With softpin we allow the userspace to take control over the GPU virtual
address space. The new capability is relected by a bump of the minor DRM
version. There are a few restrictions for userspace to take into
account:
1. The kernel reserves a bit of the address space to implement zero page
faulting and mapping of the kernel internal ring buffer. Userspace can
query the kernel for the first usable GPU VM address via
ETNAVIV_PARAM_SOFTPIN_START_ADDR.
2. We only allow softpin on GPUs, which implement proper process
separation via PPAS. If softpin is not available the softpin start
address will be set to ~0.
3. Softpin is all or nothing. A submit using softpin must not use any
address fixups via relocs.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
This builds on top of the MMU contexts introduced earlier. Instead of having
one context per GPU core, each GPU client receives its own context.
On MMUv1 this still means a single shared pagetable set is used by all
clients, but on MMUv2 there is now a distinct set of pagetables for each
client. As the command fetch is also translated via the MMU on MMUv2 the
kernel command ringbuffer is mapped into each of the client pagetables.
As the MMU context switch is a bit of a heavy operation, due to the needed
cache and TLB flushing, this patch implements a lazy way of switching the
MMU context. The kernel does not have its own MMU context, but reuses the
last client context for all of its operations. This has some visible impact,
as the GPU can now only be started once a client has submitted some work and
we got the client MMU context assigned. Also the MMU context has a different
lifetime than the general client context, as the GPU might still execute the
kernel command buffer in the context of a client even after the client has
completed all GPU work and has been terminated. Only when the GPU is runtime
suspended or switches to another clients MMU context is the old context
freed up.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
Move buffer setup and starting of the FE loop in the kernel ringbuffer
into a separate function. This is a preparation to start the FE later
in the submit process.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
This reworks the MMU handling to make it possible to have multiple MMU contexts.
A context is basically one instance of GPU page tables. Currently we have one
set of page tables per GPU, which isn't all that clever, as it has the
following two consequences:
1. All GPU clients (aka processes) are sharing the same pagetables, which means
there is no isolation between clients, but only between GPU assigned memory
spaces and the rest of the system. Better than nothing, but also not great.
2. Clients operating on the same set of buffers with different etnaviv GPU
cores, e.g. a workload using both the 2D and 3D GPU, need to map the used
buffers into the pagetable sets of each used GPU.
This patch reworks all the MMU handling to introduce the abstraction of the
MMU context. A context can be shared across different GPU cores, as long as
they have compatible MMU implementations, which is the case for all systems
with Vivante GPUs seen in the wild.
As MMUv1 is not able to change pagetables on the fly, without a
"stop the world" operation, which stops GPU, changes pagetables via CPU
interaction, restarts GPU, the implementation introduces a shared context on
MMUv1, which is returned whenever there is a request for a new context.
This patch assigns a MMU context to each GPU, so on MMUv2 systems there is
still one set of pagetables per GPU, but due to the shared context MMUv1
systems see a change in behavior as now a single pagetable set is used
across all GPU cores.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
There is no need for each GPU to have it's own cmdbuf suballocation
region. Only allocate a single one for the the etnaviv virtual device
and share it across all GPUs.
As the suballoc space is now potentially shared by more hardware jobs
running in parallel, double its size to 512KB to avoid contention.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
This allows to decouple the cmdbuf suballocator create and mapping
the region into the GPU address space. Allowing multiple AS to share
a single cmdbuf suballoc.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
Remember if the GPU has been sucessfully initialized. Only in that case
do we need to clean up various structures in the unbind path. If the
GPU hasn't been sucessfully initialized all the cleanups should happen
in the failure paths of the init function.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Guido Günther <agx@sigxcpu.org>
Drop unused includes, move more includes from the generic etnaviv_drv.h to
the units where they are actually used, sort includes.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Use devm_platform_ioremap_resource() to simplify the code a bit.
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Drop use of the deprecated drmP.h header file.
Fix fallout in all .c files.
The etnaviv_drv.h header file was made self-contained,
and missing includes was then added to the .c files that needed them.
In a few cases the list of include files was sorted.
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Russell King <linux+etnaviv@armlinux.org.uk>
Cc: Christian Gmeiner <christian.gmeiner@gmail.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: etnaviv@lists.freedesktop.org
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
When something goes wrong in the GPU init after the cmdbuf suballocator
has been constructed, we fail to destroy it properly. This causes havok
later when the GPU is unbound due to a module unload or similar.
Fixes: e66774dd6f (drm/etnaviv: add cmdbuf suballocator)
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Russell King <rmk+kernel@armlinux.org.uk>
If there is a match in the HW DB, the function is left early, before
inititalizing the idle mask. Fix this by doing the init earlier, as
only old GPUs, not present in the HW DB need a different idle mask.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
It only written and we don't infer any useful information from
it anymore. Remove it.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
The etnaviv_gpu header only needs to know about the pointer types, so
replace by a forward declaration and only include the headers where needed.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
The only event function that is called from IRQ context is event_free,
which is already using atomic bitmap operations, so we can avoid taking
the event spinlock in this function completely. As other the other
functions still using the event spinlock are all called from normal
process context, we can avoid disabling IRQs while holding the spinlock.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
This is the only place in the driver that should have to deal with
the raw hardware fences. To avoid any further confusion, consolidate
the fence handling in this file and remove any traces of this from
the header files.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
There is no need to track the currently active fence. The GPU scheduler
keeps track of all the in-flight jobs.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
From: Lucas Stach <l.stach@pengutronix.de>
"not much to de-stage this time. Changes from Philipp and Souptick to
use memset32 more and switch the fault handler to the new vm_fault_t
and two small fixes for issues that can be hit in rare corner cases
from me."
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1533563808.2809.7.camel@pengutronix.de
When the suballocator was unable to provide a suitable buffer for the MMUv1
linear window, we roll back the GPU initialization. As the GPU is runtime
resumed at that point we need to clear the kernel cmdbuf suballoc entry to
properly skip any attempt to manipulate the cmdbuf when the GPU gets shut
down in the runtime suspend later on.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The documentation of drm_sched_job_init and drm_sched_entity_push_job has
been clarified. Both functions should be called under a shared lock, to
avoid jobs getting pushed into the scheduler queue in a different order
than their sched_fence seqnos, which will confuse checks that are looking
at the seqnos to infer information about completion order.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
dma_fence_default_wait is the default now, same for the trivial
enable_signaling implementation.
Acked-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Russell King <linux+etnaviv@armlinux.org.uk>
Cc: Christian Gmeiner <christian.gmeiner@gmail.com>
Cc: etnaviv@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20180503142603.28513-8-daniel.vetter@ffwll.ch
This replaces the repetitive GPL-2.0 license text in code and header files
with the SPDX tags. Generated hardware headers aren't changed, as any changes
there need to be done in the upstream rnndb repository.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
I'm not aware of any case where tracing GPU register manipulation at the
kernel level would have been useful. It only adds more indirections and
adds to the code size.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
GPUs with support for the security features need some additional
setup to get the frontend started.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
New versions of the Vivante kernel driver don't trust the hardware feature
bits anymore, but use an internal hardware database. This also includes
more feature fields than are available in hardware.
As we can't trust the hardware feature bits to be correct anymore, we need
to replicate the HWDB in etanviv. For now only the GC7000L as found on
the i.MX8M is supported.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The slave interface clock is a clock input found on newer cores to gate
the register interface. For now we simply ungate it when the GPU is in
active state.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Split out the fault dumping, as this will get more complex in the future.
Also there is no need to read and dump the fault address from MMUs that
didn't signal a fault.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The module autoloading can be triggered through the GPU core nodes
and the necessary platform device for the DRM toplevel device will
be instantiated on module init.
Suggested-by: Rob Herring <robh@kernel.org>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Rob Herring <robh@kernel.org>
This replaces the etnaviv internal hangcheck logic with the job timeout
handling provided by the DRM scheduler. This simplifies the driver further
and allows to replay jobs after a GPU reset, so only minimal state is lost.
This introduces a user-visible change in that we don't allow jobs to run
indefinitely as long as they make progress anymore, as this introduces
quality of service issues when multiple processes are using the GPU.
Userspace is now responsible to flush jobs in a way that the finish in a
reasonable time, where reasonable is currently defined as less than 500ms.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Move the fence dependency handling to the scheduler where it belongs.
Jobs with unsignaled dependencies just get to sit in the scheduler queue
without holding any locks.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
This hooks in the DRM GPU scheduler. No improvement yet, as all the
dependency handling is still done in etnaviv_gem_submit. This just
replaces the actual GPU submit by passing through the scheduler.
Allows to get rid of the retire worker, as this is now driven by the
scheduler.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
This moves away from using the internal seqno as the userspace fence
reference. By moving to a generic ID, we can later replace the internal
fence by something different than the etnaviv seqno fence.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Some architecture ports like ARC don't provide the PHYS_OFFSET symbol.
Define it to 0 in that case, which is the most conservative default in
the usage context of the etnaviv driver.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
There is no need to hold the GPU lock while freeing the submit
object. Only move the retired submits from the GPU active list to
a temporary retire list under the GPU lock.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
As long as there is an active submit, we want the GPU to stay awake. This
is slightly complicated by the fact that we really want to wake the GPU
at the last possible moment to achieve maximum power savings.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The active count is used to check if the BO is idle, where idle is defined
as not active on the GPU and all VM mappings and reference counts dropped
to the initial state. As the idling of the mappings and references now only
happens in the submit cleanup, the active state handling must be moved to
the same location in order to keep the userspace semantics.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Less dynamic allocations and slims down the cmdbuf object to only the
required information, as everything else is already available in the
submit object.
This also simplifies buffer and mappings lifetime management, as they
are now exlusively attached to the submit object and not additionally
to the cmdbuf.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The GPU exec state may have changed at the time when the perfmon sampling
is done, as it reflects the state of the last submission, not the current
GPU execution state.
So for proper sampling we must use the submit exec_state.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
We'll need this in some places where only the submit is available. Also
this is a first step at slimming down the cmdbuf object.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
To make them available to the event worker even after the actual
command stream execution has finished.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
The object fencing has nothing to do with the actual GPU buffer submit,
so move it to the gem submit path to have a cleaner split.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Inserting the END command when suspending the GPU is changing the
command buffer state, which requires the GPU to be held.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
While the etnaviv workqueue needs to be ordered, as we rely on work items
being executed in queuing order, this is only true for a single GPU.
Having a shared workqueue for all GPUs in the system limits concurrency
artificially.
Getting each GPU its own ordered workqueue still meets our ordering
expectations and enables retire workers to run concurrently.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
There is no need to store this in the gpu struct. MMU flushes are triggered
correctly in reaction to MMU maps and unmaps, independent of the current ctx.
Any required pipe switches can be infered from the current and the desired
GPU exec state.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
There is no need to synchronize with oustanding retire jobs if the object
has gone idle. Retire jobs only ever change the object state from active to
idle, not the other way around.
The IOVA put race is uncritical, as the GEM_WAIT ioctl itself is holding
a reference to the GEM object, so the retire worker will not pull the
object into the CPU domain, which is the thing we are trying to guard
against with etnaviv_gpu_wait_obj_inactive. The ordering of the various
counts and waits may change a bit, but the userspace visible behavior at
the bounds of the syscall are unchanged.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Flush and prefetch are properly handled in the buffer code, data endianess
would need much wider changes than adding something to this single function.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
If the FE is restarted before the sync point event is cleared, the GPU
might trigger a completion IRQ for the next sync point, corrupting
the state of the currently running worker.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
