From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Sun 27 Oct 2002 - 06:07:39 GMT
On Thu, Oct 24, 2002 at 11:07:28PM -0400, tedsuo wrote:
> Hello
> How do you expect to do that *simply*, i mean it look quite
> difficult to play in that.
I found some time this weekend, so I thought, give it
a try and play a little with the kernel ...
I think, I implemented the context quota support
for ext2/ext3 by taking a more general aproach.
(had no time to test it ... yet)
[get the patches from http://www.13thfloor.at/VServer/]
- the inode structure was extended (not the raw_inode)
by the new context information (32bit).
- the raw_inode <-> inode information exchange was
modified to map the new context information with
the upper 16 uid/gid bits in the filesystem.
- the quota definition was extended by a context
quota (CTXQUOTA).
- inode allocation/ownership change for ext2/ext3
was modified to reflect the context information.
- stat was modified to report the "true" uid/gid
for contexts < 2 by applying the same mapping.
then I started to modified the quota-tools to recognize
the "new" context quota (-x) which is done for
quotacheck and reportquota not for quotaon/off ...
what should work?
- creating/chowning/destroying files in context X
with uid/gid ((X << 16) | ID) ...
- access restrictions across contexts
- quotacheck -acx on filesystems ext2/3 with
option ctxquota in mtab
- repquota -ax on filesystems ext2/3 with
option ctxquota in mtab
so what is missing?
- physical (context 0) files should appear in
every context with "context" permissions
- mount should accept the ctxquota option
- quotaon/off, quot, quota (for context)
- transfer files/dirs to/from context
- modifications for other filesystems
> if everybody get together to work on quota, maybe we will
> get quota up in the next vserver version...
I realy hope so ...
best,
Herbert