VFS commit schedule HEADS UP, and vfsx24.patch available - vfs work (expert developers only)

Sean Hogan shogun70 at westnet.com.au
Wed Nov 10 18:35:09 PST 2004


Matthew Dillon wrote:

:Wouldn't UNIONFS and NULLFS be just as suited to being implemented in 
:userland as any filesystem?
:
:I anticipated some daemon vaguely akin to amd which handles requests to 
:union and null mounted filesystems.  And portalfs, etc, etc.  Any 
:filesystem that is more-or-less a redirection service.
:
:If the user VFS API was ready you might also get more filesystem 
:developers.  Or lower quality filesystem developers.  Or a free set of 
:steak knives.
:
:cheers,
:SDH

    I would say so, but a prerequisit to the user VFS layering is kernel
    VFS threading which I haven't gotten to yet.  The current VFS patch set
    is focused on the VOP_*() API fixes and namecache conversion (which 
    themselves are prereqs to being able to properly thread VFS).
   
					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>
Um, yes.  Just like you said in your VFS roadmap back with your first 
VFS patch.  I forgot/didn't read it properly.
Apologies.
SDH






More information about the Kernel mailing list