You have been subscribed to a public bug:
Hi all,
This one is similar to #1805816 and #1678184 (one was fixed, other closed).
Symptoms:
During regular use, system starts failing after 10m - 1hr after start.
Icons start disappearing, writing to disk fails.
In-memory operations still work for a while (switching windows, streaming
video calls, typing).
After some time the entire system crashes, with a Black Screen Of Death
constantly looping:
---------------------------
EXT4-fs error (device nvme0n1p5) ext4_find_entry:1455: inode #4594258: comm
gmain: reading directory lblock 0
[... same repeats for 8 times on average]
systemd-journald[439]: Failed to write entry (9 items, 270 bytes),
ignoring: Read-only file system
[... repeats for 10 times on average]
---------------------------
Probable causes:
Updated both kernel and BIOS 2 days ago. Unable to determine, which one
caused the change.
Don't know how to determine which kernel and bios I was running before the
update.
Looks like APST issue, based on info from web and previous bug reports.
Verification:
Created a rudimentary bash script, writing to a file in a loop, incrementing
timeout between two consecutive writes each time.
Ran script using:
- different nvme_core.default_ps_max_latency_us settings
in GRUB_CMDLINE_LINUX_DEFAULT="quiet splash
nvme_core.default_ps_max_latency_us=[0|200|5500]"
- regular boot mode
- logged in to account
- on battery power
With default_ps_max_latency_us NOT SET:
writing FAILS between 57 and 70 seconds timeout between writes
With default_ps_max_latency_us=5500:
no write failure during 1hrs run
With default_ps_max_latency_us=200:
no write failure during 30m run
With latency 0:
no write failure during 10m run
This suggests an APST issue.
Machine:
Lenovo Thinkpad T570
Disk:
SAMSUNG MZVLB512HAJQ-000L7
512 GB (512110190592 bytes)
Firmware: 3L2QEXA7
Serial#: S3TNNE0K119126
System:
OS 1: Ubuntu 18.04.5 LTS
Kernel: 4.15.0-122-generic #124-Ubuntu SMP Thu Oct 15 13:03:05 UTC 2020
x86_64
OS 2: Windows 7 (on a separate partition on same disk, dualbooted with grub).
Actions taken:
Successfully checked the partitions for errors by running "Check partition"
and "Repair partition" in Disks utility in Ubuntu, running from a bootable USB.
Starting in "recovery mode" yields an error (among other suspicious behavior):
--------------------
sd 0:0:0:0: Attached scsi generic sg0 type 0
sd 0:0:0:0: [sda] Attached SCSI removable Disk
input: TPPS/2 IBM TrackPoint as
/devices/platform/18042/serio1/serio2/input/input ....
nvme nvme0: controller is down; will reset: CSTS=0xffffffff,
PCI_STATUS=0x10
nvme 0000:40:00.0: enabling device (0000 -> 0002)
nvme nvme0: Removing after probe failure status: -19
nvme0n1: detected capacity change from 512110190592 to 0
print_req_error: I/O error, dev nvme0n1, sector 1000215040
nvme nvme0: failed to set APST feature (-19)
Waiting for suspend/resume device ... Begin: Running /scripts/local-block
No devices listed in conf file were found.
No devices listed in conf file were found.
[repeats]
--------------------
Without nvme_core.default_ps_max_latency_us set: writing fails between cca.
57 - 70, on battery.
With nvme_core.default_ps_max_latency_us=5500: FIXES THE PROBLEM.
With nvme_core.default_ps_max_latency_us=200: FIXES THE PROBLEM.
With nvme_core.default_ps_max_latency_us=0: FIXES THE PROBLEM.
Previous behavior on same machine:
Same OS, with a previous Kernel has been running perfectly fine for the last
year, "on high revs" (it's a development machine).
Often running on battery alone.
sleep and wakeup without issues.
It is running Windows 7 (dual booted) without issues.
Misc info:
To test different settings, did 'sudo nano /etc/default/grub' with each of
these settings:
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash"
# GRUB_CMDLINE_LINUX_DEFAULT="quiet splash
nvme_core.default_ps_max_latency_us=5500"
# GRUB_CMDLINE_LINUX_DEFAULT="quiet splash
nvme_core.default_ps_max_latency_us=200"
# GRUB_CMDLINE_LINUX_DEFAULT="quiet splash
nvme_core.default_ps_max_latency_us=0"
followed by 'sudo update-grub' and reboot.
** Affects: linux (Ubuntu)
Importance: Undecided
Status: New
** Tags: apst nvme samsung
--
Failure to write to NVMe disk soon after boot (APST-related)
https://bugs.launchpad.net/bugs/1903388
You received this bug notification because you are a member of Kernel Packages,
which is subscribed to linux in Ubuntu.
--
Mailing list: https://launchpad.net/~kernel-packages
Post to : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help : https://help.launchpad.net/ListHelp
