On Wed, Jul 9, 2025 at 3:01 PM David Howells dhowells@redhat.com wrote:
If you keep an eye on /proc/fs/netfs/requests you should be able to see any tasks in there that get stuck. If one gets stuck, then:
After one got stuck, this is what I see in /proc/fs/netfs/requests:
REQUEST OR REF FL ERR OPS COVERAGE ======== == === == ==== === ========= 00000065 2C 2 80002020 0 0 @0000 0/0
Looking in /proc/fs/netfs/requests, you should be able to see the debug ID of the stuck request. If you can try grepping the trace log for that:
grep "R=<8-digit-hex-id>" /sys/kernel/debug/tracing/trace
kworker/u96:4-455 [008] ...1. 107.145222: netfs_sreq: R=00000065[1] WRIT PREP f=00 s=0 0/0 s=0 e=0 kworker/u96:4-455 [008] ...1. 107.145292: netfs_sreq: R=00000065[1] WRIT SUBMT f=100 s=0 0/29e1 s=0 e=0 kworker/u96:4-455 [008] ...1. 107.145311: netfs_sreq: R=00000065[1] WRIT CA-PR f=100 s=0 0/3000 s=0 e=0 kworker/u96:4-455 [008] ...1. 107.145457: netfs_sreq: R=00000065[1] WRIT CA-WR f=100 s=0 0/3000 s=0 e=0 kworker/8:1-437 [008] ...1. 107.149530: netfs_sreq: R=00000065[1] WRIT TERM f=100 s=0 3000/3000 s=2 e=0 kworker/8:1-437 [008] ...1. 107.149531: netfs_rreq: R=00000065 2C WAKE-Q f=80002020
I can reproduce this easily - it happens every time I log out of that machine when bash tries to write the bash_history file - the write() always gets stuck.
(The above was 6.15.5 plus all patches in this PR.)