From mboxrd@z Thu Jan 1 00:00:00 1970 From: pmoore@redhat.com (Paul Moore) Date: Fri, 08 Nov 2013 13:23:53 -0500 Subject: [libseccomp-discuss] [PATCH v2] seccomp: not compatible with ARM OABI In-Reply-To: References: <20131107174746.GA22344@www.outflux.net> <1464650.041viV29xe@sifl> Message-ID: <2521056.N3R8bTMgyo@sifl> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Friday, November 08, 2013 08:39:29 AM Andy Lutomirski wrote: > On Fri, Nov 8, 2013 at 8:29 AM, Paul Moore wrote: > > On Thursday, November 07, 2013 11:05:26 AM Andy Lutomirski wrote: > >> On Thu, Nov 7, 2013 at 10:56 AM, Eric Paris wrote: > >> > Isn't x32 similarly screwy? Does it work because the syscall numbers > >> > are different? > >> > >> Yes (from reading the code -- I haven't actually tried it). > > > > I've got a x32 VM that I boot occasionally to test seccomp/libseccomp. > > For the purposes of seccomp it looks exactly like x86_64, including > > sharing the same AUDIT_ARCH_X86_64 value, the only difference being the > > syscall number offset ... Assuming you're using kernel 3.9 or later. > > Previous kernels had a bug which stripped the x32 syscall offset so it was > > impossible to distinguish from x86_64 and x32 with seccomp. See the > > following commit for the details: > > Ooh -- where did you get this? (I imagine I could debootstrap such a > beast and then just chroot / nspawn / schroot in, but if there are > readily available images, that would be great. Fedora doesn't seem to > have much x32 support.) I built up a small Gentoo image: * http://distfiles.gentoo.org/releases/amd64/current-stage3 -- paul moore security and virtualization @ redhat