From 5138826b52f8bc441770069b1b305e1d1426ae9e Mon Sep 17 00:00:00 2001 From: Michael Chan Date: Tue, 25 Jan 2011 22:14:50 +0000 Subject: [PATCH 1/8] cnic: Fix big endian bug The chip's page tables did not set up properly on big endian machines, causing EEH errors on PPC machines. Reported-by: Breno Leitao Signed-off-by: Michael Chan Signed-off-by: David S. Miller --- drivers/net/cnic.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c index 263a2944566f..7ff170cbc7dc 100644 --- a/drivers/net/cnic.c +++ b/drivers/net/cnic.c @@ -699,13 +699,13 @@ static void cnic_free_dma(struct cnic_dev *dev, struct cnic_dma *dma) static void cnic_setup_page_tbl(struct cnic_dev *dev, struct cnic_dma *dma) { int i; - u32 *page_table = dma->pgtbl; + __le32 *page_table = (__le32 *) dma->pgtbl; for (i = 0; i < dma->num_pages; i++) { /* Each entry needs to be in big endian format. */ - *page_table = (u32) ((u64) dma->pg_map_arr[i] >> 32); + *page_table = cpu_to_le32((u64) dma->pg_map_arr[i] >> 32); page_table++; - *page_table = (u32) dma->pg_map_arr[i]; + *page_table = cpu_to_le32(dma->pg_map_arr[i] & 0xffffffff); page_table++; } } @@ -713,13 +713,13 @@ static void cnic_setup_page_tbl(struct cnic_dev *dev, struct cnic_dma *dma) static void cnic_setup_page_tbl_le(struct cnic_dev *dev, struct cnic_dma *dma) { int i; - u32 *page_table = dma->pgtbl; + __le32 *page_table = (__le32 *) dma->pgtbl; for (i = 0; i < dma->num_pages; i++) { /* Each entry needs to be in little endian format. */ - *page_table = dma->pg_map_arr[i] & 0xffffffff; + *page_table = cpu_to_le32(dma->pg_map_arr[i] & 0xffffffff); page_table++; - *page_table = (u32) ((u64) dma->pg_map_arr[i] >> 32); + *page_table = cpu_to_le32((u64) dma->pg_map_arr[i] >> 32); page_table++; } } From 4bb9ebc78097376b3734c6d3001a96aecac0f7bb Mon Sep 17 00:00:00 2001 From: Michael Chan Date: Tue, 25 Jan 2011 22:14:51 +0000 Subject: [PATCH 2/8] bnx2: Eliminate AER error messages on systems not supporting it On PPC for example, AER is not supported and we see unnecessary AER error message without this patch: bnx2 0003:01:00.1: pci_cleanup_aer_uncorrect_error_status failed 0xfffffffb Reported-by: Breno Leitao Signed-off-by: Michael Chan Signed-off-by: David S. Miller --- drivers/net/bnx2.c | 17 +++++++++-------- drivers/net/bnx2.h | 1 + 2 files changed, 10 insertions(+), 8 deletions(-) diff --git a/drivers/net/bnx2.c b/drivers/net/bnx2.c index 99e7652e2169..0ba59d5aeb7f 100644 --- a/drivers/net/bnx2.c +++ b/drivers/net/bnx2.c @@ -7966,11 +7966,8 @@ bnx2_init_board(struct pci_dev *pdev, struct net_device *dev) /* AER (Advanced Error Reporting) hooks */ err = pci_enable_pcie_error_reporting(pdev); - if (err) { - dev_err(&pdev->dev, "pci_enable_pcie_error_reporting " - "failed 0x%x\n", err); - /* non-fatal, continue */ - } + if (!err) + bp->flags |= BNX2_FLAG_AER_ENABLED; } else { bp->pcix_cap = pci_find_capability(pdev, PCI_CAP_ID_PCIX); @@ -8233,8 +8230,10 @@ bnx2_init_board(struct pci_dev *pdev, struct net_device *dev) return 0; err_out_unmap: - if (bp->flags & BNX2_FLAG_PCIE) + if (bp->flags & BNX2_FLAG_AER_ENABLED) { pci_disable_pcie_error_reporting(pdev); + bp->flags &= ~BNX2_FLAG_AER_ENABLED; + } if (bp->regview) { iounmap(bp->regview); @@ -8422,8 +8421,10 @@ bnx2_remove_one(struct pci_dev *pdev) kfree(bp->temp_stats_blk); - if (bp->flags & BNX2_FLAG_PCIE) + if (bp->flags & BNX2_FLAG_AER_ENABLED) { pci_disable_pcie_error_reporting(pdev); + bp->flags &= ~BNX2_FLAG_AER_ENABLED; + } free_netdev(dev); @@ -8539,7 +8540,7 @@ static pci_ers_result_t bnx2_io_slot_reset(struct pci_dev *pdev) } rtnl_unlock(); - if (!(bp->flags & BNX2_FLAG_PCIE)) + if (!(bp->flags & BNX2_FLAG_AER_ENABLED)) return result; err = pci_cleanup_aer_uncorrect_error_status(pdev); diff --git a/drivers/net/bnx2.h b/drivers/net/bnx2.h index 5488a2e82fe9..f459fb2f9add 100644 --- a/drivers/net/bnx2.h +++ b/drivers/net/bnx2.h @@ -6741,6 +6741,7 @@ struct bnx2 { #define BNX2_FLAG_JUMBO_BROKEN 0x00000800 #define BNX2_FLAG_CAN_KEEP_VLAN 0x00001000 #define BNX2_FLAG_BROKEN_STATS 0x00002000 +#define BNX2_FLAG_AER_ENABLED 0x00004000 struct bnx2_napi bnx2_napi[BNX2_MAX_MSIX_VEC]; From 389f2a18c6a2a5531ac5a155c3ba25784065b1cb Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 26 Jan 2011 00:04:18 +0000 Subject: [PATCH 3/8] econet: remove compiler warnings MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit net/econet/af_econet.c: In function ‘econet_sendmsg’: net/econet/af_econet.c:494: warning: label ‘error’ defined but not used net/econet/af_econet.c:268: warning: unused variable ‘sk’ Signed-off-by: Eric Dumazet Acked-by: Phil Blundell Signed-off-by: David S. Miller --- net/econet/af_econet.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/econet/af_econet.c b/net/econet/af_econet.c index 15dcc1a586b4..0c2826337919 100644 --- a/net/econet/af_econet.c +++ b/net/econet/af_econet.c @@ -265,13 +265,13 @@ static void ec_tx_done(struct sk_buff *skb, int result) static int econet_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg, size_t len) { - struct sock *sk = sock->sk; struct sockaddr_ec *saddr=(struct sockaddr_ec *)msg->msg_name; struct net_device *dev; struct ec_addr addr; int err; unsigned char port, cb; #if defined(CONFIG_ECONET_AUNUDP) || defined(CONFIG_ECONET_NATIVE) + struct sock *sk = sock->sk; struct sk_buff *skb; struct ec_cb *eb; #endif @@ -488,10 +488,10 @@ static int econet_sendmsg(struct kiocb *iocb, struct socket *sock, error_free_buf: vfree(userbuf); +error: #else err = -EPROTOTYPE; #endif - error: mutex_unlock(&econet_mutex); return err; From e0ce4af920eb028f38bfd680b1d733f4c7a0b7cf Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Thu, 27 Jan 2011 04:14:03 +0000 Subject: [PATCH 4/8] xen: netfront: handle incoming GSO SKBs which are not CHECKSUM_PARTIAL The Linux network stack expects all GSO SKBs to have ip_summed == CHECKSUM_PARTIAL (which implies that the frame contains a partial checksum) and the Xen network ring protocol similarly expects an SKB which has GSO set to also have NETRX_csum_blank (which also implies a partial checksum). However there have been cases of buggy guests which mark a frame as GSO but do not set csum_blank. If we detect that we a receiving such a frame (which manifests as ip_summed != PARTIAL && skb_is_gso) then force the SKB to partial and recalculate the checksum, since we cannot rely on the peer having done so if they have not set csum_blank. Add an ethtool stat to track occurances of this event. Signed-off-by: Ian Campbell Cc: Jeremy Fitzhardinge Cc: David Miller Cc: xen-devel@lists.xensource.com Cc: netdev@vger.kernel.org Signed-off-by: David S. Miller --- drivers/net/xen-netfront.c | 96 ++++++++++++++++++++++++++++++++++---- 1 file changed, 88 insertions(+), 8 deletions(-) diff --git a/drivers/net/xen-netfront.c b/drivers/net/xen-netfront.c index 546de5749824..da1f12120346 100644 --- a/drivers/net/xen-netfront.c +++ b/drivers/net/xen-netfront.c @@ -120,6 +120,9 @@ struct netfront_info { unsigned long rx_pfn_array[NET_RX_RING_SIZE]; struct multicall_entry rx_mcl[NET_RX_RING_SIZE+1]; struct mmu_update rx_mmu[NET_RX_RING_SIZE]; + + /* Statistics */ + int rx_gso_checksum_fixup; }; struct netfront_rx_info { @@ -770,11 +773,29 @@ static RING_IDX xennet_fill_frags(struct netfront_info *np, return cons; } -static int skb_checksum_setup(struct sk_buff *skb) +static int checksum_setup(struct net_device *dev, struct sk_buff *skb) { struct iphdr *iph; unsigned char *th; int err = -EPROTO; + int recalculate_partial_csum = 0; + + /* + * A GSO SKB must be CHECKSUM_PARTIAL. However some buggy + * peers can fail to set NETRXF_csum_blank when sending a GSO + * frame. In this case force the SKB to CHECKSUM_PARTIAL and + * recalculate the partial checksum. + */ + if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) { + struct netfront_info *np = netdev_priv(dev); + np->rx_gso_checksum_fixup++; + skb->ip_summed = CHECKSUM_PARTIAL; + recalculate_partial_csum = 1; + } + + /* A non-CHECKSUM_PARTIAL SKB does not require setup. */ + if (skb->ip_summed != CHECKSUM_PARTIAL) + return 0; if (skb->protocol != htons(ETH_P_IP)) goto out; @@ -788,9 +809,23 @@ static int skb_checksum_setup(struct sk_buff *skb) switch (iph->protocol) { case IPPROTO_TCP: skb->csum_offset = offsetof(struct tcphdr, check); + + if (recalculate_partial_csum) { + struct tcphdr *tcph = (struct tcphdr *)th; + tcph->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, + skb->len - iph->ihl*4, + IPPROTO_TCP, 0); + } break; case IPPROTO_UDP: skb->csum_offset = offsetof(struct udphdr, check); + + if (recalculate_partial_csum) { + struct udphdr *udph = (struct udphdr *)th; + udph->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, + skb->len - iph->ihl*4, + IPPROTO_UDP, 0); + } break; default: if (net_ratelimit()) @@ -829,13 +864,11 @@ static int handle_incoming_queue(struct net_device *dev, /* Ethernet work: Delayed to here as it peeks the header. */ skb->protocol = eth_type_trans(skb, dev); - if (skb->ip_summed == CHECKSUM_PARTIAL) { - if (skb_checksum_setup(skb)) { - kfree_skb(skb); - packets_dropped++; - dev->stats.rx_errors++; - continue; - } + if (checksum_setup(dev, skb)) { + kfree_skb(skb); + packets_dropped++; + dev->stats.rx_errors++; + continue; } dev->stats.rx_packets++; @@ -1632,12 +1665,59 @@ static void netback_changed(struct xenbus_device *dev, } } +static const struct xennet_stat { + char name[ETH_GSTRING_LEN]; + u16 offset; +} xennet_stats[] = { + { + "rx_gso_checksum_fixup", + offsetof(struct netfront_info, rx_gso_checksum_fixup) + }, +}; + +static int xennet_get_sset_count(struct net_device *dev, int string_set) +{ + switch (string_set) { + case ETH_SS_STATS: + return ARRAY_SIZE(xennet_stats); + default: + return -EINVAL; + } +} + +static void xennet_get_ethtool_stats(struct net_device *dev, + struct ethtool_stats *stats, u64 * data) +{ + void *np = netdev_priv(dev); + int i; + + for (i = 0; i < ARRAY_SIZE(xennet_stats); i++) + data[i] = *(int *)(np + xennet_stats[i].offset); +} + +static void xennet_get_strings(struct net_device *dev, u32 stringset, u8 * data) +{ + int i; + + switch (stringset) { + case ETH_SS_STATS: + for (i = 0; i < ARRAY_SIZE(xennet_stats); i++) + memcpy(data + i * ETH_GSTRING_LEN, + xennet_stats[i].name, ETH_GSTRING_LEN); + break; + } +} + static const struct ethtool_ops xennet_ethtool_ops = { .set_tx_csum = ethtool_op_set_tx_csum, .set_sg = xennet_set_sg, .set_tso = xennet_set_tso, .get_link = ethtool_op_get_link, + + .get_sset_count = xennet_get_sset_count, + .get_ethtool_stats = xennet_get_ethtool_stats, + .get_strings = xennet_get_strings, }; #ifdef CONFIG_SYSFS From 9eb710797a21fa4a9e09ae9c86c4b3ec9d291c2d Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Wed, 26 Jan 2011 00:45:42 +0000 Subject: [PATCH 5/8] dl2k: nulify fraginfo after unmap Patch fixes: "DMA-API: device driver tries to free an invalid DMA memory address" warning reported here: https://bugzilla.redhat.com/show_bug.cgi?id=639824 Reported-by: Frantisek Hanzlik Signed-off-by: Stanislaw Gruszka Signed-off-by: David S. Miller --- drivers/net/dl2k.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/dl2k.c b/drivers/net/dl2k.c index e1a8216ff692..c05db6046050 100644 --- a/drivers/net/dl2k.c +++ b/drivers/net/dl2k.c @@ -1753,8 +1753,6 @@ rio_close (struct net_device *dev) /* Free all the skbuffs in the queue. */ for (i = 0; i < RX_RING_SIZE; i++) { - np->rx_ring[i].status = 0; - np->rx_ring[i].fraginfo = 0; skb = np->rx_skbuff[i]; if (skb) { pci_unmap_single(np->pdev, @@ -1763,6 +1761,8 @@ rio_close (struct net_device *dev) dev_kfree_skb (skb); np->rx_skbuff[i] = NULL; } + np->rx_ring[i].status = 0; + np->rx_ring[i].fraginfo = 0; } for (i = 0; i < TX_RING_SIZE; i++) { skb = np->tx_skbuff[i]; From 6d3a9a685465986d7653c5abbc0b24681e7c44d7 Mon Sep 17 00:00:00 2001 From: Kurt Van Dijck Date: Wed, 26 Jan 2011 04:55:24 +0000 Subject: [PATCH 6/8] net: fix validate_link_af in rtnetlink core I'm testing an API that uses IFLA_AF_SPEC attribute. In the rtnetlink core , the set_link_af() member of the rtnl_af_ops struct receives the nested attribute (as I expected), but the validate_link_af() member receives the parent attribute. IMO, this patch fixes this. Signed-off-by: Kurt Van Dijck Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 750db57f3bb3..31459ef13ca2 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -1121,8 +1121,7 @@ static int validate_linkmsg(struct net_device *dev, struct nlattr *tb[]) return -EOPNOTSUPP; if (af_ops->validate_link_af) { - err = af_ops->validate_link_af(dev, - tb[IFLA_AF_SPEC]); + err = af_ops->validate_link_af(dev, af); if (err < 0) return err; } From c2aa3665cf8510b1665ee2f5a9525cf7be6dec4f Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 25 Jan 2011 23:18:38 +0000 Subject: [PATCH 7/8] net: add kmemcheck annotation in __alloc_skb() pskb_expand_head() triggers a kmemcheck warning when copy of skb_shared_info is done in pskb_expand_head() This is because destructor_arg field is not necessarily initialized at this point. Add kmemcheck_annotate_variable() call in __alloc_skb() to instruct kmemcheck this is a normal situation. Resolves bugzilla.kernel.org 27212 Reference: https://bugzilla.kernel.org/show_bug.cgi?id=27212 Reported-by: Christian Casteyde Signed-off-by: Eric Dumazet CC: Andrew Morton Signed-off-by: David S. Miller --- net/core/skbuff.c | 1 + 1 file changed, 1 insertion(+) diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 7cd1bc86d591..d883dcc78b6b 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -210,6 +210,7 @@ struct sk_buff *__alloc_skb(unsigned int size, gfp_t gfp_mask, shinfo = skb_shinfo(skb); memset(shinfo, 0, offsetof(struct skb_shared_info, dataref)); atomic_set(&shinfo->dataref, 1); + kmemcheck_annotate_variable(shinfo->destructor_arg); if (fclone) { struct sk_buff *child = skb + 1; From 8f2771f2b85aea4d0f9a0137ad3b63d1173c0962 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 27 Jan 2011 14:55:22 -0800 Subject: [PATCH 8/8] ipv6: Remove route peer binding assertions. They are bogus. The basic idea is that I wanted to make sure that prefixed routes never bind to peers. The test I used was whether RTF_CACHE was set. But first of all, the RTF_CACHE flag is set at different spots depending upon which ip6_rt_copy() caller you're talking about. I've validated all of the code paths, and even in the future where we bind peers more aggressively (for route metric COW'ing) we never bind to prefix'd routes, only fully specified ones. This even applies when addrconf or icmp6 routes are allocated. Signed-off-by: David S. Miller --- net/ipv6/route.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/net/ipv6/route.c b/net/ipv6/route.c index 1534508f6c68..28a85fc63cb8 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -194,7 +194,6 @@ static void ip6_dst_destroy(struct dst_entry *dst) in6_dev_put(idev); } if (peer) { - BUG_ON(!(rt->rt6i_flags & RTF_CACHE)); rt->rt6i_peer = NULL; inet_putpeer(peer); } @@ -204,9 +203,6 @@ void rt6_bind_peer(struct rt6_info *rt, int create) { struct inet_peer *peer; - if (WARN_ON(!(rt->rt6i_flags & RTF_CACHE))) - return; - peer = inet_getpeer_v6(&rt->rt6i_dst.addr, create); if (peer && cmpxchg(&rt->rt6i_peer, NULL, peer) != NULL) inet_putpeer(peer);