amdgpu: Add amdgpu_cs_syncobj_transfer to amdgpu-symbol-check

Fixes make check. Trivial.

add syncobj timeline tests v3

v2: drop DRM_SYNCOBJ_CREATE_TYPE_TIMELINE, fix timeout calculation,
    fix some warnings
v3: add export/import and cpu signal testing cases

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

wrap transfer interfaces

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

expose timeline signal/export/import interfaces v2

v2: adapt to new one transfer ioctl

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

add timeline signal/transfer ioctls v2

v2: use one transfer ioctl

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

wrap syncobj timeline query/wait APIs for amdgpu v3

v2: symbos are stored in lexical order.
v3: drop export/import and extra query indirection

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

add timeline wait/query ioctl v2

v2: drop export/import

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

add cs chunk for syncobj timeline

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

xf86drm: Fix possible memory leak with drmModeGetPropertyPtr()

In drmModeGetPropertyPtr(), from upper error path, it calls free
but with just next error path, it does not call. Fix the possible
memory leak.

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

tests/libkms-test-plane: fix possbile memory leak

The pointer p aquired with drmModeGetPlane() is not free in error
path. Fix possible memory leak by calling drmModeFreePlane() in
the error path.

Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>

libdrm: omap: Add DRM_RDWR flag to dmabuf export

Allows mmap on dmabuf fd with MAP_SHARED and PROT_WRITE.

This fixes boot failures with Android (likely w/ closed source
user-space drivers) that were caused due to mmap() returning
error.

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Acked-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Hemant Hariyani <hemanthariyani@ti.com>
[picked and updated commitmsg from http://git.ti.com/cgit/cgit.cgi/android/external-libdrm.git/]
Signed-off-by: Praneeth Bajjuri <praneeth@ti.com>
Signed-off-by: Alistair Strachan <astrachan@google.com>
[jstultz: Tweaked commit message]
Signed-off-by: John Stultz <john.stultz@linaro.org>

libdrm: reduce number of reallocations in drmModeAtomicAddProperty

When calling drmModeAtomicAddProperty allocation of memory
happens as needed in increments of 16 elements. This can be very
slow if there are multiple properties to be updated in an Atomic
Commit call.

Increase this to as many as can fit in a memory PAGE to avoid
having to reallocate memory too often.

Also this patch has a small one line perf tweak in
drmModeAtomicDuplicate() to only memcpy items to the cursor
position in order avoid copying the entire item array if its
mostly empty.

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
[jstultz: Expanded commit message]
Signed-off-by: John Stultz <john.stultz@linaro.org>

libdrm: Avoid additional drm open close

Avoid additional drm device open and close.

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>

libdrm: amdgpu: Initialize unions with memset rather than "= {0}"

Clang complains when initializing unions using "= {0}"
so instead use memset.

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>

libdrm: Use mmap64 instead of __mmap2

__mmap2 isn't supported on all platforms, mmap64 is the right way
to do this in android.

Also folds in a fix from Stéphane Marchesin <marcheu@chromium.org>

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
[jstultz: Folded in Stéphane's fix]
Signed-off-by: John Stultz <john.stultz@linaro.org>

libdrm: Android.mk: Add minimal Android platform check

Add a check to error out on Android version K(4.4) or
lower.

This is due to dependency added in a previous commit on mmap64,
which was introduced with Android L.

Cc: Sean Paul <seanpaul@chromium.org>
Cc: Alistair Strachan <astrachan@google.com>
Cc: Marissa Wall <marissaw@google.com>
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>

Bump the version to 2.4.98

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>

amdgpu: Add context priority override function.

This way we can override the priority of a single context using a
master fd.

Since we cannot usefully create an amdgpu device of a master fd
without the fd deduplication kicking in this takes a plain fd.

This can be used by e.g. radv to get high priority contexts using
a master fd from the primary node or a lease.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Andres Rodriguez <andresx7@gmail.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>

tests/util: Add armada-drm driver

This makes the test utilities work with the Armada driver without the
necessity of using the -M argument.

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

Revert "libdrm: Fix issue about differrent domainID but same BDF"

This reverts commit 56c21f877bda7b8c62c3d7925b070b5113d807f2.

There were issues pointed out during review that were not addressed.
Would love to have this re-land, once those are addressed.

tests/modetest: add QCOM_COMPRESSED to supported modifiers list

Signed-off-by: Fritz Koenig <frkoenig@google.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

fix various typos

Saw a couple of typos fixes in the patch DragonFlyBSD carries [1], so
I ran codespell (a spell checker for code) on the whole repo.

[1] https://github.com/DragonFlyBSD/DPorts/blob/master/graphics/libdrm/files/patch-xf86drm.c

Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

amdgpu/tests: drop unused local vars

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

amdgpu: Fix a structure initialization issue

struct drmPciBusInfo has been aligned to 6 bytes. So memcmp will access
the last byte which is not initialized.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

headers: Sync with drm-next

Generated using make headers_install from the drm-next
tree - git://anongit.freedesktop.org/drm/drm
branch - drm-next
commit - 14d2bd53a47a7e1cb3e03d00a6b952734cf90f3f

The changes were as follows :-

core: (drm.h, drm_fourcc.h, drm_mode.h)
- Added 'struct drm_syncobj_transfer', 'struct drm_syncobj_timeline_wait' and 'struct drm_syncobj_timeline_array'
- Added various DRM_IOCTL_SYNCOBJ_ ioctls
- Added some new RGB and YUV formats
- Added 'DRM_FORMAT_MOD_VENDOR_ALLWINNER'
- Added 'SAMSUNG' and Arm's 'AFBC' and 'ALLWINNER' format modifiers
- Added 'struct drm_mode_rect'

i915:
- Added struct 'struct i915_user_extension' and various 'struct drm_i915_gem_context_'
- Added different modes of per-process Graphics Translation Table

Changes from v1:-
- Removed the changes to 'msm_drm.h' as it breaks the build for 'freedreno' platform.

Signed-off-by: Ayan Kumar halder <ayan.halder@arm.com>
Acked-by: Eric Engestrom <eric.engestrom@intel.com>

drm/amdgpu: support test mask

support per device test mask. Skip inject test on non-server card.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: add ras tests

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: update amdgpu_drm.h from drm-next for 5.2

From drm-next commit b4e4538a0ab5079ae5dc401970e11f0ff2ba13a7

Adds support for:
- RAS queries
- context priority updates
- CS chunks support for scheduled dependencies
- IB flag for GDS max wave id

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

virtgpu: Update kernel header

Generated using make headers_install.

This brings in the in/out fence support for explicit
synchronization.

v2: don't use experimental kernel branch

Signed-off-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Rob Clark <robclark@freedesktop.org>

intel: sync i915_pciids.h with kernel

Add CML and EHL PCI IDs, and one more for ICL. This is in sync with
kernel header as of b024ab9b2d3a ("drm/i915/bios: iterate over child
devices to initialize ddi_port_info")

Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>

tests/amdgpu: minor fix for dispatch/draw test

1. clear cmd buffer
2. make amdgpu_memcpy_dispatch_test static
3. tab/space fix

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Flora Cui <flora.cui@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: add memcpy draw test

add memcpy draw test for gfx9

Signed-off-by: Flora Cui <flora.cui@amd.com>
Tested-by: Rui Teng <rui.teng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: add memset draw test

add memset draw test for gfx9

Signed-off-by: Flora Cui <flora.cui@amd.com>
Tested-by: Rui Teng <rui.teng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: add memcpy dispatch test

add memcpy dispatch test for gfx9

Signed-off-by: Flora Cui <flora.cui@amd.com>
Tested-by: Rui Teng <rui.teng@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: add memset dispatch test

add memset dispatch test for gfx9
v2: disable dispatch test for other ASICs

Signed-off-by: Flora Cui <flora.cui@amd.com>
Tested-by: Rui Teng <rui.teng@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

tests/amdgpu: add deadlock test for sdma

deadlock test for sdma will cause gpu recoverty.
disable the test for now until GPU reset recovery could survive at least
1000 times test.

v2: add modprobe parameter

Reviewed-and-tested-by: Evan Quan <evan.quan@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Flora Cui <flora.cui@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: add marketing name for AMD Radeon VII

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

xf86drm: Fix segmentation fault while parsing device info

This fixes a bug, which was introduced with commit ee798b98
"xf85drm: de-duplicate drmParse{Platform.Host1x}{Bus,Device}Info".
where accessing *compatible[i] with i>0 results in a segfault.

Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Fixes: ee798b98479709acdd8d "xf85drm: de-duplicate drmParse{Platform.Host1x}{Bus,Device}Info"
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
[Eric: add the same fix to the free() below]
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>

libkms: update list of intel_drivers for Android build

Add new iris driver, remove deprecated ilo driver.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>

configure.ac fix build error for config.h in autotools

After the commit 0926f0af54d6 ("meson,configure: include config.h
automatically"), there is build error for autotools because
config.h is not included. Fix the error by adding "-include
config.h" to CPPFLAGS instead of CFLAGS from configure.ac.

Reference: https://bugs.freedesktop.org/show_bug.cgi?id=106561
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Fixes: 0926f0af54d654b1a958 "meson,configure: include config.h automatically"

libdrm: Fix issue about differrent domainID but same BDF

For multiple GPUs which has the same BDF, but has different domain ID,
the drmOpenByBusid will return the wrong fd when startx.

The reproduce sequence as below:
1. Call drmOpenByBusid to open Card0, then will return the right fd0, and the
fd0 is master privilege;
2. Call drmOpenByBusid to open Card1. In function drmOpenByBusid, it will
open Card0 first, this time, the fd1 for opening Card0 is not master
privilege, and will call drmSetInterfaceVersion to identify the
domain ID feature, as the fd1 is not master privilege, then drmSetInterfaceVersion
will fail, and then won't compare domain ID, then return the wrong fd for Card1.

Solution:
First loop search the best match fd about drm 1.4.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

gitlab-ci: fix archlinux builds

base/archlinux has been replaced with archlinux/base, which is
maintained directly by the archlinux community.

Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>

freedreno: revert bad freedreno/atomic_ops commits

This reverts 6d2379857b6fccc1cade "xf86atomic: #undef internal define"
and b541d21a0a908bf98d44 "freedreno: remove always-defined #ifdef".

I didn't realise at the time that freedreno/freedreno_ringbuffer.h gets
installed, and then used by Mesa for instance. These two commits were
fine in libdrm, but broke Mesa which needs to use struct fd_ringbuffer
but doesn't need to access ::refcnt. The hack that I removed serves to
keep the struct at the correct size while only exposing the ::refcnt
member within libdrm.

Fixes: 6d2379857b6fccc1cade "xf86atomic: #undef internal define"
Fixes: b541d21a0a908bf98d44 "freedreno: remove always-defined #ifdef"
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>

xf86drm: fix return type for drmIsMaster()

Xserver has struct members named `bool`, which means the last commit
breaks its build with errors like this:

  error: two or more data types in declaration specifiers
  Bool bool;
       ^

Fix this by making it return a 0/1 integer, with the same semantic as
the boolean it was before.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=109587
Fixes: 17dfe3ac93217b43f93b "xf86drm: Add drmIsMaster()"
Cc: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

xf86drm: Add drmIsMaster()

We can't use drmSetMaster to query whether or not a drm fd is master
because it requires CAP_SYS_ADMIN, even if the fd *is* a master fd.

Pick DRM_IOCTL_MODE_ATTACHMODE as a long-deprecated ioctl that is
DRM_MASTER but not DRM_ROOT_ONLY as the probe by which we can detect
whether or not the fd is master.

This is useful for code that might get master by open()ing the drm device
while no other master exists, but can't call drmSetMaster itself because
it's not running as root or is in a container, where container-root isn't
real-root.

v2: Use the AUTH_MAGIC request rather than MODE_ATTACHMODE, as it's more
    clearly related to master status.

v3: [Emil] Don't expose internals, check for -EACCES.

Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> (v2)
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

intel: sync i915_pciids.h with kernel

Straight copy from the kernel file.

Add more PCI Device IDs for Coffee Lake, Ice Lake,
and Amber Lake. It also include a reorg on Whiskey Lake IDs.

Align with kernel commits:

5e0f5a58b167 ("drm/i915/cfl: Adding another PCI Device ID.")
03ca3cf8e9aa ("drm/i915/icl: Adding few more device IDs for Ice Lake")
c0c46ca461f1 ("drm/i915/aml: Add new Amber Lake PCI ID")
c1c8f6fa731b ("drm/i915: Redefine some Whiskey Lake SKUs")

Cc: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

xf85drm: de-duplicate drmParse{Platform.Host1x}{Bus,Device}Info

The functions are virtually identical, fold them up.

v2: foo -> tmp_name (Eric)

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

xf86drm: fallback to MODALIAS for OF less platform devices

Some devices can lack OF data or it may not be available in the uevent
file. Fallback to the MODALIAS data in those cases.

We strip any leading "MODALIAS=.*:" thus the resulting information is
compatible with existing code in Mesa.

v2: foo -> tmp_name

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Tested-by: Lucas Stach <l.stach@pengutronix.de> (v1)

amdgpu: add some raven marketing names

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

Bump the version to 2.4.97

amdgpu: add a faster BO list API

Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>

amdgpu: update amdgpu_drm.h

it's in kernel 5.0

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

amdgpu: update to latest marketing names from 18.50

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

README: reflow the project description to improve readability

Also, move the sentence about "who would use libdrm" into its own paragraph,
as it is something people discovering libdrm will want to know.

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

xf86atomic: #undef internal define

Thanks to the #error just above, any file including this header can only
see one state for this macro: defined, with the value `1`.
Let's just #undef it once we're done using it in here so that other
files don't misconstrue any meaning to it.

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

freedreno: remove always-defined #ifdef

While at it, let's include xf86atomic.h explicitly, instead of relying
on some other file accidentally including it before including this file.

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

android: Fix 32-bit app crashing in 64-bit Android

Seemingly the 64-bit int is always aligned to 8 in LP64.
But this is not hold in LP32.

Consequently sizeof(gralloc_drm_handle_t) are different
between LP64 (which is 18 ints) and LP32 (which is 16 ints).
As a result, 32-bit apps will crash in 64-bit OS since the
checking handle->base.numInts != GRALLOC_GBM_HANDLE_NUM_INTS
is true.

Fix it by always aligning 64-bit int to 8. Besides, to avoid
additional padding, just exchange the order of data_owner
and modifier. It aligns modifier to 8 natually.
This makes gralloc_drm_handle_t fit in 16 ints perfectly.

(v2) gralloc_drm_handle.h patch now applied in gralloc_handle.h
     and GRALLOC_HANDLE_VERSION updated to 4

Reported-by: Mauro Rossi <issor.oruam@gmail.com>
Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
Reviewed-by: Robert Foss <robert.foss@collabora.com>

gitignore: add _build

This is the directory used by meson/autotools (at least in the
.gitlab-ci configuration) so ignore the whole dir.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Emil Velikov <emil.velikov@collabora.com>

gitignore: sort file

LANG=C sort -u .gitignore | sponge .gitignore

This way it's easier to keep track of the entries.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Emil Velikov <emil.velikov@collabora.com>

tests/amdgpu/vcn: fix the nop command in IBs

Just make them properly i.e. put 0 to the Nop reg

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

libdrm: Use DRM_IOCTL_GET_PCIINFO on DragonFly

It is a cleaner and less fragile way to get PCI IDs than the one
currently used by local DPorts patches.

Signed-off-by: François Tigeot <ftigeot@wolfpond.org>

xf86drm: implement drmParseSubsystemType for DragonFly

Like on OpenBSD, the DragonFly BSD kernel only contains
pci drm drivers.

Signed-off-by: François Tigeot <ftigeot@wolfpond.org>

amdgpu/test: Enable deadlock test for CI family (gfx7)

I retested GPU recovery with  Bonaire ASIC and it works.

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

amdgpu: add VM test to exercise max/min address space

Make sure the kernel doesn't crash if we map something at the minimum/maximum address.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>

drm: Attempt to parse SPI devices as platform bus devices.

For ARM systems with tinydrm displays attached to SPI, the bus name is
/spi but we have platform device info for the rest.  Fixes
eglInitialize() failures on hx8357d since the EGL_EXT_device_drm
changes.

Acked-by: Eric Engestrom <eric.engestrom@intel.com>

Avoid hardcoded strlens in drmParseSubsystemType().

Having people count characters is error-prone, when we could just have
a computer do it.

Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>

tests: skip drmdevice test if the machine doesn't have any drm device

Error message was invalid too, negative values aren't the number of
devices, they're errno error codes.

Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>

meson: fix typo in compiler flag

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

xf86drmHash: remove unused loop variable

Reported-by: Jan Vesely <jano.vesely@gmail.com>
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

amdgpu/test: Disable deadlock tests for all non gfx8/9 ASICs.

Since only for those ASICs gpu reset is enabled by deafult.
Also update disable message and fix identation .

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu/test: Add illegal register and memory access test v2

Illegal access will cause CP hang followed by job timeout and
recovery kicking in.
Also, disable the suite for all APU ASICs until GPU
reset issues for them will be resolved and GPU reset recovery
will be enabled by default.

v2:
Add KV to deasbled APUs list and add comments regarding
necessary kernel amdgpu paramteres to run the tests.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

doc: Rename README&CONTRIBUTING to .rst

Looks much neater on the gitlab UI, e.g. on my personal libdrm fork:

https://gitlab.freedesktop.org/danvet/drm

Acked-by: Eric Engestrom <eric.engestrom@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Bump to version 2.4.96

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

freedreno/msm: fix c90 warning

Wasn't really sure how to switch the silly -Wdeclaration-after-statement
flag off on a per directory basis.  So make the code uglier instead.

freedreno/kgsl: fix build

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

freedreno/msm: handle ring-reloc to other stateobjs

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

freedreno/msm: simplify msm_ringbuffer_flush()

Now that it doesn't have to deal with ringmarkers we can simplify the
reloc handling.

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

freedreno/msm: use hashtable to track submit.cmds table

With streaming stateobjs to upload uniforms, the submit.cmds table gets
much larger, and iterating over it for each ring to ring reloc starts
getting expensive.

TODO if we have flag to pass when constructing parent rb, we could
avoid dynamically allocating this and bo_table in get_cmd() or bo2idx

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

freedreno/msm: simplify emit_reloc_ring() vfunc

Now that it doesn't have to deal with the ringmarker case, we can make
some simplifications.

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

freedreno/msm: remove reset of linked rings

The msm_cmd isn't refcount'd, so with stateobj rb's that have
independent lifecycle, this is no longer a safe thing to do.
Really, now that there is a bo-cache for rb's, fd_ringbuffer_reset()
should be deprecated because it adds a bunch of pointless complexity.

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

freedreno: remove deprecated ringmarker API

It's usage in mesa was removed more than two years ago.  And it stands
in the way of some optimizations needed to reduce the overhead of hw
stateobjs (ie. CP_SET_DRAW_STATE, where the # of cmds in the submit
ioctl goes up significantly).

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

freedreno/msm: support suballocation for stateobj rb's

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

freedreno: add flags param for rb creation

For now, we want a way for gallium to be able to provide hints for the
upcoming rb suballocation.  But could be useful for other things down
the road.

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

freedreno: expose refcnt'ing on ringbuffers

Move this out of msm_ringbuffer backend so that the gallium driver can
refcnt rb's

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

freedreno/msm: get rid of ring_bo unref hack

Since 28328298 'freedreno: move ring_cache behind fd_bo_del()' this hack
is no longer necessary.

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

xf86drmHash: remove redundant zero init

drmMalloc() is already calloc()

Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Rob Clark <robclark@freedesktop.org>

amdgpu/test: Fix deadlock tests for AI and RV v2

Seems like AI and RV requires uncashed memory mapping to be able
to pickup value written to memory by CPU after the WAIT_REG_MEM
command was already launched.
.
Enable the test for AI and RV.

v2:
Update commit description.

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

amdgpu/test: Allow BO mapping flags to be passed in tests

v2:
Call amdgpu_bo_va_op_raw directly in amdgpu_bo_alloc_and_map_raw
Move amdgpu_bo_alloc_and_map_raw into C file to avoid including
unistd.h in amdgpu_test.h

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

libdrm: Allow dynamic drm majors on linux

To determine whether a device node is a drm device node or not, the code
currently compares the node's major number to the static drm major device
number.

This breaks the standalone vmwgfx driver on XWayland dri clients,
https://cgit.freedesktop.org/mesa/vmwgfx
and any future attempt to introduce dynamic device numbers for drm.

So instead of checking for the device major, instead check for the presence
of the /sys/dev/char/<major>:<minor>/device/drm directory.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Acked-by: Dave Airlie <airlied@redhat.com>

Bump to version 2.4.95

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

intel: include i915_pciids.h in the tarball

Fixes: 4e81d4f9c9b ("intel: add generic functions to check PCI ID")
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

Revert "Bump to version 2.4.95"

This reverts commit 6e50a309afb340088d7da762f4434d409216caf2.

Bump to version 2.4.95

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

libdrm: headers: Sync with drm-next

Generated using make headers_install from the drm-next
tree - git://anongit.freedesktop.org/drm/drm
branch - drm-next
commit - 2dc7bad71cd310dc94d1c9907909324dd2b0618f

The changes were as follows :-

  core: (drm.h, drm_fourcc.h, drm_mode.h)
    - Added client capabilities for ASPECT_RATIO and WRITEBACK_CONNECTORS
    - Added Arm AFBC modifiers
    - Added BROADCOM's SAND and UIF modifiers
    - Added Qualcomm's modifiers
    - Added some picture aspect ratio and content type options
    - Added some drm mode flags
    - Added writeback connector id

  amdgpu:
    - Added GEM domain mask
    - Added some GEM flags
    - Added some hardware ip flags
    - Added chunk id and IB fence.
    - Added some query ids

  i915:
    -Added an IOCTL (I915_PARAM_MMAP_GTT_COHERENT)

  qxl:
    - Minor changes

  tegra:
    - Added some comments about struct drm_tegra* members
    - Modified DRM_IOCTL_TEGRA_CLOSE_CHANNEL

  vc4:
    - Added some members for 'struct drm_vc4_submit_cl'

Changes in v2:
    - Mentioned 'libdrm' in the commit header.

Changes in v3:
    - Removed the changes to radeon_drm.h, sis_drm.h and via_drm.h as suggested by
      Emil Velikov <emil.l.velikov@gmail.com>

Changes in v4:
    - Removed the changes to vmwgfx_drm.h as it caused a build break ie
      'make check' failed.

Change-Id: I018a06f65bf4a6a68400ab252b9cd05d041299b3
Signed-off-by: Ayan Kumar halder <ayan.halder@arm.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

gitlab-ci: pass the correct toggles to configure

Trivial typos - s/admgpu/amdgpu/;s/vmwfgx/vmwgfx/

Fixes: 4a9030dc8b7 ("add gitlab-ci builds of libdrm")
Cc: Brian Starkey <brian.starkey@arm.com>
Reported-by: Brian Starkey <brian.starkey@arm.com>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>

tests/util: Add support for sun4i-drm driver

This is the DRM driver for all Allwinner (sunxi) platforms.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

android: make symbols hidden by default

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

*-symbols-check: error out when using unset variables

It will make bugs like the one fixed with previous patch dead obvious.

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

automake: set NM before running the tests

Set/export the NM variable since it may not be set already.

Fixes: 4f08bfe96da ("*-symbol-check: Don't hard-code nm executable")
Cc: Heiko Becker <heirecka@exherbo.org>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>

CI: Capture test logs as GitLab artifacts

GitLab CI already captures all the stdout/stderr output from the build
process as the log. However, some other important information is hidden
in other log files.

Taken from Wayland, capture logs from the configuration process as well
as from every check.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Cc: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Cc: Eric Engeström <eric.engestrom@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Engestrom <eric.engestrom@intel.com>
[Emil: use wildcard to match the artefacts]
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>