Possible killer kernel regression: Intel HD Graphics 3000/Core i7

Funkliford · 2017-08-28 20:45:31

So I realize this probably isn't the right place to post it, but I don't really have the time or will to take up torch. Long story short, laptop has been plagued with overheating problems. So last night I restart and I notice the system is ungodly slow. The weird thing is CPU, memory, and disk usage were fine. Then I noticed my CPU temp just under 90c just after booting -- which even for me is pretty nuts -- so I figure it's throttling and I manage to get the fan back online but the temps are still unusally high and the performance is still shit even with throttling disabled. No clue what's happening. Eventually I manage to dmesg while logged into X (I'd be working on a virtual console because of the performance issues) and I see this being spammed.

[ 33.408367] pipe A vblank wait timed out
[ 33.408420] ------------[ cut here ]------------
[ 33.408447] WARNING: CPU: 1 PID: 49 at drivers/gpu/drm/i915/intel_display.c:12636 intel_atomic_commit_tail+0x1002/0x1010 [i915]
[ 33.408448] Modules linked in: joydev mousedev arc4 iwldvm mac80211 iwlwifi iTCO_wdt iTCO_vendor_support uvcvideo snd_hda_codec_hdmi videobuf2_vmalloc videobuf2_memops cfg80211 videobuf2_v4l2 videobuf2_core snd_hda_codec_conexant mei_wdt videodev snd_hda_codec_generic media btusb btrtl btbcm btintel bluetooth ecdh_generic intel_rapl x86_pkg_temp_thermal coretemp kvm_intel kvm irqbypass snd_hda_intel intel_cstate intel_rapl_perf snd_hda_codec snd_hda_core snd_hwdep psmouse input_leds i2c_i801 snd_pcm thinkpad_acpi pcspkr snd_timer nvram shpchp lpc_ich wmi e1000e snd soundcore mei_me rfkill ptp mei battery ac pps_core tpm_tis evdev tpm_tis_core tpm thermal mac_hid sch_fq_codel vboxnetflt(O) vboxnetadp(O) pci_stub vboxpci(O) vboxdrv(O) sg ip_tables x_tables ext4 crc16 jbd2 fscrypto mbcache algif_skcipher
[ 33.408485] af_alg dm_crypt dm_mod dax sd_mod crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel pcbc serio_raw atkbd libps2 aesni_intel aes_x86_64 crypto_simd glue_helper xhci_pci cryptd xhci_hcd ahci libahci libata sdhci_pci sdhci ehci_pci led_class ehci_hcd scsi_mod mmc_core usbcore usb_common i8042 serio i915 video button intel_gtt i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm
[ 33.408507] CPU: 1 PID: 49 Comm: kworker/u16:1 Tainted: G O 4.12.8-2-ARCH #1
[ 33.408508] Hardware name: LENOVO 4236D84/4236D84, BIOS 83ET78WW (1.48 ) 01/21/2016
[ 33.408522] Workqueue: events_unbound intel_atomic_commit_work [i915]
[ 33.408524] task: ffff8c54947ec740 task.stack: ffffb3ba40e44000
[ 33.408537] RIP: 0010:intel_atomic_commit_tail+0x1002/0x1010 [i915]
[ 33.408538] RSP: 0018:ffffb3ba40e47d88 EFLAGS: 00010282
[ 33.408539] RAX: 000000000000001c RBX: 0000000000000000 RCX: 0000000000000000
[ 33.408540] RDX: 0000000000000000 RSI: ffff8c549e24dcc8 RDI: ffff8c549e24dcc8
[ 33.408541] RBP: ffffb3ba40e47e40 R08: 0000000000000345 R09: 0000000000000004
[ 33.408541] R10: ffffb3ba40e47d88 R11: 0000000000000001 R12: 000000000000074a
[ 33.408542] R13: ffff8c5490680000 R14: ffff8c5490976000 R15: 0000000000000001
[ 33.408543] FS: 0000000000000000(0000) GS:ffff8c549e240000(0000) knlGS:0000000000000000
[ 33.408544] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 33.408545] CR2: 00007ff5017203d8 CR3: 0000000202019000 CR4: 00000000000406e0
[ 33.408546] Call Trace:
[ 33.408551] ? wake_bit_function+0x60/0x60
[ 33.408564] intel_atomic_commit_work+0x12/0x20 [i915]
[ 33.408566] process_one_work+0x1de/0x430
[ 33.408567] worker_thread+0x47/0x3f0
[ 33.408570] kthread+0x125/0x140
[ 33.408571] ? process_one_work+0x430/0x430
[ 33.408572] ? kthread_create_on_node+0x70/0x70
[ 33.408576] ret_from_fork+0x25/0x30
[ 33.408577] Code: ff ff ff 48 83 c7 08 e8 7d f9 b6 e3 4c 8b 85 70 ff ff ff 4d 85 c0 0f 85 f3 fa ff ff 8d 73 41 48 c7 c7 70 30 5f c0 e8 21 54 c3 e3 <0f> ff e9 dd fa ff ff 0f 1f 80 00 00 00 00 66 66 66 66 90 55 48
[ 33.408603] ---[ end trace b884171fac83df95 ]---

Dpwngrading to linux-lts seems to have fixed the issue entirely. Temperatures are back to normal,. and performance has returned.

Last edited by Funkliford (2017-08-28 20:48:39)

seth · 2017-08-29 06:59:45

Looks like the i915 module performes a busy wait for a (never arriving?) vblank signal?

What kinds of outputs (monitors) are attached?
Any weird module paramerts in use?
Does this also happen if you only boot into the multi-user.target (no graphical UI)?

Standard question: intel-ucode installed and enabled?

Arch Linux

#1 2017-08-28 20:45:31

Possible killer kernel regression: Intel HD Graphics 3000/Core i7

#2 2017-08-29 06:59:45

Re: Possible killer kernel regression: Intel HD Graphics 3000/Core i7

Board footer