[bpf-next,v2,1/3] bpf: Fix array bounds error with may_goto

may_goto uses an additional 8 bytes on the stack, which causes the
interpreters[] array to go out of bounds when calculating index by
stack_size.

1. If a BPF program is rewritten, re-evaluate the stack size. For non-JIT
cases, reject loading directly.

2. For non-JIT cases, calculating interpreters[idx] may still cause
out-of-bounds array access, and just warn about it.

3. For jit_requested cases, the execution of bpf_func also needs to be
warned. So Move the definition of function __bpf_prog_ret0_warn out of
the macro definition CONFIG_BPF_JIT_ALWAYS_ON

Reported-by: syzbot+d2a2c639d03ac200a4f1@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/bpf/0000000000000f823606139faa5d@google.com/
Fixes: 011832b97b311 ("bpf: Introduce may_goto instruction")
Signed-off-by: Jiayuan Chen <mrpre@163.com>
---
 kernel/bpf/core.c     | 18 ++++++++++++++----
 kernel/bpf/verifier.c |  7 +++++++
 2 files changed, 21 insertions(+), 4 deletions(-)

On Thu, Feb 13, 2025 at 9:03 AM Jiayuan Chen <mrpre@163.com> wrote:
>
> On Thu, Feb 13, 2025 at 08:02:55AM -0800, Alexei Starovoitov wrote:
> > On Thu, Feb 13, 2025 at 5:13 AM Jiayuan Chen <mrpre@163.com> wrote:
> > >
> > > may_goto uses an additional 8 bytes on the stack, which causes the
> > > interpreters[] array to go out of bounds when calculating index by
> > > stack_size.
> > >
> > > 1. If a BPF program is rewritten, re-evaluate the stack size. For non-JIT
> > > cases, reject loading directly.
> > >
> > > 2. For non-JIT cases, calculating interpreters[idx] may still cause
> > > out-of-bounds array access, and just warn about it.
> > >
> > > 3. For jit_requested cases, the execution of bpf_func also needs to be
> > > warned. So Move the definition of function __bpf_prog_ret0_warn out of
> > > the macro definition CONFIG_BPF_JIT_ALWAYS_ON
> > >
> [...]
> > > ---
> > >  EVAL6(PROG_NAME_LIST, 224, 256, 288, 320, 352, 384)
> > >  EVAL4(PROG_NAME_LIST, 416, 448, 480, 512)
> > >  };
> > > +
> > > +#define MAX_INTERPRETERS_CALLBACK (sizeof(interpreters) / sizeof(*interpreters))
> >
> > There is ARRAY_SIZE macro.
> Thanks, I will use it.
> >
> > >  #undef PROG_NAME_LIST
> > >  #define PROG_NAME_LIST(stack_size) PROG_NAME_ARGS(stack_size),
> > >  static __maybe_unused
> > > @@ -2290,17 +2293,18 @@ void bpf_patch_call_args(struct bpf_insn *insn, u32 stack_depth)
> > >         insn->code = BPF_JMP | BPF_CALL_ARGS;
> > >  }
> > >  #endif
> > > -#else
> > > +#endif
> > > +
> > >  static unsigned int __bpf_prog_ret0_warn(const void *ctx,
> > >                                          const struct bpf_insn *insn)
> > >  {
> > >         /* If this handler ever gets executed, then BPF_JIT_ALWAYS_ON
> > > -        * is not working properly, so warn about it!
> > > +        * is not working properly, or interpreter is being used when
> > > +        * prog->jit_requested is not 0, so warn about it!
> > >          */
> > >         WARN_ON_ONCE(1);
> > >         return 0;
> > >  }
> > > -#endif
> > >
> > >  bool bpf_prog_map_compatible(struct bpf_map *map,
> > >                              const struct bpf_prog *fp)
> > > @@ -2380,8 +2384,14 @@ static void bpf_prog_select_func(struct bpf_prog *fp)
> > >  {
> > >  #ifndef CONFIG_BPF_JIT_ALWAYS_ON
> > >         u32 stack_depth = max_t(u32, fp->aux->stack_depth, 1);
> > > +       u32 idx = (round_up(stack_depth, 32) / 32) - 1;
> > >
> > > -       fp->bpf_func = interpreters[(round_up(stack_depth, 32) / 32) - 1];
> > > +       if (!fp->jit_requested) {
> >
> > I don't think above check is necessary.
> > Why not just
> > if (WARN_ON_ONCE(idx >= ARRAY_SIZE(interpreters)))
> >   fp->bpf_func = __bpf_prog_ret0_warn;
> > else
> >   fp->bpf_func = interpreters[idx];
> >
>
> When jit_requested is set 1, the stack_depth can still go above 512,
> and we'd end up executing this function, where the index calculation would
> overflow, triggering an array out-of-bounds warning from USCAN or WAR().

Ok, then do:
if (!fp->jit_requested && WARN_ON_ONCE(idx >= ARRAY_SIZE(interpreters)))

> > > +               WARN_ON_ONCE(idx >= MAX_INTERPRETERS_CALLBACK);
> > > +               fp->bpf_func = interpreters[idx];

since warning and anyway proceeding to access the array out of bounds
is just wrong.