[0/7] DYNAMIC_DEBUG fixups for rc

Message ID 20221111221715.563020-1-jim.cromie@gmail.com
Headers
Series DYNAMIC_DEBUG fixups for rc |

Message

Jim Cromie Nov. 11, 2022, 10:17 p.m. UTC
  hi Jason, Greg, DRM-folk,

drm.debug-on-dyndbg has a regression due to a chicken-&-egg problem;
drm.debug is applied to enable dyndbg callsites before the dependent
modules' callsites are available to be enabled.

My "fixes" are unready, so lets just mark it BROKEN for now.

Meanwhile, heres some other fixes, a comment tweak, a proof of
non-bug, an internal simplification, and a cleanup/improvement to the
main macro (API):

Split DECLARE_DYNDBG_CLASSMAP in 1/2; REFERENCE_DYNDBG_CLASSMAP now
refers to a classmap DECLARE'd just once.  I think this gives a path
away from the coordination-by-identical-classmaps "feature" that Jani
and others thought was "weird" (my term).


Jim Cromie (7):
  drm: mark drm.debug-on-dyndbg as BROKEN for now
  drm_print: fixup improve stale comment
  test-dyndbg: fixup CLASSMAP usage error
  test-dyndbg: show that DEBUG enables prdbgs at compiletime
  dyndbg: fix readback value on LEVEL_NAMES interfaces
  dyndbg: clone DECLARE_DYNDBG_CLASSMAP to REFERENCE_DYNDBG_CLASSMAP
  dyndbg: replace classmap list with a vector

 drivers/gpu/drm/Kconfig                 |  1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c |  2 +-
 drivers/gpu/drm/display/drm_dp_helper.c |  2 +-
 drivers/gpu/drm/drm_crtc_helper.c       |  2 +-
 drivers/gpu/drm/i915/i915_params.c      |  2 +-
 drivers/gpu/drm/nouveau/nouveau_drm.c   |  2 +-
 include/drm/drm_print.h                 |  5 +-
 include/linux/dynamic_debug.h           | 10 ++++
 lib/dynamic_debug.c                     | 63 +++++++++++++------------
 lib/test_dynamic_debug.c                |  4 +-
 10 files changed, 57 insertions(+), 36 deletions(-)
  

Comments

Greg KH Nov. 17, 2022, 6:29 a.m. UTC | #1
On Fri, Nov 11, 2022 at 03:17:08PM -0700, Jim Cromie wrote:
> hi Jason, Greg, DRM-folk,
> 
> drm.debug-on-dyndbg has a regression due to a chicken-&-egg problem;
> drm.debug is applied to enable dyndbg callsites before the dependent
> modules' callsites are available to be enabled.
> 
> My "fixes" are unready, so lets just mark it BROKEN for now.
> 
> Meanwhile, heres some other fixes, a comment tweak, a proof of
> non-bug, an internal simplification, and a cleanup/improvement to the
> main macro (API):
> 
> Split DECLARE_DYNDBG_CLASSMAP in 1/2; REFERENCE_DYNDBG_CLASSMAP now
> refers to a classmap DECLARE'd just once.  I think this gives a path
> away from the coordination-by-identical-classmaps "feature" that Jani
> and others thought was "weird" (my term).
> 

Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>