A programming restriction exists for this instruction, atleast one component
of one valid vertex element must be enabled.
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Reviewed-by: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
From Gen9, by default push constant command is not committed to the shader unit
untill the corresponding shader's BTP_* command is parsed. This is the
behaviour when set shader is enabled. This patch updates the batch to follow
this requirement otherwise it results in gpu hang.
Set shader need to be disabled if legacy behaviour is required.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89959
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Tested-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Don't set the size of bindless surface state on rendercopy.
And as of doing so, take into account the workaround for setting
the command size.
This was tried during hunting for
https://bugs.freedesktop.org/show_bug.cgi?id=89959. But no
impact was found.
Cc: Arun Siluvery <arun.siluvery@linux.intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
The commit "benchmarks: Do not install to system-wide bin/" changed
the benchmark file list from bin_PROGRAMS to benchmarks_PROGRAMS.
However Android.mk was not updated, resulting in IGT failing to
build for Android.
This commit adds that change. It also adds LOCAL_MODULE_PATH to
specify where the built benchmarks should be put.
v2: I discovered that the existing definitions of LOCAL_MODULE_PATH
were creating what should have been an invalid path. Not sure how it
was ever working previously, but fixed now.
Signed-off-by: Derek Morton <derek.j.morton@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
We have to defer figuring out the gen until we have a PCI ID, which we
don't have for sure until the first execbuffer2 ioctl. We can't write
the file header until we have the gen since we need to know the size of
offsets in the file format. Move the call to write_header() into the
dump_execbuffer2() where we have the gen.
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
Put 48 bit block offsets in the AUB file and track the bo offsets with
uint64_t.
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
The test does the following
1. set_domain src GTT
2. set_caching src NONE
3. pwrite src
4. set_caching src CACHED
5. blt src->dst
6. pread dst
7. verify data matches
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
igt_plane_set_fb()+igt_display_commit() have too much overhead, and that
causes the cache to get flushed before we flip, making the test
useless, at least on machines with small LLC. Switch to
drmModeSetPlane() to reduce the chance that the cache gets flushed
before we grab the crc.
Still nowhere near 100% reliable on my IVB laptop with 3 MiB LLC,
but at least it can now hit the problem occasioanally. My desktop
IVB with 8 MiB LLC seems to hit it rather reliably.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Our invalid-flags/params testcases are meant to catch abi extensions
by just testing for the next available flag/param. Unfortunately we
need that since without those we forgot to write testcases for these
new flags way too often :(
But it's not entirely clear why this is, so document this trick with
comments.
Also gem_wait wasn't this paranoid, so change the testcase to be so.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Convenience wrapper suggested by Chris for igt_set_timeout(0, NULL).
v2: While at it add an empty line in kms_flip to make
set/reset_timeout a visual block.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
"Timed out" isn't a terribly informative message, allow users to set
something more informative. Inspired by a request from Jesse.
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Allow specification of the many different busyness modes and relocation
interfaces, along with the number of buffers to use and relocations.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Use madvise(MADV_SEQUENTIAL) to let the kernel optimise for our
straightforward sequential read pattern.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Now that we actually install the benchmarks into a sane location,
slightly abuse it to put the tracer for gem_exec_trace alongside.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
These benchmarks are first-and-foremost development tools, not aimed at
general users. As such they should not be installed into the system-wide
bin/ directory, but installed into libexec/.
v2: Now actually install beneath ${libexec}
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This slightly idealises the behaviour of clients with the aim of
measuring the kernel overhead of different workloads. This test focuses
on the cost of relocating batchbuffers.
A trace file is generated with an LD_PRELOAD intercept around
execbuffer, which we can then replay at our leisure. The replay replaces
the real buffers with a set of empty ones so the only thing that the
kernel has to do is parse the relocations. but without a real workload
we lose the impact of having to rewrite active buffers.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Instead of doing an extra getparam ioctl to discover the PCI ID, we
can snoop it when the application queries it.
Signed-off-by: Kristian Høgsberg <krh@cryptic-dragon.jf.intel.com>
DRI3 passes the drm fd over X protocol and as such we can't rely on
hooking open to discover the drm fd. Instead we look for drm ioctl
codes in the ioctl wrapper. If we don't have a drm fd and see
something that looks like a drm ioctl, we stat the fd to see if it's a
drm fd. If it is, we save it for later so we don't have to do an
extra stat on every ioctl. We can then drop the open wrapper, but we
keep the close wrapper so we can invalidate our cached drm_fd if it's
closed.
Signed-off-by: Kristian Høgsberg <krh@cryptic-dragon.jf.intel.com>
This reverts commit 0b45b0746f45deea11670a8b2c949776bbbef55c.
We want to check against our local copy of the params so we can at least
potentially catch new flag additions.
Requested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Use -1 as the invalid param since we've added more since this test was
created, causing expected EINVAL failures to return an unexpected error
code.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Recently added tools / benckmarks have the same module name as
existing tests. Android does not allow duplicate modules. This
patch appends _benchmark and _tool to the module names used when
building benckmarks and tools to prevent clashes with tests of
the same name.
Signed-off-by: Derek Morton <derek.j.morton@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A helpful function for when you want to read a whole debugfs file to a
string and don't want to worry about opening and closing file
descriptors and asserting buffer sizes.
We've been using this already for kms_frontbuffer_tracking and
kms_fbcon_fbt, so the only test with new code here is kms_fbc_crc.
Also notice that for kms_fbc_crc we had to increase the buffer size
since the file can sometimes be bigger than 64 bytes - depending on
the reason why FBC is disabled.
Of course, there are probably many other programs we can patch, but
I'm not doing this now.
v2: Add the macro to wrap sizeof() (Daniel).
v3: Add documentation for the macro too (Daniel).
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
We can't add this to igt_draw since igt_draw doesn't care whether it's
writing on a frontbuffer or not.
PS: the ENOSYS is for Kernels without the patch implementing the
IOCTL.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Just don't check the drmIoctl() return code: the "if (val)" should be
enough to prevent any problems.
v2: Don't SKIP, just proceed (Chris).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89739
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
autoconf executes the init commands after each file it creates,
causing these errors at configure time:
...
chmod: cannot access 'tools/intel_aubdump': No such file or directory
config.status: creating tools/Makefile
chmod: cannot access 'tools/intel_aubdump': No such file or directory
config.status: creating tools/intel_aubdump
config.status: creating tools/quick_dump/Makefile
...
Once it creates intel_aubdump, the errors go away. When I first added
the chmod command, I already had a tools/intel_aubdump in my tree and
didn't get the error.
Split the intel_aubdump creation into its own AC_CONFIG_FILES()
invocation to avoid this.
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
pkg-config adds /usr/include/libdrm to the include path so we only need i915_drm.h
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
This lets us capture AUB traces for platforms different from the one
we're running on.
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
This does everything the aub dump functionality in libdrm does, but
without being part of libdrm. This moves the very developer oriented
functionality out of core libdrm and adds some flexibility in how we
activate it (we can specify filename, for example). Most importantly,
this lets us dump aub files for tools and/or drivers that don't use
libdrm, without having to add that code to each of those projects.
The tool is used much like strace or valgrind. For example:
$ intel_aubdump -v --output=stuff.aub -- glxgears -geometry 500x500
will launch glxgears with its options and enable aub dumping and pass
the -v and --output=stuff.aub options to the aub dumper.
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
If we are using a streaming GGTT write into the source, we need to
trigger an early fault in order to obtain a mappable offset. This is
required when later we start reserving execbuf object top-down to try
and avoid mappable space!
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Measure the overhead of execution when doing nothing, switching between
a pair of contexts, or creating a new context every time.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
It is possible to race between unreference of the underlying BO and
importing it from prime_fd/name. Verify that the behaviour of libdrm
is consistent for prime/flink.
v2: more comments in source file, dropped extra whitespace
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Cc: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
By measuring both the query and the event round trip time, we can make a
reasonable estimate of how long it takes for the query to send the
vblank following an interrupt.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In my haste to merge the two gem_pread/gem_pwrite, I forgot to write up
the command line switch to getopt.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
