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] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAK7-dKZ9fP=2u+-KtwqFiA0SYP-0OUrREjESwvM_vwT4St8ZyA@mail.gmail.com>
Date: Mon, 8 Sep 2025 13:52:44 +0900
From: Hoyeon Lee <hoyeon.lee@...e.com>
To: Yonghong Song <yonghong.song@...ux.dev>
Cc: netdev@...r.kernel.org, Andrii Nakryiko <andrii@...nel.org>, 
	Eduard Zingerman <eddyz87@...il.com>, Alexei Starovoitov <ast@...nel.org>, 
	Daniel Borkmann <daniel@...earbox.net>, Martin KaFai Lau <martin.lau@...ux.dev>, Song Liu <song@...nel.org>, 
	John Fastabend <john.fastabend@...il.com>, KP Singh <kpsingh@...nel.org>, 
	Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>, 
	Nathan Chancellor <nathan@...nel.org>, Nick Desaulniers <nick.desaulniers+lkml@...il.com>, 
	Bill Wendling <morbo@...gle.com>, Justin Stitt <justinstitt@...gle.com>, 
	"open list:BPF [LIBRARY] (libbpf)" <bpf@...r.kernel.org>, open list <linux-kernel@...r.kernel.org>, 
	"open list:CLANG/LLVM BUILD SUPPORT:Keyword:b(?i:clang|llvm)b" <llvm@...ts.linux.dev>
Subject: Re: [RFC bpf-next v2 1/1] libbpf: add compile-time OOB warning to bpf_tail_call_static

On Sat, Sep 6, 2025 at 12:55 AM Yonghong Song <yonghong.song@...ux.dev> wrote:
>
>
>
> On 9/4/25 10:18 PM, Hoyeon Lee wrote:
> > Add a compile-time check to bpf_tail_call_static() to warn when a
> > constant slot(index) >= map->max_entries. This uses a small
> > BPF_MAP_ENTRIES() macro together with Clang's diagnose_if attribute.
> >
> > Clang front-end keeps the map type with a '(*max_entries)[N]' field,
> > so the expression
> >
> >      sizeof(*(m)->max_entries) / sizeof(**(m)->max_entries)
> >
> > is resolved to N entirely at compile time. This allows diagnose_if()
> > to emit a warning when a constant slot index is out of range.
> >
> > Out-of-bounds tail calls are currently silent no-ops at runtime, so
> > emitting a compile-time warning helps detect logic errors earlier.
> > This is currently limited to Clang (due to diagnose_if) and only for
> > constant indices, but should still catch the common cases.
> >
> > Signed-off-by: Hoyeon Lee <hoyeon.lee@...e.com>
> > ---
> > Changes in V2:
> > - add function definition for __bpf_tail_call_warn for compile error
> >
> >   tools/lib/bpf/bpf_helpers.h | 21 +++++++++++++++++++++
> >   1 file changed, 21 insertions(+)
> >
> > diff --git a/tools/lib/bpf/bpf_helpers.h b/tools/lib/bpf/bpf_helpers.h
> > index 80c028540656..98bc1536c497 100644
> > --- a/tools/lib/bpf/bpf_helpers.h
> > +++ b/tools/lib/bpf/bpf_helpers.h
> > @@ -173,6 +173,27 @@ bpf_tail_call_static(void *ctx, const void *map, const __u32 slot)
> >                    :: [ctx]"r"(ctx), [map]"r"(map), [slot]"i"(slot)
> >                    : "r0", "r1", "r2", "r3", "r4", "r5");
> >   }
> > +
> > +#if __has_attribute(diagnose_if)
> > +static __always_inline void __bpf_tail_call_warn(int oob)
> > +     __attribute__((diagnose_if(oob, "bpf_tail_call: slot >= max_entries",
> > +                                "warning"))) {};
> > +
> > +#define BPF_MAP_ENTRIES(m) \
> > +     ((__u32)(sizeof(*(m)->max_entries) / sizeof(**(m)->max_entries)))
> > +
> > +#ifndef bpf_tail_call_static
> > +#define bpf_tail_call_static(ctx, map, slot)                               \
> > +({                                                                         \
> > +     /* wrapped to avoid double evaluation. */                             \
> > +     const __u32 __slot = (slot);                                          \
> > +     __bpf_tail_call_warn(__slot >= BPF_MAP_ENTRIES(map));                 \
> > +     /* Avoid re-expand & invoke original as (bpf_tail_call_static)(..) */ \
> > +     (bpf_tail_call_static)(ctx, map, __slot);                             \
> > +})
> > +#endif /* bpf_tail_call_static */
> > +#endif
>
> I got the following error with llvm21.
>
> progs/tailcall_bpf2bpf3.c:20:3: error: bpf_tail_call: slot >= max_entries [-Werror,-Wuser-defined-warnings]
>     20 |                 bpf_tail_call_static(skb, &jmp_table,progs/tailcall_bpf2bpf2.c:17:3 10);
>        | :                ^
>   /home/yhs/work/bpf-next/tools/testing/selftests/bpf/tools/include/bpf/bpf_helpers.h:190:53: note: expanded from macro
>        'bpf_tail_call_static'
>    190 |         __bpf_tail_call_warn(__slot >= BPF_MAP_ENTRIES(map));                 \
>        |                                                            ^
> /home/yhs/work/bpf-next/tools/testing/selftests/bpf/tools/include/bpf/bpf_helpers.h:179:17: note: from 'diagnose_if'
>        attribute on '__bpf_tail_call_warn':
>    179 |         __attribute__((diagnose_if(oob, "bpf_tail_call: slot >= max_entries",
>        |                        ^           ~~~
> error: bpf_tail_call: slot >= max_entries [-Werror,-Wuser-defined-warnings]
>     17 |                 bpf_tail_call_static(skb, &jmp_table, 1);
>        |                 ^
> /home/yhs/work/bpf-next/tools/testing/selftests/bpf/tools/include/bpf/bpf_helpers.h:190:53: note: expanded from macro
>        'bpf_tail_call_static'
>    190 |         __bpf_tail_call_warn(__slot >= BPF_MAP_ENTRIES(map));                 \
>        |                                                            ^
> /home/yhs/work/bpf-next/tools/testing/selftests/bpf/tools/include/bpf/bpf_helpers.h:179:17: note: from 'diagnose_if'
>        attribute on '__bpf_tail_call_warn':
>    179 |         __attribute__((diagnose_if(oob, "bpf_tail_call: slot >= max_entries",
>        |                        ^           ~~~
>    CLNG-BPF [test_progs] tailcall_poke.bpf.o
> 1 error generated.
> make: *** [Makefile:733: /home/yhs/work/bpf-next/tools/testing/selftests/bpf/tailcall_bpf2bpf3.bpf.o] Error 1
>

Thank you for sharing build results! Checked BPF CI, and found 2 issues:

1. selftests/bpf promote warnings to errors (-Werror)
For bpf2bpf tail-call variant progs that intentionally calls OOB trigger
this diagnostic, relaxing just those files keeps CI green while still
showing the warning:

        # tools/testing/selftests/bpf/Makefile
        progs/tailcall_bpf2bpf%.c-CFLAGS := -Wno-error=user-defined-warnings

2. 'void *' maps build error  (bpf2bpf / map-in-map)
The proposed warning is meant only for typed .maps objects. When a prog
passes a void * map, BPF_MAP_ENTRIES() must not attempt member
access. A _Generic gate fixes this by filtering only for typed maps and
yielding 0U for void* families:

        # from BPF-CI build error
        # function prototype:
        #         int subprog_tail(struct __sk_buff *skb, void *jmp_table)
        progs/tailcall_bpf2bpf_hierarchy3.c:36:2: error: member
reference base type 'void' is not a structure or union
          36 | bpf_tail_call_static(skb, jmp_table, 0);
               | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

fixes:

        #define BPF_MAP_ENTRIES(m) _Generic((m), \
               void *:                 0U,      \
               const void *:           0U,      \
               volatile void *:        0U,      \
               const volatile void *:  0U,      \
               default:
((__u32)(sizeof(*(m)->max_entries) / sizeof(**(m)->max_entries))) \
       )

This avoids the compile error, but this is not a very clean solution.

As Andrii Nakryiko noted, map->max_entries can be changed at runtime,
which makes this compile-time approach misleading. I hadn’t considered
that, so this RFC seems less practical to pursue further. Still, I’m
sharing the troubleshooting above in case parts of it may be useful for
future attempts.


> > +
> >   #endif
> >   #endif
> >
> > --
> > 2.51.0
>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ