[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Hi Andrew!

From: Yura Umanets
Subject: Re: Hi Andrew!
Date: Mon, 06 May 2002 01:18:53 +0300
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7+) Gecko/20020124

Andrew Clausen wrote:

On Mon, May 06, 2002 at 01:00:24AM +0300, Yura Umanets wrote:

Andrew Clausen wrote:

BTW: you might find that it's useful to have both "smart" and
"dumb" resizing... since "smart" resizing has to modify the
entire file system (because each block is renumbered), it is
inherently slower.

I think smart resizing will be so fast as dumb is. This is because it is not nessecary to traverse through all filesystem and move every block to somewhere. I have got "the standard" relocation mechanism based on controled traverse through the balanced tree. It placed inside geom.c and tree.c. It makes the minimum of block relocations.

Yes, right.  But you still have to update every non-leaf node in
every tree, etc.

I must update certain block in cases:
1. Block is a leaf and it contains unformated items which points to the area out of new bounds. 2. Block is an internal node and it has modified due to leafs has chnaged location.
3. Block is unformated block and it lies out of new bounds.

OTOH, when growing a file system at the end, it can be REALLY fast,
for example.
Growing required just bitmap memory chunk to be reallocated and synced (very fast).

Close to O(1) time.  (You just need to grow some
bitmaps, right?).
:) right

So, it's good to make your algorithm fast for
"growing the end".  (It's a common thing system administrators
need to do)

Right, besides I see yet another reason, why I should to add new function (say reiserfs_fs_resize_smart). This reason is the interface number of libreiserfs which allows do not undate parted's reiserfs code right now :)

BTW, I have seen, you updated parted page :) By now reiserfs supported too :)


Did you hear something from that men who is working on jfs? It is likely parted will be support jfs too. Parted cool tool :))

Yep :)

Basically: I want all the file systems people to work together
more.  I think I'm succeeding now :)

I have conversation with Oleg Drokin. He made me aware about NAMESYS' usual working day, office rules, etc.

Are they ok?  Looking forward to it?

I'd say yes :) Too little constraints :)

Cool :)


reply via email to

[Prev in Thread] Current Thread [Next in Thread]