Comment 7 for bug 2028932

Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

With 6.1.0-1019-oem in -proposed, the ubuntu_bpf test can be built without any issue. Thanks!

But to note that test_verifier test reported that there is a new failure:
  #906/p precise: ST insn causing spi > allocated_stack FAIL
  Unexpected verifier log!
  EXP: parent didn't have regs=10 stack=0 marks
  RES:
  func#0 @0
  0: R1=ctx(off=0,imm=0) R10=fp0
  0: (bf) r3 = r10 ; R3=fp0 R10=fp0
  1: (55) if r3 != 0x7b goto pc+0 ; R3=fp0
  2: (7a) *(u64 *)(r3 -8) = 0 ; R3=fp0 fp-8_w=00000000
  3: (79) r4 = *(u64 *)(r10 -8) ; R4_w=P0 R10=fp0
  4: (b7) r0 = -1 ; R0=-1
  5: (2d) if r4 > r0 goto pc+0
  last_idx 5 first_idx 5
  parent didn't have regs=1 stack=0 marks: R0_rw=P-1 R1=ctx(off=0,imm=0) R3=fp0 R4_rw=P0 R10=fp0 fp-8_w=00000000
  last_idx 4 first_idx 2
  regs=1 stack=0 before 4: (b7) r0 = -1
  6: (95) exit

  from 1 to 2: safe
  processed 8 insns (limit 1000000) max_states_per_insn 0 total_states 4 peak_states 4 mark_read 1

This probably came with some other patches from stable as well. This test case was good with 6.1.0-1015.15:
 #906/p precise: ST insn causing spi > allocated_stack OK