Bug 204537 - Complete Xorg freeze due to intermittent unstable tsc on Dell Precision 5540 (Coffeelake): Broken BIOS?
Summary: Complete Xorg freeze due to intermittent unstable tsc on Dell Precision 5540 ...
Status: NEW
Alias: None
Product: Timers
Classification: Unclassified
Component: Interval Timers (show other bugs)
Hardware: Intel Linux
: P1 high
Assignee: timers_interval-timers
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-08-09 05:45 UTC by Arcadiy Ivanov
Modified: 2019-11-14 08:13 UTC (History)
2 users (show)

See Also:
Kernel Version: 5.1.20
Subsystem:
Regression: No
Bisected commit-id:


Attachments

Description Arcadiy Ivanov 2019-08-09 05:45:45 UTC
The symptoms are intermittent (Fedora 30 KDE latest updates):

1. X freezes. Mouse cursor still moves, tty still works but X will not respond to any clicks/key combinations.
2. Switching to tty and issuing a reboot command, receive the following set of errors.
3. I suspect X tries to play a sound, ends up stuck in ALSA that is itself stuck due to unstable TSC, but that's an uneducated guess.

=========
Manufacturer: Dell Inc.
Product Name: Precision 5540
Product Name: 0V030K
Version: A00
BIOS Information
        Vendor: Dell Inc.
        Version: 1.2.3
        Release Date: 07/23/2019

Kernel: 5.1.20-300.fc30.x86_64
===


Aug 09 01:22:18 hostname kernel: alsactl[1527]: segfault at 28 ip 000055e8e10bb6f0 sp 00007ffc12a3a158 error 4 in alsactl[55e8e10ad000+11000]
Aug 09 01:22:18 hostname kernel: Code: 4c 89 e6 e8 02 2c ff ff 89 ef e8 4b 2a ff ff ba 0a 00 00 00 31 f6 4c 89 e7 c6 44 24 17 00 e8 67 31 ff ff eb b2 e8 00 34 ff ff <8b> 06 41 54 55 48 89 fd 53 85 c0 7e 22 49 89 f4 31 db 66 0f 1f>
Aug 09 01:22:24 hostname kernel: clocksource: timekeeping watchdog on CPU12: Marking clocksource 'tsc' as unstable because the skew is too large:
Aug 09 01:22:24 hostname kernel: clocksource:                       'hpet' wd_now: fc44fc84 wd_last: fba84649 mask: ffffffff
Aug 09 01:22:24 hostname kernel: clocksource:                       'tsc' cs_now: 3a41f2261e40 cs_last: 3a41ab28c198 mask: ffffffffffffffff
Aug 09 01:22:24 hostname kernel: tsc: Marking TSC unstable due to clocksource watchdog
Aug 09 01:22:24 hostname kernel: TSC found unstable after boot, most likely due to broken BIOS. Use 'tsc=unstable'.
Aug 09 01:22:24 hostname kernel: sched_clock: Marking unstable (26663054300069, -369815753)<-(26662712684796, -28200423)
Aug 09 01:22:24 hostname kernel: clocksource: Switched to clocksource hpet
Comment 1 Kai-Heng Feng 2019-08-09 07:43:13 UTC
Does the issue get resolved with kernel parameter "tsc=reliable"?
Comment 2 Arcadiy Ivanov 2019-08-09 19:19:56 UTC
Do you mean 'tsc=unstable'?
Comment 3 Arcadiy Ivanov 2019-08-13 03:29:26 UTC
[    0.423575] Kernel command line: BOOT_IMAGE=(hd0,gpt2)/vmlinuz-5.1.20-300.fc30.x86_64 root=/dev/mapper/fedora-root ro resume=/dev/mapper/fedora-swap rd.lvm.lv=fedora/root rd.lvm.lv=fedora/swap rootflags=discard rd.driver.blacklist=nouveau i915.enable_guc=3 l1tf=flush acpi_osi=Linux-Dell-Video tsc=unstable
[    0.423676] tsc: Marking TSC unstable due to boot parameter

Will see if it works.
Comment 4 Arcadiy Ivanov 2019-08-16 20:13:47 UTC
`tsc=unstable` appears to be resolving this problem.

Will try with `tsc=reliable` now.
Comment 5 Kai-Heng Feng 2019-08-29 09:14:40 UTC
Can you please test this patch:
https://lkml.org/lkml/2019/8/29/233
Comment 6 Arcadiy Ivanov 2019-11-14 08:13:32 UTC
I've been running with `tsc=reliable` for several months and the problem is gone. 
I guess it is bug #203183 after all.

Note You need to log in before you can comment on or make changes to this bug.