On Thu, Sep 26, 2024 at 06:45:14PM +0530, Nautiyal, Ankit K wrote: > > On 9/26/2024 4:49 PM, Ville Syrjälä wrote: > > On Thu, Sep 26, 2024 at 12:56:31PM +0530, Ankit Nautiyal wrote: > >> Pass the current pipe into enabled_joiner_pipes(), and let it figure out > >> the proper bitmasks for us. Since the enabled_joiner_pipes now gets the > >> primary and secondary pipe wrt a given pipe, the helpers > >> to get primary pipe and secondary pipes are no longer required. > >> > >> v2: > >> -Simplify helper get_joiner_primary_pipes. (Ville) > >> -Nuke get_joiner_secondary_pipes. (Ville) > >> -Add more drm_WARNs final primary/secondary pipes. (Ville) > >> v3: > >> -Drop ultrajoiner stuff and add it in subsequent patches. (Ville) > >> > >> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@xxxxxxxxx> > >> Suggested-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> > >> --- > >> drivers/gpu/drm/i915/display/intel_display.c | 89 ++++++++++---------- > >> 1 file changed, 44 insertions(+), 45 deletions(-) > >> > >> diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c > >> index 2d6260c3bca5..ea259b142786 100644 > >> --- a/drivers/gpu/drm/i915/display/intel_display.c > >> +++ b/drivers/gpu/drm/i915/display/intel_display.c > >> @@ -3663,7 +3663,15 @@ static u8 expected_bigjoiner_secondary_pipes(u8 bigjoiner_primary_pipes) > >> return expected_secondary_pipes(bigjoiner_primary_pipes, 2); > >> } > >> > >> +static u8 get_joiner_primary_pipe(enum pipe pipe, u8 primary_pipes) > >> +{ > >> + primary_pipes &= GENMASK(pipe, 0); > >> + > >> + return primary_pipes ? BIT(fls(primary_pipes) - 1) : 0; > >> +} > >> + > >> static void enabled_joiner_pipes(struct drm_i915_private *dev_priv, > >> + enum pipe pipe, > >> u8 *primary_pipes, u8 *secondary_pipes) > >> { > >> struct intel_display *display = to_intel_display(&dev_priv->drm); > >> @@ -3703,45 +3711,38 @@ static void enabled_joiner_pipes(struct drm_i915_private *dev_priv, > >> expected_uncompressed_joiner_secondary_pipes(primary_uncompressed_joiner_pipes), > >> secondary_uncompressed_joiner_pipes); > >> > >> - *primary_pipes = primary_uncompressed_joiner_pipes | primary_bigjoiner_pipes; > >> - > >> - *secondary_pipes = secondary_uncompressed_joiner_pipes | secondary_bigjoiner_pipes; > >> - > >> - /* Joiner pipes should always be consecutive primary and secondary */ > >> - drm_WARN(display->drm, *secondary_pipes != *primary_pipes << 1, > >> - "Joiner misconfigured (primary pipes %#x, secondary pipes %#x)\n", > >> - *primary_pipes, *secondary_pipes); > >> -} > >> - > >> -static enum pipe get_joiner_primary_pipe(enum pipe pipe, u8 primary_pipes, u8 secondary_pipes) > >> -{ > >> - if ((secondary_pipes & BIT(pipe)) == 0) > >> - return pipe; > >> - > >> - /* ignore everything above our pipe */ > >> - primary_pipes &= ~GENMASK(7, pipe); > >> - > >> - /* highest remaining bit should be our primary pipe */ > >> - return fls(primary_pipes) - 1; > >> -} > >> - > >> -static u8 get_joiner_secondary_pipes(enum pipe pipe, u8 primary_pipes, u8 secondary_pipes) > >> -{ > >> - enum pipe primary_pipe, next_primary_pipe; > >> - > >> - primary_pipe = get_joiner_primary_pipe(pipe, primary_pipes, secondary_pipes); > >> - > >> - if ((primary_pipes & BIT(primary_pipe)) == 0) > >> - return 0; > >> - > >> - /* ignore our primary pipe and everything below it */ > >> - primary_pipes &= ~GENMASK(primary_pipe, 0); > >> - /* make sure a high bit is set for the ffs() */ > >> - primary_pipes |= BIT(7); > >> - /* lowest remaining bit should be the next primary pipe */ > >> - next_primary_pipe = ffs(primary_pipes) - 1; > >> - > >> - return secondary_pipes & GENMASK(next_primary_pipe - 1, primary_pipe); > >> + *primary_pipes = 0; > >> + *secondary_pipes = 0; > >> + > >> + if (uncompressed_joiner_pipes & BIT(pipe)) { > >> + *primary_pipes = get_joiner_primary_pipe(pipe, primary_uncompressed_joiner_pipes); > >> + *secondary_pipes = secondary_uncompressed_joiner_pipes & > >> + expected_uncompressed_joiner_secondary_pipes(*primary_pipes); > >> + > >> + drm_WARN(display->drm, > >> + expected_uncompressed_joiner_secondary_pipes(*primary_pipes) != > >> + *secondary_pipes, > >> + "Wrong uncompressed joiner secondary pipes for primary_pipes %#x (expected %#x, current %#x)\n", > >> + *primary_pipes, > >> + expected_uncompressed_joiner_secondary_pipes(*primary_pipes), > >> + *secondary_pipes); > >> + return; > >> + } > >> + > >> + if (bigjoiner_pipes & BIT(pipe)) { > >> + *primary_pipes = get_joiner_primary_pipe(pipe, primary_bigjoiner_pipes); > >> + *secondary_pipes = secondary_bigjoiner_pipes & > >> + expected_bigjoiner_secondary_pipes(*primary_pipes); > >> + > >> + drm_WARN(display->drm, > >> + expected_bigjoiner_secondary_pipes(*primary_pipes) != > >> + *secondary_pipes, > >> + "Wrong bigjoiner secondary pipes for primary_pipes %#x (expected %#x, current %#x)\n", > >> + *primary_pipes, > >> + expected_bigjoiner_secondary_pipes(*primary_pipes), > >> + *secondary_pipes); > >> + return; > >> + } > >> } > >> > >> static u8 hsw_panel_transcoders(struct drm_i915_private *i915) > >> @@ -3813,10 +3814,10 @@ static u8 hsw_enabled_transcoders(struct intel_crtc *crtc) > >> enabled_transcoders |= BIT(cpu_transcoder); > >> > >> /* joiner secondary -> consider the primary pipe's transcoder as well */ > >> - enabled_joiner_pipes(dev_priv, &primary_pipes, &secondary_pipes); > >> + enabled_joiner_pipes(dev_priv, crtc->pipe, &primary_pipes, &secondary_pipes); > >> if (secondary_pipes & BIT(crtc->pipe)) { > >> cpu_transcoder = (enum transcoder) > >> - get_joiner_primary_pipe(crtc->pipe, primary_pipes, secondary_pipes); > >> + ffs(get_joiner_primary_pipe(crtc->pipe, primary_pipes)) - 1; > > The get_joiner_primary_pipe() shouldn't be needed here since > > enabled_joiner_pipes() guarantees that only one bit is set. > > > I agree. Additionally, I was considering changing the input variable > name from `primary_pipes` to `primary_pipe` in enabled_joiner_pipes for > clarity. Does that make sense? Probably a good idea. It'll be a bit weird to have it as a bitmask, but feels like returning just a single enum pipe could make things more complicated. So yeah, 'u8 *primary_pipe' seems perhaps the best compromise. -- Ville Syrjälä Intel