct drm_printer *p)
>> @@ -1900,14 +1893,9 @@ void xe_lrc_snapshot_free(struct xe_lrc_snapshot *snapshot)
>> return;
>>
>> kvfree(snapshot->lrc_snapshot);
>> - if (snapshot->lrc_bo) {
>> - struct xe_vm *vm;
>> -
>> - vm = snapshot->lrc_bo->vm;
>> + if (snapshot->lrc_bo)
>> xe_bo_put(snapshot->lrc_bo);
>> - if (vm)
>> - xe_vm_put(vm);
>> - }
>> +
>> kfree(snapshot);
>> }
>>
>> --
>> 2.43.0
>>
next prev parent reply other threads:[~2025-05-29 5:28 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-05-29 4:04 [PATCH] drm/xe: Create LRC BO without VM Niranjana Vishwanathapura
2025-05-29 4:18 ` Matthew Brost
2025-05-29 5:25 ` Niranjana Vishwanathapura [this message]
2025-05-29 4:34 ` ✓ CI.Patch_applied: success for " Patchwork
2025-05-29 4:34 ` ✓ CI.checkpatch: " Patchwork
2025-05-29 4:35 ` ✓ CI.KUnit: " Patchwork
2025-05-29 4:45 ` ✓ CI.Build: " Patchwork
2025-05-29 4:48 ` ✓ CI.Hooks: " Patchwork
2025-05-29 4:49 ` ✓ CI.checksparse: " Patchwork
2025-05-29 5:19 ` ✓ Xe.CI.BAT: " Patchwork
2025-05-29 9:53 ` ✓ Xe.CI.Full: " Patchwork
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=aDfvu99Y17J1Xa8L@nvishwa1-desk1 \
--to=niranjana.vishwanathapura@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=maarten@lankhorst.se \
--cc=matthew.brost@intel.com \
--cc=stuart.summers@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