Skip to contentSkip to conversation

drm/vc4: kms: Assign a FIFO to enabled CRTCs instead of active

PATCH In review Maxime Ripard 3 participants 8 messages Sep 18, 2020 – Oct 1, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 drm/vc4: kms: Assign a FIFO to enabled CRTCs instead of active

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")
---
 drivers/gpu/drm/vc4/vc4_kms.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/vc4/vc4_kms.c +1 -1
index af3ee3dcdab6..01fa60844695 100644
@@ -643,7 +643,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 		struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
 		unsigned int matching_channels;
 
-		if (!crtc_state->active)
+		if (!crtc_state->enable)
 			continue;
 
 		/*
Signature · 7 lines
2.26.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#1 lore raw

PATCH 2/2 drm/vc4: crtc: Keep the previously assigned HVS FIFO

The HVS FIFOs are currently assigned each time we have an atomic_check for all the enabled CRTCs.

However, if we are running multiple outputs in parallel and we happen to disable the first (by index) CRTC, we end up changing the assigned FIFO of the second CRTC without disabling and reenabling the pixelvalve which ends up in a stall and eventually a VBLANK timeout.

In order to fix this, we can create a special value for our assigned channel to mark it as disabled, and if our CRTC already had an assigned channel in its previous state, we keep on using it.

Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")
---
 drivers/gpu/drm/vc4/vc4_crtc.c | 13 ++++++++++---
 drivers/gpu/drm/vc4/vc4_drv.h  |  1 +
 drivers/gpu/drm/vc4/vc4_kms.c  | 21 +++++++++++++++------
 3 files changed, 26 insertions(+), 9 deletions(-)
drivers/gpu/drm/vc4/vc4_crtc.c +10 -3
index a393f93390a2..be754120faa8 100644
@@ -852,11 +852,18 @@ void vc4_crtc_destroy_state(struct drm_crtc *crtc,
 
 void vc4_crtc_reset(struct drm_crtc *crtc)
 {
+	struct vc4_crtc_state *vc4_crtc_state;
+
 	if (crtc->state)
 		vc4_crtc_destroy_state(crtc, crtc->state);
-	crtc->state = kzalloc(sizeof(struct vc4_crtc_state), GFP_KERNEL);
-	if (crtc->state)
-		__drm_atomic_helper_crtc_reset(crtc, crtc->state);
+
+	vc4_crtc_state = kzalloc(sizeof(*vc4_crtc_state), GFP_KERNEL);
+	if (!vc4_crtc_state)
+		return;
+
+	vc4_crtc_state->assigned_channel = VC4_HVS_CHANNEL_DISABLED;
+	crtc->state = &vc4_crtc_state->base;
+	__drm_atomic_helper_crtc_reset(crtc, crtc->state);
 }
 
 static const struct drm_crtc_funcs vc4_crtc_funcs = {
drivers/gpu/drm/vc4/vc4_drv.h +1 -0
index 8c8d96b6289f..2b13f2126f13 100644
@@ -531,6 +531,7 @@ struct vc4_crtc_state {
 		unsigned int bottom;
 	} margins;
 };
+#define VC4_HVS_CHANNEL_DISABLED ((unsigned int) -1)
 
 static inline struct vc4_crtc_state *
 to_vc4_crtc_state(struct drm_crtc_state *crtc_state)
drivers/gpu/drm/vc4/vc4_kms.c +15 -6
index 01fa60844695..f452dad50c22 100644
@@ -616,7 +616,7 @@ static int
 vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 {
 	unsigned long unassigned_channels = GENMASK(NUM_CHANNELS - 1, 0);
-	struct drm_crtc_state *crtc_state;
+	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
 	struct drm_crtc *crtc;
 	int i, ret;
 
@@ -629,6 +629,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 	 * modified.
 	 */
 	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		struct drm_crtc_state *crtc_state;
 		if (!crtc->state->enable)
 			continue;
 
@@ -637,15 +638,23 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 			return PTR_ERR(crtc_state);
 	}
 
-	for_each_new_crtc_in_state(state, crtc, crtc_state, i) {
-		struct vc4_crtc_state *vc4_crtc_state =
-			to_vc4_crtc_state(crtc_state);
+	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+		struct vc4_crtc_state *new_vc4_crtc_state =
+			to_vc4_crtc_state(new_crtc_state);
 		struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
 		unsigned int matching_channels;
 
-		if (!crtc_state->enable)
+		if (old_crtc_state->enable && !new_crtc_state->enable)
+			new_vc4_crtc_state->assigned_channel = VC4_HVS_CHANNEL_DISABLED;
+
+		if (!new_crtc_state->enable)
 			continue;
 
+		if (new_vc4_crtc_state->assigned_channel != VC4_HVS_CHANNEL_DISABLED) {
+			unassigned_channels &= ~BIT(new_vc4_crtc_state->assigned_channel);
+			continue;
+		}
+
 		/*
 		 * The problem we have to solve here is that we have
 		 * up to 7 encoders, connected to up to 6 CRTCs.
@@ -674,7 +683,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 		if (matching_channels) {
 			unsigned int channel = ffs(matching_channels) - 1;
 
-			vc4_crtc_state->assigned_channel = channel;
+			new_vc4_crtc_state->assigned_channel = channel;
 			unassigned_channels &= ~BIT(channel);
 		} else {
 			return -EINVAL;
Signature · 7 lines
2.26.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
maintainer Tested-by on v1 2/2
#2 lore raw

Re: [PATCH 2/2] drm/vc4: crtc: Keep the previously assigned HVS FIFO

Hi Maxime

Thanks for the patch - it makes mode switching reliable for me! :-)

Quoted 9 lines · Maxime Ripard

The HVS FIFOs are currently assigned each time we have an atomic_check for all the enabled CRTCs.

However, if we are running multiple outputs in parallel and we happen to disable the first (by index) CRTC, we end up changing the assigned FIFO of the second CRTC without disabling and reenabling the pixelvalve which ends up in a stall and eventually a VBLANK timeout.

Quoted 6 lines

In order to fix this, we can create a special value for our assigned channel to mark it as disabled, and if our CRTC already had an assigned channel in its previous state, we keep on using it.

Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")

Review comments below though.

Quoted 15 lines
---
 drivers/gpu/drm/vc4/vc4_crtc.c | 13 ++++++++++---
 drivers/gpu/drm/vc4/vc4_drv.h  |  1 +
 drivers/gpu/drm/vc4/vc4_kms.c  | 21 +++++++++++++++------
 3 files changed, 26 insertions(+), 9 deletions(-)
drivers/gpu/drm/vc4/vc4_crtc.c +1 -0
index a393f93390a2..be754120faa8 100644
@@ -852,11 +852,18 @@ void vc4_crtc_destroy_state(struct drm_crtc *crtc,
 
 void vc4_crtc_reset(struct drm_crtc *crtc)
 {
+       struct vc4_crtc_state *vc4_crtc_state;

drivers/gpu/drm/vc4/vc4_crtc.c void vc4_crtc_destroy_state(struct drm_crtc *crtc, · line 862

Quoted 10 lines
drivers/gpu/drm/vc4/vc4_crtc.c +5 -3
+
        if (crtc->state)
                vc4_crtc_destroy_state(crtc, crtc->state);
-       crtc->state = kzalloc(sizeof(struct vc4_crtc_state), GFP_KERNEL);
-       if (crtc->state)
-               __drm_atomic_helper_crtc_reset(crtc, crtc->state);
+
+       vc4_crtc_state = kzalloc(sizeof(*vc4_crtc_state), GFP_KERNEL);
+       if (!vc4_crtc_state)
+               return;

This error path has me worried, but I may have missed it.

If crtc->state was set then it's been freed via vc4_crtc_destroy_state. However I don't see anything that has reset crtc->state to NULL.
If the new alloc fails then I believe we exit with crtc->state still set to the old freed pointer.

The old code directly set crtc->state, so it got reset to NULL by the failure.

Quoted 12 lines
drivers/gpu/drm/vc4/vc4_crtc.c +4 -0
+
+       vc4_crtc_state->assigned_channel = VC4_HVS_CHANNEL_DISABLED;
+       crtc->state = &vc4_crtc_state->base;
+       __drm_atomic_helper_crtc_reset(crtc, crtc->state);
 }
 
 static const struct drm_crtc_funcs vc4_crtc_funcs = {
drivers/gpu/drm/vc4/vc4_drv.h +0 -0
index 8c8d96b6289f..2b13f2126f13 100644

drivers/gpu/drm/vc4/vc4_drv.h struct vc4_crtc_state { · line 534

Quoted 4 lines
drivers/gpu/drm/vc4/vc4_drv.h +1 -0
@@ -531,6 +531,7 @@ struct vc4_crtc_state {
                unsigned int bottom;
        } margins;
 };
+#define VC4_HVS_CHANNEL_DISABLED ((unsigned int) -1)

Checkpatch whinge on that
CHECK: No space is necessary after a cast #55: FILE: drivers/gpu/drm/vc4/vc4_drv.h:539:
+#define VC4_HVS_CHANNEL_DISABLED ((unsigned int) -1)

CHECK: Please use a blank line after function/struct/union/enum declarations #55: FILE: drivers/gpu/drm/vc4/vc4_drv.h:539:
};
+#define VC4_HVS_CHANNEL_DISABLED ((unsigned int) -1)

Quoted 16 lines
drivers/gpu/drm/vc4/vc4_drv.h +0 -0
 static inline struct vc4_crtc_state *
 to_vc4_crtc_state(struct drm_crtc_state *crtc_state)
drivers/gpu/drm/vc4/vc4_kms.c +1 -1
index 01fa60844695..f452dad50c22 100644
@@ -616,7 +616,7 @@ static int
 vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
 {
        unsigned long unassigned_channels = GENMASK(NUM_CHANNELS - 1, 0);
-       struct drm_crtc_state *crtc_state;
+       struct drm_crtc_state *old_crtc_state, *new_crtc_state;
        struct drm_crtc *crtc;
        int i, ret;
 

drivers/gpu/drm/vc4/vc4_kms.c vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state) · line 632

Quoted 4 lines
drivers/gpu/drm/vc4/vc4_kms.c +1 -0
@@ -629,6 +629,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
         * modified.
         */
        list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+               struct drm_crtc_state *crtc_state;

Blank line between variables and code, or not in this subsystem? Checkpatch hasn't complained to me here.

Quoted 41 lines
drivers/gpu/drm/vc4/vc4_kms.c +13 -6
                if (!crtc->state->enable)
                        continue;
 
@@ -637,15 +638,23 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
                        return PTR_ERR(crtc_state);
        }
 
-       for_each_new_crtc_in_state(state, crtc, crtc_state, i) {
-               struct vc4_crtc_state *vc4_crtc_state =
-                       to_vc4_crtc_state(crtc_state);
+       for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) {
+               struct vc4_crtc_state *new_vc4_crtc_state =
+                       to_vc4_crtc_state(new_crtc_state);
                struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
                unsigned int matching_channels;
 
-               if (!crtc_state->enable)
+               if (old_crtc_state->enable && !new_crtc_state->enable)
+                       new_vc4_crtc_state->assigned_channel = VC4_HVS_CHANNEL_DISABLED;
+
+               if (!new_crtc_state->enable)
                        continue;
 
+               if (new_vc4_crtc_state->assigned_channel != VC4_HVS_CHANNEL_DISABLED) {
+                       unassigned_channels &= ~BIT(new_vc4_crtc_state->assigned_channel);
+                       continue;
+               }
+
                /*
                 * The problem we have to solve here is that we have
                 * up to 7 encoders, connected to up to 6 CRTCs.
@@ -674,7 +683,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
                if (matching_channels) {
                        unsigned int channel = ffs(matching_channels) - 1;
 
-                       vc4_crtc_state->assigned_channel = channel;
+                       new_vc4_crtc_state->assigned_channel = channel;
                        unassigned_channels &= ~BIT(channel);
                } else {
                        return -EINVAL;
--
Quoted 2 lines

2.26.2

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

maintainer Tested-by Reviewed-by on v1 1/2
#3 lore raw
In reply to Maxime Ripard

Hi Maxime

Thanks for the patch.

Quoted 10 lines · Maxime Ripard

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")
Quoted 18 lines
---
 drivers/gpu/drm/vc4/vc4_kms.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/vc4/vc4_kms.c +1 -2
index af3ee3dcdab6..01fa60844695 100644
@@ -643,7 +643,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
                struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
                unsigned int matching_channels;
 
-               if (!crtc_state->active)
+               if (!crtc_state->enable)
                        continue;
 
                /*
--
Quoted 2 lines

2.26.2

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

author Applied
#4 lore raw
Quoted 13 lines · Dave Stevenson

Hi Maxime

Thanks for the patch.

Quoted 8 lines · Maxime Ripard

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Quoted 5 lines
Quoted 2 lines
Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")

Applied, thanks!
Maxime

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#5 lore raw
In reply to Maxime Ripard

Hi Maxime,

Quoted 6 lines · Maxime Ripard

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Thanks for the patch.

There is a problem when doing page flip.
After connecting 2 HDMIs without hotplug and booting.(Same thing when using hotplug.)

When executing page flip for each of HDMI 0 and 1 in modetest operation does not work normally. (crtc irq does not occur, so timeout occurs.)
Sometimes both hdmi 0 or 1 work or not.

LOGs:
root:~> modetest -Mvc4 -s 32:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 32, crtc 64 failed to set gamma: Invalid argument
freq: 60.24Hz
freq: 60.00Hz

root:~> modetest -Mvc4 -s 38:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 38, crtc 69 failed to set gamma: Invalid argument
select timed out or error (ret 0)
select timed out or error (ret 0)

Could you please check it?

Best regards,
Hoegeun

Quoted 12 lines
Fixes 87ebcd42fb7b ("drm/vc4: crtc: Assign output to channel automatically")
---
  drivers/gpu/drm/vc4/vc4_kms.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/vc4/vc4_kms.c +0 -0
index af3ee3dcdab6..01fa60844695 100644

drivers/gpu/drm/vc4/vc4_kms.c vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state) · line 649

Quoted 8 lines
drivers/gpu/drm/vc4/vc4_kms.c +1 -1
@@ -643,7 +643,7 @@ vc4_atomic_check(struct drm_device *dev, struct drm_atomic_state *state)
  		struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
  		unsigned int matching_channels;
 
-		if (!crtc_state->active)
+		if (!crtc_state->enable)
  			continue;
 
  		/*

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#6 lore raw
Quoted 28 lines · Hoegeun Kwon

Hi Maxime,

Quoted 6 lines · Maxime Ripard

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Thanks for the patch.

There is a problem when doing page flip.
After connecting 2 HDMIs without hotplug and booting.(Same thing when using hotplug.)

When executing page flip for each of HDMI 0 and 1 in modetest operation does not work normally. (crtc irq does not occur, so timeout occurs.)
Sometimes both hdmi 0 or 1 work or not.

LOGs:
root:~> modetest -Mvc4 -s 32:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 32, crtc 64 failed to set gamma: Invalid argument
freq: 60.24Hz
freq: 60.00Hz

Quoted 7 lines

root:~> modetest -Mvc4 -s 38:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 38, crtc 69 failed to set gamma: Invalid argument
select timed out or error (ret 0)
select timed out or error (ret 0)

Could you please check it?

I'll look into it, thanks :)

Maxime

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#7 lore raw
In reply to Hoegeun Kwon
Quoted 28 lines · Hoegeun Kwon

Hi Maxime,

Quoted 6 lines · Maxime Ripard

The HVS has three FIFOs that can be assigned to a number of PixelValves through a mux.

However, changing that FIFO requires that we disable and then enable the pixelvalve, so we want to assign FIFOs to all the enabled CRTCs, and not just the active ones.

Thanks for the patch.

There is a problem when doing page flip.
After connecting 2 HDMIs without hotplug and booting.(Same thing when using hotplug.)

When executing page flip for each of HDMI 0 and 1 in modetest operation does not work normally. (crtc irq does not occur, so timeout occurs.)
Sometimes both hdmi 0 or 1 work or not.

LOGs:
root:~> modetest -Mvc4 -s 32:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 32, crtc 64 failed to set gamma: Invalid argument
freq: 60.24Hz
freq: 60.00Hz

Quoted 7 lines

root:~> modetest -Mvc4 -s 38:1280x720 -v
setting mode 1280x720-60Hz@XR24 on connectors 38, crtc 69 failed to set gamma: Invalid argument
select timed out or error (ret 0)
select timed out or error (ret 0)

Could you please check it?

So I can reproduce that bug, but I've not been able to fix it yet.

The issue seems to happen 100% of the time when you start first with the second connector, and then the first. It creates yet another muxing corner case, which is that when the first modeset runs, there's only one enabled CRTC (with the 69 here) that gets assigned the channel 0 (since it's the only one and it can run from that channel). However, when modeset exits, it doesn't disable that CRTC for some reason.

Then you enable a second CRTC (64) with the second modeset command, but it has a lower index so it gets evaluated first and gets assigned the channel 0 as well since we haven't removed the CRTC 69 from the pool yet.

I've fixed that up by first removing the channels in current use, and then allocating them in two separate passes, but it doesn't address the problem entirely, so I'll keep looking.

Maxime

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel