tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: inode open

On Jul 18, 2008, at 10:17 AM, Christos Zoulas wrote:

On Jul 18,  8:16am, (Jason Thorpe) wrote:
-- Subject: Re: inode open

| Didn't I cover that in my more detailed message?

I did not see it. You only said in the message I replied to that are
handled specially. I am curious how do you know which path to choose
to reach it?

I actually covered that in my email:

The semantics are a little weird in the presence of hard links, but those are fairly rare in Mac OS X (with the exception of Time Machine backup stores). File systems that support this typically return the last name-parent tuple used to look up a file when asked for this information. If the file has not been looked up before, they typically report the canonical name-parent tuple; in HFS+, hard links have a different CNID that the VFS is aware of, even though the file's CNID is the one reported back to userland in stat::st_ino when you stat a hard link. With Xsan, which has UFS hard link semantics, the "first" name-parent tuple (the path used to create the file, usually) is returned.


-- thorpej

Home | Main Index | Thread Index | Old Index