[lkml]   [2001]   [Dec]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [CFT] BeFS filesystem 0.6
Alexander Viro wrote:

> Umm... Obvious comments:
> a) typedef struct super_block vfs_sb; Please don't.
> b) in inode.c:
> inode->i_mode = (umode_t) raw_inode->mode;
> is wrong. It's guaranteed bug either on big- or little-endian boxen.
> Same for mtime, uid and gid. befs_count_blocks() also needs cleanup.
> c) befs_read_block()... Erm. Why bother with extra layer of abstraction?
> d) befs_readdir(). You _are_ guaranteed that inode is non-NULL, you put
> pointer to befs_dir_operations only is S_ISDIR is true and S_ISDIR is
> mutually exclusive with S_ISLNK.
> e) are there sparse files on BeFS? If yes - you want to make BH_Mapped
> conditional in get_block() (set it if block is present, don't if it's a
> hole in file).
> f) befs_arch(). You probably want to make that an option...
> g) endianness problems in read_super()...
> h) TODO: use line breaks ;-)

Hi Al. Thanks for taking the time to respond.

A & B) I originaly had plans to make the driver work under a variety of
different VFSs (maybe freebsd and atheos) and to have user-space tools
(like the old mtools for DOS) for people on other OSs. That was the
motivation behind those extra abstractions, but I've actually kinda lost
interest in the idea. I'll just take 'em out if it really offends
people's asthetic sense. Perhaps kernel drivers just weren't meant to be

B) The PPC and x86 versions of BeOS each write their filesystem metadata
in native byte-order. So this would only be a problem with disks that
were formatted on a big-endian system and read on a little-endian (or
the other way around, of course). I've been meaning to handle that case

D) Ok. I killed that test.

E) To my knowledge, there are not sparse files on BeFS.

F) Ideally, I would detect both the endianness of the filesystem and the
host, and refuse to mount if they differ (or byte-swap the metadata).
Can I rely on #ifdef __LITTLE_ENDIAN to detect the endianness I am
running on?

G) Same thing as B, I think.

H) Yeah. I despise editors that won't line-wrap for you. But this is
unix, so I guess I gotta accomodate them. Is makeing sure they don't
overflow 80 columns enough? Or should that be 78?

> befs_count_blocks() also needs cleanup.

How so, please?

Will Dyson

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:14    [W:0.036 / U:6.392 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site