drm/edid: Add helper to detect whether EDID changed

This adds a common drm helper to detect whether the EDID changed from the last known cached one. This is useful help detect that a monitor was changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during suspend), no hotplug event will be triggered so the change will not be caught at resume time. Detecting that the EDID changed allows detecting it.

Signed-off-by Paul Kocialkowski <[email protected]>

---
 drivers/gpu/drm/drm_edid.c | 45 +++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  3 +++
 2 files changed, 48 insertions(+)

drivers/gpu/drm/drm_edid.c +45 -0

index 6bb6337be920..f6ce8bc2907a 100644
@@ -5036,3 +5036,48 @@ static void drm_get_displayid(struct drm_connector *connector,
 	}
 	return;
 }
+
+/**
+ * drm_check_edid_changed - Check whether the EDID changed since the last update
+ * @connector: connector we're probing
+ * @adapter: I2C adapter to use for DDC
+ *
+ * Check whether the EDID changed since the last time it was updated in the
+ * drm blob cache.
+ *
+ * Return: A boolean indicating whether a change happened or not.
+ */
+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter)
+{
+	struct drm_property_blob *edid_blob;
+	struct edid *edid_stored;
+	struct edid *edid_read;
+	int ret = 0;
+
+	if (!connector->edid_blob_ptr)
+		return false;
+
+	edid_blob = drm_property_blob_get(connector->edid_blob_ptr);
+	if (!edid_blob)
+		return false;
+
+	if (!edid_blob->data || edid_blob->length != sizeof(struct edid))
+		goto out;
+
+	edid_stored = (struct edid *) edid_blob->data;
+
+	edid_read = drm_get_edid(connector, adapter);
+	if (!edid_read)
+		goto out;
+
+	ret = memcmp(edid_stored, edid_read, sizeof(struct edid));
+
+	kfree(edid_read);
+
+out:
+	drm_property_blob_put(edid_blob);
+
+	return ret != 0;
+}
+EXPORT_SYMBOL_GPL(drm_check_edid_changed);

include/drm/drm_edid.h +3 -0

index 1e1908a6b1d6..593a97b269c3 100644
@@ -485,4 +485,7 @@ void drm_edid_get_monitor_name(struct edid *edid, char *name,
 struct drm_display_mode *drm_mode_find_dmt(struct drm_device *dev,
 					   int hsize, int vsize, int fresh,
 					   bool rb);
+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter);
+
 #endif /* __DRM_EDID_H__ */

This adds a common drm helper to detect whether the EDID changed from the last known cached one. This is useful help detect that a monitor was changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during suspend), no hotplug event will be triggered so the change will not be caught at resume time. Detecting that the EDID changed allows detecting it.

Signed-off-by Paul Kocialkowski <[email protected]>

---
 drivers/gpu/drm/drm_edid.c | 45 +++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  3 +++
 2 files changed, 48 insertions(+)

drivers/gpu/drm/drm_edid.c +45 -0

index 6bb6337be920..f6ce8bc2907a 100644
@@ -5036,3 +5036,48 @@ static void drm_get_displayid(struct drm_connector *connector,
 	}
 	return;
 }
+
+/**
+ * drm_check_edid_changed - Check whether the EDID changed since the last update
+ * @connector: connector we're probing
+ * @adapter: I2C adapter to use for DDC
+ *
+ * Check whether the EDID changed since the last time it was updated in the
+ * drm blob cache.
+ *
+ * Return: A boolean indicating whether a change happened or not.
+ */
+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter)
+{
+	struct drm_property_blob *edid_blob;
+	struct edid *edid_stored;
+	struct edid *edid_read;
+	int ret = 0;
+
+	if (!connector->edid_blob_ptr)
+		return false;
+
+	edid_blob = drm_property_blob_get(connector->edid_blob_ptr);
+	if (!edid_blob)
+		return false;
+
+	if (!edid_blob->data || edid_blob->length != sizeof(struct edid))
+		goto out;
+
+	edid_stored = (struct edid *) edid_blob->data;
+
+	edid_read = drm_get_edid(connector, adapter);
+	if (!edid_read)
+		goto out;
+
+	ret = memcmp(edid_stored, edid_read, sizeof(struct edid));
+
+	kfree(edid_read);
+
+out:
+	drm_property_blob_put(edid_blob);
+
+	return ret != 0;
+}
+EXPORT_SYMBOL_GPL(drm_check_edid_changed);

include/drm/drm_edid.h +3 -0

index 1e1908a6b1d6..593a97b269c3 100644
@@ -485,4 +485,7 @@ void drm_edid_get_monitor_name(struct edid *edid, char *name,
 struct drm_display_mode *drm_mode_find_dmt(struct drm_device *dev,
 					   int hsize, int vsize, int fresh,
 					   bool rb);
+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter);
+
 #endif /* __DRM_EDID_H__ */

Signature · 6 lines

2.13.2

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

Quoted 6 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed from
the last known cached one. This is useful help detect that a monitor was
changed during a suspend/resume cycle.

Quoted 7 lines

When that happens (a monitor is replaced by another one during suspend),
no hotplug event will be triggered so the change will not be caught at
resume time. Detecting that the EDID changed allows detecting it.

Signed-off-by Paul Kocialkowski <[email protected]>

I can't find the older mails I've typed about this, but the plan we've discussed a while back was:
- Add a generational counter to each connector, maybe even expose it to
userspace.

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a new
edid
with the set_edid helper.

- Tada, no changes needed to drivers, and easily extensible to other things than edid!

Quoted 85 lines

---
 drivers/gpu/drm/drm_edid.c | 45
+++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  3 +++
 2 files changed, 48 insertions(+)

drivers/gpu/drm/drm_edid.c +3 -0

index 6bb6337be920..f6ce8bc2907a 100644
@@ -5036,3 +5036,48 @@ static void drm_get_displayid(struct
 drm_connector *connector,
 	}
 	return;
 }
+
+/**
+ * drm_check_edid_changed - Check whether the EDID changed since

the last update
+ * @connector: connector we're probing
+ * @adapter: I2C adapter to use for DDC
+ *
+ * Check whether the EDID changed since the last time it was updated in the
+ * drm blob cache.
+ *
+ * Return: A boolean indicating whether a change happened or not.
+ */

drivers/gpu/drm/drm_edid.c +12 -0

+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter)
+{
+	struct drm_property_blob *edid_blob;
+	struct edid *edid_stored;
+	struct edid *edid_read;
+	int ret = 0;
+
+	if (!connector->edid_blob_ptr)
+		return false;
+
+	edid_blob = drm_property_blob_get(connector-

Quoted 1 line

edid_blob_ptr);

drivers/gpu/drm/drm_edid.c +4 -0

+	if (!edid_blob)
+		return false;
+
+	if (!edid_blob->data || edid_blob->length != sizeof(struct

edid))

drivers/gpu/drm/drm_edid.c +18 -0

+		goto out;
+
+	edid_stored = (struct edid *) edid_blob->data;
+
+	edid_read = drm_get_edid(connector, adapter);
+	if (!edid_read)
+		goto out;
+
+	ret = memcmp(edid_stored, edid_read, sizeof(struct edid));
+
+	kfree(edid_read);
+
+out:
+	drm_property_blob_put(edid_blob);
+
+	return ret != 0;
+}
+EXPORT_SYMBOL_GPL(drm_check_edid_changed);

include/drm/drm_edid.h +3 -0

index 1e1908a6b1d6..593a97b269c3 100644
@@ -485,4 +485,7 @@ void drm_edid_get_monitor_name(struct edid
 *edid, char *name,
 struct drm_display_mode *drm_mode_find_dmt(struct drm_device *dev,
 					   int hsize, int vsize,
 int fresh,
 					   bool rb);
+bool drm_check_edid_changed(struct drm_connector *connector,
+			    struct i2c_adapter *adapter);
+
 #endif /* __DRM_EDID_H__ */

Signature · 6 lines

2.13.2

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

Quoted 21 lines · Daniel Vetter

Quoted 6 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed from
the last known cached one. This is useful help detect that a monitor was
changed during a suspend/resume cycle.

Quoted 7 lines

When that happens (a monitor is replaced by another one during suspend),
no hotplug event will be triggered so the change will not be caught at
resume time. Detecting that the EDID changed allows detecting it.

Signed-off-by Paul Kocialkowski <[email protected]>

I can't find the older mails I've typed about this, but the plan we've discussed a while back was:
- Add a generational counter to each connector, maybe even expose it to
userspace.

Quoted 7 lines

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a new
edid
with the set_edid helper.

- Tada, no changes needed to drivers, and easily extensible to other things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything has changed when a monitor was replaced by another one, so I don't see how adding a counter in the mix would help.

Could you provide more details about the reasoning? I feel like I'm missing something here.

Cheers,

Paul

Quoted 81 lines

Quoted 81 lines

---
 drivers/gpu/drm/drm_edid.c | 45
+++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  3 +++
 2 files changed, 48 insertions(+)

drivers/gpu/drm/drm_edid.c +3 -0

index 6bb6337be920..f6ce8bc2907a 100644
@@ -5036,3 +5036,48 @@ static void drm_get_displayid(struct
 drm_connector *connector,
    }
    return;
 }
+
+/**
+ * drm_check_edid_changed - Check whether the EDID changed since

the last update
+ * @connector: connector we're probing
+ * @adapter: I2C adapter to use for DDC
+ *
+ * Check whether the EDID changed since the last time it was updated in the
+ * drm blob cache.
+ *
+ * Return: A boolean indicating whether a change happened or not.
+ */

drivers/gpu/drm/drm_edid.c +12 -0

+bool drm_check_edid_changed(struct drm_connector *connector,
+                       struct i2c_adapter *adapter)
+{
+   struct drm_property_blob *edid_blob;
+   struct edid *edid_stored;
+   struct edid *edid_read;
+   int ret = 0;
+
+   if (!connector->edid_blob_ptr)
+           return false;
+
+   edid_blob = drm_property_blob_get(connector-

Quoted 1 line

edid_blob_ptr);

drivers/gpu/drm/drm_edid.c +4 -0

+   if (!edid_blob)
+           return false;
+
+   if (!edid_blob->data || edid_blob->length != sizeof(struct

edid))

drivers/gpu/drm/drm_edid.c +18 -0

+           goto out;
+
+   edid_stored = (struct edid *) edid_blob->data;
+
+   edid_read = drm_get_edid(connector, adapter);
+   if (!edid_read)
+           goto out;
+
+   ret = memcmp(edid_stored, edid_read, sizeof(struct edid));
+
+   kfree(edid_read);
+
+out:
+   drm_property_blob_put(edid_blob);
+
+   return ret != 0;
+}
+EXPORT_SYMBOL_GPL(drm_check_edid_changed);

include/drm/drm_edid.h +3 -1

index 1e1908a6b1d6..593a97b269c3 100644
@@ -485,4 +485,7 @@ void drm_edid_get_monitor_name(struct edid
 *edid, char *name,
 struct drm_display_mode *drm_mode_find_dmt(struct drm_device *dev,
                                       int hsize, int vsize,
 int fresh,
                                       bool rb);
+bool drm_check_edid_changed(struct drm_connector *connector,
+                       struct i2c_adapter *adapter);
+
 #endif /* __DRM_EDID_H__ */
--

2.13.2

Quoted 6 lines

Quoted 4 lines

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

--
Paul Kocialkowski <[email protected]>
Intel Finland Oy - BIC 0357606-4 - Westendinkatu 7, 02160 Espoo, Finland

Quoted 44 lines · Paul Kocialkowski

Quoted 27 lines · Daniel Vetter

Quoted 15 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed from
the last known cached one. This is useful help detect that a monitor
was
changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during suspend),
no hotplug event will be triggered so the change will not be caught
at
resume time. Detecting that the EDID changed allows detecting it.

Quoted 2 lines

Signed-off-by Paul Kocialkowski <[email protected].

com>

I can't find the older mails I've typed about this, but the plan we've
discussed a while back was:
- Add a generational counter to each connector, maybe even expose it
to
userspace.

Quoted 8 lines

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a new
edid
with the set_edid helper.

- Tada, no changes needed to drivers, and easily extensible to other
things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything has
changed when a monitor was replaced by another one, so I don't see how
adding a counter in the mix would help.

Quoted 2 lines

Could you provide more details about the reasoning? I feel like I'm missing something here.

Your bug doesn't just exist over s/r, it's just much easier to observe in s/r since users can take however long they want to with plugging in a different monitor. But the same issue exists e.g. when we go from hpd to polling because too much noise on the line.

Wrt the suspend/resume issue: What we need to do on resume is do a full reprobe of all outputs, in an async worker. Telling userspace to do this by sending an uevent was the cheapest way, but it'd be better if the kernel could do that asynchronously and inform userspace about the exact changes. And there's more to reprobe than just the edid, and we don't want to re-invent a separate reprobe path just for resume like you start in your patch series. So yeah my plan was missing:

- force a full async reprobe after resume (maybe we could reuse the poll worker for that as a one-shot).

Quoted 97 lines

Quoted 86 lines

Quoted 86 lines

---
 drivers/gpu/drm/drm_edid.c | 45
+++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |  3 +++
 2 files changed, 48 insertions(+)

drivers/gpu/drm/drm_edid.c +3 -0

index 6bb6337be920..f6ce8bc2907a 100644
@@ -5036,3 +5036,48 @@ static void drm_get_displayid(struct
 drm_connector *connector,
    }
    return;
 }
+
+/**
+ * drm_check_edid_changed - Check whether the EDID changed

since
the last update
+ * @connector: connector we're probing
+ * @adapter: I2C adapter to use for DDC
+ *
+ * Check whether the EDID changed since the last time it was updated in the
+ * drm blob cache.
+ *
+ * Return: A boolean indicating whether a change happened or not.
+ */

drivers/gpu/drm/drm_edid.c +12 -0

+bool drm_check_edid_changed(struct drm_connector *connector,
+                       struct i2c_adapter *adapter)
+{
+   struct drm_property_blob *edid_blob;
+   struct edid *edid_stored;
+   struct edid *edid_read;
+   int ret = 0;
+
+   if (!connector->edid_blob_ptr)
+           return false;
+
+   edid_blob = drm_property_blob_get(connector-

Quoted 1 line

edid_blob_ptr);

drivers/gpu/drm/drm_edid.c +4 -0

+   if (!edid_blob)
+           return false;
+
+   if (!edid_blob->data || edid_blob->length != sizeof(struct

edid))

drivers/gpu/drm/drm_edid.c +18 -0

+           goto out;
+
+   edid_stored = (struct edid *) edid_blob->data;
+
+   edid_read = drm_get_edid(connector, adapter);
+   if (!edid_read)
+           goto out;
+
+   ret = memcmp(edid_stored, edid_read, sizeof(struct edid));
+
+   kfree(edid_read);
+
+out:
+   drm_property_blob_put(edid_blob);
+
+   return ret != 0;
+}
+EXPORT_SYMBOL_GPL(drm_check_edid_changed);

include/drm/drm_edid.h +0 -0

index 1e1908a6b1d6..593a97b269c3 100644
@@ -485,4 +485,7 @@ void drm_edid_get_monitor_name(struct edid
 *edid, char *name,
 struct drm_display_mode *drm_mode_find_dmt(struct drm_device
 *dev,
                                       int hsize, int vsize,

int fresh,

                                       bool rb);

include/drm/drm_edid.h +3 -1

+bool drm_check_edid_changed(struct drm_connector *connector,
+                       struct i2c_adapter *adapter);
+
 #endif /* __DRM_EDID_H__ */
--

2.13.2

Quoted 6 lines

Quoted 4 lines

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

--
Paul Kocialkowski <[email protected]> Intel Finland Oy - BIC 0357606-4 - Westendinkatu 7, 02160 Espoo, Finland

Quoted 62 lines · Daniel Vetter

Quoted 44 lines · Paul Kocialkowski

Quoted 27 lines · Daniel Vetter

Quoted 15 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed from
the last known cached one. This is useful help detect that a monitor
was
changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during suspend),
no hotplug event will be triggered so the change will not be caught
at
resume time. Detecting that the EDID changed allows detecting it.

Quoted 2 lines

Signed-off-by Paul Kocialkowski <[email protected].

com>

I can't find the older mails I've typed about this, but the plan we've
discussed a while back was:
- Add a generational counter to each connector, maybe even expose it
to
userspace.

Quoted 8 lines

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a new
edid
with the set_edid helper.

- Tada, no changes needed to drivers, and easily extensible to other
things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything has
changed when a monitor was replaced by another one, so I don't see how
adding a counter in the mix would help.

Quoted 2 lines

Could you provide more details about the reasoning? I feel like I'm missing something here.

Your bug doesn't just exist over s/r, it's just much easier to observe in s/r since users can take however long they want to with plugging in a different monitor. But the same issue exists e.g. when we go from hpd to polling because too much noise on the line.

Wrt the suspend/resume issue: What we need to do on resume is do a full reprobe of all outputs, in an async worker. Telling userspace to do this by sending an uevent was the cheapest way, but it'd be better if the kernel could do that asynchronously and inform userspace about the exact changes. And there's more to reprobe than just the edid, and we don't want to re-invent a separate reprobe path just for resume like you start in your patch series. So yeah my plan was missing:

Quoted 2 lines

- force a full async reprobe after resume (maybe we could reuse the poll worker for that as a one-shot).

First off, I definitely agree we need a way to tell userspace exactly what has happened. I wanted to start a discussion about that in i-g-t patch "Unrelated hotplug uevent masking out actual test result" but it didn't get much traction. For testing purposes, it is unacceptable that userspace only gets notified that "something happened".

Still, as far as I know, userspace is expected to ask for a full reprobe when something has changed, and that is apparently part of the DRM spec, so we can't expect that it could query for an update on "only the things that changed".

However, one way to mitigate this is to make sure that the driver knows what changed and only updates these things when a full reprobe is requested. Is this the approach that you have in mind?

The methodology behind my series follows what is currently done: detect change in whatever way necessary, inform userspace and let it trigger full reprobe. If I'm understanding correctly, what you're suggesting is instead to reprobe what is needed on the kernel side when an associated change occurs instead of having userspace trigger it, and then let userspace aware that something changed and return the "cached" updated status when userspace asks for the subsequent reprobe. Is that correct?

Quoted 95 lines · Paul Kocialkowski

Quoted 80 lines · Daniel Vetter

Quoted 54 lines · Paul Kocialkowski

Quoted 40 lines · Daniel Vetter

Quoted 18 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed
from
the last known cached one. This is useful help detect that a monitor
was
changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during
suspend),
no hotplug event will be triggered so the change will not be caught
at
resume time. Detecting that the EDID changed allows detecting
it.

Quoted 3 lines

Signed-off-by Paul Kocialkowski <[email protected]

tel.
com>

I can't find the older mails I've typed about this, but the plan
we've
discussed a while back was:
- Add a generational counter to each connector, maybe even expose
it
to
userspace.

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a
new
edid
with the set_edid helper.

Quoted 3 lines

- Tada, no changes needed to drivers, and easily extensible to other
things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything
has
changed when a monitor was replaced by another one, so I don't see
how
adding a counter in the mix would help.

Quoted 3 lines

Could you provide more details about the reasoning? I feel like I'm
missing something here.

Your bug doesn't just exist over s/r, it's just much easier to observe
in s/r since users can take however long they want to with plugging in
a different monitor. But the same issue exists e.g. when we go from
hpd to polling because too much noise on the line.

Wrt the suspend/resume issue: What we need to do on resume is do a full reprobe of all outputs, in an async worker. Telling userspace to
do this by sending an uevent was the cheapest way, but it'd be better
if the kernel could do that asynchronously and inform userspace about
the exact changes. And there's more to reprobe than just the edid, and
we don't want to re-invent a separate reprobe path just for resume like you start in your patch series. So yeah my plan was missing:

Quoted 3 lines

- force a full async reprobe after resume (maybe we could reuse the
poll worker for that as a one-shot).

First off, I definitely agree we need a way to tell userspace exactly
what has happened. I wanted to start a discussion about that in i-g- t
patch "Unrelated hotplug uevent masking out actual test result" but it
didn't get much traction. For testing purposes, it is unacceptable that
userspace only gets notified that "something happened".

Quoted 7 lines

Still, as far as I know, userspace is expected to ask for a full reprobe
when something has changed, and that is apparently part of the DRM spec,
so we can't expect that it could query for an update on "only the things
that changed".

We can update that spec in a backwards compatible way. E.g. we can ask for
the current properties without forcing a reprobe (won't even call down into the driver), and userspace could use that to check which connector
has an incremented epoche counter since the last time it sampled things.
Then it can reprobe just that one.

Old userspace wouldn't know about this, and would keep working as-is.

Quoted 5 lines

However, one way to mitigate this is to make sure that the driver knows
what changed and only updates these things when a full reprobe is requested. Is this the approach that you have in mind?

Quoted 13 lines

The methodology behind my series follows what is currently done:
detect
change in whatever way necessary, inform userspace and let it trigger
full reprobe. If I'm understanding correctly, what you're suggesting is
instead to reprobe what is needed on the kernel side when an associated
change occurs instead of having userspace trigger it, and then let userspace aware that something changed and return the "cached" updated
status when userspace asks for the subsequent reprobe. Is that correct?

There's two things: the uapi discussion and the internal implementation,
imo their separate (but somewhat connected) topics.

- For the internal implementation of detecting edid changes I don't like
your approach of rolling a completely new detect path just for
resume.
I think we can very well integrate that into the existing probe code using the approach I've laid out.

- There's more than just edid (e.g. hdcp status, various stuff that's handled in dp aux for DP sinks), and I think a general mechanism for tracking that something changed will be useful for the internal implementation. The other plan would be that we have to wire a bool changed through the entire probe stack, and make sure it's handled correctly everywhere, which is a) a lot more work b) more fragile.
Doing
a connector->status_epoch++ everywhere we detect a change is _much_ simpler.

- For the uapi change: We already support returning the cached stuff, the
only bit that's missing is the epoch counter to let userspace know
where
it might need to do a full reprobe. Or maybe we'll just spec that a
full
reprobe isn't necessary after a hpd event (but that's unlikely to
work
out given how many bugs we'd need to fix first).

Quoted 113 lines · Daniel Vetter

Quoted 95 lines · Paul Kocialkowski

Quoted 80 lines · Daniel Vetter

Quoted 54 lines · Paul Kocialkowski

Quoted 40 lines · Daniel Vetter

Quoted 18 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed
from
the last known cached one. This is useful help detect that a monitor
was
changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during
suspend),
no hotplug event will be triggered so the change will not be caught
at
resume time. Detecting that the EDID changed allows detecting
it.

Quoted 3 lines

Signed-off-by Paul Kocialkowski <[email protected]

tel.
com>

I can't find the older mails I've typed about this, but the plan
we've
discussed a while back was:
- Add a generational counter to each connector, maybe even expose
it
to
userspace.

- Increment that counter every time something changed, e.g. connector->status in the propbe helpers, or when attaching a
new
edid
with the set_edid helper.

Quoted 3 lines

- Tada, no changes needed to drivers, and easily extensible to other
things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything
has
changed when a monitor was replaced by another one, so I don't see
how
adding a counter in the mix would help.

Quoted 3 lines

Could you provide more details about the reasoning? I feel like I'm
missing something here.

Your bug doesn't just exist over s/r, it's just much easier to observe
in s/r since users can take however long they want to with plugging in
a different monitor. But the same issue exists e.g. when we go from
hpd to polling because too much noise on the line.

Wrt the suspend/resume issue: What we need to do on resume is do a full reprobe of all outputs, in an async worker. Telling userspace to
do this by sending an uevent was the cheapest way, but it'd be better
if the kernel could do that asynchronously and inform userspace about
the exact changes. And there's more to reprobe than just the edid, and
we don't want to re-invent a separate reprobe path just for resume like you start in your patch series. So yeah my plan was missing:

Quoted 3 lines

- force a full async reprobe after resume (maybe we could reuse the
poll worker for that as a one-shot).

First off, I definitely agree we need a way to tell userspace exactly
what has happened. I wanted to start a discussion about that in i-g- t
patch "Unrelated hotplug uevent masking out actual test result" but it
didn't get much traction. For testing purposes, it is unacceptable that
userspace only gets notified that "something happened".

Quoted 7 lines

Still, as far as I know, userspace is expected to ask for a full reprobe
when something has changed, and that is apparently part of the DRM spec,
so we can't expect that it could query for an update on "only the things
that changed".

We can update that spec in a backwards compatible way. E.g. we can ask for
the current properties without forcing a reprobe (won't even call down into the driver), and userspace could use that to check which connector
has an incremented epoche counter since the last time it sampled things.
Then it can reprobe just that one.

Quoted 1 line

Old userspace wouldn't know about this, and would keep working as-is.

So the level of detail you're aiming at providing userspace is "connector foo changed" then? I agree it is better than the current "some connector(s) changed", but what I'd like to see for proper testing is a way to find out "bar for connector foo changed".

Quoted 30 lines

Quoted 5 lines

However, one way to mitigate this is to make sure that the driver knows
what changed and only updates these things when a full reprobe is requested. Is this the approach that you have in mind?

Quoted 13 lines

The methodology behind my series follows what is currently done:
detect
change in whatever way necessary, inform userspace and let it trigger
full reprobe. If I'm understanding correctly, what you're suggesting is
instead to reprobe what is needed on the kernel side when an associated
change occurs instead of having userspace trigger it, and then let userspace aware that something changed and return the "cached" updated
status when userspace asks for the subsequent reprobe. Is that correct?

There's two things: the uapi discussion and the internal implementation,
imo their separate (but somewhat connected) topics.

- For the internal implementation of detecting edid changes I don't like
your approach of rolling a completely new detect path just for
resume.
I think we can very well integrate that into the existing probe code using the approach I've laid out.

Quoted 9 lines

- There's more than just edid (e.g. hdcp status, various stuff that's handled in dp aux for DP sinks), and I think a general mechanism for tracking that something changed will be useful for the internal implementation. The other plan would be that we have to wire a bool changed through the entire probe stack, and make sure it's handled correctly everywhere, which is a) a lot more work b) more fragile.
Doing
a connector->status_epoch++ everywhere we detect a change is _much_ simpler.

So to summarize, the following would happen: an async worker would detect whether something changed, then increase the counter for that connector and notify userspace, which would trigger full reprobe of that connector only. Legacy userspace would just trigger full reprobe for all connectors.

I am still under the impression that you'd like the full reprobe to be done on the kernel's async worker, to detect that e.g. EDID changed. But then userspace is going to fully reprobe again, so it will be duplicated. Unless the kernel also keeps a reference of the last time the counter was read from userspace, to determine when to skip full reprobe when it is asked from userspace? That feels pretty similar to having a bool indicating change.

My approach here was to look specifically for the thing that can change in the async worker (only EDID with this change, but it could be extended for the other things you mentioned) as to reduce the duplication as much as possible.

Quoted 9 lines

- For the uapi change: We already support returning the cached stuff, the
only bit that's missing is the epoch counter to let userspace know
where
it might need to do a full reprobe. Or maybe we'll just spec that a
full
reprobe isn't necessary after a hpd event (but that's unlikely to
work
out given how many bugs we'd need to fix first).

Okay, thanks for the additional explanation. I think I'm getting a better grasp on your idea.

Quoted 138 lines · Paul Kocialkowski

Quoted 137 lines · Daniel Vetter

Quoted 115 lines · Paul Kocialkowski

Quoted 96 lines · Daniel Vetter

Quoted 63 lines · Paul Kocialkowski

Quoted 47 lines · Daniel Vetter

Quoted 20 lines · Paul Kocialkowski

This adds a common drm helper to detect whether the EDID changed
from
the last known cached one. This is useful help detect that a
monitor
was
changed during a suspend/resume cycle.

When that happens (a monitor is replaced by another one during
suspend),
no hotplug event will be triggered so the change will not be
caught
at
resume time. Detecting that the EDID changed allows detecting
it.

Quoted 4 lines

Signed-off-by Paul Kocialkowski <paul.kocialkowski@linu

x.in
tel.
com>

I can't find the older mails I've typed about this, but the
plan
we've
discussed a while back was:
- Add a generational counter to each connector, maybe even expose
it
to
userspace.

- Increment that counter every time something changed, e.g.
connector->status in the propbe helpers, or when
attaching a
new
edid
with the set_edid helper.

Quoted 4 lines

- Tada, no changes needed to drivers, and easily
extensible to
other
things than edid!

I don't see how it solves the problem here though. After a suspend/resume cycle, there is simply no indication that anything
has
changed when a monitor was replaced by another one, so I don't
see
how
adding a counter in the mix would help.

Quoted 4 lines

Could you provide more details about the reasoning? I feel like
I'm
missing something here.

Your bug doesn't just exist over s/r, it's just much easier to observe
in s/r since users can take however long they want to with plugging in
a different monitor. But the same issue exists e.g. when we go from
hpd to polling because too much noise on the line.

Wrt the suspend/resume issue: What we need to do on resume is do a
full reprobe of all outputs, in an async worker. Telling userspace
to
do this by sending an uevent was the cheapest way, but it'd be better
if the kernel could do that asynchronously and inform userspace
about
the exact changes. And there's more to reprobe than just the edid,
and
we don't want to re-invent a separate reprobe path just for resume
like you start in your patch series. So yeah my plan was missing:

Quoted 4 lines

- force a full async reprobe after resume (maybe we could reuse
the
poll worker for that as a one-shot).

First off, I definitely agree we need a way to tell userspace exactly
what has happened. I wanted to start a discussion about that in i-g-
t
patch "Unrelated hotplug uevent masking out actual test result" but
it
didn't get much traction. For testing purposes, it is unacceptable
that
userspace only gets notified that "something happened".

Quoted 9 lines

Still, as far as I know, userspace is expected to ask for a full reprobe
when something has changed, and that is apparently part of the DRM
spec,
so we can't expect that it could query for an update on "only the
things
that changed".

We can update that spec in a backwards compatible way. E.g. we can ask
for
the current properties without forcing a reprobe (won't even call down
into the driver), and userspace could use that to check which connector
has an incremented epoche counter since the last time it sampled things.
Then it can reprobe just that one.

Quoted 8 lines

Quoted 2 lines

Old userspace wouldn't know about this, and would keep working as- is.

So the level of detail you're aiming at providing userspace is "connector foo changed" then? I agree it is better than the current "some connector(s) changed", but what I'd like to see for proper testing
is a way to find out "bar for connector foo changed".

If you want taht level of detail you need introspection in a in-kernel selftest I think. We'd need to rather massively change/extend the uapi to
support that level of testing through the uapi, and thus far no one else
is asking for it with a real use-case.

Quoted 92 lines

Quoted 36 lines

Quoted 7 lines

However, one way to mitigate this is to make sure that the driver
knows
what changed and only updates these things when a full reprobe is
requested. Is this the approach that you have in mind?

Quoted 15 lines

The methodology behind my series follows what is currently done:
detect
change in whatever way necessary, inform userspace and let it trigger
full reprobe. If I'm understanding correctly, what you're suggesting
is
instead to reprobe what is needed on the kernel side when an associated
change occurs instead of having userspace trigger it, and then let
userspace aware that something changed and return the "cached" updated
status when userspace asks for the subsequent reprobe. Is that correct?

There's two things: the uapi discussion and the internal implementation,
imo their separate (but somewhat connected) topics.

- For the internal implementation of detecting edid changes I don't
like
your approach of rolling a completely new detect path just for
resume.
I think we can very well integrate that into the existing probe
code
using the approach I've laid out.

Quoted 15 lines

- There's more than just edid (e.g. hdcp status, various stuff that's
handled in dp aux for DP sinks), and I think a general mechanism
for
tracking that something changed will be useful for the internal implementation. The other plan would be that we have to wire a
bool
changed through the entire probe stack, and make sure it's
handled
correctly everywhere, which is a) a lot more work b) more
fragile.
Doing
a connector->status_epoch++ everywhere we detect a change is
_much_
simpler.

So to summarize, the following would happen: an async worker would detect whether something changed, then increase the counter for that connector and notify userspace, which would trigger full reprobe of that
connector only. Legacy userspace would just trigger full reprobe for all
connectors.

I am still under the impression that you'd like the full reprobe to be
done on the kernel's async worker, to detect that e.g. EDID changed. But
then userspace is going to fully reprobe again, so it will be duplicated. Unless the kernel also keeps a reference of the last time
the counter was read from userspace, to determine when to skip full reprobe when it is asked from userspace? That feels pretty similar to
having a bool indicating change.

My approach here was to look specifically for the thing that can change
in the async worker (only EDID with this change, but it could be extended for the other things you mentioned) as to reduce the duplication as much as possible.

Quoted 13 lines

- For the uapi change: We already support returning the cached stuff,
the
only bit that's missing is the epoch counter to let userspace
know
where
it might need to do a full reprobe. Or maybe we'll just spec
that a
full
reprobe isn't necessary after a hpd event (but that's unlikely
to
work
out given how many bugs we'd need to fix first).

Quoted 2 lines

Okay, thanks for the additional explanation. I think I'm getting a better grasp on your idea.

Yeah, the full reprobe isn't needed if you we spec that as the new fancy
behaviour. We already support getting the cached values through drmModeGetConnectorCurrent(), without forcing a full reprobe.

Note that really the only thing the hpd handling doesn't do is fill and
filter the mode list. I guess as part of this work we should fix that, that should take care of most of the needs for full reprobing.