Matthew Dillon <[EMAIL PROTECTED]> writes:
>:> distribute the inodes all over the cylinder group rather then concentrate
>:> all the inodes in one place.
>:
>:Yes. I have implemented most of the code. I noticed the "ls -al" is slow
>:but "ls" is OK.
>
> Yes, ls (without any options) is ok because the file type is now being
> stuffed in the directory entry, allowing ls (without any options) to
> avoid stat()ing the file.
Interesting - I made a similar mechanism in a hash-chain-based
filesystem to speed up directory listings; by storing all the commonly
accessed information about a file in the directory in a compressed format,
thus avoiding fetching the fileheader (inode) block for every file. The
speedup was impressive; I think I was getting 7-25 entries per 512-byte
sector; including just all ls -l information. The downside was increased
overhead on file-close-after-modify and create/delete, but not a lot. As a
side-benefit, recovery after a trashed FS is slightly easier since there's
more redundant information available (if the main directory sector/inode
gets whacked).
--
Randell Jesup, Worldgate Communications, ex-Scala, ex-Amiga OS team ('88-94)
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
