Re: [PATCH] drm/panel: Add prepare_prev_first flag to Visionox VTDR6130
From: Abhinav Kumar
Date: Tue Aug 29 2023 - 15:22:30 EST
On 8/29/2023 12:15 PM, Dmitry Baryshkov wrote:
On Tue, 29 Aug 2023 at 22:09, Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx> wrote:
On 8/29/2023 11:51 AM, Dmitry Baryshkov wrote:
On Tue, 29 Aug 2023 at 20:22, Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx> wrote:
On 8/29/2023 9:43 AM, Dmitry Baryshkov wrote:
On Tue, 29 Aug 2023 at 19:37, Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx> wrote:
On 8/29/2023 2:26 AM, Dmitry Baryshkov wrote:
On Tue, 29 Aug 2023 at 12:22, <neil.armstrong@xxxxxxxxxx> wrote:
On 28/08/2023 19:07, Abhinav Kumar wrote:
Hi Neil
Sorry I didnt respond earlier on this thread.
On 8/28/2023 1:49 AM, neil.armstrong@xxxxxxxxxx wrote:
Hi Jessica,
On 25/08/2023 20:37, Jessica Zhang wrote:
On 8/21/2023 3:01 AM, neil.armstrong@xxxxxxxxxx wrote:
Hi Maxime,
On 21/08/2023 10:17, Maxime Ripard wrote:
Hi,
On Fri, Aug 18, 2023 at 10:25:48AM +0200, neil.armstrong@xxxxxxxxxx wrote:
On 17/08/2023 20:35, Dmitry Baryshkov wrote:
On 16/08/2023 10:51, neil.armstrong@xxxxxxxxxx wrote:
Sending HS commands will always work on any controller, it's all
about LP commands. The Samsung panels you listed only send HS
commands so they can use prepare_prev_first and work on any
controllers.
I think there is some misunderstanding there, supported by the
description of the flag.
If I remember correctly, some hosts (sunxi) can not send DCS
commands after enabling video stream and switching to HS mode, see
[1]. Thus, as you know, most of the drivers have all DSI panel setup
commands in drm_panel_funcs::prepare() /
drm_bridge_funcs::pre_enable() callbacks, not paying attention
whether these commands are to be sent in LP or in HS mode.
Previously DSI source drivers could power on the DSI link either in
mode_set() or in pre_enable() callbacks, with mode_set() being the
hack to make panel/bridge drivers to be able to send commands from
their prepare() / pre_enable() callbacks.
With the prev_first flags being introduced, we have established that
DSI link should be enabled in DSI host's pre_enable() callback and
switched to HS mode (be it command or video) in the enable()
callback.
So far so good.
It seems coherent, I would like first to have a state of all DSI host
drivers and make this would actually work first before adding the
prev_first flag to all the required panels.
This is definitely what we should do in an ideal world, but at least for
sunxi there's no easy way for it at the moment. There's no documentation
for it and the driver provided doesn't allow this to happen.
Note that I'm not trying to discourage you or something here, I'm simply
pointing out that this will be something that we will have to take into
account. And it's possible that other drivers are in a similar
situation.
Unfortunately this change is not fully backwards-compatible. This
requires that all DSI panels sending commands from prepare() should
have the prepare_prev_first flag. In some sense, all such patches
might have Fixes: 5ea6b1702781 ("drm/panel: Add prepare_prev_first
flag to drm_panel").
This kind of migration should be done *before* any possible
regression, not the other way round.
If all panels sending commands from prepare() should have the
prepare_prev_first flag, then it should be first, check for
regressions then continue.
<snip>
I understand, but this patch doesn't qualify as a fix for
9e15123eca79 and is too late to be merged in drm-misc-next for
v6.6, and since 9e15123eca79 actually breaks some support it
should be reverted (+ deps) since we are late in the rc cycles.
If we go this way, we can never reapply these patches. There will be
no guarantee that all panel drivers are completely converted. We
already have a story without an observable end -
DRM_BRIDGE_ATTACH_NO_CONNECTOR.
I don't understand this point, who would block re-applying the patches ?
The migration to DRM_BRIDGE_ATTACH_NO_CONNECTOR was done over multiple
Linux version and went smoothly because we reverted regressing patches
and restarted when needed, I don't understand why we can't do this
here aswell.
I'd consider that the DSI driver is correct here and it is about the
panel drivers that require fixes patches. If you care about the
particular Fixes tag, I have provided one several lines above.
Unfortunately it should be done in the other way round, prepare for
migration, then migrate,
I mean if it's a required migration, then it should be done and I'll
support it from both bridge and panel PoV.
So, first this patch has the wrong Fixes tag, and I would like a
better explanation on the commit message in any case. Then I would
like to have an ack from some drm-misc maintainers before applying it
because it fixes a patch that was sent via the msm tree thus per the
drm-misc rules I cannot apply it via the drm-misc-next-fixes tree.
Sorry, it's not clear to me what you'd like our feedback on exactly?
So let me resume the situation:
- pre_enable_prev_first was introduced in [1]
- some panels made use of pre_enable_prev_first
- Visionox VTDR6130 was enabled on SM8550 systems and works on v6.5 kernels and before
- patch [2] was introduced on MSM DRM tree, breaking VTDR6130 on SM8550 systems (and probably other Video mode panels on Qcom platforms)
- this fix was sent late, and is now too late to be merged via drm-misc-next
Hi Neil and Maxime,
I agree with Neil that 9e15123eca79 was the commit that introduced the issue (since it changed the MSM DSI host behavior).
However, I'm not too keen on simply reverting that patch because
1) it's not wrong to have the dsi_power_on in pre_enable. Arguably, it actually makes more sense to power on DSI host in pre_enable than in modeset (since modeset is meant for setting the bridge mode), and
I never objected that, it's the right path to go.
Ack.
2) I think it would be good practice to keep specific bridge chip checks out of the DSI host driver.
We discussed about a plan with Maxime and Dmitry about that, and it would require adding
a proper atomic panel API to handle a "negociation" with the host controller.
May I know what type of negotiation is needed here?
That being said, what do you think about setting the default value of prepare_prev_first to true (possibly in panel_bridge_attach)?
As Dmitry pointed, all panels sending LP commands in pre_enable() should have prepare_prev_first to true.
I wanted to respond to this earlier but didnt get a chance.
From the documentation of this flag, this has nothing to do whether panels are sending the LP commands (commands sent in LP mode) OR HS commands (commands sent in HS mode).
This is more about sending the commands whether the lanes are in LP11 state before sending the ON commands.
195 * The previous controller should be prepared first, before the prepare
196 * for the panel is called. This is largely required for DSI panels
197 * where the DSI host controller should be initialised to LP-11 before
198 * the panel is powered up.
199 */
200 bool prepare_prev_first;
These are conceptually different and thats what I explained Dmitry in our call.
Sending ON commands in LP11 state is a requirement I have seen with many panels and its actually the right expectation as well to send the commands when the lanes are in a well-defined LP11 state.
From the panels which I have seen, the opposite is never true (OR i have never seen it this way).
The parade chip was the only exception and that issue was never root-caused leading us to have bridge specific handling in MSM driver.
In other words, it would be very unlikely that a panel should be broken or shouldn't work when the ON commands are sent when the lanes are in LP11 state.
So I agree with Jessica, that we should set the default value of this flag to true in the framework so that only the bridges/panels which need this to be false do that explicitly. From the examples I pointed out including MTK, even those vendors are powering on their DSI in pre_enable() which means none of these panels will work there too.
It seems to me that most panel drivers send DCS commands during pre_enable, so maybe it would make more sense to power on DSI host before panel enable() by default. Any panel that needs DSI host to be powered on later could then explicitly set the flag to false in their respective drivers.
A proper migration should be done, yes, but not as a fix on top of v6.5.
I am fine to drop this fix in favor of making the prepare_prev_first as default true but we need an agreement first. From what I can see, parade chip will be the only one which will need this to be set to false and we can make that change.
Let me know if this works as a migration plan.
Yep agreed, let's do this
The panel's prepare_prev_first should be reversed to something like not_prepare_prev_first to make it an exception.
This will break all non-DSI panels, which might depend on the current
bridge calling order.
I started looking at the explicit DSI power up sequencing, but it will
take a few more days to mature.
May I know why this would break all non-DSI panels?
Existing panel drivers might be depending on the init order. Do we
know for sure that none of DPI panels will be broken if there is a
video stream ongoing during the reset procedure?
Or the panel-edp, which I'm pretty sure will require not_prepare_prev_first.
Can you please explain why would video stream be ON in pre_enable()?
Even though we call msm_dsi_host_enable() in the DSI's pre_enable(), the
timing engine is not enabled until the encoder's enable and the first
commit after that so video stream wont be sent till then.
You are describing the MSM DSI case. I was pointing to the fact that
parent's pre_enable if called too early might conflict with the next
bridge driver in the _generic_ case. E.g. eDP or DPI. Even if is not a
full-featured video stream, this state might confuse the panel. So we
can not blindly switch the order of pre_enable callbacks for the
bridge-panel pair.
Even if the end connector was a eDP or DPI, the input to the bridge was
DSI so I think its unlikely that video stream was on before encoder's
enable but I cannot speak for all these interfaces/vendors.
So, to accommodate both worlds, does this work?
diff --git a/drivers/gpu/drm/bridge/panel.c b/drivers/gpu/drm/bridge/panel.c
index 9316384b4474..2b38388d4e56 100644
--- a/drivers/gpu/drm/bridge/panel.c
+++ b/drivers/gpu/drm/bridge/panel.c
@@ -416,7 +416,10 @@ struct drm_bridge
*devm_drm_panel_bridge_add_typed(struct device *dev,
return bridge;
}
- bridge->pre_enable_prev_first = panel->prepare_prev_first;
+ if (connector_type == DRM_MODE_CONNECTOR_DSI)
+ bridge->pre_enable_prev_first = true;
+ else
+ bridge->pre_enable_prev_first = panel->prepare_prev_first;
looks like a hack.
Nope its not, DSI spec has clear mentions about LP11 state to be used
during initialization.
"After power-up, the host processor shall observe an initialization
period, TINIT, during which it shall drive a
sustained TX-Stop state (LP-11) on all Lanes of the Link. See [MIPI04]
for descriptions of TINIT and the TX-Stop state.
Peripherals shall power up in the RX-Stop state and monitor the Link to
determine if the host processor has
asserted a TX-Stop state for at least the TINIT period. The peripheral
shall ignore all Link states prior to
detection of a TINIT event. The peripheral shall be ready to accept bus
transactions immediately following the end of the TINIT period."
I am only extending this statement by mandating that the DSI PHY /
controller be powered up so that its able to drive the lanes to LP11 state.
Now, pls explain why this is a hack
*ptr = bridge;
devres_add(dev, ptr);
drm_atomic_bridge_chain_pre_enable() is called before the encoder's enable.
drm_atomic_bridge_chain_enable() is the one which is called after the
encoder's enable().
Like I said, we dont know the full details of the parade issue but I do
not see any reason why powering on a bridge chip with the DSI lanes
being in proper LP11 state should cause an issue. Its a well defined and
documented state in DSI spec.
On the contrary, trying to turn on a bridge chip before powering on a
controller could have more issues as we do not know what state the lanes
are in when the MIPI devices (panel or bridge) are powered up.
This sets the expectation and handshake straight.