Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH] loongarch/bpf: Fix bpf load failed with
 CONFIG_BPF_JIT_ALWAYS_ON, caused by jit (BPF_ST | BPF_NOSPEC) code
To:     George Guo <guodongtai@kylinos.cn>, chenhuacai@kernel.org,
        masahiroy@kernel.org, michal.lkml@markovi.net
Cc:     kernel@xen0n.name, ndesaulniers@google.com, ast@kernel.org,
        loongarch@lists.linux.dev, linux-kernel@vger.kernel.org,
        linux-kbuild@vger.kernel.org, bpf@vger.kernel.org
References: <20230326044019.2139628-1-guodongtai@kylinos.cn>
From:   Daniel Borkmann <daniel@iogearbox.net>
Message-ID: <c1932d0d-cf3f-5005-958d-7e08dddf42c9@iogearbox.net>
Date:   Mon, 27 Mar 2023 11:29:48 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.7.2
MIME-Version: 1.0
In-Reply-To: <20230326044019.2139628-1-guodongtai@kylinos.cn>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 3/26/23 6:40 AM, George Guo wrote:
> Here just skip the code(BPF_ST | BPF_NOSPEC) that has no couterpart to the loongarch.
> 
> To verify, use ltp testcase:
> 
> Without this patch:
> $ ./bpf_prog02
> ... ...
> bpf_common.c:123: TBROK: Failed verification: ??? (524)
> 
> Summary:
> passed   0
> failed   0
> broken   1
> skipped  0
> warnings 0
> 
> With this patch:
> $ ./bpf_prog02
> ... ...
> Summary:
> passed   0
> failed   0
> broken   0
> skipped  0
> warnings 0
> 
> Signed-off-by: George Guo <guodongtai@kylinos.cn>
> ---
>   arch/loongarch/net/bpf_jit.c | 5 +++++
>   1 file changed, 5 insertions(+)
> 
> diff --git a/arch/loongarch/net/bpf_jit.c b/arch/loongarch/net/bpf_jit.c
> index 288003a9f0ca..745d344385ed 100644
> --- a/arch/loongarch/net/bpf_jit.c
> +++ b/arch/loongarch/net/bpf_jit.c
> @@ -1046,6 +1046,11 @@ static int build_body(struct jit_ctx *ctx, bool extra_pass)
>   		if (ctx->image == NULL)
>   			ctx->offset[i] = ctx->idx;
>   
> +		/* skip the code that has no couterpart to the host arch */
> +		if(insn->code == (BPF_ST | BPF_NOSPEC)) {
> +			continue;
> +		}

Small nit, but could we align with other JIT implementations and place it into similar
location for consistency? Above looks a bit out of place and it should really be part
of build_insn.

diff --git a/arch/loongarch/net/bpf_jit.c b/arch/loongarch/net/bpf_jit.c
index 288003a9f0ca..d586df48ecc6 100644
--- a/arch/loongarch/net/bpf_jit.c
+++ b/arch/loongarch/net/bpf_jit.c
@@ -1022,6 +1022,10 @@ static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx, bool ext
                 emit_atomic(insn, ctx);
                 break;

+       /* Speculation barrier */
+       case BPF_ST | BPF_NOSPEC:
+               break;
+
         default:
                 pr_err("bpf_jit: unknown opcode %02x\n", code);
                 return -EINVAL;