Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
Thanks, Thomas
On Fri, 4 Nov 2011, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
Thanks, Thomas
FWIW, Acked-by: Ilija Hadzic ihadzic@research.bell-labs.com I have a few places where I could use such an ioctl.
BTW, Thomas, in the above "since the cursor data is sent though the command stream", did you mean "since the cursor data is *not* sent though the command stream". If it was sent, through command stream, then CS ioctl would know when the cursor changes.
My understanding is that the cursor data are mmap-ed letting userland poke it at will (so the case when an "hourglass" changes into "arrow" is particularly hard to know that it happened).
-- Ilija
On 11/04/2011 03:36 PM, Ilija Hadzic wrote:
On Fri, 4 Nov 2011, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
Thanks, Thomas
FWIW, Acked-by: Ilija Hadzic ihadzic@research.bell-labs.com I have a few places where I could use such an ioctl.
Well, the idea was to reuse the DRM_IOCTL_MODE_CURSOR ictl that sets the cursor BO, and require that it is re-emitted once the cursor image has changed, since I guess that's what many X servers do anyway.
BTW, Thomas, in the above "since the cursor data is sent though the command stream", did you mean "since the cursor data is *not* sent though the command stream". If it was sent, through command stream, then CS ioctl would know when the cursor changes.
I meant the device interface, not the drm interface. So when we have a new cursor image, we need to collect the scanline data an send it through the device command stream. We don't export that possibility in the DRM interface.
My understanding is that the cursor data are mmap-ed letting userland poke it at will (so the case when an "hourglass" changes into "arrow" is particularly hard to know that it happened).
Yes. Hardware that scan out directly from the cursor BO don't really need to know, but we do.
-- Ilija
Thanks, /Thomas
On Fri, Nov 04, 2011 at 12:59:59PM +0100, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
On i915 we need the cursor in physical memory for some (old) platforms, which is seperate storage from the bo backing storage. So we have the same problem. We've solved it by intercepting pwrite ioctl calls and demanding that userspace only uses these for cursor updates. Is there a special reason you can't use such a driver-specific trick? -Daniel
On 11/04/2011 04:34 PM, Daniel Vetter wrote:
On Fri, Nov 04, 2011 at 12:59:59PM +0100, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
On i915 we need the cursor in physical memory for some (old) platforms, which is seperate storage from the bo backing storage. So we have the same problem. We've solved it by intercepting pwrite ioctl calls and demanding that userspace only uses these for cursor updates. Is there a special reason you can't use such a driver-specific trick? -Daniel
We have something similar in use today: We snoop DMAs to hardware cursor surfaces, but this gets a bit nasty when apps start to do hardware render to cursor surfaces, and we simply ignore that today.
Furthermore, maps rather than pwrites are the common usage-pattern for buffer-backed cursors on vmwgfx, and while it's possible to dirty those buffers based on page-faults, like we do with fb surfaces, we'd rather avoid having to implement and maintain that.
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
/Thomas
On Fri, Nov 4, 2011 at 11:30 PM, Thomas Hellstrom thellstrom@vmware.com wrote:
On 11/04/2011 04:34 PM, Daniel Vetter wrote:
On Fri, Nov 04, 2011 at 12:59:59PM +0100, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
On i915 we need the cursor in physical memory for some (old) platforms, which is seperate storage from the bo backing storage. So we have the same problem. We've solved it by intercepting pwrite ioctl calls and demanding that userspace only uses these for cursor updates. Is there a special reason you can't use such a driver-specific trick? -Daniel
We have something similar in use today: We snoop DMAs to hardware cursor surfaces, but this gets a bit nasty when apps start to do hardware render to cursor surfaces, and we simply ignore that today.
Furthermore, maps rather than pwrites are the common usage-pattern for buffer-backed cursors on vmwgfx, and while it's possible to dirty those buffers based on page-faults, like we do with fb surfaces, we'd rather avoid having to implement and maintain that.
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
/Thomas
dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
As far as i know nouveau keeps an internal buffer object for the cursor and relies purely on the ioctl to copy the cursor into the actual cursor memory area. So why do you need tricks?
On 11/04/2011 11:49 PM, Maarten Maathuis wrote:
On Fri, Nov 4, 2011 at 11:30 PM, Thomas Hellstromthellstrom@vmware.com wrote:
On 11/04/2011 04:34 PM, Daniel Vetter wrote:
On Fri, Nov 04, 2011 at 12:59:59PM +0100, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
On i915 we need the cursor in physical memory for some (old) platforms, which is seperate storage from the bo backing storage. So we have the same problem. We've solved it by intercepting pwrite ioctl calls and demanding that userspace only uses these for cursor updates. Is there a special reason you can't use such a driver-specific trick? -Daniel
We have something similar in use today: We snoop DMAs to hardware cursor surfaces, but this gets a bit nasty when apps start to do hardware render to cursor surfaces, and we simply ignore that today.
Furthermore, maps rather than pwrites are the common usage-pattern for buffer-backed cursors on vmwgfx, and while it's possible to dirty those buffers based on page-faults, like we do with fb surfaces, we'd rather avoid having to implement and maintain that.
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
/Thomas
dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
As far as i know nouveau keeps an internal buffer object for the cursor and relies purely on the ioctl to copy the cursor into the actual cursor memory area. So why do you need tricks?
Because I have a hard time judging whether the Intel implementation (needs tricks) or the Nouveau implementation (doesn't need tricks) should define the semantics of the ioctl, although I would prefer we could all agree on the "doesn't need tricks" semantics and put that down in writing in the drm_mode.h header file.
/Thomas
On Fri, Nov 4, 2011 at 11:59 PM, Thomas Hellstrom thellstrom@vmware.com wrote:
On 11/04/2011 11:49 PM, Maarten Maathuis wrote:
On Fri, Nov 4, 2011 at 11:30 PM, Thomas Hellstromthellstrom@vmware.com wrote:
On 11/04/2011 04:34 PM, Daniel Vetter wrote:
On Fri, Nov 04, 2011 at 12:59:59PM +0100, Thomas Hellstrom wrote:
Hi.
I have a question about the semantics of the DRM_IOCTL_MODE_CURSOR iotcl:
Some hardware (vmware's virtual in particular) may not be able to pick up the changes from a bo directly, since the cursor data is sent though the command stream. Hence we need a notification when the cursor image has changed.
Could we *require* that a cursor image change needs to be followed by an ioctl call with the flag DRM_MODE_CURSOR_BO?
On i915 we need the cursor in physical memory for some (old) platforms, which is seperate storage from the bo backing storage. So we have the same problem. We've solved it by intercepting pwrite ioctl calls and demanding that userspace only uses these for cursor updates. Is there a special reason you can't use such a driver-specific trick? -Daniel
We have something similar in use today: We snoop DMAs to hardware cursor surfaces, but this gets a bit nasty when apps start to do hardware render to cursor surfaces, and we simply ignore that today.
Furthermore, maps rather than pwrites are the common usage-pattern for buffer-backed cursors on vmwgfx, and while it's possible to dirty those buffers based on page-faults, like we do with fb surfaces, we'd rather avoid having to implement and maintain that.
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
/Thomas
dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
As far as i know nouveau keeps an internal buffer object for the cursor and relies purely on the ioctl to copy the cursor into the actual cursor memory area. So why do you need tricks?
Because I have a hard time judging whether the Intel implementation (needs tricks) or the Nouveau implementation (doesn't need tricks) should define the semantics of the ioctl, although I would prefer we could all agree on the "doesn't need tricks" semantics and put that down in writing in the drm_mode.h header file.
/Thomas
I see your point, i certainly prefer the "doesn't need tricks" approach. Maybe allow for some kind of hybrid approach, require the ioctl for every cursor change, but allow drivers to do a kind of lazy implementation that simply attaches the buffer to the hardware and let it handle it. If they want to avoid copying overhead or stuff like that. This would impose some restrictions on what you can do with a cursor buffer after calling the ioctl, specifically you would be required to leave the content alone, except for cursor updates.
On Fri, Nov 04, 2011 at 11:30:06PM +0100, Thomas Hellstrom wrote:
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
We don't handle really hanlde rendering to cursor objects. I think the "require a set_cursor after every cursor bo change" semantic is good, I've just feared that when only vmgfx needs this, no generic kms user will actually implement it. But nouveau seems to require this too, so I think at least for this case reality (and generic kms clients) will play along. -Daniel
On 11/05/2011 05:10 PM, Daniel Vetter wrote:
On Fri, Nov 04, 2011 at 11:30:06PM +0100, Thomas Hellstrom wrote:
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
We don't handle really hanlde rendering to cursor objects. I think the "require a set_cursor after every cursor bo change" semantic is good, I've just feared that when only vmgfx needs this, no generic kms user will actually implement it. But nouveau seems to require this too, so I think at least for this case reality (and generic kms clients) will play along. -Daniel
Great. I'll cook up a header file patch with some extra documentation.
/Thomas
On Fri, Nov 04, 2011 at 11:30:06PM +0100, Thomas Hellstrom wrote:
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
We don't handle really hanlde rendering to cursor objects. I think the "require a set_cursor after every cursor bo change" semantic is good, I've just feared that when only vmgfx needs this, no generic kms user will actually implement it. But nouveau seems to require this too, so I think at least for this case reality (and generic kms clients) will play along.
I know this is a bit off topic but I see two approaches to the cursor api. One is the nouveau method which is
... drm_gem_object_lookup internal_mmap gem object copy to another internally allocated buffer objec unmap gem object drm_gem_object_unreference_unlocked ...
Or the radeon approach
... drm_gem_object_lookup gem_object_pin program cursor location gem_object_unpin drm_gem_object_unreference_unlocked ...
What are the pros and cons to each method?
On Fri, Nov 11, 2011 at 7:17 PM, James Simmons jsimmons@infradead.org wrote:
On Fri, Nov 04, 2011 at 11:30:06PM +0100, Thomas Hellstrom wrote:
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
We don't handle really hanlde rendering to cursor objects. I think the "require a set_cursor after every cursor bo change" semantic is good, I've just feared that when only vmgfx needs this, no generic kms user will actually implement it. But nouveau seems to require this too, so I think at least for this case reality (and generic kms clients) will play along.
I know this is a bit off topic but I see two approaches to the cursor api. One is the nouveau method which is
... drm_gem_object_lookup internal_mmap gem object copy to another internally allocated buffer objec unmap gem object drm_gem_object_unreference_unlocked ...
Or the radeon approach
... drm_gem_object_lookup gem_object_pin program cursor location gem_object_unpin drm_gem_object_unreference_unlocked ...
What are the pros and cons to each method? _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
Copying means you can convert the cursor to another format if needed.
On Fri, Nov 11, 2011 at 8:10 PM, Maarten Maathuis madman2003@gmail.com wrote:
On Fri, Nov 11, 2011 at 7:17 PM, James Simmons jsimmons@infradead.org wrote:
On Fri, Nov 04, 2011 at 11:30:06PM +0100, Thomas Hellstrom wrote:
I'm not sure whether / how you handle the case of hardware render to cursor surfaces on i915, but it seems to me like if a lot of drivers need to implement driver specific "tricks" to meet the semantics of a generic interface, we should perhaps consider specifying those semantics in a way that helps avoid driver-specific workarounds?
We don't handle really hanlde rendering to cursor objects. I think the "require a set_cursor after every cursor bo change" semantic is good, I've just feared that when only vmgfx needs this, no generic kms user will actually implement it. But nouveau seems to require this too, so I think at least for this case reality (and generic kms clients) will play along.
I know this is a bit off topic but I see two approaches to the cursor api. One is the nouveau method which is
... drm_gem_object_lookup internal_mmap gem object copy to another internally allocated buffer objec unmap gem object drm_gem_object_unreference_unlocked ...
Or the radeon approach
... drm_gem_object_lookup gem_object_pin program cursor location gem_object_unpin drm_gem_object_unreference_unlocked ...
What are the pros and cons to each method? _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
Copying means you can convert the cursor to another format if needed.
-- Far away from the primal instinct, the song seems to fade away, the river get wider between your thoughts and the things we do and say.
Also no worries about strange side effects if people decide to render to the cursor. Although i don't know if this is really an issue.
dri-devel@lists.freedesktop.org
-
Daniel Vetter -
Ilija Hadzic -
James Simmons -
Maarten Maathuis -
Thomas Hellstrom