[Bug] OSError: [Errno 14] Bad address

[maxwowpow]

Creating an empty file produces the error OSError: [Errno 14] Bad address

mkdir /tmp/exfat
cd $_
fallocate -l 512M exfat.img
mkfs.exfat exfat.img
sudo mkdir -p /fatimagedir/
sudo mount -o umask=0022,gid=$(id -g),uid=$(id -u) /tmp/exfat/exfat.img /fatimagedir/
cd /fatimagedir
python3 -c 'open("foo", "wb", buffering=0).write(b"")'

See also: https://bbs.archlinux.org/viewtopic.php?id=294837
See also: zed-industries/zed#21595

ArchLinux
Ubuntu
Fedora 41

[relan]

I reproduced this issue with the in-kernel implementation of exFAT.

No such issue with fuse-exfat:

$ sudo losetup -f exfat.img
$ sudo fuse/mount.exfat-fuse -o loop,umask=0022,gid=$(id -g),uid=$(id -u) /dev/loop0 /fatimagedir/
FUSE exfat 1.4.0 (libfuse3)
$ cd /fatimagedir/
$ python3 -c 'open("foo", "wb", buffering=0).write(b"")'
$ ls
foo

[maxwowpow]

so the recommendation is to mount with fuse, is that correct?

[relan]

Yes, try fuse-exfat.

[maxwowpow]

it works for me as you pointed out. Lots of internet articles say though, that exfat is integrated into the kernel and the usage with fuse is "obsolete" and requires compiling from source.

Would that mean a bug in the kernel module needs to be raised? If yes would you know by chance what procedure is best for that like mailing list(s) of a file system responsible etc?

PS: I'm using fedora currently and didn't find any packages in the fedora repo.

[relan]

Lots of internet articles say though, that exfat is integrated into the kernel and the usage with fuse is "obsolete" and requires compiling from source.

Well, sometimes fuse-exfat works better than in-kernel implementation...

Would that mean a bug in the kernel module needs to be raised?

Seems so. I'd suggest checking https://fedoraproject.org/wiki/Bugzilla and https://bugzilla.kernel.org/

[sandeen]

https://git.kernel.org/pub/scm/linux/kernel/git/linkinjeon/exfat.git/commit/?h=dev solves the kernel driver issue.

[relan]

https://git.kernel.org/pub/scm/linux/kernel/git/linkinjeon/exfat.git/commit/?h=dev solves the kernel driver issue.

Great! Thanks for letting know!