Igt-dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: vitaly prosyak <vprosyak@amd.com>
To: "Srinivasan Shanmugam" <srinivasan.shanmugam@amd.com>,
	"Christian König" <christian.koenig@amd.com>,
	"Alex Deucher" <alexander.deucher@amd.com>,
	"Vitaly Prosyak" <vitaly.prosyak@amd.com>
Cc: igt-dev@lists.freedesktop.org, David Yat Sin <David.YatSin@amd.com>
Subject: Re: [PATCH 0/2] tests/amdgpu: Exercise MMIO_REMAP remap page (alloc/map + HDP flush)
Date: Sat, 20 Sep 2025 00:55:31 -0400	[thread overview]
Message-ID: <55ffa49e-e69e-4e51-a4e4-d586a19d5549@amd.com> (raw)
In-Reply-To: <20250918084655.2192395-1-srinivasan.shanmugam@amd.com>

[-- Attachment #1: Type: text/plain, Size: 2279 bytes --]

Both patches look good overall and provide valuable MMIO_REMAP subtests. 2 improvements are suggested:

1. Remove Excess Debug Logs for both patches except when igt_skip is called.
   Please reduce the verbosity by removing non-essential igt_info() calls, such as:
    igt_info("[mmio_remap] alloc request: size=%u align=%u domain=0x%x flags=0x%llx\n", ...)   etc..
 
 2. Ensure MMIO Write Posting (Patch 2/2)
   In the HDP flush subtest, please insert a memory barrier to ensure the MMIO write is posted before the read:
   mmio[flush_idx] = 0x1;
*  __sync_synchronize();  *
 (void)mmio[flush_idx];
This ensures correctness on platforms with weak memory ordering (e.g., ARM, RISC-V), where MMIO writes can be delayed or reordered.

Reviewed-by: Vitaly Prosyak vitaly.prosyak@amd.com


On 2025-09-18 04:46, Srinivasan Shanmugam wrote:
> This series adds two IGT subtests in amd_bo.c to exercise the MMIO_REMAP
> remap page:
>
>   1) mmio_remap_gem_create_close
>      - Allocates a 4 KiB, page-aligned BO in
>        AMDGPU_GEM_DOMAIN_MMIO_REMAP and immediately frees it. Skips
>        gracefully when the domain is not supported
>        (-EOPNOTSUPP/-ENODEV/-EINVAL) or system PAGE_SIZE > 4K.
>
>   2) mmio_remap_map_and_poke
>      - Allocates and CPU-maps the 4 KiB remap page, then write-1 to
>        HDP_MEM_FLUSH_CNTL (offset 0) twice, followed by a dummy read to
>        post the writes. Unmaps and frees on exit.
> 	   
> These tests help cover the basic KGD/KFD MMIO_REMAP path (allocation and
> BAR-mapped MMIO access).
>
> TODO: A kernel patch enabling this path is pending merge; once it lands,
> the Link below can serve as the stable reference.
>
> Link: https://patchwork.freedesktop.org/series/153209/#rev5
> Cc: Vitaly Prosyak <vitaly.prosyak@amd.com>
> Cc: David Yat Sin <David.YatSin@amd.com>
> Cc: Christian König <christian.koenig@amd.com>
> Cc: Alex Deucher <alexander.deucher@amd.com>
> Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
>
> Srinivasan Shanmugam (2):
>   tests/amdgpu/amd_bo: Add mmio_remap_gem_create_close subtest
>   tests/amdgpu/amd_bo: Add mmio_remap_map_and_poke (HDP flush) subtest
>
>  tests/amdgpu/amd_bo.c | 144 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 144 insertions(+)
>

[-- Attachment #2: Type: text/html, Size: 3440 bytes --]

  parent reply	other threads:[~2025-09-20  4:55 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-18  8:46 [PATCH 0/2] tests/amdgpu: Exercise MMIO_REMAP remap page (alloc/map + HDP flush) Srinivasan Shanmugam
2025-09-18  8:46 ` [PATCH 1/2] tests/amdgpu/amd_bo: Add mmio_remap_gem_create_close subtest Srinivasan Shanmugam
2025-09-18  8:46 ` [PATCH 2/2] tests/amdgpu/amd_bo: Add mmio_remap_map_and_poke (HDP flush) subtest Srinivasan Shanmugam
2025-09-18 13:46 ` ✓ Xe.CI.BAT: success for tests/amdgpu: Exercise MMIO_REMAP remap page (alloc/map + HDP flush) Patchwork
2025-09-18 18:44 ` ✓ i915.CI.BAT: " Patchwork
2025-09-18 21:47 ` ✗ Xe.CI.Full: failure " Patchwork
2025-09-19  1:27 ` ✓ i915.CI.Full: success " Patchwork
2025-09-20  4:55 ` vitaly prosyak [this message]
2025-09-22  1:37 ` [PATCH 0/2] " vitaly prosyak

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=55ffa49e-e69e-4e51-a4e4-d586a19d5549@amd.com \
    --to=vprosyak@amd.com \
    --cc=David.YatSin@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=christian.koenig@amd.com \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=srinivasan.shanmugam@amd.com \
    --cc=vitaly.prosyak@amd.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