From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:39338) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1USOxK-0006OO-D8 for qemu-devel@nongnu.org; Wed, 17 Apr 2013 05:47:11 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1USOxF-0002DI-G3 for qemu-devel@nongnu.org; Wed, 17 Apr 2013 05:47:10 -0400 Received: from mail-ea0-x22d.google.com ([2a00:1450:4013:c01::22d]:34185) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1USOxF-0002D9-AS for qemu-devel@nongnu.org; Wed, 17 Apr 2013 05:47:05 -0400 Received: by mail-ea0-f173.google.com with SMTP id k11so624146eaj.4 for ; Wed, 17 Apr 2013 02:47:04 -0700 (PDT) Sender: Paolo Bonzini From: Paolo Bonzini Date: Wed, 17 Apr 2013 11:46:49 +0200 Message-Id: <1366192012-14872-4-git-send-email-pbonzini@redhat.com> In-Reply-To: <1366192012-14872-1-git-send-email-pbonzini@redhat.com> References: <1366192012-14872-1-git-send-email-pbonzini@redhat.com> Subject: [Qemu-devel] [PATCH 3/6] iov: reorganize iov_send_recv, part 2 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: owasserm@redhat.com, quintela@redhat.com Do not touch the "bytes" argument anymore. Instead, remember the original length of the last iovec if we touch it, and restore it afterwards. This requires undoing the changes in opposite order. The previous algorithm didn't care. Reviewed-by: Juan Quintela Reviewed-by: Orit Wassermann Signed-off-by: Paolo Bonzini --- util/iov.c | 28 ++++++++++++++-------------- 1 file changed, 14 insertions(+), 14 deletions(-) diff --git a/util/iov.c b/util/iov.c index adb9a70..110d18e 100644 --- a/util/iov.c +++ b/util/iov.c @@ -145,7 +145,9 @@ ssize_t iov_send_recv(int sockfd, struct iovec *iov, unsigned iov_cnt, bool do_send) { ssize_t ret; + size_t orig_len, tail; unsigned si, ei; /* start and end indexes */ + if (bytes == 0) { /* Catch the do-nothing case early, as otherwise we will pass an * empty iovec to sendmsg/recvmsg(), and not all implementations @@ -174,31 +176,29 @@ ssize_t iov_send_recv(int sockfd, struct iovec *iov, unsigned iov_cnt, } /* Find the end position skipping `bytes' bytes: */ /* first, skip all full-sized elements */ - for (ei = 0; ei < iov_cnt && iov[ei].iov_len <= bytes; ++ei) { - bytes -= iov[ei].iov_len; + tail = bytes; + for (ei = 0; ei < iov_cnt && iov[ei].iov_len <= tail; ++ei) { + tail -= iov[ei].iov_len; } - if (bytes) { - /* second, fixup the last element, and remember - * the length we've cut from the end of it in `bytes' */ - size_t tail; + if (tail) { + /* second, fixup the last element, and remember the original + * length */ assert(ei < iov_cnt); - assert(iov[ei].iov_len > bytes); - tail = iov[ei].iov_len - bytes; - iov[ei].iov_len = bytes; - bytes = tail; /* bytes is now equal to the tail size */ - ++ei; + assert(iov[ei].iov_len > tail); + orig_len = iov[ei].iov_len; + iov[ei++].iov_len = tail; } ret = do_send_recv(sockfd, iov, ei, do_send); /* Undo the changes above */ + if (tail) { + iov[ei-1].iov_len = orig_len; + } if (offset) { iov[0].iov_base -= offset; iov[0].iov_len += offset; } - if (bytes) { - iov[ei-1].iov_len += bytes; - } return ret; } -- 1.8.1.4