syzkaller generated a BPF proglet and triggered a warning with
the following:

  0: (b7) r0 = 0
  1: (d5) if r0 s<= 0x0 goto pc+0
   R0=inv0 R1=ctx(id=0,off=0,imm=0) R10=fp0
  2: (1f) r0 -= r1
   R0=inv0 R1=ctx(id=0,off=0,imm=0) R10=fp0
  verifier internal error: known but bad sbounds

What happens is that in the first insn, r0's min/max value
are both 0 due to the immediate assignment, later in the jsle
test the bounds are updated for the min value in the false
path, meaning, they yield smin_val = 1, smax_val = 0, and when
ctx pointer is subtracted from r0, verifier bails out with the
internal error and throwing a WARN since smin_val != smax_val
for the known constant.

For min_val > max_val scenario it means that reg_set_min_max()
and reg_set_min_max_inv() (which both refine existing bounds)
demonstrated that such branch cannot be taken at runtime.

In above scenario for the case where it will be taken, the
existing [0, 0] bounds are kept intact. Meaning, the rejection
is not due to a verifier internal error, and therefore the
WARN() is not necessary either.

We could just reject such cases in adjust_{ptr,scalar}_min_max_vals()
when either known scalars have smin_val != smax_val or
umin_val != umax_val or any scalar reg with bounds
smin_val > smax_val or umin_val > umax_val. However, there
may be a small risk of breakage of buggy programs, so handle
this more gracefully and in adjust_{ptr,scalar}_min_max_vals()
just taint the dst reg as unknown scalar when we see ops with
such kind of src reg.

Reported-by: syzbot+6d362cadd45dc0a12ba4@syzkaller.appspotmail.com
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>