kernel upgrade broke juicefs for me on Fedora 44

[avtechmjc]

Hi there. Hosted juicefs has been working solidly for me for years, but a recent kernel update in Fedora 44 (and 43 as well) has broken it.

This works: kernel-7.0.13-200.fc44.x86_64
This does not: kernel-7.1.3-200.fc44.x86_64

If I run:
juicefs mount avtech-juice-vol01 /fileserver1 --foreground --verbose

I get debug output like this (you can see where I pressed Ctrl+C):

... snipped ...
2026/07/08 16:58:15.696796 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.707087 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.717383 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.728013 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.738246 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.748598 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.759492 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.769737 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.780070 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
2026/07/08 16:58:15.790296 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]

2026/07/08 16:58:15.800553 juicefs[38655] : stat /fileserver1: %!s() [checkMountpoint@mount_unix.go:466]
^C
2026/07/08 16:58:47.760541 juicefs[38718] : received signal interrupt [installHandler@mount_unix.go:1175]
2026/07/08 16:58:47.760574 juicefs[38718] : read from meta 54.156.14.70:9327: read tcp 192.168.1.17:60432->54.156.14.70:9327: i/o timeout [read@client.go:483]
2026/07/08 16:58:47.760609 juicefs[38718] : read header from 54.156.14.70:9327: read tcp 192.168.1.17:60432->54.156.14.70:9327: i/o timeout [receiveThread@client.go:1503]
2026/07/08 16:58:47.760618 juicefs[38718] : disconnect to 54.156.14.70:9327 [close@client.go:410]
2026/07/08 16:58:47.760630 juicefs[38718] : clear moved inodes and splitted dir for zone 0 [newCluster@cluster.go:213]
2026/07/08 16:58:47.760625 juicefs[38718] : heartbeat is delayed 41.04403751s [heatbeat@main.go:68]
2026/07/08 16:58:47.760594 juicefs[38718] : load config from /fileserver1: /fileserver1 is not inside JuiceFS [installHandler@mount_unix.go:1206]

Anyone have similar issues?

Downgrading the kernel restores normal operation.

[Xuyuchao-juice]

OK, I have also reproduced this issue. It seems to be related to some changes in the kernel. I am still conducting further analysis.

[Xuyuchao-juice]

This is a bug that resulted from the combined effect of JuiceFS and the Linux kernel.
The commit for the introduction of the kernel side is （torvalds/linux@dfb84c33079497bf27058b15780e1c7bba4c371b）
fixed by (torvalds/linux@9f6f44aa5a58)

If it is not convenient for you to modify the kernel, you can also apply this patch to solve this problem.