On Mon, Aug 11, 2003 at 04:47:31PM +0100, Sam Vilain wrote: > > Opinion Poll! > > let's assume each file and directory carry a tag which > says "this is a file of context N", where N is the context > number of a virtual server.
Hi Sam! maybe you should have a look at the archives ... > An idea I just had is to treat it like an extension to the user ID - > eg, if you are using 16 bit user IDs then the context + the uid is the > `system userid' of 32 bits, but with special behaviour (such as > setting a default, meaning `any context', etc) when the context part > is 0 or 1. That way, files are uniquely identifiable between > contexts. wow! great idea, and it actually works, or at least seems to, as I use it since october 2002 ... 8-) http://www.13thfloor.at/VServer/Concepts.shtml > btw, where would you put those extra bits for each inode, is there > room in the ext2/reiser/etc reserved structures? Of course you could > use the top half of the nice shiny 32-bit UIDs in Linux 2.6 :-) > This would mean adding syntax to `chown' and/or `chgrp' to specify a > context name as well as a username (eg, chown [EMAIL PROTECTED]:[EMAIL PROTECTED] > filename). http://vserver.13thfloor.at/Linux2.6/index.php?page=Per+Context+Quota there you have the chctx/lsctx tools too ... what a surprise *g* ... > It could also be a different command, chctx, as suggested elsewhere. > But personally, it looks like ownership to me. > > 2) if a program of context N encounters a file of > context M, where N != M ... > > a) on modify change the file to the new context? > b) do not allow access to files from other contexts > except context zero/one? > c) allow modification while keeping the file > in its 'original' context? > > 3) consider a program creating a (hard)link to a file > in another context (including zero/one), should ... > > a) the file change to the 'new' context? > b) the file keep the old context? > c) this operation be disallowed? > > 4) consider a program removing a link to a file with > more than one links, should the remaining links ... > > a) be still 'owned' by the removing context? > b) be changed to context zero/one? > > The behaviour should be exactly as if it were owned by a different > user. objection, the least thing to consider is root in different contexts, which you do not want to be handled like 'normal' users ... best, Herbert > -- > Sam Vilain, [EMAIL PROTECTED] > > C++, where only your friends can access your private parts. >
