Re: ✗ Xe.CI.Full: failure for PF: Update auto-provisioning

["Bernatowicz, Marcin" <marcin.bernatowicz@linux.intel.com>]

On 10/16/2025 12:55 PM, Michal Wajdeczko wrote:
>
> On 10/16/2025 12:12 AM, Patchwork wrote:
>> *Patch Details*
>> *Series:*	PF: Update auto-provisioning
>> *URL:*	https://patchwork.freedesktop.org/series/155971/ <https://patchwork.freedesktop.org/series/155971/>
>> *State:*	failure
>> *Details:*	https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/index.html <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/index.html>
>>
>>
>>    CI Bug Log - changes from xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a_FULL -> xe-pw-155971v1_FULL
>>
>>
>>      Summary
>>
>> *FAILURE*
>>
>> Serious unknown changes coming with xe-pw-155971v1_FULL absolutely need to be
>> verified manually.
>>
>> If you think the reported changes have nothing to do with the changes
>> introduced in xe-pw-155971v1_FULL, please notify your bug team (I915-ci-infra@lists.freedesktop.org) to allow them
>> to document this new failure mode, which will reduce false positives in CI.
>>
>>
>>      Participating hosts (4 -> 4)
>>
>> No changes in participating hosts
>>
>>
>>      Possible new issues
>>
>> Here are the unknown changes that may have been introduced in xe-pw-155971v1_FULL:
>>
>>
>>        IGT changes
>>
>>
>>          Possible regressions
>>
>>    *
>>
>>      igt@kms_big_fb@linear-8bpp-rotate-180:
>>
>>        o shard-bmg: PASS <https://intel-gfx-ci.01.org/tree/intel-xe/xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a/shard-bmg-3/igt@kms_big_fb@linear-8bpp-rotate-180.html> -> FAIL <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-8/igt@kms_big_fb@linear-8bpp-rotate-180.html> +22 other tests fail
> unrelated
>
> (kms_big_fb:12070) igt_fb-DEBUG: igt_create_fb_with_bo_size(width=16384, height=16384, format=C8  (0x20203843), modifier=0x0, size=0)
> (kms_big_fb:12070) igt_kms-CRITICAL: Test assertion failure function kmstest_dumb_create, file ../lib/igt_kms.c:1316:
> (kms_big_fb:12070) igt_kms-CRITICAL: Failed assertion: igt_ioctl((fd), ((((2U|1U) << (((0+8)+8)+14)) | ((('d')) << (0+8)) | (((0xB2)) << 0) | ((((sizeof(struct drm_mode_create_dumb)))) << ((0+8)+8)))), (&create)) == 0
> (kms_big_fb:12070) igt_kms-CRITICAL: Last errno: 12, Cannot allocate memory
>
>
>>    *
>>
>>      igt@kms_big_fb@linear-max-hw-stride-64bpp-rotate-0:
>>
>>        o shard-bmg: NOTRUN -> FAIL <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-4/igt@kms_big_fb@linear-max-hw-stride-64bpp-rotate-0.html> +4 other tests fail
>>    *
>>
>>      igt@kms_color@legacy-gamma@pipe-a-dp-2:
>>
>>        o shard-bmg: PASS <https://intel-gfx-ci.01.org/tree/intel-xe/xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a/shard-bmg-1/igt@kms_color@legacy-gamma@pipe-a-dp-2.html> -> SKIP <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-6/igt@kms_color@legacy-gamma@pipe-a-dp-2.html>
>>    *
>>
>>      igt@kms_cursor_legacy@cursor-vs-flip-atomic-transitions-varying-size:
>>
>>        o shard-adlp: NOTRUN -> FAIL <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-adlp-6/igt@kms_cursor_legacy@cursor-vs-flip-atomic-transitions-varying-size.html> +1 other test fail
> unrelated
>
> (kms_cursor_legacy:2966) CRITICAL: Test assertion failure function cursor_vs_flip, file ../tests/kms_cursor_legacy.c:1387:
> (kms_cursor_legacy:2966) CRITICAL: Failed assertion: igt_ioctl((display->drm_fd), ((((2U|1U) << (((0+8)+8)+14)) | ((('d')) << (0+8)) | (((0xA3)) << 0) | ((((sizeof(struct drm_mode_cursor)))) << ((0+8)+8)))), (&arg[i & 1])) == 0
> (kms_cursor_legacy:2966) CRITICAL: Last errno: 28, No space left on device
>
>
>>    *
>>
>>      igt@kms_vblank@ts-continuation-dpms-suspend@pipe-a-dp-2:
>>
>>        o shard-bmg: NOTRUN -> ABORT <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-3/igt@kms_vblank@ts-continuation-dpms-suspend@pipe-a-dp-2.html> +1 other test abort
> unrelated
>
> <4> [462.803151] xe 0000:03:00.0: [drm] Assertion `!(bo->flags & ((((1UL))) << (19))) || !iosys_map_is_null(&bo->vmap)` failed!
> platform: BATTLEMAGE subplatform: 7
> graphics: Xe2_HPG 20.01 step A0
> media: Xe2_HPM 13.01 step A1
> <4> [462.803185] WARNING: CPU: 1 PID: 3847 at drivers/gpu/drm/xe/xe_bo_evict.c:217 xe_bo_restore_and_map_ggtt+0x22c/0x260 [xe]
> <4> [462.803496] RIP: 0010:xe_bo_restore_and_map_ggtt+0x22c/0x260 [xe]
> <4> [462.803633] Call Trace:
> <4> [462.803635]  <TASK>
> <4> [462.803648]  xe_bo_apply_to_pinned+0xdf/0x220 [xe]
> <4> [462.803748]  ? __pfx_xe_bo_restore_and_map_ggtt+0x10/0x10 [xe]
> <4> [462.803848]  xe_bo_restore_late+0x2d/0x130 [xe]
> <4> [462.803943]  xe_pm_resume+0x21d/0x440 [xe]
> <4> [462.804117]  ? pci_read_config_word+0x27/0x50
> <4> [462.804126]  ? __pci_set_master+0x33/0xe0
> <4> [462.804141]  xe_pci_resume+0x60/0x70 [xe]
>
>
>
>>    *
>>
>>      igt@xe_compute_preempt@compute-preempt-many-all-ram:
>>
>>        o shard-dg2-set2: NOTRUN -> SKIP <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-dg2-436/igt@xe_compute_preempt@compute-preempt-many-all-ram.html>
> unrelated
>
> <4> [388.734979] xe 0000:03:00.0: [drm] VM worker error: -12
>
>
>>    *
>>
>>      igt@xe_evict@evict-beng-mixed-threads-small-multi-vm:
>>
>>        o shard-bmg: PASS <https://intel-gfx-ci.01.org/tree/intel-xe/xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a/shard-bmg-2/igt@xe_evict@evict-beng-mixed-threads-small-multi-vm.html> -> DMESG-FAIL <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-2/igt@xe_evict@evict-beng-mixed-threads-small-multi-vm.html>
>>    *
>>
>>      igt@xe_evict_ccs@evict-overcommit-parallel-instantfree-samefd:
>>
>>        o shard-bmg: PASS <https://intel-gfx-ci.01.org/tree/intel-xe/xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a/shard-bmg-5/igt@xe_evict_ccs@evict-overcommit-parallel-instantfree-samefd.html> -> ABORT <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-bmg-7/igt@xe_evict_ccs@evict-overcommit-parallel-instantfree-samefd.html> +6 other tests abort
>
> unrelated
>
> <1> [331.175806] BUG: kernel NULL pointer dereference, address: 0000000000000001
> <1> [331.175819] #PF: supervisor read access in kernel mode
> <1> [331.175824] #PF: error_code(0x0000) - not-present page
> <4> [331.175857] RIP: 0010:xe_vm_insert_vma+0x1c/0x1b0 [xe]
> <4> [331.176097] Call Trace:
> <4> [331.176100]  <TASK>
> <4> [331.176103]  ? vm_bind_ioctl_ops_create+0xa9/0x8d0 [xe]
> <4> [331.176361]  ? lock_is_held_type+0xa3/0x130
> <4> [331.176371]  vm_bind_ioctl_ops_parse+0x709/0xa00 [xe]
> <4> [331.176541]  xe_vm_bind_ioctl+0x1103/0x1850 [xe]
> <4> [331.176728]  ? __pfx_xe_vm_bind_ioctl+0x10/0x10 [xe]
> <4> [331.176950]  drm_ioctl_kernel+0xb3/0x120
> <4> [331.176958]  ? drm_ioctl_kernel+0xb3/0x120
> <4> [331.176965]  drm_ioctl+0x2d4/0x5a0
> <4> [331.176970]  ? __pfx_xe_vm_bind_ioctl+0x10/0x10 [xe]
> <4> [331.177144]  ? trace_hardirqs_on+0x1e/0xd0
> <4> [331.177150]  ? _raw_spin_unlock_irqrestore+0x51/0x80
> <4> [331.177156]  ? __pm_runtime_resume+0x5b/0x90
> <4> [331.177163]  ? xe_pm_runtime_get_ioctl+0x61/0x150 [xe]
> <4> [331.177249]  xe_drm_ioctl+0x53/0x80 [xe]
>
>
>>    *
>>
>>      igt@xe_sriov_auto_provisioning@fair-allocation:
>>
>>        o shard-adlp: PASS <https://intel-gfx-ci.01.org/tree/intel-xe/xe-3924-6a9cafb6723c47062cc4d593804d5bb69155d81a/shard-adlp-2/igt@xe_sriov_auto_provisioning@fair-allocation.html> -> SKIP <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-adlp-8/igt@xe_sriov_auto_provisioning@fair-allocation.html>
> but
>
> Test requirement not met in function __igt_unique____real_main316, file ../tests/intel/xe_sriov_auto_provisioning.c:343:
> Test requirement: !(ret)
> contexts_provisioned: Failed ranges check on gt0 (-34)
>
> this is now somehow expected, as it looks that some earlier test did some tweaks in the VFs provisioning causing 'auto' mode to be disabled:
>
> <7>[  344.422273] xe 0000:00:02.0: [drm:pf_update_policy_bool [xe]] PF: Tile0: GT0: policy key 0x8001 'sched_if_idle' updated to 'disabled'
> <7>[  344.422465] xe 0000:00:02.0: [drm:xe_sriov_pf_provision_set_mode [xe]] PF: mode auto changed to custom by sched_if_idle_set [xe]
>
> so there is no more VFs auto-unprovisioning, and some of VFs resources may remain allocated and must be explicitly released by the test
>
> FYI: @Marcin

There are two tests, xe_pmu and xe_sriov_scheduling, which switch to 
custom mode due to scheduling parameter modifications. These are 
addressed by the series.

>>    *
>>
>>      igt@xe_sriov_auto_provisioning@resources-released-on-vfs-disabling:
>>
>>        o shard-adlp: NOTRUN -> SKIP <https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-155971v1/shard-adlp-6/igt@xe_sriov_auto_provisioning@resources-released-on-vfs-disabling.html>
>>
>>