FUSE Passthrough Support May Land For Linux 6.9 To Help Boost I/O Performance

[Joe2021]

As virtiofs is using (at least parts of) fuse, are there gains to expect, too?

[S.Pam]

As virtiofs is using (at least parts of) fuse, are there gains to expect, too?

That's what I am hoping for. I use virtiofs as root filesystem for lots of VMs, and what's lacking is especially random io performance.

[WestCoastSuccess]

Is this expected to have performance benefits for ocamlfuse (which mounts Google Drive folders to a Linux filesystem, and works quite well, but is sometimes quite slow)?

[geearf]

trapexit I was hopeful to read about mergerfs here and I'm glad to see your comments! Will this passthrough simplify the mount options needed for performance or not really? Thanks!

[gnarlin]

The new NTFS driver is in the kernel, this patch wouldn't help the NTFS fuse driver though because it's only for when the backing storage is handled by a native Linux fs driver.

What is the state of the new NTFS driver? Is is ever going to become the default NTFS driver in distros? Does it come with a propper fsck.ntfs utility? I don't know where to look for information about this.

[trapexit]

trapexit I was hopeful to read about mergerfs here and I'm glad to see your comments! Will this passthrough simplify the mount options needed for performance or not really? Thanks!

There are no "performance" options really. They are all features that may have impact on performance. passthrough is not compatible or makes no sense with certain options so if you want to call that "simplifying" then sure... but it is just different features for different options.

[trapexit]

Is this expected to have performance benefits for ocamlfuse (which mounts Google Drive folders to a Linux filesystem, and works quite well, but is sometimes quite slow)?

No. Passthrough is about *local* files being handled by the kernel directly. For union filesystems or overlay filesystems. There is no file for ocamlfuse to pass through.

[caligula]

There are downsides to using this however. In my own project (mergerfs) I can add resiliency by catching certain errors. As far as I know when using this passthrough there won't be that opportunity any longer. If a read fails on a remote filesystem or write results in ENOSPC there is no mechanism to push that info to the FUSE server and do something about it. I think many users will be fine with that feature tradeoff though.

BTW your project (mergerfs) rocks. Most useful fuse fs I know of.

[ehansin]

No. Passthrough is about *local* files being handled by the kernel directly. For union filesystems or overlay filesystems. There is no file for ocamlfuse to pass through.

I think I have knowingly (meaning not something that something else uses "behind the scenes") only used sshfs with FUSE. After reading all of this and comments, unless I really am mistaken, not something that would apply to the sshfs usage case, same as a lot of other examples. But still seems cool where you can get this kind of performance boost where it is applicable.

[trapexit]

I think I have knowingly (meaning not something that something else uses "behind the scenes") only used sshfs with FUSE. After reading all of this and comments, unless I really am mistaken, not something that would apply to the sshfs usage case, same as a lot of other examples. But still seems cool where you can get this kind of performance boost where it is applicable.

Correct, not applicable to sshfs. This is effectively passing an existing FD to the kernel and having it work with that afterwards without going through the fuse server. You can't just pass a network socket FD or something and have that work like a regular file FD. This only is meaningful for filesystem oriented FDs.