Re: Split render/display SoCs, Mesa's renderonly, and Wayland dmabuf hints
Thanks for the feedback! Replying to both Daniels below. On Tuesday, April 20th, 2021 at 1:14 PM, Daniel Vetter wrote: > On Tue, Apr 20, 2021 at 12:18 PM Daniel Stone wrote: > > >> There are a few solutions: > >> > >> 1. Require compositors to discover the render device by trying to import > >>a buffer. For each available render device, the compositor would > >>allocate a buffer, export it as a DMA-BUF, import it to the > >>display-only device, then try to drmModeAddFB. > > > > > > I don't think this is actually tractable? Assuming that 'allocate a buffer' > > means 'obtain a gbm_device for the render node directly and allocate a > > gbm_bo from it', even with compatible formats and modifiers this will fail > > for more restrictive display hardware. imx-drm and pl111 (combined with vc4 > > on some Raspberry Pis) will fail this, since they'll take different > > allocation paths when they're bound through kmsro vs. directly, accounting > > for things like contiguous allocation. So we'd get false negatives on at > > least some platforms. Right. It would work for drivers using renderonly_create_gpu_import_for_resource, but wouldn't work for drivers using renderonly_create_kms_dumb_buffer_for_resource. Doing the reverse (creating a buffer on the display-only device, import it on the render-only device) wouldn't work either, since some drivers will perform DMA transfers under-the-hood without telling user-space. So the check would succeed even if the pipeline is inefficient. > >> 2. Allow compositors to query the render device magically opened by > >>kmsro. This could be done either via EGL_EXT_device_drm, or via a > >>new EGL extension. > > > > This would be my strong preference, and I don't entirely understand > > anholt's pushback here. The way I see it, GBM is about allocation for > > scanout, and EGL is about rendering. If, on a split GPU/display system, we > > create a gbm_device from a KMS display-only device node, then creating an > > EGLDisplay from that magically binds us to a completely different DRM GPU > > node, and anything using that EGLDisplay will use that GPU device to > > render. > > > > Being able to discover the GPU device node through the device query is > > really useful, because it tells us exactly what implicit magic EGL did > > under the hood, and about the device that EGL will use. Being able to > > discover the display node is much less useful; it does tell us how GBM will > > allocate buffers, but the user already knows which device is in use because > > they supplied it to GBM. I see the display node as a property of GBM, and > > the GPU node as a property of EGL, even if EGL does do (*waves hands*) > > stuff under the hood to ensure the two are compatible. GBM can be used for non-scanout allocation too. This is useful for clients that want to have more control over the allocation parameters, the swapchain and want to manage DMA-BUFs directly. > > If we had EGL_EXT_explicit_device, things get even more weird, I think; > > would the device query on an EGLDisplay created with a combination of a > > gbm_device native display handle and an explicit EGLDevice handle return > > the scanout device from GBM or the GPU device from EGL? On my reading, I'd > > expect it to be the latter; if the queries returned very different things > > based on whether GPU device selection was implicit (returning the KMS node) > > or explicit (GPU node), that would definitely violate the principle of > > least surprise. Right now as implemented by Mesa, things are pretty weird indeed. EGL_EXT_device_enumeration will only expose a list of one EGLDevice per DRM render node, plus one EGLDevice for software rendering. If you create an EGLDisplay from a gbm_device from a display-only device, then a new EGLDevice for this display-only device gets created and is returned by EGL_EXT_device_query. It would make more sense to me to return the EGLDevice for the render-only device in EGL_EXT_device_query. However that means we need to pass back this information through Mesa's DRI interface and throught the Gallium interface. That sounds pretty annoying, but was what I tried to suggest earlier here: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4178#note_815040 > >> 3. Allow compositors to query the kernel drivers to know which devices > >>are compatible with each other. Some uAPI to query a compatible > >>display device from a render-only device, or vice-versa, has been > >>suggested in the past. > > > > > > What does 'compatible' mean? Would an Intel iGPU and and AMD dGPU be > > compatible with each other? Would a Mali GPU bound to system memory through > > AMBA be as compatible with the display controller as it would with an AMD > > GPU on PCIE? I think a query which only exposed whether or not devices > > could share dmabufs with each other is far too generic to be helpful for > > the actual usecase we have, as well as not being useful enough for other > > usecases
Re: Split render/display SoCs, Mesa's renderonly, and Wayland dmabuf hints
On Tue, Apr 20, 2021 at 3:18 AM Daniel Stone wrote: > > Hi, > > On Mon, 19 Apr 2021 at 13:06, Simon Ser wrote: >> >> I'm working on a Wayland extension [1] that, among other things, allows >> compositors to advertise the preferred device to be used by Wayland >> clients. >> >> In general, compositors will send a render node. However, in the case >> of split render/display SoCs, things get a little bit complicated. >> >> [...] > > > Thanks for the write-up Simon! > >> >> There are a few solutions: >> >> 1. Require compositors to discover the render device by trying to import >>a buffer. For each available render device, the compositor would >>allocate a buffer, export it as a DMA-BUF, import it to the >>display-only device, then try to drmModeAddFB. > > > I don't think this is actually tractable? Assuming that 'allocate a buffer' > means 'obtain a gbm_device for the render node directly and allocate a gbm_bo > from it', even with compatible formats and modifiers this will fail for more > restrictive display hardware. imx-drm and pl111 (combined with vc4 on some > Raspberry Pis) will fail this, since they'll take different allocation paths > when they're bound through kmsro vs. directly, accounting for things like > contiguous allocation. So we'd get false negatives on at least some platforms. > >> >> 2. Allow compositors to query the render device magically opened by >>kmsro. This could be done either via EGL_EXT_device_drm, or via a >>new EGL extension. > > > This would be my strong preference, and I don't entirely understand anholt's > pushback here. The way I see it, GBM is about allocation for scanout, and EGL > is about rendering. If, on a split GPU/display system, we create a gbm_device > from a KMS display-only device node, then creating an EGLDisplay from that > magically binds us to a completely different DRM GPU node, and anything using > that EGLDisplay will use that GPU device to render. > > Being able to discover the GPU device node through the device query is really > useful, because it tells us exactly what implicit magic EGL did under the > hood, and about the device that EGL will use. Being able to discover the > display node is much less useful; it does tell us how GBM will allocate > buffers, but the user already knows which device is in use because they > supplied it to GBM. I see the display node as a property of GBM, and the GPU > node as a property of EGL, even if EGL does do (*waves hands*) stuff under > the hood to ensure the two are compatible. I guess if we're assuming that the caller definitely knows about the display device and is asking EGL for the render node in order to do smarter buffer sharing between display and render, I can see it. My objection was that getting the render node in that discussion was apparently some workaround for other brokenness, and was going to result in software that didn't work on pl111 and vc4 displays because it was trying to dodge kmsro. ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: Split render/display SoCs, Mesa's renderonly, and Wayland dmabuf hints
Just 2 comments on the kernel aspects here.
On Tue, Apr 20, 2021 at 12:18 PM Daniel Stone wrote:
>
> Hi,
>
> On Mon, 19 Apr 2021 at 13:06, Simon Ser wrote:
>>
>> I'm working on a Wayland extension [1] that, among other things, allows
>> compositors to advertise the preferred device to be used by Wayland
>> clients.
>>
>> In general, compositors will send a render node. However, in the case
>> of split render/display SoCs, things get a little bit complicated.
>>
>> [...]
>
>
> Thanks for the write-up Simon!
>
>>
>> There are a few solutions:
>>
>> 1. Require compositors to discover the render device by trying to import
>>a buffer. For each available render device, the compositor would
>>allocate a buffer, export it as a DMA-BUF, import it to the
>>display-only device, then try to drmModeAddFB.
>
>
> I don't think this is actually tractable? Assuming that 'allocate a buffer'
> means 'obtain a gbm_device for the render node directly and allocate a gbm_bo
> from it', even with compatible formats and modifiers this will fail for more
> restrictive display hardware. imx-drm and pl111 (combined with vc4 on some
> Raspberry Pis) will fail this, since they'll take different allocation paths
> when they're bound through kmsro vs. directly, accounting for things like
> contiguous allocation. So we'd get false negatives on at least some platforms.
>
>>
>> 2. Allow compositors to query the render device magically opened by
>>kmsro. This could be done either via EGL_EXT_device_drm, or via a
>>new EGL extension.
>
>
> This would be my strong preference, and I don't entirely understand anholt's
> pushback here. The way I see it, GBM is about allocation for scanout, and EGL
> is about rendering. If, on a split GPU/display system, we create a gbm_device
> from a KMS display-only device node, then creating an EGLDisplay from that
> magically binds us to a completely different DRM GPU node, and anything using
> that EGLDisplay will use that GPU device to render.
>
> Being able to discover the GPU device node through the device query is really
> useful, because it tells us exactly what implicit magic EGL did under the
> hood, and about the device that EGL will use. Being able to discover the
> display node is much less useful; it does tell us how GBM will allocate
> buffers, but the user already knows which device is in use because they
> supplied it to GBM. I see the display node as a property of GBM, and the GPU
> node as a property of EGL, even if EGL does do (*waves hands*) stuff under
> the hood to ensure the two are compatible.
>
> If we had EGL_EXT_explicit_device, things get even more weird, I think; would
> the device query on an EGLDisplay created with a combination of a gbm_device
> native display handle and an explicit EGLDevice handle return the scanout
> device from GBM or the GPU device from EGL? On my reading, I'd expect it to
> be the latter; if the queries returned very different things based on whether
> GPU device selection was implicit (returning the KMS node) or explicit (GPU
> node), that would definitely violate the principle of least surprise.
>
>>
>> 3. Allow compositors to query the kernel drivers to know which devices
>>are compatible with each other. Some uAPI to query a compatible
>>display device from a render-only device, or vice-versa, has been
>>suggested in the past.
>
>
> What does 'compatible' mean? Would an Intel iGPU and and AMD dGPU be
> compatible with each other? Would a Mali GPU bound to system memory through
> AMBA be as compatible with the display controller as it would with an AMD GPU
> on PCIE? I think a query which only exposed whether or not devices could
> share dmabufs with each other is far too generic to be helpful for the actual
> usecase we have, as well as not being useful enough for other usecases ('well
> you _can_ use dmabufs from your AMD GPU on your Mali GPU, but only if they
> were allocated in the right domain').
>
>>
>> (1) has a number of limitations and gotchas. It requires allocating
>> real buffers, this has a rather big cost for something done at
>> compositor initialization time. It requires to select a buffer format
>> and modifier compatible with both devices, so it can't be isolated in
>> a simple function (and e.g. shared between all compositors in libdrm).
>
>
> We're already going to have to do throwaway allocations to make Intel's tiled
> modes work; I'd rather not extend this out to doing throwaway allocations
> across device combinations as well as modifier lists.
>
>>
>> Some drivers will allow to drmModeAddFB buffers that can't be scanned
>> out, and will only reject the buffer at atomic commit time.
>
>
> This is 100% a KMS driver bug and should be fixed there. It's not
> catastrophic, since commits can fail for any reason or none at all and
> compositors are expected to handle this, but they should absolutely be
> rejecting buffers which can never be scanned out at all at
Re: Split render/display SoCs, Mesa's renderonly, and Wayland dmabuf hints
Hi,
On Mon, 19 Apr 2021 at 13:06, Simon Ser wrote:
> I'm working on a Wayland extension [1] that, among other things, allows
> compositors to advertise the preferred device to be used by Wayland
> clients.
>
> In general, compositors will send a render node. However, in the case
> of split render/display SoCs, things get a little bit complicated.
>
> [...]
>
Thanks for the write-up Simon!
> There are a few solutions:
>
> 1. Require compositors to discover the render device by trying to import
>a buffer. For each available render device, the compositor would
>allocate a buffer, export it as a DMA-BUF, import it to the
>display-only device, then try to drmModeAddFB.
>
I don't think this is actually tractable? Assuming that 'allocate a buffer'
means 'obtain a gbm_device for the render node directly and allocate a
gbm_bo from it', even with compatible formats and modifiers this will fail
for more restrictive display hardware. imx-drm and pl111 (combined with vc4
on some Raspberry Pis) will fail this, since they'll take different
allocation paths when they're bound through kmsro vs. directly, accounting
for things like contiguous allocation. So we'd get false negatives on at
least some platforms.
> 2. Allow compositors to query the render device magically opened by
>kmsro. This could be done either via EGL_EXT_device_drm, or via a
>new EGL extension.
>
This would be my strong preference, and I don't entirely understand
anholt's pushback here. The way I see it, GBM is about allocation for
scanout, and EGL is about rendering. If, on a split GPU/display system, we
create a gbm_device from a KMS display-only device node, then creating an
EGLDisplay from that magically binds us to a completely different DRM GPU
node, and anything using that EGLDisplay will use that GPU device to render.
Being able to discover the GPU device node through the device query is
really useful, because it tells us exactly what implicit magic EGL did
under the hood, and about the device that EGL will use. Being able to
discover the display node is much less useful; it does tell us how GBM will
allocate buffers, but the user already knows which device is in use because
they supplied it to GBM. I see the display node as a property of GBM, and
the GPU node as a property of EGL, even if EGL does do (*waves hands*)
stuff under the hood to ensure the two are compatible.
If we had EGL_EXT_explicit_device, things get even more weird, I think;
would the device query on an EGLDisplay created with a combination of a
gbm_device native display handle and an explicit EGLDevice handle return
the scanout device from GBM or the GPU device from EGL? On my reading, I'd
expect it to be the latter; if the queries returned very different things
based on whether GPU device selection was implicit (returning the KMS node)
or explicit (GPU node), that would definitely violate the principle of
least surprise.
> 3. Allow compositors to query the kernel drivers to know which devices
>are compatible with each other. Some uAPI to query a compatible
>display device from a render-only device, or vice-versa, has been
>suggested in the past.
>
What does 'compatible' mean? Would an Intel iGPU and and AMD dGPU be
compatible with each other? Would a Mali GPU bound to system memory through
AMBA be as compatible with the display controller as it would with an AMD
GPU on PCIE? I think a query which only exposed whether or not devices
could share dmabufs with each other is far too generic to be helpful for
the actual usecase we have, as well as not being useful enough for other
usecases ('well you _can_ use dmabufs from your AMD GPU on your Mali GPU,
but only if they were allocated in the right domain').
> (1) has a number of limitations and gotchas. It requires allocating
> real buffers, this has a rather big cost for something done at
> compositor initialization time. It requires to select a buffer format
> and modifier compatible with both devices, so it can't be isolated in
> a simple function (and e.g. shared between all compositors in libdrm).
>
We're already going to have to do throwaway allocations to make Intel's
tiled modes work; I'd rather not extend this out to doing throwaway
allocations across device combinations as well as modifier lists.
> Some drivers will allow to drmModeAddFB buffers that can't be scanned
> out, and will only reject the buffer at atomic commit time.
>
This is 100% a KMS driver bug and should be fixed there. It's not
catastrophic, since commits can fail for any reason or none at all and
compositors are expected to handle this, but they should absolutely be
rejecting buffers which can never be scanned out at all at AddFB time.
> (2) wouldn't work with non-EGL APIs such as Vulkan. Eric Anholt seemed
> pretty opposed to this idea, but I didn't fully understood why.
>
Well, Vulkan doesn't have GBM in the same way, right? In the Vulkan case,
we already know exactly what the GPU is, because it's
Split render/display SoCs, Mesa's renderonly, and Wayland dmabuf hints
Hi all, I'm working on a Wayland extension [1] that, among other things, allows compositors to advertise the preferred device to be used by Wayland clients. In general, compositors will send a render node. However, in the case of split render/display SoCs, things get a little bit complicated. The Wayland compositor will find the display-only DRM device (usually at /dev/dri/card0). This DRM device will have a DRM primary node, but no DRM render node. The Wayland compositor will create a GBM device from this display-only device, then create an EGL display. Under the hood, Mesa's kmsro will kick in and magically open a render node from a different device. However the compositor has no knowledge of this, and has no way to discover the render node opened by kmsro. This is an issue because the compositor has no render node to send to Wayland clients. The compositor is forced to send a primary node to clients. Clients will need to open the primary node and rely on Mesa's renderonly to once again magically open the render node under the hood. In general clients cannot be expected to be able to open primary nodes. This happens to work on systemd distributions because udev sets a special uaccess tag on the primary node that makes logind grant permissions to users physically logged in on a VT. This will fall apart on non-logind systems and on systems where no user is physically logged in. Additionally, it may be desirable to deny access to primary nodes in sandboxes. So I believe the best way forward would be for the compositor to send the render node to clients. This could prevent clients to allocate buffers suitable for scan-out, but that can be fixed with some kind of buffer constraints negotiation, like we presented at XDC 2020 [2]. There are a few solutions: 1. Require compositors to discover the render device by trying to import a buffer. For each available render device, the compositor would allocate a buffer, export it as a DMA-BUF, import it to the display-only device, then try to drmModeAddFB. 2. Allow compositors to query the render device magically opened by kmsro. This could be done either via EGL_EXT_device_drm, or via a new EGL extension. 3. Allow compositors to query the kernel drivers to know which devices are compatible with each other. Some uAPI to query a compatible display device from a render-only device, or vice-versa, has been suggested in the past. (1) has a number of limitations and gotchas. It requires allocating real buffers, this has a rather big cost for something done at compositor initialization time. It requires to select a buffer format and modifier compatible with both devices, so it can't be isolated in a simple function (and e.g. shared between all compositors in libdrm). Some drivers will allow to drmModeAddFB buffers that can't be scanned out, and will only reject the buffer at atomic commit time. (2) wouldn't work with non-EGL APIs such as Vulkan. Eric Anholt seemed pretty opposed to this idea, but I didn't fully understood why. I don't know how feasible (3) is. The kernel drivers must be able to decide whether buffers coming from another driver can be scanned out, but how early can they give an answer? Can they give an answer solely based on a DRM node, and not a DMA-BUF? Feedback is welcome. Do you agree with the premise that compositors need access to the render node? Do you have any other potential solution in mind? Which solution would you prefer? Thanks, Simon [1]: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/8 [2]: https://xdc2020.x.org/event/9/contributions/634/ ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
