On Thu, Apr 13, 2006 at 10:03:35PM -0700, EKC wrote:
> I'm using the vs2.1.1-rc16 patch. The additional nameidata parameter
> in the interface to vfs_mkdir and other vfs functions breaks the
> compilation of certain dependencies which do not supply a nameidata
> parameter(Lustre, for instance).
those changes are part of the (not really Linux-VServer
related) BME (Bind Mount Extenstions) patches, which I
maintain for several years now (they allow --bind mounts
to be read-only for example, instead of silently ignoring
the ro mount option and staying read-write)
to do those checks it was necessary to provide more
information to certain vfs functions, so that they can
tell the attributes of the mount
> If I pass in a value of NULL for nameidata when calling vfs
> operations, such as vfs_mkdir, what will the repercussions be?
if you scan the kernel code, you will see that certain
places (which do not have that information available)
already call them with a NULL argument, which basically
disables the additional checks, but I'd assume that a
proper filesystem (such as Lustre) will have that data
already available, and you just have to pass it to those
vfs functions, which would be preferred, as it would
allow BME to work on that filesystems too ...
HTH,
Herbert
> Thanks
> _______________________________________________
> Vserver mailing list
> Vserver@list.linux-vserver.org
> http://list.linux-vserver.org/mailman/listinfo/vserver
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Fri Apr 14 17:01:34 2006