[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAEf4BzY+zO5Eu6i3KUoWRhhXbKA3zsKFxAhL5txNMkrgzCL-hQ@mail.gmail.com>
Date: Thu, 20 Jun 2024 12:07:40 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Puranjay Mohan <puranjay@...nel.org>
Cc: Alexei Starovoitov <ast@...nel.org>, Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>, Martin KaFai Lau <martin.lau@...ux.dev>,
Eduard Zingerman <eddyz87@...il.com>, Song Liu <song@...nel.org>,
Yonghong Song <yonghong.song@...ux.dev>, John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>, Stanislav Fomichev <sdf@...gle.com>, Hao Luo <haoluo@...gle.com>,
Jiri Olsa <jolsa@...nel.org>, Xu Kuohai <xukuohai@...weicloud.com>,
Catalin Marinas <catalin.marinas@....com>, Will Deacon <will@...nel.org>, bpf@...r.kernel.org,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
puranjay12@...il.com
Subject: Re: [PATCH] bpf, arm64: inline bpf_get_current_task/_btf() helpers
On Wed, Jun 19, 2024 at 6:25 AM Puranjay Mohan <puranjay@...nel.org> wrote:
>
> On ARM64, the pointer to task_struct is always available in the sp_el0
> register and therefore the calls to bpf_get_current_task() and
> bpf_get_current_task_btf() can be inlined into a single MRS instruction.
>
> Here is the difference before and after this change:
>
> Before:
>
> ; struct task_struct *task = bpf_get_current_task_btf();
> 54: mov x10, #0xffffffffffff7978 // #-34440
> 58: movk x10, #0x802b, lsl #16
> 5c: movk x10, #0x8000, lsl #32
> 60: blr x10 --------------> 0xffff8000802b7978 <+0>: mrs x0, sp_el0
> 64: add x7, x0, #0x0 <-------------- 0xffff8000802b797c <+4>: ret
>
> After:
>
> ; struct task_struct *task = bpf_get_current_task_btf();
> 54: mrs x7, sp_el0
>
> This shows around 1% performance improvement in artificial microbenchmark.
>
> Signed-off-by: Puranjay Mohan <puranjay@...nel.org>
> ---
> arch/arm64/net/bpf_jit_comp.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
These are frequently used helpers, similarly to
get_smp_processor_id(), so I think it makes sense to optimize them.
Acked-by: Andrii Nakryiko <andrii@...nel.org>
> diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
> index 720336d28856..b838dab3bd26 100644
> --- a/arch/arm64/net/bpf_jit_comp.c
> +++ b/arch/arm64/net/bpf_jit_comp.c
> @@ -1244,6 +1244,13 @@ static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx,
> break;
> }
>
> + /* Implement helper call to bpf_get_current_task/_btf() inline */
> + if (insn->src_reg == 0 && (insn->imm == BPF_FUNC_get_current_task ||
> + insn->imm == BPF_FUNC_get_current_task_btf)) {
> + emit(A64_MRS_SP_EL0(r0), ctx);
> + break;
> + }
> +
> ret = bpf_jit_get_func_addr(ctx->prog, insn, extra_pass,
> &func_addr, &func_addr_fixed);
> if (ret < 0)
> @@ -2581,6 +2588,8 @@ bool bpf_jit_inlines_helper_call(s32 imm)
> {
> switch (imm) {
> case BPF_FUNC_get_smp_processor_id:
> + case BPF_FUNC_get_current_task:
> + case BPF_FUNC_get_current_task_btf:
> return true;
> default:
> return false;
> --
> 2.40.1
>
Powered by blists - more mailing lists