Coda File System

Re: venus-kernel interface

From: Troy Benjegerdes <>
Date: Thu, 30 Sep 2004 16:02:01 -0500
On Thu, Sep 30, 2004 at 10:27:11AM +0200, Ivan Popov wrote:
> > > There were precedents, like podfuk-smb based on Coda, but it did not become
> > > popular enough to be maintained, and there is no such one now...
> > 
> > Podfuk got renamed to uservfs ( Then there is a 
> podfuk-smb is SMB fs implementation via Coda kernel module, not included with
> podfuk or uservfs. It can be found e.g. at
>  * A modified podfuk to provide a Network Neighbourhood type arrangement
>  * within a Unix filesystem.
>  *
>  * Podfuk pretends to be the Coda cache manager Venus.
>   ...
>  * Based on podfuk.c by Pavel Machek.
>  *
>  * Copyright 2000 Tim Potter,
>  * Copyright 1999 Pavel Machek,
> It is an old code and it is simplistic (which I like :)
> and it looks having been fully functional.
> If somebody cheers it up to work in a modern enviroment,
> it would be useful, not least as an "educational tool" about
> how filesystems should be connected together in the name space.
> It mounts smbfs in the Coda's fashion, via one and only /smb mountpoint.
> Solaris could profit from podfuk-smb as it lacks smbfs.
> (when we fix the Solaris kernel module for current 64-bits kernels)
> For Linux and *BSD their kernel support for CIFS is certainly more efficient
> than podfuk-smb (podfuk-smb has Coda's full-file-caching limitations).

If you added linux & solaris cachefs support to the coda kernel module,
you'd get around the objections people have to podfuk.   ;)

Yes, my next project is taking a look at the cachefs & venus code, but I
seem to have more ideas than time to work on them.. 
Received on 2004-09-30 17:03:08