Commit Graph

236 Commits

Author SHA1 Message Date
Thierry Reding
c8a2036474 gpu: host1x: Unconditionally select IOMMU_IOVA
Currently configurations can be generated where IOMMU_SUPPORT is
disabled but IOMMU_IOVA is built as a module and HOST1X as built-in. In
such a case, the symbols guarded by IOMMU_IOVA will not be available
when linking the host1x driver and cause a linking failure.

Simplify this by unconditionally selecting IOMMU_IOVA, which makes sure
that it will be forced to =y if HOST1X=y. Technically we can now get
IOMMU_IOVA code built-in even if we don't use it (host1x only uses it
when IOMMU_SUPPORT is also enabled), but such configuration are of a
mostly academic nature. In all practical configurations we want IOMMU
support anyway.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-11-01 10:49:17 +01:00
Thierry Reding
06867a362d gpu: host1x: Set DMA mask based on IOMMU setup
If the Tegra DRM clients are backed by an IOMMU, push buffers are likely
to be allocated beyond the 32-bit boundary if sufficient system memory
is available. This is problematic on earlier generations of Tegra where
host1x supports a maximum of 32 address bits for the GATHER opcode. More
recent versions of Tegra (Tegra186 and later) have a wide variant of the
GATHER opcode, which allows addressing up to 64 bits of memory.

If host1x itself is behind an IOMMU as well this doesn't matter because
the IOMMU's input address space is restricted to 32 bits on generations
without support for wide GATHER opcodes.

However, if host1x is not behind an IOMMU, it won't be able to process
push buffers beyond the 32-bit boundary on Tegra generations that don't
support wide GATHER opcodes. Restrict the DMA mask to 32 bits on these
generations prevents buffers from being allocated from beyond the 32-bit
boundary.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-29 15:04:35 +01:00
Thierry Reding
af1cbfb9bf gpu: host1x: Support DMA mapping of buffers
If host1x_bo_pin() returns an SG table, create a DMA mapping for the
buffer. For buffers that the host1x client has already mapped itself,
host1x_bo_pin() returns NULL and the existing DMA address is used.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-29 15:04:35 +01:00
Thierry Reding
b78e70c04c gpu: host1x: Allocate gather copy for host1x
Currently when the gather buffers are copied, they are copied to a
buffer that is allocated for the host1x client that wants to execute the
command streams in the buffers. However, the gather buffers will be read
by the host1x device, which causes SMMU faults if the DMA API is backed
by an IOMMU.

Fix this by allocating the gather buffer copy for the host1x device,
which makes sure that it will be mapped into the host1x's IOVA space if
the DMA API is backed by an IOMMU.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-29 15:04:35 +01:00
Thierry Reding
44156eee91 gpu: host1x: Clean up debugfs on removal
The debugfs files created for host1x are never removed, causing these
files to be left dangling in debugfs. This results in a crash when any
of these files are accessed after the host1x driver has been removed,
as well as a failure to create the debugfs entries when they are added
again on driver probe.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-29 15:04:34 +01:00
Thierry Reding
80327ce3d4 gpu: host1x: Overhaul host1x_bo_{pin,unpin}() API
The host1x_bo_pin() and host1x_bo_unpin() APIs are used to pin and unpin
buffers during host1x job submission. Pinning currently returns the SG
table and the DMA address (an IOVA if an IOMMU is used or a physical
address if no IOMMU is used) of the buffer. The DMA address is only used
for buffers that are relocated, whereas the host1x driver will map
gather buffers into its own IOVA space so that they can be processed by
the CDMA engine.

This approach has a couple of issues. On one hand it's not very useful
to return a DMA address for the buffer if host1x doesn't need it. On the
other hand, returning the SG table of the buffer is suboptimal because a
single SG table cannot be shared for multiple mappings, because the DMA
address is stored within the SG table, and the DMA address may be
different for different devices.

Subsequent patches will move the host1x driver over to the DMA API which
doesn't work with a single shared SG table. Fix this by returning a new
SG table each time a buffer is pinned. This allows the buffer to be
referenced by multiple jobs for different engines.

Change the prototypes of host1x_bo_pin() and host1x_bo_unpin() to take a
struct device *, specifying the device for which the buffer should be
pinned. This is required in order to be able to properly construct the
SG table. While at it, make host1x_bo_pin() return the SG table because
that allows us to return an ERR_PTR()-encoded error code if we need to,
or return NULL to signal that we don't need the SG table to be remapped
and can simply use the DMA address as-is. At the same time, returning
the DMA address is made optional because in the example of command
buffers, host1x doesn't need to know the DMA address since it will have
to create its own mapping anyway.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-29 15:04:34 +01:00
Ben Dooks (Codethink)
33904487f1 gpu: host1x: Make host1x_cdma_wait_pushbuffer_space() static
The host1x_cdma_wait_pushbuffer_space() function is not declared or
directly called from outside the file it is in, so make it static.

Fixes the following sparse warning:

    drivers/gpu/host1x/cdma.c:235:5: warning: symbol 'host1x_cdma_wait_pushbuffer_space' was not declared. Should it be static?

Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-28 11:18:34 +01:00
Thierry Reding
caccddcfc4 gpu: host1x: Request channels for clients, not devices
A struct device doesn't carry much information that a channel might be
interested in, but the client very much does. Request channels for the
clients rather than their parent devices and store a pointer to them
in order to have that information available when needed.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-28 11:18:33 +01:00
Thierry Reding
8f45f5071a gpu: host1x: Explicitly initialize host1x_info structures
It's technically not required to explicitly initialize the fields that
will be zero by default, but it's easier to read these structures if
they are all initialized uniformly.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-28 11:18:33 +01:00
Thierry Reding
b9cd7b954a gpu: host1x: Remove gratuitous blank line
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-28 11:18:33 +01:00
Thierry Reding
d98914ebc2 gpu: host1x: Do not limit DMA segment size
host1x nor any its clients have any limitations on the DMA segment size,
so don't pretend that they do.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-10-28 11:18:08 +01:00
Dave Airlie
dfd03396d7 drm/tegra: Changes for v5.3-rc1
This contains a couple of small improvements and cleanups for the Tegra
 DRM driver.
 -----BEGIN PGP SIGNATURE-----
 
 iQJHBAABCAAxFiEEiOrDCAFJzPfAjcif3SOs138+s6EFAl0M8jsTHHRyZWRpbmdA
 bnZpZGlhLmNvbQAKCRDdI6zXfz6zoU0MEACZhCSKLvinuCw2pWdsc2A5rZN3r0Mr
 9TpdIKXYwcHO2G2gYC1mXv0wlp2pzIOr5y3VHLl1VboWXlPZjvgH0OiDtmwZk/vU
 jjFOGZO5ZjwyVNAZ8SlQrpdBNIAm3wseVfrhVURIL+9dANDgXsxWiqTkgnEmVclZ
 Kj2SpXfysTM9TJyXfQW8op6jKtP3NJ/IPEtTguNL1R9ho0phlcRYBUMEIYqtgBVE
 aRsjrrbM27OGf+JFPY3C7bW90hzgVqZLeK9R4AAMPS8iZ91k9njlFZ20LI9Yk/P1
 GBVyemiRtcG3owbli3NlfFzaeNjmM2PmcMZJsOyf6T+UH5juH2XZ/TV5O6i/jq0z
 /8DWsuFdEMX5tdP0t4B3vnbGQTYMOo/6bRsZHceLXcpKFNuJcC6lY233/Fc3D1bw
 gWm5ZHTs3SmPm2JoWmA54h2oiXU8/hGiPZGoUDHUTxf/h7DOGhK2hfaNdB6ZYXJu
 be4yS5TidFlFKi911JuXblCDeFf0VPsOfemQtJW4OvKg9FD6WmRuehYPytJ8ifB1
 dByXh5siOVMcyE0a2bsMPAvsxK6z4pPwyDz34AJUETJ0DSmSfWeEMSbCmpJwt6m5
 35IiMossSULJXWjqtc5bcK1gmbYMJSEwj+Xe33O+H67THSgo4GLxqKdb++5QK5Ky
 svluhQIjYYx3pw==
 =T2hU
 -----END PGP SIGNATURE-----

Merge tag 'drm/tegra/for-5.3-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/tegra: Changes for v5.3-rc1

This contains a couple of small improvements and cleanups for the Tegra
DRM driver.

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Thierry Reding <thierry.reding@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190621150753.19550-1-thierry.reding@gmail.com
2019-06-25 12:59:43 +10:00
Greg Kroah-Hartman
eb7cf945a8 host1x: debugfs_create_dir() can never return NULL
So there is no need to check for a value that can never happen.  No need
to check the return value all anyway, as any debugfs call can take the
result of this function as an option just fine.

Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-tegra@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-06-13 18:23:38 +02:00
Thomas Gleixner
9c92ab6191 treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 282
Based on 1 normalized pattern(s):

  this software is licensed under the terms of the gnu general public
  license version 2 as published by the free software foundation and
  may be copied distributed and modified under those terms 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

extracted by the scancode license scanner the SPDX license identifier

  GPL-2.0-only

has been chosen to replace the boilerplate/reference in 285 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>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190529141900.642774971@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-06-05 17:36:37 +02:00
Thierry Reding
31fa25f100 gpu: host1x: Do not link logical devices to DT nodes
Logical devices created by the host1x bus infrastructure don't need to
be associated with a device tree node. Doing so will cause the driver
core to attempt to hook up IOMMU operations and fail because it is not
a real device.

However, for backwards-compatibility, we need to provide various OF_*
uevent variables that were previously provided by of_device_uevent() and
which are parsed by libdrm in userspace when querying the available
devices. Do this by implementing a uevent callback for the host1x bus.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-06-05 15:06:03 +02:00
Thierry Reding
1e390478cf gpu: host1x: Increase maximum DMA segment size
Recent versions of the DMA API debug code have started to warn about
violations of the maximum DMA segment size. This is because the segment
size defaults to 64 KiB, which can easily be exceeded in large buffer
allocations such as used in DRM/KMS for framebuffers.

Technically the Tegra SMMU and ARM SMMU don't have a maximum segment
size (they map individual pages irrespective of whether they are
contiguous or not), so the choice of 4 MiB is a bit arbitrary here. The
maximum segment size is a 32-bit unsigned integer, though, so we can't
set it to the correct maximum size, which would be the size of the
aperture.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-06-05 15:06:03 +02:00
Thierry Reding
4bb923e807 gpu: host1x: Do not output error message for deferred probe
When deferring probe, avoid logging a confusing error message. While at
it, make the error message more informational.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-06-05 15:05:49 +02:00
Thomas Gleixner
9952f6918d 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 <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Steve Winslow <swinslow@gmail.com>
Reviewed-by: Richard Fontana <rfontana@redhat.com>
Reviewed-by: Alexios Zavras <alexios.zavras@intel.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190528171438.107155473@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-05-30 11:29:52 -07:00
Thomas Gleixner
ec8f24b7fa treewide: Add SPDX license identifier - Makefile/Kconfig
Add SPDX license identifiers to all Make/Kconfig files which:

 - Have no license information of any form

These files fall under the project license, GPL v2 only. The resulting SPDX
license identifier is:

  GPL-2.0-only

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-05-21 10:50:46 +02:00
Arnd Bergmann
8bbad1ba31 gpu: host1x: Program stream ID to bypass without SMMU
If SMMU support is not available, fall back to programming the bypass
stream ID (0x7f).

Fixes: de5469c21f ("gpu: host1x: Program the channel stream ID")
Suggested-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
[treding@nvidia.com: rebase this on top of a later build fix]
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-04-11 17:40:35 +02:00
Stefan Agner
e154592a1d gpu: host1x: Fix compile error when IOMMU API is not available
In case the IOMMU API is not available compiling host1x fails with
the following error:

  In file included from drivers/gpu/host1x/hw/host1x06.c:27:
  drivers/gpu/host1x/hw/channel_hw.c: In function ‘host1x_channel_set_streamid’:
  drivers/gpu/host1x/hw/channel_hw.c:118:30: error: implicit declaration of function
    ‘dev_iommu_fwspec_get’; did you mean ‘iommu_fwspec_free’?  [-Werror=implicit-function-declaration]
  struct iommu_fwspec *spec = dev_iommu_fwspec_get(channel->dev->parent);
                              ^~~~~~~~~~~~~~~~~~~~
                              iommu_fwspec_free

Fixes: de5469c21f ("gpu: host1x: Program the channel stream ID")
Signed-off-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-04-11 10:35:39 +02:00
Dmitry Osipenko
79930bafe2 gpu: host1x: Continue CDMA execution starting with a next job
Currently gathers of a hung job are getting NOP'ed and a restarted CDMA
executes the NOP'ed gathers. There shouldn't be a reason to not restart
CDMA execution starting with a next job, avoiding the unnecessary churning
with gathers NOP'ing.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:34:25 +01:00
Dmitry Osipenko
5d6f043685 gpu: host1x: Don't complete a completed job
There is a chance that the last job has been completed at the time of
CDMA timeout handler invocation. In this case there is no need to complete
the completed job.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:30:58 +01:00
Dmitry Osipenko
e8bad65938 gpu: host1x: Cancel only job that actually got stuck
Host1x doesn't have information about jobs inter-dependency, that is
something that will become available once host1x will get a proper
jobs scheduler implementation. Currently a hang job causes other unrelated
jobs to be canceled, that is a relic from downstream driver which is
irrelevant to upstream. Let's cancel only the hanging job and not to touch
other jobs in queue.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:30:24 +01:00
Thierry Reding
e1f338c0f8 gpu: host1x: Optimize CDMA push buffer memory usage
The host1x CDMA push buffer is terminated by a special opcode (RESTART)
that tells the CDMA to wrap around to the beginning of the push buffer.
To accomodate the RESTART opcode, an extra 4 bytes are allocated on top
of the 512 * 8 = 4096 bytes needed for the 512 slots (1 slot = 2 words)
that are used for other commands passed to CDMA. This requires that two
memory pages are allocated, but most of the second page (4092 bytes) is
never used.

Decrease the number of slots to 511 so that the RESTART opcode fits
within the page. Adjust the push buffer wraparound code to take into
account push buffer sizes that are not a power of two.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:59 +01:00
Thierry Reding
0e43b8da15 gpu: host1x: Use correct semantics for HOST1X_CHANNEL_DMAEND
The HOST1X_CHANNEL_DMAEND is an offset relative to the value written to
the HOST1X_CHANNEL_DMASTART register, but it is currently treated as an
absolute address. This can cause SMMU faults if the CDMA fetches past a
pushbuffer's IOMMU mapping.

Properly setting the DMAEND prevents the CDMA from fetching beyond that
address and avoid such issues. This is currently not observed because a
whole (almost) page of essentially scratch space absorbs any excessive
prefetching by CDMA. However, changing the number of slots in the push
buffer can trigger these SMMU faults.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:58 +01:00
Thierry Reding
8de896eb20 gpu: host1x: Support 40-bit addressing on Tegra186
The host1x and clients instantiated on Tegra186 support addressing 40
bits of memory.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:58 +01:00
Thierry Reding
38fabcc953 gpu: host1x: Restrict IOVA space to DMA mask
On Tegra186 and later, the ARM SMMU provides an input address space that
is 48 bits wide. However, memory clients can only address up to 40 bits.
If the geometry is used as-is, allocations of IOVA space can end up in a
region that is not addressable by the memory clients.

To fix this, restrict the IOVA space to the DMA mask of the host1x
device.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:57 +01:00
Thierry Reding
67a82dbc0a gpu: host1x: Support 40-bit addressing
Tegra186 and later support 40 bits of address space. Additional
registers need to be programmed to store the full 40 bits of push
buffer addresses.

Since command stream gathers can also reside in buffers in a 40-bit
address space, a new variant of the GATHER opcode is also introduced.
It takes two parameters: the first parameter contains the lower 32
bits of the address and the second parameter contains bits 32 to 39.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:35 +01:00
Thierry Reding
5a5fccbd8c gpu: host1x: Introduce support for wide opcodes
The CDMA push buffer can currently only handle opcodes that take a
single word parameter. However, the host1x implementation on Tegra186
and later supports opcodes that require multiple words as parameters.

Unfortunately the way the push buffer is structured, these wide opcodes
cannot simply be composed of two regular opcodes because that could
result in the wide opcode being split across the end of the push buffer
and the final RESTART opcode required to wrap the push buffer around
would break the wide opcode.

One way to fix this would be to remove the concept of slots to simplify
push buffer operations. However, that's not entirely trivial and should
be done in a separate patch. For now, simply use a different function
to push four-word opcodes into the push buffer. Technically only three
words are pushed, with the fourth word used as padding to preserve the
2-word alignment required by the slots abstraction. The fourth word is
always a NOP opcode.

Additional care must be taken when the end of the push buffer is
reached. If a four-word opcode doesn't fit into the push buffer without
being split by the boundary, NOP opcodes will be introduced and the new
wide opcode placed at the beginning of the push buffer.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:35 +01:00
Thierry Reding
de5469c21f gpu: host1x: Program the channel stream ID
When processing command streams, make sure the host1x's stream ID is
programmed for the channel so that addresses are properly translated
through the SMMU.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-07 18:28:33 +01:00
Thierry Reding
6841482b82 gpu: host1x: Set up stream ID table
In order to enable the MMIO path stream ID protection provided by the
incarnation of host1x found in Tegra186 and later, the host1x must be
provided with the list of stream ID register offsets for each of its
clients. Some clients (such as VIC) have multiple stream ID registers
that are assumed to be contiguous. The host1x is programmed with the
base offset and a limit which provide the range of registers that the
host1x needs to monitor for writes.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-04 08:35:55 +01:00
Thierry Reding
f67524caf4 gpu: host1x: Represent host1x bus devices in debugfs
This new debugfs file represents the state of host1x bus devices,
specifying the list of subdevices and marking which ones have
successfully registered.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-04 08:35:54 +01:00
Arnd Bergmann
0747a672a3 gpu: host1x: Use completion instead of semaphore
In this usage, the two are completely equivalent, but the completion
documents better what is going on, and we generally try to avoid
semaphores these days.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2019-02-04 08:35:54 +01:00
Thierry Reding
ac1bdbf22b gpu: host1x: Add Tegra194 support
The host1x hardware found on Tegra194 is mostly backwards compatible
with the version found on Tegra186, with the notable exceptions of the
increased number of syncpoints and mlocks. In addition, some rarely
used features such as syncpoint wait bases were dropped and some
registers had to move around to accomodate the increased number of
syncpoints.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-11-29 17:11:49 +01:00
Thierry Reding
2fc777ba84 gpu: host1x: Fix syncpoint ID field size on Tegra186
The number of syncpoints on Tegra186 is 576 and therefore no longer fits
into 8 bits. Increase the size of the syncpoint ID field to 10 in order
to accomodate all syncpoints.

Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-11-27 17:18:39 +01:00
Thierry Reding
b7c61d511d gpu: host1x: Resize channel register region on Tegra186 and later
The register region allocated per channel was decreased from 16384 bytes
to 256 bytes on Tegra186 and later. Resize the region to make sure every
channel (instead of only the first) is properly programmed.

Suggested-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-11-27 17:18:26 +01:00
Dmitry Osipenko
e31c8ea5af gpu: host1x: Detach Host1x from IOMMU DMA domain on arm32
Host1x is getting attached to an implicit IOMMU DMA domain if
CONFIG_ARM_DMA_USE_IOMMU=y. Since Host1x driver manages IOMMU by
itself, Host1x device must be detached from the implicit domain using
arch-specific IOMMU-API.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-09-26 17:11:14 +02:00
Thierry Reding
50bac83c80 gpu: host1x: Remove spurious tab
All other assignments have a single space around the = sign, so remove
the spurious tab for consistency.

Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-09-26 16:09:46 +02:00
Dmitry Osipenko
ec58923215 gpu: host1x: Check whether size of unpin isn't 0
Only gather pins are mapped by the Host1x driver, regular BO relocations
are not. Check whether size of unpin isn't 0, otherwise IOVA allocation at
0x0 could be erroneously released.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-07-09 10:31:30 +02:00
Dmitry Osipenko
4466b1f0e0 gpu: host1x: Skip IOMMU initialization if firewall is enabled
Host1x's CDMA can't access the command buffers if IOMMU and Host1x
firewall are enabled in the kernels config because firewall doesn't map
the copied buffer into IOVA space. Fix this by skipping IOMMU
initialization if firewall is enabled as firewall merges sparse cmdbufs
into a single contiguous buffer and hence IOMMU isn't needed in this case.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-07-09 10:31:10 +02:00
Linus Torvalds
135c5504a6 drm for v4.18-rc1
-----BEGIN PGP SIGNATURE-----
 
 iQIcBAABAgAGBQJbFzauAAoJEAx081l5xIa+jzwP/AwfTreH3XBmlLeMO8kwkAMW
 AiH1u3KHrlitN1U85x90dC8hVuuV2Kv9pEXQo1me/TAL/QCb9VZYCSf2dHHkkMwD
 1UdwxAQW7soBbTRo9k0zuVJpRGSYIiYfqDh3L6KmTC3UF0tUJm53VOWCLG/DNrtn
 XhEnGnlCUNbABZMMpavEvlwxPtvaYFlp6M+MmwRPIx32SrPJ3vSaBzxwWxOaFmjU
 xiRdK+GpAbCV8yGeCCSkDgEe/TdWGUhZxoC9dLb0H9ex7ip8uZ0W4D+VTHPFrhQX
 6nCpqUbp7BQTsbVSd1pAVsAv45scmSgWbKcqfC0NKSVcsHcJZBR0tQOF9OvnGZcf
 o1Hv/beTqJ++IcG2rEIwyJTGxAGfZ0YSb0evTC9VcszaYo+b3+G283bdztIjzDeS
 0QCTeLHYbZRHPITWVULNpMWy3TkJv32IdFhQfYSnD8/OGQIxLNhh4FFOtHnOmxSF
 N8dnzOLKXhXMo/NgOL+UMNnbgLqIyOtEXCPDLuOQJNv/SOp8662m/A0yRjQNR6M2
 gsPmR7dxQIwwJMyqrkLDOF411ABZohulquYgwLgG938MRPmTpPWOR72PtpGF4hAW
 HLg+3HHBd1N/A1mlJUMAbUn2eMUACZBUIycE9u+U/geRgve/OQnzJH/FKGP2EJ4R
 pf6CruEva+6GRR5GVzuM
 =twst
 -----END PGP SIGNATURE-----

Merge tag 'drm-next-2018-06-06-1' of git://anongit.freedesktop.org/drm/drm

Pull drm updates from Dave Airlie:
 "This starts to support NVIDIA volta hardware with nouveau, and adds
  amdgpu support for the GPU in the Kabylake-G (the intel + radeon
  single package chip), along with some initial Intel icelake enabling.

  Summary:

  New Drivers:
   - v3d - driver for broadcom V3D V3.x+ hardware
   - xen-front - XEN PV display frontend

  core:
   - handle zpos normalization in the core
   - stop looking at legacy pointers in atomic paths
   - improved scheduler documentation
   - improved aspect ratio validation
   - aspect ratio support for 64:27 and 256:135
   - drop unused control node code.

  i915:
   - Icelake (ICL) enabling
   - GuC/HuC refactoring
   - PSR/PSR2 enabling and fixes
   - DPLL management refactoring
   - DP MST fixes
   - NV12 enabling
   - HDCP improvements
   - GEM/Execlist/reset improvements
   - GVT improvements
   - stolen memory first 4k fix

  amdgpu:
   - Vega 20 support
   - VEGAM support (Kabylake-G)
   - preOS scanout buffer reservation
   - power management gfxoff support for raven
   - SR-IOV fixes
   - Vega10 power profiles and clock voltage control
   - scatter/gather display support on CZ/ST

  amdkfd:
   - GFX9 dGPU support
   - userptr memory mapping

  nouveau:
   - major refactoring for Volta GV100 support

  tda998x:
   - HDMI i2c CEC support

  etnaviv:
   - removed unused logging code
   - license text cleanups
   - MMU handling improvements
   - timeout fence fix for 50 days uptime

  tegra:
   - IOMMU support in gr2d/gr3d drivers
   - zpos support

  vc4:
   - syncobj support
   - CTM, plane alpha and async cursor support

  analogix_dp:
   - HPD and aux chan fixes

  sun4i:
   - MIPI DSI support

  tilcdc:
   - clock divider fixes for OMAP-l138 LCDK board

  rcar-du:
   - R8A77965 support
   - dma-buf fences fixes
   - hardware indexed crtc/du group handling
   - generic zplane property support

  atmel-hclcdc:
   - generic zplane property support

  mediatek:
   - use generic video mode function

  exynos:
   - S5PV210 FIMD variant support
   - IPP v2 framework
   - more HW overlays support"

* tag 'drm-next-2018-06-06-1' of git://anongit.freedesktop.org/drm/drm: (1286 commits)
  drm/amdgpu: fix 32-bit build warning
  drm/exynos: fimc: signedness bug in fimc_setup_clocks()
  drm/exynos: scaler: fix static checker warning
  drm/amdgpu: Use dev_info() to report amdkfd is not supported for this ASIC
  drm/amd/display: Remove use of division operator for long longs
  drm/amdgpu: Update GFX info structure to match what vega20 used
  drm/amdgpu/pp: remove duplicate assignment
  drm/sched: add rcu_barrier after entity fini
  drm/amdgpu: move VM BOs on LRU again
  drm/amdgpu: consistenly use VM moved flag
  drm/amdgpu: kmap PDs/PTs in amdgpu_vm_update_directories
  drm/amdgpu: further optimize amdgpu_vm_handle_moved
  drm/amdgpu: cleanup amdgpu_vm_validate_pt_bos v2
  drm/amdgpu: rework VM state machine lock handling v2
  drm/amdgpu: Add runtime VCN PG support
  drm/amdgpu: Enable VCN static PG by default on RV
  drm/amdgpu: Add VCN static PG support on RV
  drm/amdgpu: Enable VCN CG by default on RV
  drm/amdgpu: Add static CG control for VCN on RV
  drm/exynos: Fix default value for zpos plane property
  ...
2018-06-06 08:16:33 -07:00
Thierry Reding
326bbd79fd gpu: host1x: Use not explicitly sized types
The number of words and the offset in a gather don't need to be
explicitly sized, so make them unsigned int instead.

Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:51:37 +02:00
Thierry Reding
06490bb99e gpu: host1x: Rename relocarray -> relocs for consistency
All other array variables use a plural, and this is the only one using
the *array suffix. This is confusing, so rename it for consistency.

Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:51:25 +02:00
Thierry Reding
ac330f45c7 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 <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:51:01 +02:00
Thierry Reding
d4ad3ad9b8 gpu: host1x: Cleanup loop variable usage
Use unsigned int where possible and don't unnecessarily initialize the
loop variable.

Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:50:40 +02:00
Thierry Reding
bf3d41ccab gpu: host1x: Store pointer to client in jobs
Rather than storing some identifier derived from the application
context that can't be used concretely anywhere, store a pointer to the
client directly so that accesses can be made directly through that
client object.

Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:50:24 +02:00
Thierry Reding
24c94e166d 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 <digetx@gmail.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 21:50:04 +02:00
Emil Goode
2f8a6da866 gpu: host1x: Fix compiler errors by converting to dma_addr_t
The compiler is complaining with the following errors:

drivers/gpu/host1x/cdma.c:94:48: error:
	passing argument 3 of ‘dma_alloc_wc’ from incompatible pointer type
	[-Werror=incompatible-pointer-types]

drivers/gpu/host1x/cdma.c:113:48: error:
	passing argument 3 of ‘dma_alloc_wc’ from incompatible pointer type
	[-Werror=incompatible-pointer-types]

The expected pointer type of the third argument to dma_alloc_wc() is
dma_addr_t but phys_addr_t is passed.

Change the phys member of struct push_buffer to be dma_addr_t so that we
pass the correct type to dma_alloc_wc().
Also check pb->mapped for non-NULL in the destroy function as that is the
right way of checking if dma_alloc_wc() was successful.

Signed-off-by: Emil Goode <emil.fsw@goode.io>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-18 11:23:48 +02:00
Thierry Reding
f40e1590c5 gpu: host1x: Acquire a reference to the IOVA cache
The IOVA API uses a memory cache to allocate IOVA nodes from. To make
sure that this cache is available, obtain a reference to it and release
the reference when the cache is no longer needed.

On 64-bit ARM this is hidden by the fact that the DMA mapping API gets
that reference and never releases it. On 32-bit ARM, this is papered
over by the Tegra DRM driver (the sole user of the host1x API requiring
the cache) acquiring a reference to the IOVA cache for its own purposes.
However, there may be additional users of this API in the future, so fix
this upfront to avoid surprises.

Fixes: 404bfb78da ("gpu: host1x: Add IOMMU support")
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
2018-05-17 17:44:48 +02:00