Re: [PATCH bpf-next v2 1/4] bpf: Add bpf_call_rcu() kfunc

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] bpf: build failure due to missing bpf_call_rcu_tasks_trace implementation
--

commit 06a7a5ada7a6b07d90668fb28558d25075dfcf41
Author: Puranjay Mohan <puranjay@kernel.org>

bpf: Add bpf_call_rcu() kfunc

This commit introduces the bpf_call_rcu() kfunc to allow BPF programs to
queue callbacks that execute after an RCU grace period. It adds the
necessary BTF fields, map validations, and verifier support to safely
invoke these callbacks.

> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 6c6b8d8520cdf..f91c997bc8445 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -12526,6 +12583,8 @@ enum special_kfunc_type {
>  	KF___bpf_trap,
>  	KF_bpf_task_work_schedule_signal,
>  	KF_bpf_task_work_schedule_resume,
> +	KF_bpf_call_rcu,
> +	KF_bpf_call_rcu_tasks_trace,

[Severity: Low]
Will this cause a build failure? Looking at the addition of
KF_bpf_call_rcu_tasks_trace to enum special_kfunc_type in
kernel/bpf/verifier.c, the kfunc is added to the special list here, but the
actual implementation of bpf_call_rcu_tasks_trace does not seem to be
included in this patch.

I understand the implementation is added later in the series in commit
d37bdb9e0af5ddd44775e974f01a56e73c98e9d3 (bpf: Add
bpf_call_rcu_tasks_trace() kfunc), but shouldn't each patch be
self-contained and compilable to preserve git bisect functionality?

>  	KF_bpf_arena_alloc_pages,
>  	KF_bpf_arena_free_pages,
>  	KF_bpf_arena_reserve_pages,

[ ... ]

> @@ -12619,6 +12678,8 @@ BTF_ID(func, bpf_dynptr_file_discard)
>  BTF_ID(func, __bpf_trap)
>  BTF_ID(func, bpf_task_work_schedule_signal)
>  BTF_ID(func, bpf_task_work_schedule_resume)
> +BTF_ID(func, bpf_call_rcu)
> +BTF_ID(func, bpf_call_rcu_tasks_trace)

[Severity: Low]
Does exposing this function to resolve_btfids without its implementation
present in this commit break the build? Since the symbol is missing in
vmlinux at this point in the series, resolve_btfids might fail to find it
and cause a regression during compilation.

>  BTF_ID(func, bpf_arena_alloc_pages)
>  BTF_ID(func, bpf_arena_free_pages)
>  BTF_ID(func, bpf_arena_reserve_pages)

[ ... ]

> @@ -12690,6 +12751,12 @@ static bool is_bpf_rbtree_add_kfunc(u32 func_id)
>  	       func_id == special_kfunc_list[KF_bpf_rbtree_add_impl];
>  }
>  
> +static bool is_call_rcu_kfunc(u32 func_id)
> +{
> +	return func_id == special_kfunc_list[KF_bpf_call_rcu] ||
> +	       func_id == special_kfunc_list[KF_bpf_call_rcu_tasks_trace];
> +}
> +

[Severity: Low]
Can this check for special_kfunc_list[KF_bpf_call_rcu_tasks_trace] be
deferred to the commit that actually introduces the function, so the
verifier checks remain strictly aligned with the available features?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260915114240.3269184-1-puranjay@kernel.org?part=1