Linux-RISC-V Archive on lore.kernel.org
 help / color / Atom feed
From: Palmer Dabbelt <palmerdabbelt@google.com>
To: Bjorn Topel <bjorn.topel@gmail.com>
Cc: daniel@iogearbox.net, netdev@vger.kernel.org, ast@kernel.org,
	lukenels@cs.washington.edu, Bjorn Topel <bjorn.topel@gmail.com>,
	linux-riscv@lists.infradead.org, bpf@vger.kernel.org,
	xi.wang@gmail.com
Subject: Re: [PATCH bpf-next v2 4/9] riscv, bpf: add support for far jumps and exits
Date: Mon, 23 Dec 2019 10:18:32 -0800 (PST)
Message-ID: <mhng-c035e490-421b-4df0-9875-ec3059b8f749@palmerdabbelt-glaptop> (raw)
In-Reply-To: <20191216091343.23260-5-bjorn.topel@gmail.com>

On Mon, 16 Dec 2019 01:13:38 PST (-0800), Bjorn Topel wrote:
> This commit add support for far (offset > 21b) jumps and exits.
>
> Reviewed-by: Luke Nelson <lukenels@cs.washington.edu>
> Cc: Xi Wang <xi.wang@gmail.com>
> Signed-off-by: Björn Töpel <bjorn.topel@gmail.com>
> ---
>  arch/riscv/net/bpf_jit_comp.c | 37 ++++++++++++++++-------------------
>  1 file changed, 17 insertions(+), 20 deletions(-)
>
> diff --git a/arch/riscv/net/bpf_jit_comp.c b/arch/riscv/net/bpf_jit_comp.c
> index c38c95df3440..2fc0f24ad30f 100644
> --- a/arch/riscv/net/bpf_jit_comp.c
> +++ b/arch/riscv/net/bpf_jit_comp.c
> @@ -496,16 +496,6 @@ static int is_12b_check(int off, int insn)
>  	return 0;
>  }
>
> -static int is_21b_check(int off, int insn)
> -{
> -	if (!is_21b_int(off)) {
> -		pr_err("bpf-jit: insn=%d 21b < offset=%d not supported yet!\n",
> -		       insn, (int)off);
> -		return -1;
> -	}
> -	return 0;
> -}
> -
>  static void emit_imm(u8 rd, s64 val, struct rv_jit_context *ctx)
>  {
>  	/* Note that the immediate from the add is sign-extended,
> @@ -820,6 +810,21 @@ static void emit_sext_32_rd(u8 *rd, struct rv_jit_context *ctx)
>  	*rd = RV_REG_T2;
>  }
>
> +static void emit_jump_and_link(u8 rd, int rvoff, struct rv_jit_context *ctx)
> +{
> +	s64 upper, lower;
> +
> +	if (is_21b_int(rvoff)) {
> +		emit(rv_jal(rd, rvoff >> 1), ctx);
> +		return;
> +	}
> +
> +	upper = (rvoff + (1 << 11)) >> 12;
> +	lower = rvoff & 0xfff;
> +	emit(rv_auipc(RV_REG_T1, upper), ctx);
> +	emit(rv_jalr(rd, RV_REG_T1, lower), ctx);
> +}

What constrains these jumps to always be 32-bit PC relative?  We have some
issues in the module loader with references to kernel symbols being too far
away to the loaded modules, it seems like similar issues could creep in here.

>  static bool is_signed_bpf_cond(u8 cond)
>  {
>  	return cond == BPF_JSGT || cond == BPF_JSLT ||
> @@ -1101,13 +1106,7 @@ static int emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx,
>  	/* JUMP off */
>  	case BPF_JMP | BPF_JA:
>  		rvoff = rv_offset(i, off, ctx);
> -		if (!is_21b_int(rvoff)) {
> -			pr_err("bpf-jit: insn=%d offset=%d not supported yet!\n",
> -			       i, rvoff);
> -			return -1;
> -		}
> -
> -		emit(rv_jal(RV_REG_ZERO, rvoff >> 1), ctx);
> +		emit_jump_and_link(RV_REG_ZERO, rvoff, ctx);
>  		break;
>
>  	/* IF (dst COND src) JUMP off */
> @@ -1245,9 +1244,7 @@ static int emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx,
>  			break;
>
>  		rvoff = epilogue_offset(ctx);
> -		if (is_21b_check(rvoff, i))
> -			return -1;
> -		emit(rv_jal(RV_REG_ZERO, rvoff >> 1), ctx);
> +		emit_jump_and_link(RV_REG_ZERO, rvoff, ctx);
>  		break;
>
>  	/* dst = imm64 */


  parent reply index

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-12-16  9:13 [PATCH bpf-next v2 0/9] riscv: BPF JIT fix, optimizations and far jumps support Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 1/9] riscv, bpf: fix broken BPF tail calls Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 2/9] riscv, bpf: add support for far branching Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 3/9] riscv, bpf: add support for far branching when emitting tail call Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 4/9] riscv, bpf: add support for far jumps and exits Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 5/9] riscv, bpf: optimize BPF tail calls Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 6/9] riscv, bpf: provide RISC-V specific JIT image alloc/free Björn Töpel
2019-12-16 15:09   ` Daniel Borkmann
2019-12-18  6:23     ` Björn Töpel
2020-01-04  1:32       ` Paul Walmsley
2020-01-07 10:24         ` Björn Töpel
2020-01-07 10:47           ` Paul Walmsley
2020-02-02 13:37   ` Alex Ghiti
2020-02-03 12:28     ` Björn Töpel
2020-02-03 20:57       ` Alex Ghiti
2019-12-16  9:13 ` [PATCH bpf-next v2 7/9] riscv, bpf: optimize calls Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 8/9] riscv, bpf: add missing uapi header for BPF_PROG_TYPE_PERF_EVENT programs Björn Töpel
2019-12-16  9:13 ` [PATCH bpf-next v2 9/9] riscv, perf: add arch specific perf_arch_bpf_user_pt_regs Björn Töpel
2019-12-19 15:07 ` [PATCH bpf-next v2 0/9] riscv: BPF JIT fix, optimizations and far jumps support Daniel Borkmann
2019-12-19 22:02 ` [PATCH bpf-next v2 1/9] riscv, bpf: fix broken BPF tail calls Palmer Dabbelt
2019-12-23 18:03 ` [PATCH bpf-next v2 2/9] riscv, bpf: add support for far branching Palmer Dabbelt
2020-01-07  8:13   ` Björn Töpel
2020-01-23  2:08   ` Palmer Dabbelt
2019-12-23 18:18 ` [PATCH bpf-next v2 3/9] riscv, bpf: add support for far branching when emitting tail call Palmer Dabbelt
2019-12-23 18:18 ` Palmer Dabbelt [this message]
2019-12-23 18:29 ` [PATCH bpf-next v2 5/9] riscv, bpf: optimize BPF tail calls Palmer Dabbelt
2019-12-23 18:30 ` [PATCH bpf-next v2 6/9] riscv, bpf: provide RISC-V specific JIT image alloc/free Palmer Dabbelt
2019-12-23 18:58 ` [PATCH bpf-next v2 7/9] riscv, bpf: optimize calls Palmer Dabbelt
2020-01-07 10:14   ` Björn Töpel
2020-01-28  2:15   ` Palmer Dabbelt
2020-02-03 12:11     ` Björn Töpel
2019-12-23 18:58 ` [PATCH bpf-next v2 8/9] riscv, bpf: add missing uapi header for BPF_PROG_TYPE_PERF_EVENT programs Palmer Dabbelt
2019-12-23 18:58 ` [PATCH bpf-next v2 9/9] riscv, perf: add arch specific perf_arch_bpf_user_pt_regs Palmer Dabbelt

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=mhng-c035e490-421b-4df0-9875-ec3059b8f749@palmerdabbelt-glaptop \
    --to=palmerdabbelt@google.com \
    --cc=ast@kernel.org \
    --cc=bjorn.topel@gmail.com \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=linux-riscv@lists.infradead.org \
    --cc=lukenels@cs.washington.edu \
    --cc=netdev@vger.kernel.org \
    --cc=xi.wang@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Linux-RISC-V Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-riscv/0 linux-riscv/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-riscv linux-riscv/ https://lore.kernel.org/linux-riscv \
		linux-riscv@lists.infradead.org
	public-inbox-index linux-riscv

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.infradead.lists.linux-riscv


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git