Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Souza, Jose" <jose.souza@intel.com>
To: "intel-xe@lists.freedesktop.org" <intel-xe@lists.freedesktop.org>,
	"Auld,  Matthew" <matthew.auld@intel.com>,
	"Yadav, Sanjay Kumar" <sanjay.kumar.yadav@intel.com>
Cc: "Summers, Stuart" <stuart.summers@intel.com>
Subject: Re: [PATCH] drm/xe/query: Only advertise NO_COMPRESSION hint on devices with flat CCS
Date: Fri, 6 Feb 2026 13:28:51 +0000	[thread overview]
Message-ID: <d05696f532f44ce6b13002ddb07d4255a99b1025.camel@intel.com> (raw)
In-Reply-To: <847e5437-994c-4fa2-a316-a6b9f92d73bb@intel.com>

On Thu, 2026-02-05 at 14:17 +0000, Matthew Auld wrote:
> On 05/02/2026 13:43, Souza, Jose wrote:
> > On Thu, 2026-02-05 at 13:50 +0530, Sanjay Yadav wrote:
> > > Devices like CRI have GRAPHICS_VER >= 20 but don't have
> > > compression
> > > support their PAT table (xe3p_xpc_pat_table) has no compression-
> > > enabled
> > > entries.
> > > 
> > > Only advertise HAS_NO_COMPRESSION_HINT flag when the device
> > > actually
> > > has
> > > flat CCS and thus compression capability.
> > 
> > Is this change required to fix a bug? Semantically, it doesn't make
> > much sense. If a device doesn't support compression and we pass
> > DRM_XE_GEM_CREATE_FLAG_NO_COMPRESSION during BO creation, I don't
> > see a
> > problem with it. You could optimize and always enable this flag
> > internally for this device, but passing it during creation isn't
> > wrong.
> 
> Yeah, there is no real bug. But having HAS_NO_COMPRESSION return
> false 
> to reflect that tagging buffers with NO_COMPRESSION is not 
> relevant/needed on this device, since compression as a feature is
> either 
> disabled or does not even exist in the hw, also seems reasonable.
> 
> The motivation was originally IGT, just to make it a bit simpler to 
> detect when this flag is bogus, on certain platforms. And from UMD
> pov 
> having it more accurately reflect whether using this flag is relevant
> or 
> not seemed reasonable.
> 
> We can try to so fix the IGT issue in another way, without touching
> this 
> flag, if preferred.

If the reason is IGT I would fix it there. Do less changes in KMD as
possible but up to you.
Mesa is ready for both cases.

> 
> > 
> > > 
> > > Fixes: 78d91ba6bd796 ("drm/xe/uapi: Add NO_COMPRESSION BO flag
> > > and
> > > query capability")
> > > Cc: Matthew Auld <matthew.auld@intel.com>
> > > Cc: José Roberto de Souza <jose.souza@intel.com>
> > > Cc: Stuart Summers <stuart.summers@intel.com>
> > > Signed-off-by: Sanjay Yadav <sanjay.kumar.yadav@intel.com>
> > > ---
> > >   drivers/gpu/drm/xe/xe_query.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/gpu/drm/xe/xe_query.c
> > > b/drivers/gpu/drm/xe/xe_query.c
> > > index 34db266b723f..c152ebfbd7d8 100644
> > > --- a/drivers/gpu/drm/xe/xe_query.c
> > > +++ b/drivers/gpu/drm/xe/xe_query.c
> > > @@ -335,7 +335,7 @@ static int query_config(struct xe_device *xe,
> > > struct drm_xe_device_query *query)
> > >   	if (xe->info.has_usm &&
> > > IS_ENABLED(CONFIG_DRM_XE_GPUSVM))
> > >   		config->info[DRM_XE_QUERY_CONFIG_FLAGS] |=
> > >   			DRM_XE_QUERY_CONFIG_FLAG_HAS_CPU_ADDR_MI
> > > RROR
> > > ;
> > > -	if (GRAPHICS_VER(xe) >= 20)
> > > +	if (GRAPHICS_VER(xe) >= 20 &&
> > > xe_device_has_flat_ccs(xe))
> > >   		config->info[DRM_XE_QUERY_CONFIG_FLAGS] |=
> > >   			DRM_XE_QUERY_CONFIG_FLAG_HAS_NO_COMPRESS
> > > ION_
> > > HINT;
> > >   	config->info[DRM_XE_QUERY_CONFIG_FLAGS] |=

  reply	other threads:[~2026-02-06 13:28 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-05  8:20 [PATCH] drm/xe/query: Only advertise NO_COMPRESSION hint on devices with flat CCS Sanjay Yadav
2026-02-05  8:31 ` ✓ CI.KUnit: success for " Patchwork
2026-02-05  9:29 ` ✓ Xe.CI.BAT: " Patchwork
2026-02-05  9:39 ` [PATCH] " Vivekanandan, Balasubramani
2026-02-05  9:49   ` Matthew Auld
2026-02-05  9:56 ` Matthew Auld
2026-02-05 13:43 ` Souza, Jose
2026-02-05 14:17   ` Matthew Auld
2026-02-06 13:28     ` Souza, Jose [this message]
2026-02-06  4:33 ` ✓ Xe.CI.FULL: success for " 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=d05696f532f44ce6b13002ddb07d4255a99b1025.camel@intel.com \
    --to=jose.souza@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=matthew.auld@intel.com \
    --cc=sanjay.kumar.yadav@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