Hi Maxime,
Thank you for the review.
> -----Original Message-----
> From: Maxime Ripard <mripard@xxxxxxxxxx>
> Sent: Thursday, March 14, 2024 5:05 AM
> To: Klymenko, Anatoliy <Anatoliy.Klymenko@xxxxxxx>
> Cc: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>; Maarten Lankhorst
> <maarten.lankhorst@xxxxxxxxxxxxxxx>; Thomas Zimmermann
> <tzimmermann@xxxxxxx>; David Airlie <airlied@xxxxxxxxx>; Daniel Vetter
> <daniel@xxxxxxxx>; Simek, Michal <michal.simek@xxxxxxx>; Andrzej Hajda
> <andrzej.hajda@xxxxxxxxx>; Neil Armstrong <neil.armstrong@xxxxxxxxxx>; Robert
> Foss <rfoss@xxxxxxxxxx>; Jonas Karlman <jonas@xxxxxxxxx>; Jernej Skrabec
> <jernej.skrabec@xxxxxxxxx>; Rob Herring <robh+dt@xxxxxxxxxx>; Krzysztof
> Kozlowski <krzysztof.kozlowski+dt@xxxxxxxxxx>; Conor Dooley
> <conor+dt@xxxxxxxxxx>; Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>; dri-
> devel@xxxxxxxxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx; linux-
> media@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2 8/8] drm: xlnx: Intoduce TPG CRTC driver
>
> Hi,
>
> On Tue, Mar 12, 2024 at 05:55:05PM -0700, Anatoliy Klymenko wrote:
> > DO NOT MERGE. REFERENCE ONLY.
> >
> > Add CRTC driver based on AMD/Xilinx Video Test Pattern Generator IP.
> > TPG based FPGA design represents minimalistic harness useful for
> > testing links between FPGA based CRTC and external DRM encoders, both
> > FPGA and hardened IP based.
> >
> > Add driver for AMD/Xilinx Video Timing Controller. The VTC, working in
> > generator mode, suplements TPG with video timing signals.
> >
> > Signed-off-by: Anatoliy Klymenko <anatoliy.klymenko@xxxxxxx>
>
> As I said previously, we don't want to have unused APIs, so this patch should be in
> a good enough state to be merged if we want to merge the whole API.
>
This is understandable, but even having this API just reviewed by the community will open the path forward for aligning AMD/Xilinx downstream DRM drivers with the upstream kernel.
> > +/*
> > +---------------------------------------------------------------------
> > +--------
> > + * DRM CRTC
> > + */
> > +
> > +static enum drm_mode_status xlnx_tpg_crtc_mode_valid(struct drm_crtc
> *crtc,
> > + const struct
> drm_display_mode *mode) {
> > + return MODE_OK;
> > +}
> > +
> > +static int xlnx_tpg_crtc_check(struct drm_crtc *crtc,
> > + struct drm_atomic_state *state) {
> > + struct drm_crtc_state *crtc_state =
> drm_atomic_get_new_crtc_state(state, crtc);
> > + int ret;
> > +
> > + if (!crtc_state->enable)
> > + goto out;
> > +
> > + ret = drm_atomic_helper_check_crtc_primary_plane(crtc_state);
> > + if (ret)
> > + return ret;
> > +
> > +out:
> > + return drm_atomic_add_affected_planes(state, crtc); }
> > +
>
> [...]
>
> > +
> > +static u32 xlnx_tpg_crtc_select_output_bus_format(struct drm_crtc *crtc,
> > + struct drm_crtc_state
> *crtc_state,
> > + const u32 *in_bus_fmts,
> > + unsigned int
> num_in_bus_fmts) {
> > + struct xlnx_tpg *tpg = crtc_to_tpg(crtc);
> > + unsigned int i;
> > +
> > + for (i = 0; i < num_in_bus_fmts; ++i)
> > + if (in_bus_fmts[i] == tpg->output_bus_format)
> > + return tpg->output_bus_format;
> > +
> > + return 0;
> > +}
> > +
> > +static const struct drm_crtc_helper_funcs xlnx_tpg_crtc_helper_funcs = {
> > + .mode_valid = xlnx_tpg_crtc_mode_valid,
> > + .atomic_check = xlnx_tpg_crtc_check,
> > + .atomic_enable = xlnx_tpg_crtc_enable,
> > + .atomic_disable = xlnx_tpg_crtc_disable,
> > + .select_output_bus_format = xlnx_tpg_crtc_select_output_bus_format,
> > +};
>
> From that code, it's not clear to me how the CRTC is going to be able to get what
> the format is.
>
It's coming from DT "bus-format" property. The idea is that this property will reflect the FPGA design variation synthesized.
> It looks like you hardcode it here, but what if there's several that would fit the
> bill? Is the CRTC expected to store it into its private structure?
>
It's impractical from the resources utilization point of view to support multiple runtime options for FPGA-based CRTCs output signal format, so the bus-format will be runtime fixed but can vary between differently synthesized instances.
> If so, I would expect it to be in the crtc state, and atomic_enable to just reuse
> whatever is in the state.
>
This could be totally valid for different kinds of CRTCs, although for this particular case, the bus-fomat choice is runtime immutable.
> Maxime
Thank you,
Anatoliy
