Subj : CVS commit src/sbbs3/nope To : Deuce From : Digital Man Date : Sun Dec 26 2004 02:58 am Re: CVS commit src/sbbs3/nope By: Deuce to Digital Man on Sun Dec 26 2004 01:26 am > > Hmm... sure you can. If one thread has file open DENYNONE, I sure hope th > > another thread or process running on the same system will be denied open > > access. > > > > > As for being different than Linux, FreeBSD doesn't use the SMBFS that > > > uses. > > > > Oh. It's interesting that it exhibits some of the same flaws. Did they st > > from the same source at some point? > > re: 1: Nope. advisory locking only, *nix has no DENY* stuff at all... all > file/record locking prevents is anyone else from getting a lock... not from > opening, reading, writing, etc. The only way to get a lock-related error fr > open() is if you use some non-standard extensions ie: In FreeBSD, you can us > the O_SHLOCK and O_EXLOCK flags... which get a lock during the open(). Right. But I'm talking about sopen(), not open(). Our sopen() implementation attempts a file lock and it's this lock that returning the error with errno= EDEADLK. > re: 2: I don't believe so, I think it was something BSD/OS donated... but th > problems in resolving quirks would probobly drive people to the same "fixes" > and it's not impossible that the BSD people have glanced at how Linux did it > > The biggest problem with locking between Win32 and *nix is that neither OS i > capable of everything the other OS can do. Windows is missing support for t > braindead *nix record locking semantics, and *nix has no portable concept of > mandatory locking. There are other differences too... basically, neither OS > concept of locking of any kind is supported completely on the other. Well SMB/CIFS has standardized, cross-platform, file and record locking, which should (obviously) be supported by all smb/cifs clients and servers (e.g. smbfs), regardless of the OS on which they're hosted. digital man Snapple "Real Fact" #109: Smelling apples and/or bananas can help you loose weight. .