Kui-Feng,
You are right. Maybe simply "load_word" and "load_byte" would be a
better name here. WDYT?
-Jordan
On Fri, Apr 12, 2024 at 6:01 PM Kui-Feng Lee <sinquersw@xxxxxxxxx> wrote:
>
>
>
> On 4/12/24 09:52, Jordan Rife wrote:
> > Without this fix, the bind4 and bind6 programs will reject bind attempts
> > on big endian systems. This patch ensures that CI tests pass for the
> > s390x architecture.
> >
> > Signed-off-by: Jordan Rife <jrife@xxxxxxxxxx>
> > ---
> > .../testing/selftests/bpf/progs/bind4_prog.c | 18 ++++++++++--------
> > .../testing/selftests/bpf/progs/bind6_prog.c | 18 ++++++++++--------
> > tools/testing/selftests/bpf/progs/bind_prog.h | 19 +++++++++++++++++++
> > 3 files changed, 39 insertions(+), 16 deletions(-)
> > create mode 100644 tools/testing/selftests/bpf/progs/bind_prog.h
> >
> > diff --git a/tools/testing/selftests/bpf/progs/bind4_prog.c b/tools/testing/selftests/bpf/progs/bind4_prog.c
> > index a487f60b73ac4..2bc052ecb6eef 100644
> > --- a/tools/testing/selftests/bpf/progs/bind4_prog.c
> > +++ b/tools/testing/selftests/bpf/progs/bind4_prog.c
> > @@ -12,6 +12,8 @@
> > #include <bpf/bpf_helpers.h>
> > #include <bpf/bpf_endian.h>
> >
> > +#include "bind_prog.h"
> > +
> > #define SERV4_IP 0xc0a801feU /* 192.168.1.254 */
> > #define SERV4_PORT 4040
> > #define SERV4_REWRITE_IP 0x7f000001U /* 127.0.0.1 */
> > @@ -118,23 +120,23 @@ int bind_v4_prog(struct bpf_sock_addr *ctx)
> >
> > // u8 narrow loads:
> > user_ip4 = 0;
> > - user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[0] << 0;
> > - user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[1] << 8;
> > - user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[2] << 16;
> > - user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[3] << 24;
> > + user_ip4 |= load_byte_ntoh(ctx->user_ip4, 0, sizeof(user_ip4));
> > + user_ip4 |= load_byte_ntoh(ctx->user_ip4, 1, sizeof(user_ip4));
> > + user_ip4 |= load_byte_ntoh(ctx->user_ip4, 2, sizeof(user_ip4));
> > + user_ip4 |= load_byte_ntoh(ctx->user_ip4, 3, sizeof(user_ip4));
> > if (ctx->user_ip4 != user_ip4)
> > return 0;
> >
> > user_port = 0;
> > - user_port |= ((volatile __u8 *)&ctx->user_port)[0] << 0;
> > - user_port |= ((volatile __u8 *)&ctx->user_port)[1] << 8;
> > + user_port |= load_byte_ntoh(ctx->user_port, 0, sizeof(user_port));
> > + user_port |= load_byte_ntoh(ctx->user_port, 1, sizeof(user_port));
> > if (ctx->user_port != user_port)
> > return 0;
> >
> > // u16 narrow loads:
> > user_ip4 = 0;
> > - user_ip4 |= ((volatile __u16 *)&ctx->user_ip4)[0] << 0;
> > - user_ip4 |= ((volatile __u16 *)&ctx->user_ip4)[1] << 16;
> > + user_ip4 |= load_word_ntoh(ctx->user_ip4, 0, sizeof(user_ip4));
> > + user_ip4 |= load_word_ntoh(ctx->user_ip4, 1, sizeof(user_ip4));
> > if (ctx->user_ip4 != user_ip4)
> > return 0;
> >
> > diff --git a/tools/testing/selftests/bpf/progs/bind6_prog.c b/tools/testing/selftests/bpf/progs/bind6_prog.c
> > index d62cd9e9cf0ea..194583e3375bf 100644
> > --- a/tools/testing/selftests/bpf/progs/bind6_prog.c
> > +++ b/tools/testing/selftests/bpf/progs/bind6_prog.c
> > @@ -12,6 +12,8 @@
> > #include <bpf/bpf_helpers.h>
> > #include <bpf/bpf_endian.h>
> >
> > +#include "bind_prog.h"
> > +
> > #define SERV6_IP_0 0xfaceb00c /* face:b00c:1234:5678::abcd */
> > #define SERV6_IP_1 0x12345678
> > #define SERV6_IP_2 0x00000000
> > @@ -129,25 +131,25 @@ int bind_v6_prog(struct bpf_sock_addr *ctx)
> > // u8 narrow loads:
> > for (i = 0; i < 4; i++) {
> > user_ip6 = 0;
> > - user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[0] << 0;
> > - user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[1] << 8;
> > - user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[2] << 16;
> > - user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[3] << 24;
> > + user_ip6 |= load_byte_ntoh(ctx->user_ip6[i], 0, sizeof(user_ip6));
> > + user_ip6 |= load_byte_ntoh(ctx->user_ip6[i], 1, sizeof(user_ip6));
> > + user_ip6 |= load_byte_ntoh(ctx->user_ip6[i], 2, sizeof(user_ip6));
> > + user_ip6 |= load_byte_ntoh(ctx->user_ip6[i], 3, sizeof(user_ip6));
> > if (ctx->user_ip6[i] != user_ip6)
> > return 0;
> > }
> >
> > user_port = 0;
> > - user_port |= ((volatile __u8 *)&ctx->user_port)[0] << 0;
> > - user_port |= ((volatile __u8 *)&ctx->user_port)[1] << 8;
> > + user_port |= load_byte_ntoh(ctx->user_port, 0, sizeof(user_port));
> > + user_port |= load_byte_ntoh(ctx->user_port, 1, sizeof(user_port));
> > if (ctx->user_port != user_port)
> > return 0;
> >
> > // u16 narrow loads:
> > for (i = 0; i < 4; i++) {
> > user_ip6 = 0;
> > - user_ip6 |= ((volatile __u16 *)&ctx->user_ip6[i])[0] << 0;
> > - user_ip6 |= ((volatile __u16 *)&ctx->user_ip6[i])[1] << 16;
> > + user_ip6 |= load_word_ntoh(ctx->user_ip6[i], 0, sizeof(user_ip6));
> > + user_ip6 |= load_word_ntoh(ctx->user_ip6[i], 1, sizeof(user_ip6));
> > if (ctx->user_ip6[i] != user_ip6)
> > return 0;
> > }
> > diff --git a/tools/testing/selftests/bpf/progs/bind_prog.h b/tools/testing/selftests/bpf/progs/bind_prog.h
> > new file mode 100644
> > index 0000000000000..0fdc466aec346
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/progs/bind_prog.h
> > @@ -0,0 +1,19 @@
> > +/* SPDX-License-Identifier: GPL-2.0 */
> > +#ifndef __BIND_PROG_H__
> > +#define __BIND_PROG_H__
> > +
> > +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
> > +#define load_byte_ntoh(src, b, s) \
> > + (((volatile __u8 *)&(src))[b] << 8 * b)
> > +#define load_word_ntoh(src, w, s) \
> > + (((volatile __u16 *)&(src))[w] << 16 * w)
> > +#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
> > +#define load_byte_ntoh(src, b, s) \
> > + (((volatile __u8 *)&(src))[(b) + (sizeof(src) - (s))] << 8 * ((s) - (b) - 1))
> > +#define load_word_ntoh(src, w, s) \
> > + (((volatile __u16 *)&(src))[w] << 16 * (((s) / 2) - (w) - 1))
> These names, load_byte_ntoh() and load_word_ntoh(), are miss-leading.
>
> They don't actually do byte-order conversion from network order to host
> order. Network order is big endian. 0xdeadbeef in u32 should be stored
> as the sequence of
>
> 0xde, 0xad, 0xbe, 0xef
>
> The little endian implementation of load_word_ntoh() provided here will
> return 0xadde and 0xefbe0000. However, a network order to host order
> conversion should return 0xbeef and 0xdead0000 for little endian.
>
> The little endian implementation of load_byte_ntoh() here returns 0xde,
> 0xad00, 0xbe0000, and 0xef000000. However, a network to host order
> conversion should return 0xef, 0xbe00, 0xad0000, and 0xde00000.
>
> So, they just access raw data following the host byte order, not
> providing any byte order conversion.
>
>
> > +#else
> > +# error "Fix your compiler's __BYTE_ORDER__?!"
> > +#endif
> > +
> > +#endif
