Re: [PATCH] General filesystem cache

Re: [PATCH] General filesystem cache

[Linus Torvalds]

On Fri, 18 Jul 2003, David Howells wrote:
> 
> Here's a patch to add a quasi-filesystem ("CacheFS") that turns a block device
> into a general cache for any other filesystem that cares to make use of its
> facilities.
> 
> This is primarily intended for use with my AFS filesystem, but I've designed
> it such that it needs to know nothing about the filesystem it's backing, and
> so it may also be useful for NFS, SMB and ISO9660 for example.

Ok. Sounds good. In fact, it's something I've wanted for a while, since 
it's also potentially the solution to performance-critical things like 
virtual filesystems based on revision control logic etc (traditionally 
done with fake NFS servers).

I did a very very quick scan, and didn't see anything that raised my 
hackles. But it's late in the 2.6.x game, and as a result I'm not going to 
apply it until I get a lot of feedback from actual users too.

		Linus

Re: [PATCH] General filesystem cache

[David Chow]

Is it a file system or a patch to the VFS? I love to see this patch. It 
is quite useful for me. Is it available for 2.4?

David Chow

Linus Torvalds wrote:

>On Fri, 18 Jul 2003, David Howells wrote:
>  
>
>>Here's a patch to add a quasi-filesystem ("CacheFS") that turns a block device
>>into a general cache for any other filesystem that cares to make use of its
>>facilities.
>>
>>This is primarily intended for use with my AFS filesystem, but I've designed
>>it such that it needs to know nothing about the filesystem it's backing, and
>>so it may also be useful for NFS, SMB and ISO9660 for example.
>>    
>>
>
>Ok. Sounds good. In fact, it's something I've wanted for a while, since 
>it's also potentially the solution to performance-critical things like 
>virtual filesystems based on revision control logic etc (traditionally 
>done with fake NFS servers).
>
>I did a very very quick scan, and didn't see anything that raised my 
>hackles. But it's late in the 2.6.x game, and as a result I'm not going to 
>apply it until I get a lot of feedback from actual users too.
>  
>

Re: [PATCH] General filesystem cache

[David Howells]

> 
> Is it a file system or a patch to the VFS? I love to see this patch. It 
> is quite useful for me. Is it available for 2.4?

Not yet. There are plans to backport it at some point - possibly by me though
I may be able to twist Jeff Garzik's arm sufficiently that he'll do the work.

David

Re: [PATCH] General filesystem cache

[Jeff Garzik]

Linus Torvalds wrote:
> On Fri, 18 Jul 2003, David Howells wrote:
> 
>>Here's a patch to add a quasi-filesystem ("CacheFS") that turns a block device
>>into a general cache for any other filesystem that cares to make use of its
>>facilities.
>>
>>This is primarily intended for use with my AFS filesystem, but I've designed
>>it such that it needs to know nothing about the filesystem it's backing, and
>>so it may also be useful for NFS, SMB and ISO9660 for example.
> 
> 
> Ok. Sounds good. In fact, it's something I've wanted for a while, since 
> it's also potentially the solution to performance-critical things like 
> virtual filesystems based on revision control logic etc (traditionally 
> done with fake NFS servers).


I've been pushing David to keep it general enough to work for NFS, which 
has been a long term goal of mine...  (and it appears my nagging worked)

Thanks David!

	Jeff

Re: [PATCH] General filesystem cache

[David Howells]

> 
> I've been pushing David to keep it general enough to work for NFS, which 
> has been a long term goal of mine...  (and it appears my nagging worked)

Didn't require any nagging on your part. By the way, can I nag you in return
into backporting to 2.4? :-)

David

Re: [PATCH] General filesystem cache

[David Howells]

> 
> Ok. Sounds good. In fact, it's something I've wanted for a while, since 
> it's also potentially the solution to performance-critical things like 
> virtual filesystems based on revision control logic etc (traditionally 
> done with fake NFS servers).

It really does make a difference to my AFS client:-)

> I did a very very quick scan, and didn't see anything that raised my 
> hackles. But it's late in the 2.6.x game, and as a result I'm not going to 
> apply it until I get a lot of feedback from actual users too.

I can understand that... I've been developing it on and off for a while (it
has had to take the backseat to other things unfortunately, depending on RH's
whims).

BTW, will you be at OLS? I'm only going to that not being invited to the KS
(though David Woodhouse has my mobile number;-).

David

Re: [PATCH] General filesystem cache

[David Howells]

> 
> Is this patch available from somewhere?  I looked on the MARC archives but 
> haven't been able to find the original mail which includes the patch.

I don't know. The message containing the patches doesn't seem to have made it
into either the LKML or the linux-fsdevel mailing lists.

The basic code can be found in:

	http://cvs.infradead.org/cgi-bin/cvsweb.cgi/afs/

In directories:

	Documentations/filesystems/
	include/linux/
	fs/cachefs/

I'll see about putting the patch up for download when I get to OLS.

David

Re: [PATCH] General filesystem cache

[Rob Landley]

On Sunday 20 July 2003 05:38, David Howells wrote:
> > Is this patch available from somewhere?  I looked on the MARC archives
> > but haven't been able to find the original mail which includes the patch.
>
> I don't know. The message containing the patches doesn't seem to have made
> it into either the LKML or the linux-fsdevel mailing lists.
>
> The basic code can be found in:
>
> 	http://cvs.infradead.org/cgi-bin/cvsweb.cgi/afs/
>
> In directories:
>
> 	Documentations/filesystems/
> 	include/linux/
> 	fs/cachefs/
>
> I'll see about putting the patch up for download when I get to OLS.
>
> David

Random question:

Way back when I used union mount code (under OS/2) that would mount the first 
filesystem read-only, and allow you to cache changes to it in a second 
filesystem.  (So you could do a build of slightly self-modifying code, a lot 
like the Linux build in 2.4 was, against a read-only tree.) 

Would this be a good tool for that sort of application?  (There's all sorts of 
other "trial run" type applications; running stuff against database snapshots 
and then seeing what the deltas are, etc...)

Rob