Re: unsigned short for nlink_t

Russell Coker (russell@coker.com.au)
Mon, 6 Dec 1999 13:07:07 +0100


On Mon, 06 Dec 1999, Alexander Viro wrote:
>On 5 Dec 1999, Trond Myklebust wrote:
>
>> > Physical layout of many filesystems doesn't support that. Ditto
>> > for NFS, IIRC.
>>
>> Both the NFSv2 and v3 protocols allow an unsigned 32-bit value for
>> 'nlinks'.
>
>Oops. Sorry. IOW, Sun decide that there was nothing around to pack
>together with nlinks and with 32bit alignment... Anyway, neither s5fs, nor
>FFS and its derivatives (UFS, ext2) have enough place in inode for that
>stuff. So...

Ext2 is currently being worked on significantly. The new one may be called
Ext3 or Journalled Ext2. In any case it's a significant change and changing
the storage of links should be possible.
ReiserFS is developing much faster and this could probably be added in a
smallish amount of time.
What about NetApp Filers and other NFS serving devices. I would expect a
NetApp to be able to do this - unless the client was a Linux machine...

As work is currently in progress on 64bit files, 32bit link counts would
probably be a good idea too. These changes should fit well together because
many of the same system calls have to be changed...

-- 
The ultimate result is that some innovations that would truly benefit
consumers never occur for the sole reason that they do not coincide with
Microsoft's self-interest.
-- Judge Thomas Penfield Jackson, U.S. District Judge

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/