Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received

From: Kuogee Hsieh
Date: Thu Jan 13 2022 - 12:51:47 EST

Next message: G, GurucharanX: "RE: [Intel-wired-lan] [PATCH] ice: Remove useless DMA-32 fallback configuration"
Previous message: G, GurucharanX: "RE: [Intel-wired-lan] [PATCH] i40e: Remove useless DMA-32 fallback configuration"
In reply to: Stephen Boyd: "Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received"
Next in thread: Stephen Boyd: "Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On 1/12/2022 8:13 PM, Stephen Boyd wrote:

Quoting Kuogee Hsieh (2022-01-12 14:17:54)

On 1/12/2022 12:00 PM, Stephen Boyd wrote:

Quoting Kuogee Hsieh (2022-01-11 10:43:23)

Current DP drivers have regulators, clocks, irq and phy are grouped
together within a function and executed not in a symmetric manner.
This increase difficulty of code maintenance and limited code scalability.
This patch divides the driver life cycle of operation into four states,
resume (including booting up), dongle plugin, dongle unplugged and suspend.
Regulators, core clocks and irq are grouped together and enabled at resume
(or booting up) so that the DP controller is armed and ready to receive HPD
plugin interrupts. HPD plugin interrupt is generated when a dongle plugs
into DUT (device under test). Once HPD plugin interrupt is received, DP
controller will initialize phy so that dpcd read/write will function and
following link training can be proceeded successfully. DP phy will be
disabled after main link is teared down at end of unplugged HPD interrupt
handle triggered by dongle unplugged out of DUT. Finally regulators, code
clocks and irq are disabled at corresponding suspension.

0) Please note that dongles are behavior differently.

1) Apple dongle will generate plug-in interrupt only if no hdmi monitor
atatched to dongle. it will generate irq-hpd interrupt once hdmi monitor
connect to dongle later.

2) Apple dongle will generate plugged-in interrupt followed by irq-hpd
interrupt if dongle has hdmi monitor attached when connects to DUT.

3) other dongle will not generate plug-in interrupt unless dongle has
hdmi monitor attached when connects to DUT. It only generate plug-in
interrupt only and no irq-hpd interrupt generated on this case.

Ok. The test scenarios can be reworded in terms of plugin irq and
irq-hpd if that makes it easier.

4) Note: phy_initialized only associated with plugged-in interrupt

5) irq-hpd interrupt must happen after plugged-in interrupt and before
unplugged interrupt

More precisely it's that plugged-in interrupt must be handled before
irq-hpd but plugged-in and irq-hpd can both be pending at the device
concurrently unless they're masked and unmasked in some particular
order. I thought the driver ensures that only irq-hpd is unmasked once
the plugged in irq is handled. Can you confirm? Similarly, unplugged irq
is unmasked after plugged in irq is handled, but irq-hpd and unplugged
can both be pending if the irq handler is delayed?

we went through this before,

1) plugin_handler ==> do plug staffs + unmask irq_hpd at end of handler

2) unplug_handler ==> mask both plugin and irq_hpd + do unplug staffs + unmask plugin irq at end of handler

3) priority order of handling irq when multiple irqs pending ==> plugin -> irq_hpd -> unplug

I will fill up below question with Apple dongle case with the order of
event happen timing.

I'll write out the various scenarios

#1
dongle plugged in with HDMI cable attached
driver probe

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) generate plugged-in interrupt triggers handler
4) dp_display_phy_init() ==> phy_initialized = true;

#2
dongle unplugged
driver probe

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;

#3
dongle plugged in without HDMI cable attached
driver probe

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) generate plug-in interrupt triggers handler
4) dp_display_phy_init() ==> phy_initialized = true;

Note: same as case #1

#4
driver probe
dongle plugged in without HDMI cable attached

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plugged in
4) generate plug-in interrupt triggers handler
5) dp_display_phy_init() ==> phy_initialized = true;

#5
driver probe
dongle plugged in with HDMI cable attached

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plugged in
4) generate plug-in interrupt trigger handler
5) dp_display_phy_init() ==> phy_initialized = true;

Note: same as case #4

#6
driver probe
dongle plugged in
suspend
resume

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plug in
4) generate plug-in interrupt triggers handler
5) dp_display_phy_init() ==> phy_initialized = true;
6) suspend
7) dp_display_host_deinit() ==> core_initialized = false;
8) dp_display_host_phy_exit() ==> phy_initialize = false;
9) resume
10) dp_display_host_init() ==> core_initialized = true;
11) generate plug-in interrupt
12) dp_display_phy_init() ==> phy_initialize = true;

#7
driver probe
dongle plugged in
suspend
dongle unplugged
resume

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plugged in
4) generate plug-in interrupt triggers handler
5) dp_display_phy_init() ==> phy_initialized = true;
6) suspend
7) dp_display_host_deinit() ==> core_initialized = false;
8) dp_display_host_phy_exit() ==> phy_initialize = false;

Why is the order of operations swapped? During probe core_initialized
is done first and then phy_initialized but then on suspend
core_initialized is done first again before phy_initialized. That's
asymmetric.

ok, will fix this

9) dongle unplugged
10) resume
11) dp_display_host_init() ==> core_initialized = true;

#8
driver probe
dongle plugged in without HDMI cable attached
suspend
resume

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plug in
4) generate plug-in interrupt triggers handler
5) dp_display_phy_init() ==> phy_initialized = true;
6) suspend
7) dp_display_host_deinit() ==> core_initialized = false;
8) dp_display_host_phy_exit() ==> phy_initialize = false;
9) resume
10) dp_display_host_init() ==> core_initialized = true;
11) generate plug-in interrupt
12) dp_display_phy_init() ==> phy_initialize = true;

NOTE: same case #6

#9
driver probe
dongle plugged in without HDMI cable attached
suspend
HDMI cable attached during suspend
resume

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) dongle plugged in
4) generate plug-in interrupt triggers handler
5) dp_display_phy_init() ==> phy_initialized = true;
6) suspend
7) dp_display_host_deinit() ==> core_initialized = false;
8) dp_display_host_phy_exit() ==> phy_initialize = false;
9) HDMI cable attached
10) resume
11) dp_display_host_init() ==> core_initialized = true;
12) generate plug-in interrupt
13) dp_display_phy_init() ==> phy_initialize = true;

What's the state of the phy and core initialized variable at the end of
each of these scenarios? Please fill out the truth table.

+-----------------+------------------------
| false | true |
+-----------------+------------------------
phy_initialized | | |
+-----------------+------------------------
core_initialized | | #1, |
+-----------------+------------------------

I guess we also need eDP scenarios, but that's probably simpler

#10
eDP panel connected
driver probe

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) generate plug-in interrupt triggers handler

I think this is more like "dp_display_config_hpd() is called by hpd
kthread"?

yes, correct as below (eDP will call phy_init immediately

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_config_hpd()
3) dp_display_host_init() ==> core_initialized = true;
4) dp_display_host_phy_init() ==> phy_initialize = true
5) generate plug-in interrupt triggers handler

4) dp_display_phy_init() ==> phy_initialized = true;

#11
eDP panel disconnected
driver probe

NOTE: eDP panel can not be disconnected

The panel can certainly be disconnected in the sense that the ribbon
cable to the panel is busted or not working properly. That's what this
scenario is for.

#12
eDP panel disconnected
driver probe
suspend
resume

NOTE: assume edp panel connected
1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_host_init() ==> core_initialized = true;
3) generate plug-in interrupt triggers handler
4) dp_display_phy_init() ==> phy_initialized = true;
5) suspend
6) dp_display_host_deinit() ==> core_initialized = false;
7) dp_display_host_phy_exit() ==> phy_initialize = false;
8) resume
9) dp_display_host_init() ==> core_initialized = true;
10) generate plug-in interrupt
11) dp_display_phy_init() ==> phy_initialize = true;

1) driver probe ==> core_initialized = false; phy_initialized = false;
2) dp_display_config_hpd()
3) dp_display_host_init() ==> core_initialized = true;
4) dp_display_host_phy_init() ==> phy_initialize = true
5) generate plug-in interrupt triggers handler
6) dp_display_phy_init() ==> phy_initialized = true;
7) suspend
8) dp_display_host_deinit() ==> core_initialized = false;
9) dp_display_host_phy_exit() ==> phy_initialize = false;
10) resume
11) dp_display_host_init() ==> core_initialized = true;
11) dp_display_phy_init() ==> phy_initialize = true;
12) generate plug-in interrupt triggers handler

Thanks. It really helps to see the various scenarios.

diff --git a/drivers/gpu/drm/msm/dp/dp_display.c b/drivers/gpu/drm/msm/dp/dp_display.c
index 7cc4d21..f6bb4bc 100644
--- a/drivers/gpu/drm/msm/dp/dp_display.c
+++ b/drivers/gpu/drm/msm/dp/dp_display.c
@@ -83,6 +83,7 @@ struct dp_display_private {

/* state variables */
bool core_initialized;
+ bool phy_initialized;
bool hpd_irq_on;
bool audio_supported;

@@ -372,21 +373,46 @@ static int dp_display_process_hpd_high(struct dp_display_private *dp)
return rc;
}

-static void dp_display_host_init(struct dp_display_private *dp, int reset)
+static void dp_display_host_phy_init(struct dp_display_private *dp)
{
- bool flip = false;
+ DRM_DEBUG_DP("core_init=%d phy_init=%d\n",
+ dp->core_initialized, dp->phy_initialized);

+ if (!dp->phy_initialized) {
+ dp_ctrl_phy_init(dp->ctrl);
+ dp->phy_initialized = true;
+ }
+}
+
+static void dp_display_host_phy_exit(struct dp_display_private *dp)
+{
+ DRM_DEBUG_DP("core_init=%d phy_init=%d\n",
+ dp->core_initialized, dp->phy_initialized);
+
+ if (dp->phy_initialized) {
+ dp_ctrl_phy_exit(dp->ctrl);
+ dp->phy_initialized = false;
+ }
+}
+
+static void dp_display_host_init(struct dp_display_private *dp)
+{
DRM_DEBUG_DP("core_initialized=%d\n", dp->core_initialized);
if (dp->core_initialized) {
DRM_DEBUG_DP("DP core already initialized\n");
return;
}

- if (dp->usbpd->orientation == ORIENTATION_CC2)
- flip = true;
+ dp_power_init(dp->power, false);
+ dp_ctrl_reset_irq_ctrl(dp->ctrl, true);
+
+ /*
+ * eDP is the embedded primary display and has its own phy
+ * initialize phy immediately

Question still stands why we can't wait for hpd high from the eDP panel.
Also, I think "has its own phy" means that it's not part of a combo
USB+DP phy? Can you please clarify?

+ */
+ if (dp->dp_display.connector_type == DRM_MODE_CONNECTOR_eDP)
+ dp_display_host_phy_init(dp);

- dp_power_init(dp->power, flip);
- dp_ctrl_host_init(dp->ctrl, flip, reset);
dp_aux_init(dp->aux);
dp->core_initialized = true;
}
@@ -1306,20 +1330,23 @@ static int dp_pm_resume(struct device *dev)
dp->hpd_state = ST_DISCONNECTED;

/* turn on dp ctrl/phy */
- dp_display_host_init(dp, true);
+ dp_display_host_init(dp);

dp_catalog_ctrl_hpd_config(dp->catalog);

- /*
- * set sink to normal operation mode -- D0
- * before dpcd read
- */
- dp_link_psm_config(dp->link, &dp->panel->link_info, false);

if (dp_catalog_link_is_connected(dp->catalog)) {
+ /*
+ * set sink to normal operation mode -- D0
+ * before dpcd read
+ */
+ dp_display_host_phy_init(dp);
+ dp_link_psm_config(dp->link, &dp->panel->link_info, false);
sink_count = drm_dp_read_sink_count(dp->aux);
if (sink_count < 0)
sink_count = 0;
+
+ dp_display_host_phy_exit(dp);

Why is the phy exited on resume when the link is still connected? Is
this supposed to be done only when the sink_count is 0? And how does
this interact with eDP where the phy is initialized by the call to
dp_display_host_init() earlier in this function.

}

dp->link->sink_count = sink_count;

Any response to the above two comments?

Next message: G, GurucharanX: "RE: [Intel-wired-lan] [PATCH] ice: Remove useless DMA-32 fallback configuration"
Previous message: G, GurucharanX: "RE: [Intel-wired-lan] [PATCH] i40e: Remove useless DMA-32 fallback configuration"
In reply to: Stephen Boyd: "Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received"
Next in thread: Stephen Boyd: "Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]