On 2007-Jun-01 11:52:07 -0400, Garance A Drosehn <gad_at_freebsd.org> wrote: >> - Peter Wemm has been talking about moving us to 64-bit inode numbers > > I suspect this should wait. It would probably be better to group > together all the other changes to filesystem stat-ish data that we've > also been talking about for years, and do them all in the same major > release. Agreed. Every time we change struct stat, we need to create legacy syscalls or libc code so the fewer changes the better. > I don't know at what level you mean to move to 64-bit inode > numbers, but if (for instance) you meant a 64-bit value for st_ino, Since an inode number needs to fit into an ino_t, I would expect that ino_t and hence st_ino would both become 64-bit. > then I'd also like to see a 64-bit value for st_dev at the same time. I can see the reason for having more than 2^32 inodes in a filesystem. It's not as obvious why you would need a 64-bit dev_t. You're never going to have more than 2^32 devices attached to a system and I would suggest that it's unlikely that you would have more than 255 active drivers on one system. What other struct stat changes are up for discussion? -- Peter Jeremy
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:11 UTC