Messages in this thread | | | Date | Tue, 19 Mar 2019 10:29:14 +0900 | From | Sergey Senozhatsky <> | Subject | Re: [PATCH] tcp: don't use __constant_cpu_to_be32 |
| |
On (03/18/19 09:32), Stanislav Fomichev wrote: [..] > > -#define bpf_htons(x) \ > > - (__builtin_constant_p(x) ? \ > > - __bpf_constant_htons(x) : __bpf_htons(x)) > > -#define bpf_ntohs(x) \ > > - (__builtin_constant_p(x) ? \ > > - __bpf_constant_ntohs(x) : __bpf_ntohs(x)) > > -#define bpf_htonl(x) \ > > - (__builtin_constant_p(x) ? \ > > - __bpf_constant_htonl(x) : __bpf_htonl(x)) > > -#define bpf_ntohl(x) \ > > - (__builtin_constant_p(x) ? \ > > - __bpf_constant_ntohl(x) : __bpf_ntohl(x)) > > +#define bpf_htons(x) __bpf_htons((x)) > > +#define bpf_ntohs(x) __bpf_ntohs((x)) > > +#define bpf_htonl(x) __bpf_htonl((x)) > > +#define bpf_ntohl(x) __bpf_ntohl((x)) > At this point we can probably drop __bpf_xxx as well? > Care to resend with proper description when bpf-next opens?
OK.
-ss
| |