coredumps with zfs send/receive with 2.1.7

[mabod]

I am experiencing coredumps when doing send/receive with the latest 2.1.7 release. This is on Endeavouros with zfs-dksm/zfs-utils PKGBUILD files modified to pull version 2.1.7.

I was initially thinking that this here was also my issue: #14252
The solution to this issue is reverting patch c8d2ab0
But that does not solve my problem.

I am getting coredumps when doing send/receive between two pools resp. datasets. I am using syncoid to do that:

    syncoid --sendoptions="L" --mbuffer-size=512M --no-sync-snap zHome/home zstore/data/BACKUP/rakete_home/home
    NEWEST SNAPSHOT: 2022-12-03--23:23
    Sending incremental zHome/home@2022-11-29--08:05 ... 2022-12-03--23:23 (~ 34 KB):
0,00 B 0:00:00 [0,00 B/s] [>                                                                   ]  0%            
    cannot receive: failed to read from stream
CRITICAL ERROR:  zfs send -L  -I 'zHome/home'@'2022-11-29--08:05' 'zHome/home'@'2022-12-03--23:23' | mbuffer  -q -s 128k -m 512M 2>/dev/null | pv -p -t -e -r -b -s 35264 |  zfs receive  -s -F 'zstore/data/BACKUP/rakete_home/home' 2>&1 failed: 256 at /usr/bin/syncoid line 817.

# zfs get recordsize,compression,encryption zHome/home zstore/data/BACKUP/rakete_home/home
NAME                                 PROPERTY     VALUE           SOURCE
zHome/home                           recordsize   128K            inherited from zHome
zHome/home                           compression  lz4             inherited from zHome
zHome/home                           encryption   off             default
zstore/data/BACKUP/rakete_home/home  recordsize   1M              inherited from zstore
zstore/data/BACKUP/rakete_home/home  compression  zstd            inherited from zstore/data/BACKUP
zstore/data/BACKUP/rakete_home/home  encryption   aes-256-gcm     -

I am not able to do a git bisect. Unloading the zfs module does not work. Although I exported all pools it keeps telling me that the module is busy.

I am also not able to get a full coredump: "Resource limits disable core dumping for process 9485 (zfs)." I do not know why that is. gnome-shell dumped core yesterday and the day before and I do have a coredump file for that.

Dez 12 11:25:14 rakete kernel: traps: zfs[9485] general protection fault ip:7f50bc70dbd0 sp:7ffce1cd88c0 error:0 in libzfs.so.4.1.0[7f50bc6d2000+43000]
Dez 12 11:25:15 rakete systemd[1]: Created slice Slice /system/systemd-coredump.
Dez 12 11:25:15 rakete systemd[1]: Started Process Core Dump (PID 9489/UID 0).
Dez 12 11:25:15 rakete systemd-coredump[9490]: Resource limits disable core dumping for process 9485 (zfs).
Dez 12 11:25:15 rakete systemd-coredump[9490]: [🡕] Process 9485 (zfs) of user 0 dumped core.
Dez 12 11:25:15 rakete systemd[1]: systemd-coredump@0-9489-0.service: Deactivated successfully.

Any help is appreciated

[mabod]

I guess I made a mistake with reverting the patch c8d2ab0

In the meantime I did more trials and now I have it working. For the last dozens of send/receive I did not have any issue. No locks, no core dumps.

I am closing this issue.