Our two baytrails starting reporting the following warning randomly when running igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b, starting from CI_DRM_2783: [ 446.057678] Unclaimed read from register 0x1f0034 [ 446.057755] ------------[ cut here ]------------ [ 446.057816] WARNING: CPU: 0 PID: 2936 at drivers/gpu/drm/i915/intel_uncore.c:792 __unclaimed_reg_debug+0x3e/0x50 [i915] [ 446.057820] Modules linked in: vgem snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_codec intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul snd_hwdep ghash_clmulni_intel snd_hda_core snd_pcm i915 r8169 mii lpc_ich i2c_hid i2c_designware_platform i2c_designware_core prime_numbers [ 446.057915] CPU: 0 PID: 2936 Comm: kworker/u8:14 Tainted: G U 4.12.0-CI-CI_DRM_2804+ #1 [ 446.057919] Hardware name: GIGABYTE GB-BXBT-1900/MZBAYAB-00, BIOS F6 02/17/2015 [ 446.057928] Workqueue: events_unbound async_run_entry_fn [ 446.057936] task: ffff88010d2e8040 task.stack: ffffc90000430000 [ 446.057991] RIP: 0010:__unclaimed_reg_debug+0x3e/0x50 [i915] [ 446.057996] RSP: 0018:ffffc90000433978 EFLAGS: 00010096 [ 446.058004] RAX: 0000000000000025 RBX: 0000000000000000 RCX: 0000000000000002 [ 446.058009] RDX: 0000000000000000 RSI: ffffffff81cba24f RDI: ffffffff81c99957 [ 446.058013] RBP: ffffc90000433990 R08: 0000000000000000 R09: 0000000000000001 [ 446.058017] R10: ffffc90000433908 R11: fc8c7a4200000000 R12: 00000000001f0034 [ 446.058021] R13: 0000000000000001 R14: 00000000ffffffff R15: ffff8801321c0b98 [ 446.058026] FS: 0000000000000000(0000) GS:ffff88013fc00000(0000) knlGS:0000000000000000 [ 446.058031] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 446.058035] CR2: 00007f0b0d26f028 CR3: 0000000135678000 CR4: 00000000001006f0 [ 446.058039] Call Trace: [ 446.058099] fwtable_read32+0x278/0x2c0 [i915] [ 446.058153] vlv_program_watermarks+0x41e/0x770 [i915] [ 446.058213] vlv_optimize_watermarks+0xa1/0xc0 [i915] [ 446.058273] intel_atomic_commit_tail+0x339/0xf70 [i915] [ 446.058343] intel_atomic_commit+0x3fb/0x500 [i915] [ 446.058353] ? drm_atomic_check_only+0x370/0x560 [ 446.058405] ? intel_runtime_pm_put+0x51/0xa0 [i915] [ 446.058416] drm_atomic_commit+0x46/0x50 [ 446.058425] drm_atomic_helper_commit_duplicated_state+0xbf/0xd0 [ 446.058482] __intel_display_resume+0x81/0xc0 [i915] [ 446.058541] intel_display_resume+0xca/0xf0 [i915] [ 446.058601] i915_pm_restore+0xef/0x190 [i915] [ 446.058652] i915_pm_resume+0x9/0x10 [i915] [ 446.058659] pci_pm_resume+0x5f/0x90 [ 446.058668] dpm_run_callback+0x6f/0x330 [ 446.058673] ? pci_pm_thaw+0x90/0x90 [ 446.058682] device_resume+0xac/0x1e0 [ 446.058691] ? dpm_watchdog_set+0x60/0x60 [ 446.058703] async_resume+0x18/0x40 [ 446.058709] async_run_entry_fn+0x34/0x160 [ 446.058719] process_one_work+0x1fe/0x670 [ 446.058732] worker_thread+0x49/0x3b0 [ 446.058744] kthread+0x10f/0x150 [ 446.058750] ? process_one_work+0x670/0x670 [ 446.058756] ? kthread_create_on_node+0x40/0x40 [ 446.058765] ret_from_fork+0x27/0x40 [ 446.058781] Code: de ff ff 38 d8 76 2d 45 84 ed 48 c7 c0 9d 89 18 a0 48 c7 c6 a7 89 18 a0 48 0f 45 f0 44 89 e2 48 c7 c7 b0 89 18 a0 e8 7b a6 0c e1 <0f> ff 83 2d 45 32 12 00 01 5b 41 5c 41 5d 5d c3 66 90 55 48 89 [ 446.059026] ---[ end trace cc38744b313e72cf ]--- This may be linked to this bug: https://bugs.freedesktop.org/show_bug.cgi?id=101516
Just adding a me too comment here. FWIW here is my almost identical backtrace: [ 56.955016] PM: early resume of devices complete after 1068.631 msecs [ 56.956396] pcieport 0000:00:1c.0: System wakeup disabled by ACPI [ 56.957697] rtc_cmos 00:04: System wakeup disabled by ACPI [ 56.958261] Suspended for 0.888 seconds [ 56.964250] Unclaimed read from register 0x1f0034 [ 56.964347] ------------[ cut here ]------------ [ 56.964428] WARNING: CPU: 2 PID: 1799 at drivers/gpu/drm/i915/intel_uncore.c:801 __unclaimed_reg_debug+0x4e/0x60 [i915] ... [ 56.964590] Workqueue: events_unbound async_run_entry_fn [ 56.964594] task: ffff89ad698b0000 task.stack: ffff9ba401454000 [ 56.964662] RIP: 0010:__unclaimed_reg_debug+0x4e/0x60 [i915] ... [ 56.964688] Call Trace: [ 56.964758] fwtable_read32+0x17a/0x1d0 [i915] [ 56.964818] vlv_program_watermarks+0x3c4/0x610 [i915] [ 56.964883] ? intel_hdmi_get_hw_state+0x27/0xd0 [i915] [ 56.964941] vlv_optimize_watermarks+0x95/0xb0 [i915] [ 56.965008] intel_atomic_commit_tail+0x2e2/0x1030 [i915] [ 56.965016] ? tracing_record_cmdline+0x32/0x120 [ 56.965022] ? __schedule+0x23e/0x860 [ 56.965090] intel_atomic_commit+0x399/0x4b0 [i915] [ 56.965125] ? drm_atomic_check_only+0x37f/0x540 [drm] [ 56.965155] drm_atomic_commit+0x4b/0x50 [drm] [ 56.965174] drm_atomic_helper_commit_duplicated_state+0xc2/0xd0 [drm_kms_helper] [ 56.965245] __intel_display_resume+0x85/0xc0 [i915] [ 56.965312] intel_display_resume+0xf7/0x120 [i915] [ 56.965370] i915_drm_resume+0xe1/0x180 [i915] [ 56.965427] i915_pm_resume+0x1e/0x30 [i915] [ 56.965434] pci_pm_resume+0x65/0xa0 [ 56.965440] dpm_run_callback+0x57/0x140 [ 56.965444] ? pci_pm_thaw+0x90/0x90 [ 56.965447] device_resume+0xe1/0x200 [ 56.965450] async_resume+0x1d/0x50 [ 56.965455] async_run_entry_fn+0x39/0x170 [ 56.965460] process_one_work+0x193/0x3c0 ... [ 57.062736] PM: resume of devices complete after 107.709 msecs [ 57.063468] PM: resume devices took 0.108 seconds [ 57.063478] PM: Finishing wakeup. Regards, Hans
I've not thought through the issue in detail, but I think one potential way to fix this would to add the MODESET domain to the display power well. That should also mean that we could again eliminate 886015a0ad43 ("drm/i915: reintroduce VLV/CHV PFI programming power domain workaround").
(In reply to Ville Syrjala from comment #2) > I've not thought through the issue in detail, but I think one potential way > to fix this would to add the MODESET domain to the display power well. That > should also mean that we could again eliminate 886015a0ad43 ("drm/i915: > reintroduce VLV/CHV PFI programming power domain workaround"). Hi Ville, I have been playing with this for a while, mostly blindly, since I don't have documentation on power domains. In fact, I had tried this suggestion beforebut I gave another try and it didn't make a difference: https://lists.freedesktop.org/archives/intel-gfx-trybot/2017-October/023994.html
Seems first patch in https://patchwork.freedesktop.org/series/32739/ fixes it..
Quick note, we're still hitting this issue on BYT with IGT-Version: 1.20-g7aac0e8 (x86_64) (Linux: 4.14.0-rc7-drm-intel-qa-ww44-commit-ec9f758+ x86_64).
https://intel-gfx-ci.01.org/tree/drm-tip/igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b.html Seems fixed in CI_DRM_3320, most likely commit: commit 1a1f12872edcd5e425b668a35fb23548cfa918ef Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Tue Nov 7 14:03:38 2017 +0000 drm/i915: Prevent unbounded wm results in g4x_compute_wm()
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.