On 16/03/18 15:14, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-03-16 22:11:11)
On 13/03/18 05:31, Chris Wilson wrote:
Execute the same batch on each engine and check that the composite fence
across all engines completes only after the batch is completed on every
engine.
Signed
On 22/03/18 05:42, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2018-03-22 12:36:58)
On 22/03/2018 11:39, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2018-03-22 11:17:11)
trigger_reset(fd);
+
+ /* HACK for CI */
+ igt_assert(igt_nsec_elapsed(&ts) < 5e9);
igt_seconds_
On 22/03/18 11:14, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-03-22 17:32:46)
On 22/03/18 05:42, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2018-03-22 12:36:58)
On 22/03/2018 11:39, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2018-03-22 11:17:11)
trigger_reset(fd
On 03/04/18 04:36, Tvrtko Ursulin wrote:
From: Tvrtko Ursulin
Reset and unwedge stress testing is supposed to trigger wedging or resets
at incovenient times and then re-use the context so either the context or
driver tracking might get confused and break.
v2:
* Renamed for more sensible na
On 03/04/18 11:24, Antonio Argenziano wrote:
On 03/04/18 04:36, Tvrtko Ursulin wrote:
From: Tvrtko Ursulin
Reset and unwedge stress testing is supposed to trigger wedging or resets
at incovenient times and then re-use the context so either the context or
driver tracking might get confused
On 04/04/18 02:58, Tvrtko Ursulin wrote:
On 03/04/2018 19:34, Antonio Argenziano wrote:
On 03/04/18 11:24, Antonio Argenziano wrote:
On 03/04/18 04:36, Tvrtko Ursulin wrote:
From: Tvrtko Ursulin
Reset and unwedge stress testing is supposed to trigger wedging or
resets
at incovenient
On 12/04/18 01:43, Chris Wilson wrote:
Modesetting requires DRM_MASTER privileges, so use
drm_open_driver_master() to assert that we do acquire them.
References: https://bugs.freedesktop.org/show_bug.cgi?id=105997
Signed-off-by: Chris Wilson
Acked-by: Antonio Argenziano
On 17/05/18 08:37, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-05-17 16:08:14)
On 17/05/18 01:23, Chris Wilson wrote:
Confirm we have the available HW before asserting it succeeds.
Signed-off-by: Chris Wilson
---
tests/gem_cpu_reloc.c | 1 +
1 file changed, 1 insertion
On 17/05/18 09:52, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-05-17 17:29:26)
On 17/05/18 08:37, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-05-17 16:08:14)
On 17/05/18 01:23, Chris Wilson wrote:
Confirm we have the available HW before asserting it succeeds.
Signed
places. Also,
we have a gem_has_blt() function which is slightly different. Which is
making things quite confusing for me... I wish I had a better name but I
think we should add at least a comment to differentiate the two.
With that:
Reviewed-by: Antonio Argenziano
Thanks,
Antonio
+
On 17/05/18 02:56, Chris Wilson wrote:
If the blitter is not available, we cannot use it as a source for dirty
rectangles. We shall have to rely on the other engines to create GPU
dirty instead.
v2: Try using lots of subgroup+fixtures
Signed-off-by: Chris Wilson
TEST_MODE_ITER_BEG
spend.
References: https://bugs.freedesktop.org/show_bug.cgi?id=106640
Signed-off-by: Chris Wilson
Cc: Tomi Sarvela
LGTM:
Reviewed-by: Antonio Argenziano
---
lib/igt_core.c | 34 --
lib/igt_core.h | 1 +
tests/drv_suspend.c
; flushing it to idle in the
process!
Signed-off-by: Chris Wilson
Cc: Tvrtko Ursulin
LGTM.
Reviewed-by: Antonio Argenziano
---
tests/perf_pmu.c | 11 ++-
1 file changed, 2 insertions(+), 9 deletions(-)
diff --git a/tests/perf_pmu.c b/tests/perf_pmu.c
index 590e6526b..9af192dd8
On 30/05/18 03:33, Chris Wilson wrote:
As we measure the ring size, we never expect to find we can not submit
no batches at all. Assert against the unexpected.
Signed-off-by: Chris Wilson
Cc: Antonio Argenziano
---
lib/i915/gem_ring.c | 1 +
1 file changed, 1 insertion(+)
diff --git a
On 30/05/18 03:33, Chris Wilson wrote:
After hitting the SIGINT from execbuf, wait until the next timer signal
before trying again. This aligns the start of the ioctl to the timer,
hopefully maximising the amount of time we have for processing before
the next signal -- trying to prevent the cas
On 30/05/18 03:33, Chris Wilson wrote:
Check twice for the signal interrupting the execbuf, because the real
world is messy.
References: https://bugs.freedesktop.org/show_bug.cgi?id=106695
Signed-off-by: Chris Wilson
Cc: Antonio Argenziano
LGTM.
Reviewed-by: Antonio Argenziano
On 30/05/18 12:52, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-05-30 18:30:36)
On 30/05/18 03:33, Chris Wilson wrote:
After hitting the SIGINT from execbuf, wait until the next timer signal
before trying again. This aligns the start of the ioctl to the timer,
hopefully maximising
On 06/06/18 10:42, Chris Wilson wrote:
The current method of checking for a failed module load is flawed, as we
only report the error on probing it is not being reported back by
modprobe. So we have to dig inside the module_parameters while the
module is still loaded to discover the error.
v2:
On 15/06/18 11:56, Chris Wilson wrote:
As we hang ctx0 quite frequently, it needs to be harden against being
banned.
Signed-off-by: Chris Wilson
Cc: Mika Kuoppala
For some reason I thought we had more hang stress tests.
Reviewed-by: Antonio Argenziano
---
tests/gem_eio.c | 2 +-
1
er of 5us, which is often larger
than the target latency.
Signed-off-by: Chris Wilson
Reviewed-by: Antonio Argenziano
---
lib/igt_dummyload.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/lib/igt_dummyload.c b/lib/igt_dummyload.c
index d32b421c6..b090b8004 1
intaining HW focused tests.
Cc: Latvala Petri
Antonio Argenziano (1):
tests/gem_bad_address: Fix and update gem_bad_address
Makefile.am | 4 ++
configure.ac | 11 ++
tests_hw/Makefile.am | 36 ++
tests
tests.
Signed-off-by: Antonio Argenziano
---
Makefile.am | 4 ++
configure.ac | 11 ++
tests_hw/Makefile.am | 36 ++
tests_hw/Makefile.sources | 8
{tests => tests_hw}/gem_bad_addres
Test expects pending batches to be discarded after a reset. That is no
longer the case. Fixed to expect a normal execution.
Cc: Michel Thierry
Signed-off-by: Antonio Argenziano
---
tests/gem_reset_stats.c | 24 +++-
1 file changed, 7 insertions(+), 17 deletions(-)
diff
ned-off-by: Antonio Argenziano
---
tests/gem_reset_stats.c | 24 +++-
1 file changed, 7 insertions(+), 17 deletions(-)
diff --git a/tests/gem_reset_stats.c b/tests/gem_reset_stats.c
index c4ce4ac2..73afeeb2 100644
--- a/tests/gem_reset_stats.c
+++ b/tests/gem_reset_stats.c
@@ -23
On 20/06/17 04:45, Arkadiusz Hiler wrote:
On Fri, Jun 09, 2017 at 10:54:13AM -0700, Michel Thierry wrote:
On 6/9/2017 10:02 AM, Antonio Argenziano wrote:
Test expects pending batches to be discarded after a reset. That is no
longer the case. Fixed to expect a normal execution.
You could
On 20/06/17 11:25, Michel Thierry wrote:
From: Chris Wilson
The hang detector relies on a uevent for notification and aborting the
test. As proposed, fine-grained resets may not produce a global uevent
and so this hang detection becomes void. As we don't expect any hang, we
can just reduce th
On 28/06/17 11:37, Michel Thierry wrote:
Platforms with per-engine reset enabled (i915.reset=2) are unlikely to
perform a full chip reset, keeping the reset_count unmodified. In order
to keep the expectations of this test, enforce that full GPU reset is
enabled (i915.reset=1).
Later on, we can
On 10/10/17 01:55, Abdiel Janulgue wrote:
This improves the GEM tests section of I-G-T to make it more
suitable for CI testing
Can you provide a little more details on what is the rationale behind
this choice. Are the tests being removed only because they do not fit
the CI guidelines or is
On 10/10/17 13:19, Chris Wilson wrote:
Quoting Antonio Argenziano (2017-10-10 18:12:17)
On 10/10/17 01:55, Abdiel Janulgue wrote:
This improves the GEM tests section of I-G-T to make it more
suitable for CI testing
Can you provide a little more details on what is the rationale behind
This patch adds a context creation ioctl wrapper that returns the error
for the caller to consume. Multiple tests that implemented this already,
have been changed to use the new library function.
Signed-off-by: Antonio Argenziano
---
benchmarks/gem_exec_ctx.c | 16
benchmarks
: Antonio Argenziano
Cc: Chris Wilson
---
benchmarks/gem_exec_ctx.c | 6 +++---
benchmarks/gem_exec_trace.c | 4 ++--
lib/ioctl_wrappers.c| 44 ++--
lib/ioctl_wrappers.h| 2 ++
tests/gem_ctx_switch.c | 13 -
tests/gem_eio.c
When a client gets a fixed amount of its contexts banned, will be
blocked from creating any new contexts itself. This patch adds a test
for this feature.
Signed-off-by: Antonio Argenziano
Cc: Michel Thierry
---
tests/gem_ctx_create.c | 10 +-
tests/gem_reset_stats.c | 47
On 12/10/17 15:10, Antonio Argenziano wrote:
When a client gets a fixed amount of its contexts banned, will be
blocked from creating any new contexts itself. This patch adds a test
for this feature.
Signed-off-by: Antonio Argenziano
Cc: Michel Thierry
---
tests/gem_ctx_create.c | 10
: Antonio Argenziano
Cc: Chris Wilson
---
benchmarks/gem_exec_ctx.c | 6 +++---
benchmarks/gem_exec_trace.c | 4 ++--
lib/ioctl_wrappers.c| 46 +++--
lib/ioctl_wrappers.h| 2 ++
tests/gem_ctx_create.c | 10 +-
tests
A client that submits 'bad' contexts will be banned eventually while
other clients are not affected. Add a test for this.
v2
- Do not use a fixed number of contexts to ban client (Chris)
Signed-off-by: Antonio Argenziano
Cc: Michel Thierry
Cc: Chris Wilson
gem_has_contexts to check for contexts support and change
gem_require_contexts to skip if contests support is not available.
(Chris)
Signed-off-by: Antonio Argenziano
Cc: Chris Wilson
Cc: Michał Winiarski
---
benchmarks/gem_exec_ctx.c | 6 ++---
benchmarks/gem_exec_trace.c
gem_has_contexts to check for contexts support and change
gem_require_contexts to skip if contests support is not available.
(Chris)
v4:
- Cosmetic changes and use lib function in gem_ctx_create where
possible. (Michal)
Signed-off-by: Antonio Argenziano
Cc
On 13/02/18 01:26, Chris Wilson wrote:
icl offers a much reduced context space, and in its simplest setup we
cannot allocate one context per priority level, so trim the number and
reuse the same context for multiple priority requests.
Signed-off-by: Chris Wilson
---
tests/gem_exec_schedule.
On 13/02/18 15:10, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-02-13 22:47:35)
On 13/02/18 01:26, Chris Wilson wrote:
icl offers a much reduced context space, and in its simplest setup we
cannot allocate one context per priority level, so trim the number and
reuse the same context
On 07/02/18 01:50, Chris Wilson wrote:
Build up a large stockpile of requests, ~500,000, and feed them into the
system at 20KHz whilst simultaneously triggering set-wedged in order to
try and race i915_gem_set_wedged() against the engine->submit_request()
callback.
Signed-off-by: Chris Wilson
On 22/02/18 00:50, Chris Wilson wrote:
__igt_spin_batch_new() may be used inside a background helper which is
competing against the GPU being reset. As such, we cannot even assert
that the spin->handle is busy immediately after submission as it may
have already been reset by another client writ
On 28/02/18 07:51, Chris Wilson wrote:
Just a small variant to apply a continuous context-switch load to all
engines.
---
tests/gem_ctx_switch.c | 83 ++
1 file changed, 83 insertions(+)
diff --git a/tests/gem_ctx_switch.c b/tests/gem_ctx_swit
On 28/02/18 07:51, Chris Wilson wrote:
Execute the same batch on each engine and check that the composite fence
across all engines completes only after the batch is completed on every
engine.
Signed-off-by: Chris Wilson
LGTM.
Reviewed-by: Antonio Argenziano
---
tests/gem_exec_fence.c
On 28/02/18 23:51, Chris Wilson wrote:
Just a small variant to apply a continuous context-switch load to all
engines.
v2: Adapt to for_each_physical_engine() and sane gem_context_create()
Signed-off-by: Chris Wilson
Cc: Antonio Argenziano
LGTM.
Reviewed-by: Antonio Argenziano
Signed-off-by: Chris Wilson
Cc: Antonio Argenziano
Reviewed-by: Antonio Argenziano
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
ssert(found);
Test changes look fine to me, failures on CI seems to be caused by the
test not waiting for reset to happen only before we would have not
caught it.
Reviwed-by: Antonio Argenziano
}
static void test_error_state_capture(unsigned ring_id,
___
On 07/03/18 14:49, Chris Wilson wrote:
Exercise some new API that allows applications to request that
individual contexts are executed within a desired frequency range.
Signed-off-by: Chris Wilson
---
tests/Makefile.am | 2 +-
tests/Makefile.sources | 1 +
tests/gem_ctx_freq.c
On 07/03/18 17:18, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-03-08 00:55:47)
On 07/03/18 14:49, Chris Wilson wrote:
+static void single(int fd, const struct intel_execution_engine *e)
+{
+ const unsigned int engine = e->exec_id | e->flags;
+ uint32
: Antonio Argenziano
---
tests/Makefile.am | 1 +
tests/Makefile.sources | 1 +
tests/gem_ctx_freq.c | 604 +
tests/meson.build | 1 +
4 files changed, 607 insertions(+)
create mode 100644 tests/gem_ctx_freq.c
+static void
On 08/03/18 17:03, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-03-09 00:45:42)
On 08/03/18 09:13, Chris Wilson wrote:
Exercise some new API that allows applications to request that
individual contexts are executed within a desired frequency range.
v2: Split single/continuous
after each invalid request
v4: Check the frequencies reported by the kernel across the entire
range.
v5: Rewrite sandwich to create a sandwich between multiple concurrent
engines.
Signed-off-by: Chris Wilson
Cc: Praveen Paneri
Cc: Sagar A Kamble
Cc: Antonio Argenziano
LGTM.
Reviewed-by: Antonio
for a long time before actually preempting.
The testcase added here, stimulates this scenario where a high priority
request is sent while another process keeps submitting requests and
filling its ringbuffer.
Cc: Chris Wilson
Signed-off-by: Antonio Argenziano
---
tests/gem_ringfill.c | 140
On 15/08/17 15:35, Chris Wilson wrote:
Quoting Antonio Argenziano (2017-08-15 22:44:21)
Sending as RFC to get feedback on what would be the correct behaviour of
the driver and, therefore, if the test is valid.
It's not a preemption specific bug. It is fair to say that any client
blo
.
- Bound processes to same cpu and wake with signals.
Cc: Chris Wilson
Signed-off-by: Antonio Argenziano
---
tests/gem_exec_schedule.c | 113 +-
1 file changed, 112 insertions(+), 1 deletion(-)
diff --git a/tests/gem_exec_schedule.c b/tests
This patch introduces a fixture at the end of the test to destroy
objects that have been created and stop the hang detector.
Cc: Chris Wilson
Signed-off-by: Antonio Argenziano
---
tests/gem_exec_reuse.c | 7 +++
1 file changed, 7 insertions(+)
diff --git a/tests/gem_exec_reuse.c b/tests
On 08/12/17 09:07, Antonio Argenziano wrote:
On 08/12/17 08:46, Chris Wilson wrote:
Quoting Antonio Argenziano (2017-12-08 16:27:17)
The test expected IOCTL 'I915_GET_RESET_STATS' would return an error
when not root. That is no longer true in the driver and therefore
th
Instead of returning a zero value for non root users, return an EPERM
error.
Signed-off-by: Antonio Argenziano
Cc: Chris Wilson
Cc: Michel Thierry
---
drivers/gpu/drm/i915/i915_drv.c | 2 +-
drivers/gpu/drm/i915/i915_gem_context.c | 9 -
2 files changed, 5 insertions(+), 6
On 16/01/18 09:27, Chris Wilson wrote:
Quoting Antonio Argenziano (2018-01-16 16:09:35)
Instead of returning a zero value for non root users, return an EPERM
error.
What? reset-stats are per-context, private to the client, with the
exception of the global value which is solely protected by
On 13/12/17 16:48, Antonio Argenziano wrote:
The test expected IOCTL 'I915_GET_RESET_STATS' would return an error
when not root. That is no longer true in the driver since commit
4c9c0d09741d ("drm/i915: Fix retrieval of hangcheck stats") and therefore
the test was inco
hanged the behaviour in the Driver to the
commit message. (Michel)
v3:
- Reuse get_reset_count instead of implementing a new function.
(Michel)
Cc: Michel Thierry
Cc: Arkadiusz Hiler
Cc: Chris Wilson
Signed-off-by: Antonio Argenziano
---
tests/gem_reset_stats.c |
On 21/01/18 12:20, Chris Wilson wrote:
Apply a random load to one or all engines in order to apply stress to
RPS as it tries to constantly adjust the GPU frequency to meet the
changing workload.
Signed-off-by: Chris Wilson
---
tests/Makefile.sources | 1 +
tests/gem_exec_load.c | 124 ++
do {\
+ uint64_t elapsed = igt_nsec_elapsed(&tv) >> 20; \
Maybe tv_ and elapsed_ just for consistency.
Reviewed-by: Anto
On 29/01/18 06:41, Chris Wilson wrote:
Previously, we relied on only running the hangcheck while somebody was
waiting on the GPU, in order to minimise the amount of time hangcheck
had to run. (If nobody was watching the GPU, nobody would notice if the
GPU wasn't responding -- eventually somebod
: Chris Wilson
LGTM.
Reviewed-by: Antonio Argenziano
---
tests/prime_mmap_coherency.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/tests/prime_mmap_coherency.c b/tests/prime_mmap_coherency.c
index 192b43489..04b15dddb 100644
--- a/tests/prime_mmap_coherency.c
+++ b/tests
: Chris Wilson
Acked-by: Antonio Argenziano
---
tests/Makefile.sources | 1 +
tests/gem_exec_load.c | 176 +
2 files changed, 177 insertions(+)
create mode 100644 tests/gem_exec_load.c
diff --git a/tests/Makefile.sources b/tests
On 09/02/18 06:30, Chris Wilson wrote:
During the startup, we try to determine if the system supports rc6 prior
to testing. However, the startup check asserts the residency debugfs
exists, instead of testing for a requirement.
Signed-off-by: Chris Wilson
Reviewed-by: Antonio Argenziano
+1428,20 @@ void igt_require_gem(int fd)
* sequences of batches.
*/
err = igt_debugfs_dir(fd);
I would prefer to use two variables, one for the debugfs fd and one for
-errno for the IOCTL below.
With or without the new variable,
Reviewed-by: Antonio Argenziano
Added functions for flushing a batch buffer as privileged.
Signed-off-by: Antonio Argenziano
---
lib/intel_batchbuffer.c | 45 +
lib/intel_batchbuffer.h | 4 +++-
2 files changed, 44 insertions(+), 5 deletions(-)
diff --git a/lib/intel_batchbuffer.c
The test purpose is to write at an invalid GTT location. To do so, the
store instruction used in the test has been updated to use the correct
value and extra unneeded flags have been removed. The batch buffer is
also sent as privileged now.
Signed-off-by: Antonio Argenziano
---
tests
The intent of the test is to provide an incorrect input to a command.
Since the driver doesn't do any command validation, this test is removed
as non applicable.
Signed-off-by: Antonio Argenziano
---
tests/Makefile.sources | 1 -
tests/gem_bad_address.c
("drm/i915/bxt: Broxton decoupled MMIO).
Reported-by: Antonio Argenziano
Signed-off-by: Michel Thierry
Tested-By: Antonio Argenziano
---
drivers/gpu/drm/i915/i915_debugfs.c | 4
1 file changed, 4 insertions(+)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c
b/drivers/gpu/drm/i915
101 - 171 of 171 matches
Mail list logo
