From: Jim Cromie <jim.cromie@gmail.com>
To: jbaron@akamai.com, gregkh@linuxfoundation.org,
ukaszb@chromium.org, louis.chauvet@bootlin.com,
linux-kernel@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org, amd-gfx@lists.freedesktop.org,
intel-gvt-dev@lists.freedesktop.org,
intel-gfx@lists.freedesktop.org, daniel.vetter@ffwll.ch,
tvrtko.ursulin@linux.intel.com, jani.nikula@intel.com,
ville.syrjala@linux.intel.com, Jim Cromie <jim.cromie@gmail.com>
Subject: [PATCH v3 00/54] Fix CONFIG_DRM_USE_DYNAMIC_DEBUG=y
Date: Wed, 2 Apr 2025 11:41:02 -0600 [thread overview]
Message-ID: <20250402174156.1246171-1-jim.cromie@gmail.com> (raw)
This series fixes dynamic-debug's support for DRM debug-categories.
Classmaps-v1 evaded full review, and got committed in 2 chunks:
b7b4eebdba7b..6ea3bf466ac6 # core dyndbg changes
0406faf25fb1..ee7d633f2dfb # drm adoption
Then DRM-CI found a regression when booting with drm.debug=<initval>;
the pr_debug callsites' static-keys under the drm-dbgs in drm.ko got
enabled, but those in drivers & helpers did not, since they weren't
yet modprobed, and didn't yet exist.
new in v3:
I've flip-flopped again (evolved) on class "protection"
Simply put, /sys/module/drm/parameters/debug is ABI.
If that is true, its an inviolate user expectation that their settings
are authoritative, and we cannot slide dynamic-debug in underneath
without accepting that constraint, special-cases and interface
annoyances not withstanding.
That said, w/o a PARAM, there is no ABI, so classmap use-cases without
one are left unprotected, so that they can be adjusted without
explicit "class FOO" queries. This should satisfy the
keep-both-pieces crowd.
Note also: theres only 1 reason to want to get under the ABI: you want
to turn off some of a class's debugs, because they enlarge and clutter
the logs you must review.
# iirc, these are high volume
:#> echo module driverX class DRM_UT_VBL -p > control
But this isnt actually needed in this case; DRM_UT_VBL is already
pretty specific, and controllable directly from drm.debug.
Someday "mfsl" flags might apply to drm-dbg, but all drm's logging
prefixes are currently hardcoded, and may be semi-ABI, since
log-watchers might key on particular formats (however suboptimal that
might be).
This change is expressed separately, in a new patch:
0028-dyndbg-restore-classmap-protection-when-theres-a-.patch
Minor changes:
. more late cleanup patches pulled forward in parts
. split doc of 2 parsing features
. merge parsing features, tests, docs
. added v2 reviewed-by tags.
. dropped extraneous patches
For your testing convenience, its here:
https://github.com/jimc/linux.git
commit 5cbe833ce591cdb076310b7dcf479fd566f10a22 (HEAD -> dd-fix-614, ghlinux/dd-fix-614)
-v2 is here:
https://lore.kernel.org/lkml/20250320185238.447458-1-jim.cromie@gmail.com/
Design Review / Restatemment:
My primary design goal for dyndbg-classmaps was DIRECT support of DRM,
with NO api changes, starting with the enum drm_debug_category typing,
and keeping as much of the existing macro stack as unaltered as possible.
Some UUID-ish thing was OO-scope; pessimizing unseen optimizations on
compile-time constant ints (that could fit in a byte), across DRM's
macro stack, with ~5k expansions, would be ungood.
Immediate consequences:
= unique integers are *hard* to coordinate across the whole kernel.
= 0 is not special - DRM_UT_CORE is already used.
other users probably want it too
= 0..N are special - in several ways:
1. DRM_UT_<*> is 0..10
its the natural range, everybody wants it.
2. they're exposed in sysfs knobs.
ie: echo 0x1ff > /sys/module/drm/parameters/debug
all classmap-param users get 0..N on their knob.
= DRM_UT_<*> has only 10 categories/classes
small enough for DIRECT storage inside the _ddebug descriptor
= DRM's enum gives us both class-names and class/category values.
classnames === stringified DRM_UT_<*>
class_ids === bare DRM_UT_<*>
all future classmap users are expected to follow this model.
To accommodate the "special" contraints, we need to "privatize" the
class_id's, per module/group. We can do this in a few steps:
= modules declare their map: classnames => Ids.
author defined, user oriented
classnames/strings is wide, expressive, *flat* interface
2-levels used: DRM_UT_*, no 3-levels yet;
_UT could vary, but has no meaning now.
We can use the known (and desired) ordering/contiguity of the 0..Ns
here, and of the array of classnames, to simplify the specification of
the map:
_CLASSMAP_DEFINE (_base-ID/start-of-range, list-of-classnames)
no sparse maps, no reorder-maps, nothing extra to think about. Just
consecutive classnames mapped to consecutive class_ids / categories.
= dyndbg refuses direct selection by class_ids.
this privatizes the numbers
"speak my class-name if you expect a response".
= dyndbg requires "grammar access only"
no other control exists, not contemplated.
ie: "class DRM_UT_CORE +p" >control
ie: "name the class to change it"
class-param handler writes cmd-strings per changed class bit.
= classname --> class_id lookup "validates" the user.
unknown classnames can be discarded, per module.
other modules get the same,
no other class-user would accidentally know "DRM_UT_CORE"
= with private class_ids per module.
a small 0..N range is enough.
32 is a practical limit for a single classmap,
if you think "echo 0x12345678 > $sysknob" is practical.
16 is more practical, DRM has 10.
we have 63.
This allows ...
= multiple classmaps are supported
as long as the mapped class_ids dont conflict.
a fair requirement for a narrow, future use-case.
this is now verified in this patchset.
7x 8-bit classes is workable (if you'd like)
= "legacy" class.
all existing (un-class'd) pr_debugs.
class_id is a 6-bit integer,
so all existing pr_debugs need a default.
63 = 2^6-1 is our _DFLT
0..62 is then available for named classes. (0 is not special)
63 is "legacy".
Jim Cromie (54):
vmlinux.lds.h: fixup HEADERED_SECTION{,_BY} macros
docs/dyndbg: update examples \012 to \n
docs/dyndbg: explain flags parse 1st
test-dyndbg: fixup CLASSMAP usage error
dyndbg: reword "class unknown," to "class:_UNKNOWN_"
dyndbg: make ddebug_class_param union members same size
dyndbg: drop NUM_TYPE_ARRAY
dyndbg: reduce verbose/debug clutter
dyndbg: refactor param_set_dyndbg_classes and below
dyndbg: tighten fn-sig of ddebug_apply_class_bitmap
dyndbg: replace classmap list with a vector
dyndbg: macrofy a 2-index for-loop pattern
dyndbg,module: make proper substructs in _ddebug_info
dyndbg: hoist classmap-filter-by-modname up to ddebug_add_module
dyndbg: ddebug_table.mod_name down to _ddebug_info
dyndbg-API: remove DD_CLASS_TYPE_(DISJOINT|LEVEL)_NAMES and code
dyndbg-API: replace DECLARE_DYNDBG_CLASSMAP
selftests-dyndbg: add tools/testing/selftests/dynamic_debug/*
dyndbg: detect class_id reservation conflicts
dyndbg: check DYNAMIC_DEBUG_CLASSMAP_DEFINE args at compile-time
dyndbg-test: change do_prints testpoint to accept a loopct
dyndbg-API: promote DYNAMIC_DEBUG_CLASSMAP_PARAM to API
dyndbg: treat comma as a token separator
dyndbg: split multi-query strings with %
selftests-dyndbg: add test_mod_submod
dyndbg: change __dynamic_func_call_cls* macros into expressions
dyndbg: drop "protection" of class'd pr_debugs from legacy queries
dyndbg: restore classmap protection when theres a controlling_param
docs/dyndbg: add classmap info to howto
drm: use correct ccflags-y spelling
drm-dyndbg: adapt drm core to use dyndbg classmaps-v2
drm-dyndbg: adapt DRM to invoke DYNAMIC_DEBUG_CLASSMAP_PARAM
drm-print: fix config-dependent unused variable
drm-dyndbg: DRM_CLASSMAP_USE in amdgpu driver
drm-dyndbg: DRM_CLASSMAP_USE in i915 driver
drm-dyndbg: DRM_CLASSMAP_USE in drm_crtc_helper
drm-dyndbg: DRM_CLASSMAP_USE in drm_dp_helper
drm-dyndbg: DRM_CLASSMAP_USE in nouveau
drm-dyndbg: add DRM_CLASSMAP_USE to Xe driver
drm-dyndbg: add DRM_CLASSMAP_USE to virtio_gpu
drm-dyndbg: add DRM_CLASSMAP_USE to simpledrm
drm-dyndbg: add DRM_CLASSMAP_USE to bochs
drm-dyndbg: add DRM_CLASSMAP_USE to etnaviv
drm-dyndbg: add DRM_CLASSMAP_USE to gma500 driver
drm-dyndbg: add DRM_CLASSMAP_USE to radeon
drm-dyndbg: add DRM_CLASSMAP_USE to vmwgfx driver
drm-dyndbg: add DRM_CLASSMAP_USE to vkms driver
drm-dyndbg: add DRM_CLASSMAP_USE to udl driver
drm-dyndbg: add DRM_CLASSMAP_USE to mgag200 driver
drm-dyndbg: add DRM_CLASSMAP_USE to the gud driver
drm-dyndbg: add DRM_CLASSMAP_USE to the qxl driver
drm-dyndbg: add DRM_CLASSMAP_USE to the drm_gem_shmem_helper driver
drm: restore CONFIG_DRM_USE_DYNAMIC_DEBUG un-BROKEN
drm: RFC - make drm_dyndbg_user.o for drm-*_helpers, drivers
.../admin-guide/dynamic-debug-howto.rst | 143 +++-
MAINTAINERS | 3 +-
drivers/gpu/drm/Kconfig | 3 +-
drivers/gpu/drm/Makefile | 12 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 12 +-
drivers/gpu/drm/display/drm_dp_helper.c | 12 +-
drivers/gpu/drm/drm_crtc_helper.c | 12 +-
drivers/gpu/drm/drm_dyndbg_user.c | 11 +
drivers/gpu/drm/drm_gem_shmem_helper.c | 1 +
drivers/gpu/drm/drm_print.c | 38 +-
drivers/gpu/drm/etnaviv/etnaviv_drv.c | 2 +
drivers/gpu/drm/gma500/psb_drv.c | 2 +
drivers/gpu/drm/gud/gud_drv.c | 2 +
drivers/gpu/drm/i915/i915_params.c | 12 +-
drivers/gpu/drm/mgag200/mgag200_drv.c | 2 +
drivers/gpu/drm/nouveau/nouveau_drm.c | 12 +-
drivers/gpu/drm/qxl/qxl_drv.c | 2 +
drivers/gpu/drm/radeon/radeon_drv.c | 2 +
drivers/gpu/drm/tiny/bochs.c | 2 +
drivers/gpu/drm/tiny/simpledrm.c | 2 +
drivers/gpu/drm/udl/udl_main.c | 2 +
drivers/gpu/drm/virtio/virtgpu_drv.c | 2 +
drivers/gpu/drm/vkms/vkms_drv.c | 2 +
drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 2 +
drivers/gpu/drm/xe/xe_drm_client.c | 2 +
include/asm-generic/vmlinux.lds.h | 12 +-
include/drm/drm_print.h | 12 +
include/linux/dynamic_debug.h | 216 ++++--
kernel/module/main.c | 15 +-
lib/Kconfig.debug | 24 +-
lib/Makefile | 3 +
lib/dynamic_debug.c | 637 +++++++++++-------
lib/test_dynamic_debug.c | 203 ++++--
lib/test_dynamic_debug_submod.c | 21 +
tools/testing/selftests/Makefile | 1 +
.../testing/selftests/dynamic_debug/Makefile | 9 +
tools/testing/selftests/dynamic_debug/config | 7 +
.../dynamic_debug/dyndbg_selftest.sh | 373 ++++++++++
38 files changed, 1353 insertions(+), 477 deletions(-)
create mode 100644 drivers/gpu/drm/drm_dyndbg_user.c
create mode 100644 lib/test_dynamic_debug_submod.c
create mode 100644 tools/testing/selftests/dynamic_debug/Makefile
create mode 100644 tools/testing/selftests/dynamic_debug/config
create mode 100755 tools/testing/selftests/dynamic_debug/dyndbg_selftest.sh
--
2.49.0
next reply other threads:[~2025-04-03 8:54 UTC|newest]
Thread overview: 81+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-02 17:41 Jim Cromie [this message]
2025-04-02 17:41 ` [PATCH v3 01/54] vmlinux.lds.h: fixup HEADERED_SECTION{,_BY} macros Jim Cromie
2025-04-15 9:59 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 02/54] docs/dyndbg: update examples \012 to \n Jim Cromie
2025-04-15 9:59 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 03/54] docs/dyndbg: explain flags parse 1st Jim Cromie
2025-04-02 17:41 ` [PATCH v3 04/54] test-dyndbg: fixup CLASSMAP usage error Jim Cromie
2025-04-02 17:41 ` [PATCH v3 05/54] dyndbg: reword "class unknown, " to "class:_UNKNOWN_" Jim Cromie
2025-04-02 17:41 ` [PATCH v3 06/54] dyndbg: make ddebug_class_param union members same size Jim Cromie
2025-04-02 17:41 ` [PATCH v3 07/54] dyndbg: drop NUM_TYPE_ARRAY Jim Cromie
2025-04-02 17:41 ` [PATCH v3 08/54] dyndbg: reduce verbose/debug clutter Jim Cromie
2025-04-02 17:41 ` [PATCH v3 09/54] dyndbg: refactor param_set_dyndbg_classes and below Jim Cromie
2025-04-02 17:41 ` [PATCH v3 10/54] dyndbg: tighten fn-sig of ddebug_apply_class_bitmap Jim Cromie
2025-04-02 17:41 ` [PATCH v3 11/54] dyndbg: replace classmap list with a vector Jim Cromie
2025-04-15 9:59 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 12/54] dyndbg: macrofy a 2-index for-loop pattern Jim Cromie
2025-04-02 17:41 ` [PATCH v3 13/54] dyndbg, module: make proper substructs in _ddebug_info Jim Cromie
2025-04-15 9:59 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 14/54] dyndbg: hoist classmap-filter-by-modname up to ddebug_add_module Jim Cromie
2025-04-15 10:00 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 15/54] dyndbg: ddebug_table.mod_name down to _ddebug_info Jim Cromie
2025-04-02 17:41 ` [PATCH v3 16/54] dyndbg-API: remove DD_CLASS_TYPE_(DISJOINT|LEVEL)_NAMES and code Jim Cromie
2025-04-15 10:00 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 17/54] dyndbg-API: replace DECLARE_DYNDBG_CLASSMAP Jim Cromie
2025-04-15 10:01 ` Louis Chauvet
2025-04-15 19:38 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 18/54] selftests-dyndbg: add tools/testing/selftests/dynamic_debug/* Jim Cromie
2025-04-15 10:02 ` Louis Chauvet
2025-04-15 19:46 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 19/54] dyndbg: detect class_id reservation conflicts Jim Cromie
2025-04-15 10:03 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 20/54] dyndbg: check DYNAMIC_DEBUG_CLASSMAP_DEFINE args at compile-time Jim Cromie
2025-04-15 10:04 ` Louis Chauvet
2025-04-15 19:54 ` jim.cromie
2025-04-16 6:54 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 21/54] dyndbg-test: change do_prints testpoint to accept a loopct Jim Cromie
2025-04-15 10:04 ` Louis Chauvet
2025-04-15 20:15 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 22/54] dyndbg-API: promote DYNAMIC_DEBUG_CLASSMAP_PARAM to API Jim Cromie
2025-04-15 10:06 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 23/54] dyndbg: treat comma as a token separator Jim Cromie
2025-04-15 10:04 ` Louis Chauvet
2025-04-15 20:17 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 24/54] dyndbg: split multi-query strings with % Jim Cromie
2025-04-02 17:41 ` [PATCH v3 25/54] selftests-dyndbg: add test_mod_submod Jim Cromie
2025-04-02 17:41 ` [PATCH v3 26/54] dyndbg: change __dynamic_func_call_cls* macros into expressions Jim Cromie
2025-04-15 10:06 ` Louis Chauvet
2025-04-15 22:49 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 27/54] dyndbg: drop "protection" of class'd pr_debugs from legacy queries Jim Cromie
2025-04-15 10:06 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 28/54] dyndbg: restore classmap protection when theres a controlling_param Jim Cromie
2025-06-20 14:41 ` Sean Paul
2025-07-29 22:00 ` jim.cromie
2025-04-02 17:41 ` [PATCH v3 29/54] docs/dyndbg: add classmap info to howto Jim Cromie
2025-04-15 10:06 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 30/54] drm: use correct ccflags-y spelling Jim Cromie
2025-04-02 17:41 ` [PATCH v3 31/54] drm-dyndbg: adapt drm core to use dyndbg classmaps-v2 Jim Cromie
2025-04-02 17:41 ` [PATCH v3 32/54] drm-dyndbg: adapt DRM to invoke DYNAMIC_DEBUG_CLASSMAP_PARAM Jim Cromie
2025-04-02 17:41 ` [PATCH v3 33/54] drm-print: fix config-dependent unused variable Jim Cromie
2025-04-02 17:41 ` [PATCH v3 34/54] drm-dyndbg: DRM_CLASSMAP_USE in amdgpu driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 35/54] drm-dyndbg: DRM_CLASSMAP_USE in i915 driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 36/54] drm-dyndbg: DRM_CLASSMAP_USE in drm_crtc_helper Jim Cromie
2025-04-02 17:41 ` [PATCH v3 37/54] drm-dyndbg: DRM_CLASSMAP_USE in drm_dp_helper Jim Cromie
2025-04-02 17:41 ` [PATCH v3 38/54] drm-dyndbg: DRM_CLASSMAP_USE in nouveau Jim Cromie
2025-04-02 17:41 ` [PATCH v3 39/54] drm-dyndbg: add DRM_CLASSMAP_USE to Xe driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 40/54] drm-dyndbg: add DRM_CLASSMAP_USE to virtio_gpu Jim Cromie
2025-04-02 17:41 ` [PATCH v3 41/54] drm-dyndbg: add DRM_CLASSMAP_USE to simpledrm Jim Cromie
2025-04-02 17:41 ` [PATCH v3 42/54] drm-dyndbg: add DRM_CLASSMAP_USE to bochs Jim Cromie
2025-04-02 17:41 ` [PATCH v3 43/54] drm-dyndbg: add DRM_CLASSMAP_USE to etnaviv Jim Cromie
2025-04-02 17:41 ` [PATCH v3 44/54] drm-dyndbg: add DRM_CLASSMAP_USE to gma500 driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 45/54] drm-dyndbg: add DRM_CLASSMAP_USE to radeon Jim Cromie
2025-04-02 17:41 ` [PATCH v3 46/54] drm-dyndbg: add DRM_CLASSMAP_USE to vmwgfx driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 47/54] drm-dyndbg: add DRM_CLASSMAP_USE to vkms driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 48/54] drm-dyndbg: add DRM_CLASSMAP_USE to udl driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 49/54] drm-dyndbg: add DRM_CLASSMAP_USE to mgag200 driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 50/54] drm-dyndbg: add DRM_CLASSMAP_USE to the gud driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 51/54] drm-dyndbg: add DRM_CLASSMAP_USE to the qxl driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 52/54] drm-dyndbg: add DRM_CLASSMAP_USE to the drm_gem_shmem_helper driver Jim Cromie
2025-04-02 17:41 ` [PATCH v3 53/54] drm: restore CONFIG_DRM_USE_DYNAMIC_DEBUG un-BROKEN Jim Cromie
2025-04-15 10:00 ` Louis Chauvet
2025-04-02 17:41 ` [PATCH v3 54/54] drm: RFC - make drm_dyndbg_user.o for drm-*_helpers, drivers Jim Cromie
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=20250402174156.1246171-1-jim.cromie@gmail.com \
--to=jim.cromie@gmail.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=daniel.vetter@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=gregkh@linuxfoundation.org \
--cc=intel-gfx@lists.freedesktop.org \
--cc=intel-gvt-dev@lists.freedesktop.org \
--cc=jani.nikula@intel.com \
--cc=jbaron@akamai.com \
--cc=linux-kernel@vger.kernel.org \
--cc=louis.chauvet@bootlin.com \
--cc=tvrtko.ursulin@linux.intel.com \
--cc=ukaszb@chromium.org \
--cc=ville.syrjala@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;
as well as URLs for NNTP newsgroup(s).