drm: replace HASH_DEBUG with DEBUG

... and remove the useless SL_DEBUG and RANDOM_DEBUG

v2: Rebase on earlier changes.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/random: return non-zero on test failure

... and wire it up to make check

v2: s/rand - state->check/rand != state->check/. (Jan)

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/random: extract test out of xf86drmRandom.c

With follow up commits we can clear it up and wire to
make check

v2:
 - Use xf86drmRandom.h for common struct.(Jan)
 - Add test to .gitignore.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/hash: return non-zero on failure

... and wire up to `make check' now that it's useful.

v2: Really return non-zero on failure.
v3: Initialise ret.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu> (v2)

tests/hash: style fixes

v2: Rebase on earlier changes. Keep count initialisation as is.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/hash: misc compilation fixes

Get the test from completely broken to working like a charm.

 - Use the same variable type for both HashInsert and HashLookup.
 - Use correct storage type for the HashLookup return value.
 - Remove useless backward iteration of HashLookup(i).

v2:
 - Use void * instead of unsigned long.
 - Change value to key << 16 | key.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/hash: extract test out of xf86drmHash.c

This way with follow up commits we can fix it and wire it up to
make check

v2:
 - Use xf86drmHash.h for common structs.(Jan)
 - Add test to .gitignore.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests/drmsl: Extract tests out of xf86drmSL.c

v2: merge tests creation and xf86drmSL cleanup
    rename tests/drmsltest -> tests/drmsl
    move the test out of libudev test block
v3: run test even on noudev builds

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>

drmSL: Fix neighbor lookup

Commit e4a519635f75bde38aeb5b09f2ff4efbf73453e9:
    Tidy up compile warnings by cleaning up types.

removed call to SLLocate which gutted the function of all functionality.
This patch restores the original behavior, with an additional fix
that zeros the update array in case SLLocate bails early.

v2: zero the update array instead of checking the return value.
    SLLocate returns NULL both on failure and if the element is greater
    than everything in the list
v3: Improve commit message

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>

android: remove explicit include to libpciaccess

Both android-x86 and android-ia versions of libpciacccess correctly
"export" the include. If anyone else is wrapping up their own version
they should do so as well.

Remove this fixed location hack from the build.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Chih-Wei Huang <cwhuang@linux.org.tw>

tests/exynos: Fix missing static keyword

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>

Remove drmSetDebugMsgFunction and related infrastructure

Not used anywhere

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

Fix unused function warnings

v2: Remove the handler function instead of commenting out
    split debugmsg function removal to a separate patch

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

tests: add rockchip to modetest, kmstest, vbltest and proptest

There is a rockchip drm kms driver.
Add "rockchip" to the static lists of driver names in the the standard
set of tests.

Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
[Emil Velikov: Resolve trivial conflicts.]
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

proptest: install it with --enable-install-test-programs

--enable-install-test-programs allows tests to be installed in $bindir.
This is disabled by default, but very useful when cross compiling.

Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

autotools: remove ${srcdir} from the includes

Already handled by the build system.

v2: s/compiler/build system/

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

android: get rid of LIBDRM_TOP

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>

android: simplify the including rule of subdirs

Use android build system functions to include Android.mk
of subdirs.

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
[Emil Velikov: Resolve trivial conflicts.]
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

android: add the missing tag "optional" to libkms

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

android: remove LOCAL_COPY_HEADERS* variables

With earlier changes we've implicitly add the relevant directories
to the includes list, via LOCAL_EXPORT_C_INCLUDES_DIRS.

v2: Update the top Android.mk as well.

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

android: remove ${srcdir} from the includes

Already handled by the build system.

v2: s/compiler/build system/. Spotted by Chih-Wei.

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

android: simplify LOCAL_C_INCLUDES

Each of the libdrm_${hw} modules pull libdrm for linking as such:

libdrm's LOCAL_EXPORT_C_INCLUDE_DIRS are added to the includes list.
The former of which is already set to ${top} and ${top}/include/drm.

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

android: correcly set LOCAL_EXPORT_C_INCLUDE_DIRS

 - Don't add ${hw}/${hw}, but ${hw} to the includes path. The former
does not exist.
 - Set the variable for libkms.

Inspired by the work of from Chih-Wei from the Android-x86 project.

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

RELEASING: Fix annouce typo

That's the only typo :set spell found.

v2: Fix typo in commit message (Ilia Mirkin)

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

RELEASING: Fix the step numbering

v2: Really fix the numbering (Emil Velikov)

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

RELEASING: Fix releasing instructions to match the latest release.sh

It seems that the tests don't need DRM master anymore? at least make
distcheck passes when X is running.

release.sh is also invoked with just the path to the libdrm git checkout
and we don't want to pass additional arguments that will be treated as
additional modules we want to release.

Also, make a note that release.sh will run make distcheck for you, so we
don't strickly need to run it beforehand.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

intel: Merge latest i915_drm.h

The main incentive to do this is to get I915_PARAM_REVISION.

v2: Rebase on top of some changes that were made to the header without
    copying the whole file from the kernel source.

Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Neil Roberts <neil@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

build: Bump version number to 2.4.60 before release

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

intel: Export total subslice and EU counts

Update kernel interface with new I915_GETPARAM ioctl entries for
subslice total and EU total. Add a wrapping function for each
parameter. Userspace drivers need these values when constructing
GPGPU commands. This kernel query method is intended to replace
the PCI ID-based tables that userspace drivers currently maintain.
The kernel driver can employ fuse register reads as needed to
ensure the most accurate determination of GT config attributes.
This first became important with Cherryview in which the config
could differ between devices with the same PCI ID.

The kernel detection of these values is device-specific. Userspace
drivers should continue to maintain ID-based tables for older
devices which return ENODEV when using this query.

v2: remove unnecessary include of <stdbool.h> and increment the
    I915_GETPARAM indices to match updated kernel patch.

For: VIZ-4636
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Jeff McGee <jeff.mcgee@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>

configure.ac: error out if building freedreno_kgsl without freedreno

The former is a subset of the latter. Error out early so the user is
aware that they are doing something very wrong.

Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

configure.ac: fix help string copy/pasta

The message "enabled on x86" was meant for the intel libdrm.
Take the opportunity to mention how libkms is autodetected.

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

configure.ac: fix host_cpu/atomics detection

Previous code was busted, as it wasn't checking directly for what it was
meant to, and at the end changing the user's selection if host_cpu
heuristics were involved.

Simplify things by adding a macro that does the long message printing
for us, and check for only what we need.

This fixes commit 36cff14bb03(configure: omap, freedreno and tegra
require atomics) which incorrectly assumed that the code was working
fine, and effectively made impossible to enable freedreno due to it's
host_cpu detection.

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

autogen.sh: handle out-of-tree invokation

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

configure: Stop using AM_MAINTAINER_MODE

AM_MAINTAINER_MODE can be used to disable generation of rebuild rules.
This is not something we want to condone/support, considering it can
cause greater problems than the perceived benefits. Additionally the
Automake manual leans towards avoiding the use of AM_MAINTAINER_MODE.

http://www.gnu.org/software/automake/manual/html_node/maintainer_002dmode.html

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

exynos: fimg2d: follow-up fix for G2D_COEFF_MODE_GB_COLOR

Also add the register field formatting info provided by
Inki Dae <inki.dae@samsung.com>.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Suggested-by: Inki Dae <inki.dae@samsung.com>

exynos: add fimg2d header to common includes

The reason for this change is to let userspace use the header.
Currently 'make install' does not install it.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: add exynos prefix to fimg2d header

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: use structure initialization instead of memset

Keeps the code cleaner, since the structs have to be initialized
once anyway.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>
[evelikov: squash trivial conflict]
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Conflicts:
tests/exynos/exynos_fimg2d_test.c

exynos: honor the repeat mode in g2d_copy_with_scale

This is useful when the default repeat mode, which is 'repeat'
produces artifacts at the borders of the copied image.
Choose the 'pad' mode to make use of the color of the destination
image.

In my usage case the destination is the framebuffer, which is
solid filled with a background color. Scaling with 'pad' mode
would then just do the right thing and also produces nice
borders on the output.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: add g2d_scale_and_blend

This is a combination of g2d_copy_with_scale and g2d_scale.
It is a pretty common operation to scale one buffer and then
blend it on top of another, so provide a direct way to that
operation.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

tests/exynos: fimg2d: add a checkerboard test

This makes it easier to spot memory corruptions which don't become
visible when using a plain buffer filled with a solid color (so
corruptions that are just a permutation of the bytes in the buffer).

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

modetest: include into the build when libkms is not selected.

With commit d7c0a08bc57(modetest: Allocate dumb buffers with the correct
bpp) we moved away from the libkms dependency. As such we are safe with
including the Makefile/subdir, even as we opt out of building the
library.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

Add static qualifier to local functions

v2: Don't bother marking dead functions static
    (handler, xf86VDrvMsgVerb, drmSetDebugMsgFunction)

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

Fix unused, and unused-but-set variables warnings

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <eil.l.velikov@gmail.com>

dristat: Handle DRM_CONSISTENT

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <eil.l.velikov@gmail.com>

Fix type-limits, pointer-arith and sign-compare warnings

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <eil.l.velikov@gmail.com>

tests: String literals are const char *

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Emil Velikov <eil.l.velikov@gmail.com>

On Solaris, #include <sys/mkdev.h> in xf86drm.c

Needed on Solaris for the definitions of major() & minor() used in
drmGetNodeTypeFromFd() and makedev() used in drmOpenMinor()

Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

nouveau: Do not add most bo's to the global bo list.

Only add wrapped bo's and bo's that have been exported through flink or dma-buf.
This avoids a lock in the common case, and decreases traversal needed for importing
a dma-buf or flink.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>
Tested-By: Emil Velikov <emil.l.velikov@gmail.com>

nouveau: make nouveau importing global buffers completely thread-safe, with tests

While I've closed off most races in a previous patch, a small race still existed
where importing then unreffing cound cause an invalid bo. Add a test for this case.

Racing sequence fixed:

- thread 1 releases bo, refcount drops to zero, blocks on acquiring nvdev->lock.
- thread 2 increases refcount to 1.
- thread 2 decreases refcount to zero, blocks on acquiring nvdev->lock.

At this point the 2 threads will clean up the same bo.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>
Reviewed-By: Emil Velikov <emil.l.velikov@gmail.com>

Use __sync_add_and_fetch instead of __sync_fetch_and_add for atomic_dec_and_test

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>
Acked-By: Emil Velikov <emil.l.velikov@gmail.com>

Add atomic_inc_return to atomics.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>
Acked-By: Emil Velikov <emil.l.velikov@gmail.com>

exynos: fimg2d: whitespace fix in g2d_flush

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

exynos: fimg2d: introduce G2D_OP_INTERPOLATE

This sets up the blending equation in the following way:
out = src * src_alpha + dst * (1 - src_alpha)

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: fimg2d: unify register style

Register defines all use uppercase hex codes.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: fimg2d: fix comment for G2D_COEFF_MODE_GB_COLOR

The coefficient mode enables use of global color, not alpha.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: fimg2d: remove TRUE/FALSE from header

The fimg2d header was defining TRUE and FALSE, but actually
these defines are just used once. Remove them, since they
don't make the code better readable/understandable.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

tests/exynos: improve error handling

Check for a useable connector and also if the resolution is sane
(width and height are both non-zero).

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: introduce g2d_add_base_addr helper function

In almost all functions the base address register is written, so it
makes sense to have a helper function for this.

v3: Wrap line as pointed out by Emil Velikov
    <emil.l.velikov@gmail.com>.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

tests/exynos: introduce wait_for_user_input

Currently getchar() is used to pause execution after each test.
The user isn't informed if one is supposed to do anything for
the tests to continue, so print a simple message to make this
more clear.

v3: Compactify printf calls as pointed out by
    Emil Velikov <emil.l.velikov@gmail.com>.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

tests/exynos: disable the G2D userptr/blend test

v2: Move the commit description into the patch itself.
v3: Use common commenting style as pointed out by
    Emil Velikov <emil.l.velikov@gmail.com>.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

tests/exynos: fix typos and change wording

No functional changes.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

exynos: replace G2D_DOUBLE_TO_FIXED macro with function

This also avoids the floating point conversion steps and just
uses pure integer arithmetic.
Since the G2D hardware scaling approach is a bit unintuitive,
document it in the function as well.

v2: Explicitly mention the normalization constant.
v3: Use common commenting style as pointed out by
    Emil Velikov <emil.l.velikov@gmail.com>.

Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Joonyoung Shim <jy0922.shim@samsung.com>

drm: add drmGet(Primary|Render)DeviceNameFromFd functions

Currently most places assume reliable primary(master) <> render node
mapping. Although this may work in some cases, it is not correct.

Add a couple of helpers that hide the details and provide the name of
the master or render device name, given an fd. The latter may belong to
either the master, control or render node device.

v2:
 - Rename Device and Primary to Master (aka the /dev/dri/cardX device).
 - Check for the file via readdir_r() rather than stat().
 - Wrap the check into a single function.
 - Return NULL for non-linux platforms.

v3:
 - Don't segfault if name is NULL.
 - Update function names, as suggested by Frank Binns.

v4:
 - Update commit message to reflect the function name changes.

Cc: Frank Binns <frank.binns@imgtec.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: David Herrmann <dh.herrmann@googlemail.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>

tests: automake: keep the libs link at the final stage

Currently the static archive libdrmtest.la links against libdrm.la.
Only to have both added to the executables' LDADD. Simplify things a
bit, by doing the link in the final module (the executables/tests in
this case).

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

tests: move the SUBDIR at the top of the makefile

None of the subdirs require the modules built in the local makefile, so
moving them at the top makes things a hell lot easier to read.

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

automake: drop the NULL variable from the makefile

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

configure: update help strings

Mention that freedreno and intel are both auto-detected (depending on
the presence of atomics), plus host_cpu "defined".

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

configure: omap, freedreno and tegra require atomics

They have used them since day one, but the check was never there.

v2:
 - Update the freedreno message (API is no longer experimental).
 - Move the freedreno host_cpu detection next to the intel one.

Cc: Rob Clark <robdclark@gmail.com>
Cc: Thierry Reding <treding@nvidia.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

automake: wrap an insanely long line

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

Revert "intel: Fix documentation for drm_intel_gem_bo_wait()"

This reverts commit 080b4929b7452dc1fea32ac1d32e7e571e7fb38b.

Chris noticed that "negative values wait forever" is indeed intended
behaviour and the issue is just that we didn't have a testcase (fixed
now) and that a regression slipped through (fixed and on track for all
stable kernels).

So lets undo the documentation change for consistency, since working
around kernel regressions isn't good. Practical impact is nil anyway.

v2: Add a note to docs that some kernels have been broken.

v3: Remove the random garbage included by accident.

Cc: Kristian Høgsberg <krh@bitplanet.net>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

intel: Fix documentation for drm_intel_gem_bo_wait()

The kernel doesn't actually wait indefinately when passed a negative,
timeout, it returns immediately.  Document this and suggest using INT64_MAX
for indefinite waits.

Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

tests/radeon: set the list* functions as inline

To silence the chatty compiler.
As a future work we may want to merge these with libdrm_lists.h

Cc: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

exynos_fimg2d_test: remove unused variables

Cc: Inki Dae <inki.dae@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests: remove unused variables

As kindly pointed out by GCC.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

autotools: add WARN_CFLAGS to all targets

... minus test/ttmtest. The latter is not really hooked up with the
actual build.

This will give us 66 warnings on a distribution build of which
 - 12 -Wunused-variable
 - 11 -Wunused-function
 - 19 -Wmissing-prototypes
and a few -Wswitch-enum, -Wtype-limits etc.

Adding the CFLAGS gives some exposure to these so that we can fix them.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests: fix implicit funciton declaration errors

ioctl() and strcmp() were used without the relevent header being
included.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

exynos_fimg2d_test: fix implicit funciton declaration errors

As one adds WARN_CFLAGS to the build the compiler throws a couple of
lovely error messages. Add the relevant includes to fix them.

  error: implicit declaration of function ‘time’
  error: implicit declaration of function ‘getopt’

Cc: Inki Dae <inki.dae@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

autotools: add AM_DISTCHECK_CONFIGURE_FLAGS

To make sure that the release/distribution tarball is not broken for all
the targets. Currently the experimental APIs are disabled by default
amongst others.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>

tests: add support for imx-drm

Since imx-drm has graduated from staging it seems to be a good idea to
recognize it by default in the libdrm tests.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

Add new drmGetNodeTypeFromFd function

Add a helper function that returns the type of device node from an fd.

Signed-off-by: Frank Binns <frank.binns@imgtec.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

android: remove duplicate libdrm in LOCAL_SHARED_LIBRARIES

v2: Fold libpciaccess and libdrm into a single local_shared_libraries

Acked-by: Jan Vesely <jan.vesely@rutgers.edu>
Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

libdrm: fix the Android 64bit build

With 64bit bionic mmap now handles 64bit offset, thus we no longer
need the __mmap2 trick.

Fix from Chih-Wei Huang, over at the google forums.

Cc: Chih-Wei Huang <cwhuang@android-x86.org>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

nouveau: fix unlock nouveau_bo_name_ref()

Signed-off-by: Jérôme Glisse <jglisse@redhat.com>

Add new drmOpenOnceWithType function (v2)

v2: call drmOpenOnceWithType in drmOpenOnce, and drop unused param
for drmOpenOnceWithType

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

Add new drmOpenWithType function (v4)

v2: Add drmGetMinorBase, and call drmOpenWithType in drmOpen
v3: Pass 'type' to drmOpenByBusid and drmOpenDevice in drmOpenByName
v4: Renumber node type definitions, and return -1 for unsupported type

Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v3)
Reviewed-by: Frank Binns <frank.binns@imgtec.com>

Stop undefining _ATOMIC_TYPE in Solaris/NetBSD section of xf86atomic.h

Since the atomic_set() macro references _ATOMIC_TYPE, undefining it
broke that macro, resulting in build errors of:
"intel_bufmgr_gem.c", line 775: undefined symbol: _ATOMIC_TYPE

Now that we're leaving the macro in the namespace, renamed it to start
with LIBDRM_ instead of _ to avoid polluting others namespace.

Regression introduced in 87fdd32c8779648 Add NetBSD atomic ops support.

Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-By: Thomas Klausner <wiz@NetBSD.org>

xf86drm: Fix ioctl struct clearing in drmGetVersion

Oops, fumbled that one.

Reported-by: Jan Vesely <jan.vesely@rutgers.edu>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

tests: remove intel-specific tests

These all moved to igt meanwhile.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

xf86drm: Unconditionally clear ioctl structs

We really have to do this to avoid surprises when extending the ABI
later on. Especially when growing the structures.

A bit overkill to update all the old legacy ioctl wrappers, but can't
hurt really either.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

drm: use drmIoctl everywhere

Well just core drm. All the other callers in there that still use
direct calls to ioctl have some custom retry logic already, so should
be good already.

All the other offenders (tests, freedreno/kgsl, ...) don't really
matter (e.g. kgsl is the blob library and so not a drm thing) or are
again special exceptions with their own retry loops.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

xf86drmMode: Unconditionally clear ioctl structs

We really have to do this to avoid surprises when extending the ABI
later on. Especially when growing the structures.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

intel: Unconditionally clear ioctl structs

We really have to do this to avoid surprises when extending the ABI
later on. Especially when growing the structures.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Fix gcc -Wextra warnings

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Ian Romanick <idr@freedesktop.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

random: Use unsigned long for seed

v2: Remove unrelated change in main()

This is more consistent with the rest, and avoids potential undefined
behavior (signed overflow) ind drmRandom()

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Ian Romanick <idr@freedesktop.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

exynos: remove DRM_EXYNOS_GEM_{MAP_OFFSET/MMAP} ioctls

This patch removes the ioctls which are removed from the linux kernel.

Signed-off-by: Hyungwon Hwang <human.hwang@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Rob Clark <robclark@freedesktop.org>

exynos: Don't use DRM_EXYNOS_GEM_{MAP_OFFSET/MMAP} ioctls

The ioctl DRM_EXYNOS_GEM_MAP_OFFSET and DRM_EXYNOS_GEM_MMAP are removed from
the linux kernel. This patch modifies libdrm and libkms to use drm generic
ioctls instead of the removed ioctls.

v2: The original patch was erroneous. In case the MODE_MAP_DUMB ioctl failed
    it would return the retvalue as a void-pointer. Users of libdrm would then
    happily use that ptr, eventually leading to a segfault. Change this to
    return NULL in that case and also restore the previous behaviour of logging
    to stderr.
    The other error was that 'bo->vaddr' was never filled with the mapped
    buffer address. Hence exynos_bo_map still returned NULL even if the
    buffer mapping succeeded.

Signed-off-by: Hyungwon Hwang <human.hwang@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Rob Clark <robclark@freedesktop.org>

nouveau: Remove unused static function.

Signed-off-by: Thomas Klausner <wiz@NetBSD.org>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>

intel: Only define variable when it's used.

Signed-off-by: Thomas Klausner <wiz@NetBSD.org>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>

Add NetBSD atomic ops support.

Signed-off-by: Thomas Klausner <wiz@NetBSD.org>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@ubuntu.com>