Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

9P2000 is available on Linux with the CONFIG_9P_FS and related kernel compile options, which say things like:

"If you say Y here, you will get experimental support for Plan 9 resource sharing via the 9P2000 protocol. See http://v9fs.sf.net for more information."



It's also the native host-guest file sharing protocol used by KVM: https://www.linux-kvm.org/page/9p_virtio

It works in at least Ubuntu & Fedora out of the box, and is used by the virt-manager GUI to set up shares, so not too experimental to enable by default I guess.


That's been experimental for a long time - https://github.com/torvalds/linux/tree/master/fs/9p

I guess file systems don't go into staging/?


On Arch Linux, it's been enabled by default for several years, and I've been using it daily for that time.

On x86_64, it's plenty stable, but I see why it's still experimental: On i686 there's a bug where files are mysteriously missing from directory listings. I tried to debug it once; it's reading them from the protocol correctly, but they're being dropped in the VFS layer, which is a bit more complicated than the network protocol.


Maybe, once FUSE came along, 9pfuse made more sense than putting 9p in the kernel?

https://9fans.github.io/plan9port/man/man4/9pfuse.html


I was just wondering, as I began to read the article: which would perform better - in-kernel 9P, or FUSE?

Intuitively I get the idea that FUSE is going to be the better performer overall because of lack of network overhead.

But I wonder how much worse 9P would be.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: