Re: [PATCH bpf-next v3 1/2] bpf: Skip bounds adjustment for conditional jumps on same scalar register

[KaFai Wan <kafai.wan@linux.dev>]

On Fri, 2025-10-31 at 09:37 -0700, Alexei Starovoitov wrote:
> On Fri, Oct 31, 2025 at 8:44 AM KaFai Wan <kafai.wan@linux.dev> wrote:
> > 
> > When conditional jumps are performed on the same scalar register
> > (e.g., r0 <= r0, r0 > r0, r0 < r0), the BPF verifier incorrectly
> > attempts to adjust the register's min/max bounds. This leads to
> > invalid range bounds and triggers a BUG warning.
> > 
> > The problematic BPF program:
> >    0: call bpf_get_prandom_u32
> >    1: w8 = 0x80000000
> >    2: r0 &= r8
> >    3: if r0 > r0 goto <exit>
> > 
> > The instruction 3 triggers kernel warning:
> >    3: if r0 > r0 goto <exit>
> >    true_reg1: range bounds violation u64=[0x1, 0x0] s64=[0x1, 0x0] u32=[0x1, 0x0] s32=[0x1, 0x0]
> > var_off=(0x0, 0x0)
> >    true_reg2: const tnum out of sync with range bounds u64=[0x0, 0xffffffffffffffff]
> > s64=[0x8000000000000000, 0x7fffffffffffffff] var_off=(0x0, 0x0)
> > 
> > Comparing a register with itself should not change its bounds and
> > for most comparison operations, comparing a register with itself has
> > a known result (e.g., r0 == r0 is always true, r0 < r0 is always false).
> > 
> > Fix this by:
> > 1. Enhance is_scalar_branch_taken() to properly handle branch direction
> >    computation for same register comparisons across all BPF jump operations
> > 2. Adds early return in reg_set_min_max() to avoid bounds adjustment
> >    for unknown branch directions (e.g., BPF_JSET) on the same register
> > 
> > The fix ensures that unnecessary bounds adjustments are skipped, preventing
> > the verifier bug while maintaining correct branch direction analysis.
> > 
> > Reported-by: Kaiyan Mei <M202472210@hust.edu.cn>
> > Reported-by: Yinhao Hu <dddddd@hust.edu.cn>
> > Closes: https://lore.kernel.org/all/1881f0f5.300df.199f2576a01.Coremail.kaiyanm@hust.edu.cn/
> > Signed-off-by: KaFai Wan <kafai.wan@linux.dev>
> > ---
> >  kernel/bpf/verifier.c | 33 +++++++++++++++++++++++++++++++++
> >  1 file changed, 33 insertions(+)
> > 
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 542e23fb19c7..a571263f4ebe 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -15995,6 +15995,8 @@ static int is_scalar_branch_taken(struct bpf_reg_state *reg1, struct
> > bpf_reg_sta
> > 
> >         switch (opcode) {
> >         case BPF_JEQ:
> > +               if (reg1 == reg2)
> > +                       return 1;
> >                 /* constants, umin/umax and smin/smax checks would be
> >                  * redundant in this case because they all should match
> >                  */
> > @@ -16021,6 +16023,8 @@ static int is_scalar_branch_taken(struct bpf_reg_state *reg1, struct
> > bpf_reg_sta
> >                 }
> >                 break;
> >         case BPF_JNE:
> > +               if (reg1 == reg2)
> > +                       return 0;
> >                 /* constants, umin/umax and smin/smax checks would be
> >                  * redundant in this case because they all should match
> >                  */
> > @@ -16047,6 +16051,12 @@ static int is_scalar_branch_taken(struct bpf_reg_state *reg1, struct
> > bpf_reg_sta
> >                 }
> >                 break;
> >         case BPF_JSET:
> > +               if (reg1 == reg2) {
> > +                       if (tnum_is_const(t1))
> > +                               return t1.value != 0;
> > +                       else
> > +                               return (smin1 <= 0 && smax1 >= 0) ? -1 : 1;
> > +               }
> >                 if (!is_reg_const(reg2, is_jmp32)) {
> >                         swap(reg1, reg2);
> >                         swap(t1, t2);
> > @@ -16059,48 +16069,64 @@ static int is_scalar_branch_taken(struct bpf_reg_state *reg1, struct
> > bpf_reg_sta
> >                         return 0;
> >                 break;
> >         case BPF_JGT:
> > +               if (reg1 == reg2)
> > +                       return 0;
> >                 if (umin1 > umax2)
> >                         return 1;
> >                 else if (umax1 <= umin2)
> >                         return 0;
> >                 break;
> >         case BPF_JSGT:
> > +               if (reg1 == reg2)
> > +                       return 0;
> 
> This is uglier than the previous version.
> reg1 == reg2 is a syzbot territory.
> We shouldn't uglify the code everywhere because of it.
> 
ok, will update in v4.
> pw-bot: cr

-- 
Thanks,
KaFai