Messages in this thread | | | From | (Eric W. Biederman) | Subject | Re: [PATCH -mm 5/7] add user namespace | Date | Thu, 13 Jul 2006 21:45:49 -0600 |
| |
Dave Hansen <haveblue@us.ibm.com> writes:
> On Thu, 2006-07-13 at 13:02 -0600, Eric W. Biederman wrote: >> All comparisons of a user equality need to be of the tuple (user namespace, > user id). >> Any comparison that does not do that is an optimization. > ... >> So my impression was that Cedric's patchset was overoptimized because >> it did not change most of the uid comparisons, to (user namespace, user id). > > I might just be tempted to call them bugs so people understand what I'm > talking about ;) > >> Because you can have access to files created in another user namespace it >> is very unlikely that optimization will apply very frequently. The easy > scenario >> to get access to a file descriptor from another context is to consider unix >> domain sockets. > > OK, so you're saying that the lack of checks will cause problems rarely, > and that passing a fd across a unix domain sockets is one of the times > when you _could_ encounter this problem?
I think for filesystems like /proc and /sys that there will normally be problems. However many of those problems can be rationalized away as a reasonable optimization, or are not immediately apparent.
Passing a file descriptor between process in a unix domain socket is a case where I can easily construct scenarios where there are indisputable problems. It is one of my standard thought experiments to see if a namespace is sound.
Eric - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |