On Tue, 2023-11-28 at 03:33 +0200, Eduard Zingerman wrote:
[...]
> Also, I think there are some tests that do oob stack read in branches
> that should be proven unreachable, with expectation that if certain
> verifier logic does not work as expected stack access would serve as a
> canary. Have no idea how to identify these tests, though.
I looked through all test cases I ever added (not so many) and it
seems that only one test case should be updated:
diff --git a/tools/testing/selftests/bpf/progs/iters.c b/tools/testing/selftests/bpf/progs/iters.c
index b2181f850d3e..3aca3dc145b5 100644
--- a/tools/testing/selftests/bpf/progs/iters.c
+++ b/tools/testing/selftests/bpf/progs/iters.c
@@ -846,7 +846,7 @@ __naked int delayed_precision_mark(void)
"call %[bpf_iter_num_next];"
"if r0 == 0 goto 2f;"
"if r6 != 42 goto 3f;"
- "r7 = -32;"
+ "r7 = -33;"
"call %[bpf_get_prandom_u32];"
"r6 = r0;"
"goto 1b;\n"
Here oob access is replaced by unaligned, this does not affect error
message, but should be future proof in case if widening logic would
get smarter.
