* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 10:57 ` Ville Syrjälä
@ 2026-04-17 12:18 ` Julian Orth
2026-04-17 12:36 ` Julian Orth
` (2 subsequent siblings)
3 siblings, 0 replies; 25+ messages in thread
From: Julian Orth @ 2026-04-17 12:18 UTC (permalink / raw)
To: Ville Syrjälä
Cc: Michel Dänzer, Nicolas Frattaroli, Maarten Lankhorst,
Maxime Ripard, Thomas Zimmermann, David Airlie, Simona Vetter,
Louis Chauvet, Haneen Mohammed, Melissa Wen, Daniel Stone,
Ian Forbes, Dmitry Baryshkov, dri-devel, linux-kernel, kernel,
wayland-devel, Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 12:58 PM Ville Syrjälä
<ville.syrjala@linux.intel.com> wrote:
>
> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > On 4/16/26 15:16, Julian Orth wrote:
> > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > <nicolas.frattaroli@collabora.com> wrote:
> > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > >>> <nicolas.frattaroli@collabora.com> wrote:
> > >>>>
> > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > >>>> without it being passed the actual connector that caused it. This takes
> > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > >>>> propagating the connector ID when changing the connector's status.
> > >>>>
> > >>>> The motivation is that user-space applications such as Weston would
> > >>>> previously receive non-connector-specific hotplug events, and then have
> > >>>> to figure out themselves which connector needs to have a modeset
> > >>>> executed on. This notably did not work when the hotplug events came in
> > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > >>>> through the other connectors that could've caused it.
> > >>>
> > >>> Have you considered adding a u64 serial number as a DRM connector
> > >>> property that is incremented every time the connector changes in some
> > >>> way? Userspace could then check this serial number to see if the
> > >>> connector has changed since the last time it queried the serial
> > >>> number.
> > >>
> > >> The connector internally already has an epoch_counter member which
> > >> could be used for this. However, for the particular thing this
> > >> series fixes, I don't think exposing it through the uAPI is necessary
> > >> or desirable. Sending hotplug events specific to the connector does
> > >> not need any additional handling on the userspace side as long as it
> > >> already listens to the per-connector hotplug events in order to
> > >> avoid the pitfall described in the cover letter.
> > >
> > > I currently do not handle per-connector hotplug events. Instead,
> > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > kernel state for the device. That is
> > >
> > > - DRM_IOCTL_MODE_GETRESOURCES
> > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > >
> > > Once I have the new state, I compare it against the desired compositor
> > > state and perform a modeset if necessary.
> >
> > mutter is doing something similar as well.
> >
> >
> > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> >
> > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
>
> GPU reset should relight the display on its own really. That's what
> i915 does, albeit somewhat badly at the moment.
>
> > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
>
> There's nothing preventing the kernel from doing extra atomic
> commits whenever it wants. But if you want to punt the thing to
> userspace then the kernel must set the link-status property to
> bad, and then fire the hotplug uevent.
>
> > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
>
> Only for connectors where things changed, or the link-status shows bad.
>
> >
> >
> > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
>
> The actual argument is that you should not defer the hotplug
> handling when things get disconnected, mainly because of crap type-c
> firmware.
>
> I think the userspace behaviour there was that you get a disconnected,
> defer processing it, and then you get a reconnect, and then decide that
> nothing actually changed and a modeset is not needed after all. That is
> not correct IMO. Clearly a ->disconnect->reconnect should count as a
> change in the connector's state, and a full modeset is thus required.
The connector state, AIUI from userspace, is the set of all DRM
connector properties. If the true state changes in a way that requires
userspace to act but the DRM properties are unchanged, then I would
argue that that is a kernel bug.
The link-status property seems like it is perfect for this.
> The kernel can of course then decide that the full modeset is not
> actually required and skip the modeset part during the commit. But
> userspace cannot make that determination.
>
> I suppose what we could maybe do there to force userspace's hand is set
> the link-status to bad already when the thing gets disconnected, and keep
> it like that until a full disable+re-enable cycle has been done. Then
> userspace could not think that the ->disconnect->reconnect is a NOP
> (which I still think is incorrect behaviour).
>
> --
> Ville Syrjälä
> Intel
^ permalink raw reply [flat|nested] 25+ messages in thread* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 10:57 ` Ville Syrjälä
2026-04-17 12:18 ` Julian Orth
@ 2026-04-17 12:36 ` Julian Orth
2026-04-17 14:36 ` Ville Syrjälä
2026-04-17 12:42 ` Nicolas Frattaroli
2026-04-17 14:17 ` Michel Dänzer
3 siblings, 1 reply; 25+ messages in thread
From: Julian Orth @ 2026-04-17 12:36 UTC (permalink / raw)
To: Ville Syrjälä
Cc: Michel Dänzer, Nicolas Frattaroli, Maarten Lankhorst,
Maxime Ripard, Thomas Zimmermann, David Airlie, Simona Vetter,
Louis Chauvet, Haneen Mohammed, Melissa Wen, Daniel Stone,
Ian Forbes, Dmitry Baryshkov, dri-devel, linux-kernel, kernel,
wayland-devel, Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 12:58 PM Ville Syrjälä
<ville.syrjala@linux.intel.com> wrote:
>
> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > On 4/16/26 15:16, Julian Orth wrote:
> > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > <nicolas.frattaroli@collabora.com> wrote:
> > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > >>> <nicolas.frattaroli@collabora.com> wrote:
> > >>>>
> > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > >>>> without it being passed the actual connector that caused it. This takes
> > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > >>>> propagating the connector ID when changing the connector's status.
> > >>>>
> > >>>> The motivation is that user-space applications such as Weston would
> > >>>> previously receive non-connector-specific hotplug events, and then have
> > >>>> to figure out themselves which connector needs to have a modeset
> > >>>> executed on. This notably did not work when the hotplug events came in
> > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > >>>> through the other connectors that could've caused it.
> > >>>
> > >>> Have you considered adding a u64 serial number as a DRM connector
> > >>> property that is incremented every time the connector changes in some
> > >>> way? Userspace could then check this serial number to see if the
> > >>> connector has changed since the last time it queried the serial
> > >>> number.
> > >>
> > >> The connector internally already has an epoch_counter member which
> > >> could be used for this. However, for the particular thing this
> > >> series fixes, I don't think exposing it through the uAPI is necessary
> > >> or desirable. Sending hotplug events specific to the connector does
> > >> not need any additional handling on the userspace side as long as it
> > >> already listens to the per-connector hotplug events in order to
> > >> avoid the pitfall described in the cover letter.
> > >
> > > I currently do not handle per-connector hotplug events. Instead,
> > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > kernel state for the device. That is
> > >
> > > - DRM_IOCTL_MODE_GETRESOURCES
> > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > >
> > > Once I have the new state, I compare it against the desired compositor
> > > state and perform a modeset if necessary.
> >
> > mutter is doing something similar as well.
> >
> >
> > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> >
> > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
>
> GPU reset should relight the display on its own really. That's what
> i915 does, albeit somewhat badly at the moment.
>
> > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
>
> There's nothing preventing the kernel from doing extra atomic
> commits whenever it wants. But if you want to punt the thing to
> userspace then the kernel must set the link-status property to
> bad, and then fire the hotplug uevent.
>
> > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
>
> Only for connectors where things changed, or the link-status shows bad.
>
> >
> >
> > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
>
> The actual argument is that you should not defer the hotplug
> handling when things get disconnected, mainly because of crap type-c
> firmware.
>
> I think the userspace behaviour there was that you get a disconnected,
> defer processing it, and then you get a reconnect, and then decide that
> nothing actually changed and a modeset is not needed after all. That is
> not correct IMO. Clearly a ->disconnect->reconnect should count as a
> change in the connector's state, and a full modeset is thus required.
> The kernel can of course then decide that the full modeset is not
> actually required and skip the modeset part during the commit. But
> userspace cannot make that determination.
>
> I suppose what we could maybe do there to force userspace's hand is set
> the link-status to bad already when the thing gets disconnected, and keep
> it like that until a full disable+re-enable cycle has been done. Then
> userspace could not think that the ->disconnect->reconnect is a NOP
> (which I still think is incorrect behaviour).
I don't think the kernel can make the assumption that userspace sees
any udev events. For example, say such a disconnect/reconnect cycle
happens that leaves the connector in a state that would require two
modeset. Nobody acts on these modesets and some time later the
compositor starts. Obviously the compositor is not going to receive
any udev events for the disconnect/reconnect cycle that happened long
ago. This would leave the connector in a permanently unusable state.
>
> --
> Ville Syrjälä
> Intel
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 12:36 ` Julian Orth
@ 2026-04-17 14:36 ` Ville Syrjälä
0 siblings, 0 replies; 25+ messages in thread
From: Ville Syrjälä @ 2026-04-17 14:36 UTC (permalink / raw)
To: Julian Orth
Cc: Michel Dänzer, Nicolas Frattaroli, Maarten Lankhorst,
Maxime Ripard, Thomas Zimmermann, David Airlie, Simona Vetter,
Louis Chauvet, Haneen Mohammed, Melissa Wen, Daniel Stone,
Ian Forbes, Dmitry Baryshkov, dri-devel, linux-kernel, kernel,
wayland-devel, Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 02:36:11PM +0200, Julian Orth wrote:
> On Fri, Apr 17, 2026 at 12:58 PM Ville Syrjälä
> <ville.syrjala@linux.intel.com> wrote:
> >
> > On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > > On 4/16/26 15:16, Julian Orth wrote:
> > > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > > <nicolas.frattaroli@collabora.com> wrote:
> > > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > > >>> <nicolas.frattaroli@collabora.com> wrote:
> > > >>>>
> > > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > > >>>> without it being passed the actual connector that caused it. This takes
> > > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > > >>>> propagating the connector ID when changing the connector's status.
> > > >>>>
> > > >>>> The motivation is that user-space applications such as Weston would
> > > >>>> previously receive non-connector-specific hotplug events, and then have
> > > >>>> to figure out themselves which connector needs to have a modeset
> > > >>>> executed on. This notably did not work when the hotplug events came in
> > > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > > >>>> through the other connectors that could've caused it.
> > > >>>
> > > >>> Have you considered adding a u64 serial number as a DRM connector
> > > >>> property that is incremented every time the connector changes in some
> > > >>> way? Userspace could then check this serial number to see if the
> > > >>> connector has changed since the last time it queried the serial
> > > >>> number.
> > > >>
> > > >> The connector internally already has an epoch_counter member which
> > > >> could be used for this. However, for the particular thing this
> > > >> series fixes, I don't think exposing it through the uAPI is necessary
> > > >> or desirable. Sending hotplug events specific to the connector does
> > > >> not need any additional handling on the userspace side as long as it
> > > >> already listens to the per-connector hotplug events in order to
> > > >> avoid the pitfall described in the cover letter.
> > > >
> > > > I currently do not handle per-connector hotplug events. Instead,
> > > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > > kernel state for the device. That is
> > > >
> > > > - DRM_IOCTL_MODE_GETRESOURCES
> > > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > > >
> > > > Once I have the new state, I compare it against the desired compositor
> > > > state and perform a modeset if necessary.
> > >
> > > mutter is doing something similar as well.
> > >
> > >
> > > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> > >
> > > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> >
> > GPU reset should relight the display on its own really. That's what
> > i915 does, albeit somewhat badly at the moment.
> >
> > > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> >
> > There's nothing preventing the kernel from doing extra atomic
> > commits whenever it wants. But if you want to punt the thing to
> > userspace then the kernel must set the link-status property to
> > bad, and then fire the hotplug uevent.
> >
> > > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
> >
> > Only for connectors where things changed, or the link-status shows bad.
> >
> > >
> > >
> > > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
> >
> > The actual argument is that you should not defer the hotplug
> > handling when things get disconnected, mainly because of crap type-c
> > firmware.
> >
> > I think the userspace behaviour there was that you get a disconnected,
> > defer processing it, and then you get a reconnect, and then decide that
> > nothing actually changed and a modeset is not needed after all. That is
> > not correct IMO. Clearly a ->disconnect->reconnect should count as a
> > change in the connector's state, and a full modeset is thus required.
> > The kernel can of course then decide that the full modeset is not
> > actually required and skip the modeset part during the commit. But
> > userspace cannot make that determination.
> >
> > I suppose what we could maybe do there to force userspace's hand is set
> > the link-status to bad already when the thing gets disconnected, and keep
> > it like that until a full disable+re-enable cycle has been done. Then
> > userspace could not think that the ->disconnect->reconnect is a NOP
> > (which I still think is incorrect behaviour).
>
> I don't think the kernel can make the assumption that userspace sees
> any udev events. For example, say such a disconnect/reconnect cycle
> happens that leaves the connector in a state that would require two
> modeset. Nobody acts on these modesets and some time later the
> compositor starts. Obviously the compositor is not going to receive
> any udev events for the disconnect/reconnect cycle that happened long
> ago. This would leave the connector in a permanently unusable state.
I'm not sure why those outputs would even be enabled if there is
no one around to do kms stuff? Though I suppose there might be some
race conditions if you are switching drm masters exactly when the
uevents are sent.
We do have a sort of plan to always do the disconnect disable from
the kernel, but we can't just disable everything the normal way
as we're not allowed to corrupt the uapi kms state. So the idea
is we'd disable everything internally, and fake it so that
userspace still sees it as enabled. But no one has actually
implemented this so far.
A potential short term solution might be to disable+re-enable
(in our current fallback tbt mode) automagically on disconnect.
But that extra re-enable is not something we actually want in
case userspace is going to do its job properly. Instead of
having one disable modeset on disconnect, we'd always end up
with disable+re-enable+disable. Though I suppose one option
would be to defer the internal disable+re-enable a little bit
to give userspace some time to do its job.
--
Ville Syrjälä
Intel
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 10:57 ` Ville Syrjälä
2026-04-17 12:18 ` Julian Orth
2026-04-17 12:36 ` Julian Orth
@ 2026-04-17 12:42 ` Nicolas Frattaroli
2026-04-17 14:16 ` Ville Syrjälä
2026-04-17 14:17 ` Michel Dänzer
3 siblings, 1 reply; 25+ messages in thread
From: Nicolas Frattaroli @ 2026-04-17 12:42 UTC (permalink / raw)
To: Michel Dänzer, Ville Syrjälä
Cc: Julian Orth, Maarten Lankhorst, Maxime Ripard, Thomas Zimmermann,
David Airlie, Simona Vetter, Louis Chauvet, Haneen Mohammed,
Melissa Wen, Daniel Stone, Ian Forbes, Dmitry Baryshkov,
dri-devel, linux-kernel, kernel, wayland-devel, Marius Vlad,
Imre Deak
On Friday, 17 April 2026 12:57:58 Central European Summer Time Ville Syrjälä wrote:
> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > On 4/16/26 15:16, Julian Orth wrote:
> > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > <nicolas.frattaroli@collabora.com> wrote:
> > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > >>> <nicolas.frattaroli@collabora.com> wrote:
> > >>>>
> > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > >>>> without it being passed the actual connector that caused it. This takes
> > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > >>>> propagating the connector ID when changing the connector's status.
> > >>>>
> > >>>> The motivation is that user-space applications such as Weston would
> > >>>> previously receive non-connector-specific hotplug events, and then have
> > >>>> to figure out themselves which connector needs to have a modeset
> > >>>> executed on. This notably did not work when the hotplug events came in
> > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > >>>> through the other connectors that could've caused it.
> > >>>
> > >>> Have you considered adding a u64 serial number as a DRM connector
> > >>> property that is incremented every time the connector changes in some
> > >>> way? Userspace could then check this serial number to see if the
> > >>> connector has changed since the last time it queried the serial
> > >>> number.
> > >>
> > >> The connector internally already has an epoch_counter member which
> > >> could be used for this. However, for the particular thing this
> > >> series fixes, I don't think exposing it through the uAPI is necessary
> > >> or desirable. Sending hotplug events specific to the connector does
> > >> not need any additional handling on the userspace side as long as it
> > >> already listens to the per-connector hotplug events in order to
> > >> avoid the pitfall described in the cover letter.
> > >
> > > I currently do not handle per-connector hotplug events. Instead,
> > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > kernel state for the device. That is
> > >
> > > - DRM_IOCTL_MODE_GETRESOURCES
> > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > >
> > > Once I have the new state, I compare it against the desired compositor
> > > state and perform a modeset if necessary.
> >
> > mutter is doing something similar as well.
> >
> >
> > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> >
> > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
>
> GPU reset should relight the display on its own really. That's what
> i915 does, albeit somewhat badly at the moment.
>
> > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
>
> There's nothing preventing the kernel from doing extra atomic
> commits whenever it wants. But if you want to punt the thing to
> userspace then the kernel must set the link-status property to
> bad, and then fire the hotplug uevent.
>
> > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
>
> Only for connectors where things changed, or the link-status shows bad.
>
> >
> >
> > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
>
> The actual argument is that you should not defer the hotplug
> handling when things get disconnected, mainly because of crap type-c
> firmware.
>
> I think the userspace behaviour there was that you get a disconnected,
> defer processing it, and then you get a reconnect, and then decide that
> nothing actually changed and a modeset is not needed after all. That is
> not correct IMO. Clearly a ->disconnect->reconnect should count as a
> change in the connector's state, and a full modeset is thus required.
> The kernel can of course then decide that the full modeset is not
> actually required and skip the modeset part during the commit. But
> userspace cannot make that determination.
So just to loop around to the patches here: is sending per-connector
hotplug events not acceptable? Your review[1] on v5 indicated you had
a problem with the implementation, not a fundamental issue with the
behaviour the patch tried to change.
> I suppose what we could maybe do there to force userspace's hand is set
> the link-status to bad already when the thing gets disconnected, and keep
> it like that until a full disable+re-enable cycle has been done. Then
> userspace could not think that the ->disconnect->reconnect is a NOP
> (which I still think is incorrect behaviour).
The documentation of drm_connector_set_link_status_property explicitly
states:
* Note: Drivers cannot rely on userspace to support this property and
* issue a modeset. As such, they may choose to handle issues (like
* re-training a link) without userspace's intervention.
which I think conflicts with your suggestion.
Kind regards,
Nicolas Frattaroli
https://lore.kernel.org/dri-devel/aROKwmZxFt52g4ed@intel.com/ [1]
^ permalink raw reply [flat|nested] 25+ messages in thread* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 12:42 ` Nicolas Frattaroli
@ 2026-04-17 14:16 ` Ville Syrjälä
2026-04-17 15:00 ` Nicolas Frattaroli
0 siblings, 1 reply; 25+ messages in thread
From: Ville Syrjälä @ 2026-04-17 14:16 UTC (permalink / raw)
To: Nicolas Frattaroli
Cc: Michel Dänzer, Julian Orth, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 02:42:36PM +0200, Nicolas Frattaroli wrote:
> On Friday, 17 April 2026 12:57:58 Central European Summer Time Ville Syrjälä wrote:
> > On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > > On 4/16/26 15:16, Julian Orth wrote:
> > > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > > <nicolas.frattaroli@collabora.com> wrote:
> > > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > > >>> <nicolas.frattaroli@collabora.com> wrote:
> > > >>>>
> > > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > > >>>> without it being passed the actual connector that caused it. This takes
> > > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > > >>>> propagating the connector ID when changing the connector's status.
> > > >>>>
> > > >>>> The motivation is that user-space applications such as Weston would
> > > >>>> previously receive non-connector-specific hotplug events, and then have
> > > >>>> to figure out themselves which connector needs to have a modeset
> > > >>>> executed on. This notably did not work when the hotplug events came in
> > > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > > >>>> through the other connectors that could've caused it.
> > > >>>
> > > >>> Have you considered adding a u64 serial number as a DRM connector
> > > >>> property that is incremented every time the connector changes in some
> > > >>> way? Userspace could then check this serial number to see if the
> > > >>> connector has changed since the last time it queried the serial
> > > >>> number.
> > > >>
> > > >> The connector internally already has an epoch_counter member which
> > > >> could be used for this. However, for the particular thing this
> > > >> series fixes, I don't think exposing it through the uAPI is necessary
> > > >> or desirable. Sending hotplug events specific to the connector does
> > > >> not need any additional handling on the userspace side as long as it
> > > >> already listens to the per-connector hotplug events in order to
> > > >> avoid the pitfall described in the cover letter.
> > > >
> > > > I currently do not handle per-connector hotplug events. Instead,
> > > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > > kernel state for the device. That is
> > > >
> > > > - DRM_IOCTL_MODE_GETRESOURCES
> > > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > > >
> > > > Once I have the new state, I compare it against the desired compositor
> > > > state and perform a modeset if necessary.
> > >
> > > mutter is doing something similar as well.
> > >
> > >
> > > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> > >
> > > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> >
> > GPU reset should relight the display on its own really. That's what
> > i915 does, albeit somewhat badly at the moment.
> >
> > > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> >
> > There's nothing preventing the kernel from doing extra atomic
> > commits whenever it wants. But if you want to punt the thing to
> > userspace then the kernel must set the link-status property to
> > bad, and then fire the hotplug uevent.
> >
> > > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
> >
> > Only for connectors where things changed, or the link-status shows bad.
> >
> > >
> > >
> > > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
> >
> > The actual argument is that you should not defer the hotplug
> > handling when things get disconnected, mainly because of crap type-c
> > firmware.
> >
> > I think the userspace behaviour there was that you get a disconnected,
> > defer processing it, and then you get a reconnect, and then decide that
> > nothing actually changed and a modeset is not needed after all. That is
> > not correct IMO. Clearly a ->disconnect->reconnect should count as a
> > change in the connector's state, and a full modeset is thus required.
> > The kernel can of course then decide that the full modeset is not
> > actually required and skip the modeset part during the commit. But
> > userspace cannot make that determination.
>
> So just to loop around to the patches here: is sending per-connector
> hotplug events not acceptable? Your review[1] on v5 indicated you had
> a problem with the implementation, not a fundamental issue with the
> behaviour the patch tried to change.
What I was saying is that we already have the epoch_counter. If
there are gaps in the implementation then those should just be
fixed. We don't want a second implementation of the same
mechanism.
>
> > I suppose what we could maybe do there to force userspace's hand is set
> > the link-status to bad already when the thing gets disconnected, and keep
> > it like that until a full disable+re-enable cycle has been done. Then
> > userspace could not think that the ->disconnect->reconnect is a NOP
> > (which I still think is incorrect behaviour).
>
> The documentation of drm_connector_set_link_status_property explicitly
> states:
>
> * Note: Drivers cannot rely on userspace to support this property and
> * issue a modeset. As such, they may choose to handle issues (like
> * re-training a link) without userspace's intervention.
>
> which I think conflicts with your suggestion.
In i915 we do the retraining in kernel. IIRC we only really use the bad
link stuff once retraining has failed hard enough that userspace
intervention is needed, ie. when we cannot reduce the link parameters
anymore without userspace selecting a mode with a lower resolution.
I think some more recent drivers may have opted to forego all in kernel
retraining and just punt it all to userspace. I think the link-status
property has existed for long enough that any userspace that doesn't
support it could simply be considered broken.
Actually I know of a single "userspace" that to my knowledge doesn't
implement the link-status stuff, and that is the in kernel fb helper.
That should probably be remedied...
--
Ville Syrjälä
Intel
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 14:16 ` Ville Syrjälä
@ 2026-04-17 15:00 ` Nicolas Frattaroli
2026-04-17 15:19 ` Ville Syrjälä
0 siblings, 1 reply; 25+ messages in thread
From: Nicolas Frattaroli @ 2026-04-17 15:00 UTC (permalink / raw)
To: Ville Syrjälä
Cc: Michel Dänzer, Julian Orth, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On Friday, 17 April 2026 16:16:01 Central European Summer Time Ville Syrjälä wrote:
> On Fri, Apr 17, 2026 at 02:42:36PM +0200, Nicolas Frattaroli wrote:
> > On Friday, 17 April 2026 12:57:58 Central European Summer Time Ville Syrjälä wrote:
> > > On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > > > On 4/16/26 15:16, Julian Orth wrote:
> > > > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > > > <nicolas.frattaroli@collabora.com> wrote:
> > > > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > > > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > > > >>> <nicolas.frattaroli@collabora.com> wrote:
> > > > >>>>
> > > > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > > > >>>> without it being passed the actual connector that caused it. This takes
> > > > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > > > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > > > >>>> propagating the connector ID when changing the connector's status.
> > > > >>>>
> > > > >>>> The motivation is that user-space applications such as Weston would
> > > > >>>> previously receive non-connector-specific hotplug events, and then have
> > > > >>>> to figure out themselves which connector needs to have a modeset
> > > > >>>> executed on. This notably did not work when the hotplug events came in
> > > > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > > > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > > > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > > > >>>> through the other connectors that could've caused it.
> > > > >>>
> > > > >>> Have you considered adding a u64 serial number as a DRM connector
> > > > >>> property that is incremented every time the connector changes in some
> > > > >>> way? Userspace could then check this serial number to see if the
> > > > >>> connector has changed since the last time it queried the serial
> > > > >>> number.
> > > > >>
> > > > >> The connector internally already has an epoch_counter member which
> > > > >> could be used for this. However, for the particular thing this
> > > > >> series fixes, I don't think exposing it through the uAPI is necessary
> > > > >> or desirable. Sending hotplug events specific to the connector does
> > > > >> not need any additional handling on the userspace side as long as it
> > > > >> already listens to the per-connector hotplug events in order to
> > > > >> avoid the pitfall described in the cover letter.
> > > > >
> > > > > I currently do not handle per-connector hotplug events. Instead,
> > > > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > > > kernel state for the device. That is
> > > > >
> > > > > - DRM_IOCTL_MODE_GETRESOURCES
> > > > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > > > >
> > > > > Once I have the new state, I compare it against the desired compositor
> > > > > state and perform a modeset if necessary.
> > > >
> > > > mutter is doing something similar as well.
> > > >
> > > >
> > > > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> > > >
> > > > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> > >
> > > GPU reset should relight the display on its own really. That's what
> > > i915 does, albeit somewhat badly at the moment.
> > >
> > > > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> > >
> > > There's nothing preventing the kernel from doing extra atomic
> > > commits whenever it wants. But if you want to punt the thing to
> > > userspace then the kernel must set the link-status property to
> > > bad, and then fire the hotplug uevent.
> > >
> > > > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
> > >
> > > Only for connectors where things changed, or the link-status shows bad.
> > >
> > > >
> > > >
> > > > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
> > >
> > > The actual argument is that you should not defer the hotplug
> > > handling when things get disconnected, mainly because of crap type-c
> > > firmware.
> > >
> > > I think the userspace behaviour there was that you get a disconnected,
> > > defer processing it, and then you get a reconnect, and then decide that
> > > nothing actually changed and a modeset is not needed after all. That is
> > > not correct IMO. Clearly a ->disconnect->reconnect should count as a
> > > change in the connector's state, and a full modeset is thus required.
> > > The kernel can of course then decide that the full modeset is not
> > > actually required and skip the modeset part during the commit. But
> > > userspace cannot make that determination.
> >
> > So just to loop around to the patches here: is sending per-connector
> > hotplug events not acceptable? Your review[1] on v5 indicated you had
> > a problem with the implementation, not a fundamental issue with the
> > behaviour the patch tried to change.
>
> What I was saying is that we already have the epoch_counter. If
> there are gaps in the implementation then those should just be
> fixed. We don't want a second implementation of the same
> mechanism.
epoch_counter is explicitly documented as "used to detect any other changes
in connector, besides status". pending_hp is changed if the status changes.
If we use epoch_counter for this, I think we'll need to keep track of the
last epoch_counter an hpd was sent for each connector, so we wouldn't win
anything in reduced state being tracked.
> >
> > > I suppose what we could maybe do there to force userspace's hand is set
> > > the link-status to bad already when the thing gets disconnected, and keep
> > > it like that until a full disable+re-enable cycle has been done. Then
> > > userspace could not think that the ->disconnect->reconnect is a NOP
> > > (which I still think is incorrect behaviour).
> >
> > The documentation of drm_connector_set_link_status_property explicitly
> > states:
> >
> > * Note: Drivers cannot rely on userspace to support this property and
> > * issue a modeset. As such, they may choose to handle issues (like
> > * re-training a link) without userspace's intervention.
> >
> > which I think conflicts with your suggestion.
>
> In i915 we do the retraining in kernel. IIRC we only really use the bad
> link stuff once retraining has failed hard enough that userspace
> intervention is needed, ie. when we cannot reduce the link parameters
> anymore without userspace selecting a mode with a lower resolution.
>
> I think some more recent drivers may have opted to forego all in kernel
> retraining and just punt it all to userspace. I think the link-status
> property has existed for long enough that any userspace that doesn't
> support it could simply be considered broken.
Is breaking userspace a better choice than sending per-connector hotplug
events rather than a single global hotplug event? I get where you're
coming from, in that userspace should not be trying to optimise away
modesets, as that's something the kernel can do in a race-free way. I'm
just not familiar enough with the landscape to know whether setting
link-status to bad will actually result in userspace doing a modeset,
and whether drivers in practice then optimise that modeset away if it
wasn't needed.
> Actually I know of a single "userspace" that to my knowledge doesn't
> implement the link-status stuff, and that is the in kernel fb helper.
> That should probably be remedied...
Kind regards,
Nicolas Frattaroli
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 15:00 ` Nicolas Frattaroli
@ 2026-04-17 15:19 ` Ville Syrjälä
0 siblings, 0 replies; 25+ messages in thread
From: Ville Syrjälä @ 2026-04-17 15:19 UTC (permalink / raw)
To: Nicolas Frattaroli
Cc: Michel Dänzer, Julian Orth, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 05:00:10PM +0200, Nicolas Frattaroli wrote:
> On Friday, 17 April 2026 16:16:01 Central European Summer Time Ville Syrjälä wrote:
> > On Fri, Apr 17, 2026 at 02:42:36PM +0200, Nicolas Frattaroli wrote:
> > > On Friday, 17 April 2026 12:57:58 Central European Summer Time Ville Syrjälä wrote:
> > > > On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> > > > > On 4/16/26 15:16, Julian Orth wrote:
> > > > > > On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> > > > > > <nicolas.frattaroli@collabora.com> wrote:
> > > > > >> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> > > > > >>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> > > > > >>> <nicolas.frattaroli@collabora.com> wrote:
> > > > > >>>>
> > > > > >>>> This series addresses a shortcoming whereby a hot plug event is sent
> > > > > >>>> without it being passed the actual connector that caused it. This takes
> > > > > >>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> > > > > >>>> path. It also adds support for the vkms driver (using ConfigFS) for
> > > > > >>>> propagating the connector ID when changing the connector's status.
> > > > > >>>>
> > > > > >>>> The motivation is that user-space applications such as Weston would
> > > > > >>>> previously receive non-connector-specific hotplug events, and then have
> > > > > >>>> to figure out themselves which connector needs to have a modeset
> > > > > >>>> executed on. This notably did not work when the hotplug events came in
> > > > > >>>> too fast, resulting in Weston missing an on-off-on transition of a
> > > > > >>>> connector, seeing that its state was unchanged from "on" so can't be the
> > > > > >>>> one that was hotplugged, and skipping reinitialising it as it looks
> > > > > >>>> through the other connectors that could've caused it.
> > > > > >>>
> > > > > >>> Have you considered adding a u64 serial number as a DRM connector
> > > > > >>> property that is incremented every time the connector changes in some
> > > > > >>> way? Userspace could then check this serial number to see if the
> > > > > >>> connector has changed since the last time it queried the serial
> > > > > >>> number.
> > > > > >>
> > > > > >> The connector internally already has an epoch_counter member which
> > > > > >> could be used for this. However, for the particular thing this
> > > > > >> series fixes, I don't think exposing it through the uAPI is necessary
> > > > > >> or desirable. Sending hotplug events specific to the connector does
> > > > > >> not need any additional handling on the userspace side as long as it
> > > > > >> already listens to the per-connector hotplug events in order to
> > > > > >> avoid the pitfall described in the cover letter.
> > > > > >
> > > > > > I currently do not handle per-connector hotplug events. Instead,
> > > > > > whenever I get a UDEV change event for a device, I re-fetch the entire
> > > > > > kernel state for the device. That is
> > > > > >
> > > > > > - DRM_IOCTL_MODE_GETRESOURCES
> > > > > > - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> > > > > > - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> > > > > > - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> > > > > > - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> > > > > >
> > > > > > Once I have the new state, I compare it against the desired compositor
> > > > > > state and perform a modeset if necessary.
> > > > >
> > > > > mutter is doing something similar as well.
> > > > >
> > > > >
> > > > > Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> > > > >
> > > > > The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> > > >
> > > > GPU reset should relight the display on its own really. That's what
> > > > i915 does, albeit somewhat badly at the moment.
> > > >
> > > > > A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> > > >
> > > > There's nothing preventing the kernel from doing extra atomic
> > > > commits whenever it wants. But if you want to punt the thing to
> > > > userspace then the kernel must set the link-status property to
> > > > bad, and then fire the hotplug uevent.
> > > >
> > > > > If this "modeset required after hotplug event" rule is confirmed, it means that after a hotplug event without connector ID, the compositor must do a modeset for all connectors.
> > > >
> > > > Only for connectors where things changed, or the link-status shows bad.
> > > >
> > > > >
> > > > >
> > > > > P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
> > > >
> > > > The actual argument is that you should not defer the hotplug
> > > > handling when things get disconnected, mainly because of crap type-c
> > > > firmware.
> > > >
> > > > I think the userspace behaviour there was that you get a disconnected,
> > > > defer processing it, and then you get a reconnect, and then decide that
> > > > nothing actually changed and a modeset is not needed after all. That is
> > > > not correct IMO. Clearly a ->disconnect->reconnect should count as a
> > > > change in the connector's state, and a full modeset is thus required.
> > > > The kernel can of course then decide that the full modeset is not
> > > > actually required and skip the modeset part during the commit. But
> > > > userspace cannot make that determination.
> > >
> > > So just to loop around to the patches here: is sending per-connector
> > > hotplug events not acceptable? Your review[1] on v5 indicated you had
> > > a problem with the implementation, not a fundamental issue with the
> > > behaviour the patch tried to change.
> >
> > What I was saying is that we already have the epoch_counter. If
> > there are gaps in the implementation then those should just be
> > fixed. We don't want a second implementation of the same
> > mechanism.
>
> epoch_counter is explicitly documented as "used to detect any other changes
> in connector, besides status".
That's just some nonsense that got in by accident. It was always
meant to be a "did anything for this connector change?" type of thing.
Looks like the bogus docs were part of the original commit, but
it even disagrees with the commit message itself.
> pending_hp is changed if the status changes.
>
> If we use epoch_counter for this, I think we'll need to keep track of the
> last epoch_counter an hpd was sent for each connector, so we wouldn't win
> anything in reduced state being tracked.
What's wrong with just looking at the epoch before and after the detect?
>
> > >
> > > > I suppose what we could maybe do there to force userspace's hand is set
> > > > the link-status to bad already when the thing gets disconnected, and keep
> > > > it like that until a full disable+re-enable cycle has been done. Then
> > > > userspace could not think that the ->disconnect->reconnect is a NOP
> > > > (which I still think is incorrect behaviour).
> > >
> > > The documentation of drm_connector_set_link_status_property explicitly
> > > states:
> > >
> > > * Note: Drivers cannot rely on userspace to support this property and
> > > * issue a modeset. As such, they may choose to handle issues (like
> > > * re-training a link) without userspace's intervention.
> > >
> > > which I think conflicts with your suggestion.
> >
> > In i915 we do the retraining in kernel. IIRC we only really use the bad
> > link stuff once retraining has failed hard enough that userspace
> > intervention is needed, ie. when we cannot reduce the link parameters
> > anymore without userspace selecting a mode with a lower resolution.
> >
> > I think some more recent drivers may have opted to forego all in kernel
> > retraining and just punt it all to userspace. I think the link-status
> > property has existed for long enough that any userspace that doesn't
> > support it could simply be considered broken.
>
> Is breaking userspace a better choice than sending per-connector hotplug
> events rather than a single global hotplug event? I get where you're
> coming from, in that userspace should not be trying to optimise away
> modesets, as that's something the kernel can do in a race-free way. I'm
> just not familiar enough with the landscape to know whether setting
> link-status to bad will actually result in userspace doing a modeset,
> and whether drivers in practice then optimise that modeset away if it
> wasn't needed.
I'm not suggesting using link-status for normal hotplug events
where something on the connector actually changed. In those cases
userspace should really be doing a full modeset anyway.
>
> > Actually I know of a single "userspace" that to my knowledge doesn't
> > implement the link-status stuff, and that is the in kernel fb helper.
> > That should probably be remedied...
>
> Kind regards,
> Nicolas Frattaroli
>
--
Ville Syrjälä
Intel
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 10:57 ` Ville Syrjälä
` (2 preceding siblings ...)
2026-04-17 12:42 ` Nicolas Frattaroli
@ 2026-04-17 14:17 ` Michel Dänzer
2026-04-17 14:55 ` Ville Syrjälä
3 siblings, 1 reply; 25+ messages in thread
From: Michel Dänzer @ 2026-04-17 14:17 UTC (permalink / raw)
To: Ville Syrjälä
Cc: Julian Orth, Nicolas Frattaroli, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On 4/17/26 12:57, Ville Syrjälä wrote:
> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
>> On 4/16/26 15:16, Julian Orth wrote:
>>> On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
>>> <nicolas.frattaroli@collabora.com> wrote:
>>>> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
>>>>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
>>>>> <nicolas.frattaroli@collabora.com> wrote:
>>>>>>
>>>>>> This series addresses a shortcoming whereby a hot plug event is sent
>>>>>> without it being passed the actual connector that caused it. This takes
>>>>>> into consideration both the polling path and the HPD (Hot Plug Detect)
>>>>>> path. It also adds support for the vkms driver (using ConfigFS) for
>>>>>> propagating the connector ID when changing the connector's status.
>>>>>>
>>>>>> The motivation is that user-space applications such as Weston would
>>>>>> previously receive non-connector-specific hotplug events, and then have
>>>>>> to figure out themselves which connector needs to have a modeset
>>>>>> executed on. This notably did not work when the hotplug events came in
>>>>>> too fast, resulting in Weston missing an on-off-on transition of a
>>>>>> connector, seeing that its state was unchanged from "on" so can't be the
>>>>>> one that was hotplugged, and skipping reinitialising it as it looks
>>>>>> through the other connectors that could've caused it.
>>>>>
>>>>> Have you considered adding a u64 serial number as a DRM connector
>>>>> property that is incremented every time the connector changes in some
>>>>> way? Userspace could then check this serial number to see if the
>>>>> connector has changed since the last time it queried the serial
>>>>> number.
>>>>
>>>> The connector internally already has an epoch_counter member which
>>>> could be used for this. However, for the particular thing this
>>>> series fixes, I don't think exposing it through the uAPI is necessary
>>>> or desirable. Sending hotplug events specific to the connector does
>>>> not need any additional handling on the userspace side as long as it
>>>> already listens to the per-connector hotplug events in order to
>>>> avoid the pitfall described in the cover letter.
>>>
>>> I currently do not handle per-connector hotplug events. Instead,
>>> whenever I get a UDEV change event for a device, I re-fetch the entire
>>> kernel state for the device. That is
>>>
>>> - DRM_IOCTL_MODE_GETRESOURCES
>>> - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
>>> - DRM_IOCTL_MODE_GETCONNECTOR for each connector
>>> - DRM_IOCTL_MODE_GETPROPERTY for each connector property
>>> - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
>>>
>>> Once I have the new state, I compare it against the desired compositor
>>> state and perform a modeset if necessary.
>>
>> mutter is doing something similar as well.
>>
>>
>> Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
>>
>> The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
>
> GPU reset should relight the display on its own really. That's what
> i915 does, albeit somewhat badly at the moment.
>
>> A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
>
> There's nothing preventing the kernel from doing extra atomic
> commits whenever it wants.
I made that same argument at first.
Then it occurred to me the kernel-internal atomic modeset commit could cause spurious EBUSY failures for user-space atomic commits overlapping with it.
> But if you want to punt the thing to userspace then the kernel must set the link-status
> property to bad, and then fire the hotplug uevent.
I later suggested using the link-status property for this as well.
Checking with others on IRC and reading documentation / code though, I realized my recollection of its semantics was wrong, it actually doesn't look suitable for this. In particular, the expected user space response to link-status bad is to set a *different* mode (since the mode may be relevant for the link failure), not the same one which was already set.
>> P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
>
> The actual argument is that you should not defer the hotplug
> handling when things get disconnected, mainly because of crap type-c
> firmware.
>
> I think the userspace behaviour there was that you get a disconnected,
> defer processing it, and then you get a reconnect, and then decide that
> nothing actually changed and a modeset is not needed after all. That is
> not correct IMO. Clearly a ->disconnect->reconnect should count as a
> change in the connector's state, and a full modeset is thus required.
While that makes sense to me in principle, as Julian pointed out, the kernel can't rely on user space seeing the intermittent disconnected state.
Seems like another argument for a serial number property. That would make the need for a modeset unambiguous.
Note that the referenced issue is about a different scenario though:
1. mutter does modeset for DPMS off
2. hotplug events during DPMS off (presumably triggered by the monitor scanning its inputs)
3. mutter does modeset for DPMS on
The monitor stays off after step 3. The argument in the comment I referenced is that mutter should repeat step 1 before step 3.
--
Earthling Michel Dänzer \ GNOME / Xwayland / Mesa developer
https://redhat.com \ Libre software enthusiast
^ permalink raw reply [flat|nested] 25+ messages in thread* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 14:17 ` Michel Dänzer
@ 2026-04-17 14:55 ` Ville Syrjälä
2026-04-17 16:40 ` Michel Dänzer
0 siblings, 1 reply; 25+ messages in thread
From: Ville Syrjälä @ 2026-04-17 14:55 UTC (permalink / raw)
To: Michel Dänzer
Cc: Julian Orth, Nicolas Frattaroli, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 04:17:45PM +0200, Michel Dänzer wrote:
> On 4/17/26 12:57, Ville Syrjälä wrote:
> > On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> >> On 4/16/26 15:16, Julian Orth wrote:
> >>> On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> >>> <nicolas.frattaroli@collabora.com> wrote:
> >>>> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> >>>>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> >>>>> <nicolas.frattaroli@collabora.com> wrote:
> >>>>>>
> >>>>>> This series addresses a shortcoming whereby a hot plug event is sent
> >>>>>> without it being passed the actual connector that caused it. This takes
> >>>>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> >>>>>> path. It also adds support for the vkms driver (using ConfigFS) for
> >>>>>> propagating the connector ID when changing the connector's status.
> >>>>>>
> >>>>>> The motivation is that user-space applications such as Weston would
> >>>>>> previously receive non-connector-specific hotplug events, and then have
> >>>>>> to figure out themselves which connector needs to have a modeset
> >>>>>> executed on. This notably did not work when the hotplug events came in
> >>>>>> too fast, resulting in Weston missing an on-off-on transition of a
> >>>>>> connector, seeing that its state was unchanged from "on" so can't be the
> >>>>>> one that was hotplugged, and skipping reinitialising it as it looks
> >>>>>> through the other connectors that could've caused it.
> >>>>>
> >>>>> Have you considered adding a u64 serial number as a DRM connector
> >>>>> property that is incremented every time the connector changes in some
> >>>>> way? Userspace could then check this serial number to see if the
> >>>>> connector has changed since the last time it queried the serial
> >>>>> number.
> >>>>
> >>>> The connector internally already has an epoch_counter member which
> >>>> could be used for this. However, for the particular thing this
> >>>> series fixes, I don't think exposing it through the uAPI is necessary
> >>>> or desirable. Sending hotplug events specific to the connector does
> >>>> not need any additional handling on the userspace side as long as it
> >>>> already listens to the per-connector hotplug events in order to
> >>>> avoid the pitfall described in the cover letter.
> >>>
> >>> I currently do not handle per-connector hotplug events. Instead,
> >>> whenever I get a UDEV change event for a device, I re-fetch the entire
> >>> kernel state for the device. That is
> >>>
> >>> - DRM_IOCTL_MODE_GETRESOURCES
> >>> - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> >>> - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> >>> - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> >>> - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> >>>
> >>> Once I have the new state, I compare it against the desired compositor
> >>> state and perform a modeset if necessary.
> >>
> >> mutter is doing something similar as well.
> >>
> >>
> >> Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> >>
> >> The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> >
> > GPU reset should relight the display on its own really. That's what
> > i915 does, albeit somewhat badly at the moment.
> >
> >> A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> >
> > There's nothing preventing the kernel from doing extra atomic
> > commits whenever it wants.
>
> I made that same argument at first.
>
> Then it occurred to me the kernel-internal atomic modeset commit could cause spurious EBUSY failures for user-space atomic commits overlapping with it.
Not so. Everything just gets blocked on the kms mutexes and the -EBUSY
stuff never even comes into play.
>
>
> > But if you want to punt the thing to userspace then the kernel must set the link-status
> > property to bad, and then fire the hotplug uevent.
>
> I later suggested using the link-status property for this as well.
>
> Checking with others on IRC and reading documentation / code though, I realized my recollection of its semantics was wrong, it actually doesn't look suitable for this. In particular, the expected user space response to link-status bad is to set a *different* mode (since the mode may be relevant for the link failure), not the same one which was already set.
The expected response it to retry with the same mode, at least if
that mode is still on the connector's mode list. If the mode got
pruned then there's perhaps no point in trying it again.
That's if the mode was on the connector's mode list to begin with.
If it was never there then just blindly retrying it again seems
like the correct option to me.
>
>
> >> P.S. https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/14420#note_2984697 even argued that two modesets are required after a hotplug event, one which turns things off and another one which turns them on again. I don't agree with that though, a single modeset should suffice.
> >
> > The actual argument is that you should not defer the hotplug
> > handling when things get disconnected, mainly because of crap type-c
> > firmware.
> >
> > I think the userspace behaviour there was that you get a disconnected,
> > defer processing it, and then you get a reconnect, and then decide that
> > nothing actually changed and a modeset is not needed after all. That is
> > not correct IMO. Clearly a ->disconnect->reconnect should count as a
> > change in the connector's state, and a full modeset is thus required.
>
> While that makes sense to me in principle, as Julian pointed out, the kernel can't rely on user space seeing the intermittent disconnected state.
>
> Seems like another argument for a serial number property. That would make the need for a modeset unambiguous.
I guess it could at least optimize the "has this connector changed?"
check for the single uevent without connector IDs case.
And I suppose it could avoid race conditions with switching drm masters
when the new master can't be sure it's seen all the uevents. Another
thing that would avoid the issues here is if each new master was
required to do one full modeset on everything. But maybe that was never
a rule?
But for the case where the kernel only sends the per-connector uevents
and the drm master sees all of them the serial number would do nothing.
In that case the kernel wouldn't even send the uevent unless the
serial number has changed.
>
>
> Note that the referenced issue is about a different scenario though:
>
> 1. mutter does modeset for DPMS off
> 2. hotplug events during DPMS off (presumably triggered by the monitor scanning its inputs)
> 3. mutter does modeset for DPMS on
>
> The monitor stays off after step 3. The argument in the comment I referenced is that mutter should repeat step 1 before step 3.
That can't be it. We don't do anything if you try to disable an already
disabled output. And the analysis of the logs indicated that the disable
(DPMS off) was completely missing.
--
Ville Syrjälä
Intel
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 14:55 ` Ville Syrjälä
@ 2026-04-17 16:40 ` Michel Dänzer
2026-04-17 18:50 ` Ville Syrjälä
0 siblings, 1 reply; 25+ messages in thread
From: Michel Dänzer @ 2026-04-17 16:40 UTC (permalink / raw)
To: Ville Syrjälä
Cc: Julian Orth, Nicolas Frattaroli, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On 4/17/26 16:55, Ville Syrjälä wrote:
> On Fri, Apr 17, 2026 at 04:17:45PM +0200, Michel Dänzer wrote:
>> On 4/17/26 12:57, Ville Syrjälä wrote:
>>> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
>>>> On 4/16/26 15:16, Julian Orth wrote:
>>>>> On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
>>>>> <nicolas.frattaroli@collabora.com> wrote:
>>>>>> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
>>>>>>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
>>>>>>> <nicolas.frattaroli@collabora.com> wrote:
>>>>>>>>
>>>>>>>> This series addresses a shortcoming whereby a hot plug event is sent
>>>>>>>> without it being passed the actual connector that caused it. This takes
>>>>>>>> into consideration both the polling path and the HPD (Hot Plug Detect)
>>>>>>>> path. It also adds support for the vkms driver (using ConfigFS) for
>>>>>>>> propagating the connector ID when changing the connector's status.
>>>>>>>>
>>>>>>>> The motivation is that user-space applications such as Weston would
>>>>>>>> previously receive non-connector-specific hotplug events, and then have
>>>>>>>> to figure out themselves which connector needs to have a modeset
>>>>>>>> executed on. This notably did not work when the hotplug events came in
>>>>>>>> too fast, resulting in Weston missing an on-off-on transition of a
>>>>>>>> connector, seeing that its state was unchanged from "on" so can't be the
>>>>>>>> one that was hotplugged, and skipping reinitialising it as it looks
>>>>>>>> through the other connectors that could've caused it.
>>>>>>>
>>>>>>> Have you considered adding a u64 serial number as a DRM connector
>>>>>>> property that is incremented every time the connector changes in some
>>>>>>> way? Userspace could then check this serial number to see if the
>>>>>>> connector has changed since the last time it queried the serial
>>>>>>> number.
>>>>>>
>>>>>> The connector internally already has an epoch_counter member which
>>>>>> could be used for this. However, for the particular thing this
>>>>>> series fixes, I don't think exposing it through the uAPI is necessary
>>>>>> or desirable. Sending hotplug events specific to the connector does
>>>>>> not need any additional handling on the userspace side as long as it
>>>>>> already listens to the per-connector hotplug events in order to
>>>>>> avoid the pitfall described in the cover letter.
>>>>>
>>>>> I currently do not handle per-connector hotplug events. Instead,
>>>>> whenever I get a UDEV change event for a device, I re-fetch the entire
>>>>> kernel state for the device. That is
>>>>>
>>>>> - DRM_IOCTL_MODE_GETRESOURCES
>>>>> - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
>>>>> - DRM_IOCTL_MODE_GETCONNECTOR for each connector
>>>>> - DRM_IOCTL_MODE_GETPROPERTY for each connector property
>>>>> - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
>>>>>
>>>>> Once I have the new state, I compare it against the desired compositor
>>>>> state and perform a modeset if necessary.
>>>>
>>>> mutter is doing something similar as well.
>>>>
>>>>
>>>> Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
>>>>
>>>> The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
>>>
>>> GPU reset should relight the display on its own really. That's what
>>> i915 does, albeit somewhat badly at the moment.
>>>
>>>> A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
>>>
>>> There's nothing preventing the kernel from doing extra atomic
>>> commits whenever it wants.
>>
>> I made that same argument at first.
>>
>> Then it occurred to me the kernel-internal atomic modeset commit could cause spurious EBUSY failures for user-space atomic commits overlapping with it.
>
> Not so. Everything just gets blocked on the kms mutexes and the -EBUSY
> stuff never even comes into play.
Makes things easier in that case. I followed up accordingly in the other thread.
>>> But if you want to punt the thing to userspace then the kernel must set the link-status
>>> property to bad, and then fire the hotplug uevent.
>>
>> I later suggested using the link-status property for this as well.
>>
>> Checking with others on IRC and reading documentation / code though, I realized my recollection of its semantics was wrong, it actually doesn't look suitable for this. In particular, the expected user space response to link-status bad is to set a *different* mode (since the mode may be relevant for the link failure), not the same one which was already set.
>
> The expected response it to retry with the same mode, at least if
> that mode is still on the connector's mode list. If the mode got
> pruned then there's perhaps no point in trying it again.
On re-reading the link-status documentation in drivers/gpu/drm/drm_connector.c, I agree.
I'm afraid you're optimistic in terms of user space support for link-status though.
Neither mutter nor weston even look at it yet.
KWin reacts to link-status bad with a modeset, not sure it handles the current mode disappearing though. Seems good enough for the scenarios we're discussing here.
wlroots seems to handle it properly.
(Any compositors which don't handle it yet need to be fixed, just pointing out it might not make a difference with existing releases of some of them)
>>> I think the userspace behaviour there was that you get a disconnected,
>>> defer processing it, and then you get a reconnect, and then decide that
>>> nothing actually changed and a modeset is not needed after all. That is
>>> not correct IMO. Clearly a ->disconnect->reconnect should count as a
>>> change in the connector's state, and a full modeset is thus required.
>>
>> While that makes sense to me in principle, as Julian pointed out, the kernel can't rely on user space seeing the intermittent disconnected state.
>>
>> Seems like another argument for a serial number property. That would make the need for a modeset unambiguous.
>
> [...]
>
> But for the case where the kernel only sends the per-connector uevents
> and the drm master sees all of them the serial number would do nothing.
> In that case the kernel wouldn't even send the uevent unless the
> serial number has changed.
The serial number is still necessary in cases where status flip-flops connected → disconnected → connected, user space never sees the disconnected state though.
>> Note that the referenced issue is about a different scenario though:
>>
>> 1. mutter does modeset for DPMS off
>> 2. hotplug events during DPMS off (presumably triggered by the monitor scanning its inputs)
>> 3. mutter does modeset for DPMS on
>>
>> The monitor stays off after step 3. The argument in the comment I referenced is that mutter should repeat step 1 before step 3.
>
> That can't be it. We don't do anything if you try to disable an already
> disabled output. And the analysis of the logs indicated that the disable
> (DPMS off) was completely missing.
The mutter debug log in https://gitlab.gnome.org/GNOME/mutter/-/issues/4145#note_2457199 disagrees.
--
Earthling Michel Dänzer \ GNOME / Xwayland / Mesa developer
https://redhat.com \ Libre software enthusiast
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH RESEND v7 0/2] Pass down hot-plug CONNECTOR ID to user-space
2026-04-17 16:40 ` Michel Dänzer
@ 2026-04-17 18:50 ` Ville Syrjälä
0 siblings, 0 replies; 25+ messages in thread
From: Ville Syrjälä @ 2026-04-17 18:50 UTC (permalink / raw)
To: Michel Dänzer
Cc: Julian Orth, Nicolas Frattaroli, Maarten Lankhorst, Maxime Ripard,
Thomas Zimmermann, David Airlie, Simona Vetter, Louis Chauvet,
Haneen Mohammed, Melissa Wen, Daniel Stone, Ian Forbes,
Dmitry Baryshkov, dri-devel, linux-kernel, kernel, wayland-devel,
Marius Vlad, Imre Deak
On Fri, Apr 17, 2026 at 06:40:51PM +0200, Michel Dänzer wrote:
> On 4/17/26 16:55, Ville Syrjälä wrote:
> > On Fri, Apr 17, 2026 at 04:17:45PM +0200, Michel Dänzer wrote:
> >> On 4/17/26 12:57, Ville Syrjälä wrote:
> >>> On Fri, Apr 17, 2026 at 09:49:36AM +0200, Michel Dänzer wrote:
> >>>> On 4/16/26 15:16, Julian Orth wrote:
> >>>>> On Thu, Apr 16, 2026 at 9:46 AM Nicolas Frattaroli
> >>>>> <nicolas.frattaroli@collabora.com> wrote:
> >>>>>> On Wednesday, 15 April 2026 20:57:53 Central European Summer Time Julian Orth wrote:
> >>>>>>> On Wed, Apr 15, 2026 at 8:19 PM Nicolas Frattaroli
> >>>>>>> <nicolas.frattaroli@collabora.com> wrote:
> >>>>>>>>
> >>>>>>>> This series addresses a shortcoming whereby a hot plug event is sent
> >>>>>>>> without it being passed the actual connector that caused it. This takes
> >>>>>>>> into consideration both the polling path and the HPD (Hot Plug Detect)
> >>>>>>>> path. It also adds support for the vkms driver (using ConfigFS) for
> >>>>>>>> propagating the connector ID when changing the connector's status.
> >>>>>>>>
> >>>>>>>> The motivation is that user-space applications such as Weston would
> >>>>>>>> previously receive non-connector-specific hotplug events, and then have
> >>>>>>>> to figure out themselves which connector needs to have a modeset
> >>>>>>>> executed on. This notably did not work when the hotplug events came in
> >>>>>>>> too fast, resulting in Weston missing an on-off-on transition of a
> >>>>>>>> connector, seeing that its state was unchanged from "on" so can't be the
> >>>>>>>> one that was hotplugged, and skipping reinitialising it as it looks
> >>>>>>>> through the other connectors that could've caused it.
> >>>>>>>
> >>>>>>> Have you considered adding a u64 serial number as a DRM connector
> >>>>>>> property that is incremented every time the connector changes in some
> >>>>>>> way? Userspace could then check this serial number to see if the
> >>>>>>> connector has changed since the last time it queried the serial
> >>>>>>> number.
> >>>>>>
> >>>>>> The connector internally already has an epoch_counter member which
> >>>>>> could be used for this. However, for the particular thing this
> >>>>>> series fixes, I don't think exposing it through the uAPI is necessary
> >>>>>> or desirable. Sending hotplug events specific to the connector does
> >>>>>> not need any additional handling on the userspace side as long as it
> >>>>>> already listens to the per-connector hotplug events in order to
> >>>>>> avoid the pitfall described in the cover letter.
> >>>>>
> >>>>> I currently do not handle per-connector hotplug events. Instead,
> >>>>> whenever I get a UDEV change event for a device, I re-fetch the entire
> >>>>> kernel state for the device. That is
> >>>>>
> >>>>> - DRM_IOCTL_MODE_GETRESOURCES
> >>>>> - DRM_IOCTL_MODE_OBJ_GETPROPERTIES for each connector, crtc, plane
> >>>>> - DRM_IOCTL_MODE_GETCONNECTOR for each connector
> >>>>> - DRM_IOCTL_MODE_GETPROPERTY for each connector property
> >>>>> - DRM_IOCTL_MODE_GETPROPBLOB for the EDID
> >>>>>
> >>>>> Once I have the new state, I compare it against the desired compositor
> >>>>> state and perform a modeset if necessary.
> >>>>
> >>>> mutter is doing something similar as well.
> >>>>
> >>>>
> >>>> Note that some are arguing a modeset is always required after a hotplug event, even if the state hasn't changed.
> >>>>
> >>>> The most convincing argument I've seen is the scenario of a GPU reset, after which a modeset is required to light up the displays again.
> >>>
> >>> GPU reset should relight the display on its own really. That's what
> >>> i915 does, albeit somewhat badly at the moment.
> >>>
> >>>> A hotplug event seems the only mechanism available for the kernel to request a modeset from the compositor. (The kernel may not be able to reliably do the modeset on its own, e.g. due to interactions with user-space atomic commits)
> >>>
> >>> There's nothing preventing the kernel from doing extra atomic
> >>> commits whenever it wants.
> >>
> >> I made that same argument at first.
> >>
> >> Then it occurred to me the kernel-internal atomic modeset commit could cause spurious EBUSY failures for user-space atomic commits overlapping with it.
> >
> > Not so. Everything just gets blocked on the kms mutexes and the -EBUSY
> > stuff never even comes into play.
>
> Makes things easier in that case. I followed up accordingly in the other thread.
>
>
> >>> But if you want to punt the thing to userspace then the kernel must set the link-status
> >>> property to bad, and then fire the hotplug uevent.
> >>
> >> I later suggested using the link-status property for this as well.
> >>
> >> Checking with others on IRC and reading documentation / code though, I realized my recollection of its semantics was wrong, it actually doesn't look suitable for this. In particular, the expected user space response to link-status bad is to set a *different* mode (since the mode may be relevant for the link failure), not the same one which was already set.
> >
> > The expected response it to retry with the same mode, at least if
> > that mode is still on the connector's mode list. If the mode got
> > pruned then there's perhaps no point in trying it again.
>
> On re-reading the link-status documentation in drivers/gpu/drm/drm_connector.c, I agree.
>
> I'm afraid you're optimistic in terms of user space support for link-status though.
>
> Neither mutter nor weston even look at it yet.
>
> KWin reacts to link-status bad with a modeset, not sure it handles the current mode disappearing though. Seems good enough for the scenarios we're discussing here.
>
> wlroots seems to handle it properly.
>
> (Any compositors which don't handle it yet need to be fixed, just pointing out it might not make a difference with existing releases of some of them)
>
>
> >>> I think the userspace behaviour there was that you get a disconnected,
> >>> defer processing it, and then you get a reconnect, and then decide that
> >>> nothing actually changed and a modeset is not needed after all. That is
> >>> not correct IMO. Clearly a ->disconnect->reconnect should count as a
> >>> change in the connector's state, and a full modeset is thus required.
> >>
> >> While that makes sense to me in principle, as Julian pointed out, the kernel can't rely on user space seeing the intermittent disconnected state.
> >>
> >> Seems like another argument for a serial number property. That would make the need for a modeset unambiguous.
> >
> > [...]
> >
> > But for the case where the kernel only sends the per-connector uevents
> > and the drm master sees all of them the serial number would do nothing.
> > In that case the kernel wouldn't even send the uevent unless the
> > serial number has changed.
>
> The serial number is still necessary in cases where status flip-flops connected → disconnected → connected, user space never sees the disconnected state though.
I suppose that could happen if the flip flop is fast enough that the
status has changed back to connected by the time userspace gets to
process the first uevent. Although the fact that the uevent was sent
in the first place (assuming it's the per-connector one) would imply
that something did in fact change. But I do agree that having a clear
signal in the form of the changed serial number would avoid a lot of
weird guesswork in userspace.
> >> Note that the referenced issue is about a different scenario though:
> >>
> >> 1. mutter does modeset for DPMS off
> >> 2. hotplug events during DPMS off (presumably triggered by the monitor scanning its inputs)
> >> 3. mutter does modeset for DPMS on
> >>
> >> The monitor stays off after step 3. The argument in the comment I referenced is that mutter should repeat step 1 before step 3.
> >
> > That can't be it. We don't do anything if you try to disable an already
> > disabled output. And the analysis of the logs indicated that the disable
> > (DPMS off) was completely missing.
> The mutter debug log in https://gitlab.gnome.org/GNOME/mutter/-/issues/4145#note_2457199 disagrees.
I only see two modeset commits in that log. Both with CRTC 134 being
enabled with connector 250. So far I can't find any DPMS off in there.
Am I just blind?
--
Ville Syrjälä
Intel
^ permalink raw reply [flat|nested] 25+ messages in thread