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: Avoid declare variables in switch-case
To:     Huacai Chen <chenhuacai@kernel.org>,
        WANG Xuerui <kernel@xen0n.name>
References: <20221013154000.3462836-1-chenhuacai@loongson.cn>
 <8a8fa581-94a9-649d-8c01-f1afd4bc9514@xen0n.name>
 <CAAhV-H5U1c_wfWLuxMaHD6c9-k+g-iSqgtcJVwceoL13J7hEiA@mail.gmail.com>
Cc:     Huacai Chen <chenhuacai@loongson.cn>, loongarch@lists.linux.dev,
        Xuefeng Li <lixuefeng@loongson.cn>,
        Guo Ren <guoren@kernel.org>,
        Jiaxun Yang <jiaxun.yang@flygoat.com>,
        linux-kernel@vger.kernel.org
From:   Tiezhu Yang <yangtiezhu@loongson.cn>
Message-ID: <ae6260e5-0eb6-a615-7032-6481cd186f3f@loongson.cn>
Date:   Fri, 14 Oct 2022 10:18:38 +0800
User-Agent: Mozilla/5.0 (X11; Linux mips64; rv:45.0) Gecko/20100101
 Thunderbird/45.4.0
MIME-Version: 1.0
In-Reply-To: <CAAhV-H5U1c_wfWLuxMaHD6c9-k+g-iSqgtcJVwceoL13J7hEiA@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 10/14/2022 09:13 AM, Huacai Chen wrote:
> Hi, Xuerui,
>
> On Fri, Oct 14, 2022 at 12:43 AM WANG Xuerui <kernel@xen0n.name> wrote:
>>
>> On 10/13/22 23:40, Huacai Chen wrote:
>>> Not all compilers support declare variables in switch-case, so move
>>> declarations to the beginning of a function. Otherwise we may get such
>>> build errors:

...

>>>
>>>   static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx, bool extra_pass)
>>>   {
>>> -     const bool is32 = BPF_CLASS(insn->code) == BPF_ALU ||
>>> -                       BPF_CLASS(insn->code) == BPF_JMP32;
>>> +     u8 t0 = -1;
>> Here "t0" seems to be a versatile temp value, while the "t1" below is
>> the actual GPR $t1. What about renaming "t0" to something like "tmp" to
>> reduce confusion? I believe due to things like "t0 = LOONGARCH_GPR_ZERO"
>> the "t0" is 100% not an actual mapping to $t0.
> I rename t7 to t0 because there is no t3-t6, t7 looks very strange.
> But from emit_cond_jmp() the 3rd and 4th parameters have no difference
> so I suppose t0 is just OK, then whether rename it to tmp depends on
> Tiezhu's opinion.
>

Use "tmp" seems better due to it is a temp value.

>>> +     u64 func_addr;
>>> +     bool func_addr_fixed;
>>> +     int i = insn - ctx->prog->insnsi;
>>> +     int ret, jmp_offset;
>>>       const u8 code = insn->code;
>>>       const u8 cond = BPF_OP(code);
>>>       const u8 t1 = LOONGARCH_GPR_T1;
>>> @@ -400,8 +402,8 @@ static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx, bool ext
>>>       const u8 dst = regmap[insn->dst_reg];
>>>       const s16 off = insn->off;
>>>       const s32 imm = insn->imm;
>>> -     int jmp_offset;
>>> -     int i = insn - ctx->prog->insnsi;
>>> +     const u64 imm64 = (u64)(insn + 1)->imm << 32 | (u32)insn->imm;
>>> +     const bool is32 = BPF_CLASS(insn->code) == BPF_ALU || BPF_CLASS(insn->code) == BPF_JMP32;
>> Please consider reducing diff damage and not touching parts not directly
>> affected by this change. For example this "is32" declaration and
>> initialization was moved although not related to this change.

It looks reasonable, one change per patch is better.

> I think defining variables from simple to complex and grouping them
> can make life easier. :)
>

No strong opinion on this, I am OK either way.

Thanks,
Tiezhu