Re: kernel/bpf/verifier.c: 4 * possible unintended fallthrough ?

Re: kernel/bpf/verifier.c: 4 * possible unintended fallthrough ?

[Alexei Starovoitov]

On Tue, Dec 20, 2016 at 3:20 AM, David Binderman <dcb314@hotmail.com> wrote:
> Hello there,
>
> I just tried to compile kernel-4.9 with a recent development
> version of gcc. It said
>
> kernel/bpf/verifier.c:1907:23: warning: this statement may fall through [-Wimplicit-fallthrough=]
> kernel/bpf/verifier.c:1918:23: warning: this statement may fall through [-Wimplicit-fallthrough=]
> kernel/bpf/verifier.c:1859:24: warning: this statement may fall through [-Wimplicit-fallthrough=]
> kernel/bpf/verifier.c:1869:24: warning: this statement may fall through [-Wimplicit-fallthrough=]
>
> Source code for the first one is
>
>     case BPF_JGT:
>         /* Unsigned comparison, the minimum value is 0. */
>         true_reg->min_value = 0;
>     case BPF_JSGT:
>
> Suggest either add the missing break or document the fallthrough
> with a comment something like /* FALLTHROUGH */

I've tried 4.9 and 5.2 and don't see this warning.
Is this 6.x gcc?
I suspect it will have such warnings all over the kernel.

Re: kernel/bpf/verifier.c: 4 * possible unintended fallthrough ?

[Josef Bacik]

On Tue, Dec 20, 2016 at 11:34 AM, David Binderman <dcb314@hotmail.com> 
wrote:
> 
> Hello there,
> 
>> From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
>> I've tried 4.9 and 5.2 and don't see this warning.
> 
> As expected - I used a development version of gcc.
> Latest released version is 6.2
> 
>> Is this 6.x gcc?
> 
> 7.0 would be more accurate.
> 
>> I suspect it will have such warnings all over the kernel.
> 
> Indeed it has hundreds, but the subject under discussion is file
> kernel/bpf/verifier.c.
> 
> I am still not sure if I have found a fallthrough bug or not.

You haven't, this is intended so is a useless warning.  Thanks,

Josef