fix Jumbo frames on strict alignment architectures by allocating a new mbuf and

copying the Ethernet header to the new mbuf. The new mbuf is then prepended
into the existing mbuf chain.

From FreeBSD

ok reyk@ pascoe@ jason@

1 files changed, 56 insertions, 61 deletions

diff --git a/sys/dev/pci/if_em.c b/sys/dev/pci/if_em.c
index b2f4fb72cd7..253199aeb92 100644
--- a/sys/dev/pci/if_em.c
+++ b/sys/dev/pci/if_em.c
@@ -31,7 +31,7 @@ POSSIBILITY OF SUCH DAMAGE.
 
 ***************************************************************************/
 
-/* $OpenBSD: if_em.c,v 1.130 2006/05/28 23:38:49 brad Exp $ */
+/* $OpenBSD: if_em.c,v 1.131 2006/05/31 02:09:18 brad Exp $ */
 /* $FreeBSD: if_em.c,v 1.46 2004/09/29 18:28:28 mlaier Exp $ */
 
 #include <dev/pci/if_em.h>
@@ -143,6 +143,9 @@ void em_clean_transmit_interrupts(struct em_softc *);
 int  em_allocate_receive_structures(struct em_softc *);
 int  em_allocate_transmit_structures(struct em_softc *);
 void em_process_receive_interrupts(struct em_softc *, int);
+#ifdef __STRICT_ALIGNMENT
+int  em_fixup_rx(struct em_softc *);
+#endif
 void em_receive_checksum(struct em_softc *, struct em_rx_desc *,
 			 struct mbuf *);
 void em_transmit_checksum_setup(struct em_softc *, struct mbuf *,
@@ -265,9 +268,6 @@ em_attach(struct device *parent, struct device *self, void *aux)
 	 * Set the max frame size assuming standard Ethernet
 	 * sized frames.
 	 */
-#ifdef __STRICT_ALIGNMENT
-	sc->hw.max_frame_size = ETHER_MAX_LEN;
-#else
 	switch (sc->hw.mac_type) {
 		case em_82573:
 		{
@@ -294,7 +294,6 @@ em_attach(struct device *parent, struct device *self, void *aux)
 			sc->hw.max_frame_size =
 			    MAX_JUMBO_FRAME_SIZE;
 	}
-#endif
 
 	sc->hw.min_frame_size = 
 	    ETHER_MIN_LEN + ETHER_CRC_LEN;
@@ -2051,11 +2050,6 @@ em_get_buf(int i, struct em_softc *sc, struct mbuf *nmp)
 		mp->m_next = NULL;
 	}
 
-#ifdef __STRICT_ALIGNMENT
-	if (ifp->if_mtu <= ETHERMTU)
-		m_adj(mp, ETHER_ALIGN);
-#endif
-
 	rx_buffer = &sc->rx_buffer_area[i];
 
 	/*
@@ -2225,10 +2219,8 @@ em_initialize_receive_unit(struct em_softc *sc)
 		break;
 	}
 
-#ifndef __STRICT_ALIGNMENT
 	if (sc->hw.mac_type != em_82573)
 		reg_rctl |= E1000_RCTL_LPE;
-#endif
 
 	/* Enable 82543 Receive Checksum Offload for TCP and UDP */
 	if (sc->hw.mac_type >= em_82543) {
@@ -2368,55 +2360,6 @@ em_process_receive_interrupts(struct em_softc *sc, int count)
 			/* Assign correct length to the current fragment */
 			mp->m_len = len;
 
-#ifdef __STRICT_ALIGNMENT
-			/*
-			 * The Ethernet payload is not 32-bit aligned when
-			 * Jumbo packets are enabled, so on architectures with
-			 * strict alignment we need to shift the entire packet
-			 * ETHER_ALIGN bytes. Ugh.
-			 */
-			if (ifp->if_mtu > ETHERMTU) {
-				unsigned char tmp_align_buf[ETHER_ALIGN];
-				int tmp_align_buf_len = 0;
-
-				if (prev_len_adj > sc->align_buf_len)
-					prev_len_adj -= sc->align_buf_len;
-				else
-					prev_len_adj = 0;
-
-				if (mp->m_len > MCLBYTES - ETHER_ALIGN) {
-					bcopy(mp->m_data +
-					    (MCLBYTES - ETHER_ALIGN),
-					    &tmp_align_buf,
-					    ETHER_ALIGN);
-					tmp_align_buf_len = mp->m_len -
-					    (MCLBYTES - ETHER_ALIGN);
-					mp->m_len -= ETHER_ALIGN;
-				} 
-
-				if (mp->m_len) {
-					bcopy(mp->m_data,
-					    mp->m_data + ETHER_ALIGN,
-					    mp->m_len);
-					if (!sc->align_buf_len)
-						mp->m_data += ETHER_ALIGN;
-				}
-
-				if (sc->align_buf_len) {
-					mp->m_len += sc->align_buf_len;
-					bcopy(&sc->align_buf,
-					    mp->m_data,
-					    sc->align_buf_len);
-				}
-
-				if (tmp_align_buf_len) 
-					bcopy(&tmp_align_buf,
-					    &sc->align_buf,
-					    tmp_align_buf_len);
-				sc->align_buf_len = tmp_align_buf_len;
-			}
-#endif /* __STRICT_ALIGNMENT */
-
 			if (sc->fmp == NULL) {
 				mp->m_pkthdr.len = mp->m_len;
 				sc->fmp = mp;	 /* Store the first mbuf */
@@ -2442,6 +2385,9 @@ em_process_receive_interrupts(struct em_softc *sc, int count)
 				ifp->if_ipackets++;
 				em_receive_checksum(sc, current_desc,
 					    sc->fmp);
+#ifdef __STRICT_ALIGNMENT
+				em_fixup_rx(sc);
+#endif
 				m = sc->fmp;
 				sc->fmp = NULL;
 				sc->lmp = NULL;
@@ -2489,6 +2435,55 @@ em_process_receive_interrupts(struct em_softc *sc, int count)
 	E1000_WRITE_REG(&sc->hw, RDT, i);
 }
 
+
+
+#ifdef __STRICT_ALIGNMENT
+/*
+ * When Jumbo frames are enabled we should realign the entire payload on
+ * strict alignment architecures. This is a serious design mistake of the
+ * 8254x chipset as it nullifies DMA operations. 8254x allows the RX buffer
+ * size to be 2048/4096/8192/16384. What we really want is 2048 - ETHER_ALIGN
+ * to align its payload. On non strict alignment architectures 8254x still
+ * performs unaligned memory access which will reduce the performance too. To
+ * avoid copying over an entire frame to align, we allocate a new mbuf and
+ * copy the Ethernet header to the new mbuf. The new mbuf is then prepended
+ * into the existing mbuf chain.
+ *
+ * Be aware, best performance of the 8254x chipset is achived only when Jumbo
+ * frames are not used at all on strict alignment architectures.
+ */
+int
+em_fixup_rx(struct em_softc *sc)
+{
+	struct mbuf *m, *n;
+	int error = 0;
+
+	m = sc->fmp;
+	if (m->m_len <= (MCLBYTES - ETHER_HDR_LEN)) {
+		bcopy(m->m_data, m->m_data + ETHER_HDR_LEN, m->m_len);
+		m->m_data += ETHER_HDR_LEN;
+	} else {
+		MGETHDR(n, M_DONTWAIT, MT_DATA);
+		if (n != NULL) {
+			bcopy(m->m_data, n->m_data, ETHER_HDR_LEN);
+			m->m_data += ETHER_HDR_LEN;
+			m->m_len -= ETHER_HDR_LEN;
+			n->m_len = ETHER_HDR_LEN;
+			M_MOVE_PKTHDR(n, m);
+			n->m_next = m;
+			sc->fmp = n;
+		} else {
+			sc->dropped_pkts++;
+			m_freem(sc->fmp);
+			sc->fmp = NULL;
+			error = ENOMEM;
+		}
+	}
+
+	return (error);
+}
+#endif
+
 /*********************************************************************
  *
  *  Verify that the hardware indicated that the checksum is valid.