2005-01-06 21:43:49

by Adrian Bunk

Subject: 2.6: fs/coda/dir.c: coda_hasmknod seems to be buggy

Function coda_mknod in fs/coda/dir.c (I checked 2.6.10-mm2) stats with:

<-- snip -->

...
static int coda_mknod(struct inode *dir, struct dentry *de, int mode, dev_t rdev)
{
int error=0;
const char *name=de->d_name.name;
int length=de->d_name.len;
struct inode *inode;
struct CodaFid newfid;
struct coda_vattr attrs;

if ( coda_hasmknod == 0 )
return -EIO;

<more code>
...

<-- snip -->


The problem is that not a single place in the kernel assigns any value
to coda_hasmknod.


cu
Adrian

--

"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed

2005-01-07 00:14:08

by Jan Harkes

Subject: Re: 2.6: fs/coda/dir.c: coda_hasmknod seems to be buggy

On Thu, Jan 06, 2005 at 10:40:53PM +0100, Adrian Bunk wrote:
> Function coda_mknod in fs/coda/dir.c (I checked 2.6.10-mm2) stats with:

Thanks,

Device nodes do not make much sense in a distributed filesystem. I never
removed this code just in case anyone (i.e. not Coda) was using it, but
you just proved that there can be no other user of this code.

I'll remove the has_mknod related parts.

Jan