LXer: JFS File-System Can Now Handle SSD TRIM Discard

LXer · 10-03-2012, 11:40 PM

Published at LXer:

Yet another change for the Linux 3.7 kernel is that the JFS file-system now supports the SSD TRIM command. The JFS file-system now supports TRIM for informing solid-state disks about blocks of data that are no longer in use and can be wiped internally. There's already been Linux support for the TRIM command on SSDs with EXT4, Btrfs, FAT, XFS, and other file-systems, but now it's come to the aging JFS file-system.

Read More...

H_TeXMeX_H · 10-04-2012, 05:29 AM

That's great news, because JFS is one of the fastest filesystems on SSDs.
Although I am boycotting phoronix, this older benchmark is reasonably good:
http://www.phoronix.com/scan.php?pag...38_large&num=1

Of course, nobody cares, everyone will use ext4, because it's the default.

catkin · 10-04-2012, 05:42 AM

Quote:

Originally Posted by H_TeXMeX_H

Of course, nobody cares, everyone will use ext4, because it's the default.

I care H_TeXMeX_H because I use JFS for all file systems except UFDs and boot -- and my next computer will have an SSD

H_TeXMeX_H · 10-04-2012, 05:54 AM

Well, you can see if it's true, if it is as fast as the benchmarks say.

Jeebizz · 10-04-2012, 06:23 PM

So JFS is still being worked on!? Thats good news considering the last changes in it's changelog:

Quote:

http://jfs.cvs.sourceforge.net/*chec...?revision=1.33

New in 1.1.15 - 2011-03-04
* Several fixes for large filesystems where 64-bit variables are needed
* Fix incorrect size check on directories
* Make the timestamp format consistent

New in 1.1.14 - 2009-04-06
* Fix journal replay to handle records that span 3 pages
* Increase the maximum inline journal size from 32 MB to 128 MB
* Don't call ioctl(BLKFLSBUF) on a ramdisk. It wipes out everything.
* Fix fsck abort

Considering those in changelog I had my doubts, and was seriously thinking about moving on since even XFS has waay more activity than JFS.