On Sun, 8 Jul 2007, Linus Torvalds wrote: > > This would seem to be the minimal change, and I think it's right.
Side note: I considered just changing do_utimes() to use ATIME_SET/MTIME_SET instead, which would simplify the logic a lot.. But it turns out that the semantics for A/MTIME_SET is that only the owner of the file can do that. So the "!times" case is simply _fundamentally_ different from a permissions check standpoint, and there we have to check for "writable" rather than anything else (although ownership will override it). So rather than simplify the thing, I had to add code. Oh, well. I do think that it would be even nicer to just have a function that fills in the "struct nameidata" from the dfd. I think we should be able to: the "struct file" really does have the "f_path" thing with both dentry and mnt information, and that would clean up that whole ugly "filepointer-vs-nd" thing a lot. That would be an "Al cleanup", though. What do you think, Al? Linus - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/