Events: available bytes calculation via ioctl(FIONREAD).

This makes it possible to avoid looping for a long time while working
with a fast enough peer when data are added to the socket buffer faster
than we are able to read and process them (ticket #1431).  This is
basically what we already do on FreeBSD with kqueue, where information
about the number of bytes in the socket buffer is returned by
the kevent() call.

With other event methods rev->available is now set to -1 when the socket
is ready for reading.  Later in ngx_recv() and ngx_recv_chain(), if
full buffer is received, real number of bytes in the socket buffer is
retrieved using ioctl(FIONREAD).  Reading more than this number of bytes
ensures that even with edge-triggered event methods the event will be
triggered again, so it is safe to stop processing of the socket and
switch to other connections.

Using ioctl(FIONREAD) only after reading a full buffer is an optimization.
With this approach we only call ioctl(FIONREAD) when there are at least
two recv()/readv() calls.

3 files changed, 77 insertions, 2 deletions

diff --git a/src/os/unix/ngx_readv_chain.c b/src/os/unix/ngx_readv_chain.c
index 454cfdcad..a3577ce19 100644
--- a/src/os/unix/ngx_readv_chain.c
+++ b/src/os/unix/ngx_readv_chain.c
@@ -60,7 +60,7 @@ ngx_readv_chain(ngx_connection_t *c, ngx_chain_t *chain, off_t limit)
                        "readv: eof:%d, avail:%d",
                        rev->pending_eof, rev->available);
 
-        if (!rev->available && !rev->pending_eof) {
+        if (rev->available == 0 && !rev->pending_eof) {
             return NGX_AGAIN;
         }
     }
@@ -165,6 +165,40 @@ ngx_readv_chain(ngx_connection_t *c, ngx_chain_t *chain, off_t limit)
 
 #endif
 
+#if (NGX_HAVE_FIONREAD)
+
+            if (rev->available >= 0) {
+                rev->available -= n;
+
+                /*
+                 * negative rev->available means some additional bytes
+                 * were received between kernel notification and readv(),
+                 * and therefore ev->ready can be safely reset even for
+                 * edge-triggered event methods
+                 */
+
+                if (rev->available < 0) {
+                    rev->available = 0;
+                    rev->ready = 0;
+                }
+
+                ngx_log_debug1(NGX_LOG_DEBUG_EVENT, c->log, 0,
+                               "readv: avail:%d", rev->available);
+
+            } else if (n == size) {
+
+                if (ngx_socket_nread(c->fd, &rev->available) == -1) {
+                    n = ngx_connection_error(c, ngx_socket_errno,
+                                             ngx_socket_nread_n " failed");
+                    break;
+                }
+
+                ngx_log_debug1(NGX_LOG_DEBUG_EVENT, c->log, 0,
+                               "readv: avail:%d", rev->available);
+            }
+
+#endif
+
 #if (NGX_HAVE_EPOLLRDHUP)
 
             if ((ngx_event_flags & NGX_USE_EPOLL_EVENT)
diff --git a/src/os/unix/ngx_recv.c b/src/os/unix/ngx_recv.c
index c85fd453c..ddfae4dbe 100644
--- a/src/os/unix/ngx_recv.c
+++ b/src/os/unix/ngx_recv.c
@@ -57,7 +57,7 @@ ngx_unix_recv(ngx_connection_t *c, u_char *buf, size_t size)
                        "recv: eof:%d, avail:%d",
                        rev->pending_eof, rev->available);
 
-        if (!rev->available && !rev->pending_eof) {
+        if (rev->available == 0 && !rev->pending_eof) {
             rev->ready = 0;
             return NGX_AGAIN;
         }
@@ -116,6 +116,40 @@ ngx_unix_recv(ngx_connection_t *c, u_char *buf, size_t size)
 
 #endif
 
+#if (NGX_HAVE_FIONREAD)
+
+            if (rev->available >= 0) {
+                rev->available -= n;
+
+                /*
+                 * negative rev->available means some additional bytes
+                 * were received between kernel notification and recv(),
+                 * and therefore ev->ready can be safely reset even for
+                 * edge-triggered event methods
+                 */
+
+                if (rev->available < 0) {
+                    rev->available = 0;
+                    rev->ready = 0;
+                }
+
+                ngx_log_debug1(NGX_LOG_DEBUG_EVENT, c->log, 0,
+                               "recv: avail:%d", rev->available);
+
+            } else if ((size_t) n == size) {
+
+                if (ngx_socket_nread(c->fd, &rev->available) == -1) {
+                    n = ngx_connection_error(c, ngx_socket_errno,
+                                             ngx_socket_nread_n " failed");
+                    break;
+                }
+
+                ngx_log_debug1(NGX_LOG_DEBUG_EVENT, c->log, 0,
+                               "recv: avail:%d", rev->available);
+            }
+
+#endif
+
 #if (NGX_HAVE_EPOLLRDHUP)
 
             if ((ngx_event_flags & NGX_USE_EPOLL_EVENT)
diff --git a/src/os/unix/ngx_socket.h b/src/os/unix/ngx_socket.h
index fcc515335..ec66a6f83 100644
--- a/src/os/unix/ngx_socket.h
+++ b/src/os/unix/ngx_socket.h
@@ -38,6 +38,13 @@ int ngx_blocking(ngx_socket_t s);
 
 #endif
 
+#if (NGX_HAVE_FIONREAD)
+
+#define ngx_socket_nread(s, n)  ioctl(s, FIONREAD, n)
+#define ngx_socket_nread_n      "ioctl(FIONREAD)"
+
+#endif
+
 int ngx_tcp_nopush(ngx_socket_t s);
 int ngx_tcp_push(ngx_socket_t s);