I don't have the hardware, so this is only compile-tested. Patches 1 and 2 should be safe. Patch 1 is a bugfix, although given the bitness sensitivity of the MPX data structures, any user code that hits the bug is probably doomed to failure anyway. Patch 2 is a hardening change that adds almost no complexity (it's mostly just reordering some code) that will make it considerably harder to exploit a possibly insn decoder vulnerability using threads that modify MPX instructions that send #BR before the trap handler runs.
Patch 3 will be much more controversial, since it's a complete ABI break. The ABI in question has never appeared in a released kernel, though. If patch 3 is not okay, then I want to fix the prctl implementation to at least validate its arguments. Andy Lutomirski (3): x86, mpx: Check user mode bitness correctly when decoding instructions x86, mpx: Short-circuit the instruction decoder for unexpected opcodes x86, mpx: Change the MPX enable/disable API to arch_prctl arch/x86/include/asm/ptrace.h | 5 +++++ arch/x86/include/uapi/asm/prctl.h | 15 +++++++++++---- arch/x86/kernel/process_64.c | 10 ++++++++++ arch/x86/mm/mpx.c | 28 ++++++++++++++++++---------- include/uapi/linux/prctl.h | 6 ------ kernel/sys.c | 12 ------------ 6 files changed, 44 insertions(+), 32 deletions(-) -- 2.1.0 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
