News for hackers. Stuff that matters.
Re:Tmpfs inode corruption: introducing inode64 by esjeon on Saturday September 04, 2021 @03:12AM (#28412820)
Please read the article. TMPFS was internally using 32bit unsigned int for counting inode number, which got overflow-ed. Kernel already has 64bit wide ino_t, the dev migrated to it.
Re:Tmpfs inode corruption: introducing inode64 by puppet-master on Saturday September 04, 2021 @05:09AM (#28413407)
The question was clear enough: per the article tmpfs is able to track two distinct files with the same inode number, implying it does not use the inode internally to differentiate between files
Re:Tmpfs inode corruption: introducing inode64 by Tobu on Saturday September 04, 2021 @06:46AM (#28413895)
I'm not sure where the confusion is coming from.

If userspace wants to distinguish files, it normally does so with a (ino_t, dev_t) pair. With some nuance wrt inode generations if you aren't holding on to files yet want to guard against ino reuse, and some funkyness with overlayfs.

Most filesystems don't expose a way to access files by ino only, so the internal implementation generally doesn't key anything by them or rely on them itself; they're just a kind of of uniqueness cookie. But keying by them somewhere in the kernel implementation is feasible, if you scope it properly.

Re:Tmpfs inode corruption: introducing inode64 by adrianmonk on Saturday September 04, 2021 @02:08PM (#28417562)
It may be that, in modern times, filesystems don't actually key things with i-node number. But wasn't that the original purpose? Otherwise, why the design where directories are basically a mapping of names to i-node numbers?

Also, the article does say this:

> On non-volatile filesystems, inode numbers are typically finite and often have some kind of implementation-defined semantic meaning (eg. implying where the inode is disk).

If you've got other means to know where the i-node is on disk, then why would a filesystem bother encoding that into the i-node numbering scheme?

Maybe the article is wrong here. I don't know. But if the question is where the confusion is coming from, the article definitely seems to be painting a picture where normal filesystems look up things based on i-node number and tmpfs is relatively unique in not doing so.

Re:Tmpfs inode corruption: introducing inode64 by formerly_proven on Saturday September 04, 2021 @02:22PM (#28417678)
Yes, the OG Unix file system simply had a fixed-size (at partition time, this is still the case for ext4 and XFS iirc) array of struct inode on the disk, and the inode number (iirc it was just idx or something like that in the code) is simply the index into that array. An OG Unix directory was simply a file with the directory flag set and the file was just an array of struct { char name[30]; int ino; } (~something like that). You actually used to be able to just open() a directory and directly read directory entries (dirents henceforth) from it, just like a file - exactly like a file, because it WAS a file.

ext and XFS still largely work like this, though directories are now hashtables and I think XFS supports multiple independent arrays for inodes (or just for extent allocation? I don't remember). NTFS also looks a lot like a Unix file system with some weird growths on it, and not a whole lot like FAT.

Re:Tmpfs inode corruption: introducing inode64 by yyyk on Saturday September 04, 2021 @02:29PM (#28417747)
There's also some funkyness with btrfs subvolumes:

https://news.ycombinator.com/item?id=28274054