From: Rodrigo Vivi <rodrigo.vivi@intel.com>
To: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>,
<intel-xe@lists.freedesktop.org>,
Thomas Hellstrom <thomas.hellstrom@intel.com>
Subject: Re: [PATCH 5/7] drm/xe: Define verbose debug macros
Date: Tue, 9 Sep 2025 10:01:17 -0400 [thread overview]
Message-ID: <aMAzLSQuB75QzxDt@intel.com> (raw)
In-Reply-To: <da185812-efc4-4085-832a-3e10b6bc68a5@intel.com>
On Thu, Sep 04, 2025 at 12:19:33AM +0200, Michal Wajdeczko wrote:
>
>
> On 9/4/2025 12:06 AM, Lucas De Marchi wrote:
> > On Wed, Sep 03, 2025 at 11:36:10PM +0200, Michal Wajdeczko wrote:
> >> We might want to add (and in some cases already have) some debug
> >> level messages that are too verbose and need to be guarded by the
> >> CONFIG_DRM_XE_DEBUG. To simplify that define verbose variant of
> >> our printk debug macros that will be compiled out on non-debug
> >> builds, in same way like we already did with xe_asserts() macros.
> >
> > thinking out loud... aren't we going too overboard with this?
> > Looking at other places in the kernel we have pr_debug() and that's
> > awesome as we can simply enable it via dyndbg when we want, with the
> > granularity that matters for the issue we are looking at.
>
> even if some dbg logs are disabled by default and not printing
> anything until explicitly enabled by dyndbg or drm.debug flags,
> they still add some code/text to the driver footprint.
>
> and this patch just allows to add more logs for debug builds
> and strip them in production where we know no one will use them
>
> but still, this is just a proposal
I'm honestly on the fence here... I like the end result of the log messages
with gt and tile very clear. But I don't like the macros very much and
would be totally in favor of some conversion over the pr_debug to follow
the rest of the kernel and avoid the NIH, like i915_utils...
>
> >
> > Lucas De Marchi
> >
> >>
> >> Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
> >> ---
> >> drivers/gpu/drm/xe/xe_gt_printk.h | 6 ++++++
> >> drivers/gpu/drm/xe/xe_printk.h | 6 ++++++
> >> drivers/gpu/drm/xe/xe_tile_printk.h | 6 ++++++
> >> 3 files changed, 18 insertions(+)
> >>
> >> diff --git a/drivers/gpu/drm/xe/xe_gt_printk.h b/drivers/gpu/drm/xe/xe_gt_printk.h
> >> index 8dc4217ddb4e..69c4553d0f80 100644
> >> --- a/drivers/gpu/drm/xe/xe_gt_printk.h
> >> +++ b/drivers/gpu/drm/xe/xe_gt_printk.h
> >> @@ -35,6 +35,12 @@
> >> #define xe_gt_dbg(_gt, _fmt, ...) \
> >> xe_gt_printk((_gt), dbg, _fmt, ##__VA_ARGS__)
> >>
> >> +#if IS_ENABLED(CONFIG_DRM_XE_DEBUG)
> >> +#define xe_gt_dbg_verbose xe_gt_dbg
> >> +#else
> >> +#define xe_gt_dbg_verbose(_gt, ...) typecheck(struct xe_gt *, (_gt))
> >> +#endif
> >> +
> >> #define xe_gt_WARN_type(_gt, _type, _condition, _fmt, ...) \
> >> xe_tile_WARN##_type((_gt)->tile, _condition, _fmt, ## __VA_ARGS__)
> >>
> >> diff --git a/drivers/gpu/drm/xe/xe_printk.h b/drivers/gpu/drm/xe/xe_printk.h
> >> index 691d3611b889..9ffa26c869da 100644
> >> --- a/drivers/gpu/drm/xe/xe_printk.h
> >> +++ b/drivers/gpu/drm/xe/xe_printk.h
> >> @@ -36,6 +36,12 @@
> >> #define xe_dbg(_xe, _fmt, ...) \
> >> xe_printk((_xe), dbg, _fmt, ##__VA_ARGS__)
> >>
> >> +#if IS_ENABLED(CONFIG_DRM_XE_DEBUG)
> >> +#define xe_dbg_verbose xe_dbg
> >> +#else
> >> +#define xe_dbg_verbose(_xe, ...) typecheck(struct xe_device *, (_xe))
> >> +#endif
> >> +
> >> #define xe_WARN_type(_xe, _type, _condition, _fmt, ...) \
> >> drm_WARN##_type(&(_xe)->drm, _condition, _fmt, ## __VA_ARGS__)
> >>
> >> diff --git a/drivers/gpu/drm/xe/xe_tile_printk.h b/drivers/gpu/drm/xe/xe_tile_printk.h
> >> index ed951cdb08cb..33ebfd9722d3 100644
> >> --- a/drivers/gpu/drm/xe/xe_tile_printk.h
> >> +++ b/drivers/gpu/drm/xe/xe_tile_printk.h
> >> @@ -34,6 +34,12 @@
> >> #define xe_tile_dbg(_tile, _fmt, ...) \
> >> xe_tile_printk((_tile), dbg, _fmt, ##__VA_ARGS__)
> >>
> >> +#if IS_ENABLED(CONFIG_DRM_XE_DEBUG)
> >> +#define xe_tile_dbg_verbose xe_tile_dbg
> >> +#else
> >> +#define xe_tile_dbg_verbose(_tile, ...) typecheck(struct xe_tile *, (_tile))
> >> +#endif
> >> +
> >> #define xe_tile_WARN_type(_tile, _type, _condition, _fmt, ...) \
> >> xe_WARN##_type((_tile)->xe, _condition, _fmt, ## __VA_ARGS__)
> >>
> >> --
> >> 2.47.1
> >>
>
next prev parent reply other threads:[~2025-09-09 14:02 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-03 21:36 [PATCH 0/7] RFC: Add dedicated printk macros for tile and device Michal Wajdeczko
2025-09-03 21:36 ` [PATCH 1/7] drm/xe: Keep xe_gt_err() macro definitions together Michal Wajdeczko
2025-09-03 21:59 ` Lucas De Marchi
2025-09-03 21:36 ` [PATCH 2/7] drm/xe: Drop "gt_" prefix from xe_gt_WARN() macros Michal Wajdeczko
2025-09-03 22:00 ` Lucas De Marchi
2025-09-03 21:36 ` [PATCH 3/7] drm/xe: Prepare format for GT-oriented messages in one place Michal Wajdeczko
2025-09-09 13:55 ` Rodrigo Vivi
2025-09-03 21:36 ` [PATCH 4/7] drm/xe: Add dedicated printk macros for tile and device Michal Wajdeczko
2025-09-09 13:57 ` Rodrigo Vivi
2025-09-03 21:36 ` [PATCH 5/7] drm/xe: Define verbose debug macros Michal Wajdeczko
2025-09-03 22:06 ` Lucas De Marchi
2025-09-03 22:19 ` Michal Wajdeczko
2025-09-09 14:01 ` Rodrigo Vivi [this message]
2025-09-03 21:36 ` [PATCH 6/7] drm/xe: Use tile-oriented messages in GGTT code Michal Wajdeczko
2025-09-09 13:58 ` Rodrigo Vivi
2025-09-03 21:36 ` [PATCH 7/7] drm/xe: Use tile-oriented messages in LMTT code Michal Wajdeczko
2025-09-09 13:58 ` Rodrigo Vivi
2025-09-03 21:43 ` ✗ CI.checkpatch: warning for RFC: Add dedicated printk macros for tile and device Patchwork
2025-09-03 21:44 ` ✓ CI.KUnit: success " Patchwork
2025-09-03 22:21 ` ✗ Xe.CI.BAT: failure " Patchwork
2025-09-04 4:49 ` ✗ 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=aMAzLSQuB75QzxDt@intel.com \
--to=rodrigo.vivi@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=lucas.demarchi@intel.com \
--cc=michal.wajdeczko@intel.com \
--cc=thomas.hellstrom@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