Received-SPF: pass (google.com: domain of linux-kernel+bounces-133447-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Precedence: bulk
MIME-Version: 1.0
References: <20240405124348.27644-1-puranjay12@gmail.com>
In-Reply-To: <20240405124348.27644-1-puranjay12@gmail.com>
From: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Date: Fri, 5 Apr 2024 11:04:22 -0700
Message-ID: <CAEf4BzZ2Tz5-GwbQKYg7KoGwqN8ewPBakmghHaH20MfoATe74g@mail.gmail.com>
Subject: Re: [PATCH bpf-next] riscv, bpf: add internal-only MOV instruction to
 resolve per-CPU addrs
To: Puranjay Mohan <puranjay12@gmail.com>
Cc: Alexei Starovoitov <ast@kernel.org>, Daniel Borkmann <daniel@iogearbox.net>, 
	Andrii Nakryiko <andrii@kernel.org>, Martin KaFai Lau <martin.lau@linux.dev>, 
	Eduard Zingerman <eddyz87@gmail.com>, Song Liu <song@kernel.org>, 
	Yonghong Song <yonghong.song@linux.dev>, John Fastabend <john.fastabend@gmail.com>, 
	KP Singh <kpsingh@kernel.org>, Stanislav Fomichev <sdf@google.com>, Hao Luo <haoluo@google.com>, 
	Jiri Olsa <jolsa@kernel.org>, =?UTF-8?B?QmrDtnJuIFTDtnBlbA==?= <bjorn@kernel.org>, 
	Paul Walmsley <paul.walmsley@sifive.com>, Palmer Dabbelt <palmer@dabbelt.com>, 
	Albert Ou <aou@eecs.berkeley.edu>, bpf@vger.kernel.org, 
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org, 
	Pu Lehui <pulehui@huawei.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Apr 5, 2024 at 5:44=E2=80=AFAM Puranjay Mohan <puranjay12@gmail.com=
> wrote:
>
> Support an instruction for resolving absolute addresses of per-CPU
> data from their per-CPU offsets. This instruction is internal-only and
> users are not allowed to use them directly. They will only be used for
> internal inlining optimizations for now between BPF verifier and BPF
> JITs.
>
> RISC-V uses generic per-cpu implementation where the offsets for CPUs
> are kept in an array called __per_cpu_offset[cpu_number]. RISCV stores
> the address of the task_struct in TP register. The first element in
> tast_struct is struct thread_info, and we can get the cpu number by
> reading from the TP register + offsetof(struct thread_info, cpu).
>
> Once we have the cpu number in a register we read the offset for that
> cpu from address: &__per_cpu_offset + cpu_number << 3. Then we add this
> offset to the destination register.
>
> To measure the improvement from this change, the benchmark in [1] was
> used on Qemu:
>
> Before:
> glob-arr-inc   :    1.127 =C2=B1 0.013M/s
> arr-inc        :    1.121 =C2=B1 0.004M/s
> hash-inc       :    0.681 =C2=B1 0.052M/s
>
> After:
> glob-arr-inc   :    1.138 =C2=B1 0.011M/s
> arr-inc        :    1.366 =C2=B1 0.006M/s
> hash-inc       :    0.676 =C2=B1 0.001M/s

TBH, I don't trust benchmarks done inside QEMU. Can you try running
this on some real hardware?

>
> [1] https://github.com/anakryiko/linux/commit/8dec900975ef
>
> Signed-off-by: Puranjay Mohan <puranjay12@gmail.com>
> ---
>  arch/riscv/net/bpf_jit_comp64.c | 24 ++++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
>
> diff --git a/arch/riscv/net/bpf_jit_comp64.c b/arch/riscv/net/bpf_jit_com=
p64.c
> index 15e482f2c657..e95bd1d459a4 100644
> --- a/arch/riscv/net/bpf_jit_comp64.c
> +++ b/arch/riscv/net/bpf_jit_comp64.c
> @@ -12,6 +12,7 @@
>  #include <linux/stop_machine.h>
>  #include <asm/patch.h>
>  #include <asm/cfi.h>
> +#include <asm/percpu.h>
>  #include "bpf_jit.h"
>
>  #define RV_FENTRY_NINSNS 2
> @@ -1089,6 +1090,24 @@ int bpf_jit_emit_insn(const struct bpf_insn *insn,=
 struct rv_jit_context *ctx,
>                         emit_or(RV_REG_T1, rd, RV_REG_T1, ctx);
>                         emit_mv(rd, RV_REG_T1, ctx);
>                         break;
> +               } else if (insn_is_mov_percpu_addr(insn)) {
> +                       if (rd !=3D rs)
> +                               emit_mv(rd, rs, ctx);

Is this an unconditional move instruction? in x86-64, EMIT_mov checks
whether source and destination registers are the same and doesn't emit
anything if they match (which makes sense, right)?

> +#ifdef CONFIG_SMP
> +                               /* Load current CPU number in T1 */
> +                               emit_ld(RV_REG_T1, offsetof(struct thread=
_info, cpu), RV_REG_TP,
> +                                       ctx);

nit: maybe keep this on the same line?

> +                               /* << 3 because offsets are 8 bytes */
> +                               emit_slli(RV_REG_T1, RV_REG_T1, 3, ctx);
> +                               /* Load address of __per_cpu_offset array=
 in T2 */
> +                               emit_imm(RV_REG_T2, (u64)&__per_cpu_offse=
t, ctx);
> +                               /* Add offset of current CPU to  __per_cp=
u_offset */
> +                               emit_add(RV_REG_T1, RV_REG_T2, RV_REG_T1,=
 ctx);
> +                               /* Load __per_cpu_offset[cpu] in T1 */
> +                               emit_ld(RV_REG_T1, 0, RV_REG_T1, ctx);
> +                               /* Add the offset to Rd */
> +                               emit_add(rd, rd, RV_REG_T1, ctx);
> +#endif
>                 }
>                 if (imm =3D=3D 1) {
>                         /* Special mov32 for zext */
> @@ -2038,3 +2057,8 @@ bool bpf_jit_supports_arena(void)
>  {
>         return true;
>  }
> +
> +bool bpf_jit_supports_percpu_insn(void)
> +{
> +       return true;
> +}
> --
> 2.40.1
>