drm/i915: Fix dbuf slice config lookup
authorVille Syrjälä <ville.syrjala@linux.intel.com>
Mon, 7 Feb 2022 13:26:59 +0000 (15:26 +0200)
committerTvrtko Ursulin <tvrtko.ursulin@intel.com>
Mon, 14 Feb 2022 09:52:56 +0000 (09:52 +0000)
Apparently I totally fumbled the loop condition when I
removed the ARRAY_SIZE() stuff from the dbuf slice config
lookup. Comparing the loop index with the active_pipes bitmask
is utter nonsense, what we want to do is check to see if the
mask is zero or not.

Note that the code actually ended up working correctly despite
the fumble, up until commit eef173954432 ("drm/i915: Allow
!join_mbus cases for adlp+ dbuf configuration") when things
broke for real.

Cc: stable@vger.kernel.org
Fixes: 05e8155afe35 ("drm/i915: Use a sentinel to terminate the dbuf slice arrays")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220207132700.481-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit a28fde308c3c1c174249ff9559b57f24e6850086)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
drivers/gpu/drm/i915/intel_pm.c

index 3edba7f..da6ce24 100644 (file)
@@ -4870,7 +4870,7 @@ static u8 compute_dbuf_slices(enum pipe pipe, u8 active_pipes, bool join_mbus,
 {
        int i;
 
-       for (i = 0; i < dbuf_slices[i].active_pipes; i++) {
+       for (i = 0; dbuf_slices[i].active_pipes != 0; i++) {
                if (dbuf_slices[i].active_pipes == active_pipes &&
                    dbuf_slices[i].join_mbus == join_mbus)
                        return dbuf_slices[i].dbuf_mask[pipe];