amdgpu: fix "add AMDGPU_VA_RANGE_HIGH"

The range is stored as exclusive, not inclusive. Subtracts one to get
the inclusive interval for the calculation. This fixes crashes when 32bit
addresses are in use.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>

amdgpu: add AMDGPU_VA_RANGE_HIGH

Return high addresses if requested and available.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

amdgpu: mostly revert "use the high VA range if possible v2"

This reverts commit 07ea20d5beb24315b721adf83bbfa72ce016e146.

Unfortunately it turned out that this change broke some corner cases in
Mesa.

Revert it for now, but keep the high range in separate VA managers.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

freedreno: add interface to get buffer address

Needed for clover/OpenCL.  Fortunately the kernel interface is already
in place.

Include a stub _put_iova() so mesa can tell us when it no longer needs
the buffer to be pinned.  There is no kernel interface for this (yet),
but at least if we want to unpin buffers we won't need mesa changes.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

*-symbol-check: Don't hard-code nm executable

Helpful if your nm executable has a prefix based on the
architecture, for example.

Signed-off-by: Heiko Becker <heirecka@exherbo.org>
Cc: Timo Gurr <timo.gurr@gmail.com>
[Eric: v2: rebase and add Meson support]
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

android: fix gralloc_handle_create() problems

There's a number of problems with gralloc_handle_create starting with it
doesn't even compile. More importantly, it doesn't really create (i.e.
allocate) a handle. It allocates a native_handle_t, copies it to a
struct gralloc_handle_t on the stack and returns the struct (not a ptr).
So the caller still has to allocate a struct gralloc_handle_t to hold
the returned struct.

Rework gralloc_handle_create() to allocate a new handle and return the
pointer to the allocated handle. Callers should free the handle with
native_handle_close() and native_handle_delete(). In the interest of
making gralloc_handle_t opaque, return a native_handle_t ptr instead.

Reviewed-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>

android: add helper to convert buffer_handle_t to gralloc_handle_t ptr

Clients frequently need to convert a buffer_handle_t (aka
native_handle_t *) to a gralloc_handle_t ptr. This is a simple cast, but
add an inline function to do the conversion.

Reviewed-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>

android: fix mis-named alloc_handle_t

Fix a typo where alloc_handle_t should be gralloc_handle_t. One still
remains in gralloc_handle_create, but a subsequent commit will fix that
along with other problems in gralloc_handle_create.

Reviewed-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>

android: revert making handle magic and version members const

Const members are problematic for dynamically allocating struct
gralloc_handle_t, so just drop the const modifier.

Reviewed-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>

meson/configure.ac: pthread-stubs not present on OpenBSD

pthread-stubs is no longer required on OpenBSD and has been removed.
libpthread parts involved moved to libc.

Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
[Eric: add meson equivalent]
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

meson: do not use cairo/valgrind if disabled

-Dcairo-tests=false currently results into enabling cairo support if it
was found. Same for valgrind.

v2:
* Use underscore-prefixed variables to not change type of variable
* Use empty array for "fake" dependency instead of real empty object

v3:
* Fix typo

Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Signed-off-by: Igor Gnatenko <ignatenko@redhat.com>

amdgpu: Fix mistake in initial hole size calculation.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>

drm/tegra: Sanitize format modifiers

The existing format modifier definitions were merged prematurely, and
recent work has unveiled that the definitions are suboptimal in several
ways:

  - The format specifiers, except for one, are not Tegra specific, but
    the names don't reflect that.
  - The number space is split into two, reserving 32 bits for some
    "parameter" which most of the modifiers are not going to have.
  - Symbolic names for the modifiers are not using the standard
    DRM_FORMAT_MOD_* prefix, which makes them awkward to use.
  - The vendor prefix NV is somewhat ambiguous.

Fortunately, nobody's started using these modifiers, so we can still fix
the above issues. Do so by using the standard prefix. Also, remove TEGRA
from the name of those modifiers that exist on NVIDIA GPUs as well. In
case of the block linear modifiers, make the "parameter" smaller (4
bits, though only 6 values are valid) and don't let that leak into any
of the other modifiers.

Finally, also use the more canonical NVIDIA instead of the ambiguous NV
prefix.

This is based on commit 5843f4e02fbe86a59981e35adc6cabebee46fdc0 from
Linux v4.16-rc1 and also updates modetest to use the new defines.

Acked-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>

drm/fourcc: Fix fourcc_mod_code() definition

Avoid compiler warnings when the val parameter is an expression.

This is based on commit 5843f4e02fbe86a59981e35adc6cabebee46fdc0 from
Linux v4.16-rc1.

Acked-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>

RELEASING: mention meson

meson: bump the version number

configure.ac: bump version to 2.4.90

android: Change gralloc_handle_t members to be fixed width

In order to lessen future alignment issues, lets switch to
fixed width integers where possible.

This excludes the data_owner since it is a pid_t which
in theory could be larger than 32 bits.

Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>

android: Remove member name from gralloc_handle_t

The name member of gralloc_handle_t is no longer needed and has been removed.
The version field has also been bumped.

Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>

android: Mark gralloc_handle_t magic variable as const

Mark magic member of gralloc_handle_t as const.

Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>

android: Add version variable to gralloc_handle_t

The version variable will be used for versioning of this
struct and the corresponding accessor functions.

Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>

android: Move gralloc handle struct to libdrm

This struct is used in mesa and drm_hwcomposer.
Versions of if have been implemented in several grallocs:
drm_gralloc, gbm_gralloc, minigbm and intel-minigbm.

Other than the 1:1 move of the struct a new generic name
has been chosen and variables have had comments added to them.

Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Rob Herring <robh@kernel.org>

amdgpu: Add amdgpu_query_sw_info to amdgpu-symbol-check

Fixes make check. Trivial.

amdgpu: add amdgpu_query_sw_info for querying high bits of 32-bit address space

Reviewed-by: Christian König <christian.koenig@amd.com>

meson: include headers in root directory in ext_libdrm

Which is used in wraps.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

amdgpu: clean up non list code path for vamgr v2

v2: Add missing "goto out"

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>

Revert "amdgpu: clean up non list code path for vamgr"

This reverts commit 41b94a3fb6e87d057fad78568d920d29489e5060.

It caused crashes with radeonsi in at least glxgears and Xorg.

tests/amdgpu: add bo eviction test

for(( i=1; i < 100; i++))
do
     echo "Hello, Welcome $i times "
     sudo ./amdgpu_test -s 1 -t 5
done

with above stricpt, run in two terminals, will reproduce Felix's swap leeking issue.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>

amdgpu: clean up non list code path for vamgr

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

amdgpu: fix inefficient vamgr algorithm

issue: UMD allocates top 4GB, but don't do anything, just reserve top 4GB space,
but the performance of VP13 drops from 162fps to 99fps.

root cause:
our va hole list of vamgr is too long by time going.

fix:
reusing old hole as much as possible can make the list shortest.

result:
performance recovers as non-list path, next patch will remove non-list code path.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

fix return value for syncobj wait

otherwise -ETIME is missed.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

drm: Fix 32-bit drmSyncobjWait.

Otherwise we get an EFAULT, at least on a 64-bit kernel.

Fixes: 2048a9e7 "drm: add drmSyncobjWait wrapper"
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

meson: fix libdrm_nouveau pkgconfig include directories

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

tests/amdgpu: add missing config.h include

Otherwise we'll end up without the macros set during configure stage.
And effectively error out in sanity tests such as the mmap static
assert.

To reproduce, do a multilib build - 32bit build on 64bit machine.

Cc: Fabio Pedretti <pedretti.fabio@gmail.com>
Cc: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Fixes: 33dcc29f7cc ("amdgpu: Add VMID reservation per GPU context test.")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104819
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

meson,configure: turn undefined preprocessor tokens warnings into errors

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

exynos/tests: use #ifdef for never-defined token

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

configure: always define HAVE_LIBDRM_ATOMIC_PRIMITIVES and HAVE_LIB_ATOMIC_OPS

Fixes #if undefined warnings

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

xf86drmHash: remove always-false #if guards

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: add warning when using undefined preprocessor tokens

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson: cleanup whitespace

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

xf86atomic: fix -Wundef warning

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson: sort HAVE_* defines

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

always define HAVE_VALGRIND

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

always define HAVE_CAIRO

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

always define HAVE_FREEDRENO_KGSL

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: always define HAVE_{INTEL,VMWGFX,NOUVEAU,EXYNOS,VC4,RADEON}

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: remove unused HAVE_ETNAVIV define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: remove unused HAVE_FREEDRENO define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: remove unused HAVE_TEGRA define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson,configure: remove unused HAVE_OMAP define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

configure: remove unused HAVE_INSTALL_TESTS define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

configure: remove unused HAVE_CUNIT define

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

meson: add missing HAVE_RADEON

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>

tests/etnaviv: drop unused `return 0`

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

tests/util: drop unused parameters

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

tests/util: fix signed/unsigned comparisons

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

tests/amdgpu: drop unused variables

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

tests/amdgpu: add parentheses to make operation priority explicit

While at it, align with the other half on the next line.

Cc: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

remove unnecessary double-semicolon

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

freedreno: clamp priority based on # of rings

In case of a kernel that is new enough to support multiple submit-
queues, but with an adreno generation which doesn't support multiple
prioritized ringbuffers, we'd attempt to open a submit-queue with
prio=1 (medium), which is rejected by the kernel.

This could happen either w/ an older mesa (which uses fd_pipe_new())
or a newer mesa which defaults to prio=1 if no pipe context priority
flags are set.

The simple answer to fix both cases is to clamp the requested priority
according to the number of rings.  This might not do exactly what you
want, if we hypothetically had 2 rings (it would result in requested
medium priority being high priority instead of low priority).  But the
number of rings (for hw gen's that support this) is purely a software
construct, so the easy answer there is to have the kernel advertise at
least 3 rings if it supports more than one.  There isn't really any
reason to do otherwise.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

amdgpu: Disable VM test suite by default for SI ASICs

Hangs my Cape Verde.

Acked-by: Christian König <christian.koenig@amd.com>

amdgpu: Disable deadlock test suite by default for SI ASICs

Hangs my Cape Verde.

Acked-by: Christian König <christian.koenig@amd.com>

amdgpu: Fix segfault in deadlock test.

If amdgpu_cs_query_fence_status terminates prematurely the BO
sometimes is unmapped before helper thread writes a vlaue
into it causing a segfault.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>

amdgpu: Update deadlock test to not assert on ECANCELED

Kernel will abort jobs for guilty (causing GPU hang) context
with -ECANCELED don't assert if that the case.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>

meson: set the minimum version correctly

Currently we ask for 0.42, but we actually require 0.43 because we pass
file objects as arguments to tests. If someone needs version 0.42 it
wouldn't be hard, just a lot of replacing files() with strings.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

meson: set proper pkg-config version for libdrm_freedreno

Copy and paste error from exynos.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

modetest: Fix to check return value of asprintf()

There is warning about ignoring return value of 'asprintf'. Fix to
check return value of asprintf().

Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

amdgpu: fix high VA mask

That constant needs to be 64bits.

Fixes: amdgpu: use the high VA range if possible v2

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

meson: fix the install path of amdgpu.ids

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>

amdgpu: Symlink .editorconfig to tests/amdgpu

In order to use consistent editorconfig settings in both amdgpu
directories.

Reviewed-by: Christian König <christian.koenig@amd.com>

amdgpu: Don't dereference device_handle after amdgpu_device_deinitialize

Fixes use after free:

==2537== Invalid read of size 4
==2537==    at 0x1162C9: suite_deadlock_tests_enable (deadlock_tests.c:101)
==2537==    by 0x10B157: amdgpu_disable_suits (amdgpu_test.c:421)
==2537==    by 0x10B157: main (amdgpu_test.c:560)
==2537==  Address 0x5e44f24 is 452 bytes inside a block of size 1,016 free'd
==2537==    at 0x4C2BE1B: free (vg_replace_malloc.c:530)
==2537==    by 0x504CD8B: amdgpu_device_reference (amdgpu_device.c:164)
==2537==    by 0x504CD8B: amdgpu_device_deinitialize (amdgpu_device.c:307)
==2537==    by 0x1162BB: suite_deadlock_tests_enable (deadlock_tests.c:97)
==2537==    by 0x10B157: amdgpu_disable_suits (amdgpu_test.c:421)
==2537==    by 0x10B157: main (amdgpu_test.c:560)
==2537==  Block was alloc'd at
==2537==    at 0x4C2CC05: calloc (vg_replace_malloc.c:711)
==2537==    by 0x504CA5E: amdgpu_device_initialize (amdgpu_device.c:212)
==2537==    by 0x116298: suite_deadlock_tests_enable (deadlock_tests.c:93)
==2537==    by 0x10B157: amdgpu_disable_suits (amdgpu_test.c:421)
==2537==    by 0x10B157: main (amdgpu_test.c:560)

Reviewed-by: Christian König <christian.koenig@amd.com>

amdgpu: Don't print error message if parse_one_line returned -EAGAIN

It means it just didn't find an entry for the GPU in the amdgpu.ids file.

Fixes spurious

 amdgpu_parse_asic_ids: Cannot parse ASIC IDs: Resource temporarily unavailable

error messages in that case.

Reported-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

README: Add note about meson

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

autotools: Include meson.build files in tarball

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

Add meson build system

This patch adds a complete meson build system, including tests and
install. It has the necessary hooks to allow it be used as a subproject
for other meson based builds such as mesa.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-and-tested-by: Igor Gnatenko <i.gnatenko.brain@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>

tests/amdgpu: execute copy linear on all the available rings

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: execute const fill on all the available rings

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: execute write linear on all the available rings

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

intel: Add more Coffeelake PCI IDs

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

test/amdgpu: fix compiler warnings

SWAP_32() should mask first and then shift.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

amdgpu: use the high VA range if possible v2

Retire the low range on Vega10 this frees up everything below 0xffff800000000000 for HMM.

v2: keep the 32bit range working.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

headers: sync up amdgpu_drm.h with drm-next

Sync up amdgpu changes from drm-next.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

amdgpu: fix 32bit VA manager max address

The range is exclusive not inclusive.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

amdgpu: fix not to add amdgpu.ids when building without amdgpu

The amdgpu.ids is only required when building with amdgpu support.
Fix not to add it without amdgpu.

Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

configure.ac: bump version for release

drm: Add CrtcGetSequence and CrtcQueueSequence IOCTLs [v2]

These provide a crtc-id based interface to get the current sequence
(frame) number and to queue an event to be delivered at a specific sequence.

v2: Remove FIRST_PIXEL_OUT flag. This has been removed from the
    proposed kernel API

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

drm: Add drm mode lease ioctl wrappers [v3]

drmModeCreateLease
drmModeListLessees
drmModeGetLease
drmModeRevokeLease

Changes for v2:

Remove lessee id from GetLease
Remove lessor_id from ListLeases
Add revoke
Renumber to track kernel rebase on drm-next

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

amdgpu: Add syncobj reset & signal wrappers.

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

drm: Add drmSyncobjReset & drmSyncobjSignal wrappers.

anv already uses the ioctls but does not use libdrm, so these were
not wrapped yet.

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

etnaviv: fix BO cache to properly work with different flags

Currently if the oldest BO in a bucket has different flags than what we
look for we'll miss the cache.Fix this by iterating over the cached BOs
until we find the oldest one with matching flags. This improves the hit
ratio for some of the buckets.

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>

tests/amdgpu: Add return CUE_SUCCESS to suite_vcn_tests_clean.

    fixes: 806d0803600000faecb4025d8e9c7490cb097c25 (amdgpu: Use new suite/test disabling functionality.)
    bug: https://bugs.freedesktop.org/show_bug.cgi?id=104280

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

etnaviv: support performance monitor requests

Add etna_cmd_stream_perf(..) to submit perform requests.
Userspace can submit pmrs via submit ioctl to sample perfmon
signals.

v3:
 - mark perfmon bos as RW

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>

etnaviv: add permon support

Query all domains and their signals and provide it this information
via struct etna_perfmon and the corresponding api functions.

v2:
 - code style changes
 - etna_perfmon_create(..): add missing clean up in error case

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>

etnaviv: sync uapi header

Import the etnaviv header changes from kernel commit 05916bed1 (drm-next)

The drm_etnaviv_gem_submit structure was extended to include performance
monitor requests. Also two new ioctls got added to be able to readout
performance monitor domains and their signals.

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Acked-by: Lucas Stach <l.stach@pengutronix.de>

amdgpu: Remove dummy CU_ASSERT_EQUAL.

Fixes test failure on rhel.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Qiang Yu <Qiang.Yu@amd.com>

amdgpu.ids: Refresh from AMD 17.40 release

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>

amdgpu: Only remember the device's marketing name

There's no point in keeping around the full table of marketing names,
when amdgpu_get_marketing_name only ever returns the device's marketing
name.

Acked-by: Slava Abramov <slava.abramov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: Simplify error handling in parse_one_line

* Move empty/commented line check before the strdup and return -EAGAIN
  directly
* Initialize r = -EAGAIN and remove redundant assignments
* Set r = -ENOMEM if last strdup fails, and remove redundant goto

Acked-by: Slava Abramov <slava.abramov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: Clean up amdgpu_parse_asic_ids error handling

* Move error message printing into amdgpu_parse_asic_ids and make it
  return void
* Print only "Invalid format" error message if parse_one_line returns
  -EINVAL
* Use strerror instead of printing the (negative) error code in hex

Acked-by: Slava Abramov <slava.abramov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: Add explicit dependency test.

The test is as following:

1) Create context A & B
2) Send a command submission using context A which fires up a compute shader.
3) The shader wait a bit and then write a value to a memory location.
4) Send a command submission using context B which writes another value to the same memory location, but having an explicit dependency on the first command submission.
5) Wait with the CPU for both submissions to finish and inspect the written value.

Test passes if the value seen in the memory location after both submissions is from command B.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>