linux-kernel - Re: [PATCH bpf-next] bpf, x86: inline bpf_get_current_task() for x86

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <0f0bd124a42723acf87b427cc69356a0e4b1e339.camel@gmail.com>
Date: Mon, 29 Dec 2025 11:58:03 -0800
From: Eduard Zingerman <eddyz87@...il.com>
To: Menglong Dong <menglong8.dong@...il.com>, ast@...nel.org
Cc: davem@...emloft.net, dsahern@...nel.org, daniel@...earbox.net, 
	andrii@...nel.org, martin.lau@...ux.dev, song@...nel.org,
 yonghong.song@...ux.dev, 	john.fastabend@...il.com, kpsingh@...nel.org,
 sdf@...ichev.me, haoluo@...gle.com, 	jolsa@...nel.org, tglx@...utronix.de,
 mingo@...hat.com, bp@...en8.de, 	dave.hansen@...ux.intel.com,
 jiang.biao@...ux.dev, x86@...nel.org, hpa@...or.com, 
	netdev@...r.kernel.org, bpf@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH bpf-next] bpf, x86: inline bpf_get_current_task() for
 x86_64

On Thu, 2025-12-25 at 18:44 +0800, Menglong Dong wrote:
> Inline bpf_get_current_task() and bpf_get_current_task_btf() for x86_64
> to obtain better performance. The instruction we use here is:
> 
>   65 48 8B 04 25 [offset] // mov rax, gs:[offset]
> 
> Not sure if there is any side effect here.
> 
> Signed-off-by: Menglong Dong <dongml2@...natelecom.cn>
> ---

The change makes sense to me.
Could you please address the compilation error reported by kernel test robot?
Could you please also add a tests case using __jited annotation like
in verifier_ldsx.c?

>  arch/x86/net/bpf_jit_comp.c | 33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
> 
> diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c
> index b69dc7194e2c..7f38481816f0 100644
> --- a/arch/x86/net/bpf_jit_comp.c
> +++ b/arch/x86/net/bpf_jit_comp.c
> @@ -1300,6 +1300,19 @@ static void emit_st_r12(u8 **pprog, u32 size, u32 dst_reg, int off, int imm)
>  	emit_st_index(pprog, size, dst_reg, X86_REG_R12, off, imm);
>  }
>  
> +static void emit_ldx_percpu_r0(u8 **pprog, const void __percpu *ptr)
> +{
> +	u8 *prog = *pprog;
> +
> +	/* mov rax, gs:[offset] */
> +	EMIT2(0x65, 0x48);
> +	EMIT2(0x8B, 0x04);
> +	EMIT1(0x25);
> +	EMIT((u32)(unsigned long)ptr, 4);
> +
> +	*pprog = prog;
> +}
> +
>  static int emit_atomic_rmw(u8 **pprog, u32 atomic_op,
>  			   u32 dst_reg, u32 src_reg, s16 off, u8 bpf_size)
>  {
> @@ -2435,6 +2448,15 @@ st:			if (is_imm8(insn->off))
>  		case BPF_JMP | BPF_CALL: {
>  			u8 *ip = image + addrs[i - 1];
>  
> +			if (insn->src_reg == 0 && (insn->imm == BPF_FUNC_get_current_task ||
> +						   insn->imm == BPF_FUNC_get_current_task_btf)) {
> +				if (IS_ENABLED(CONFIG_USE_X86_SEG_SUPPORT))
> +					emit_ldx_percpu_r0(&prog, &const_current_task);
> +				else
> +					emit_ldx_percpu_r0(&prog, &current_task);

Nit: arch/x86/include/asm/current.h says that current_task and const_current_task are aliases.
     In that case, why would we need two branches here?

> +				break;
> +			}
> +
>  			func = (u8 *) __bpf_call_base + imm32;
>  			if (src_reg == BPF_PSEUDO_CALL && tail_call_reachable) {
>  				LOAD_TAIL_CALL_CNT_PTR(stack_depth);
> @@ -4067,3 +4089,14 @@ bool bpf_jit_supports_timed_may_goto(void)
>  {
>  	return true;
>  }
> +
> +bool bpf_jit_inlines_helper_call(s32 imm)
> +{
> +	switch (imm) {
> +	case BPF_FUNC_get_current_task:
> +	case BPF_FUNC_get_current_task_btf:
> +		return true;
> +	default:
> +		return false;
> +	}
> +}