Bug 14610 - Lockup in kernel 5.10 when copying large folder
Summary: Lockup in kernel 5.10 when copying large folder
Status: NEW
Alias: None
Product: CifsVFS
Classification: Unclassified
Component: kernel fs (show other bugs)
Version: 5.x
Hardware: All Linux
: P5 normal
Target Milestone: ---
Assignee: Steve French
QA Contact: cifs QA contact
Depends on:
Reported: 2021-01-08 17:29 UTC by Tom Mason
Modified: 2021-01-08 17:29 UTC (History)
0 users

See Also:

server side config file (9.19 KB, text/plain)
2021-01-08 17:29 UTC, Tom Mason
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Tom Mason 2021-01-08 17:29:26 UTC
Created attachment 16384 [details]
server side config file

I have been having an issue with CIFS mounts when running kernel 5.10 on the client (using the 5.10 build from https://wiki.ubuntu.com/Kernel/MainlineBuilds).

When doing a lot of IO in a file copy tool I'm writing, the mount would lock up, and calls to open() would hang, sometimes form multiple seconds, and in a few cases indefinitely (I left it for multiple minutes). Accessing the share from a separate process (eg, running ls in a separate shell) would normally work fine, but on a few occasions the share locked up completely and wouldn't re-mount until I rebooted. When the lockups happen I get the following in dmesg, sometimes just once, and sometimes repeating thousands of times:

Jan  2 00:13:50 pooka kernel: [ 9548.025373] CIFS: VFS: \\ No task to wake, unknown frame received! NumMids 1
   29 Jan  2 00:13:50 pooka kernel: [ 9548.025380] 00000000: 424d53fe 00000040 00000000 00000012  .SMB@...........
   30 Jan  2 00:13:50 pooka kernel: [ 9548.025382] 00000010: 00000001 00000000 ffffffff ffffffff  ................
   31 Jan  2 00:13:50 pooka kernel: [ 9548.025383] 00000020: 00000000 00000000 00000000 00000000  ................
   32 Jan  2 00:13:50 pooka kernel: [ 9548.025384] 00000030: 00000000 00000000 00000000 00000000  ................

The kernel messages will show up for me by running a copy of a large directory using standard cp, as well. The problem is not present when running kernel 5.8 (stock for my distro).

The machines are connected directly via ethernet, I am pretty certain that connection loss is not part of the problem. The client is running Ubuntu 20.10, with kernel 5.10, and the client is stock Ubuntu 20.04 LTS.

Here's the mount line from /etc/fstab on the client:
//                       /files_zfs cifs rw,relatime,cache=strict,user=guest,pass=,uid=0,file_mode=0755,dir_mode=0755,uid=1000,gid=1000 0 0

I have attached the server side config file.