BPF List
 help / color / mirror / Atom feed
From: Eduard Zingerman <eddyz87@gmail.com>
To: bpf@vger.kernel.org, ast@kernel.org, andrii@kernel.org
Cc: daniel@iogearbox.net, martin.lau@linux.dev, kernel-team@fb.com,
	yonghong.song@linux.dev, eddyz87@gmail.com
Subject: [PATCH bpf v1 2/2] selftests/bpf: widen_imprecise_scalars() and different stack depth
Date: Thu, 13 Nov 2025 18:57:30 -0800	[thread overview]
Message-ID: <20251114025730.772723-2-eddyz87@gmail.com> (raw)
In-Reply-To: <20251114025730.772723-1-eddyz87@gmail.com>

A test case for a situation when widen_imprecise_scalars() is called
with old->allocated_stack > cur->allocated_stack. Test structure:

    def widening_stack_size_bug():
      r1 = 0
      for r6 in 0..1:
        iterator_with_diff_stack_depth(r1)
        r1 = 42

    def iterator_with_diff_stack_depth(r1):
      if r1 != 42:
        use 128 bytes of stack
      iterator based loop

iterator_with_diff_stack_depth() is verified with r1 == 0 first and
r1 == 42 next. Causing stack usage of 128 bytes on a first visit and 8
bytes on a second. Such arrangement triggered a KASAN error in
widen_imprecise_scalars().

Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
---
 .../selftests/bpf/progs/iters_looping.c       | 53 +++++++++++++++++++
 1 file changed, 53 insertions(+)

diff --git a/tools/testing/selftests/bpf/progs/iters_looping.c b/tools/testing/selftests/bpf/progs/iters_looping.c
index 05fa5ce7fc59..d00fd570255a 100644
--- a/tools/testing/selftests/bpf/progs/iters_looping.c
+++ b/tools/testing/selftests/bpf/progs/iters_looping.c
@@ -161,3 +161,56 @@ int simplest_loop(void *ctx)
 
 	return 0;
 }
+
+__used
+static void iterator_with_diff_stack_depth(int x)
+{
+	struct bpf_iter_num iter;
+
+	asm volatile (
+		"if r1 == 42 goto 0f;"
+		"*(u64 *)(r10 - 128) = 0;"
+	"0:"
+		/* create iterator */
+		"r1 = %[iter];"
+		"r2 = 0;"
+		"r3 = 10;"
+		"call %[bpf_iter_num_new];"
+	"1:"
+		/* consume next item */
+		"r1 = %[iter];"
+		"call %[bpf_iter_num_next];"
+		"if r0 == 0 goto 2f;"
+		"goto 1b;"
+	"2:"
+		/* destroy iterator */
+		"r1 = %[iter];"
+		"call %[bpf_iter_num_destroy];"
+		:
+		: __imm_ptr(iter), ITER_HELPERS
+		: __clobber_common, "r6"
+	);
+}
+
+SEC("socket")
+__success
+__naked int widening_stack_size_bug(void *ctx)
+{
+	/*
+	 * Depending on iterator_with_diff_stack_depth() parameter value,
+	 * subprogram stack depth is either 8 or 128 bytes. Arrange values so
+	 * that it is 128 on a first call and 8 on a second. This triggered a
+	 * bug in verifier's widen_imprecise_scalars() logic.
+	 */
+	asm volatile (
+		"r6 = 0;"
+		"r1 = 0;"
+	"1:"
+		"call iterator_with_diff_stack_depth;"
+		"r1 = 42;"
+		"r6 += 1;"
+		"if r6 < 2 goto 1b;"
+		"r0 = 0;"
+		"exit;"
+		::: __clobber_all);
+}
-- 
2.51.1


  reply	other threads:[~2025-11-14  2:57 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-11-14  2:57 [PATCH bpf v1 1/2] bpf: account for current allocated stack depth in widen_imprecise_scalars() Eduard Zingerman
2025-11-14  2:57 ` Eduard Zingerman [this message]
2025-11-14 17:57 ` patchwork-bot+netdevbpf

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20251114025730.772723-2-eddyz87@gmail.com \
    --to=eddyz87@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=kernel-team@fb.com \
    --cc=martin.lau@linux.dev \
    --cc=yonghong.song@linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox