On 24 July 2012 00:49, Jing Huang <jing.huang....@gmail.com> wrote:
>
> Signed-off-by: Jing Huang <jing.huang....@gmail.com>
> Reviewed-by: Peter Maydell <peter.mayd...@linaro.org>
> ---
> linux-user/syscall.c | 15 +++++++++++++--
> 1 files changed, 13 insertions(+), 2 deletions(-)
>
> diff --git a/linux-user/syscall.c b/linux-user/syscall.c
> index 539af3f..3319bb8 100644
> --- a/linux-user/syscall.c
> +++ b/linux-user/syscall.c
> @@ -1256,7 +1256,6 @@ static inline abi_long
> host_to_target_sockaddr(abi_ulong target_addr,
> return 0;
> }
>
> -/* ??? Should this also swap msgh->name? */
> static inline abi_long target_to_host_cmsg(struct msghdr *msgh,
> struct target_msghdr *target_msgh)
> {
This is removing the comment from the wrong function: there's
an identical one over host_to_target_cmsg() which is the one
most immediately addressed by this patch. You can remove both
comments, though.
> @@ -1873,10 +1872,22 @@ static abi_long do_sendrecvmsg(int fd, abi_ulong
> target_msg,
> if (!is_error(ret)) {
> len = ret;
> ret = host_to_target_cmsg(msgp, &msg);
> - if (!is_error(ret))
> + if (!is_error(ret)) {
> + msgp->msg_namelen = tswap32(msg.msg_namelen);
> + if (msg.msg_name != NULL) {
> + ret = host_to_target_sockaddr(tswapal(msgp->msg_name),
> + msg.msg_name,
> msg.msg_namelen);
> + if (ret) {
> + goto out;
> + }
> + }
> +
> ret = len;
> + }
> }
> }
> +
> +out:
> unlock_iovec(vec, target_vec, count, !send);
> unlock_user_struct(msgp, target_msg, send ? 0 : 1);
> return ret;
> --
> 1.7.8.6
>
-- PMM
