[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <9830ef1c-d9cc-46ce-b292-07c8222d72ac@huawei.com>
Date: Sun, 8 Feb 2026 09:58:03 +0800
From: Pu Lehui <pulehui@...wei.com>
To: Menglong Dong <menglong.dong@...ux.dev>, Menglong Dong
<menglong8.dong@...il.com>
CC: <bjorn@...nel.org>, <ast@...nel.org>, <daniel@...earbox.net>,
<andrii@...nel.org>, <martin.lau@...ux.dev>, <eddyz87@...il.com>,
<song@...nel.org>, <yonghong.song@...ux.dev>, <john.fastabend@...il.com>,
<kpsingh@...nel.org>, <sdf@...ichev.me>, <haoluo@...gle.com>,
<jolsa@...nel.org>, <puranjay@...nel.org>, <pjw@...nel.org>,
<palmer@...belt.com>, <aou@...s.berkeley.edu>, <alex@...ti.fr>,
<bpf@...r.kernel.org>, <linux-riscv@...ts.infradead.org>,
<linux-kernel@...r.kernel.org>, <jiang.biao@...ux.dev>
Subject: Re: [PATCH bpf-next v3 1/3] bpf, riscv: introduce
emit_store_stack_imm64() for trampoline
On 2026/2/7 18:53, Menglong Dong wrote:
> On 2026/2/7 09:13, Pu Lehui wrote:
>>
>> On 2026/2/6 20:20, Menglong Dong wrote:
>>> Introduce a helper to store 64-bit immediate on the trampoline stack with
>>> a help of a register.
>>>
>>> Signed-off-by: Menglong Dong <dongml2@...natelecom.cn>
>>> Tested-by: Björn Töpel <bjorn@...nel.org>
>>> Acked-by: Björn Töpel <bjorn@...nel.org>
>>> ---
>>> arch/riscv/net/bpf_jit_comp64.c | 25 ++++++++++++++-----------
>>> 1 file changed, 14 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/arch/riscv/net/bpf_jit_comp64.c b/arch/riscv/net/bpf_jit_comp64.c
>>> index 37888abee70c..e4f45e2e7e2f 100644
>>> --- a/arch/riscv/net/bpf_jit_comp64.c
>>> +++ b/arch/riscv/net/bpf_jit_comp64.c
>>> @@ -926,6 +926,14 @@ static void restore_stack_args(int nr_stack_args, int args_off, int stk_arg_off,
>>> }
>>> }
>>>
>>> +static void emit_store_stack_imm64(u8 reg, int stack_off, u64 imm64,
>>> + struct rv_jit_context *ctx)
>>
>> Some nit. The first parameter can be removed by using a fixed RV_REG_T1.
>> Also, placing imm64 before stack_off might looks better.
>
> Hi, Lehui. When I implement the emit_store_stack_imm64() in x86,
> Andrii suggested that we'd better use the register explicitly to indicate
> the register is used. So maybe it's better to keep this part still?
No problem, the impact is minimal. And in that case, just leave imm64
and stack_off as they are. :)
>
> I can place the imm64 before stack_off.
>
> Thanks!
> Menglong Dong
>
>>
>>> +{
>>> + /* Load imm64 into reg and store it at [FP + stack_off]. */
>>> + emit_imm(reg, (s64)imm64, ctx);
>>> + emit_sd(RV_REG_FP, stack_off, reg, ctx);
>>> +}
>>> +
>>> static int invoke_bpf_prog(struct bpf_tramp_link *l, int args_off, int retval_off,
>>> int run_ctx_off, bool save_ret, struct rv_jit_context *ctx)
>>> {
>>> @@ -933,12 +941,10 @@ static int invoke_bpf_prog(struct bpf_tramp_link *l, int args_off, int retval_of
>>> struct bpf_prog *p = l->link.prog;
>>> int cookie_off = offsetof(struct bpf_tramp_run_ctx, bpf_cookie);
>>>
>>> - if (l->cookie) {
>>> - emit_imm(RV_REG_T1, l->cookie, ctx);
>>> - emit_sd(RV_REG_FP, -run_ctx_off + cookie_off, RV_REG_T1, ctx);
>>> - } else {
>>> + if (l->cookie)
>>> + emit_store_stack_imm64(RV_REG_T1, -run_ctx_off + cookie_off, l->cookie, ctx);
>>> + else
>>> emit_sd(RV_REG_FP, -run_ctx_off + cookie_off, RV_REG_ZERO, ctx);
>>> - }
>>>
>>> /* arg1: prog */
>>> emit_imm(RV_REG_A0, (const s64)p, ctx);
>>> @@ -1123,13 +1129,10 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
>>> emit_sd(RV_REG_FP, -sreg_off, RV_REG_S1, ctx);
>>>
>>> /* store ip address of the traced function */
>>> - if (flags & BPF_TRAMP_F_IP_ARG) {
>>> - emit_imm(RV_REG_T1, (const s64)func_addr, ctx);
>>> - emit_sd(RV_REG_FP, -ip_off, RV_REG_T1, ctx);
>>> - }
>>> + if (flags & BPF_TRAMP_F_IP_ARG)
>>> + emit_store_stack_imm64(RV_REG_T1, -ip_off, (u64)func_addr, ctx); >
>>> - emit_li(RV_REG_T1, nr_arg_slots, ctx);
>>> - emit_sd(RV_REG_FP, -nregs_off, RV_REG_T1, ctx);
>>> + emit_store_stack_imm64(RV_REG_T1, -nregs_off, nr_arg_slots, ctx);
>>>
>>> store_args(nr_arg_slots, args_off, ctx);
>>>
>>
>>
>
>
>
>
>
>
Powered by blists - more mailing lists