Re: [PATCHv7 bpf-next 10/29] bpf: Add multi tracing attach types

[Jiri Olsa <olsajiri@gmail.com>]

On Wed, Jun 03, 2026 at 11:37:51AM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [High] Missing `BPF_TRACE_FENTRY_MULTI` and `BPF_TRACE_FEXIT_MULTI` cases in `btf_validate_prog_ctx_type` causes valid `.multi` programs using `u64 *ctx` to be rejected.
> --
> 
> commit 0330f287f02c3a32c9ab8bf7f4f5a6ad6cf7366f
> Author: Jiri Olsa <jolsa@kernel.org>
> 
> bpf: Add multi tracing attach types
> 
> This commit introduces new program attach types BPF_TRACE_FENTRY_MULTI and 
> BPF_TRACE_FEXIT_MULTI to support multi-tracing attachment. Base verifier 
> support is added to recognize these types and accept them as sleepable 
> programs, while deferring actual target validation to link creation.
> 
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 8ed484cb1a8a4..521ec2a9051ed 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> [ ... ]
> > @@ -16314,6 +16314,8 @@ static bool return_retval_range(struct bpf_verifier_env *env, struct bpf_retval_
> >  		case BPF_TRACE_FENTRY:
> >  		case BPF_TRACE_FEXIT:
> >  		case BPF_TRACE_FSESSION:
> > +		case BPF_TRACE_FENTRY_MULTI:
> > +		case BPF_TRACE_FEXIT_MULTI:
> 
> [Severity: High]
> Are we missing an update to btf_validate_prog_ctx_type() in kernel/bpf/btf.c
> to handle these new multi types?
> 
> Currently, BPF_TRACE_FENTRY and BPF_TRACE_FEXIT are explicitly whitelisted
> there to allow u64 * as a context type:
> 
> kernel/bpf/btf.c:btf_validate_prog_ctx_type() {
>     ...
> 		case BPF_TRACE_FENTRY:
> 		case BPF_TRACE_FEXIT:
> 		case BPF_MODIFY_RETURN:
> 		case BPF_TRACE_FSESSION:
> 			/* allow u64* as ctx */
> 			if (btf_is_int(t) && t->size == 8)
> 				return 0;
> 			break;
>     ...
> }
> 
> Without adding the new multi types to this switch statement, the verifier
> will fall back to the default canonical type (void *) and will reject valid
> multi programs that use BPF_PROG and u64 *ctx.

yes, we disabled direct ctx access for tracing multi programs 

jirka