Patch "bpf: reject out-of-bounds stack pointer calculation" has been added to the 4.9-stable tree

[<gregkh@linuxfoundation.org>]

This is a note to let you know that I've just added the patch titled

    bpf: reject out-of-bounds stack pointer calculation

to the 4.9-stable tree which can be found at:
    http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
     bpf-reject-out-of-bounds-stack-pointer-calculation.patch
and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@vger.kernel.org> know about it.


>From foo@baz Fri Dec 22 16:57:35 CET 2017
From: Daniel Borkmann <daniel@iogearbox.net>
Date: Fri, 22 Dec 2017 16:29:04 +0100
Subject: bpf: reject out-of-bounds stack pointer calculation
To: gregkh@linuxfoundation.org
Cc: ast@kernel.org, daniel@iogearbox.net, jannh@google.com, stable@vger.kernel.org
Message-ID: <20171222152905.3455-4-daniel@iogearbox.net>

From: Daniel Borkmann <daniel@iogearbox.net>


From: Jann Horn <jannh@google.com>

Reject programs that compute wildly out-of-bounds stack pointers.
Otherwise, pointers can be computed with an offset that doesn't fit into an
`int`, causing security issues in the stack memory access check (as well as
signed integer overflow during offset addition).

This is a fix specifically for the v4.9 stable tree because the mainline
code looks very different at this point.

Fixes: 7bca0a9702edf ("bpf: enhance verifier to understand stack pointer arithmetic")
Signed-off-by: Jann Horn <jannh@google.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 kernel/bpf/verifier.c |   22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -1861,10 +1861,28 @@ static int check_alu_op(struct bpf_verif
 			   ((BPF_SRC(insn->code) == BPF_X &&
 			     regs[insn->src_reg].type == CONST_IMM) ||
 			    BPF_SRC(insn->code) == BPF_K)) {
-			if (BPF_SRC(insn->code) == BPF_X)
+			if (BPF_SRC(insn->code) == BPF_X) {
+				/* check in case the register contains a big
+				 * 64-bit value
+				 */
+				if (regs[insn->src_reg].imm < -MAX_BPF_STACK ||
+				    regs[insn->src_reg].imm > MAX_BPF_STACK) {
+					verbose("R%d value too big in R%d pointer arithmetic\n",
+						insn->src_reg, insn->dst_reg);
+					return -EACCES;
+				}
 				dst_reg->imm += regs[insn->src_reg].imm;
-			else
+			} else {
+				/* safe against overflow: addition of 32-bit
+				 * numbers in 64-bit representation
+				 */
 				dst_reg->imm += insn->imm;
+			}
+			if (dst_reg->imm > 0 || dst_reg->imm < -MAX_BPF_STACK) {
+				verbose("R%d out-of-bounds pointer arithmetic\n",
+					insn->dst_reg);
+				return -EACCES;
+			}
 			return 0;
 		} else if (opcode == BPF_ADD &&
 			   BPF_CLASS(insn->code) == BPF_ALU64 &&


Patches currently in stable-queue which might be from daniel@iogearbox.net are

queue-4.9/bpf-fix-branch-pruning-logic.patch
queue-4.9/bpf-adjust-insn_aux_data-when-patching-insns.patch
queue-4.9/bpf-fix-incorrect-sign-extension-in-check_alu_op.patch
queue-4.9/bpf-reject-out-of-bounds-stack-pointer-calculation.patch