With multiple semaphores support, we can use a GPU job to handle
multiple signal semaphores in the end of a cmd buffer batch. It
means, the last job in the last cmd buffer will be in change of
signalling semaphores as long as it meets some conditions:
1 - A GPU-job signals semaphores whenever we only have submitted
jobs for the same queue (there is no syncobj created for any
other type). Otherwise, we emit a noop job that waits on the
completion of all jobs submitted and then signals semaphores.
2 - A CPU-job is never in charge of signalling semaphores. We
process it first and emit a noop job that depends on all jobs
previously submitted to signal semaphores.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
With multiple semaphore support, we can improve the way we handle
wait semaphores considering different job types and cmd buffer
batch scenarios, that means:
- A GPU job depends on wait semaphores whenever it is the first job
submitted to a queue in this command buffer batch (the `first` flag
for the job's queue type is set).
- For the first CPU job, if there are wait semaphores, we should
wait for the CPU and GPU being idle to process the job.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
The order in which a GPU job is scheduled is guaranteed within the
same queue type (CL, TFU, CSD), but the order of completion of jobs
from different queues cannot be guaranteed. Since we have multiple
semaphores support now, we can track the completion of the last job
submitted to each queue and therefore better determine when gpu is
idle. We do it using an array of syncobj (last_job_syncs) for each
GPU queue (CL, TFU, CSD). With this, job serialization also become
more accurate. We also keep tracking the very last job submitted
(last_job_sync became an element of the last_job_syncs array as
V3DV_QUEUE_ANY) for the case we don't have multisync support.
To help in handling wait semaphores, we set a flag per queue to
indicate we are starting a new cmd buffer batch and a job submitted
to this queue will be the first.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
In addition to keep a copy of wait semaphores, extend
v3dv_submit_info_semaphores to hold a copy of signal semaphores too.
With a copy of wait and signal semaphores, we can enable GPU jobs to
handle more than one wait and signal semaphores.
By now, we don't change the way as we signalling semaphores when all
jobs complete, i.e., we still use the master thread to signal
semaphores. In this context, no GPU job is actually in charge of
signalling, but the support for multiple signal semaphores is done
here.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
Whenever v3d kernel-driver supports multisync extension, use it to
enable more than one semaphores in a tfu job.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
Whenever v3d kernel-driver supports multisync extension, use it to
enable more than one semaphores in cl submission. In CL, we have two
kind of job (bin and render), therefore, we need also to determine
the stage to sync, that means to add job dependencies/wait
semaphores.
Also, as we currently process all signal semaphores of a cmd buffer
batch together in the submit master thread (when the last wait
thread completes), there isn't now a situation in which GPU jobs
need to handle signal VkSemaphores.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
Instead of a boolean (sem_wait) in v3dv_event_wait_cpu_job_info,
that is used to determine wait condition for jobs put in a wait
thread, copy the wait semaphores data and store it as struct
v3dv_submit_info_semaphores. In the following patches we enable
multiple semaphores in GPU jobs, and therefore we need this data
to add wait semaphores as job dependencies for pending jobs
submitted from a wait thread.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
Instead of pass pSubmit to queue_submit_cmd_buffer, create a struct
v3dv_submit_info_semaphores to wrap semaphores data from VkSubmitInfo.
In the next commit, this struct will help to handle wait condition for
jobs submitted in a wait event context, since we need to hold this
data when handle wait events and pass it to queue_submit_job() called
from wait threads. The main goal is to allow multiple wait semaphores
in a job submission. Later, this struct will be extended to include a
copy of signal semaphores too.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
is_wait_thread is passed, but not actually used; and cpu_queue_handle_idle
is in charge to handle wait threads spawned before this one.
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13178>
We had some with unlikely, some without it. Let's just put unlikely to
all of them.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13071>
Dumps the command list, excluding the binary resources.
v2 (Juan):
- Make this option independent from `cl`
v3 (Iago):
- Rename option name
- Fix style issues
- Do not print BO ranges
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12803>
This fixes an issue related with testing this with a kernel with the
performance counters enabled: it introduces a "pad" field that in the CL
submission structure that is not initialized.
Fixes: ca13868098 ("drm-uapi: add v3d performance counters")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12390>
When multiview is enabled, queries must begin and end in the
same subpass and N consecutive queries are implicitly used,
where N is the number of views enabled in the subpass.
Implementations decide how results are split across queries.
In our case, only one query is really used, but we still need
to flag all N queries as available by the time we flag the one
we use so that the application doesn't receive unexpected errors
when trying to retrieve values from them.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12034>
The idea would be to move all the code that uses cl_emit,
cl_emit_with_prepack, v3dx_pack, and any enum/structure definition
defined on the v3d pack headers.
All those methods would be defined on v3dvx_private (that would be the
equivalent to v3dx_context.h on v3d).
This commit includes the definition of v3dX for the current version
supported (42), a function calling wrapper, and the move for v3dv_queue
methods as a reference.
About the function calling wrapper, I took the idea from anv. We don't
have on v3d, but we added it because we foresee that we will need that
functionality more often. So without that macro, in order to call the
correct version of the method from the general code we would need to
do like we do on v3d, and doing something like this:
if (devinfo->ver >= 42)
return v3d42_pack_sampler_state(sampler, pCreateInfo);
else
return v3d33_pack_sampler_state(sampler, pCreateInfo);
So with the macro we can just do this:
v3dv_X(device, pack_sampler_state)(sampler, pCreateInfo).
Note that as mentioned, that is to be used on the general code, so a
runtime decision. If we are already on version-dependant code (so at
v3dx_queue for example) we just use v3dX, as at that point is a build
time decision.
Also, fwiw, I don't like too much the name of that macro, but I was
not able to think on a better one.
v2: merge job_emit_noop_bin and job_emit_noop_render (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11310>
This provides most of the implementation, but there are some
things we cannot enable until we improve of kernel submit
interface, namely:
We don't expose capacity to export SYNC_FD, although we do
have the implementation in place. This requires that we
improve our kernel interface and event wait implementation
first so we can cover the corner case where the application
submits a command buffer that includes a VkCmdWaitForEvents
and tries to export a SYNC_FD from its signal semaphores or
fence before it the event is signaled and the command buffer
is sent to the kernel for execution in full.
Likewise, we can't currently import semaphores. This is because
our current kernel submit interface can only take one syncobj.
We have been working around this so far by waiting on the last
syncobj produced from the device whenever we had to wait on any
semaphores (which is obviously suboptimal already), but this
won't work as soon as we allow importing external semaphores,
as those could (and would typically) be produced from a
different device.
Once we address the kernel bits, we should come back and enable
SYNC_FD exports as well as semaphore imports.
Relevant CTS tests:
dEQP-VK.api.external.fence.*
dEQP-VK.api.external.semaphore.*
dEQP-VK.synchronization.cross_instance.*
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11105>
We can (and should) close the descriptor immediately after the import.
Gets the following CTS test to pass without requiring to increase limits
for open file descriptors:
dEQP-VK.synchronization.basic.binary_semaphore.chain
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11105>
Check if v3dv_ioctl() or v3dv_bo_map() fail, and print a proper error
message.
This check happens in the rest of the code, so it makes sense to apply
here too.
Fixes CID#1468162 "Unchecked return value (CHECKED_RETURN)".
v2:
- Fix message error (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10380>
Dedicated BOs waste memory and are also a significant cause of CPU
overhead when applications use hundreds of them per frame due to
all the work the kernel has to do to page in all these BOs for a job.
The UE4 Vehicle demo was hitting this causing it to freeze and stutter
under 1fps.
The hardware allows us to setup groups of 16 queries in consecutive
4-byte addresses, requiring only that each group of 16 queries is
aligned to a 1024 byte boundary. With this change, we allocate all
the queries in a pool in a single BO and we assign them different
offsets based on the above restriction. This eliminates the freezes
and stutters in the Vehicle sample.
One caveat of this solution is that we can only wait or test for
completion of a query by testing if the GPU is still using its BO,
which basically means that we can only wait for all active queries
in a pool to complete and not just the ones being requested by the
API. Since the Vulkan recommendation is to use a different query
pool per frame this should not be a big issue though.
If this ever becomes a problem (for example if an application does't
follow the recommendation and instead allocates a single pool and
splits its queries between frames), we could try to group queries
in a pool into a number of BOs to try and find a balance, but for
now this should work fine in most cases.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10253>
Instead, wait for the specific queries being reset to
not be in use by the GPU.
This takes query pool resets in the UE4 Shooter demo from
50-60ms down to 0.5-2ms.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8554>
We were allocating twice the size we need for this array. This was
probably caused by a copy and paste error from the GL driver which
grows this dynamically as BOs are added to the job.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7733>
Used as reference Hyujun's commit
5d3fdbc52b, that does the same for
turnip.
This commit also replaces in several cases alloc for zalloc, and adds
checks on more Destroy methods if the object to be free is NULL or
not. Most of them were needed to avoid crashes/weird behaviour due
trying to use un-initialized data. Note that now that vk_object_free
iterates over a array, making it more against un-initialized or just
NULL data.
Additionally, using zalloc we can also remove some memset to 0. In
fact we needed to remove them, as if not, they would override the
vk_object_base object to 0 (the alternative would me doing a memset
computing a pointer offset, but that's is not needed as we can just
use zalloc).
v2:
* Call memset(0) on reused descriptor sets when calling
ResetDescriptorPool, not when reallocating them (Iago)
* Add null check when calling DestroyImageView (detected by a full CTS run)
v3: Fixed rebase conflicts after last meta copy/clear changes
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7627>
So far, we have only been supporting X11, so we assumed that we were running
inside X11 and would always try to get an authenticated fd from Xorg during
device initialization. While this works for desktop Raspbian, it is not
really correct and it is not what we want to do when we start considering
other WSIs.
Initially, one could think we can still do this by guarding the WSI code
under the proper instance extension check. This, however, doesn't work
reliably, as the Vulkan loader can call vkEnumerateDevices without enabling
surface extensions on the instance, which then can lead to us not
initializing any display_fd and failing with VK_ERROR_INITIALIZATION_FAILED,
which is not correct, so while we can try to acquire the display_fd here,
it might not always work, and we should definitely not fail initialization
of the physical device for that.
Instead, with this change we move acquisition of display_fd to swapchain
creation time where required extensions need to be enabled in the instance.
This was also suggested by Daniel Stone during review of a work-in-progress
implementation for the Wayland WSI.
There is a special case to consider though: applications like Zink that
don't use Vulkan's swapchains at all but still allocate images that they
intend to use for WSI. We need to handle these by checking that we have
indeed acquired a display_fd before doing any memory allocation for WSI,
and acquiring one at that time if that's not the case.
This change also removes the render_fd and display_fd fields from the
logical device (which we were copying from the physical device), because
now there is no guarantee that we have acquired a display_fd at the
time we create a logical device. Instead, we now put a reference to the
physical device on the logical device from which we can access these.
Finally, this also fixes a regression introduced with VK_KHR_display, where
if that extension is enabled but we are running inside a compositor, we would
acquire a display_fd that is not authenticated and try to use that instead
of acquiring an authenticated display_fd from the display server.
Fixes: b1188c9451 (v3dv: VK_KHR_display extension support)
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7546>
V3D doesn't provide any means to acquire timestamps from the GPU
so we have to implement these in the CPU.
v2: enable timestampComputeAndGraphics and set timestampPeriod (Piñeiro)
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7373>
Fix defect reported by Coverity Scan.
Assign instead of compare (PW.ASSIGN_WHERE_COMPARE_MEANT)
assign_where_compare_meant: use of "=" where "==" may have been intended
Fixes: ca86c7c65a ("v3dv: assert command buffers are executable when submitting to a queue")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7156>
If we have a semaphore wait the job cannot be started before the semaphore
has been signaled, so we need to wait before starting the binning stage.
Fixes CTS failures in:
dEQP-VK.synchronization.op.single_queue.binary_semaphore.*
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
So far we have been getting away with finishing the current job in the
presence of a pipeline barrier and relying on the RCL serialization,
but of course this is not always enough.
This patch addresses synchronization across different GPU units
(i.e. draw indirect after compute), as well as cases where we need to
sync before binning.
Fixes CTS failures in:
dEQP-VK.synchronization.op.single_queue.barrier.*
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
The hardware can't do this, so we need to record a CPU job that will
map the indirect buffer at queue submission time, read the dispatch
parameters and then submit a regular dispatch.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
Both the API user and the driver may attempt to map a BO, possibly
only partially and using different ranges. This is a problem because
we only have a single map per BO. Fix this by making sure that when
a BO is mapped, we always map its entire range. This way if a BO
has been mapped before, we know that map is still valid no matter the
region we need to access now.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
The blit shader path for buffer to image copies is pretty bad,
since it needs to produce a tiled image from the linear buffer
prior to emitting the blit copy.
This patch adds a new preferential path where we implement the
copy using the CPU, similar to what the GL driver does for
texture uploads. This makes vkQuake2 at least 4x faster when
dynamic lights are enabled (which triggers dynamic texture
updates).
We also tested a GPU path where we use a shader that takes the
linear buffer as a UBO and copies directly from it. This also
shows a clear performance gain, but still worse than the CPU
implementation.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
Asserting on them makes it easier to identify applications and tests that
try to use unimplemented features.
Also, there are some APIs that relate to optional features we don't
(or can't) support, such as sparse, so for these we just provide
the trivial implementation.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
There are basically two types of scenarios to consider:
- Secondary command buffers that run inside a render pass.
- Secondary command buffers that run outside a render pass.
For the former we want to record their commands into a binning command
list that we can branch to when executed into a primary command
buffer. This means this kind of command buffers don't spawn new jobs,
just the default one where they record the binning commands which
won't include the frame setup, which will be provided by the primary
they will be executed in.
For the latter we don't require anything special, we just record as
many jobs as we need as usual and link that job list from the primary
job list when executed.
This handles most scenarios except:
- vkCmdWaitForEvents
- VkCmdClearAttachments
Both of these can spawn new jobs inside a render pass, which is not
what we want for secondary command buffers. We will address this is
follow-up patches.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
This reverts a previous half-attempt at an implementation of events
using a BO to hold the event state, and provides a full
implementation. V3D doesn't have any built-in GPU functionality to
wait on any kind of events, so we need to implement this in the driver
an therefore we no longer need to use a BO for the event state.
Instead, we implement GPU waits by using a CPU job for the wait
operation that spawns a wait thread if the wait operation doesn't have
all its events signaled by the time it is processed. To implement the
semantics of the wait correctly, any jobs in the same command buffer
that come after the wait will not be emitted until the wait thread
completes.
If a submit spawns any wait threads for a command buffer we can't
signal any semaphores for it until all the wait threads complete and
we know that all the jobs for those command buffers have been
submitted. The same applies to the submit fence, if present.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
While very limited in scope, this might be the most efficient way to blit
when applicable. In fact, we might also want to use this for the image copy
commands when possible instead of the TLB.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
This is similar to the scenario where we have a submit without
any command buffers, even if we don't have any actual GPU work to do
we still might need to signal fences/semaphored and possibly wait on
previous jobs to finish, so we need to submit something to the kernel
to get all that done right.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
The design for queries in Vulkan requires that some commands execute
in the GPU as part of a command buffer. Unfortunately, V3D doesn't
really have supprt for this, which means that we need to execute them
in the CPU but we still need to make it look as if they happened
inside the comamnd buffer from the point of view of the user, which
adds certain hassle.
The above means that in some cases we need to do CPU waits for certain
parts of the command buffer to execute so we can then run the CPU
code. For exmaple, we need to wait before executing a query resets
just in case the GPU is using them, and we have to do a CPU wait wait
for previous GPU jobs to complete before copying query results if the
user has asked us to do that. In the future, we may want to have
submission thread instead so we don't block the main thread in these
scenarios.
Because we now need to execute some tasks in the CPU as part of a
command buffer, this introduces the concept of job types, there is one
type for all GPU jobs, and then we have one type for each kind of job
that needs to execute in the CPU. CPU jobs are executed by the queue
in order just like GPU jobs, only that they are exclusively CPU tasks.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
