linux-dev/include/net, branch master

netlink: introduce bigendian integer types

2022-11-02T04:29:06Z

Jakub reported that the addition of the "network_byte_order" member in struct nla_policy increases size of 32bit platforms. Instead of scraping the bit from elsewhere Johannes suggested to add explicit NLA_BE types instead, so do this here. NLA_POLICY_MAX_BE() macro is removed again, there is no need for it: NLA_POLICY_MAX(NLA_BE.., ..) will do the right thing. NLA_BE64 can be added later. Fixes: 08724ef69907 ("netlink: introduce NLA_POLICY_MAX_BE") Reported-by: Jakub Kicinski Suggested-by: Johannes Berg Signed-off-by: Florian Westphal Link: https://lore.kernel.org/r/20221031123407.9158-1-fw@strlen.de Signed-off-by: Jakub Kicinski

net: remove SOCK_SUPPORT_ZC from sockmap

2022-10-29T03:21:25Z

sockmap replaces ->sk_prot with its own callbacks, we should remove SOCK_SUPPORT_ZC as the new proto doesn't support msghdr::ubuf_info. Cc: # 6.0 Reported-by: Jakub Kicinski Fixes: e993ffe3da4bc ("net: flag sockets supporting msghdr originated zerocopy") Signed-off-by: Pavel Begunkov Signed-off-by: Jakub Kicinski

netlink: hide validation union fields from kdoc

2022-10-29T03:19:15Z

Mark the validation fields as private, users shouldn't set them directly and they are too complicated to explain in a more succinct way (there's already a long explanation in the comment above). The strict_start_type field is set directly and has a dedicated comment so move that above the "private" section. Link: https://lore.kernel.org/r/20221027212107.2639255-1-kuba@kernel.org Signed-off-by: Jakub Kicinski

genetlink: piggy back on resv_op to default to a reject policy

2022-10-25T02:08:46Z

To keep backward compatibility we used to leave attribute parsing to the family if no policy is specified. This becomes tedious as we move to more strict validation. Families must define reject all policies if they don't want any attributes accepted. Piggy back on the resv_start_op field as the switchover point. AFAICT only ethtool has added new commands since the resv_start_op was defined, and it has per-op policies so this should be a no-op. Nonetheless the patch should still go into v6.1 for consistency. Link: https://lore.kernel.org/all/20221019125745.3f2e7659@kernel.org/ Link: https://lore.kernel.org/r/20221021193532.1511293-1-kuba@kernel.org Signed-off-by: Jakub Kicinski

net-memcg: avoid stalls when under memory pressure

2022-10-24T17:35:09Z

As Shakeel explains the commit under Fixes had the unintended side-effect of no longer pre-loading the cached memory allowance. Even tho we previously dropped the first packet received when over memory limit - the consecutive ones would get thru by using the cache. The charging was happening in batches of 128kB, so we'd let in 128kB (truesize) worth of packets per one drop. After the change we no longer force charge, there will be no cache filling side effects. This causes significant drops and connection stalls for workloads which use a lot of page cache, since we can't reclaim page cache under GFP_NOWAIT. Some of the latency can be recovered by improving SACK reneg handling but nowhere near enough to get back to the pre-5.15 performance (the application I'm experimenting with still sees 5-10x worst latency). Apply the suggested workaround of using GFP_ATOMIC. We will now be more permissive than previously as we'll drop _no_ packets in softirq when under pressure. But I can't think of any good and simple way to address that within networking. Link: https://lore.kernel.org/all/20221012163300.795e7b86@kernel.org/ Suggested-by: Shakeel Butt Fixes: 4b1327be9fe5 ("net-memcg: pass in gfp_t mask to mem_cgroup_charge_skmem()") Acked-by: Shakeel Butt Acked-by: Roman Gushchin Link: https://lore.kernel.org/r/20221021160304.1362511-1-kuba@kernel.org Signed-off-by: Jakub Kicinski

Merge tag 'net-6.1-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

2022-10-21T00:24:59Z

Pull networking fixes from Paolo Abeni: "Including fixes from netfilter. Current release - regressions: - revert "net: fix cpu_max_bits_warn() usage in netif_attrmask_next{,_and}" - revert "net: sched: fq_codel: remove redundant resource cleanup in fq_codel_init()" - dsa: uninitialized variable in dsa_slave_netdevice_event() - eth: sunhme: uninitialized variable in happy_meal_init() Current release - new code bugs: - eth: octeontx2: fix resource not freed after malloc Previous releases - regressions: - sched: fix return value of qdisc ingress handling on success - sched: fix race condition in qdisc_graft() - udp: update reuse->has_conns under reuseport_lock. - tls: strp: make sure the TCP skbs do not have overlapping data - hsr: avoid possible NULL deref in skb_clone() - tipc: fix an information leak in tipc_topsrv_kern_subscr - phylink: add mac_managed_pm in phylink_config structure - eth: i40e: fix DMA mappings leak - eth: hyperv: fix a RX-path warning - eth: mtk: fix memory leaks Previous releases - always broken: - sched: cake: fix null pointer access issue when cake_init() fails" * tag 'net-6.1-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (43 commits) net: phy: dp83822: disable MDI crossover status change interrupt net: sched: fix race condition in qdisc_graft() net: hns: fix possible memory leak in hnae_ae_register() wwan_hwsim: fix possible memory leak in wwan_hwsim_dev_new() sfc: include vport_id in filter spec hash and equal() genetlink: fix kdoc warnings selftests: add selftest for chaining of tc ingress handling to egress net: Fix return value of qdisc ingress handling on success net: sched: sfb: fix null pointer access issue when sfb_init() fails Revert "net: sched: fq_codel: remove redundant resource cleanup in fq_codel_init()" net: sched: cake: fix null pointer access issue when cake_init() fails ethernet: marvell: octeontx2 Fix resource not freed after malloc netfilter: nf_tables: relax NFTA_SET_ELEM_KEY_END set flags requirements netfilter: rpfilter/fib: Set ->flowic_uid correctly for user namespaces. ionic: catch NULL pointer issue on reconfig net: hsr: avoid possible NULL deref in skb_clone() bnxt_en: fix memory leak in bnxt_nvm_test() ip6mr: fix UAF issue in ip6mr_sk_done() when addrconf_init_net() failed udp: Update reuse->has_conns under reuseport_lock. net: ethernet: mediatek: ppe: Remove the unused function mtk_foe_entry_usable() ...

genetlink: fix kdoc warnings

2022-10-19T20:08:30Z

Address a bunch of kdoc warnings: include/net/genetlink.h:81: warning: Function parameter or member 'module' not described in 'genl_family' include/net/genetlink.h:243: warning: expecting prototype for struct genl_info. Prototype was for struct genl_dumpit_info instead include/net/genetlink.h:419: warning: Function parameter or member 'net' not described in 'genlmsg_unicast' include/net/genetlink.h:438: warning: expecting prototype for gennlmsg_data(). Prototype was for genlmsg_data() instead include/net/genetlink.h:244: warning: Function parameter or member 'op' not described in 'genl_dumpit_info' Link: https://lore.kernel.org/r/20221018231310.1040482-1-kuba@kernel.org Signed-off-by: Jakub Kicinski

udp: Update reuse->has_conns under reuseport_lock.

2022-10-18T08:17:18Z

When we call connect() for a UDP socket in a reuseport group, we have to update sk->sk_reuseport_cb->has_conns to 1. Otherwise, the kernel could select a unconnected socket wrongly for packets sent to the connected socket. However, the current way to set has_conns is illegal and possible to trigger that problem. reuseport_has_conns() changes has_conns under rcu_read_lock(), which upgrades the RCU reader to the updater. Then, it must do the update under the updater's lock, reuseport_lock, but it doesn't for now. For this reason, there is a race below where we fail to set has_conns resulting in the wrong socket selection. To avoid the race, let's split the reader and updater with proper locking. cpu1 cpu2 +----+ +----+ __ip[46]_datagram_connect() reuseport_grow() . . |- reuseport_has_conns(sk, true) |- more_reuse = __reuseport_alloc(more_socks_size) | . | | |- rcu_read_lock() | |- reuse = rcu_dereference(sk->sk_reuseport_cb) | | | | | /* reuse->has_conns == 0 here */ | | |- more_reuse->has_conns = reuse->has_conns | |- reuse->has_conns = 1 | /* more_reuse->has_conns SHOULD BE 1 HERE */ | | | | | |- rcu_assign_pointer(reuse->socks[i]->sk_reuseport_cb, | | | more_reuse) | `- rcu_read_unlock() `- kfree_rcu(reuse, rcu) | |- sk->sk_state = TCP_ESTABLISHED Note the likely(reuse) in reuseport_has_conns_set() is always true, but we put the test there for ease of review. [0] For the record, usually, sk_reuseport_cb is changed under lock_sock(). The only exception is reuseport_grow() & TCP reqsk migration case. 1) shutdown() TCP listener, which is moved into the latter part of reuse->socks[] to migrate reqsk. 2) New listen() overflows reuse->socks[] and call reuseport_grow(). 3) reuse->max_socks overflows u16 with the new listener. 4) reuseport_grow() pops the old shutdown()ed listener from the array and update its sk->sk_reuseport_cb as NULL without lock_sock(). shutdown()ed TCP sk->sk_reuseport_cb can be changed without lock_sock(), but, reuseport_has_conns_set() is called only for UDP under lock_sock(), so likely(reuse) never be false in reuseport_has_conns_set(). [0]: https://lore.kernel.org/netdev/CANn89iLja=eQHbsM_Ta2sQF0tOGU8vAGrh_izRuuHjuO1ouUag@mail.gmail.com/ Fixes: acdcecc61285 ("udp: correct reuseport selection with connected sockets") Signed-off-by: Kuniyuki Iwashima Link: https://lore.kernel.org/r/20221014182625.89913-1-kuniyu@amazon.com Signed-off-by: Paolo Abeni

Merge tag 'random-6.1-rc1-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/crng/random

2022-10-16T22:27:07Z

Pull more random number generator updates from Jason Donenfeld: "This time with some large scale treewide cleanups. The intent of this pull is to clean up the way callers fetch random integers. The current rules for doing this right are: - If you want a secure or an insecure random u64, use get_random_u64() - If you want a secure or an insecure random u32, use get_random_u32() The old function prandom_u32() has been deprecated for a while now and is just a wrapper around get_random_u32(). Same for get_random_int(). - If you want a secure or an insecure random u16, use get_random_u16() - If you want a secure or an insecure random u8, use get_random_u8() - If you want secure or insecure random bytes, use get_random_bytes(). The old function prandom_bytes() has been deprecated for a while now and has long been a wrapper around get_random_bytes() - If you want a non-uniform random u32, u16, or u8 bounded by a certain open interval maximum, use prandom_u32_max() I say "non-uniform", because it doesn't do any rejection sampling or divisions. Hence, it stays within the prandom_*() namespace, not the get_random_*() namespace. I'm currently investigating a "uniform" function for 6.2. We'll see what comes of that. By applying these rules uniformly, we get several benefits: - By using prandom_u32_max() with an upper-bound that the compiler can prove at compile-time is ≤65536 or ≤256, internally get_random_u16() or get_random_u8() is used, which wastes fewer batched random bytes, and hence has higher throughput. - By using prandom_u32_max() instead of %, when the upper-bound is not a constant, division is still avoided, because prandom_u32_max() uses a faster multiplication-based trick instead. - By using get_random_u16() or get_random_u8() in cases where the return value is intended to indeed be a u16 or a u8, we waste fewer batched random bytes, and hence have higher throughput. This series was originally done by hand while I was on an airplane without Internet. Later, Kees and I worked on retroactively figuring out what could be done with Coccinelle and what had to be done manually, and then we split things up based on that. So while this touches a lot of files, the actual amount of code that's hand fiddled is comfortably small" * tag 'random-6.1-rc1-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/crng/random: prandom: remove unused functions treewide: use get_random_bytes() when possible treewide: use get_random_u32() when possible treewide: use get_random_{u8,u16}() when possible, part 2 treewide: use get_random_{u8,u16}() when possible, part 1 treewide: use prandom_u32_max() when possible, part 2 treewide: use prandom_u32_max() when possible, part 1

Merge tag 'net-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

2022-10-13T17:51:01Z

Pull networking fixes from Jakub Kicinski: "Including fixes from netfilter, and wifi. Current release - regressions: - Revert "net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs", it may cause crashes when the qdisc is reconfigured - inet: ping: fix splat due to packet allocation refactoring in inet - tcp: clean up kernel listener's reqsk in inet_twsk_purge(), fix UAF due to races when per-netns hash table is used Current release - new code bugs: - eth: adin1110: check in netdev_event that netdev belongs to driver - fixes for PTR_ERR() vs NULL bugs in driver code, from Dan and co. Previous releases - regressions: - ipv4: handle attempt to delete multipath route when fib_info contains an nh reference, avoid oob access - wifi: fix handful of bugs in the new Multi-BSSID code - wifi: mt76: fix rate reporting / throughput regression on mt7915 and newer, fix checksum offload - wifi: iwlwifi: mvm: fix double list_add at iwl_mvm_mac_wake_tx_queue (other cases) - wifi: mac80211: do not drop packets smaller than the LLC-SNAP header on fast-rx Previous releases - always broken: - ieee802154: don't warn zero-sized raw_sendmsg() - ipv6: ping: fix wrong checksum for large frames - mctp: prevent double key removal and unref - tcp/udp: fix memory leaks and races around IPV6_ADDRFORM - hv_netvsc: fix race between VF offering and VF association message Misc: - remove -Warray-bounds silencing in the drivers, compilers fixed" * tag 'net-6.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (73 commits) sunhme: fix an IS_ERR() vs NULL check in probe net: marvell: prestera: fix a couple NULL vs IS_ERR() checks kcm: avoid potential race in kcm_tx_work tcp: Clean up kernel listener's reqsk in inet_twsk_purge() net: phy: micrel: Fixes FIELD_GET assertion openvswitch: add nf_ct_is_confirmed check before assigning the helper tcp: Fix data races around icsk->icsk_af_ops. ipv6: Fix data races around sk->sk_prot. tcp/udp: Call inet6_destroy_sock() in IPv6 sk->sk_destruct(). udp: Call inet6_destroy_sock() in setsockopt(IPV6_ADDRFORM). tcp/udp: Fix memory leak in ipv6_renew_options(). mctp: prevent double key removal and unref selftests: netfilter: Fix nft_fib.sh for all.rp_filter=1 netfilter: rpfilter/fib: Populate flowic_l3mdev field selftests: netfilter: Test reverse path filtering net/mlx5: Make ASO poll CQ usable in atomic context tcp: cdg: allow tcp_cdg_release() to be called multiple times inet: ping: fix recent breakage ipv6: ping: fix wrong checksum for large frames net: ethernet: ti: am65-cpsw: set correct devlink flavour for unused ports ...