| Message ID | 20221031055440.3594315-2-npiggin@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Remove STACK_FRAME_OVERHEAD | expand |
On Mon, Oct 31, 2022 at 03:54:22PM +1000, Nicholas Piggin wrote: > Could the user set r1 to be equal to the address matching the first > interrupt frame - STACK_INT_FRAME_SIZE, which is in the previous page > due to the kernel redzone, and induce the kernel to load the marker from > there? Possibly it could cause a crash at least. Yes, the user can set r1 to anything, it is just a general purpose register. This isn't a valid thing to do of course, the ABI requires r1 to point at a valid stack at all times, but it is an obvious attack point if we do not harden against this :-) Segher
diff --git a/arch/powerpc/perf/callchain.c b/arch/powerpc/perf/callchain.c index 082f6d0308a4..8718289c051d 100644 --- a/arch/powerpc/perf/callchain.c +++ b/arch/powerpc/perf/callchain.c @@ -61,6 +61,7 @@ perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *re next_sp = fp[0]; if (next_sp == sp + STACK_INT_FRAME_SIZE && + validate_sp(sp, current, STACK_INT_FRAME_SIZE) && fp[STACK_FRAME_MARKER] == STACK_FRAME_REGS_MARKER) { /* * This looks like an interrupt frame for an
The interrupt frame detection and loads from the hypothetical pt_regs are not bounds-checked. The next-frame validation only bounds-checks STACK_FRAME_OVERHEAD, which does not include the pt_regs. Add another test for this. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- Could the user set r1 to be equal to the address matching the first interrupt frame - STACK_INT_FRAME_SIZE, which is in the previous page due to the kernel redzone, and induce the kernel to load the marker from there? Possibly it could cause a crash at least. It also seems a bit rude to put a fancy next-frame-validation out in perf/ rather than with the rest of the frame validation code. Thanks, Nick arch/powerpc/perf/callchain.c | 1 + 1 file changed, 1 insertion(+)