On Sat, Sep 28, 2013 at 7:35 AM, Djalal Harouni <tixxdz@xxxxxxxxxx> wrote: > On Thu, Sep 26, 2013 at 03:43:24PM -0500, Kees Cook wrote: >> On Wed, Sep 25, 2013 at 3:14 PM, Djalal Harouni <tixxdz@xxxxxxxxxx> wrote: >> > The /proc/*/stack contains sensitive information and currently its mode >> > is 0444. Change this to 0400 so the VFS will be able to block >> > unprivileged processes to get file descriptors on arbitrary privileged >> > /proc/*/stack files. >> > >> > The /proc/*/stack is a /procfs ONE file that shares the same ->open() >> > file operation with other ONE files. Doing a ptrace_may_access() check >> > during open() might break userspace from accessing other ONE files >> > like /proc/*/stat and /proc/*/statm. >> > >> > Therfore make it 0400 for now, and improve its check during ->read() >> > in the next following patch. >> > >> > Cc: Kees Cook <keescook@xxxxxxxxxxxx> >> > Cc: Eric W. Biederman <ebiederm@xxxxxxxxxxxx> >> > Signed-off-by: Djalal Harouni <tixxdz@xxxxxxxxxx> >> >> While the rest of the series is being discussed, I think it would be >> nice to at least get this into the tree. Fixing this reduces which >> processes are exposed to ASLR leaks. The rest of the series closes the >> remaining holes. >> >> I would if it would be valuable adding a test for the identified leak >> conditions to some test suite? LTP perhaps? > I'm not familiar with LTP, but I guess a small program that perform I/O > redirection and execve a suid-exec will do it? It's actually a giant program. :) http://ltp.sourceforge.net/ > I'll try to add code comment in fs/proc/base.c Mostly I'm just thinking that this is a rather fragile area, and it would be nice to have a set of tests that verify we haven't broken any of the known test-cases. -Kees -- Kees Cook Chrome OS Security -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
