lkml.org 
[lkml]   [2024]   [Feb]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH 4/4] drm/atomic-helper: Add select_output_bus_format callback
Date
Hi Maxime,

Thanks for the review.

> -----Original Message-----
> From: Maxime Ripard <mripard@kernel.org>
> Sent: Wednesday, February 28, 2024 7:30 AM
> To: Klymenko, Anatoliy <Anatoliy.Klymenko@amd.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>; Maarten Lankhorst
> <maarten.lankhorst@linux.intel.com>; Thomas Zimmermann
> <tzimmermann@suse.de>; David Airlie <airlied@gmail.com>; Daniel Vetter
> <daniel@ffwll.ch>; Simek, Michal <michal.simek@amd.com>; Andrzej Hajda
> <andrzej.hajda@intel.com>; Neil Armstrong <neil.armstrong@linaro.org>; Robert
> Foss <rfoss@kernel.org>; Jonas Karlman <jonas@kwiboo.se>; Jernej Skrabec
> <jernej.skrabec@gmail.com>; dri-devel@lists.freedesktop.org; linux-arm-
> kernel@lists.infradead.org; linux-kernel@vger.kernel.org
> Subject: Re: [PATCH 4/4] drm/atomic-helper: Add select_output_bus_format
> callback
>
> Hi,
>
> On Mon, Feb 26, 2024 at 08:44:45PM -0800, Anatoliy Klymenko wrote:
> > Add select_output_bus_format to CRTC atomic helpers callbacks. This
> > callback Will allow CRTC to participate in media bus format
> > negotiation over connected DRM bridge chain and impose CRTC-specific
> restrictions.
> > A good example is CRTC implemented as FPGA soft IP. This kind of CRTC
> > will most certainly support a single output media bus format, as
> > supporting multiple runtime options consumes extra FPGA resources. A
> > variety of options for FPGA are usually achieved by synthesizing IP
> > with different parameters.
> >
> > Incorporate select_output_bus_format callback into the format
> > negotiation stage to fix the input bus format of the first DRM bridge in the
> chain.
> >
> > Signed-off-by: Anatoliy Klymenko <anatoliy.klymenko@amd.com>
> > ---
> > drivers/gpu/drm/drm_bridge.c | 19 +++++++++++++++++--
> > include/drm/drm_modeset_helper_vtables.h | 31
> > +++++++++++++++++++++++++++++++
> > 2 files changed, 48 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/drm_bridge.c
> > b/drivers/gpu/drm/drm_bridge.c index 521a71c61b16..453ae3d174b4 100644
> > --- a/drivers/gpu/drm/drm_bridge.c
> > +++ b/drivers/gpu/drm/drm_bridge.c
> > @@ -32,6 +32,7 @@
> > #include <drm/drm_edid.h>
> > #include <drm/drm_encoder.h>
> > #include <drm/drm_file.h>
> > +#include <drm/drm_modeset_helper_vtables.h>
> > #include <drm/drm_of.h>
> > #include <drm/drm_print.h>
> >
> > @@ -879,7 +880,8 @@ static int select_bus_fmt_recursive(struct drm_bridge
> *first_bridge,
> > unsigned int i, num_in_bus_fmts = 0;
> > struct drm_bridge_state *cur_state;
> > struct drm_bridge *prev_bridge;
> > - u32 *in_bus_fmts;
> > + struct drm_crtc *crtc = crtc_state->crtc;
> > + u32 *in_bus_fmts, in_fmt;
> > int ret;
> >
> > prev_bridge = drm_bridge_get_prev_bridge(cur_bridge);
> > @@ -933,7 +935,20 @@ static int select_bus_fmt_recursive(struct drm_bridge
> *first_bridge,
> > return -ENOMEM;
> >
> > if (first_bridge == cur_bridge) {
> > - cur_state->input_bus_cfg.format = in_bus_fmts[0];
> > + in_fmt = in_bus_fmts[0];
> > + if (crtc->helper_private &&
> > + crtc->helper_private->select_output_bus_format) {
> > + in_fmt = crtc->helper_private-
> >select_output_bus_format(
> > + crtc,
> > + crtc->state,
> > + in_bus_fmts,
> > + num_in_bus_fmts);
> > + if (!in_fmt) {
> > + kfree(in_bus_fmts);
> > + return -ENOTSUPP;
> > + }
> > + }
> > + cur_state->input_bus_cfg.format = in_fmt;
>
> I don't think we should start poking at the CRTC internals, but we should rather
> provide a helper here.

Makes sense, thank you. ACK.

>
> > cur_state->output_bus_cfg.format = out_bus_fmt;
> > kfree(in_bus_fmts);
> > return 0;
> > diff --git a/include/drm/drm_modeset_helper_vtables.h
> > b/include/drm/drm_modeset_helper_vtables.h
> > index 881b03e4dc28..7c21ae1fe3ad 100644
> > --- a/include/drm/drm_modeset_helper_vtables.h
> > +++ b/include/drm/drm_modeset_helper_vtables.h
> > @@ -489,6 +489,37 @@ struct drm_crtc_helper_funcs {
> > bool in_vblank_irq, int *vpos, int *hpos,
> > ktime_t *stime, ktime_t *etime,
> > const struct drm_display_mode *mode);
> > +
> > + /**
> > + * @select_output_bus_format
> > + *
> > + * Called by the first connected DRM bridge to negotiate input media
> > + * bus format. CRTC is expected to pick preferable media formats from
> > + * the list supported by the DRM bridge chain.
>
> There's nothing restricting it to bridges here. Please rephrase this to remove the
> bridge mention. The user is typically going to be the encoder, and bridges are just
> an automagic implementation of an encoder.
>

OK. I'll fix than in the next version.

> And generally speaking, I'd really like to have an implementation available before
> merging this.
>

Well, 2 instances of this callback implementations exist as drafts, as this is the new API. A little bit of a chicken and egg problem. I'll try to groom at least one of them into upstreamable shape and attach it to the patch set.

> Maxime

-Anatoliy

\
 
 \ /
  Last update: 2024-05-27 15:27    [W:0.072 / U:0.480 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site