Re: [PATCH] riscv: bpf: Fix uninitialized symbol 'retval_off'

From: Chenghao Duan <duanchenghao@kylinos.cn>
To: Pu Lehui <pulehui@huawei.com>
Cc: ast@kernel.org, bjorn@kernel.org, puranjay@kernel.org,
	paul.walmsley@sifive.com, palmer@dabbelt.com,
	aou@eecs.berkeley.edu, daniel@iogearbox.net, andrii@kernel.org,
	martin.lau@linux.dev, eddyz87@gmail.com, song@kernel.org,
	yonghong.song@linux.dev, john.fastabend@gmail.com,
	kpsingh@kernel.org, sdf@fomichev.me, haoluo@google.com,
	jolsa@kernel.org, alex@ghiti.fr, bpf@vger.kernel.org,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] riscv: bpf: Fix uninitialized symbol 'retval_off'
Date: Wed, 20 Aug 2025 18:35:30 +0800	[thread overview]
Message-ID: <20250820103530.GA1475460@chenghao-pc> (raw)
In-Reply-To: <239193b7-7dab-45b0-ab13-06bfe3f96f22@huawei.com>

On Wed, Aug 20, 2025 at 06:10:07PM +0800, Pu Lehui wrote:
> 
> 
> On 2025/8/20 17:26, Chenghao Duan wrote:
> > On Wed, Aug 20, 2025 at 02:52:01PM +0800, Pu Lehui wrote:
> > > 
> > > 
> > > On 2025/8/20 14:25, Chenghao Duan wrote:
> > > > In __arch_prepare_bpf_trampoline(), retval_off is only meaningful when
> > > > save_ret is true, so the current logic is correct. However, in the
> > > 
> > > lgtm, and same for `ip_off`, pls patch it together.
> > 
> > I also checked at the time that ip_off is only initialized and assigned
> > when flags & BPF_TRAMP_F_IP_ARG is true. However, I noticed that the use
> > of ip_off also requires this condition, so the compiler did not issue a
> > warning.
> > 
> > Chenghao
> > 
> > > 
> > > > original logic, retval_off is only initialized under certain
> 
> Can you show how to replay this warning? I guess the warning path is as
> follow. Compiler didn't know fmod_ret prog need BPF_TRAMP_F_CALL_ORIG.
> 
> ```
> if (fmod_ret->nr_links) {
> 	...
> 	emit_sd(RV_REG_FP, -retval_off, RV_REG_ZERO, ctx);
> }
> ```
> 

Exactly, the compiler sees the unconditional use of retval_off.

Chenghao

> > > > conditions, which may cause a build warning.
> > > > 
> > > > So initialize retval_off unconditionally to fix it.
> > > > 
> > > > Signed-off-by: Chenghao Duan <duanchenghao@kylinos.cn>
> > > > ---
> > > >    arch/riscv/net/bpf_jit_comp64.c | 5 ++---
> > > >    1 file changed, 2 insertions(+), 3 deletions(-)
> > > > 
> > > > diff --git a/arch/riscv/net/bpf_jit_comp64.c b/arch/riscv/net/bpf_jit_comp64.c
> > > > index 10e01ff06312..49bbda8372b0 100644
> > > > --- a/arch/riscv/net/bpf_jit_comp64.c
> > > > +++ b/arch/riscv/net/bpf_jit_comp64.c
> > > > @@ -1079,10 +1079,9 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
> > > >    	stack_size += 16;
> > > >    	save_ret = flags & (BPF_TRAMP_F_CALL_ORIG | BPF_TRAMP_F_RET_FENTRY_RET);
> > > > -	if (save_ret) {
> > > > +	if (save_ret)
> > > >    		stack_size += 16; /* Save both A5 (BPF R0) and A0 */
> > > > -		retval_off = stack_size;
> > > > -	}
> > > > +	retval_off = stack_size;
> > > >    	stack_size += nr_arg_slots * 8;
> > > >    	args_off = stack_size;