Get rid if em_align. This approach used to make sense, but now that the

hardware rx mtu always gets set to the maximum supported value we will hit
it for every received packet.  Instead, use a larger mbuf cluster size on
strict alignment architectures such that we can always m_adj to make sure the
packets are properly aligned.  This wastes some memory but simplifies things
considerably.  Hopefully we can reduce the spillage in the near future by
taking advantage of recent improvements in the pool code.

ok mpi@, mikeb@, dlg@

2 files changed, 16 insertions, 69 deletions

diff --git a/sys/dev/pci/if_em.c b/sys/dev/pci/if_em.c
index 0e530f24d70..dace553b0dd 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 1.300 2015/08/21 09:16:06 kettenis Exp $ */
+/* $OpenBSD: if_em.c,v 1.301 2015/08/26 09:17:20 kettenis Exp $ */
 /* $FreeBSD: if_em.c,v 1.46 2004/09/29 18:28:28 mlaier Exp $ */
 
 #include <dev/pci/if_em.h>
@@ -229,11 +229,6 @@ void em_disable_aspm(struct em_softc *);
 void em_txeof(struct em_softc *);
 int  em_allocate_receive_structures(struct em_softc *);
 int  em_allocate_transmit_structures(struct em_softc *);
-#ifdef __STRICT_ALIGNMENT
-void em_realign(struct em_softc *, struct mbuf *, u_int16_t *);
-#else
-#define em_realign(a, b, c) /* a, b, c */
-#endif
 int  em_rxfill(struct em_softc *);
 void em_rxeof(struct em_softc *);
 void em_receive_checksum(struct em_softc *, struct em_rx_desc *,
@@ -708,7 +703,7 @@ em_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
 
 	case SIOCGIFRXR:
 		error = if_rxr_ioctl((struct if_rxrinfo *)ifr->ifr_data,
-		    NULL, MCLBYTES, &sc->rx_ring);
+		    NULL, EM_MCLBYTES, &sc->rx_ring);
 		break;
 
 	default:
@@ -2528,14 +2523,15 @@ em_get_buf(struct em_softc *sc, int i)
 		return (ENOBUFS);
 	}
 
-	m = MCLGETI(NULL, M_DONTWAIT, NULL, MCLBYTES);
+	m = MCLGETI(NULL, M_DONTWAIT, NULL, EM_MCLBYTES);
 	if (!m) {
 		sc->mbuf_cluster_failed++;
 		return (ENOBUFS);
 	}
-	m->m_len = m->m_pkthdr.len = MCLBYTES;
-	if (sc->hw.max_frame_size <= (MCLBYTES - ETHER_ALIGN))
-		m_adj(m, ETHER_ALIGN);
+	m->m_len = m->m_pkthdr.len = EM_MCLBYTES;
+#ifdef __STRICT_ALIGNMENT
+	m_adj(m, ETHER_ALIGN);
+#endif
 
 	error = bus_dmamap_load_mbuf(sc->rxtag, pkt->map, m, BUS_DMA_NOWAIT);
 	if (error) {
@@ -2584,8 +2580,8 @@ em_allocate_receive_structures(struct em_softc *sc)
 
 	rx_buffer = sc->rx_buffer_area;
 	for (i = 0; i < sc->num_rx_desc; i++, rx_buffer++) {
-		error = bus_dmamap_create(sc->rxtag, MCLBYTES, 1,
-		    MCLBYTES, 0, BUS_DMA_NOWAIT, &rx_buffer->map);
+		error = bus_dmamap_create(sc->rxtag, EM_MCLBYTES, 1,
+		    EM_MCLBYTES, 0, BUS_DMA_NOWAIT, &rx_buffer->map);
 		if (error != 0) {
 			printf("%s: em_allocate_receive_structures: "
 			    "bus_dmamap_create failed; error %u\n",
@@ -2785,53 +2781,6 @@ em_free_receive_structures(struct em_softc *sc)
 	}
 }
 
-#ifdef __STRICT_ALIGNMENT
-void
-em_realign(struct em_softc *sc, struct mbuf *m, u_int16_t *prev_len_adj)
-{
-	unsigned char tmp_align_buf[ETHER_ALIGN];
-	int tmp_align_buf_len = 0;
-
-	/*
-	 * 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 (sc->hw.max_frame_size <= (MCLBYTES - ETHER_ALIGN))
-		return;
-
-	if (*prev_len_adj > sc->align_buf_len)
-		*prev_len_adj -= sc->align_buf_len;
-	else
-		*prev_len_adj = 0;
-
-	if (m->m_len > (MCLBYTES - ETHER_ALIGN)) {
-		bcopy(m->m_data + (MCLBYTES - ETHER_ALIGN),
-		    &tmp_align_buf, ETHER_ALIGN);
-		tmp_align_buf_len = m->m_len -
-		    (MCLBYTES - ETHER_ALIGN);
-		m->m_len -= ETHER_ALIGN;
-	} 
-
-	if (m->m_len) {
-		bcopy(m->m_data, m->m_data + ETHER_ALIGN, m->m_len);
-		if (!sc->align_buf_len)
-			m->m_data += ETHER_ALIGN;
-	}
-
-	if (sc->align_buf_len) {
-		m->m_len += sc->align_buf_len;
-		bcopy(&sc->align_buf, m->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 */
-
 int
 em_rxfill(struct em_softc *sc)
 {
@@ -2967,8 +2916,6 @@ em_rxeof(struct em_softc *sc)
 			/* Assign correct length to the current fragment */
 			m->m_len = len;
 
-			em_realign(sc, m, &prev_len_adj); /* STRICT_ALIGN */
-
 			if (sc->fmp == NULL) {
 				m->m_pkthdr.len = m->m_len;
 				sc->fmp = m;	 /* Store the first mbuf */
diff --git a/sys/dev/pci/if_em.h b/sys/dev/pci/if_em.h
index da4144784fc..f58051644b7 100644
--- a/sys/dev/pci/if_em.h
+++ b/sys/dev/pci/if_em.h
@@ -32,7 +32,7 @@ POSSIBILITY OF SUCH DAMAGE.
 ***************************************************************************/
 
 /* $FreeBSD: if_em.h,v 1.26 2004/09/01 23:22:41 pdeuskar Exp $ */
-/* $OpenBSD: if_em.h,v 1.55 2015/08/21 09:16:06 kettenis Exp $ */
+/* $OpenBSD: if_em.h,v 1.56 2015/08/26 09:17:20 kettenis Exp $ */
 
 #ifndef _EM_H_DEFINED_
 #define _EM_H_DEFINED_
@@ -266,6 +266,12 @@ typedef int	boolean_t;
 #define EM_RXBUFFER_8192	8192
 #define EM_RXBUFFER_16384	16384
 
+#ifdef __STRICT_ALIGNMENT
+#define EM_MCLBYTES		(EM_RXBUFFER_2048 + ETHER_ALIGN)
+#else
+#define EM_MCLBYTES		EM_RXBUFFER_2048
+#endif
+
 #define EM_MAX_SCATTER		64
 #define EM_TSO_SIZE		65535
 
@@ -324,12 +330,6 @@ struct em_softc {
 	struct timeout	timer_handle;
 	struct timeout	tx_fifo_timer_handle;
 
-#ifdef __STRICT_ALIGNMENT
-	/* Used for carrying forward alignment adjustments */
-	unsigned char	align_buf[ETHER_ALIGN];	/* tail of unaligned packet */
-	u_int8_t	align_buf_len;		/* bytes in tail */
-#endif /* __STRICT_ALIGNMENT */
-
 	/* Info about the board itself */
 	u_int32_t	part_num;
 	u_int8_t	link_active;