Chris Lamb: >> disorderfs: rewinddir doesn't work > > Are we sure it's directly rewinddir? ISTR FUSE does funny things with > inode numbers and — at a very quick glance — might mean that things > like (d->d_ino == st1.st_ino) are misleading >
What sorts of funny things are you referring to? My test program is adapted from posix/tst-dir.c in glibc. In both the code, there are many other calls to stat() comparing inode numbers and they all work on a disorderfs (until we get to rewinddir). Going by a quick scan of random internet documents (TM) it seems like FUSE does let you manage inode numbers "properly", so disorderfs could try that. Or do you mean that "disorderfs doesn't deal with inode numbers at all so I don't see why this is happening"? I didn't study the code in detail yet. X -- GPG: ed25519/56034877E1F87C35 GPG: rsa4096/1318EFAC5FBBDBCE https://github.com/infinity0/pubkeys.git
