Daniel Borkmann writes:
> On 01/25/2019 01:10 AM, Jiong Wang wrote: >> This patch teach verifier about the new BPF_JMP32 instruction class. >> Verifier need to treat it similar as the existing BPF_JMP class. >> A BPF_JMP32 insn needs to go through all checks that have been done on >> BPF_JMP. >> >> Also, verifier is doing runtime optimizations based on the extra info >> conditional jump instruction could offer, especially when the comparison is >> between constant and register that the value range of the register could be >> improved based on the comparison results. These code are updated >> accordingly. >> >> Acked-by: Jakub Kicinski <jakub.kicin...@netronome.com> >> Signed-off-by: Jiong Wang <jiong.w...@netronome.com> > > Series looks good to me, but if I spot this correctly one thing that has > not been addressed here is proper rebase on top of Jakub's dead code > removal, e.g. in opt_hard_wire_dead_code_branches() where we check in > insn_is_cond_jump() for jump opcodes it still only tests for BPF_JMP > class whereas BPF_JMP32 handling needs to be taught here as well. Thanks for catching this. Yes, insn_is_cond_jump() should be updated for JMP32 as well. JMP32 is guaranteed to be with condition jump operation only otherwise the earlier do_check will complain use of reserved encoding bits. I am going to teach insn_is_cond_jump to return true for JMP32. And search the commits, there is another similar new helper function in nfp driver jit. Will fix both places, and re-spin v4. Thanks. Regards, Jiong
