Coda File System

Re: a bug?

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Tue, 5 May 1998 08:37:10 -0400 (EDT)
Very well spotted indeed!!

Thanks!

- Peter -

On Tue, 5 May 1998, Bruce Janson wrote:

> Hi Peter,
>     While writing our venus-like cache-handler the following (presumably
> minor) bug became apparent:
> 
> In
> 	Linux kernel 2.1.98,
> 	file linux/fs/coda/coda_linux.c,
> 	function coda_flags_to_cflags(),
> 	lines 96 to 97:
> 
> 	..
> 	if ( flags & (O_RDONLY | O_RDWR) )
> 		coda_flags |= C_O_READ;
> 	..
> 
> But in linux/include/asm/fcntl.h, line 7:
> 
> 	..
> 	#define O_RDONLY            00
> 	..
> 
> Thus C_O_READ is often not set in coda_flags when I suspect
> that it should be.
> Presumably venus treats a zero flags value (such as in open
> and close calls) as if just the C_O_READ bit were set.
> Perhaps this should be made explicit?
> 
> Regards,
> bruce.
> 
Received on 1998-05-05 09:41:10