On 07/03/2025 07:21, Philipp Stanner wrote:
On Thu, 2025-03-06 at 17:10 +0000, Tvrtko Ursulin wrote:
On 06/03/2025 15:18, Tvrtko Ursulin wrote:
On 06/03/2025 14:56, Tvrtko Ursulin wrote:
On 06/03/2025 12:37, Philipp Stanner wrote:
On Tue, 2025-03-04 at 13:10 +0000, Tvrtko Ursulin wrote:
There has repeatedly been quite a bit of apprehension when
any change
to the DRM
scheduler is proposed, with two main reasons being code base
is
considered
fragile, not well understood and not very well documented,
and
secondly the lack
of systematic testing outside the vendor specific tests
suites and/or
test
farms.
This series is an attempt to dislodge this status quo by
adding some
unit tests
using the kunit framework.
General approach is that there is a mock "hardware" backend
which can
be
controlled from tests, which in turn allows exercising
various
scheduler code
paths.
Only some simple basic tests get added in the series and
hopefully it
is easy to
understand what tests are doing.
An obligatory "screenshot" for reference:
[14:29:37] ============ drm_sched_basic_tests (3 subtests)
============
[14:29:38] [PASSED] drm_sched_basic_submit
[14:29:38] ================== drm_sched_basic_test
===================
[14:29:38] [PASSED] A queue of jobs in a single entity
[14:29:38] [PASSED] A chain of dependent jobs across multiple
entities
[14:29:38] [PASSED] Multiple independent job queues
[14:29:38] [PASSED] Multiple inter-dependent job queues
[14:29:38] ============== [PASSED] drm_sched_basic_test
===============
[14:29:38] [PASSED] drm_sched_basic_entity_cleanup
[14:29:38] ============== [PASSED] drm_sched_basic_tests
==============
[14:29:38] ======== drm_sched_basic_timeout_tests (1 subtest)
=========
[14:29:40] [PASSED] drm_sched_basic_timeout
[14:29:40] ========== [PASSED] drm_sched_basic_timeout_tests
==========
[14:29:40] ======= drm_sched_basic_priority_tests (2
subtests)
========
[14:29:42] [PASSED] drm_sched_priorities
[14:29:42] [PASSED] drm_sched_change_priority
[14:29:42] ========= [PASSED] drm_sched_basic_priority_tests
==========
[14:29:42] ====== drm_sched_basic_modify_sched_tests (1
subtest)
======
[14:29:43] [PASSED] drm_sched_test_modify_sched
[14:29:43] ======= [PASSED]
drm_sched_basic_modify_sched_tests
========
[14:29:43]
============================================================
[14:29:43] Testing complete. Ran 10 tests: passed: 10
[14:29:43] Elapsed time: 13.330s total, 0.001s configuring,
4.005s
building, 9.276s running
Yo,
so I tried to test this all this in QEMU and I am encountering
some
explosions when I activate the scheduler tests. Just DRM tests
boot
fine.
I'm using a kernel on relatively current drm-misc-next:
44d2f310f008
I apply your series, then
make defconfig
make menuconfig # switch on kunit framework and scheduler tests
install everything + initramfs
Boot then causes errors as below. Just using the DRM kunit
tests works
fine.
Excerpt of the first fault:
[ 1.040513] # kunit_device: pass:3 fail:0 skip:0 total:3
[ 1.040867] # Totals: pass:3 fail:0 skip:0 total:3
[ 1.041296] ok 7 kunit_device
[ 1.041936] KTAP version 1
[ 1.042186] # Subtest: kunit_fault
[ 1.042517] # module: kunit_test
[ 1.042517] 1..1
[ 1.043147] BUG: kernel NULL pointer dereference, address:
0000000000000000
[ 1.043765] #PF: supervisor write access in kernel mode
[ 1.044189] #PF: error_code(0x0002) - not-present page
[ 1.044617] PGD 0 P4D 0
[ 1.044818] Oops: Oops: 0002 [#1] PREEMPT SMP PTI
[ 1.045380] CPU: 7 UID: 0 PID: 214 Comm: kunit_try_catch
Tainted:
G N 6.14.0-rc4-00387-g33e4632926a0 #8
[ 1.046262] Tainted: [N]=TEST
[ 1.046521] Hardware name: QEMU Standard PC (i440FX + PIIX,
1996),
BIOS 1.16.3-2.fc40 04/01/2014
[ 1.047224] RIP: 0010:kunit_test_null_dereference+0x37/0x80
[ 1.047706] Code: 80 b5 49 c7 c0 50 7f 56 b4 ba 01 00 00 00
65 48
8b 04 25 28 00 00 00 48 89 44 24 08 31 c0 48 8d 4c 24 07 48 c7
c6 80
8a 26 b5 <c7> 04 25 00 00 00 00 00 00 00 00 48 c7 87 70 01 00
00 a6
e9 8c b5
[ 1.049204] RSP: 0000:ffffa609807c7ec8 EFLAGS: 00010246
[ 1.049642] RAX: 0000000000000000 RBX: ffff91d982623000 RCX:
ffffa609807c7ecf
[ 1.050213] RDX: 0000000000000001 RSI: ffffffffb5268a80 RDI:
ffffa60980013c68
[ 1.050799] RBP: ffff91d98105afc0 R08: ffffffffb4567f50 R09:
ffffffffb5807ce8
[ 1.051375] R10: 0000000000000000 R11: 0000000000000001 R12:
ffff91d98105afc0
[ 1.051941] R13: ffff91d983c749c0 R14: ffffffffb45685e0 R15:
ffff91d982623000
[ 1.052543] FS: 0000000000000000(0000)
GS:ffff91e48f9c0000(0000)
knlGS:0000000000000000
[ 1.053187] CS: 0010 DS: 0000 ES: 0000 CR0:
0000000080050033
[ 1.053649] CR2: 0000000000000000 CR3: 00000004cee30000 CR4:
00000000000006f0
[ 1.054214] Call Trace:
[ 1.054427] <TASK>
[ 1.054597] ? __die+0x1e/0x60
[ 1.054844] ? page_fault_oops+0x17b/0x4a0
[ 1.055174] ? search_extable+0x26/0x30
[ 1.055482] ? kunit_test_null_dereference+0x37/0x80
[ 1.055888] ? search_module_extables+0x14/0x50
[ 1.056255] ? exc_page_fault+0x6b/0x150
[ 1.056571] ? asm_exc_page_fault+0x26/0x30
[ 1.056898] ?
__pfx_kunit_generic_run_threadfn_adapter+0x10/0x10
[ 1.057387] ? __pfx_kunit_fail_assert_format+0x10/0x10
[ 1.057799] ? kunit_test_null_dereference+0x37/0x80
[ 1.058195] ? __kthread_parkme+0x33/0x80
[ 1.058523] kunit_generic_run_threadfn_adapter+0x1c/0x40
[ 1.058949] kthread+0xe9/0x1f0
[ 1.059206] ? __pfx_kthread+0x10/0x10
[ 1.059513] ret_from_fork+0x2f/0x50
[ 1.059798] ? __pfx_kthread+0x10/0x10
[ 1.060095] ret_from_fork_asm+0x1a/0x30
[ 1.060421] </TASK>
[ 1.060597] Modules linked in:
[ 1.060841] CR2: 0000000000000000
[ 1.061104] ---[ end trace 0000000000000000 ]---
[ 1.061481] RIP: 0010:kunit_test_null_dereference+0x37/0x80
I attach my kernel config and the full log file.
What's awkward is that it does not seem to be related directly
to
sched, but only faults with sched.
Could you try to reproduce this, Tvrtko?
Any chance that between the two runs you somehow manage to enable
CONFIG_KUNIT_FAULT_TEST?
Hmm in the sea of kunit_test_null_dereference there was a drm sched
related fail too. Investigating.
Well this was quite embarrassing - I had an use after free due
relying
on scheduler fences for querying job status. That's what I get for
over-relying on KASAN...
Interesting, so KASAN had a false negative with those?
No, just that it cannot see UAF if it doesn't happen due timing. I was
referring to the false sense of security of "I am running with KASAN
therefore there cannot possibly be bugs." :)
I've fixed it by tracking the job completion status in the mock job
object directly and sent v4 out.
Also interestingly, for me testing under qemu failed to catch it.
Only
one out of two real hw test machines hit it.
That's very awkward. How can my QEMU be that different from your QEMU?
Did you use the commit I provided above as the base of your kernel,
too?
It was just about timing. Scheduler free job worker had to run before
the test dereferenced job->s_fence one last time. I replaced that with a
direct status check on the container object in the new version.
Regards,
Tvrtko
Excellent that you gave it
a spin and caught it before merge, thanks for that!
No worries, that's our mission. Cool that you found it so quickly!
P.
Regards,
Tvrtko
Regards,
Tvrtko
Thanks
P.
v2:
* Parameterize a bunch of similar tests.
* Improve test commentary.
* Rename TDR test to timeout. (Christian)
* Improve quality and consistency of naming. (Philipp)
RFC v2 -> series v1:
* Rebased for drm_sched_init changes.
* Fixed modular build.
* Added some comments.
* Filename renames. (Philipp)
v2:
* Dealt with a bunch of checkpatch warnings.
v3:
* Some mock API renames, kerneldoc grammar fixes and
indentation
fixes.
Cc: Christian König <christian.koenig@xxxxxxx>
Cc: Danilo Krummrich <dakr@xxxxxxxxxx>
Cc: Matthew Brost <matthew.brost@xxxxxxxxx>
Cc: Philipp Stanner <phasta@xxxxxxxxxx>
Tvrtko Ursulin (5):
drm: Move some options to separate new Kconfig
drm/scheduler: Add scheduler unit testing infrastructure
and some
basic tests
drm/scheduler: Add a simple timeout test
drm/scheduler: Add basic priority tests
drm/scheduler: Add a basic test for modifying entities
scheduler
list
drivers/gpu/drm/Kconfig | 109 +----
drivers/gpu/drm/Kconfig.debug | 115 +++++
drivers/gpu/drm/scheduler/.kunitconfig | 12 +
drivers/gpu/drm/scheduler/Makefile | 2 +
drivers/gpu/drm/scheduler/tests/Makefile | 7 +
.../gpu/drm/scheduler/tests/mock_scheduler.c | 323
+++++++++++++
drivers/gpu/drm/scheduler/tests/sched_tests.h | 223
+++++++++
drivers/gpu/drm/scheduler/tests/tests_basic.c | 426
++++++++++++++++++
8 files changed, 1113 insertions(+), 104 deletions(-)
create mode 100644 drivers/gpu/drm/Kconfig.debug
create mode 100644 drivers/gpu/drm/scheduler/.kunitconfig
create mode 100644 drivers/gpu/drm/scheduler/tests/Makefile
create mode 100644
drivers/gpu/drm/scheduler/tests/mock_scheduler.c
create mode 100644
drivers/gpu/drm/scheduler/tests/sched_tests.h
create mode 100644
drivers/gpu/drm/scheduler/tests/tests_basic.c
