mbox series

[bpf,0/2] selftests: bpf: more sub-register zero extension unit tests

Message ID 1559123829-9318-1-git-send-email-jiong.wang@netronome.com
Headers show
Series selftests: bpf: more sub-register zero extension unit tests | expand

Message

Jiong Wang May 29, 2019, 9:57 a.m. UTC 
JIT back-ends need to guarantee high 32-bit cleared whenever one eBPF insn
write low 32-bit sub-register only. It is possible that some JIT back-ends
have failed doing this and are silently generating wrong image.

This set completes the unit tests, so bug on this could be exposed.

Jiong Wang (2):
  selftests: bpf: move sub-register zero extension checks into subreg.c
  selftests: bpf: complete sub-register zero extension checks

 tools/testing/selftests/bpf/verifier/basic_instr.c |  39 --
 tools/testing/selftests/bpf/verifier/subreg.c      | 533 +++++++++++++++++++++
 2 files changed, 533 insertions(+), 39 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/verifier/subreg.c

Comments

Daniel Borkmann May 29, 2019, 12:59 p.m. UTC | #1 
On 05/29/2019 11:57 AM, Jiong Wang wrote:
> JIT back-ends need to guarantee high 32-bit cleared whenever one eBPF insn
> write low 32-bit sub-register only. It is possible that some JIT back-ends
> have failed doing this and are silently generating wrong image.
> 
> This set completes the unit tests, so bug on this could be exposed.
> 
> Jiong Wang (2):
>   selftests: bpf: move sub-register zero extension checks into subreg.c
>   selftests: bpf: complete sub-register zero extension checks
> 
>  tools/testing/selftests/bpf/verifier/basic_instr.c |  39 --
>  tools/testing/selftests/bpf/verifier/subreg.c      | 533 +++++++++++++++++++++
>  2 files changed, 533 insertions(+), 39 deletions(-)
>  create mode 100644 tools/testing/selftests/bpf/verifier/subreg.c

Looks good, thanks for following up! Applied, thanks!