From be325f08c432ae5ac6d6594d163e1899cdf202df Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 9 Dec 2024 10:07:45 +0000 Subject: [PATCH 1/3] rtnetlink: add ndo_fdb_dump_context rtnl_fdb_dump() and various ndo_fdb_dump() helpers share a hidden layout of cb->ctx. Before switching rtnl_fdb_dump() to for_each_netdev_dump() in the following patch, make this more explicit. Signed-off-by: Eric Dumazet Reviewed-by: Ido Schimmel Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20241209100747.2269613-2-edumazet@google.com Signed-off-by: Jakub Kicinski --- .../ethernet/freescale/dpaa2/dpaa2-switch.c | 3 ++- drivers/net/ethernet/mscc/ocelot_net.c | 3 ++- drivers/net/vxlan/vxlan_core.c | 5 ++-- include/linux/rtnetlink.h | 7 +++++ net/bridge/br_fdb.c | 3 ++- net/core/rtnetlink.c | 26 ++++++++++--------- net/dsa/user.c | 3 ++- 7 files changed, 32 insertions(+), 18 deletions(-) diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c index a293b08f36d4..147a93bf9fa9 100644 --- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c +++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-switch.c @@ -780,13 +780,14 @@ struct ethsw_dump_ctx { static int dpaa2_switch_fdb_dump_nl(struct fdb_dump_entry *entry, struct ethsw_dump_ctx *dump) { + struct ndo_fdb_dump_context *ctx = (void *)dump->cb->ctx; int is_dynamic = entry->type & DPSW_FDB_ENTRY_DINAMIC; u32 portid = NETLINK_CB(dump->cb->skb).portid; u32 seq = dump->cb->nlh->nlmsg_seq; struct nlmsghdr *nlh; struct ndmsg *ndm; - if (dump->idx < dump->cb->args[2]) + if (dump->idx < ctx->fdb_idx) goto skip; nlh = nlmsg_put(dump->skb, portid, seq, RTM_NEWNEIGH, diff --git a/drivers/net/ethernet/mscc/ocelot_net.c b/drivers/net/ethernet/mscc/ocelot_net.c index 558e03301aa8..8d48468cddd7 100644 --- a/drivers/net/ethernet/mscc/ocelot_net.c +++ b/drivers/net/ethernet/mscc/ocelot_net.c @@ -758,12 +758,13 @@ static int ocelot_port_fdb_do_dump(const unsigned char *addr, u16 vid, bool is_static, void *data) { struct ocelot_dump_ctx *dump = data; + struct ndo_fdb_dump_context *ctx = (void *)dump->cb->ctx; u32 portid = NETLINK_CB(dump->cb->skb).portid; u32 seq = dump->cb->nlh->nlmsg_seq; struct nlmsghdr *nlh; struct ndmsg *ndm; - if (dump->idx < dump->cb->args[2]) + if (dump->idx < ctx->fdb_idx) goto skip; nlh = nlmsg_put(dump->skb, portid, seq, RTM_NEWNEIGH, diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c index 43cf672b7b9f..0c356e0a61ef 100644 --- a/drivers/net/vxlan/vxlan_core.c +++ b/drivers/net/vxlan/vxlan_core.c @@ -1352,6 +1352,7 @@ static int vxlan_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb, struct net_device *dev, struct net_device *filter_dev, int *idx) { + struct ndo_fdb_dump_context *ctx = (void *)cb->ctx; struct vxlan_dev *vxlan = netdev_priv(dev); unsigned int h; int err = 0; @@ -1364,7 +1365,7 @@ static int vxlan_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb, struct vxlan_rdst *rd; if (rcu_access_pointer(f->nh)) { - if (*idx < cb->args[2]) + if (*idx < ctx->fdb_idx) goto skip_nh; err = vxlan_fdb_info(skb, vxlan, f, NETLINK_CB(cb->skb).portid, @@ -1381,7 +1382,7 @@ static int vxlan_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb, } list_for_each_entry_rcu(rd, &f->remotes, list) { - if (*idx < cb->args[2]) + if (*idx < ctx->fdb_idx) goto skip; err = vxlan_fdb_info(skb, vxlan, f, diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h index 811ce44113f6..c43cffb014a7 100644 --- a/include/linux/rtnetlink.h +++ b/include/linux/rtnetlink.h @@ -178,6 +178,13 @@ void rtnetlink_init(void); void __rtnl_unlock(void); void rtnl_kfree_skbs(struct sk_buff *head, struct sk_buff *tail); +/* Shared by rtnl_fdb_dump() and various ndo_fdb_dump() helpers. */ +struct ndo_fdb_dump_context { + unsigned long s_h; + unsigned long s_idx; + unsigned long fdb_idx; +}; + extern int ndo_dflt_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb, struct net_device *dev, diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c index 82bac2426631..902694c0ce64 100644 --- a/net/bridge/br_fdb.c +++ b/net/bridge/br_fdb.c @@ -955,6 +955,7 @@ int br_fdb_dump(struct sk_buff *skb, struct net_device *filter_dev, int *idx) { + struct ndo_fdb_dump_context *ctx = (void *)cb->ctx; struct net_bridge *br = netdev_priv(dev); struct net_bridge_fdb_entry *f; int err = 0; @@ -970,7 +971,7 @@ int br_fdb_dump(struct sk_buff *skb, rcu_read_lock(); hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) { - if (*idx < cb->args[2]) + if (*idx < ctx->fdb_idx) goto skip; if (filter_dev && (!f->dst || f->dst->dev != filter_dev)) { if (filter_dev != dev) diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index ab5f201bf0ab..453cc8bf18fb 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -4762,15 +4762,16 @@ static int nlmsg_populate_fdb(struct sk_buff *skb, int *idx, struct netdev_hw_addr_list *list) { + struct ndo_fdb_dump_context *ctx = (void *)cb->ctx; struct netdev_hw_addr *ha; - int err; u32 portid, seq; + int err; portid = NETLINK_CB(cb->skb).portid; seq = cb->nlh->nlmsg_seq; list_for_each_entry(ha, &list->list, list) { - if (*idx < cb->args[2]) + if (*idx < ctx->fdb_idx) goto skip; err = nlmsg_populate_fdb_fill(skb, dev, ha->addr, 0, @@ -4909,10 +4910,9 @@ static int valid_fdb_dump_legacy(const struct nlmsghdr *nlh, static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) { - struct net_device *dev; - struct net_device *br_dev = NULL; - const struct net_device_ops *ops = NULL; - const struct net_device_ops *cops = NULL; + const struct net_device_ops *ops = NULL, *cops = NULL; + struct ndo_fdb_dump_context *ctx = (void *)cb->ctx; + struct net_device *dev, *br_dev = NULL; struct net *net = sock_net(skb->sk); struct hlist_head *head; int brport_idx = 0; @@ -4922,6 +4922,8 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) int err = 0; int fidx = 0; + NL_ASSERT_CTX_FITS(struct ndo_fdb_dump_context); + if (cb->strict_check) err = valid_fdb_dump_strict(cb->nlh, &br_idx, &brport_idx, cb->extack); @@ -4939,8 +4941,8 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) ops = br_dev->netdev_ops; } - s_h = cb->args[0]; - s_idx = cb->args[1]; + s_h = ctx->s_h; + s_idx = ctx->s_idx; for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) { idx = 0; @@ -4992,7 +4994,7 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) cops = NULL; /* reset fdb offset to 0 for rest of the interfaces */ - cb->args[2] = 0; + ctx->fdb_idx = 0; fidx = 0; cont: idx++; @@ -5000,9 +5002,9 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) } out: - cb->args[0] = h; - cb->args[1] = idx; - cb->args[2] = fidx; + ctx->s_h = h; + ctx->s_idx = idx; + ctx->fdb_idx = fidx; return skb->len; } diff --git a/net/dsa/user.c b/net/dsa/user.c index 06c30a9e29ff..c736c019e2af 100644 --- a/net/dsa/user.c +++ b/net/dsa/user.c @@ -515,12 +515,13 @@ dsa_user_port_fdb_do_dump(const unsigned char *addr, u16 vid, bool is_static, void *data) { struct dsa_user_dump_ctx *dump = data; + struct ndo_fdb_dump_context *ctx = (void *)dump->cb->ctx; u32 portid = NETLINK_CB(dump->cb->skb).portid; u32 seq = dump->cb->nlh->nlmsg_seq; struct nlmsghdr *nlh; struct ndmsg *ndm; - if (dump->idx < dump->cb->args[2]) + if (dump->idx < ctx->fdb_idx) goto skip; nlh = nlmsg_put(dump->skb, portid, seq, RTM_NEWNEIGH, From 53970a05f799087e2dd2005973609188504e7fcc Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 9 Dec 2024 10:07:46 +0000 Subject: [PATCH 2/3] rtnetlink: switch rtnl_fdb_dump() to for_each_netdev_dump() This is the last netdev iterator still using net->dev_index_head[]. Convert to modern for_each_netdev_dump() for better scalability, and use common patterns in our stack. Following patch in this series removes the pad field in struct ndo_fdb_dump_context. Signed-off-by: Eric Dumazet Reviewed-by: Ido Schimmel Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20241209100747.2269613-3-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/linux/rtnetlink.h | 4 +- net/core/rtnetlink.c | 100 +++++++++++++++----------------------- 2 files changed, 41 insertions(+), 63 deletions(-) diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h index c43cffb014a7..5546571c2553 100644 --- a/include/linux/rtnetlink.h +++ b/include/linux/rtnetlink.h @@ -180,8 +180,8 @@ void rtnl_kfree_skbs(struct sk_buff *head, struct sk_buff *tail); /* Shared by rtnl_fdb_dump() and various ndo_fdb_dump() helpers. */ struct ndo_fdb_dump_context { - unsigned long s_h; - unsigned long s_idx; + unsigned long ifindex; + unsigned long pad; unsigned long fdb_idx; }; diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 453cc8bf18fb..8fe252c298a2 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -4914,13 +4914,10 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) struct ndo_fdb_dump_context *ctx = (void *)cb->ctx; struct net_device *dev, *br_dev = NULL; struct net *net = sock_net(skb->sk); - struct hlist_head *head; int brport_idx = 0; int br_idx = 0; - int h, s_h; - int idx = 0, s_idx; - int err = 0; int fidx = 0; + int err; NL_ASSERT_CTX_FITS(struct ndo_fdb_dump_context); @@ -4941,69 +4938,50 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb) ops = br_dev->netdev_ops; } - s_h = ctx->s_h; - s_idx = ctx->s_idx; + for_each_netdev_dump(net, dev, ctx->ifindex) { + if (brport_idx && (dev->ifindex != brport_idx)) + continue; - for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) { - idx = 0; - head = &net->dev_index_head[h]; - hlist_for_each_entry(dev, head, index_hlist) { - - if (brport_idx && (dev->ifindex != brport_idx)) + if (!br_idx) { /* user did not specify a specific bridge */ + if (netif_is_bridge_port(dev)) { + br_dev = netdev_master_upper_dev_get(dev); + cops = br_dev->netdev_ops; + } + } else { + if (dev != br_dev && + !netif_is_bridge_port(dev)) continue; - if (!br_idx) { /* user did not specify a specific bridge */ - if (netif_is_bridge_port(dev)) { - br_dev = netdev_master_upper_dev_get(dev); - cops = br_dev->netdev_ops; - } - } else { - if (dev != br_dev && - !netif_is_bridge_port(dev)) - continue; - - if (br_dev != netdev_master_upper_dev_get(dev) && - !netif_is_bridge_master(dev)) - continue; - cops = ops; - } - - if (idx < s_idx) - goto cont; - - if (netif_is_bridge_port(dev)) { - if (cops && cops->ndo_fdb_dump) { - err = cops->ndo_fdb_dump(skb, cb, - br_dev, dev, - &fidx); - if (err == -EMSGSIZE) - goto out; - } - } - - if (dev->netdev_ops->ndo_fdb_dump) - err = dev->netdev_ops->ndo_fdb_dump(skb, cb, - dev, NULL, - &fidx); - else - err = ndo_dflt_fdb_dump(skb, cb, dev, NULL, - &fidx); - if (err == -EMSGSIZE) - goto out; - - cops = NULL; - - /* reset fdb offset to 0 for rest of the interfaces */ - ctx->fdb_idx = 0; - fidx = 0; -cont: - idx++; + if (br_dev != netdev_master_upper_dev_get(dev) && + !netif_is_bridge_master(dev)) + continue; + cops = ops; } + + if (netif_is_bridge_port(dev)) { + if (cops && cops->ndo_fdb_dump) { + err = cops->ndo_fdb_dump(skb, cb, br_dev, dev, + &fidx); + if (err == -EMSGSIZE) + break; + } + } + + if (dev->netdev_ops->ndo_fdb_dump) + err = dev->netdev_ops->ndo_fdb_dump(skb, cb, dev, NULL, + &fidx); + else + err = ndo_dflt_fdb_dump(skb, cb, dev, NULL, &fidx); + if (err == -EMSGSIZE) + break; + + cops = NULL; + + /* reset fdb offset to 0 for rest of the interfaces */ + ctx->fdb_idx = 0; + fidx = 0; } -out: - ctx->s_h = h; - ctx->s_idx = idx; ctx->fdb_idx = fidx; return skb->len; From 53a6d8912372fc23ea82cc7a49eb59047aa0a650 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 9 Dec 2024 10:07:47 +0000 Subject: [PATCH 3/3] rtnetlink: remove pad field in ndo_fdb_dump_context I chose to remove this field in a separate patch to ease potential bisection, in case one ndo_fdb_dump() is still using the old way (cb->args[2] instead of ctx->fdb_idx) Signed-off-by: Eric Dumazet Reviewed-by: Ido Schimmel Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20241209100747.2269613-4-edumazet@google.com Signed-off-by: Jakub Kicinski --- include/linux/rtnetlink.h | 1 - 1 file changed, 1 deletion(-) diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h index 5546571c2553..3b9d132cbc9e 100644 --- a/include/linux/rtnetlink.h +++ b/include/linux/rtnetlink.h @@ -181,7 +181,6 @@ void rtnl_kfree_skbs(struct sk_buff *head, struct sk_buff *tail); /* Shared by rtnl_fdb_dump() and various ndo_fdb_dump() helpers. */ struct ndo_fdb_dump_context { unsigned long ifindex; - unsigned long pad; unsigned long fdb_idx; };