MIME-Version: 1.0
References: <20220108051121.28632-1-yichun@openresty.com>
In-Reply-To: <20220108051121.28632-1-yichun@openresty.com>
From:   Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date:   Tue, 11 Jan 2022 09:51:13 -0800
Message-ID: <CAADnVQLRtVtfw3GxiHskLRBV8BKgeEVOP8qbje-mRNKn9rMOFw@mail.gmail.com>
Subject: Re: [PATCH] bpf: core: Fix the call ins's offset s32 -> s16 truncation
To:     "Yichun Zhang (agentzh)" <yichun@openresty.com>
Cc:     Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@kernel.org>,
        Network Development <netdev@vger.kernel.org>,
        bpf <bpf@vger.kernel.org>, LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Jan 7, 2022 at 9:11 PM Yichun Zhang (agentzh)
<yichun@openresty.com> wrote:
>
> The BPF interpreter always truncates the BPF CALL instruction's 32-bit
> jump offset to 16-bit. Large BPF programs run by the interpreter often
> hit this issue and result in weird behaviors when jumping to the wrong
> destination instructions.
>
> The BPF JIT compiler does not have this bug.
>
> Fixes: 1ea47e01ad6ea ("bpf: add support for bpf_call to interpreter")
> Signed-off-by: Yichun Zhang (agentzh) <yichun@openresty.com>
> ---
>  kernel/bpf/core.c | 15 ++++++++-------
>  1 file changed, 8 insertions(+), 7 deletions(-)
>
> diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
> index 2405e39d800f..dc3c90992f33 100644
> --- a/kernel/bpf/core.c
> +++ b/kernel/bpf/core.c
> @@ -59,6 +59,9 @@
>  #define CTX    regs[BPF_REG_CTX]
>  #define IMM    insn->imm
>
> +static u64 (*interpreters_args[])(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5,
> +                                 const struct bpf_insn *insn);
> +
>  /* No hurry in this branch
>   *
>   * Exported for the bpf jit load helper.
> @@ -1560,10 +1563,10 @@ static u64 ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn)
>                 CONT;
>
>         JMP_CALL_ARGS:
> -               BPF_R0 = (__bpf_call_base_args + insn->imm)(BPF_R1, BPF_R2,
> -                                                           BPF_R3, BPF_R4,
> -                                                           BPF_R5,
> -                                                           insn + insn->off + 1);
> +               BPF_R0 = (interpreters_args[insn->off])(BPF_R1, BPF_R2,
> +                                                       BPF_R3, BPF_R4,
> +                                                       BPF_R5,
> +                                                       insn + insn->imm + 1);
>                 CONT;
>
>         JMP_TAIL_CALL: {
> @@ -1810,9 +1813,7 @@ EVAL4(PROG_NAME_LIST, 416, 448, 480, 512)
>  void bpf_patch_call_args(struct bpf_insn *insn, u32 stack_depth)
>  {
>         stack_depth = max_t(u32, stack_depth, 1);
> -       insn->off = (s16) insn->imm;
> -       insn->imm = interpreters_args[(round_up(stack_depth, 32) / 32) - 1] -
> -               __bpf_call_base_args;
> +       insn->off = (round_up(stack_depth, 32) / 32) - 1;
>         insn->code = BPF_JMP | BPF_CALL_ARGS;

Neat. Please add a test case and resubmit.