Announcement

Collapse
No announcement yet.

Casefolding For Bcachefs File-System Posted

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by Eberhardt View Post

    The patch is actually quite small and uses the unicode casefolding functions from the linux/unicode.h header, so technically a lot of code is already shared.
    what he is asking for is something like file/casecompacting.c

    and then one common function for all filesystems

    function (casecompacting)
    Blah blah blah

    function (casef2fs)
    casecompacting

    function (casebcachefs)
    casecompacting

    function (caseext4)
    casecompacting

    and then you can remove the codelines in the individual filesystem and have them call the helper function from the vfs level so you onlly need to code the basic usecase once instead of each individual and can save lines of code and have a basic place, wich then later can be ussed to implement it in future filesystems or existing if they want to contribute, and if 3 pepole from 3 diffrent fs´s work for the same cause they may come up with a better solution for all instead each individual implemention.

    Comment


    • #12
      Originally posted by ireri View Post
      That is not true, at least Debian has a list of btrfs problems compiled, with only a portion related to raid. There are other not universal problems that affect certain use cases, like (but not limited to) deduplication and metadata loss with some <v5.x kernels.
      do you have a link to this list?
      also, kernel older than 5.x are VERY old: a big amount of work has been done in the btrfs code since then.

      Comment


      • #13
        Originally posted by Britoid View Post

        There's not any reliability problems with btrfs outside of raid 5/6, iirc the steam desk uses btrfs for the root fs but ext4 for the game data partition.

        Some games however, don't run on XFS/btrfs and want an ext4 fs. It's also not the fastest FS right now
        Oh yes, the good old "I don't have problems, therefore problems don't exist" argument.

        Comment


        • #14
          Make case insensitivity a mount option to apply to the whole partition, until then it's useless.

          In any case, can we stop for a moment and appreciate this dude's website? Lol

          Last edited by anarki2; 13 August 2023, 03:15 PM.

          Comment


          • #15
            Originally posted by erniv2 View Post

            what he is asking for is something like file/casecompacting.c

            and then one common function for all filesystems

            function (casecompacting)
            Blah blah blah

            function (casef2fs)
            casecompacting

            function (casebcachefs)
            casecompacting

            function (caseext4)
            casecompacting

            and then you can remove the codelines in the individual filesystem and have them call the helper function from the vfs level so you onlly need to code the basic usecase once instead of each individual and can save lines of code and have a basic place, wich then later can be ussed to implement it in future filesystems or existing if they want to contribute, and if 3 pepole from 3 diffrent fs´s work for the same cause they may come up with a better solution for all instead each individual implemention.
            It's not possible to do transparently at the vfs level
            The hashing and comparison stuff happens at the fs level.

            It would be possible if you dropped the original case entirely and made everything lowercase -- but that's not how case-insensitive filesystems like NTFS, etc work.

            You can have FROG in a casefolded filesystem, but query it with FROG frOG, frog.
            But the file is still "FROG", it's not "frog".

            The underlying filesystem needs to be aware of casefolding for so many other reasons too.

            Comment


            • #16
              Originally posted by anarki2 View Post
              Make case insensitivity a mount option to apply to the whole partition, until then it's useless.

              In any case, can we stop for a moment and appreciate this dude's website? Lol

              https://froggi.es/
              It's directory based. You could literally apply `chattr +F /` if you wanted. There is nothing stopping you.

              Comment


              • #17
                Originally posted by Britoid View Post
                Some games however, don't run on XFS/btrfs and want an ext4 fs.
                Such as?

                There is the inode32 thing, but bcachefs has an inode32 compat mode.

                Comment


                • #18
                  Originally posted by brucethemoose View Post
                  Lz4 compression and easy system rollbacks would be very useful, and f2fs doesn't have either.
                  Not only does F2FS have lz4 compression, it has highly granular compression controls for tuning compression types based on file extension. Edit to add: If you want snapshot roll backs on flash based storage, you're probably better off with nilfs2... at least once they get a working fsck.

                  As for BtrFS... aside from legit data integrity questions with the filesystem there's several, including btrfs_check being incomplete and unreliable according to the Arch Wiki along with it's quite atrocious performance compared to alternatives and arguably an inappropriate choice for single device storage on portable systems, I've seen SteamPlay/Linux games in Steam that crash on any FS other than ext4. No idea why, I never followed up on what was going on - it shouldn't matter what the underlying FS is, but it does.

                  Comment


                  • #19
                    Unless I'm missing something the reason some games may crash on btrfs is precisely because btrfs doesn't do case folding, if the game is sloppy with path resolution it can't find files then shits itself. More microsoft baking a ticking compatibility timebomb into ntfs than anything btrfs has done wrong.

                    Comment


                    • #20
                      Originally posted by MadWatch View Post
                      Why is that?
                      case sensitivity
                      32bit inodes

                      Comment

                      Working...
                      X