Skip to contentSkip to conversation

Squashfs 2.1 released (compressed filesystem)

Announce Phillip Lougher 4 participants 8 messages Dec 10, 2004 – Dec 14, 2004 VFS

lore
Share & tools
lore
#0 lore raw

Hi all,

I'm pleased to announce the release of Squashfs 2.1. This release introduces indexed directories which considerably speed up directory lookup (ls, find, etc.) for directories which are greater than 8K in size. All directories are now also sorted alphabetically which further speeds up directory lookup.

Many smaller improvements have also been made in this release, and there are, for the first time, the results of some tests of Squashfs lookup and I/O performance against Zisofs, Cloop, and CRAMFS.

For further details please go to the project page http://squashfs.sourceforge.net.

Regards

Phillip Lougher

#1 lore raw
Quoted 2 lines · Phillip Lougher

I'm pleased to announce the release of Squashfs 2.1.

Are you going to submit this fs for inclusion in the main kernel tree? I'm getting tired of maintaining it as part of the Gentoo kernel patch set :)

(Gentoo uses it for their LiveCD images)

thanks,

greg k-h

#2 lore raw
Quoted 8 lines · Greg KH
Quoted 1 line · Phillip Lougher

I'm pleased to announce the release of Squashfs 2.1.

Are you going to submit this fs for inclusion in the main kernel tree? I'm getting tired of maintaining it as part of the Gentoo kernel patch set :)

Good question... When I originally released Squashfs (Oct 2002) the 2.5 kernel had just gone into the feature freeze and I was waiting for the unfreeze that would happen when 2.7 arrived. As the official stance on additions to 2.6 has been relaxed I have thought about submitting it.

I need to tidy up the code a bit before I submit it to the merciless scrutiny of LKML :-) Nothing bad (I hope) but there's lots of long lines and I do know LKML'mers like 80 columns.

I'm planning on adding ea/acl support (people have been asking for them), I can't decide whether to submit it now or wait until I've finished them. Suggestions and advice would be welcome.

Slightly off topic, I've noticed you're the kernel maintainer for Gentoo. You mentioned somewhere you're down to 4 kernel patches, including Squashfs? :) A lot of people/projects are now using Squashfs and it would help a lot of people (and me) if it did get into the kernel. Plus it would be a nice thing for me anyway to have finally got it included.

Regards

Phillip

#3 lore raw
Quoted 15 lines · Phillip Lougher
Quoted 8 lines · Greg KH
Quoted 1 line · Phillip Lougher

I'm pleased to announce the release of Squashfs 2.1.

Are you going to submit this fs for inclusion in the main kernel tree? I'm getting tired of maintaining it as part of the Gentoo kernel patch set :)

Good question... When I originally released Squashfs (Oct 2002) the 2.5 kernel had just gone into the feature freeze and I was waiting for the unfreeze that would happen when 2.7 arrived. As the official stance on additions to 2.6 has been relaxed I have thought about submitting it.

Quoted 7 lines

I need to tidy up the code a bit before I submit it to the merciless scrutiny of LKML :-) Nothing bad (I hope) but there's lots of long lines and I do know LKML'mers like 80 columns.

I'm planning on adding ea/acl support (people have been asking for them), I can't decide whether to submit it now or wait until I've finished them. Suggestions and advice would be welcome.

Submit it now, it's self contained (doesn't patch anything outside it's own directory), and is useful in it's current state for a lot of people.

Quoted 3 lines

Slightly off topic, I've noticed you're the kernel maintainer for Gentoo. You mentioned somewhere you're down to 4 kernel patches, including Squashfs? :)

Yes, squashfs is one of the four non-bugfix patches in the Gentoo kernel 2.6 kernel package.

For those who are interested, the other three are:
- fbsplash, the "next-gen" version of bootsplash. No more image decoders in kernelspace (compared to bootsplash), and a few more features.
- speakup, a driver package for visually-impaired users.
- vesafb-tng, a "next-gen" version of vesafb-tng.
- inotify.

Ok, there are now 5 patches, I forgot we addeed inotify recently to make the Gnome developers happy.

Quoted 3 lines

A lot of people/projects are now using Squashfs and it would help a lot of people (and me) if it did get into the kernel. Plus it would be a nice thing for me anyway to have finally got it included.

I agree, it would also reduce your workload of keeping an out-of-the-kernel fs patchset.

thanks,

greg k-h

#4 lore raw

nat/masquerade with 2.6.x

Hello,
I have a local network with private directions, and a adsl conection. I want to do masquerade. Then I do:
echo 1 > /proc/sys/net/ipv4/ip_forward
iptables --table nat --append POSTROUTING --out-interface eth1 -j
MASQUERADE
where eth1 is the interface conected to internet. This work perfectly with the kernels 2.4.x.
But one week ago I installed the kernel 2.6.8 and when I do :
echo 1 > /proc/sys/net/ipv4/ip_forward
iptables --table nat --append POSTROUTING --out-interface
$ifc_internet -j MASQUERADE
the hosts of the internal network can do ping to internet, and this is normal, but they can not open any web or conection the msn , they only can do ping. This is very stranger. I try the kernels 2.6.7,2.6.7,2.6.8 and 2.6.9 and they do no work.

I know that the dns is working because when I do "ping www.google.es" from internal host this work. There are not other rules in the FORWARD chain, look:

Chain INPUT (policy ACCEPT 2 packets, 100 bytes)
pkts bytes target prot opt in out
source destination

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
pkts bytes target prot opt in out
source destination

Chain OUTPUT (policy ACCEPT 5958 packets, 2480411 bytes)
pkts bytes target prot opt in out
source destination

and if i do iptables -t nat -L -nvx, then:

Chain PREROUTING (policy ACCEPT 0 packets, 0 bytes)
pkts bytes target prot opt in out
source destination

Chain POSTROUTING (policy ACCEPT 1 packets, 60 bytes)
pkts bytes target prot opt in out
source destination
0 0 MASQUERADE all -- * eth1
0.0.0.0/0 0.0.0.0/0

Chain OUTPUT (policy ACCEPT 1 packets, 60 bytes)
pkts bytes target prot opt in out
source destination

And this is all,
Can somebody help me, please?
Sorry for my bad english.

#5 lore raw

proc/sys/net/ipv4/tcp_bic

How affect the value of the file proc/sys/net/ipv4/tcp_bic if I want to do masquerade? Is possible that if proc/sys/net/ipv4/tcp_bic=1 the tcp connections don't work fine if I do masquerade?

Thanks.

#6 lore raw
In reply to Phillip Lougher

Hi Phillip,

Quoted 3 lines

Many smaller improvements have also been made in this release, and there are, for the first time, the results of some tests of Squashfs lookup and I/O performance against Zisofs, Cloop, and CRAMFS.

I've checked you tests, numbers are appealing :-)

BTW, you need to apply the following patch to build mksquashfs with gcc < 3.

Thanks for still improving this great FS !
Willy

squashfs-tools/mksquashfs.c renamed from squashfs-tools/mksquashfs.c.orig +1 -2
@@ -526,11 +526,10 @@
 	else if(type == SQUASHFS_LDIR_TYPE) {
 		int i;
 		unsigned char *p;
+		squashfs_ldir_inode_header *dir = &inode_header.ldir, *inodep;
 
 		if(byte_size >= 1 << 27)
 			BAD_ERROR("directory greater than 2^27-1 bytes!\n");
-
-		squashfs_ldir_inode_header *dir = &inode_header.ldir, *inodep;
 
 		inode = get_inode(sizeof(*dir) + i_size);
 		inodep = (squashfs_ldir_inode_header *) inode;
#7 lore raw
Quoted 6 lines · Willy Tarreau

I've checked you tests, numbers are appealing :-)

BTW, you need to apply the following patch to build mksquashfs with gcc < 3.

Thanks for still improving this great FS !
Willy

Hi

Thanks for your comments and thanks for the patch. I also received reports that the kernel patches didn't build with gcc < 3. This is because I used unamed structures and unions which are not supported by gcc < 3. The "fun" of dealing with all the differing gcc versions in use out there :-) I'll do another release with the code changes.

Phillip