Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 14 Jan 2011 22:36:33 +0000 (UTC)
From:      Pyun YongHyeon <yongari@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-8@freebsd.org
Subject:   svn commit: r217429 - stable/8/sys/dev/usb/net
Message-ID:  <201101142236.p0EMaXqi004444@svn.freebsd.org>

next in thread | raw e-mail | index | archive | help
Author: yongari
Date: Fri Jan 14 22:36:33 2011
New Revision: 217429
URL: http://svn.freebsd.org/changeset/base/217429

Log:
  MFC r216284:
    r184610 changed the way how TX frames are handled on AX88178 and
    AX88772 controllers. ASIX added a new feature for AX88178/AX88772
    controllers which allows combining multiple TX frames into a single
    big frame. This was to overcome one of USB limitation where it
    can't generate more than 8k interrupts/sec which in turn means USB
    ethernet controllers can not send more than 8k packets per second.
    Using ASIX's feature greatly enhanced TX performance(more than 3~4
    times) compared to 7.x driver. However it seems r184610 removed
    boundary checking for buffered frames which in turn caused
    instability issues under certain conditions. In addition, using
    ASIX's feature triggered another issue which made USB controller
    hang under certain conditions. Restarting ethernet controller
    didn't help under this hang condition and unplugging and replugging
    the controller was the only solution. I believe there is a silicon
    bug in TX frame combining feature on AX88178/AX88772 controllers.
  
    To address these issues, reintroduce the boundary checking for both
    AX88178 and AX88772 after copying a frame to USB buffer and do not
    use ASIX's multiple frame combining feature. Instead, use USB
    controller's multi-frame transmit capability to enhance TX
    performance as suggested by Hans[1].
    This should fix a long standing axe(4) instability issues reported
    on AX88772 and AX88178 controllers. While I'm here remove
    unnecessary TX frame length check since upper stack always
    guarantee the size of a frame to be less than MCLBYTES.
  
    Special thanks to Derrick Brashear who tried numerous patches
    during last 4 months and waited real fix with patience. Without
    this enthusiastic support, patience and H/W donation I couldn't fix
    it since I was not able to trigger the issue on my box.
  
    Suggested by:	hselasky [1]
    Tested by:	Derrick Brashear (shadow <> gmail dot com>
    H/W donated by:	Derrick Brashear (shadow <> gmail dot com>
    PR:		usb/140883

Modified:
  stable/8/sys/dev/usb/net/if_axe.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/dev/usb/net/if_axe.c
==============================================================================
--- stable/8/sys/dev/usb/net/if_axe.c	Fri Jan 14 22:33:12 2011	(r217428)
+++ stable/8/sys/dev/usb/net/if_axe.c	Fri Jan 14 22:36:33 2011	(r217429)
@@ -200,7 +200,8 @@ static const struct usb_config axe_confi
 		.type = UE_BULK,
 		.endpoint = UE_ADDR_ANY,
 		.direction = UE_DIR_OUT,
-		.bufsize = AXE_BULK_BUF_SIZE,
+		.frames = 16,
+		.bufsize = 16 * MCLBYTES,
 		.flags = {.pipe_bof = 1,.force_short_xfer = 1,},
 		.callback = axe_bulk_write_callback,
 		.timeout = 10000,	/* 10 seconds */
@@ -939,7 +940,7 @@ axe_bulk_write_callback(struct usb_xfer 
 	struct ifnet *ifp = uether_getifp(&sc->sc_ue);
 	struct usb_page_cache *pc;
 	struct mbuf *m;
-	int pos;
+	int nframes, pos;
 
 	switch (USB_GET_STATE(xfer)) {
 	case USB_ST_TRANSFERRED:
@@ -956,40 +957,34 @@ tr_setup:
 			 */
 			return;
 		}
-		pos = 0;
-		pc = usbd_xfer_get_frame(xfer, 0);
-
-		while (1) {
 
+		for (nframes = 0; nframes < 16 &&
+		    !IFQ_DRV_IS_EMPTY(&ifp->if_snd); nframes++) {
 			IFQ_DRV_DEQUEUE(&ifp->if_snd, m);
-
-			if (m == NULL) {
-				if (pos > 0)
-					break;	/* send out data */
-				return;
-			}
-			if (m->m_pkthdr.len > MCLBYTES) {
-				m->m_pkthdr.len = MCLBYTES;
-			}
+			if (m == NULL)
+				break;
+			usbd_xfer_set_frame_offset(xfer, nframes * MCLBYTES,
+			    nframes);
+			pos = 0;
+			pc = usbd_xfer_get_frame(xfer, nframes);
 			if (AXE_IS_178_FAMILY(sc)) {
-
 				hdr.len = htole16(m->m_pkthdr.len);
 				hdr.ilen = ~hdr.len;
-
 				usbd_copy_in(pc, pos, &hdr, sizeof(hdr));
-
 				pos += sizeof(hdr);
-
-				/*
-				 * NOTE: Some drivers force a short packet
-				 * by appending a dummy header with zero
-				 * length at then end of the USB transfer.
-				 * This driver uses the
-				 * USB_FORCE_SHORT_XFER flag instead.
-				 */
+				usbd_m_copy_in(pc, pos, m, 0, m->m_pkthdr.len);
+				pos += m->m_pkthdr.len;
+				if ((pos % 512) == 0) {
+					hdr.len = 0;
+					hdr.ilen = 0xffff;
+					usbd_copy_in(pc, pos, &hdr,
+					    sizeof(hdr));
+					pos += sizeof(hdr);
+				}
+			} else {
+				usbd_m_copy_in(pc, pos, m, 0, m->m_pkthdr.len);
+				pos += m->m_pkthdr.len;
 			}
-			usbd_m_copy_in(pc, pos, m, 0, m->m_pkthdr.len);
-			pos += m->m_pkthdr.len;
 
 			/*
 			 * XXX
@@ -1010,22 +1005,16 @@ tr_setup:
 
 			m_freem(m);
 
-			if (AXE_IS_178_FAMILY(sc)) {
-				if (pos > (AXE_BULK_BUF_SIZE - MCLBYTES - sizeof(hdr))) {
-					/* send out frame(s) */
-					break;
-				}
-			} else {
-				/* send out frame */
-				break;
-			}
+			/* Set frame length. */
+			usbd_xfer_set_frame_len(xfer, nframes, pos);
+		}
+		if (nframes != 0) {
+			usbd_xfer_set_frames(xfer, nframes);
+			usbd_transfer_submit(xfer);
+			ifp->if_drv_flags |= IFF_DRV_OACTIVE;
 		}
-
-		usbd_xfer_set_frame_len(xfer, 0, pos);
-		usbd_transfer_submit(xfer);
-		ifp->if_drv_flags |= IFF_DRV_OACTIVE;
 		return;
-
+		/* NOTREACHED */
 	default:			/* Error */
 		DPRINTFN(11, "transfer error, %s\n",
 		    usbd_errstr(error));



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201101142236.p0EMaXqi004444>