On Sun, Jul 1, 2018 at 11:13 PM, Jernej Škrabec <jernej.skrabec@xxxxxxxx> wrote:
> Dne nedelja, 01. julij 2018 ob 15:52:55 CEST je Chen-Yu Tsai napisal(a):
>> On Sun, Jul 1, 2018 at 6:41 PM, Jernej Škrabec <jernej.skrabec@xxxxxxxx>
> wrote:
>> > Dne četrtek, 28. junij 2018 ob 08:51:07 CEST je Chen-Yu Tsai napisal(a):
>> >> On Thu, Jun 28, 2018 at 1:15 PM, Jernej Škrabec <jernej.skrabec@xxxxxxxx>
>> >
>> > wrote:
>> >> > Dne četrtek, 28. junij 2018 ob 04:50:09 CEST je Chen-Yu Tsai
> napisal(a):
>> >> >> On Mon, Jun 25, 2018 at 8:03 PM, Jernej Skrabec
>> >> >> <jernej.skrabec@xxxxxxxx>
>> >> >
>> >> > wrote:
>> >> >> > Add all entries needed for HDMI to function properly.
>> >> >> >
>> >> >> > Since R40 has highly configurable pipeline, both mixers and both
>> >> >> > TCON
>> >> >> > TVs are added. Board specific DT should then connect them together
>> >> >> > trough TCON TOP muxers to best fit the purpose of the board.
>> >> >> >
>> >> >> > Signed-off-by: Jernej Skrabec <jernej.skrabec@xxxxxxxx>
>> >> >> > ---
>> >> >> >
>> >> >> > arch/arm/boot/dts/sun8i-r40.dtsi | 269
>> >> >> > +++++++++++++++++++++++++++++++
>> >> >> > 1 file changed, 269 insertions(+)
>> >> >> >
>> >> >> > diff --git a/arch/arm/boot/dts/sun8i-r40.dtsi
>> >> >> > b/arch/arm/boot/dts/sun8i-r40.dtsi index 173dcc1652d2..a2a75fb04caf
>> >> >> > 100644
>> >> >> > --- a/arch/arm/boot/dts/sun8i-r40.dtsi
>> >> >> > +++ b/arch/arm/boot/dts/sun8i-r40.dtsi
>> >> >> > @@ -42,8 +42,11 @@
>> >> >> >
>> >> >> > */
>> >> >> >
>> >> >> > #include <dt-bindings/interrupt-controller/arm-gic.h>
>> >> >> >
>> >> >> > +#include <dt-bindings/clock/sun8i-de2.h>
>> >> >> >
>> >> >> > #include <dt-bindings/clock/sun8i-r40-ccu.h>
>> >> >> >
>> >> >> > +#include <dt-bindings/clock/sun8i-tcon-top.h>
>> >> >
>> >> > Maxime, above line breaks compilation for build robot, sorry.
>> >> >
>> >> >> > #include <dt-bindings/reset/sun8i-r40-ccu.h>
>> >> >> >
>> >> >> > +#include <dt-bindings/reset/sun8i-de2.h>
>> >> >> >
>> >> >> > / {
>> >> >> >
>> >> >> > #address-cells = <1>;
>> >> >> >
>> >> >> > @@ -99,12 +102,76 @@
>> >> >> >
>> >> >> > };
>> >> >> >
>> >> >> > };
>> >> >> >
>> >> >> > + de: display-engine {
>> >> >> > + compatible = "allwinner,sun8i-r40-display-engine",
>> >> >> > + "allwinner,sun8i-h3-display-engine";
>> >> >>
>> >> >> Given that the display pipeline looks different, they should not be
>> >> >> compatible.
>> >> >
>> >> > Ok.
>> >> >
>> >> >> > + allwinner,pipelines = <&mixer0>, <&mixer1>;
>> >> >> > + status = "disabled";
>> >> >> > + };
>> >> >> > +
>> >> >> >
>> >> >> > soc {
>> >> >> >
>> >> >> > compatible = "simple-bus";
>> >> >> > #address-cells = <1>;
>> >> >> > #size-cells = <1>;
>> >> >> > ranges;
>> >> >> >
>> >> >> > + display_clocks: clock@1000000 {
>> >> >> > + compatible = "allwinner,sun8i-r40-de2-clk",
>> >> >> > + "allwinner,sun8i-h3-de2-clk";
>> >> >> > + reg = <0x01000000 0x100000>;
>> >> >> > + clocks = <&ccu CLK_DE>,
>> >> >> > + <&ccu CLK_BUS_DE>;
>> >> >> > + clock-names = "mod",
>> >> >> > + "bus";
>> >> >> > + resets = <&ccu RST_BUS_DE>;
>> >> >> > + #clock-cells = <1>;
>> >> >> > + #reset-cells = <1>;
>> >> >> > + };
>> >> >> > +
>> >> >> > + mixer0: mixer@1100000 {
>> >> >> > + compatible =
>> >> >> > "allwinner,sun8i-r40-de2-mixer-0";
>> >> >> > + reg = <0x01100000 0x100000>;
>> >> >> > + clocks = <&display_clocks CLK_BUS_MIXER0>,
>> >> >> > + <&display_clocks CLK_MIXER0>;
>> >> >> > + clock-names = "bus",
>> >> >> > + "mod";
>> >> >> > + resets = <&display_clocks RST_MIXER0>;
>> >> >> > +
>> >> >> > + ports {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > +
>> >> >> > + mixer0_out: port@1 {
>> >> >> > + reg = <1>;
>> >> >> > + mixer0_out_tcon_top:
>> >> >> > endpoint {
>> >> >> > + remote-endpoint =
>> >> >> > <&tcon_top_mixer0_in_mixer0>; +
>> >> >> > };
>> >> >> > + };
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + mixer1: mixer@1200000 {
>> >> >> > + compatible =
>> >> >> > "allwinner,sun8i-r40-de2-mixer-1";
>> >> >> > + reg = <0x01200000 0x100000>;
>> >> >> > + clocks = <&display_clocks CLK_BUS_MIXER1>,
>> >> >> > + <&display_clocks CLK_MIXER1>;
>> >> >> > + clock-names = "bus",
>> >> >> > + "mod";
>> >> >> > + resets = <&display_clocks RST_WB>;
>> >> >> > +
>> >> >> > + ports {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > +
>> >> >> > + mixer1_out: port@1 {
>> >> >> > + reg = <1>;
>> >> >> > + mixer1_out_tcon_top:
>> >> >> > endpoint {
>> >> >> > + remote-endpoint =
>> >> >> > <&tcon_top_mixer1_in_mixer1>; +
>> >> >> > };
>> >> >> > + };
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> >
>> >> >> > nmi_intc: interrupt-controller@1c00030 {
>> >> >> >
>> >> >> > compatible = "allwinner,sun7i-a20-sc-nmi";
>> >> >> > interrupt-controller;
>> >> >> >
>> >> >> > @@ -451,6 +518,163 @@
>> >> >> >
>> >> >> > #size-cells = <0>;
>> >> >> >
>> >> >> > };
>> >> >> >
>> >> >> > + tcon_top: tcon-top@1c70000 {
>> >> >> > + compatible = "allwinner,sun8i-r40-tcon-top";
>> >> >> > + reg = <0x01c70000 0x1000>;
>> >> >> > + clocks = <&ccu CLK_BUS_TCON_TOP>,
>> >> >> > + <&ccu CLK_TCON_TV0>,
>> >> >> > + <&ccu CLK_TVE0>,
>> >> >> > + <&ccu CLK_TCON_TV1>,
>> >> >> > + <&ccu CLK_TVE1>,
>> >> >> > + <&ccu CLK_DSI_DPHY>;
>> >> >> > + clock-names = "bus",
>> >> >> > + "tcon-tv0",
>> >> >> > + "tve0",
>> >> >> > + "tcon-tv1",
>> >> >> > + "tve1",
>> >> >> > + "dsi";
>> >> >> > + clock-output-names = "tcon-top-tv0",
>> >> >> > + "tcon-top-tv1",
>> >> >> > + "tcon-top-dsi";
>> >> >> > + resets = <&ccu RST_BUS_TCON_TOP>;
>> >> >> > + #clock-cells = <1>;
>> >> >> > +
>> >> >> > + ports {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > +
>> >> >> > + tcon_top_mixer0_in: port@0 {
>> >> >> > + reg = <0>;
>> >> >> > +
>> >> >> > + tcon_top_mixer0_in_mixer0:
>> >> >> > endpoint { +
>> >> >> > remote-endpoint = <&mixer0_out_tcon_top>; +
>> >> >> >
>> >> >> > };
>> >> >> >
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_mixer0_out: port@1 {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > + reg = <1>;
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer0_out_tcon_lcd0:
>> >> >> > endpoint@0 { + reg =
>> >> >> > <0>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer0_out_tcon_lcd1:
>> >> >> > endpoint@1 { + reg =
>> >> >> > <1>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer0_out_tcon_tv0:
>> >> >> > endpoint@2 { + reg =
>> >> >> > <2>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer0_out_tcon_tv1:
>> >> >> > endpoint@3 { + reg =
>> >> >> > <3>;
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_mixer1_in: port@2 {
>> >> >> > + reg = <2>;
>> >> >> > +
>> >> >> > + tcon_top_mixer1_in_mixer1:
>> >> >> > endpoint { +
>> >> >> > remote-endpoint = <&mixer1_out_tcon_top>; +
>> >> >> >
>> >> >> > };
>> >> >> >
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_mixer1_out: port@3 {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > + reg = <3>;
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer1_out_tcon_lcd0:
>> >> >> > endpoint@0 { + reg =
>> >> >> > <0>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer1_out_tcon_lcd1:
>> >> >> > endpoint@1 { + reg =
>> >> >> > <1>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer1_out_tcon_tv0:
>> >> >> > endpoint@2 { + reg =
>> >> >> > <2>;
>> >> >> > + };
>> >> >> > +
>> >> >> > +
>> >> >> > tcon_top_mixer1_out_tcon_tv1:
>> >> >> > endpoint@3 { + reg =
>> >> >> > <3>;
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_hdmi_in: port@4 {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > + reg = <4>;
>> >> >> > +
>> >> >> > + tcon_top_hdmi_in_tcon_tv0:
>> >> >> > endpoint@0 { + reg =
>> >> >> > <0>;
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_hdmi_in_tcon_tv1:
>> >> >> > endpoint@1 { + reg =
>> >> >> > <1>;
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_top_hdmi_out: port@5 {
>> >> >> > + reg = <5>;
>> >> >> > +
>> >> >> > + tcon_top_hdmi_out_hdmi:
>> >> >> > endpoint {
>> >> >> > + remote-endpoint =
>> >> >> > <&hdmi_in_tcon_top>; + };
>> >> >> > + };
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_tv0: lcd-controller@1c73000 {
>> >> >> > + compatible = "allwinner,sun8i-r40-tcon-tv",
>> >> >> > + "allwinner,sun8i-a83t-tcon-tv";
>> >> >> > + reg = <0x01c73000 0x1000>;
>> >> >> > + interrupts = <GIC_SPI 51
>> >> >> > IRQ_TYPE_LEVEL_HIGH>;
>> >> >> > + clocks = <&ccu CLK_BUS_TCON_TV0>, <&tcon_top
>> >> >> > 0>;
>> >> >> > + clock-names = "ahb", "tcon-ch1";
>> >> >> > + resets = <&ccu RST_BUS_TCON_TV0>;
>> >> >> > + reset-names = "lcd";
>> >> >> > +
>> >> >> > + ports {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > +
>> >> >> > + tcon_tv0_in: port@0 {
>> >> >> > + reg = <0>;
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_tv0_out: port@1 {
>> >> >> > + reg = <1>;
>> >> >> > + };
>> >> >> > + };
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_tv1: lcd-controller@1c74000 {
>> >> >> > + compatible = "allwinner,sun8i-r40-tcon-tv",
>> >> >> > + "allwinner,sun8i-a83t-tcon-tv";
>> >> >> > + reg = <0x01c74000 0x1000>;
>> >> >> > + interrupts = <GIC_SPI 52
>> >> >> > IRQ_TYPE_LEVEL_HIGH>;
>> >> >> > + clocks = <&ccu CLK_BUS_TCON_TV1>, <&tcon_top
>> >> >> > 1>;
>> >> >> > + clock-names = "ahb", "tcon-ch1";
>> >> >> > + resets = <&ccu RST_BUS_TCON_TV1>;
>> >> >> > + reset-names = "lcd";
>> >> >> > +
>> >> >> > + ports {
>> >> >> > + #address-cells = <1>;
>> >> >> > + #size-cells = <0>;
>> >> >> > +
>> >> >> > + tcon_tv1_in: port@0 {
>> >> >> > + reg = <0>;
>> >> >> > + };
>> >> >> > +
>> >> >> > + tcon_tv1_out: port@1 {
>> >> >> > + reg = <1>;
>> >> >>
>> >> >> You are missing the remote-endpoints for all the TCON-TOP <-> TCON
>> >> >> connections. Also, on the driver side, there's no code to handle
>> >> >> dynamically mapping mixers to the TCONs that are being used. In the
>> >> >> past
>> >> >> we
>> >> >> had simple 1:1 mappings. This is no longer the case, and it needs to
>> >> >> be
>> >> >> dealt with.
>> >> >
>> >> > How would TCON TOP driver know how to set muxes? There are no
>> >> > appropriate
>> >> > bingings for muxes, except for V4L2 subsystem, which doesn't really
>> >> > work
>> >> > here.
>> >>
>> >> This will end up being a bunch of custom functions exported from the TCON
>> >> TOP driver to the TCON driver. As for bindings, the stuff you already
>> >> have
>> >> is mostly enough. You do have to specify the endpoint ID vs component
>> >> mapping, so you can find the correct one.
>> >>
>> >> For example, you would specify that the IDs for TCON LCDs be 0 and 1, and
>> >> TCON TVs be 2 and 3. Matching the actual register values is a nice
>> >> convenience. These would be used by the driver as the TCON ID.
>> >
>> > So something that's already done (minus full connections).
>> >
>> >> Also, we might want to consider them as two pairs of two TCONs (LCD +
>> >> TV).
>> >> The CRTC in DRM land is actually the mixer + TCON on our platform. This
>> >> means we only have two CRTCs. So we could have CRTC 0 = mixer 0 + TCON
>> >> LCD
>> >> 0 + TCON TV 0. The "TCON_TVx_OUTSEL" and "DE_PORTx PERH Select" muxes
>> >> would
>> >> be set at run time in the mode set function, selecting either LCD or TV
>> >> based on what encoder is attached.
>> >
>> > That proposal would still limit some combinations. For example, that would
>> > put DSI always on mixer1, since it can be connected to only LCD TCON 1.
>> > It can also cause undesired combination in laptop solutions. Consider
>> > that PCB designer connected LCD1 pins to panel for some reason. Panel is
>> > definetly main screen and should definetly be connected to mixer0, but in
>> > that case, it would be to mixer1.
>>
>> There's no reason we can't have dynamically assigned mixer+tcon pairs, but
>> that would mean implementing additional scheduling code that we don't have
>> yet. The current sunxi-drm and CRTC code assume static mappings.
>
> I just remembered that your proposed solution while should work on R40,
> doesn't really fit in H6 case (which is the main reason for this series).
>
> There are only LCD TCON 0 and TV TCON 0. Their HW IDs are the same as on R40
> (1 and 3). Additionaly, there is also HDMI mux as on R40, although there is
> only one TV TCON, which can be used only for HDMI (TV output is connected to
> LCD TCON through AC200 bridge, put on same die or glued in same package, not
> really sure and doesn't really matter).
OK. So even there's only one TV TCON, you still have to program all muxes with
the correct value, to prevent a) a bad value left by the bootloader and b) bad
reset default values.
> I would like to write something which is also going to work for H6 out of the
> box. For H6, following combination would work well:
> mixer0 = LCD TCON 1 + TV TCON 0
> mixer1 = LCD TCON 0 + TV TCON 1
>
> Or alternatively, we can leave that to mux callback in TCON which would take
> care for best variant for given SoC.
I think that is the ultimate goal. However, the muxing would be a bit
complicated. For all the past SoCs, we were dealing with muxing the
downstream encoder. Now we will be muxing mixer and tcon connections
as well. You would also have to dynamically switch around the mixer
and/or tcon pointers in sun4i_crtc so existing code can still work.
> Once we concur on that, I'll try to implement something.
>
>>
>> We could do this as a second step if you're up to it.
>
> For my ultimate goal, which is 1st class Kodi experience on mainline, current
> proposal works in any case (mixer1 is enough). I just don't like solutions
> which are not universal.
>
>>
>> > Additionaly, since HDMI would became floating between TV TCON 0 and 1,
>> > whoever would write board DT would need to know this and enable only TV
>> > TCON1 if LCD is desired to be connected to mixer0 (or vice versa).
>>
>> There's no reason not to have all TCONs enabled by default. We would
>> consider the display-engine node controlling whether everything actually
>> gets used.
>> > If we really want universal solution with full connections, addtional
>> > property has to be defined and used for that.
>> >
>> >> This limitation is a software one, and should not bleed over into the
>> >> hardware representation.
>> >>
>> >> > Additionaly, how would HDMI know which TCON belongs to it to
>> >> > appropriately
>> >> > set possible_crtcs?
>> >>
>> >> HDMI is connected to the two TCON TVs through the TCON TOP mux. We handle
>> >> TCON output muxing in the TCON driver, using the set_mux callback in the
>> >> quirks. For R40, this callback would probably call into the TCON TOP
>> >> driver
>> >> asking it to set the mux to some value.
>> >>
>> >> > Currently, my idea is that board DT creates wanted connections. Since
>> >> > there is only one valid connection for each mux, driver knows eactly
>> >> > what
>> >> > to write into mux register. HDMI driver can simply check which TCON
>> >> > connection is valid in HDMI input mux and select it in possible_crtcs.
>> >>
>> >> But that is not how the actual hardware looks like. The device tree
>> >> should
>> >> model the hardware, not a subset of it just because one thinks the
>> >> implementation is difficult or won't be used at all.
>> >>
>> >> Furthermore, I think you have it backwards. possible_crtcs is generated
>> >> based on (in our case) the connections between TCON and HDMI based on the
>> >> device tree graph. So if you have both hooked up, both will show up in
>> >> possible_crtcs, but only one crtc will actually be selected to feed the
>> >> HDMI encoder. If you really need to access the current crtc, the
>> >> drm_encoder struct contains a pointer to it.
>> >>
>> >> In DE 1.0 driver, we leave all the muxing to the TCON driver. See
>> >>
>> >>
>> >> https://elixir.bootlin.com/linux/latest/source/drivers/gpu/drm/sun4i/sun4
>> >> i_
>> >> tcon.c#L537
>> >>
>> >> So in a sense, the HDMI encoder should be and is hooked up to both TCONs,
>> >> but only one is active at any given time.
>> >
>> > So something like that I had in v1 series. I'll implement something
>> > similar. That would also mean we need R40 specific TV TCON compatible.
>> > I'll restore that.
>> >
>> > Just a question on future situation when TVE driver will be implemented.
>> > Suppose that R40 board has TVE0 and HDMI as outputs. This would mean that
>> > TVE0 has possible crtcs set as 0b01 and HDMI 0b11. Will be DRM framework
>> > smart enough that it will put HDMI on second crtc because TVE0 can be
>> > connected to only to first crtc?
>>
>> Yes. If the possible_clones setting for the encoder is not set, the DRM
>> framework will not do mirroring, and will keep each active encoder on a
>> separate crtc.
>>
>> >> > Please also note that mixer0 and mixer1 don't have same capabilities
>> >> > and
>> >> > you generally want mixer0 to be connected to main output. This is in
>> >> > contrast to DE1 SoCs, where both backends and both frontends have same
>> >> > capability.
>> >>
>> >> Yes. But who's to say the two display outputs can't be reversed or
>> >> swapped
>> >> around? With fixed singular connections, you also rule out mirrored
>> >> output.
>> >
>> > I don't think there is standard way to swap around mixers at runtime,
>> > expecially if crtcs are represented as mixer + TCON pair.
>>
>> As I mentioned above, we will have to do this on our own.
>>
>> > I'm not sure what do you mean with mirrored output or at least why
>> > singular
>> > connections would prevent mirroring. HW mirroring needs DRM writeback
>> > support which is currently in RFC phase if I'm not mistaken. Once
>> > implemented in DRM framework and in sun4i-drm, it would be possible only
>> > to mirror mixer0 -> mixer1, because mixer1 doesn't support writeback (at
>> > least on existing SoCs).
>> What I meant was it might be possible to drive two TCONs with one mixer, or
>> two encoders with one TCON. I guess the latter is not possible since each
>> TCON only has one channel. Not sure about the former.
>
> I think neither of this two variants are possible, because outputs would have
> to have exactly the same resolution and in second case even the same pixel
> clock, which is very unlikely.
That is what mirroring is supposed to mean, right? For instance you could
have VGA on one and HDMI on the other, at the same resolution and dotclock.
As far as I'm concerned, the software doesn't need to have all features
covered. But that doesn't mean the hardware representation can lack
features as well. We already did that once, and now we have some legacy
graph parsing code to handle that.
Regards
ChenYu
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
