gpu: host1x: Don't rely on dma_fence_wait_timeout return value

dma_fence_wait_timeout (along with a host of other jiffies-based
timeouting functions) returns zero both in case of timeout and when
th

gpu: host1x: Don't rely on dma_fence_wait_timeout return value

dma_fence_wait_timeout (along with a host of other jiffies-based
timeouting functions) returns zero both in case of timeout and when
the wait completes during the last jiffy before timeout. As such,
we can't rely on it to distinguish between success and timeout.

To prevent confusing callers by returning -EAGAIN before the timeout
period has elapsed, check if the fence got signaled again after
the wait.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: External timeout/cancellation for fences

Currently all fences have a 30 second timeout to ensure they are
cleaned up if the fence never completes otherwise. However, this
one size fits

gpu: host1x: External timeout/cancellation for fences

Currently all fences have a 30 second timeout to ensure they are
cleaned up if the fence never completes otherwise. However, this
one size fits all solution doesn't actually fit in every case,
such as syncpoint waiting where we want to be able to have timeouts
longer than 30 seconds. As such, we want to be able to give control
over fence cancellation to the caller (and maybe eventually get rid
of the internal timeout altogether).

Here we add this cancellation mechanism by essentially adding a
function for entering the timeout path by function call, and changing
the syncpoint wait function to use it.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Implement syncpoint wait using DMA fences

In anticipation of removal of the intr API, move host1x_syncpt_wait
to use DMA fences instead. As of this patch, this means that waits
have a 3

gpu: host1x: Implement syncpoint wait using DMA fences

In anticipation of removal of the intr API, move host1x_syncpt_wait
to use DMA fences instead. As of this patch, this means that waits
have a 30 second maximum timeout because of the implicit timeout
we have with fences, but that will be lifted in a follow-up patch.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Always return syncpoint value when waiting

The new TegraDRM UAPI uses syncpoint waiting with timeout set to
zero to indicate reading the syncpoint value. To support that we
need to retu

gpu: host1x: Always return syncpoint value when waiting

The new TegraDRM UAPI uses syncpoint waiting with timeout set to
zero to indicate reading the syncpoint value. To support that we
need to return the syncpoint value always when waiting.

Fixes: 44e961381354 ("drm/tegra: Implement syncpoint wait UAPI")
Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Fix hang on Tegra186+

Tegra186+ hangs if host1x hardware is disabled at a kernel boot time
because we touch hardware before runtime PM is resumed. Move sync point
assignment initializat

gpu: host1x: Fix hang on Tegra186+

Tegra186+ hangs if host1x hardware is disabled at a kernel boot time
because we touch hardware before runtime PM is resumed. Move sync point
assignment initialization to the RPM-resume callback. Older SoCs were
unaffected because they skip that sync point initialization.

Tested-by: Jon Hunter <[email protected]> # T186
Reported-by: Jon Hunter <[email protected]> # T186
Fixes: 6b6776e2ab8a ("gpu: host1x: Add initial runtime PM and OPP support")
Signed-off-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Add initial runtime PM and OPP support

Add runtime PM and OPP support to the Host1x driver. For the starter we
will keep host1x always-on because dynamic power management require a majo

gpu: host1x: Add initial runtime PM and OPP support

Add runtime PM and OPP support to the Host1x driver. For the starter we
will keep host1x always-on because dynamic power management require a major
refactoring of the driver code since lot's of code paths are missing the
RPM handling and we're going to remove some of these paths in the future.

Reviewed-by: Ulf Hansson <[email protected]>
Tested-by: Peter Geis <[email protected]> # Ouya T30
Tested-by: Paul Fertser <[email protected]> # PAZ00 T20
Tested-by: Nicolas Chauvet <[email protected]> # PAZ00 T20 and TK1 T124
Tested-by: Matt Merhar <[email protected]> # Ouya T30
Signed-off-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Add no-recovery mode

Add a new property for jobs to enable or disable recovery i.e.
CPU increments of syncpoints to max value on job timeout. This
allows for a more solid model for hang

gpu: host1x: Add no-recovery mode

Add a new property for jobs to enable or disable recovery i.e.
CPU increments of syncpoints to max value on job timeout. This
allows for a more solid model for hanged jobs, where userspace
doesn't need to guess if a syncpoint increment happened because
the job completed, or because job timeout was triggered.

On job timeout, we stop the channel, NOP all future jobs on the
channel using the same syncpoint, mark the syncpoint as locked
and resume the channel from the next job, if any.

The future jobs are NOPed, since because we don't do the CPU
increments, the value of the syncpoint is no longer synchronized,
and any waiters would become confused if a future job incremented
the syncpoint. The syncpoint is marked locked to ensure that any
future jobs cannot increment the syncpoint either, until the
application has recognized the situation and reallocated the
syncpoint.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Reserve VBLANK syncpoints at initialization

On T20-T148 chips, the bootloader can set up a boot splash
screen with DC configured to increment syncpoint 26/27
at VBLANK. Because of this

gpu: host1x: Reserve VBLANK syncpoints at initialization

On T20-T148 chips, the bootloader can set up a boot splash
screen with DC configured to increment syncpoint 26/27
at VBLANK. Because of this we shouldn't allow these syncpoints
to be allocated until DC has been reset and will no longer
increment them in the background.

As such, on these chips, reserve those two syncpoints at
initialization, and only mark them free once the DC
driver has indicated it's safe to do so.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Reset max value when freeing a syncpoint

With job recovery becoming optional, syncpoints may have a mismatch
between their value and max value when freed. As such, when freeing,
set the

gpu: host1x: Reset max value when freeing a syncpoint

With job recovery becoming optional, syncpoints may have a mismatch
between their value and max value when freed. As such, when freeing,
set the max value to the current value of the syncpoint so that it
is in a sane state for the next user.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Cleanup and refcounting for syncpoints

Add reference counting for allocated syncpoints to allow keeping
them allocated while jobs are referencing them. Additionally,
clean up various pl

gpu: host1x: Cleanup and refcounting for syncpoints

Add reference counting for allocated syncpoints to allow keeping
them allocated while jobs are referencing them. Additionally,
clean up various places using syncpoint IDs to use host1x_syncpt
pointers instead.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Use HW-equivalent syncpoint expiration check

Make syncpoint expiration checks always use the same logic used by
the hardware. This ensures that there are no race conditions that
could o

gpu: host1x: Use HW-equivalent syncpoint expiration check

Make syncpoint expiration checks always use the same logic used by
the hardware. This ensures that there are no race conditions that
could occur because of the hardware triggering a syncpoint interrupt
and then the driver disagreeing.

One situation where this could occur is if a job incremented a
syncpoint too many times -- then the hardware would trigger an
interrupt, but the driver would assume that a syncpoint value
greater than the syncpoint's max value is in the future, and not
clean up the job.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Remove cancelled waiters immediately

Before this patch, cancelled waiters would only be cleaned up
once their threshold value was reached. Make host1x_intr_put_ref
process the cancellat

gpu: host1x: Remove cancelled waiters immediately

Before this patch, cancelled waiters would only be cleaned up
once their threshold value was reached. Make host1x_intr_put_ref
process the cancellation immediately to fix this.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Allow syncpoints without associated client

Syncpoints don't need to be associated with any client,
so remove the property, and expose host1x_syncpt_alloc.
This will allow allocating syn

gpu: host1x: Allow syncpoints without associated client

Syncpoints don't need to be associated with any client,
so remove the property, and expose host1x_syncpt_alloc.
This will allow allocating syncpoints without prior knowledge
of the engine that it will be used with.

Signed-off-by: Mikko Perttunen <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Rename "parent" to "host"

Rename the host1x clients' parent to "host" because that more closely
describes what it is. The parent can be confused with the parent device
in terms of the d

gpu: host1x: Rename "parent" to "host"

Rename the host1x clients' parent to "host" because that more closely
describes what it is. The parent can be confused with the parent device
in terms of the device hierarchy. Subsequent patches will add a new
member that refers to the parent in that hierarchy.

Signed-off-by: Thierry Reding <[email protected]>

show more ...

treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 201

Based on 1 normalized pattern(s):

this program is free software you can redistribute it and or modify
it under the terms and c

treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 201

Based on 1 normalized pattern(s):

this program is free software you can redistribute it and or modify
it under the terms and conditions of the gnu general public license
version 2 as published by the free software foundation this program
is distributed in the hope 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 228 file(s).

Signed-off-by: Thomas Gleixner <[email protected]>
Reviewed-by: Allison Randal <[email protected]>
Reviewed-by: Steve Winslow <[email protected]>
Reviewed-by: Richard Fontana <[email protected]>
Reviewed-by: Alexios Zavras <[email protected]>
Cc: [email protected]
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>

show more ...

gpu: host1x: Drop unnecessary host1x argument

Functions taking a pointer to a host1x syncpoint as an argument don't
need to specify a pointer to a host1x instance because it can be
obtained from the

gpu: host1x: Drop unnecessary host1x argument

Functions taking a pointer to a host1x syncpoint as an argument don't
need to specify a pointer to a host1x instance because it can be
obtained from the syncpoint.

Reviewed-by: Dmitry Osipenko <[email protected]>
Tested-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Cleanup loop variable usage

Use unsigned int where possible and don't unnecessarily initialize the
loop variable.

Reviewed-by: Dmitry Osipenko <[email protected]>
Tested-by: Dmitry Osip

gpu: host1x: Cleanup loop variable usage

Use unsigned int where possible and don't unnecessarily initialize the
loop variable.

Reviewed-by: Dmitry Osipenko <[email protected]>
Tested-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Remove wait check support

The job submission userspace ABI doesn't support this and there are no
plans to implement it, so all of this code is dead and can be removed.

Reviewed-by: Dmi

gpu: host1x: Remove wait check support

The job submission userspace ABI doesn't support this and there are no
plans to implement it, so all of this code is dead and can be removed.

Reviewed-by: Dmitry Osipenko <[email protected]>
Tested-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Enable Tegra186 syncpoint protection

Since Tegra186 the Host1x hardware allows syncpoints to be assigned to
specific channels, preventing any other channels from incrementing
them.

Ena

gpu: host1x: Enable Tegra186 syncpoint protection

Since Tegra186 the Host1x hardware allows syncpoints to be assigned to
specific channels, preventing any other channels from incrementing
them.

Enable this feature where available and assign syncpoints to channels
when submitting a job. Syncpoints are currently never unassigned from
channels since that would require extra work and is unnecessary with
the current channel allocation model.

Signed-off-by: Mikko Perttunen <[email protected]>
Reviewed-by: Dmitry Osipenko <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: syncpt: Request syncpoints per client

Rather than request syncpoints for a struct device *, request them for a
struct host1x_client *. This is important because subsequent patches are
g

gpu: host1x: syncpt: Request syncpoints per client

Rather than request syncpoints for a struct device *, request them for a
struct host1x_client *. This is important because subsequent patches are
going to break the assumption that host1x will always be the parent for
devices requesting a syncpoint. It's also a more natural choice because
host1x clients are really the only ones that will know how to deal with
syncpoints.

Note that host1x clients are always guaranteed to be children of host1x,
regardless of their location in the device tree.

Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Flesh out kerneldoc

Improve kerneldoc for the public parts of the host1x infrastructure in
preparation for adding driver-specific part to the GPU documentation.

Acked-by: Daniel Vetter

gpu: host1x: Flesh out kerneldoc

Improve kerneldoc for the public parts of the host1x infrastructure in
preparation for adding driver-specific part to the GPU documentation.

Acked-by: Daniel Vetter <[email protected]>
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Fix potential out-of-bounds access

The check for valid syncpoint IDs is off by one. While at it, rewrite
the check to make it more easily understandable.

Signed-off-by: Thierry Reding

gpu: host1x: Fix potential out-of-bounds access

The check for valid syncpoint IDs is off by one. While at it, rewrite
the check to make it more easily understandable.

Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Add locking to syncpt

Currently syncpoints are not locked by mutex and this causes races
if we are aggressively freeing and allocating syncpoints.

This patch adds missing mutex protect

gpu: host1x: Add locking to syncpt

Currently syncpoints are not locked by mutex and this causes races
if we are aggressively freeing and allocating syncpoints.

This patch adds missing mutex protection to syncpoint structures.

Signed-off-by: Arto Merilainen <[email protected]>
Reviewed-by: Shridhar Rasal <[email protected]>
Signed-off-by: Mikko Perttunen <[email protected]>
[[email protected]: use better label names, don't reset local variable]
Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Remove useless local variable

The local 'val' variable is used to store a value and immediately return
it to its caller, and hence serves no purpose. Just drop it and directly
return th

gpu: host1x: Remove useless local variable

The local 'val' variable is used to store a value and immediately return
it to its caller, and hence serves no purpose. Just drop it and directly
return the value.

Signed-off-by: Thierry Reding <[email protected]>

show more ...

gpu: host1x: Whitespace cleanup for readability

Insert a number of blank lines in places where they increase readability
of the code. Also collapse various variable declarations to shorten some
func

gpu: host1x: Whitespace cleanup for readability

Insert a number of blank lines in places where they increase readability
of the code. Also collapse various variable declarations to shorten some
functions and finally rewrite some code for readability.

Signed-off-by: Thierry Reding <[email protected]>

show more ...