mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Qiliang Yuan <odys.yuan@gmail.com>
To: Alexei Starovoitov <ast@kernel.org>,
	 Daniel Borkmann <daniel@iogearbox.net>,
	 John Fastabend <john.fastabend@gmail.com>,
	 Andrii Nakryiko <andrii@kernel.org>,
	Eduard Zingerman <eddyz87@gmail.com>,
	 Kumar Kartikeya Dwivedi <memxor@gmail.com>,
	 Martin KaFai Lau <martin.lau@linux.dev>,
	Song Liu <song@kernel.org>,
	 Yonghong Song <yonghong.song@linux.dev>,
	Jiri Olsa <jolsa@kernel.org>,
	 Emil Tsalapatis <emil@etsalapatis.com>,
	 Ihor Solodrai <ihor.solodrai@linux.dev>,
	 Luis Gerhorst <luis.gerhorst@fau.de>,
	 Henriette Herzog <henriette.herzog@rub.de>,
	 Anton Protopopov <a.s.protopopov@gmail.com>,
	 Xu Kuohai <xukuohai@huawei.com>,
	Hengqi Chen <hengqi.chen@gmail.com>,
	 Shuah Khan <shuah@kernel.org>
Cc: bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
	 linux-kselftest@vger.kernel.org,
	Qiliang Yuan <odys.yuan@gmail.com>
Subject: [PATCH bpf-next 4/5] selftests/bpf: Cover a nospec jump that becomes a nop
Date: Mon, 02 Nov 2026 07:37:04 -0500	[thread overview]
Message-ID: <20261102-bpf-blind-nops-fixes-v1-4-844300e42cdb@gmail.com> (raw)
In-Reply-To: <20261102-bpf-blind-nops-fixes-v1-0-844300e42cdb@gmail.com>

A conditional jump that is unsafe on a mispredicted path gets a barrier,
even when it later becomes a 'goto' over dead code only.

No test has such a jump, so nothing checks that bpf_opt_remove_nops()
keeps it with its barrier.

Compare a pointer on the mispredicted path at a jump that always goes
over dead code, and check that the barrier and the jump stay for a prog
loaded without CAP_PERFMON.

Reported-by: Alexei Starovoitov <ast@kernel.org>
Closes: https://lore.kernel.org/bpf/DMKHZS5B5CNO.186F6YH1T6YU4@gmail.com/
Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
---
 .../testing/selftests/bpf/progs/verifier_unpriv.c  | 35 ++++++++++++++++++++++
 1 file changed, 35 insertions(+)

diff --git a/tools/testing/selftests/bpf/progs/verifier_unpriv.c b/tools/testing/selftests/bpf/progs/verifier_unpriv.c
index a54b5a58b945b..b9d96366e656d 100644
--- a/tools/testing/selftests/bpf/progs/verifier_unpriv.c
+++ b/tools/testing/selftests/bpf/progs/verifier_unpriv.c
@@ -1072,4 +1072,39 @@ __naked void nospec_backward_call(void)
 "	::: __clobber_all);
 }
 
+/*
+ * Compare a pointer at l0 on the mispredicted path, so that the jump gets a
+ * barrier. On the real path r1 is 0, and the jump only goes over dead code and
+ * becomes a nop. Keep it with its barrier.
+ */
+SEC("socket")
+__description("unpriv: nospec on a jump that becomes a nop")
+__success __success_unpriv
+__caps_unpriv(CAP_BPF)
+__retval(0)
+#ifdef SPEC_V1
+__xlated_unpriv("r1 = 0")
+__xlated_unpriv("nospec")
+__xlated_unpriv("goto pc+0")
+__xlated_unpriv("r0 = 0")
+__xlated_unpriv("exit")
+#endif
+__naked void nospec_jump_becomes_nop(void)
+{
+	asm volatile ("					\
+	call %[bpf_get_prandom_u32];			\
+	r0 &= 0xff;					\
+	r1 = 0;						\
+	if r0 < 0x100 goto l0_%=;			\
+	/* executed only speculatively */		\
+	r1 = r10;					\
+l0_%=:	if r1 == 0 goto l1_%=;				\
+	r0 = 1;						\
+l1_%=:	r0 = 0;						\
+	exit;						\
+"	:
+	: __imm(bpf_get_prandom_u32)
+	: __clobber_all);
+}
+
 char _license[] SEC("license") = "GPL";

-- 
2.43.0


  parent reply	other threads:[~2026-11-02 12:37 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-02 12:37 [PATCH bpf-next 0/5] bpf: Fix three bugs around constant blinding and nop removal Qiliang Yuan
2026-11-02 12:37 ` [PATCH bpf-next 1/5] bpf: Put the line info back when the JIT falls back after blinding Qiliang Yuan
2026-11-02 12:37 ` [PATCH bpf-next 2/5] bpf: Keep the barrier of a nospec insn when removing nops Qiliang Yuan
2026-11-02 12:37 ` [PATCH bpf-next 3/5] bpf: Finalize the verifier log after the runtime is selected Qiliang Yuan
2026-11-02 12:37 ` Qiliang Yuan [this message]
2026-11-02 12:37 ` [PATCH bpf-next 5/5] selftests/bpf: Test log_true_size with an error from blinding Qiliang Yuan

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=20261102-bpf-blind-nops-fixes-v1-4-844300e42cdb@gmail.com \
    --to=odys.yuan@gmail.com \
    --cc=a.s.protopopov@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=eddyz87@gmail.com \
    --cc=emil@etsalapatis.com \
    --cc=hengqi.chen@gmail.com \
    --cc=henriette.herzog@rub.de \
    --cc=ihor.solodrai@linux.dev \
    --cc=john.fastabend@gmail.com \
    --cc=jolsa@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=luis.gerhorst@fau.de \
    --cc=martin.lau@linux.dev \
    --cc=memxor@gmail.com \
    --cc=shuah@kernel.org \
    --cc=song@kernel.org \
    --cc=xukuohai@huawei.com \
    --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

all inboxes | Powered by JetHome®