The motivation at this point is mainly native userspace mesa driver in a
VM guest. The one remaining synchronous "hotpath" is buffer allocation,
because guest needs to wait to know the bo's iova before it can start
emitting cmdstream/state that references the new bo. By allocating the
iova in the guest userspace, we no longer need to wait for a response
from the host, but can just rely on the allocation request being
processed before the cmdstream submission. Allocation failures (OoM,
etc) would just be treated as context-lost (ie. GL_GUILTY_CONTEXT_RESET)
or subsequent allocations (or readpix, etc) can raise GL_OUT_OF_MEMORY.
v2: Fix inuse check
v3: Change mismatched iova case to -EBUSY
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Link: https://lore.kernel.org/r/20220411215849.297838-11-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
With userspace allocated iova (next patch), we can have a race condition
where userspace observes the fence completion and deletes the vma before
retire_submit() gets around to unpinning the vma. To handle this, add a
fenced unpin which drops the refcount but tracks the fence, and update
msm_gem_vma_inuse() to check any previously unsignaled fences.
v2: Fix inuse underflow (duplicate unpin)
v3: Fix msm_job_run() vs submit_cleanup() race condition
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220411215849.297838-10-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
There was only a single user, which could just as easily stash the iova
when pinning.
v2: fix prepare->prepare->cleanup->cleanup sequences
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220411215849.297838-7-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Get rid of all the unnecessary conversion between address/size and page
offsets. It just confuses things.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220411215849.297838-6-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Prep for a following patch, where it gets a bit more complicated.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220411215849.297838-5-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
These belong more cleanly in the gem header.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220411215849.297838-2-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Other processes don't need to know about faults that they are isolated
from by virtue of address space isolation. They are only interested in
whether some of their state might have been corrupted.
But to be safe, also track unattributed faults. This case should really
never happen unless there is a kernel bug (and that would never happen,
right?)
v2: Instead of adding a new param, just change the behavior of the
existing param to match what userspace actually wants [anholt]
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5934
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220201161618.778455-3-robdclark@gmail.com
Reviewed-by: Emma Anholt <emma@anholt.net>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.
The respective msm functions are being removed. The file_operations
structure fops is now being created by the helper macro
DEFINE_DRM_GEM_FOPS().
v2:
* rebase onto latest upstream
* remove declaration of msm_gem_mmap_obj() from msm_fbdev.c
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20210706084753.8194-1-tzimmermann@suse.de
[squash in missing VM_DONTEXPAND flag]
Signed-off-by: Rob Clark <robdclark@chromium.org>
This was only used to detect userspace including the same bo multiple
times in a submit. But ww_mutex can already tell us this.
When we drop struct_mutex around the submit ioctl, we'd otherwise need
to lock the bo before adding it to the bo_list. But since ww_mutex can
already tell us this, it is simpler just to remove the bo_list.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20210728010632.2633470-11-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
For existing adrenos, there is one or more ringbuffer, depending on
whether preemption is supported. When preemption is supported, each
ringbuffer has it's own priority. A submitqueue (which maps to a
gl context or vk queue in userspace) is mapped to a specific ring-
buffer at creation time, based on the submitqueue's priority.
Each ringbuffer has it's own drm_gpu_scheduler. Each submitqueue
maps to a drm_sched_entity. And each submit maps to a drm_sched_job.
Closes: https://gitlab.freedesktop.org/drm/msm/-/issues/4
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20210728010632.2633470-10-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Previously the (non-fd) fence returned from submit ioctl was a raw
seqno, which is scoped to the ring. But from UABI standpoint, the
ioctls related to seqno fences all specify a submitqueue. We can
take advantage of that to replace the seqno fences with a cyclic idr
handle.
This is in preperation for moving to drm scheduler, at which point
the submit ioctl will return after queuing the submit job to the
scheduler, but before the submit is written into the ring (and
therefore before a ring seqno has been assigned). Which means we
need to replace the dma_fence that userspace may need to wait on
with a scheduler fence.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20210728010632.2633470-8-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Move all the locked/active/pinned state handling to msm_gem_submit.c.
In particular, for drm/scheduler, we'll need to do all this before
pushing the submit job to the scheduler. But while we're at it we can
get rid of the dupicate pin and refcnt.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20210728010632.2633470-7-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
No idea why we were still using this. It certainly hasn't been needed
for some time. So drop the pointless twin codepaths.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20210728010632.2633470-4-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Fix a couple incorrect or misspelt comments, and add submitqueue doc
comment.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20210728010632.2633470-2-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Wire up support to stall the SMMU on iova fault, and collect a devcore-
dump snapshot for easier debugging of faults.
Currently this is a6xx-only, but mostly only because so far it is the
only one using adreno-smmu-priv.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Jordan Crouse <jordan@cosmicpenguin.net>
Link: https://lore.kernel.org/r/20210610214431.539029-6-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Our initial logic for excluding dma-bufs was not quite right. In
particular we want msm_gem_get/put_pages() path used for exported
dma-bufs to increment/decrement the pin-count.
Also, in case the importer is vmap'ing the dma-buf, we need to be
sure to update the object's status, because it is now no longer
potentially evictable.
Fixes: 63f17ef834 drm/msm: Support evicting GEM objects to swap
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20210426235326.1230125-1-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Objects that are potential for swapping out are (1) willneed (ie. if
they are purgable/MADV_WONTNEED we can just free the pages without them
having to land in swap), (2) not on an active list, (3) not dma-buf
imported or exported, and (4) not vmap'd. This repurposes the purged
list for objects that do not have backing pages (either because they
have not been pinned for the first time yet, or in a later patch because
they have been unpinned/evicted.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20210405174532.1441497-7-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Currently nearly everything, other than newly allocated objects which
are not yet backed by pages, is pinned and resident in RAM. But it will
be nice to have some stats on what is unpinned once that is supported.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20210405174532.1441497-6-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
If you mess something up, you don't really need to see the same warn on
splat 4000 times pumped out a slow debug UART port..
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20210405174532.1441497-2-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
The last patch lost the breakdown of active vs inactive GEM objects in
$debugfs/gem. But we can add some better stats to summarize not just
active vs inactive, but also purgable/purged to make up for that.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20210401012722.527712-5-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
In normal cases the gem obj lock is acquired first before mm_lock. The
exception is iterating the various object lists. In the shrinker path,
deadlock is avoided by using msm_gem_trylock() and skipping over objects
that cannot be locked. But for debugfs the straightforward thing is to
split things out into a separate list of all objects protected by it's
own lock.
Fixes: d984457b31 ("drm/msm: Add priv->mm_lock to protect active/inactive lists")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20210401012722.527712-4-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
When the system is under heavy memory pressure, we can end up with lots
of concurrent calls into the shrinker. Keeping a running tab on what we
can shrink avoids grabbing a lock in shrinker->count(), and avoids
shrinker->scan() getting called when not profitable.
Also, we can keep purged objects in their own list to avoid re-traversing
them to help cut down time in the critical section further.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20210401012722.527712-3-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Unused since commit c951a9b284 ("drm/msm: Remove msm_gem_free_work")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20210401012722.527712-2-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
* Shutdown hook for GPU (to ensure GPU is idle before iommu goes away)
* GPU cooling device support
* DSI 7nm and 10nm phy/pll updates
* Additional sm8150/sm8250 DPU support (merge_3d and DSPP color
processing)
* Various DP fixes
* A whole bunch of W=1 fixes from Lee Jones
* GEM locking re-work (no more trylock_recursive in shrinker!)
* LLCC (system cache) support
* Various other fixes/cleanups
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGt0G=H3_RbF_GAQv838z5uujSmFd+7fYhL6Yg=23LwZ=g@mail.gmail.com
Previously we only held obj lock in the _active_get() path, and relied
on atomic_dec_return() to not be racy in the _active_put() path where
obj lock was not held.
But this is a false sense of security. Unlike obj lifetime refcnt,
where you do not expect to *increase* the refcnt after the last put
(which would mean that something has gone horribly wrong with the
object liveness reference counting), the active_count can increase
again from zero. Racing _active_put()s and _active_get()s could leave
the obj on the wrong mm list.
But in the retire path, immediately after the _active_put(), the
_unpin_iova() would acquire obj lock. So just move the locking earlier
and rely on that to protect obj->active_count.
Fixes: c5c1643cef ("drm/msm: Drop struct_mutex from the retire path")
Signed-off-by: Rob Clark <robdclark@chromium.org>
The obj->lock is sufficient for what we need.
This *does* have the implication that userspace can try to shoot
themselves in the foot by racing madvise(DONTNEED) with submit. But
the result will be about the same if they did madvise(DONTNEED) before
the submit ioctl, ie. they might not get want they want if they race
with shrinker. But iova fault handling is robust enough, so userspace
is only shooting it's own foot.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Now that we don't need struct_mutex in the free path, we can get rid of
the asynchronous free altogether.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
It cannot be atomically updated with obj->active_count, and the only
purpose is a useless WARN_ON() (which becomes a buggy WARN_ON() once
retire_submits() is not serialized with incoming submits via
struct_mutex)
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Before we remove dev->struct_mutex from the retire path, we have to deal
with the situation of a submit retiring before the submit ioctl returns.
To deal with this, ring->submits will hold a reference to the submit,
which is dropped when the submit is retired. And the submit ioctl path
holds it's own ref, which it drops when it is done with the submit.
Also, add to submit list *after* getting/pinning bo's, to prevent badness
in case the completed fence is corrupted, and retire_worker mistakenly
believes the submit is done too early.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Jordan Crouse <jcrouse@codeaurora.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
This also converts the special msm_gem_get_vaddr_active() to expect the
lock to already be held. There are two call-sites for this, one already
has the lock held, so it is more straightforward to just open-code the
locking for the other caller.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
We cannot switch to using obj->resv for locking without first moving all
the copy_from_user() ahead of submit_lock_objects(). Otherwise in the
mm fault path we aquire mm->mmap_sem before obj lock, but in the submit
path the order is reversed.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Move grabbing the bo lock into shrinker, with a msm_gem_trylock() to
skip over bo's that are already locked. This gets rid of the nested
lock classes.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
When we cut-over to using dma_resv_lock/etc instead of msm_obj->lock,
we'll need these for the submit path (where resv->lock is already held).
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
This will make it easier to transition over to obj->resv locking for
everything that is per-bo locking.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
In the case where we have a back-to-back submission that shares the same
BO, this BO will be prematurely moved to inactive_list while retiring the
first submit. But it will be still part of the second submit which is
being processed by the GPU. Now, if the shrinker happens to be triggered at
this point, it will result in a premature purging of this BO.
To fix this, we need to refcount BO while doing submit and retire. Then,
it should be moved to inactive list when this refcount becomes 0.
Signed-off-by: Akhil P Oommen <akhilpo@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
In $debugfs/gem we already show any vma(s) associated with an object.
Also show process names if the vma's address space is a per-process
address space.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Jordan Crouse <jcrouse@codeaurora.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Each submitqueue is attached to a context. Add a pointer to the
context to the submitqueue at create time and refcount it so
that it stays around through the life of the queue.
Co-developed-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
This flag sets IOMMU_PRIV, which is required for some a6xx GMU objects.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Jordan Crouse <jcrouse@codeauorora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Also log buffers with the DUMP flag set, to ensure we capture all useful
cmdstream in crashdump state with modern mesa.
Otherwise we miss out on the contents of "state object" cmdstream
buffers.
v2: add missing 'inline'
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Jordan Crouse <jcrouse@codeaurora.org>
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Be more consistent with the naming of the other DMA-buf objects.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/323401/
Based on 1 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license version 2 as
published by the free software foundation this program is
distributed in the hope that it will be useful but without any
warranty without even the implied warranty of merchantability or
fitness for a particular purpose see the gnu general public license
for more details you should have received a copy of the gnu general
public license along with this program if not see http www gnu org
licenses
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-only
has been chosen to replace the boilerplate/reference in 503 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Alexios Zavras <alexios.zavras@intel.com>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Enrico Weigelt <info@metux.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190602204653.811534538@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pass the index of the MMU domain in struct msm_file_private instead
of assuming gpu->id throughout the submit path. This clears the way
to change ctx->aspace to a per-instance pagetable.
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
