Date: Sun, 14 Sep 2014 22:42:43 +0800 From: "dongshan" <thomasyang1206@126.com> To: <net@freebsd.org> Subject: netmap BUG found on big endian machine Message-ID: <001e01cfd02a$255cb430$70161c90$@126.com>
next in thread | raw e-mail | index | archive | help
Hi,=20
=20
I have tested netmap on freescale PowerPC board, it is a big =
endian
machine, when I test the example app pkt-gen=A3=AC I found one bug, =
really it is
bug of netmap souce code.
The commit id is =
=A1=B0d39c4411a2129926d262f9faffacaf876392c7cd=A1=B1, the
bug locates in =A1=B0netmap_mem2.c, netmap_mem_rings_create()=A1=B1, you =
should use=20
=A1=B0*(uint32_t *)(uintptr_t)&ring->nr_buf_size =3D
NETMAP_BDG_BUF_SIZE(na->nm_mem);=A1=B1 substituting
=A1=B0*(uint16_t *)(uintptr_t)&ring->nr_buf_size =3D
NETMAP_BDG_BUF_SIZE(na->nm_mem);=A1=B1
Because the outcome of big endian and little endian is different. =
I
guess the source code is written and tested on intel arch machine. But =
it is
true it can=A1=AFt run big endian machine directly.
=20
=20
Best regards,
Dongshan
=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?001e01cfd02a$255cb430$70161c90$>