exposing the pstate interface through sysfs might make sense with no dynamic reclocking, we shouldn't expose it as sysfs interface, because we actually want to dynamic reclock the card later on.
Doing it in debugfs on the other hand should be fine, also we need something like that later on, when we want to benchmark our dynamic reclocking implementation (we need to disable it at runtime and manucally clock to pstates). I also have some debugfs cstate interface patches, but they actually add new functionality and aren't as trivial as this one, so these have to wait. In the end I wish we have a pstate, cstate and a disable_dyn_reclock interface in debugfs exactly for those benchmarking. Karol Herbst (6): debugfs: add infrastructure to add files with other fops than only read debugfs: rename functions to indicate they are used inside drm debugfs: we need a ctrl object for debugfs debugfs: add copy of sysfs pstate interface ported to debugfs sysfs: remove sysfs interface sysfs: remove pstate options and remove files drm/nouveau/Kbuild | 1 - drm/nouveau/nouveau_debugfs.c | 230 +++++++++++++++++++++++++++++++++++++++++- drm/nouveau/nouveau_debugfs.h | 35 ++++++- drm/nouveau/nouveau_drm.c | 10 +- drm/nouveau/nouveau_drm.h | 2 +- drm/nouveau/nouveau_sysfs.c | 198 ------------------------------------ drm/nouveau/nouveau_sysfs.h | 21 ---- 7 files changed, 262 insertions(+), 235 deletions(-) delete mode 100644 drm/nouveau/nouveau_sysfs.c delete mode 100644 drm/nouveau/nouveau_sysfs.h -- 2.6.1 _______________________________________________ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau