Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Ghimiray, Himal Prasad" <himal.prasad.ghimiray@intel.com>
To: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
	intel-xe@lists.freedesktop.org
Cc: Matthew Auld <matthew.auld@intel.com>
Subject: Re: [PATCH v6 6/9] drm/xe/xe2: Update emit_pte to use compression enabled PAT index
Date: Fri, 8 Dec 2023 10:36:04 +0530	[thread overview]
Message-ID: <2d4e34eb-075c-458f-87ea-02d6c569105d@intel.com> (raw)
In-Reply-To: <ea7ef5d1-c1bc-caac-6106-ab7a5b66d97f@linux.intel.com>


On 07-12-2023 18:17, Thomas Hellström wrote:
>
> On 12/7/23 10:19, Himal Prasad Ghimiray wrote:
>> For indirect accessed buffer use compression enabled PAT index.
>>
>> v2:
>>   - Fix parameter name.
>>
>> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
>> Cc: Matthew Auld <matthew.auld@intel.com>
>> Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
>> ---
>>   drivers/gpu/drm/xe/tests/xe_migrate.c |  2 +-
>>   drivers/gpu/drm/xe/xe_migrate.c       | 20 ++++++++++++++------
>>   2 files changed, 15 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/xe/tests/xe_migrate.c 
>> b/drivers/gpu/drm/xe/tests/xe_migrate.c
>> index 83d6a66ed369..f77477f7e9fa 100644
>> --- a/drivers/gpu/drm/xe/tests/xe_migrate.c
>> +++ b/drivers/gpu/drm/xe/tests/xe_migrate.c
>> @@ -330,7 +330,7 @@ static void xe_migrate_sanity_test(struct 
>> xe_migrate *m, struct kunit *test)
>>       else
>>           xe_res_first_sg(xe_bo_sg(pt), 0, pt->size, &src_it);
>>   -    emit_pte(m, bb, NUM_KERNEL_PDE - 1, xe_bo_is_vram(pt),
>> +    emit_pte(m, bb, NUM_KERNEL_PDE - 1, xe_bo_is_vram(pt), false,
>>            &src_it, XE_PAGE_SIZE, pt);
>>         run_sanity_job(m, xe, bb, bb->len, "Writing PTE for our fake 
>> PT", test);
>> diff --git a/drivers/gpu/drm/xe/xe_migrate.c 
>> b/drivers/gpu/drm/xe/xe_migrate.c
>> index 98dca906a023..1bfb249680f4 100644
>> --- a/drivers/gpu/drm/xe/xe_migrate.c
>> +++ b/drivers/gpu/drm/xe/xe_migrate.c
>> @@ -416,15 +416,23 @@ static u32 pte_update_size(struct xe_migrate *m,
>>     static void emit_pte(struct xe_migrate *m,
>>                struct xe_bb *bb, u32 at_pt,
>> -             bool is_vram,
>> +             bool is_vram, bool is_comp_pte,
>>                struct xe_res_cursor *cur,
>>                u32 size, struct xe_bo *bo)
>>   {
>> -    u16 pat_index = tile_to_xe(m->tile)->pat.idx[XE_CACHE_WB];
>> +    struct xe_device *xe = tile_to_xe(m->tile);
>> +
>> +    u16 pat_index;
>>       u32 ptes;
>>       u64 ofs = at_pt * XE_PAGE_SIZE;
>>       u64 cur_ofs;
>>   +    /* Indirect access needs compression enabled uncached PAT 
>> index */
>> +    if (GRAPHICS_VERx100(xe) >= 2000)
>> +        pat_index = is_comp_pte ? 12 : xe->pat.idx[XE_CACHE_NONE];
>
> Please use a relevant define instead of  "12".
Sure.
>
>> +    else
>> +        pat_index = xe->pat.idx[XE_CACHE_WB];
>> +
>>       /*
>>        * FIXME: Emitting VRAM PTEs to L0 PTs is forbidden. Currently
>>        * we're only emitting VRAM PTEs during sanity tests, so when
>> @@ -722,19 +730,19 @@ struct dma_fence *xe_migrate_copy(struct 
>> xe_migrate *m,
>>           }
>>             if (!src_is_vram)
>> -            emit_pte(m, bb, src_L0_pt, src_is_vram, &src_it, src_L0,
>> +            emit_pte(m, bb, src_L0_pt, src_is_vram, true, &src_it, 
>> src_L0,
>>                    src_bo);
>>           else
>>               xe_res_next(&src_it, src_L0);
>>             if (!dst_is_vram)
>> -            emit_pte(m, bb, dst_L0_pt, dst_is_vram, &dst_it, src_L0,
>> +            emit_pte(m, bb, dst_L0_pt, dst_is_vram, true, &dst_it, 
>> src_L0,
>>                    dst_bo);
>>           else
>>               xe_res_next(&dst_it, src_L0);
>>             if (copy_system_ccs)
>> -            emit_pte(m, bb, ccs_pt, false, &ccs_it, ccs_size, src_bo);
>> +            emit_pte(m, bb, ccs_pt, false, false, &ccs_it, ccs_size, 
>> src_bo);
>>             bb->cs[bb->len++] = MI_BATCH_BUFFER_END;
>>           update_idx = bb->len;
>> @@ -975,7 +983,7 @@ struct dma_fence *xe_migrate_clear(struct 
>> xe_migrate *m,
>>             /* Preemption is enabled again by the ring ops. */
>>           if (!clear_vram) {
>> -            emit_pte(m, bb, clear_L0_pt, clear_vram, &src_it, clear_L0,
>> +            emit_pte(m, bb, clear_L0_pt, clear_vram, true, &src_it, 
>> clear_L0,
>>                    bo);
>>           } else {
>>               xe_res_next(&src_it, clear_L0);

  reply	other threads:[~2023-12-08  5:06 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-12-07  9:19 [Intel-xe] [PATCH v6 0/9] Enable compression handling on LNL Himal Prasad Ghimiray
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 1/9] drm/xe/xe2: Determine bios enablement for flat ccs on igfx Himal Prasad Ghimiray
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 2/9] drm/xe/xe2: Allocate extra pages for ccs during bo create Himal Prasad Ghimiray
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 3/9] drm/xe/xe2: Updates on XY_CTRL_SURF_COPY_BLT Himal Prasad Ghimiray
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 4/9] drm/xe/xe_migrate: Use NULL 1G PTE mapped at 255GiB VA for ccs clear Himal Prasad Ghimiray
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 5/9] drm/xe/xe2: Update chunk size for each iteration of ccs copy Himal Prasad Ghimiray
2023-12-07 12:44   ` Thomas Hellström
2023-12-08  5:05     ` Ghimiray, Himal Prasad
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 6/9] drm/xe/xe2: Update emit_pte to use compression enabled PAT index Himal Prasad Ghimiray
2023-12-07 12:47   ` Thomas Hellström
2023-12-08  5:06     ` Ghimiray, Himal Prasad [this message]
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 7/9] drm/xe/xe2: Handle flat ccs move for igfx Himal Prasad Ghimiray
2023-12-07 12:58   ` Thomas Hellström
2023-12-11  5:15     ` Ghimiray, Himal Prasad
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 8/9] drm/xe/xe2: Modify xe_bo_test for system memory Himal Prasad Ghimiray
2023-12-07 13:00   ` Thomas Hellström
2023-12-07  9:19 ` [Intel-xe] [PATCH v6 9/9] drm/xe/xe2: Support flat ccs Himal Prasad Ghimiray
2023-12-07 13:01   ` Thomas Hellström
2023-12-08  5:08     ` Ghimiray, Himal Prasad
2023-12-07 10:20 ` ✓ CI.Patch_applied: success for Enable compression handling on LNL. (rev7) Patchwork
2023-12-07 10:20 ` ✓ CI.checkpatch: " Patchwork
2023-12-07 10:21 ` ✓ CI.KUnit: " Patchwork
2023-12-07 10:28 ` ✓ CI.Build: " Patchwork
2023-12-07 10:29 ` ✓ CI.Hooks: " Patchwork
2023-12-07 10:30 ` ✓ CI.checksparse: " Patchwork
2023-12-07 11:07 ` ✗ CI.BAT: failure " 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=2d4e34eb-075c-458f-87ea-02d6c569105d@intel.com \
    --to=himal.prasad.ghimiray@intel.com \
    --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