linux/net/netlink/af_netlink.c

2954 lines
70 KiB

netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
af_netlink: Fix shift out of bounds in group mask calculation When a netlink message is received, netlink_recvmsg() fills in the address of the sender. One of the fields is the 32-bit bitfield nl_groups, which carries the multicast group on which the message was received. The least significant bit corresponds to group 1, and therefore the highest group that the field can represent is 32. Above that, the UB sanitizer flags the out-of-bounds shift attempts. Which bits end up being set in such case is implementation defined, but it's either going to be a wrong non-zero value, or zero, which is at least not misleading. Make the latter choice deterministic by always setting to 0 for higher-numbered multicast groups. To get information about membership in groups >= 32, userspace is expected to use nl_pktinfo control messages[0], which are enabled by NETLINK_PKTINFO socket option. [0] https://lwn.net/Articles/147608/ The way to trigger this issue is e.g. through monitoring the BRVLAN group: # bridge monitor vlan & # ip link add name br type bridge Which produces the following citation: UBSAN: shift-out-of-bounds in net/netlink/af_netlink.c:162:19 shift exponent 32 is too large for 32-bit type 'int' Fixes: f7fa9b10edbb ("[NETLINK]: Support dynamic number of multicast groups per netlink family") Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Link: https://lore.kernel.org/r/2bef6aabf201d1fc16cca139a744700cff9dcb04.1647527635.git.petrm@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-03-17 15:53:06 +01:00
netlink: Fix kernel-infoleak-after-free in __skb_datagram_iter syzbot reported the following uninit-value access issue [1]: netlink_to_full_skb() creates a new `skb` and puts the `skb->data` passed as a 1st arg of netlink_to_full_skb() onto new `skb`. The data size is specified as `len` and passed to skb_put_data(). This `len` is based on `skb->end` that is not data offset but buffer offset. The `skb->end` contains data and tailroom. Since the tailroom is not initialized when the new `skb` created, KMSAN detects uninitialized memory area when copying the data. This patch resolved this issue by correct the len from `skb->end` to `skb->len`, which is the actual data offset. BUG: KMSAN: kernel-infoleak-after-free in instrument_copy_to_user include/linux/instrumented.h:114 [inline] BUG: KMSAN: kernel-infoleak-after-free in copy_to_user_iter lib/iov_iter.c:24 [inline] BUG: KMSAN: kernel-infoleak-after-free in iterate_ubuf include/linux/iov_iter.h:29 [inline] BUG: KMSAN: kernel-infoleak-after-free in iterate_and_advance2 include/linux/iov_iter.h:245 [inline] BUG: KMSAN: kernel-infoleak-after-free in iterate_and_advance include/linux/iov_iter.h:271 [inline] BUG: KMSAN: kernel-infoleak-after-free in _copy_to_iter+0x364/0x2520 lib/iov_iter.c:186 instrument_copy_to_user include/linux/instrumented.h:114 [inline] copy_to_user_iter lib/iov_iter.c:24 [inline] iterate_ubuf include/linux/iov_iter.h:29 [inline] iterate_and_advance2 include/linux/iov_iter.h:245 [inline] iterate_and_advance include/linux/iov_iter.h:271 [inline] _copy_to_iter+0x364/0x2520 lib/iov_iter.c:186 copy_to_iter include/linux/uio.h:197 [inline] simple_copy_to_iter+0x68/0xa0 net/core/datagram.c:532 __skb_datagram_iter+0x123/0xdc0 net/core/datagram.c:420 skb_copy_datagram_iter+0x5c/0x200 net/core/datagram.c:546 skb_copy_datagram_msg include/linux/skbuff.h:3960 [inline] packet_recvmsg+0xd9c/0x2000 net/packet/af_packet.c:3482 sock_recvmsg_nosec net/socket.c:1044 [inline] sock_recvmsg net/socket.c:1066 [inline] sock_read_iter+0x467/0x580 net/socket.c:1136 call_read_iter include/linux/fs.h:2014 [inline] new_sync_read fs/read_write.c:389 [inline] vfs_read+0x8f6/0xe00 fs/read_write.c:470 ksys_read+0x20f/0x4c0 fs/read_write.c:613 __do_sys_read fs/read_write.c:623 [inline] __se_sys_read fs/read_write.c:621 [inline] __x64_sys_read+0x93/0xd0 fs/read_write.c:621 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0x44/0x110 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x63/0x6b Uninit was stored to memory at: skb_put_data include/linux/skbuff.h:2622 [inline] netlink_to_full_skb net/netlink/af_netlink.c:181 [inline] __netlink_deliver_tap_skb net/netlink/af_netlink.c:298 [inline] __netlink_deliver_tap+0x5be/0xc90 net/netlink/af_netlink.c:325 netlink_deliver_tap net/netlink/af_netlink.c:338 [inline] netlink_deliver_tap_kernel net/netlink/af_netlink.c:347 [inline] netlink_unicast_kernel net/netlink/af_netlink.c:1341 [inline] netlink_unicast+0x10f1/0x1250 net/netlink/af_netlink.c:1368 netlink_sendmsg+0x1238/0x13d0 net/netlink/af_netlink.c:1910 sock_sendmsg_nosec net/socket.c:730 [inline] __sock_sendmsg net/socket.c:745 [inline] ____sys_sendmsg+0x9c2/0xd60 net/socket.c:2584 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2638 __sys_sendmsg net/socket.c:2667 [inline] __do_sys_sendmsg net/socket.c:2676 [inline] __se_sys_sendmsg net/socket.c:2674 [inline] __x64_sys_sendmsg+0x307/0x490 net/socket.c:2674 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0x44/0x110 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x63/0x6b Uninit was created at: free_pages_prepare mm/page_alloc.c:1087 [inline] free_unref_page_prepare+0xb0/0xa40 mm/page_alloc.c:2347 free_unref_page_list+0xeb/0x1100 mm/page_alloc.c:2533 release_pages+0x23d3/0x2410 mm/swap.c:1042 free_pages_and_swap_cache+0xd9/0xf0 mm/swap_state.c:316 tlb_batch_pages_flush mm/mmu_gather.c:98 [inline] tlb_flush_mmu_free mm/mmu_gather.c:293 [inline] tlb_flush_mmu+0x6f5/0x980 mm/mmu_gather.c:300 tlb_finish_mmu+0x101/0x260 mm/mmu_gather.c:392 exit_mmap+0x49e/0xd30 mm/mmap.c:3321 __mmput+0x13f/0x530 kernel/fork.c:1349 mmput+0x8a/0xa0 kernel/fork.c:1371 exit_mm+0x1b8/0x360 kernel/exit.c:567 do_exit+0xd57/0x4080 kernel/exit.c:858 do_group_exit+0x2fd/0x390 kernel/exit.c:1021 __do_sys_exit_group kernel/exit.c:1032 [inline] __se_sys_exit_group kernel/exit.c:1030 [inline] __x64_sys_exit_group+0x3c/0x50 kernel/exit.c:1030 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0x44/0x110 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x63/0x6b Bytes 3852-3903 of 3904 are uninitialized Memory access of size 3904 starts at ffff88812ea1e000 Data copied to user address 0000000020003280 CPU: 1 PID: 5043 Comm: syz-executor297 Not tainted 6.7.0-rc5-syzkaller-00047-g5bd7ef53ffe5 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 11/10/2023 Fixes: 1853c9496460 ("netlink, mmap: transform mmap skb into full skb on taps") Reported-and-tested-by: syzbot+34ad5fab48f7bf510349@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=34ad5fab48f7bf510349 [1] Signed-off-by: Ryosuke Yasuoka <ryasuoka@redhat.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20240221074053.1794118-1-ryasuoka@redhat.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-02-21 16:40:48 +09:00
netlink: remove mmapped netlink support mmapped netlink has a number of unresolved issues: - TX zerocopy support had to be disabled more than a year ago via commit 4682a0358639b29cf ("netlink: Always copy on mmap TX.") because the content of the mmapped area can change after netlink attribute validation but before message processing. - RX support was implemented mainly to speed up nfqueue dumping packet payload to userspace. However, since commit ae08ce0021087a5d812d2 ("netfilter: nfnetlink_queue: zero copy support") we avoid one copy with the socket-based interface too (via the skb_zerocopy helper). The other problem is that skbs attached to mmaped netlink socket behave different from normal skbs: - they don't have a shinfo area, so all functions that use skb_shinfo() (e.g. skb_clone) cannot be used. - reserving headroom prevents userspace from seeing the content as it expects message to start at skb->head. See for instance commit aa3a022094fa ("netlink: not trim skb for mmaped socket when dump"). - skbs handed e.g. to netlink_ack must have non-NULL skb->sk, else we crash because it needs the sk to check if a tx ring is attached. Also not obvious, leads to non-intuitive bug fixes such as 7c7bdf359 ("netfilter: nfnetlink: use original skbuff when acking batches"). mmaped netlink also didn't play nicely with the skb_zerocopy helper used by nfqueue and openvswitch. Daniel Borkmann fixed this via commit 6bb0fef489f6 ("netlink, mmap: fix edge-case leakages in nf queue zero-copy")' but at the cost of also needing to provide remaining length to the allocation function. nfqueue also has problems when used with mmaped rx netlink: - mmaped netlink doesn't allow use of nfqueue batch verdict messages. Problem is that in the mmap case, the allocation time also determines the ordering in which the frame will be seen by userspace (A allocating before B means that A is located in earlier ring slot, but this also means that B might get a lower sequence number then A since seqno is decided later. To fix this we would need to extend the spinlocked region to also cover the allocation and message setup which isn't desirable. - nfqueue can now be configured to queue large (GSO) skbs to userspace. Queing GSO packets is faster than having to force a software segmentation in the kernel, so this is a desirable option. However, with a mmap based ring one has to use 64kb per ring slot element, else mmap has to fall back to the socket path (NL_MMAP_STATUS_COPY) for all large packets. To use the mmap interface, userspace not only has to probe for mmap netlink support, it also has to implement a recv/socket receive path in order to handle messages that exceed the size of an rx ring element. Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Ken-ichirou MATSUZAWA <chamaken@gmail.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Cc: Patrick McHardy <kaber@trash.net> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-02-18 15:03:24 +01:00
netlink: specify netlink packet direction for nlmon In order to facilitate development for netlink protocol dissector, fill the unused field skb->pkt_type of the cloned skb with a hint of the address space of the new owner (receiver) socket in the notion of "to kernel" resp. "to user". At the time we invoke __netlink_deliver_tap_skb(), we already have set the new skb owner via netlink_skb_set_owner_r(), so we can use that for netlink_is_kernel() probing. In normal PF_PACKET network traffic, this field denotes if the packet is destined for us (PACKET_HOST), if it's broadcast (PACKET_BROADCAST), etc. As we only have 3 bit reserved, we can use the value (= 6) of PACKET_FASTROUTE as it's _not used_ anywhere in the whole kernel and not supported anywhere, and packets of such type were never exposed to user space, so there are no overlapping users of such kind. Thus, as wished, that seems the only way to make both PACKET_* values non-overlapping and therefore device agnostic. By using those two flags for netlink skbs on nlmon devices, they can be made available and picked up via sll_pkttype (previously unused in netlink context) in struct sockaddr_ll. We now have these two directions: - PACKET_USER (= 6) -> to user space - PACKET_KERNEL (= 7) -> to kernel space Partial `ip a` example strace for sa_family=AF_NETLINK with detected nl msg direction: syscall: direction: sendto(3, ...) = 40 /* to kernel */ recvmsg(3, ...) = 3404 /* to user */ recvmsg(3, ...) = 1120 /* to user */ recvmsg(3, ...) = 20 /* to user */ sendto(3, ...) = 40 /* to kernel */ recvmsg(3, ...) = 168 /* to user */ recvmsg(3, ...) = 144 /* to user */ recvmsg(3, ...) = 20 /* to user */ Signed-off-by: Daniel Borkmann <dborkman@redhat.com> Signed-off-by: Jakub Zawadzki <darkjames-ws@darkjames.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-12-23 14:35:56 +01:00
netlink: annotate data-races around sk->sk_err syzbot caught another data-race in netlink when setting sk->sk_err. Annotate all of them for good measure. BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg write to 0xffff8881613bb220 of 4 bytes by task 28147 on cpu 0: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff8881613bb220 of 4 bytes by task 28146 on cpu 1: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0x00000016 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 28146 Comm: syz-executor.0 Not tainted 6.6.0-rc3-syzkaller-00055-g9ed22ae6be81 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/06/2023 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20231003183455.3410550-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-10-03 18:34:55 +00:00
netlink: fix splat in skb_clone with large messages Since (c05cdb1 netlink: allow large data transfers from user-space), netlink splats if it invokes skb_clone on large netlink skbs since: * skb_shared_info was not correctly initialized. * skb->destructor is not set in the cloned skb. This was spotted by trinity: [ 894.990671] BUG: unable to handle kernel paging request at ffffc9000047b001 [ 894.991034] IP: [<ffffffff81a212c4>] skb_clone+0x24/0xc0 [...] [ 894.991034] Call Trace: [ 894.991034] [<ffffffff81ad299a>] nl_fib_input+0x6a/0x240 [ 894.991034] [<ffffffff81c3b7e6>] ? _raw_read_unlock+0x26/0x40 [ 894.991034] [<ffffffff81a5f189>] netlink_unicast+0x169/0x1e0 [ 894.991034] [<ffffffff81a601e1>] netlink_sendmsg+0x251/0x3d0 Fix it by: 1) introducing a new netlink_skb_clone function that is used in nl_fib_input, that sets our special skb->destructor in the cloned skb. Moreover, handle the release of the large cloned skb head area in the destructor path. 2) not allowing large skbuffs in the netlink broadcast path. I cannot find any reasonable use of the large data transfer using netlink in that path, moreover this helps to skip extra skb_clone handling. I found two more netlink clients that are cloning the skbs, but they are not in the sendmsg path. Therefore, the sole client cloning that I found seems to be the fib frontend. Thanks to Eric Dumazet for helping to address this issue. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-06-28 03:04:23 +02:00
2024-01-22 09:18:07 +08:00
netlink: fix splat in skb_clone with large messages Since (c05cdb1 netlink: allow large data transfers from user-space), netlink splats if it invokes skb_clone on large netlink skbs since: * skb_shared_info was not correctly initialized. * skb->destructor is not set in the cloned skb. This was spotted by trinity: [ 894.990671] BUG: unable to handle kernel paging request at ffffc9000047b001 [ 894.991034] IP: [<ffffffff81a212c4>] skb_clone+0x24/0xc0 [...] [ 894.991034] Call Trace: [ 894.991034] [<ffffffff81ad299a>] nl_fib_input+0x6a/0x240 [ 894.991034] [<ffffffff81c3b7e6>] ? _raw_read_unlock+0x26/0x40 [ 894.991034] [<ffffffff81a5f189>] netlink_unicast+0x169/0x1e0 [ 894.991034] [<ffffffff81a601e1>] netlink_sendmsg+0x251/0x3d0 Fix it by: 1) introducing a new netlink_skb_clone function that is used in nl_fib_input, that sets our special skb->destructor in the cloned skb. Moreover, handle the release of the large cloned skb head area in the destructor path. 2) not allowing large skbuffs in the netlink broadcast path. I cannot find any reasonable use of the large data transfer using netlink in that path, moreover this helps to skip extra skb_clone handling. I found two more netlink clients that are cloning the skbs, but they are not in the sendmsg path. Therefore, the sole client cloning that I found seems to be the fib frontend. Thanks to Eric Dumazet for helping to address this issue. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-06-28 03:04:23 +02:00
netlink: disable IRQs for netlink_lock_table() Syzbot reports that in mac80211 we have a potential deadlock between our "local->stop_queue_reasons_lock" (spinlock) and netlink's nl_table_lock (rwlock). This is because there's at least one situation in which we might try to send a netlink message with this spinlock held while it is also possible to take the spinlock from a hardirq context, resulting in the following deadlock scenario reported by lockdep: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); This seems valid, we can take the queue_stop_reason_lock in any kind of context ("CPU0"), and call ieee80211_report_ack_skb() with the spinlock held and IRQs disabled ("CPU1") in some code path (ieee80211_do_stop() via ieee80211_free_txskb()). Short of disallowing netlink use in scenarios like these (which would be rather complex in mac80211's case due to the deep callchain), it seems the only fix for this is to disable IRQs while nl_table_lock is held to avoid hitting this scenario, this disallows the "CPU0" portion of the reported deadlock. Note that the writer side (netlink_table_grab()) already disables IRQs for this lock. Unfortunately though, this seems like a huge hammer, and maybe the whole netlink table locking should be reworked. Reported-by: syzbot+69ff9dff50dcfe14ddd4@syzkaller.appspotmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-05-17 16:38:09 +02:00
netlink: disable IRQs for netlink_lock_table() Syzbot reports that in mac80211 we have a potential deadlock between our "local->stop_queue_reasons_lock" (spinlock) and netlink's nl_table_lock (rwlock). This is because there's at least one situation in which we might try to send a netlink message with this spinlock held while it is also possible to take the spinlock from a hardirq context, resulting in the following deadlock scenario reported by lockdep: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); This seems valid, we can take the queue_stop_reason_lock in any kind of context ("CPU0"), and call ieee80211_report_ack_skb() with the spinlock held and IRQs disabled ("CPU1") in some code path (ieee80211_do_stop() via ieee80211_free_txskb()). Short of disallowing netlink use in scenarios like these (which would be rather complex in mac80211's case due to the deep callchain), it seems the only fix for this is to disable IRQs while nl_table_lock is held to avoid hitting this scenario, this disallows the "CPU0" portion of the reported deadlock. Note that the writer side (netlink_table_grab()) already disables IRQs for this lock. Unfortunately though, this seems like a huge hammer, and maybe the whole netlink table locking should be reworked. Reported-by: syzbot+69ff9dff50dcfe14ddd4@syzkaller.appspotmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-05-17 16:38:09 +02:00
netlink: disable IRQs for netlink_lock_table() Syzbot reports that in mac80211 we have a potential deadlock between our "local->stop_queue_reasons_lock" (spinlock) and netlink's nl_table_lock (rwlock). This is because there's at least one situation in which we might try to send a netlink message with this spinlock held while it is also possible to take the spinlock from a hardirq context, resulting in the following deadlock scenario reported by lockdep: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); This seems valid, we can take the queue_stop_reason_lock in any kind of context ("CPU0"), and call ieee80211_report_ack_skb() with the spinlock held and IRQs disabled ("CPU1") in some code path (ieee80211_do_stop() via ieee80211_free_txskb()). Short of disallowing netlink use in scenarios like these (which would be rather complex in mac80211's case due to the deep callchain), it seems the only fix for this is to disable IRQs while nl_table_lock is held to avoid hitting this scenario, this disallows the "CPU0" portion of the reported deadlock. Note that the writer side (netlink_table_grab()) already disables IRQs for this lock. Unfortunately though, this seems like a huge hammer, and maybe the whole netlink table locking should be reworked. Reported-by: syzbot+69ff9dff50dcfe14ddd4@syzkaller.appspotmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-05-17 16:38:09 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-27 17:06:00 -08:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: annotate data races around nlk->portid syzbot reminds us netlink_getname() runs locklessly [1] This first patch annotates the race against nlk->portid. Following patches take care of the remaining races. [1] BUG: KCSAN: data-race in netlink_getname / netlink_insert write to 0xffff88814176d310 of 4 bytes by task 2315 on cpu 1: netlink_insert+0xf1/0x9a0 net/netlink/af_netlink.c:583 netlink_autobind+0xae/0x180 net/netlink/af_netlink.c:856 netlink_sendmsg+0x444/0x760 net/netlink/af_netlink.c:1895 sock_sendmsg_nosec net/socket.c:714 [inline] sock_sendmsg net/socket.c:734 [inline] ____sys_sendmsg+0x38f/0x500 net/socket.c:2476 ___sys_sendmsg net/socket.c:2530 [inline] __sys_sendmsg+0x19a/0x230 net/socket.c:2559 __do_sys_sendmsg net/socket.c:2568 [inline] __se_sys_sendmsg net/socket.c:2566 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2566 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88814176d310 of 4 bytes by task 2316 on cpu 0: netlink_getname+0xcd/0x1a0 net/netlink/af_netlink.c:1144 __sys_getsockname+0x11d/0x1b0 net/socket.c:2026 __do_sys_getsockname net/socket.c:2041 [inline] __se_sys_getsockname net/socket.c:2038 [inline] __x64_sys_getsockname+0x3e/0x50 net/socket.c:2038 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0xc9a49780 Reported by Kernel Concurrency Sanitizer on: CPU: 0 PID: 2316 Comm: syz-executor.2 Not tainted 6.2.0-rc3-syzkaller-00030-ge8f60cd7db24-dirty #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-01-20 12:59:53 +00:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: make sure -EBUSY won't escape from netlink_insert Linus reports the following deadlock on rtnl_mutex; triggered only once so far (extract): [12236.694209] NetworkManager D 0000000000013b80 0 1047 1 0x00000000 [12236.694218] ffff88003f902640 0000000000000000 ffffffff815d15a9 0000000000000018 [12236.694224] ffff880119538000 ffff88003f902640 ffffffff81a8ff84 00000000ffffffff [12236.694230] ffffffff81a8ff88 ffff880119c47f00 ffffffff815d133a ffffffff81a8ff80 [12236.694235] Call Trace: [12236.694250] [<ffffffff815d15a9>] ? schedule_preempt_disabled+0x9/0x10 [12236.694257] [<ffffffff815d133a>] ? schedule+0x2a/0x70 [12236.694263] [<ffffffff815d15a9>] ? schedule_preempt_disabled+0x9/0x10 [12236.694271] [<ffffffff815d2c3f>] ? __mutex_lock_slowpath+0x7f/0xf0 [12236.694280] [<ffffffff815d2cc6>] ? mutex_lock+0x16/0x30 [12236.694291] [<ffffffff814f1f90>] ? rtnetlink_rcv+0x10/0x30 [12236.694299] [<ffffffff8150ce3b>] ? netlink_unicast+0xfb/0x180 [12236.694309] [<ffffffff814f5ad3>] ? rtnl_getlink+0x113/0x190 [12236.694319] [<ffffffff814f202a>] ? rtnetlink_rcv_msg+0x7a/0x210 [12236.694331] [<ffffffff8124565c>] ? sock_has_perm+0x5c/0x70 [12236.694339] [<ffffffff814f1fb0>] ? rtnetlink_rcv+0x30/0x30 [12236.694346] [<ffffffff8150d62c>] ? netlink_rcv_skb+0x9c/0xc0 [12236.694354] [<ffffffff814f1f9f>] ? rtnetlink_rcv+0x1f/0x30 [12236.694360] [<ffffffff8150ce3b>] ? netlink_unicast+0xfb/0x180 [12236.694367] [<ffffffff8150d344>] ? netlink_sendmsg+0x484/0x5d0 [12236.694376] [<ffffffff810a236f>] ? __wake_up+0x2f/0x50 [12236.694387] [<ffffffff814cad23>] ? sock_sendmsg+0x33/0x40 [12236.694396] [<ffffffff814cb05e>] ? ___sys_sendmsg+0x22e/0x240 [12236.694405] [<ffffffff814cab75>] ? ___sys_recvmsg+0x135/0x1a0 [12236.694415] [<ffffffff811a9d12>] ? eventfd_write+0x82/0x210 [12236.694423] [<ffffffff811a0f9e>] ? fsnotify+0x32e/0x4c0 [12236.694429] [<ffffffff8108cb70>] ? wake_up_q+0x60/0x60 [12236.694434] [<ffffffff814cba09>] ? __sys_sendmsg+0x39/0x70 [12236.694440] [<ffffffff815d4797>] ? entry_SYSCALL_64_fastpath+0x12/0x6a It seems so far plausible that the recursive call into rtnetlink_rcv() looks suspicious. One way, where this could trigger is that the senders NETLINK_CB(skb).portid was wrongly 0 (which is rtnetlink socket), so the rtnl_getlink() request's answer would be sent to the kernel instead to the actual user process, thus grabbing rtnl_mutex() twice. One theory would be that netlink_autobind() triggered via netlink_sendmsg() internally overwrites the -EBUSY error to 0, but where it is wrongly originating from __netlink_insert() instead. That would reset the socket's portid to 0, which is then filled into NETLINK_CB(skb).portid later on. As commit d470e3b483dc ("[NETLINK]: Fix two socket hashing bugs.") also puts it, -EBUSY should not be propagated from netlink_insert(). It looks like it's very unlikely to reproduce. We need to trigger the rhashtable_insert_rehash() handler under a situation where rehashing currently occurs (one /rare/ way would be to hit ht->elasticity limits while not filled enough to expand the hashtable, but that would rather require a specifically crafted bind() sequence with knowledge about destination slots, seems unlikely). It probably makes sense to guard __netlink_insert() in any case and remap that error. It was suggested that EOVERFLOW might be better than an already overloaded ENOMEM. Reference: http://thread.gmane.org/gmane.linux.network/372676 Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-07 00:26:41 +02:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: annotate data races around nlk->bound While existing code is correct, KCSAN is reporting a data-race in netlink_insert / netlink_sendmsg [1] It is correct to read nlk->bound without a lock, as netlink_autobind() will acquire all needed locks. [1] BUG: KCSAN: data-race in netlink_insert / netlink_sendmsg write to 0xffff8881031c8b30 of 1 bytes by task 18752 on cpu 0: netlink_insert+0x5cc/0x7f0 net/netlink/af_netlink.c:597 netlink_autobind+0xa9/0x150 net/netlink/af_netlink.c:842 netlink_sendmsg+0x479/0x7c0 net/netlink/af_netlink.c:1892 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmsg+0x1ed/0x270 net/socket.c:2475 __do_sys_sendmsg net/socket.c:2484 [inline] __se_sys_sendmsg net/socket.c:2482 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2482 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff8881031c8b30 of 1 bytes by task 18751 on cpu 1: netlink_sendmsg+0x270/0x7c0 net/netlink/af_netlink.c:1891 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] __sys_sendto+0x2a8/0x370 net/socket.c:2019 __do_sys_sendto net/socket.c:2031 [inline] __se_sys_sendto net/socket.c:2027 [inline] __x64_sys_sendto+0x74/0x90 net/socket.c:2027 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00 -> 0x01 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 18751 Comm: syz-executor.0 Not tainted 5.14.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: da314c9923fe ("netlink: Replace rhash_portid with bound") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-04 14:24:15 -07:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
genetlink: synchronize socket closing and family removal In addition to the problem Jeff Layton reported, I looked at the code and reproduced the same warning by subscribing and removing the genl family with a socket still open. This is a fairly tricky race which originates in the fact that generic netlink allows the family to go away while sockets are still open - unlike regular netlink which has a module refcount for every open socket so in general this cannot be triggered. Trying to resolve this issue by the obvious locking isn't possible as it will result in deadlocks between unregistration and group unbind notification (which incidentally lockdep doesn't find due to the home grown locking in the netlink table.) To really resolve this, introduce a "closing socket" reference counter (for generic netlink only, as it's the only affected family) in the core netlink code and use that in generic netlink to wait for all the sockets that are being closed at the same time as a generic netlink family is removed. This fixes the race that when a socket is closed, it will should call the unbind, but if the family is removed at the same time the unbind will not find it, leading to the warning. The real problem though is that in this case the unbind could actually find a new family that is registered to have a multicast group with the same ID, and call its mcast_unbind() leading to confusing. Also remove the warning since it would still trigger, but is now no longer a problem. This also moves the code in af_netlink.c to before unreferencing the module to avoid having the same problem in the normal non-genl case. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-01-16 11:37:14 +01:00
netlink: fix for too early rmmod Netlink code does module autoload if protocol userspace is asking for is not ready. However, module can dissapear right after it was autoloaded. Example: modprobe/rmmod stress-testing and xfrm_user.ko providing NETLINK_XFRM. netlink_create() in such situation _will_ create userspace socket and _will_not_ pin module. Now if module was removed and we're going to call ->netlink_rcv into nothing: BUG: unable to handle kernel paging request at ffffffffa02f842a ^^^^^^^^^^^^^^^^ modules are loaded near these addresses here IP: [<ffffffffa02f842a>] 0xffffffffa02f842a PGD 161f067 PUD 1623063 PMD baa12067 PTE 0 Oops: 0010 [#1] PREEMPT SMP DEBUG_PAGEALLOC last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host0/target0:0:0/0:0:0:0/block/sda/uevent CPU 1 Pid: 11515, comm: ip Not tainted 2.6.33-rc5-netns-00594-gaaa5728-dirty #6 P5E/P5E RIP: 0010:[<ffffffffa02f842a>] [<ffffffffa02f842a>] 0xffffffffa02f842a RSP: 0018:ffff8800baa3db48 EFLAGS: 00010292 RAX: ffff8800baa3dfd8 RBX: ffff8800be353640 RCX: 0000000000000000 RDX: ffffffff81959380 RSI: ffff8800bab7f130 RDI: 0000000000000001 RBP: ffff8800baa3db58 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000011 R13: ffff8800be353640 R14: ffff8800bcdec240 R15: ffff8800bd488010 FS: 00007f93749656f0(0000) GS:ffff880002300000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffffffffa02f842a CR3: 00000000ba82b000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process ip (pid: 11515, threadinfo ffff8800baa3c000, task ffff8800bab7eb30) Stack: ffffffff813637c0 ffff8800bd488000 ffff8800baa3dba8 ffffffff8136397d <0> 0000000000000000 ffffffff81344adc 7fffffffffffffff 0000000000000000 <0> ffff8800baa3ded8 ffff8800be353640 ffff8800bcdec240 0000000000000000 Call Trace: [<ffffffff813637c0>] ? netlink_unicast+0x100/0x2d0 [<ffffffff8136397d>] netlink_unicast+0x2bd/0x2d0 netlink_unicast_kernel: nlk->netlink_rcv(skb); [<ffffffff81344adc>] ? memcpy_fromiovec+0x6c/0x90 [<ffffffff81364263>] netlink_sendmsg+0x1d3/0x2d0 [<ffffffff8133975b>] sock_sendmsg+0xbb/0xf0 [<ffffffff8106cdeb>] ? __lock_acquire+0x27b/0xa60 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff8106db22>] ? __lock_release+0x82/0x170 [<ffffffff810a190e>] ? might_fault+0xbe/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff81344c77>] ? verify_iovec+0x47/0xd0 [<ffffffff8133a509>] sys_sendmsg+0x1a9/0x360 [<ffffffff813c2be5>] ? _raw_spin_unlock_irqrestore+0x65/0x70 [<ffffffff8106aced>] ? trace_hardirqs_on+0xd/0x10 [<ffffffff813c2bc2>] ? _raw_spin_unlock_irqrestore+0x42/0x70 [<ffffffff81197004>] ? __up_read+0x84/0xb0 [<ffffffff8106ac95>] ? trace_hardirqs_on_caller+0x145/0x190 [<ffffffff813c207f>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff8100262b>] system_call_fastpath+0x16/0x1b Code: Bad RIP value. RIP [<ffffffffa02f842a>] 0xffffffffa02f842a RSP <ffff8800baa3db48> CR2: ffffffffa02f842a If module was quickly removed after autoloading, return -E. Return -EPROTONOSUPPORT if module was quickly removed after autoloading. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-30 10:05:05 +00:00
netlink: fix for too early rmmod Netlink code does module autoload if protocol userspace is asking for is not ready. However, module can dissapear right after it was autoloaded. Example: modprobe/rmmod stress-testing and xfrm_user.ko providing NETLINK_XFRM. netlink_create() in such situation _will_ create userspace socket and _will_not_ pin module. Now if module was removed and we're going to call ->netlink_rcv into nothing: BUG: unable to handle kernel paging request at ffffffffa02f842a ^^^^^^^^^^^^^^^^ modules are loaded near these addresses here IP: [<ffffffffa02f842a>] 0xffffffffa02f842a PGD 161f067 PUD 1623063 PMD baa12067 PTE 0 Oops: 0010 [#1] PREEMPT SMP DEBUG_PAGEALLOC last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host0/target0:0:0/0:0:0:0/block/sda/uevent CPU 1 Pid: 11515, comm: ip Not tainted 2.6.33-rc5-netns-00594-gaaa5728-dirty #6 P5E/P5E RIP: 0010:[<ffffffffa02f842a>] [<ffffffffa02f842a>] 0xffffffffa02f842a RSP: 0018:ffff8800baa3db48 EFLAGS: 00010292 RAX: ffff8800baa3dfd8 RBX: ffff8800be353640 RCX: 0000000000000000 RDX: ffffffff81959380 RSI: ffff8800bab7f130 RDI: 0000000000000001 RBP: ffff8800baa3db58 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000011 R13: ffff8800be353640 R14: ffff8800bcdec240 R15: ffff8800bd488010 FS: 00007f93749656f0(0000) GS:ffff880002300000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffffffffa02f842a CR3: 00000000ba82b000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process ip (pid: 11515, threadinfo ffff8800baa3c000, task ffff8800bab7eb30) Stack: ffffffff813637c0 ffff8800bd488000 ffff8800baa3dba8 ffffffff8136397d <0> 0000000000000000 ffffffff81344adc 7fffffffffffffff 0000000000000000 <0> ffff8800baa3ded8 ffff8800be353640 ffff8800bcdec240 0000000000000000 Call Trace: [<ffffffff813637c0>] ? netlink_unicast+0x100/0x2d0 [<ffffffff8136397d>] netlink_unicast+0x2bd/0x2d0 netlink_unicast_kernel: nlk->netlink_rcv(skb); [<ffffffff81344adc>] ? memcpy_fromiovec+0x6c/0x90 [<ffffffff81364263>] netlink_sendmsg+0x1d3/0x2d0 [<ffffffff8133975b>] sock_sendmsg+0xbb/0xf0 [<ffffffff8106cdeb>] ? __lock_acquire+0x27b/0xa60 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff8106db22>] ? __lock_release+0x82/0x170 [<ffffffff810a190e>] ? might_fault+0xbe/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff81344c77>] ? verify_iovec+0x47/0xd0 [<ffffffff8133a509>] sys_sendmsg+0x1a9/0x360 [<ffffffff813c2be5>] ? _raw_spin_unlock_irqrestore+0x65/0x70 [<ffffffff8106aced>] ? trace_hardirqs_on+0xd/0x10 [<ffffffff813c2bc2>] ? _raw_spin_unlock_irqrestore+0x42/0x70 [<ffffffff81197004>] ? __up_read+0x84/0xb0 [<ffffffff8106ac95>] ? trace_hardirqs_on_caller+0x145/0x190 [<ffffffff813c207f>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff8100262b>] system_call_fastpath+0x16/0x1b Code: Bad RIP value. RIP [<ffffffffa02f842a>] 0xffffffffa02f842a RSP <ffff8800baa3db48> CR2: ffffffffa02f842a If module was quickly removed after autoloading, return -E. Return -EPROTONOSUPPORT if module was quickly removed after autoloading. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-30 10:05:05 +00:00
netlink: fix an use-after-free issue for nlk groups ChunYu found a netlink use-after-free issue by syzkaller: [28448.842981] BUG: KASAN: use-after-free in __nla_put+0x37/0x40 at addr ffff8807185e2378 [28448.969918] Call Trace: [...] [28449.117207] __nla_put+0x37/0x40 [28449.132027] nla_put+0xf5/0x130 [28449.146261] sk_diag_fill.isra.4.constprop.5+0x5a0/0x750 [netlink_diag] [28449.176608] __netlink_diag_dump+0x25a/0x700 [netlink_diag] [28449.202215] netlink_diag_dump+0x176/0x240 [netlink_diag] [28449.226834] netlink_dump+0x488/0xbb0 [28449.298014] __netlink_dump_start+0x4e8/0x760 [28449.317924] netlink_diag_handler_dump+0x261/0x340 [netlink_diag] [28449.413414] sock_diag_rcv_msg+0x207/0x390 [28449.432409] netlink_rcv_skb+0x149/0x380 [28449.467647] sock_diag_rcv+0x2d/0x40 [28449.484362] netlink_unicast+0x562/0x7b0 [28449.564790] netlink_sendmsg+0xaa8/0xe60 [28449.661510] sock_sendmsg+0xcf/0x110 [28449.865631] __sys_sendmsg+0xf3/0x240 [28450.000964] SyS_sendmsg+0x32/0x50 [28450.016969] do_syscall_64+0x25c/0x6c0 [28450.154439] entry_SYSCALL64_slow_path+0x25/0x25 It was caused by no protection between nlk groups' free in netlink_release and nlk groups' accessing in sk_diag_dump_groups. The similar issue also exists in netlink_seq_show(). This patch is to defer nlk groups' free in deferred_put_nlk_sk. Reported-by: ChunYu Wang <chunwang@redhat.com> Acked-by: Florian Westphal <fw@strlen.de> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-09-06 11:47:12 +08:00
genetlink: synchronize socket closing and family removal In addition to the problem Jeff Layton reported, I looked at the code and reproduced the same warning by subscribing and removing the genl family with a socket still open. This is a fairly tricky race which originates in the fact that generic netlink allows the family to go away while sockets are still open - unlike regular netlink which has a module refcount for every open socket so in general this cannot be triggered. Trying to resolve this issue by the obvious locking isn't possible as it will result in deadlocks between unregistration and group unbind notification (which incidentally lockdep doesn't find due to the home grown locking in the netlink table.) To really resolve this, introduce a "closing socket" reference counter (for generic netlink only, as it's the only affected family) in the core netlink code and use that in generic netlink to wait for all the sockets that are being closed at the same time as a generic netlink family is removed. This fixes the race that when a socket is closed, it will should call the unbind, but if the family is removed at the same time the unbind will not find it, leading to the warning. The real problem though is that in this case the unbind could actually find a new family that is registered to have a multicast group with the same ID, and call its mcast_unbind() leading to confusing. Also remove the warning since it would still trigger, but is now no longer a problem. This also moves the code in af_netlink.c to before unreferencing the module to avoid having the same problem in the normal non-genl case. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-01-16 11:37:14 +01:00
[PATCH] Notifier chain update: API changes The kernel's implementation of notifier chains is unsafe. There is no protection against entries being added to or removed from a chain while the chain is in use. The issues were discussed in this thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 We noticed that notifier chains in the kernel fall into two basic usage classes: "Blocking" chains are always called from a process context and the callout routines are allowed to sleep; "Atomic" chains can be called from an atomic context and the callout routines are not allowed to sleep. We decided to codify this distinction and make it part of the API. Therefore this set of patches introduces three new, parallel APIs: one for blocking notifiers, one for atomic notifiers, and one for "raw" notifiers (which is really just the old API under a new name). New kinds of data structures are used for the heads of the chains, and new routines are defined for registration, unregistration, and calling a chain. The three APIs are explained in include/linux/notifier.h and their implementation is in kernel/sys.c. With atomic and blocking chains, the implementation guarantees that the chain links will not be corrupted and that chain callers will not get messed up by entries being added or removed. For raw chains the implementation provides no guarantees at all; users of this API must provide their own protections. (The idea was that situations may come up where the assumptions of the atomic and blocking APIs are not appropriate, so it should be possible for users to handle these things in their own way.) There are some limitations, which should not be too hard to live with. For atomic/blocking chains, registration and unregistration must always be done in a process context since the chain is protected by a mutex/rwsem. Also, a callout routine for a non-raw chain must not try to register or unregister entries on its own chain. (This did happen in a couple of places and the code had to be changed to avoid it.) Since atomic chains may be called from within an NMI handler, they cannot use spinlocks for synchronization. Instead we use RCU. The overhead falls almost entirely in the unregister routine, which is okay since unregistration is much less frequent that calling a chain. Here is the list of chains that we adjusted and their classifications. None of them use the raw API, so for the moment it is only a placeholder. ATOMIC CHAINS ------------- arch/i386/kernel/traps.c: i386die_chain arch/ia64/kernel/traps.c: ia64die_chain arch/powerpc/kernel/traps.c: powerpc_die_chain arch/sparc64/kernel/traps.c: sparc64die_chain arch/x86_64/kernel/traps.c: die_chain drivers/char/ipmi/ipmi_si_intf.c: xaction_notifier_list kernel/panic.c: panic_notifier_list kernel/profile.c: task_free_notifier net/bluetooth/hci_core.c: hci_notifier net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_chain net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_expect_chain net/ipv6/addrconf.c: inet6addr_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_expect_chain net/netlink/af_netlink.c: netlink_chain BLOCKING CHAINS --------------- arch/powerpc/platforms/pseries/reconfig.c: pSeries_reconfig_chain arch/s390/kernel/process.c: idle_chain arch/x86_64/kernel/process.c idle_notifier drivers/base/memory.c: memory_chain drivers/cpufreq/cpufreq.c cpufreq_policy_notifier_list drivers/cpufreq/cpufreq.c cpufreq_transition_notifier_list drivers/macintosh/adb.c: adb_client_list drivers/macintosh/via-pmu.c sleep_notifier_list drivers/macintosh/via-pmu68k.c sleep_notifier_list drivers/macintosh/windfarm_core.c wf_client_list drivers/usb/core/notify.c usb_notifier_list drivers/video/fbmem.c fb_notifier_list kernel/cpu.c cpu_chain kernel/module.c module_notify_list kernel/profile.c munmap_notifier kernel/profile.c task_exit_notifier kernel/sys.c reboot_notifier_list net/core/dev.c netdev_chain net/decnet/dn_dev.c: dnaddr_chain net/ipv4/devinet.c: inetaddr_chain It's possible that some of these classifications are wrong. If they are, please let us know or submit a patch to fix them. Note that any chain that gets called very frequently should be atomic, because the rwsem read-locking used for blocking chains is very likely to incur cache misses on SMP systems. (However, if the chain's callout routines may sleep then the chain cannot be atomic.) The patch set was written by Alan Stern and Chandra Seetharaman, incorporating material written by Keith Owens and suggestions from Paul McKenney and Andrew Morton. [jes@sgi.com: restructure the notifier chain initialization macros] Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Jes Sorensen <jes@sgi.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 01:16:30 -08:00
netlink: terminate outstanding dump on socket close Netlink supports iterative dumping of data. It provides the families the following ops: - start - (optional) kicks off the dumping process - dump - actual dump helper, keeps getting called until it returns 0 - done - (optional) pairs with .start, can be used for cleanup The whole process is asynchronous and the repeated calls to .dump don't actually happen in a tight loop, but rather are triggered in response to recvmsg() on the socket. This gives the user full control over the dump, but also means that the user can close the socket without getting to the end of the dump. To make sure .start is always paired with .done we check if there is an ongoing dump before freeing the socket, and if so call .done. The complication is that sockets can get freed from BH and .done is allowed to sleep. So we use a workqueue to defer the call, when needed. Unfortunately this does not work correctly. What we defer is not the cleanup but rather releasing a reference on the socket. We have no guarantee that we own the last reference, if someone else holds the socket they may release it in BH and we're back to square one. The whole dance, however, appears to be unnecessary. Only the user can interact with dumps, so we can clean up when socket is closed. And close always happens in process context. Some async code may still access the socket after close, queue notification skbs to it etc. but no dumps can start, end or otherwise make progress. Delete the workqueue and flush the dump state directly from the release handler. Note that further cleanup is possible in -next, for instance we now always call .done before releasing the main module reference, so dump doesn't have to take a reference of its own. Reported-by: syzkaller <syzkaller@googlegroups.com> Fixes: ed5d7788a934 ("netlink: Do not schedule work from sk_destruct") Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20241106015235.2458807-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-05 17:52:34 -08:00
netlink: terminate outstanding dump on socket close Netlink supports iterative dumping of data. It provides the families the following ops: - start - (optional) kicks off the dumping process - dump - actual dump helper, keeps getting called until it returns 0 - done - (optional) pairs with .start, can be used for cleanup The whole process is asynchronous and the repeated calls to .dump don't actually happen in a tight loop, but rather are triggered in response to recvmsg() on the socket. This gives the user full control over the dump, but also means that the user can close the socket without getting to the end of the dump. To make sure .start is always paired with .done we check if there is an ongoing dump before freeing the socket, and if so call .done. The complication is that sockets can get freed from BH and .done is allowed to sleep. So we use a workqueue to defer the call, when needed. Unfortunately this does not work correctly. What we defer is not the cleanup but rather releasing a reference on the socket. We have no guarantee that we own the last reference, if someone else holds the socket they may release it in BH and we're back to square one. The whole dance, however, appears to be unnecessary. Only the user can interact with dumps, so we can clean up when socket is closed. And close always happens in process context. Some async code may still access the socket after close, queue notification skbs to it etc. but no dumps can start, end or otherwise make progress. Delete the workqueue and flush the dump state directly from the release handler. Note that further cleanup is possible in -next, for instance we now always call .done before releasing the main module reference, so dump doesn't have to take a reference of its own. Reported-by: syzkaller <syzkaller@googlegroups.com> Fixes: ed5d7788a934 ("netlink: Do not schedule work from sk_destruct") Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20241106015235.2458807-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-05 17:52:34 -08:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
netlink: Convert netlink_lookup() to use RCU protected hash table Heavy Netlink users such as Open vSwitch spend a considerable amount of time in netlink_lookup() due to the read-lock on nl_table_lock. Use of RCU relieves the lock contention. Makes use of the new resizable hash table to avoid locking on the lookup. The hash table will grow if entries exceeds 75% of table size up to a total table size of 64K. It will automatically shrink if usage falls below 30%. Also splits nl_table_lock into a separate mutex to protect hash table mutations and allow synchronize_rcu() to sleep while waiting for readers during expansion and shrinking. Before: 9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup 6.42% kpktgend_0 [pktgen] [k] mod_cur_headers 6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker 6.23% kpktgend_0 [kernel.kallsyms] [k] memset 4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup 4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy 3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract 2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2 After: 15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup 8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker 7.92% kpktgend_0 [pktgen] [k] mod_cur_headers 5.11% kpktgend_0 [kernel.kallsyms] [k] memset 4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract 4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock 3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2 [...] 0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup Signed-off-by: Thomas Graf <tgraf@suug.ch> Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-08-02 11:47:45 +02:00
[NETLINK]: Fix two socket hashing bugs. 1) netlink_release() should only decrement the hash entry count if the socket was actually hashed. This was causing hash->entries to underflow, which resulting in all kinds of troubles. On 64-bit systems, this would cause the following conditional to erroneously trigger: err = -ENOMEM; if (BITS_PER_LONG > 32 && unlikely(hash->entries >= UINT_MAX)) goto err; 2) netlink_autobind() needs to propagate the error return from netlink_insert(). Otherwise, callers will not see the error as they should and thus try to operate on a socket with a zero pid, which is very bad. However, it should not propagate -EBUSY. If two threads race to autobind the socket, that is fine. This is consistent with the autobind behavior in other protocols. So bug #1 above, combined with this one, resulted in hangs on netlink_sendmsg() calls to the rtnetlink socket. We'd try to do the user sendmsg() with the socket's pid set to zero, later we do a socket lookup using that pid (via the value we stashed away in NETLINK_CB(skb).pid), but that won't give us the user socket, it will give us the rtnetlink socket. So when we try to wake up the receive queue, we dive back into rtnetlink_rcv() which tries to recursively take the rtnetlink semaphore. Thanks to Jakub Jelink for providing backtraces. Also, thanks to Herbert Xu for supplying debugging patches to help track this down, and also finding a mistake in an earlier version of this fix. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-26 15:31:51 -07:00
netlink: Only check file credentials for implicit destinations It was possible to get a setuid root or setcap executable to write to it's stdout or stderr (which has been set made a netlink socket) and inadvertently reconfigure the networking stack. To prevent this we check that both the creator of the socket and the currentl applications has permission to reconfigure the network stack. Unfortunately this breaks Zebra which always uses sendto/sendmsg and creates it's socket without any privileges. To keep Zebra working don't bother checking if the creator of the socket has privilege when a destination address is specified. Instead rely exclusively on the privileges of the sender of the socket. Note from Andy: This is exactly Eric's code except for some comment clarifications and formatting fixes. Neither I nor, I think, anyone else is thrilled with this approach, but I'm hesitant to wait on a better fix since 3.15 is almost here. Note to stable maintainers: This is a mess. An earlier series of patches in 3.15 fix a rather serious security issue (CVE-2014-0181), but they did so in a way that breaks Zebra. The offending series includes: commit aa4cf9452f469f16cea8c96283b641b4576d4a7b Author: Eric W. Biederman <ebiederm@xmission.com> Date: Wed Apr 23 14:28:03 2014 -0700 net: Add variants of capable for use on netlink messages If a given kernel version is missing that series of fixes, it's probably worth backporting it and this patch. if that series is present, then this fix is critical if you care about Zebra. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Andy Lutomirski <luto@amacapital.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-30 11:04:00 -07:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: annotate data races around nlk->bound While existing code is correct, KCSAN is reporting a data-race in netlink_insert / netlink_sendmsg [1] It is correct to read nlk->bound without a lock, as netlink_autobind() will acquire all needed locks. [1] BUG: KCSAN: data-race in netlink_insert / netlink_sendmsg write to 0xffff8881031c8b30 of 1 bytes by task 18752 on cpu 0: netlink_insert+0x5cc/0x7f0 net/netlink/af_netlink.c:597 netlink_autobind+0xa9/0x150 net/netlink/af_netlink.c:842 netlink_sendmsg+0x479/0x7c0 net/netlink/af_netlink.c:1892 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmsg+0x1ed/0x270 net/socket.c:2475 __do_sys_sendmsg net/socket.c:2484 [inline] __se_sys_sendmsg net/socket.c:2482 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2482 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff8881031c8b30 of 1 bytes by task 18751 on cpu 1: netlink_sendmsg+0x270/0x7c0 net/netlink/af_netlink.c:1891 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] __sys_sendto+0x2a8/0x370 net/socket.c:2019 __do_sys_sendto net/socket.c:2031 [inline] __se_sys_sendto net/socket.c:2027 [inline] __x64_sys_sendto+0x74/0x90 net/socket.c:2027 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00 -> 0x01 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 18751 Comm: syz-executor.0 Not tainted 5.14.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: da314c9923fe ("netlink: Replace rhash_portid with bound") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-04 14:24:15 -07:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: don't call ->netlink_bind with table lock held When I added support to allow generic netlink multicast groups to be restricted to subscribers with CAP_NET_ADMIN I was unaware that a genl_bind implementation already existed in the past. It was reverted due to ABBA deadlock: 1. ->netlink_bind gets called with the table lock held. 2. genetlink bind callback is invoked, it grabs the genl lock. But when a new genl subsystem is (un)registered, these two locks are taken in reverse order. One solution would be to revert again and add a comment in genl referring 1e82a62fec613, "genetlink: remove genl_bind"). This would need a second change in mptcp to not expose the raw token value anymore, e.g. by hashing the token with a secret key so userspace can still associate subflow events with the correct mptcp connection. However, Paolo Abeni reminded me to double-check why the netlink table is locked in the first place. I can't find one. netlink_bind() is already called without this lock when userspace joins a group via NETLINK_ADD_MEMBERSHIP setsockopt. Same holds for the netlink_unbind operation. Digging through the history, commit f773608026ee1 ("netlink: access nlk groups safely in netlink bind and getname") expanded the lock scope. commit 3a20773beeeeade ("net: netlink: cap max groups which will be considered in netlink_bind()") ... removed the nlk->ngroups access that the lock scope extension was all about. Reduce the lock scope again and always call ->netlink_bind without the table lock. The Fixes tag should be vs. the patch mentioned in the link below, but that one got squash-merged into the patch that came earlier in the series. Fixes: 4d54cc32112d8d ("mptcp: avoid lock_fast usage in accept path") Link: https://lore.kernel.org/mptcp/20210213000001.379332-8-mathew.j.martineau@linux.intel.com/T/#u Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: Xin Long <lucien.xin@gmail.com> Cc: Johannes Berg <johannes.berg@intel.com> Cc: Sean Tranchetti <stranche@codeaurora.org> Cc: Paolo Abeni <pabeni@redhat.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-04-16 21:29:13 +02:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: don't call ->netlink_bind with table lock held When I added support to allow generic netlink multicast groups to be restricted to subscribers with CAP_NET_ADMIN I was unaware that a genl_bind implementation already existed in the past. It was reverted due to ABBA deadlock: 1. ->netlink_bind gets called with the table lock held. 2. genetlink bind callback is invoked, it grabs the genl lock. But when a new genl subsystem is (un)registered, these two locks are taken in reverse order. One solution would be to revert again and add a comment in genl referring 1e82a62fec613, "genetlink: remove genl_bind"). This would need a second change in mptcp to not expose the raw token value anymore, e.g. by hashing the token with a secret key so userspace can still associate subflow events with the correct mptcp connection. However, Paolo Abeni reminded me to double-check why the netlink table is locked in the first place. I can't find one. netlink_bind() is already called without this lock when userspace joins a group via NETLINK_ADD_MEMBERSHIP setsockopt. Same holds for the netlink_unbind operation. Digging through the history, commit f773608026ee1 ("netlink: access nlk groups safely in netlink bind and getname") expanded the lock scope. commit 3a20773beeeeade ("net: netlink: cap max groups which will be considered in netlink_bind()") ... removed the nlk->ngroups access that the lock scope extension was all about. Reduce the lock scope again and always call ->netlink_bind without the table lock. The Fixes tag should be vs. the patch mentioned in the link below, but that one got squash-merged into the patch that came earlier in the series. Fixes: 4d54cc32112d8d ("mptcp: avoid lock_fast usage in accept path") Link: https://lore.kernel.org/mptcp/20210213000001.379332-8-mathew.j.martineau@linux.intel.com/T/#u Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: Xin Long <lucien.xin@gmail.com> Cc: Johannes Berg <johannes.berg@intel.com> Cc: Sean Tranchetti <stranche@codeaurora.org> Cc: Paolo Abeni <pabeni@redhat.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-04-16 21:29:13 +02:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: annotate data races around nlk->bound While existing code is correct, KCSAN is reporting a data-race in netlink_insert / netlink_sendmsg [1] It is correct to read nlk->bound without a lock, as netlink_autobind() will acquire all needed locks. [1] BUG: KCSAN: data-race in netlink_insert / netlink_sendmsg write to 0xffff8881031c8b30 of 1 bytes by task 18752 on cpu 0: netlink_insert+0x5cc/0x7f0 net/netlink/af_netlink.c:597 netlink_autobind+0xa9/0x150 net/netlink/af_netlink.c:842 netlink_sendmsg+0x479/0x7c0 net/netlink/af_netlink.c:1892 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmsg+0x1ed/0x270 net/socket.c:2475 __do_sys_sendmsg net/socket.c:2484 [inline] __se_sys_sendmsg net/socket.c:2482 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2482 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff8881031c8b30 of 1 bytes by task 18751 on cpu 1: netlink_sendmsg+0x270/0x7c0 net/netlink/af_netlink.c:1891 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] __sys_sendto+0x2a8/0x370 net/socket.c:2019 __do_sys_sendto net/socket.c:2031 [inline] __se_sys_sendto net/socket.c:2027 [inline] __x64_sys_sendto+0x74/0x90 net/socket.c:2027 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00 -> 0x01 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 18751 Comm: syz-executor.0 Not tainted 5.14.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: da314c9923fe ("netlink: Replace rhash_portid with bound") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-04 14:24:15 -07:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: annotate data races around nlk->bound While existing code is correct, KCSAN is reporting a data-race in netlink_insert / netlink_sendmsg [1] It is correct to read nlk->bound without a lock, as netlink_autobind() will acquire all needed locks. [1] BUG: KCSAN: data-race in netlink_insert / netlink_sendmsg write to 0xffff8881031c8b30 of 1 bytes by task 18752 on cpu 0: netlink_insert+0x5cc/0x7f0 net/netlink/af_netlink.c:597 netlink_autobind+0xa9/0x150 net/netlink/af_netlink.c:842 netlink_sendmsg+0x479/0x7c0 net/netlink/af_netlink.c:1892 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmsg+0x1ed/0x270 net/socket.c:2475 __do_sys_sendmsg net/socket.c:2484 [inline] __se_sys_sendmsg net/socket.c:2482 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2482 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff8881031c8b30 of 1 bytes by task 18751 on cpu 1: netlink_sendmsg+0x270/0x7c0 net/netlink/af_netlink.c:1891 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] __sys_sendto+0x2a8/0x370 net/socket.c:2019 __do_sys_sendto net/socket.c:2031 [inline] __se_sys_sendto net/socket.c:2027 [inline] __x64_sys_sendto+0x74/0x90 net/socket.c:2027 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00 -> 0x01 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 18751 Comm: syz-executor.0 Not tainted 5.14.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: da314c9923fe ("netlink: Replace rhash_portid with bound") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-04 14:24:15 -07:00
net: make getname() functions return length rather than use int* parameter Changes since v1: Added changes in these files: drivers/infiniband/hw/usnic/usnic_transport.c drivers/staging/lustre/lnet/lnet/lib-socket.c drivers/target/iscsi/iscsi_target_login.c drivers/vhost/net.c fs/dlm/lowcomms.c fs/ocfs2/cluster/tcp.c security/tomoyo/network.c Before: All these functions either return a negative error indicator, or store length of sockaddr into "int *socklen" parameter and return zero on success. "int *socklen" parameter is awkward. For example, if caller does not care, it still needs to provide on-stack storage for the value it does not need. None of the many FOO_getname() functions of various protocols ever used old value of *socklen. They always just overwrite it. This change drops this parameter, and makes all these functions, on success, return length of sockaddr. It's always >= 0 and can be differentiated from an error. Tests in callers are changed from "if (err)" to "if (err < 0)", where needed. rpc_sockname() lost "int buflen" parameter, since its only use was to be passed to kernel_getsockname() as &buflen and subsequently not used in any way. Userspace API is not changed. text data bss dec hex filename 30108430 2633624 873672 33615726 200ef6e vmlinux.before.o 30108109 2633612 873672 33615393 200ee21 vmlinux.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: David S. Miller <davem@davemloft.net> CC: linux-kernel@vger.kernel.org CC: netdev@vger.kernel.org CC: linux-bluetooth@vger.kernel.org CC: linux-decnet-user@lists.sourceforge.net CC: linux-wireless@vger.kernel.org CC: linux-rdma@vger.kernel.org CC: linux-sctp@vger.kernel.org CC: linux-nfs@vger.kernel.org CC: linux-x25@vger.kernel.org Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-12 20:00:20 +01:00
netlink: annotate data races around nlk->portid syzbot reminds us netlink_getname() runs locklessly [1] This first patch annotates the race against nlk->portid. Following patches take care of the remaining races. [1] BUG: KCSAN: data-race in netlink_getname / netlink_insert write to 0xffff88814176d310 of 4 bytes by task 2315 on cpu 1: netlink_insert+0xf1/0x9a0 net/netlink/af_netlink.c:583 netlink_autobind+0xae/0x180 net/netlink/af_netlink.c:856 netlink_sendmsg+0x444/0x760 net/netlink/af_netlink.c:1895 sock_sendmsg_nosec net/socket.c:714 [inline] sock_sendmsg net/socket.c:734 [inline] ____sys_sendmsg+0x38f/0x500 net/socket.c:2476 ___sys_sendmsg net/socket.c:2530 [inline] __sys_sendmsg+0x19a/0x230 net/socket.c:2559 __do_sys_sendmsg net/socket.c:2568 [inline] __se_sys_sendmsg net/socket.c:2566 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2566 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88814176d310 of 4 bytes by task 2316 on cpu 0: netlink_getname+0xcd/0x1a0 net/netlink/af_netlink.c:1144 __sys_getsockname+0x11d/0x1b0 net/socket.c:2026 __do_sys_getsockname net/socket.c:2041 [inline] __se_sys_getsockname net/socket.c:2038 [inline] __x64_sys_getsockname+0x3e/0x50 net/socket.c:2038 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0xc9a49780 Reported by Kernel Concurrency Sanitizer on: CPU: 0 PID: 2316 Comm: syz-executor.2 Not tainted 6.2.0-rc3-syzkaller-00030-ge8f60cd7db24-dirty #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-01-20 12:59:53 +00:00
net: make getname() functions return length rather than use int* parameter Changes since v1: Added changes in these files: drivers/infiniband/hw/usnic/usnic_transport.c drivers/staging/lustre/lnet/lnet/lib-socket.c drivers/target/iscsi/iscsi_target_login.c drivers/vhost/net.c fs/dlm/lowcomms.c fs/ocfs2/cluster/tcp.c security/tomoyo/network.c Before: All these functions either return a negative error indicator, or store length of sockaddr into "int *socklen" parameter and return zero on success. "int *socklen" parameter is awkward. For example, if caller does not care, it still needs to provide on-stack storage for the value it does not need. None of the many FOO_getname() functions of various protocols ever used old value of *socklen. They always just overwrite it. This change drops this parameter, and makes all these functions, on success, return length of sockaddr. It's always >= 0 and can be differentiated from an error. Tests in callers are changed from "if (err)" to "if (err < 0)", where needed. rpc_sockname() lost "int buflen" parameter, since its only use was to be passed to kernel_getsockname() as &buflen and subsequently not used in any way. Userspace API is not changed. text data bss dec hex filename 30108430 2633624 873672 33615726 200ef6e vmlinux.before.o 30108109 2633612 873672 33615393 200ee21 vmlinux.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: David S. Miller <davem@davemloft.net> CC: linux-kernel@vger.kernel.org CC: netdev@vger.kernel.org CC: linux-bluetooth@vger.kernel.org CC: linux-decnet-user@lists.sourceforge.net CC: linux-wireless@vger.kernel.org CC: linux-rdma@vger.kernel.org CC: linux-sctp@vger.kernel.org CC: linux-nfs@vger.kernel.org CC: linux-x25@vger.kernel.org Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-12 20:00:20 +01:00
netlink: fix splat in skb_clone with large messages Since (c05cdb1 netlink: allow large data transfers from user-space), netlink splats if it invokes skb_clone on large netlink skbs since: * skb_shared_info was not correctly initialized. * skb->destructor is not set in the cloned skb. This was spotted by trinity: [ 894.990671] BUG: unable to handle kernel paging request at ffffc9000047b001 [ 894.991034] IP: [<ffffffff81a212c4>] skb_clone+0x24/0xc0 [...] [ 894.991034] Call Trace: [ 894.991034] [<ffffffff81ad299a>] nl_fib_input+0x6a/0x240 [ 894.991034] [<ffffffff81c3b7e6>] ? _raw_read_unlock+0x26/0x40 [ 894.991034] [<ffffffff81a5f189>] netlink_unicast+0x169/0x1e0 [ 894.991034] [<ffffffff81a601e1>] netlink_sendmsg+0x251/0x3d0 Fix it by: 1) introducing a new netlink_skb_clone function that is used in nl_fib_input, that sets our special skb->destructor in the cloned skb. Moreover, handle the release of the large cloned skb head area in the destructor path. 2) not allowing large skbuffs in the netlink broadcast path. I cannot find any reasonable use of the large data transfer using netlink in that path, moreover this helps to skip extra skb_clone handling. I found two more netlink clients that are cloning the skbs, but they are not in the sendmsg path. Therefore, the sole client cloning that I found seems to be the fib frontend. Thanks to Eric Dumazet for helping to address this issue. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-06-28 03:04:23 +02:00
netlink: fix splat in skb_clone with large messages Since (c05cdb1 netlink: allow large data transfers from user-space), netlink splats if it invokes skb_clone on large netlink skbs since: * skb_shared_info was not correctly initialized. * skb->destructor is not set in the cloned skb. This was spotted by trinity: [ 894.990671] BUG: unable to handle kernel paging request at ffffc9000047b001 [ 894.991034] IP: [<ffffffff81a212c4>] skb_clone+0x24/0xc0 [...] [ 894.991034] Call Trace: [ 894.991034] [<ffffffff81ad299a>] nl_fib_input+0x6a/0x240 [ 894.991034] [<ffffffff81c3b7e6>] ? _raw_read_unlock+0x26/0x40 [ 894.991034] [<ffffffff81a5f189>] netlink_unicast+0x169/0x1e0 [ 894.991034] [<ffffffff81a601e1>] netlink_sendmsg+0x251/0x3d0 Fix it by: 1) introducing a new netlink_skb_clone function that is used in nl_fib_input, that sets our special skb->destructor in the cloned skb. Moreover, handle the release of the large cloned skb head area in the destructor path. 2) not allowing large skbuffs in the netlink broadcast path. I cannot find any reasonable use of the large data transfer using netlink in that path, moreover this helps to skip extra skb_clone handling. I found two more netlink clients that are cloning the skbs, but they are not in the sendmsg path. Therefore, the sole client cloning that I found seems to be the fib frontend. Thanks to Eric Dumazet for helping to address this issue. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-06-28 03:04:23 +02:00
netlink: avoid infinite retry looping in netlink_unicast() netlink_attachskb() checks for the socket's read memory allocation constraints. Firstly, it has: rmem < READ_ONCE(sk->sk_rcvbuf) to check if the just increased rmem value fits into the socket's receive buffer. If not, it proceeds and tries to wait for the memory under: rmem + skb->truesize > READ_ONCE(sk->sk_rcvbuf) The checks don't cover the case when skb->truesize + sk->sk_rmem_alloc is equal to sk->sk_rcvbuf. Thus the function neither successfully accepts these conditions, nor manages to reschedule the task - and is called in retry loop for indefinite time which is caught as: rcu: INFO: rcu_sched self-detected stall on CPU rcu: 0-....: (25999 ticks this GP) idle=ef2/1/0x4000000000000000 softirq=262269/262269 fqs=6212 (t=26000 jiffies g=230833 q=259957) NMI backtrace for cpu 0 CPU: 0 PID: 22 Comm: kauditd Not tainted 5.10.240 #68 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.17.0-4.fc42 04/01/2014 Call Trace: <IRQ> dump_stack lib/dump_stack.c:120 nmi_cpu_backtrace.cold lib/nmi_backtrace.c:105 nmi_trigger_cpumask_backtrace lib/nmi_backtrace.c:62 rcu_dump_cpu_stacks kernel/rcu/tree_stall.h:335 rcu_sched_clock_irq.cold kernel/rcu/tree.c:2590 update_process_times kernel/time/timer.c:1953 tick_sched_handle kernel/time/tick-sched.c:227 tick_sched_timer kernel/time/tick-sched.c:1399 __hrtimer_run_queues kernel/time/hrtimer.c:1652 hrtimer_interrupt kernel/time/hrtimer.c:1717 __sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1113 asm_call_irq_on_stack arch/x86/entry/entry_64.S:808 </IRQ> netlink_attachskb net/netlink/af_netlink.c:1234 netlink_unicast net/netlink/af_netlink.c:1349 kauditd_send_queue kernel/audit.c:776 kauditd_thread kernel/audit.c:897 kthread kernel/kthread.c:328 ret_from_fork arch/x86/entry/entry_64.S:304 Restore the original behavior of the check which commit in Fixes accidentally missed when restructuring the code. Found by Linux Verification Center (linuxtesting.org). Fixes: ae8f160e7eb2 ("netlink: Fix wraparounds of sk->sk_rmem_alloc.") Cc: stable@vger.kernel.org Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru> Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com> Link: https://patch.msgid.link/20250728080727.255138-1-pchelkin@ispras.ru Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2025-07-28 11:06:47 +03:00
netlink: remove mmapped netlink support mmapped netlink has a number of unresolved issues: - TX zerocopy support had to be disabled more than a year ago via commit 4682a0358639b29cf ("netlink: Always copy on mmap TX.") because the content of the mmapped area can change after netlink attribute validation but before message processing. - RX support was implemented mainly to speed up nfqueue dumping packet payload to userspace. However, since commit ae08ce0021087a5d812d2 ("netfilter: nfnetlink_queue: zero copy support") we avoid one copy with the socket-based interface too (via the skb_zerocopy helper). The other problem is that skbs attached to mmaped netlink socket behave different from normal skbs: - they don't have a shinfo area, so all functions that use skb_shinfo() (e.g. skb_clone) cannot be used. - reserving headroom prevents userspace from seeing the content as it expects message to start at skb->head. See for instance commit aa3a022094fa ("netlink: not trim skb for mmaped socket when dump"). - skbs handed e.g. to netlink_ack must have non-NULL skb->sk, else we crash because it needs the sk to check if a tx ring is attached. Also not obvious, leads to non-intuitive bug fixes such as 7c7bdf359 ("netfilter: nfnetlink: use original skbuff when acking batches"). mmaped netlink also didn't play nicely with the skb_zerocopy helper used by nfqueue and openvswitch. Daniel Borkmann fixed this via commit 6bb0fef489f6 ("netlink, mmap: fix edge-case leakages in nf queue zero-copy")' but at the cost of also needing to provide remaining length to the allocation function. nfqueue also has problems when used with mmaped rx netlink: - mmaped netlink doesn't allow use of nfqueue batch verdict messages. Problem is that in the mmap case, the allocation time also determines the ordering in which the frame will be seen by userspace (A allocating before B means that A is located in earlier ring slot, but this also means that B might get a lower sequence number then A since seqno is decided later. To fix this we would need to extend the spinlocked region to also cover the allocation and message setup which isn't desirable. - nfqueue can now be configured to queue large (GSO) skbs to userspace. Queing GSO packets is faster than having to force a software segmentation in the kernel, so this is a desirable option. However, with a mmap based ring one has to use 64kb per ring slot element, else mmap has to fall back to the socket path (NL_MMAP_STATUS_COPY) for all large packets. To use the mmap interface, userspace not only has to probe for mmap netlink support, it also has to implement a recv/socket receive path in order to handle messages that exceed the size of an rx ring element. Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Ken-ichirou MATSUZAWA <chamaken@gmail.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Cc: Patrick McHardy <kaber@trash.net> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-02-18 15:03:24 +01:00
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-05 23:56:36 -08:00
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-05 23:56:36 -08:00
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-27 17:06:00 -08:00
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-05 23:56:36 -08:00
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-05 23:56:36 -08:00
mm, page_alloc: distinguish between being unable to sleep, unwilling to sleep and avoiding waking kswapd __GFP_WAIT has been used to identify atomic context in callers that hold spinlocks or are in interrupts. They are expected to be high priority and have access one of two watermarks lower than "min" which can be referred to as the "atomic reserve". __GFP_HIGH users get access to the first lower watermark and can be called the "high priority reserve". Over time, callers had a requirement to not block when fallback options were available. Some have abused __GFP_WAIT leading to a situation where an optimisitic allocation with a fallback option can access atomic reserves. This patch uses __GFP_ATOMIC to identify callers that are truely atomic, cannot sleep and have no alternative. High priority users continue to use __GFP_HIGH. __GFP_DIRECT_RECLAIM identifies callers that can sleep and are willing to enter direct reclaim. __GFP_KSWAPD_RECLAIM to identify callers that want to wake kswapd for background reclaim. __GFP_WAIT is redefined as a caller that is willing to enter direct reclaim and wake kswapd for background reclaim. This patch then converts a number of sites o __GFP_ATOMIC is used by callers that are high priority and have memory pools for those requests. GFP_ATOMIC uses this flag. o Callers that have a limited mempool to guarantee forward progress clear __GFP_DIRECT_RECLAIM but keep __GFP_KSWAPD_RECLAIM. bio allocations fall into this category where kswapd will still be woken but atomic reserves are not used as there is a one-entry mempool to guarantee progress. o Callers that are checking if they are non-blocking should use the helper gfpflags_allow_blocking() where possible. This is because checking for __GFP_WAIT as was done historically now can trigger false positives. Some exceptions like dm-crypt.c exist where the code intent is clearer if __GFP_DIRECT_RECLAIM is used instead of the helper due to flag manipulations. o Callers that built their own GFP flags instead of starting with GFP_KERNEL and friends now also need to specify __GFP_KSWAPD_RECLAIM. The first key hazard to watch out for is callers that removed __GFP_WAIT and was depending on access to atomic reserves for inconspicuous reasons. In some cases it may be appropriate for them to use __GFP_HIGH. The second key hazard is callers that assembled their own combination of GFP flags instead of starting with something like GFP_KERNEL. They may now wish to specify __GFP_KSWAPD_RECLAIM. It's almost certainly harmless if it's missed in most cases as other activity will wake kswapd. Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Christoph Lameter <cl@linux.com> Cc: David Rientjes <rientjes@google.com> Cc: Vitaly Wool <vitalywool@gmail.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-11-06 16:28:21 -08:00
netlink: annotate data-races around sk->sk_err syzbot caught another data-race in netlink when setting sk->sk_err. Annotate all of them for good measure. BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg write to 0xffff8881613bb220 of 4 bytes by task 28147 on cpu 0: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff8881613bb220 of 4 bytes by task 28146 on cpu 1: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0x00000016 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 28146 Comm: syz-executor.0 Not tainted 6.6.0-rc3-syzkaller-00055-g9ed22ae6be81 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/06/2023 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20231003183455.3410550-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-10-03 18:34:55 +00:00
netlink: fix potential deadlock in netlink_set_err() syzbot reported a possible deadlock in netlink_set_err() [1] A similar issue was fixed in commit 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") in netlink_lock_table() This patch adds IRQ safety to netlink_set_err() and __netlink_diag_dump() which were not covered by cited commit. [1] WARNING: possible irq lock inversion dependency detected 6.4.0-rc6-syzkaller-00240-g4e9f0ec38852 #0 Not tainted syz-executor.2/23011 just changed the state of lock: ffffffff8e1a7a58 (nl_table_lock){.+.?}-{2:2}, at: netlink_set_err+0x2e/0x3a0 net/netlink/af_netlink.c:1612 but this lock was taken by another, SOFTIRQ-safe lock in the past: (&local->queue_stop_reason_lock){..-.}-{2:2} and interrupts could create inverse lock ordering between them. other info that might help us debug this: Possible interrupt unsafe locking scenario: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); *** DEADLOCK *** Fixes: 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") Reported-by: syzbot+a7d200a347f912723e5c@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=a7d200a347f912723e5c Link: https://lore.kernel.org/netdev/000000000000e38d1605fea5747e@google.com/T/#u Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Johannes Berg <johannes.berg@intel.com> Link: https://lore.kernel.org/r/20230621154337.1668594-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-06-21 15:43:37 +00:00
netlink: fix potential deadlock in netlink_set_err() syzbot reported a possible deadlock in netlink_set_err() [1] A similar issue was fixed in commit 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") in netlink_lock_table() This patch adds IRQ safety to netlink_set_err() and __netlink_diag_dump() which were not covered by cited commit. [1] WARNING: possible irq lock inversion dependency detected 6.4.0-rc6-syzkaller-00240-g4e9f0ec38852 #0 Not tainted syz-executor.2/23011 just changed the state of lock: ffffffff8e1a7a58 (nl_table_lock){.+.?}-{2:2}, at: netlink_set_err+0x2e/0x3a0 net/netlink/af_netlink.c:1612 but this lock was taken by another, SOFTIRQ-safe lock in the past: (&local->queue_stop_reason_lock){..-.}-{2:2} and interrupts could create inverse lock ordering between them. other info that might help us debug this: Possible interrupt unsafe locking scenario: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); *** DEADLOCK *** Fixes: 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") Reported-by: syzbot+a7d200a347f912723e5c@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=a7d200a347f912723e5c Link: https://lore.kernel.org/netdev/000000000000e38d1605fea5747e@google.com/T/#u Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Johannes Berg <johannes.berg@intel.com> Link: https://lore.kernel.org/r/20230621154337.1668594-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-06-21 15:43:37 +00:00
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-27 17:06:00 -08:00
netlink: fix potential deadlock in netlink_set_err() syzbot reported a possible deadlock in netlink_set_err() [1] A similar issue was fixed in commit 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") in netlink_lock_table() This patch adds IRQ safety to netlink_set_err() and __netlink_diag_dump() which were not covered by cited commit. [1] WARNING: possible irq lock inversion dependency detected 6.4.0-rc6-syzkaller-00240-g4e9f0ec38852 #0 Not tainted syz-executor.2/23011 just changed the state of lock: ffffffff8e1a7a58 (nl_table_lock){.+.?}-{2:2}, at: netlink_set_err+0x2e/0x3a0 net/netlink/af_netlink.c:1612 but this lock was taken by another, SOFTIRQ-safe lock in the past: (&local->queue_stop_reason_lock){..-.}-{2:2} and interrupts could create inverse lock ordering between them. other info that might help us debug this: Possible interrupt unsafe locking scenario: CPU0 CPU1 ---- ---- lock(nl_table_lock); local_irq_disable(); lock(&local->queue_stop_reason_lock); lock(nl_table_lock); <Interrupt> lock(&local->queue_stop_reason_lock); *** DEADLOCK *** Fixes: 1d482e666b8e ("netlink: disable IRQs for netlink_lock_table()") Reported-by: syzbot+a7d200a347f912723e5c@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=a7d200a347f912723e5c Link: https://lore.kernel.org/netdev/000000000000e38d1605fea5747e@google.com/T/#u Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Johannes Berg <johannes.berg@intel.com> Link: https://lore.kernel.org/r/20230621154337.1668594-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-06-21 15:43:37 +00:00
netlink: remove mmapped netlink support mmapped netlink has a number of unresolved issues: - TX zerocopy support had to be disabled more than a year ago via commit 4682a0358639b29cf ("netlink: Always copy on mmap TX.") because the content of the mmapped area can change after netlink attribute validation but before message processing. - RX support was implemented mainly to speed up nfqueue dumping packet payload to userspace. However, since commit ae08ce0021087a5d812d2 ("netfilter: nfnetlink_queue: zero copy support") we avoid one copy with the socket-based interface too (via the skb_zerocopy helper). The other problem is that skbs attached to mmaped netlink socket behave different from normal skbs: - they don't have a shinfo area, so all functions that use skb_shinfo() (e.g. skb_clone) cannot be used. - reserving headroom prevents userspace from seeing the content as it expects message to start at skb->head. See for instance commit aa3a022094fa ("netlink: not trim skb for mmaped socket when dump"). - skbs handed e.g. to netlink_ack must have non-NULL skb->sk, else we crash because it needs the sk to check if a tx ring is attached. Also not obvious, leads to non-intuitive bug fixes such as 7c7bdf359 ("netfilter: nfnetlink: use original skbuff when acking batches"). mmaped netlink also didn't play nicely with the skb_zerocopy helper used by nfqueue and openvswitch. Daniel Borkmann fixed this via commit 6bb0fef489f6 ("netlink, mmap: fix edge-case leakages in nf queue zero-copy")' but at the cost of also needing to provide remaining length to the allocation function. nfqueue also has problems when used with mmaped rx netlink: - mmaped netlink doesn't allow use of nfqueue batch verdict messages. Problem is that in the mmap case, the allocation time also determines the ordering in which the frame will be seen by userspace (A allocating before B means that A is located in earlier ring slot, but this also means that B might get a lower sequence number then A since seqno is decided later. To fix this we would need to extend the spinlocked region to also cover the allocation and message setup which isn't desirable. - nfqueue can now be configured to queue large (GSO) skbs to userspace. Queing GSO packets is faster than having to force a software segmentation in the kernel, so this is a desirable option. However, with a mmap based ring one has to use 64kb per ring slot element, else mmap has to fall back to the socket path (NL_MMAP_STATUS_COPY) for all large packets. To use the mmap interface, userspace not only has to probe for mmap netlink support, it also has to implement a recv/socket receive path in order to handle messages that exceed the size of an rx ring element. Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Ken-ichirou MATSUZAWA <chamaken@gmail.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Cc: Patrick McHardy <kaber@trash.net> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-02-18 15:03:24 +01:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: Add new socket option to enable strict checking on dumps Add a new socket option, NETLINK_DUMP_STRICT_CHK, that userspace can use via setsockopt to request strict checking of headers and attributes on dump requests. To get dump features such as kernel side filtering based on data in the header or attributes appended to the dump request, userspace must call setsockopt() for NETLINK_DUMP_STRICT_CHK and a non-zero value. Since the netlink sock and its flags are private to the af_netlink code, the strict checking flag is passed to dump handlers via a flag in the netlink_callback struct. For old userspace on new kernel there is no impact as all of the data checks in later patches are wrapped in a check on the new strict flag. For new userspace on old kernel, the setsockopt will fail and even if new userspace sets data in the headers and appended attributes the kernel will silently ignore it. Moving forward when the setsockopt succeeds, the new userspace on old kernel means the dump request can pass an attribute the kernel does not understand. The dump will then fail as the older kernel does not understand it. New userspace on new kernel setting the socket option gets the benefit of the improved data dump. Kernel side the NETLINK_DUMP_STRICT_CHK uapi is converted to a generic NETLINK_F_STRICT_CHK flag which can potentially be leveraged for tighter checking on the NEW, DEL, and SET commands. Signed-off-by: David Ahern <dsahern@gmail.com> Acked-by: Christian Brauner <christian@brauner.io> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-10-07 20:16:27 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 16:37:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Add new socket option to enable strict checking on dumps Add a new socket option, NETLINK_DUMP_STRICT_CHK, that userspace can use via setsockopt to request strict checking of headers and attributes on dump requests. To get dump features such as kernel side filtering based on data in the header or attributes appended to the dump request, userspace must call setsockopt() for NETLINK_DUMP_STRICT_CHK and a non-zero value. Since the netlink sock and its flags are private to the af_netlink code, the strict checking flag is passed to dump handlers via a flag in the netlink_callback struct. For old userspace on new kernel there is no impact as all of the data checks in later patches are wrapped in a check on the new strict flag. For new userspace on old kernel, the setsockopt will fail and even if new userspace sets data in the headers and appended attributes the kernel will silently ignore it. Moving forward when the setsockopt succeeds, the new userspace on old kernel means the dump request can pass an attribute the kernel does not understand. The dump will then fail as the older kernel does not understand it. New userspace on new kernel setting the socket option gets the benefit of the improved data dump. Kernel side the NETLINK_DUMP_STRICT_CHK uapi is converted to a generic NETLINK_F_STRICT_CHK flag which can potentially be leveraged for tighter checking on the NEW, DEL, and SET commands. Signed-off-by: David Ahern <dsahern@gmail.com> Acked-by: Christian Brauner <christian@brauner.io> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-10-07 20:16:27 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Use copy_to_user() for optval in netlink_getsockopt(). Brad Spencer provided a detailed report [0] that when calling getsockopt() for AF_NETLINK, some SOL_NETLINK options set only 1 byte even though such options require at least sizeof(int) as length. The options return a flag value that fits into 1 byte, but such behaviour confuses users who do not initialise the variable before calling getsockopt() and do not strictly check the returned value as char. Currently, netlink_getsockopt() uses put_user() to copy data to optlen and optval, but put_user() casts the data based on the pointer, char *optval. As a result, only 1 byte is set to optval. To avoid this behaviour, we need to use copy_to_user() or cast optval for put_user(). Note that this changes the behaviour on big-endian systems, but we document that the size of optval is int in the man page. $ man 7 netlink ... Socket options To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int. Fixes: 9a4595bc7e67 ("[NETLINK]: Add set/getsockopt options to support more than 32 groups") Fixes: be0c22a46cfb ("netlink: add NETLINK_BROADCAST_ERROR socket option") Fixes: 38938bfe3489 ("netlink: add NETLINK_NO_ENOBUFS socket flag") Fixes: 0a6a3a23ea6e ("netlink: add NETLINK_CAP_ACK socket option") Fixes: 2d4bc93368f5 ("netlink: extended ACK reporting") Fixes: 89d35528d17d ("netlink: Add new socket option to enable strict checking on dumps") Reported-by: Brad Spencer <bspencer@blackberry.com> Link: https://lore.kernel.org/netdev/ZD7VkNWFfp22kTDt@datsun.rim.net/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Johannes Berg <johannes@sipsolutions.net> Link: https://lore.kernel.org/r/20230421185255.94606-1-kuniyu@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-21 11:52:55 -07:00
netlink: Only check file credentials for implicit destinations It was possible to get a setuid root or setcap executable to write to it's stdout or stderr (which has been set made a netlink socket) and inadvertently reconfigure the networking stack. To prevent this we check that both the creator of the socket and the currentl applications has permission to reconfigure the network stack. Unfortunately this breaks Zebra which always uses sendto/sendmsg and creates it's socket without any privileges. To keep Zebra working don't bother checking if the creator of the socket has privilege when a destination address is specified. Instead rely exclusively on the privileges of the sender of the socket. Note from Andy: This is exactly Eric's code except for some comment clarifications and formatting fixes. Neither I nor, I think, anyone else is thrilled with this approach, but I'm hesitant to wait on a better fix since 3.15 is almost here. Note to stable maintainers: This is a mess. An earlier series of patches in 3.15 fix a rather serious security issue (CVE-2014-0181), but they did so in a way that breaks Zebra. The offending series includes: commit aa4cf9452f469f16cea8c96283b641b4576d4a7b Author: Eric W. Biederman <ebiederm@xmission.com> Date: Wed Apr 23 14:28:03 2014 -0700 net: Add variants of capable for use on netlink messages If a given kernel version is missing that series of fixes, it's probably worth backporting it and this patch. if that series is present, then this fix is critical if you care about Zebra. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Andy Lutomirski <luto@amacapital.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-30 11:04:00 -07:00
net: netlink: af_netlink: Prevent empty skb by adding a check on len. Adding a check on len parameter to avoid empty skb. This prevents a division error in netem_enqueue function which is caused when skb->len=0 and skb->data_len=0 in the randomized corruption step as shown below. skb->data[prandom_u32() % skb_headlen(skb)] ^= 1<<(prandom_u32() % 8); Crash Report: [ 343.170349] netdevsim netdevsim0 netdevsim3: set [1, 0] type 2 family 0 port 6081 - 0 [ 343.216110] netem: version 1.3 [ 343.235841] divide error: 0000 [#1] PREEMPT SMP KASAN NOPTI [ 343.236680] CPU: 3 PID: 4288 Comm: reproducer Not tainted 5.16.0-rc1+ [ 343.237569] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014 [ 343.238707] RIP: 0010:netem_enqueue+0x1590/0x33c0 [sch_netem] [ 343.239499] Code: 89 85 58 ff ff ff e8 5f 5d e9 d3 48 8b b5 48 ff ff ff 8b 8d 50 ff ff ff 8b 85 58 ff ff ff 48 8b bd 70 ff ff ff 31 d2 2b 4f 74 <f7> f1 48 b8 00 00 00 00 00 fc ff df 49 01 d5 4c 89 e9 48 c1 e9 03 [ 343.241883] RSP: 0018:ffff88800bcd7368 EFLAGS: 00010246 [ 343.242589] RAX: 00000000ba7c0a9c RBX: 0000000000000001 RCX: 0000000000000000 [ 343.243542] RDX: 0000000000000000 RSI: ffff88800f8edb10 RDI: ffff88800f8eda40 [ 343.244474] RBP: ffff88800bcd7458 R08: 0000000000000000 R09: ffffffff94fb8445 [ 343.245403] R10: ffffffff94fb8336 R11: ffffffff94fb8445 R12: 0000000000000000 [ 343.246355] R13: ffff88800a5a7000 R14: ffff88800a5b5800 R15: 0000000000000020 [ 343.247291] FS: 00007fdde2bd7700(0000) GS:ffff888109780000(0000) knlGS:0000000000000000 [ 343.248350] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 343.249120] CR2: 00000000200000c0 CR3: 000000000ef4c000 CR4: 00000000000006e0 [ 343.250076] Call Trace: [ 343.250423] <TASK> [ 343.250713] ? memcpy+0x4d/0x60 [ 343.251162] ? netem_init+0xa0/0xa0 [sch_netem] [ 343.251795] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.252443] netem_enqueue+0xe28/0x33c0 [sch_netem] [ 343.253102] ? stack_trace_save+0x87/0xb0 [ 343.253655] ? filter_irq_stacks+0xb0/0xb0 [ 343.254220] ? netem_init+0xa0/0xa0 [sch_netem] [ 343.254837] ? __kasan_check_write+0x14/0x20 [ 343.255418] ? _raw_spin_lock+0x88/0xd6 [ 343.255953] dev_qdisc_enqueue+0x50/0x180 [ 343.256508] __dev_queue_xmit+0x1a7e/0x3090 [ 343.257083] ? netdev_core_pick_tx+0x300/0x300 [ 343.257690] ? check_kcov_mode+0x10/0x40 [ 343.258219] ? _raw_spin_unlock_irqrestore+0x29/0x40 [ 343.258899] ? __kasan_init_slab_obj+0x24/0x30 [ 343.259529] ? setup_object.isra.71+0x23/0x90 [ 343.260121] ? new_slab+0x26e/0x4b0 [ 343.260609] ? kasan_poison+0x3a/0x50 [ 343.261118] ? kasan_unpoison+0x28/0x50 [ 343.261637] ? __kasan_slab_alloc+0x71/0x90 [ 343.262214] ? memcpy+0x4d/0x60 [ 343.262674] ? write_comp_data+0x2f/0x90 [ 343.263209] ? __kasan_check_write+0x14/0x20 [ 343.263802] ? __skb_clone+0x5d6/0x840 [ 343.264329] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.264958] dev_queue_xmit+0x1c/0x20 [ 343.265470] netlink_deliver_tap+0x652/0x9c0 [ 343.266067] netlink_unicast+0x5a0/0x7f0 [ 343.266608] ? netlink_attachskb+0x860/0x860 [ 343.267183] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.267820] ? write_comp_data+0x2f/0x90 [ 343.268367] netlink_sendmsg+0x922/0xe80 [ 343.268899] ? netlink_unicast+0x7f0/0x7f0 [ 343.269472] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.270099] ? write_comp_data+0x2f/0x90 [ 343.270644] ? netlink_unicast+0x7f0/0x7f0 [ 343.271210] sock_sendmsg+0x155/0x190 [ 343.271721] ____sys_sendmsg+0x75f/0x8f0 [ 343.272262] ? kernel_sendmsg+0x60/0x60 [ 343.272788] ? write_comp_data+0x2f/0x90 [ 343.273332] ? write_comp_data+0x2f/0x90 [ 343.273869] ___sys_sendmsg+0x10f/0x190 [ 343.274405] ? sendmsg_copy_msghdr+0x80/0x80 [ 343.274984] ? slab_post_alloc_hook+0x70/0x230 [ 343.275597] ? futex_wait_setup+0x240/0x240 [ 343.276175] ? security_file_alloc+0x3e/0x170 [ 343.276779] ? write_comp_data+0x2f/0x90 [ 343.277313] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.277969] ? write_comp_data+0x2f/0x90 [ 343.278515] ? __fget_files+0x1ad/0x260 [ 343.279048] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.279685] ? write_comp_data+0x2f/0x90 [ 343.280234] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.280874] ? sockfd_lookup_light+0xd1/0x190 [ 343.281481] __sys_sendmsg+0x118/0x200 [ 343.281998] ? __sys_sendmsg_sock+0x40/0x40 [ 343.282578] ? alloc_fd+0x229/0x5e0 [ 343.283070] ? write_comp_data+0x2f/0x90 [ 343.283610] ? write_comp_data+0x2f/0x90 [ 343.284135] ? __sanitizer_cov_trace_pc+0x21/0x60 [ 343.284776] ? ktime_get_coarse_real_ts64+0xb8/0xf0 [ 343.285450] __x64_sys_sendmsg+0x7d/0xc0 [ 343.285981] ? syscall_enter_from_user_mode+0x4d/0x70 [ 343.286664] do_syscall_64+0x3a/0x80 [ 343.287158] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 343.287850] RIP: 0033:0x7fdde24cf289 [ 343.288344] Code: 01 00 48 81 c4 80 00 00 00 e9 f1 fe ff ff 0f 1f 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d b7 db 2c 00 f7 d8 64 89 01 48 [ 343.290729] RSP: 002b:00007fdde2bd6d98 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 343.291730] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fdde24cf289 [ 343.292673] RDX: 0000000000000000 RSI: 00000000200000c0 RDI: 0000000000000004 [ 343.293618] RBP: 00007fdde2bd6e20 R08: 0000000100000001 R09: 0000000000000000 [ 343.294557] R10: 0000000100000001 R11: 0000000000000246 R12: 0000000000000000 [ 343.295493] R13: 0000000000021000 R14: 0000000000000000 R15: 00007fdde2bd7700 [ 343.296432] </TASK> [ 343.296735] Modules linked in: sch_netem ip6_vti ip_vti ip_gre ipip sit ip_tunnel geneve macsec macvtap tap ipvlan macvlan 8021q garp mrp hsr wireguard libchacha20poly1305 chacha_x86_64 poly1305_x86_64 ip6_udp_tunnel udp_tunnel libblake2s blake2s_x86_64 libblake2s_generic curve25519_x86_64 libcurve25519_generic libchacha xfrm_interface xfrm6_tunnel tunnel4 veth netdevsim psample batman_adv nlmon dummy team bonding tls vcan ip6_gre ip6_tunnel tunnel6 gre tun ip6t_rpfilter ipt_REJECT nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set ebtable_nat ebtable_broute ip6table_nat ip6table_mangle ip6table_security ip6table_raw iptable_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 iptable_mangle iptable_security iptable_raw ebtable_filter ebtables rfkill ip6table_filter ip6_tables iptable_filter ppdev bochs drm_vram_helper drm_ttm_helper ttm drm_kms_helper cec parport_pc drm joydev floppy parport sg syscopyarea sysfillrect sysimgblt i2c_piix4 qemu_fw_cfg fb_sys_fops pcspkr [ 343.297459] ip_tables xfs virtio_net net_failover failover sd_mod sr_mod cdrom t10_pi ata_generic pata_acpi ata_piix libata virtio_pci virtio_pci_legacy_dev serio_raw virtio_pci_modern_dev dm_mirror dm_region_hash dm_log dm_mod [ 343.311074] Dumping ftrace buffer: [ 343.311532] (ftrace buffer empty) [ 343.312040] ---[ end trace a2e3db5a6ae05099 ]--- [ 343.312691] RIP: 0010:netem_enqueue+0x1590/0x33c0 [sch_netem] [ 343.313481] Code: 89 85 58 ff ff ff e8 5f 5d e9 d3 48 8b b5 48 ff ff ff 8b 8d 50 ff ff ff 8b 85 58 ff ff ff 48 8b bd 70 ff ff ff 31 d2 2b 4f 74 <f7> f1 48 b8 00 00 00 00 00 fc ff df 49 01 d5 4c 89 e9 48 c1 e9 03 [ 343.315893] RSP: 0018:ffff88800bcd7368 EFLAGS: 00010246 [ 343.316622] RAX: 00000000ba7c0a9c RBX: 0000000000000001 RCX: 0000000000000000 [ 343.317585] RDX: 0000000000000000 RSI: ffff88800f8edb10 RDI: ffff88800f8eda40 [ 343.318549] RBP: ffff88800bcd7458 R08: 0000000000000000 R09: ffffffff94fb8445 [ 343.319503] R10: ffffffff94fb8336 R11: ffffffff94fb8445 R12: 0000000000000000 [ 343.320455] R13: ffff88800a5a7000 R14: ffff88800a5b5800 R15: 0000000000000020 [ 343.321414] FS: 00007fdde2bd7700(0000) GS:ffff888109780000(0000) knlGS:0000000000000000 [ 343.322489] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 343.323283] CR2: 00000000200000c0 CR3: 000000000ef4c000 CR4: 00000000000006e0 [ 343.324264] Kernel panic - not syncing: Fatal exception in interrupt [ 343.333717] Dumping ftrace buffer: [ 343.334175] (ftrace buffer empty) [ 343.334653] Kernel Offset: 0x13600000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 343.336027] Rebooting in 86400 seconds.. Reported-by: syzkaller <syzkaller@googlegroups.com> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Link: https://lore.kernel.org/r/20211129175328.55339-1-harshit.m.mogalapalli@oracle.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-29 09:53:27 -08:00
netlink: Only check file credentials for implicit destinations It was possible to get a setuid root or setcap executable to write to it's stdout or stderr (which has been set made a netlink socket) and inadvertently reconfigure the networking stack. To prevent this we check that both the creator of the socket and the currentl applications has permission to reconfigure the network stack. Unfortunately this breaks Zebra which always uses sendto/sendmsg and creates it's socket without any privileges. To keep Zebra working don't bother checking if the creator of the socket has privilege when a destination address is specified. Instead rely exclusively on the privileges of the sender of the socket. Note from Andy: This is exactly Eric's code except for some comment clarifications and formatting fixes. Neither I nor, I think, anyone else is thrilled with this approach, but I'm hesitant to wait on a better fix since 3.15 is almost here. Note to stable maintainers: This is a mess. An earlier series of patches in 3.15 fix a rather serious security issue (CVE-2014-0181), but they did so in a way that breaks Zebra. The offending series includes: commit aa4cf9452f469f16cea8c96283b641b4576d4a7b Author: Eric W. Biederman <ebiederm@xmission.com> Date: Wed Apr 23 14:28:03 2014 -0700 net: Add variants of capable for use on netlink messages If a given kernel version is missing that series of fixes, it's probably worth backporting it and this patch. if that series is present, then this fix is critical if you care about Zebra. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Andy Lutomirski <luto@amacapital.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-30 11:04:00 -07:00
netlink: annotate data races around nlk->bound While existing code is correct, KCSAN is reporting a data-race in netlink_insert / netlink_sendmsg [1] It is correct to read nlk->bound without a lock, as netlink_autobind() will acquire all needed locks. [1] BUG: KCSAN: data-race in netlink_insert / netlink_sendmsg write to 0xffff8881031c8b30 of 1 bytes by task 18752 on cpu 0: netlink_insert+0x5cc/0x7f0 net/netlink/af_netlink.c:597 netlink_autobind+0xa9/0x150 net/netlink/af_netlink.c:842 netlink_sendmsg+0x479/0x7c0 net/netlink/af_netlink.c:1892 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmsg+0x1ed/0x270 net/socket.c:2475 __do_sys_sendmsg net/socket.c:2484 [inline] __se_sys_sendmsg net/socket.c:2482 [inline] __x64_sys_sendmsg+0x42/0x50 net/socket.c:2482 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff8881031c8b30 of 1 bytes by task 18751 on cpu 1: netlink_sendmsg+0x270/0x7c0 net/netlink/af_netlink.c:1891 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] __sys_sendto+0x2a8/0x370 net/socket.c:2019 __do_sys_sendto net/socket.c:2031 [inline] __se_sys_sendto net/socket.c:2027 [inline] __x64_sys_sendto+0x74/0x90 net/socket.c:2027 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00 -> 0x01 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 18751 Comm: syz-executor.0 Not tainted 5.14.0-rc1-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: da314c9923fe ("netlink: Replace rhash_portid with bound") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-04 14:24:15 -07:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00
netlink: fix splat in skb_clone with large messages Since (c05cdb1 netlink: allow large data transfers from user-space), netlink splats if it invokes skb_clone on large netlink skbs since: * skb_shared_info was not correctly initialized. * skb->destructor is not set in the cloned skb. This was spotted by trinity: [ 894.990671] BUG: unable to handle kernel paging request at ffffc9000047b001 [ 894.991034] IP: [<ffffffff81a212c4>] skb_clone+0x24/0xc0 [...] [ 894.991034] Call Trace: [ 894.991034] [<ffffffff81ad299a>] nl_fib_input+0x6a/0x240 [ 894.991034] [<ffffffff81c3b7e6>] ? _raw_read_unlock+0x26/0x40 [ 894.991034] [<ffffffff81a5f189>] netlink_unicast+0x169/0x1e0 [ 894.991034] [<ffffffff81a601e1>] netlink_sendmsg+0x251/0x3d0 Fix it by: 1) introducing a new netlink_skb_clone function that is used in nl_fib_input, that sets our special skb->destructor in the cloned skb. Moreover, handle the release of the large cloned skb head area in the destructor path. 2) not allowing large skbuffs in the netlink broadcast path. I cannot find any reasonable use of the large data transfer using netlink in that path, moreover this helps to skip extra skb_clone handling. I found two more netlink clients that are cloning the skbs, but they are not in the sendmsg path. Therefore, the sole client cloning that I found seems to be the fib frontend. Thanks to Eric Dumazet for helping to address this issue. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-06-28 03:04:23 +02:00
netlink: Only check file credentials for implicit destinations It was possible to get a setuid root or setcap executable to write to it's stdout or stderr (which has been set made a netlink socket) and inadvertently reconfigure the networking stack. To prevent this we check that both the creator of the socket and the currentl applications has permission to reconfigure the network stack. Unfortunately this breaks Zebra which always uses sendto/sendmsg and creates it's socket without any privileges. To keep Zebra working don't bother checking if the creator of the socket has privilege when a destination address is specified. Instead rely exclusively on the privileges of the sender of the socket. Note from Andy: This is exactly Eric's code except for some comment clarifications and formatting fixes. Neither I nor, I think, anyone else is thrilled with this approach, but I'm hesitant to wait on a better fix since 3.15 is almost here. Note to stable maintainers: This is a mess. An earlier series of patches in 3.15 fix a rather serious security issue (CVE-2014-0181), but they did so in a way that breaks Zebra. The offending series includes: commit aa4cf9452f469f16cea8c96283b641b4576d4a7b Author: Eric W. Biederman <ebiederm@xmission.com> Date: Wed Apr 23 14:28:03 2014 -0700 net: Add variants of capable for use on netlink messages If a given kernel version is missing that series of fixes, it's probably worth backporting it and this patch. if that series is present, then this fix is critical if you care about Zebra. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Andy Lutomirski <luto@amacapital.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-30 11:04:00 -07:00
netlink: annotate lockless accesses to nlk->max_recvmsg_len syzbot reported a data-race in data-race in netlink_recvmsg() [1] Indeed, netlink_recvmsg() can be run concurrently, and netlink_dump() also needs protection. [1] BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg read to 0xffff888141840b38 of 8 bytes by task 23057 on cpu 0: netlink_recvmsg+0xea/0x730 net/netlink/af_netlink.c:1988 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] __sys_recvfrom+0x1ee/0x2e0 net/socket.c:2194 __do_sys_recvfrom net/socket.c:2212 [inline] __se_sys_recvfrom net/socket.c:2208 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2208 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff888141840b38 of 8 bytes by task 23037 on cpu 1: netlink_recvmsg+0x114/0x730 net/netlink/af_netlink.c:1989 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] ____sys_recvmsg+0x156/0x310 net/socket.c:2720 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x0000000000000000 -> 0x0000000000001000 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 23037 Comm: syz-executor.2 Not tainted 6.3.0-rc4-syzkaller-00195-g5a57b48fdfcb #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023 Fixes: 9063e21fb026 ("netlink: autosize skb lengthes") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/20230403214643.768555-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 21:46:43 +00:00
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
netlink: annotate lockless accesses to nlk->max_recvmsg_len syzbot reported a data-race in data-race in netlink_recvmsg() [1] Indeed, netlink_recvmsg() can be run concurrently, and netlink_dump() also needs protection. [1] BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg read to 0xffff888141840b38 of 8 bytes by task 23057 on cpu 0: netlink_recvmsg+0xea/0x730 net/netlink/af_netlink.c:1988 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] __sys_recvfrom+0x1ee/0x2e0 net/socket.c:2194 __do_sys_recvfrom net/socket.c:2212 [inline] __se_sys_recvfrom net/socket.c:2208 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2208 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff888141840b38 of 8 bytes by task 23037 on cpu 1: netlink_recvmsg+0x114/0x730 net/netlink/af_netlink.c:1989 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] ____sys_recvmsg+0x156/0x310 net/socket.c:2720 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x0000000000000000 -> 0x0000000000001000 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 23037 Comm: syz-executor.2 Not tainted 6.3.0-rc4-syzkaller-00195-g5a57b48fdfcb #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023 Fixes: 9063e21fb026 ("netlink: autosize skb lengthes") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/20230403214643.768555-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 21:46:43 +00:00
netlink: annotate accesses to nlk->cb_running Both netlink_recvmsg() and netlink_native_seq_show() read nlk->cb_running locklessly. Use READ_ONCE() there. Add corresponding WRITE_ONCE() to netlink_dump() and __netlink_dump_start() syzbot reported: BUG: KCSAN: data-race in __netlink_dump_start / netlink_recvmsg write to 0xffff88813ea4db59 of 1 bytes by task 28219 on cpu 0: __netlink_dump_start+0x3af/0x4d0 net/netlink/af_netlink.c:2399 netlink_dump_start include/linux/netlink.h:308 [inline] rtnetlink_rcv_msg+0x70f/0x8c0 net/core/rtnetlink.c:6130 netlink_rcv_skb+0x126/0x220 net/netlink/af_netlink.c:2577 rtnetlink_rcv+0x1c/0x20 net/core/rtnetlink.c:6192 netlink_unicast_kernel net/netlink/af_netlink.c:1339 [inline] netlink_unicast+0x56f/0x640 net/netlink/af_netlink.c:1365 netlink_sendmsg+0x665/0x770 net/netlink/af_netlink.c:1942 sock_sendmsg_nosec net/socket.c:724 [inline] sock_sendmsg net/socket.c:747 [inline] sock_write_iter+0x1aa/0x230 net/socket.c:1138 call_write_iter include/linux/fs.h:1851 [inline] new_sync_write fs/read_write.c:491 [inline] vfs_write+0x463/0x760 fs/read_write.c:584 ksys_write+0xeb/0x1a0 fs/read_write.c:637 __do_sys_write fs/read_write.c:649 [inline] __se_sys_write fs/read_write.c:646 [inline] __x64_sys_write+0x42/0x50 fs/read_write.c:646 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88813ea4db59 of 1 bytes by task 28222 on cpu 1: netlink_recvmsg+0x3b4/0x730 net/netlink/af_netlink.c:2022 sock_recvmsg_nosec+0x4c/0x80 net/socket.c:1017 ____sys_recvmsg+0x2db/0x310 net/socket.c:2718 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00 -> 0x01 Fixes: 16b304f3404f ("netlink: Eliminate kmalloc in netlink dump operation.") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-05-09 16:56:34 +00:00
netlink: annotate data-races around sk->sk_err syzbot caught another data-race in netlink when setting sk->sk_err. Annotate all of them for good measure. BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg write to 0xffff8881613bb220 of 4 bytes by task 28147 on cpu 0: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff8881613bb220 of 4 bytes by task 28146 on cpu 1: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0x00000016 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 28146 Comm: syz-executor.0 Not tainted 6.6.0-rc3-syzkaller-00055-g9ed22ae6be81 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/06/2023 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20231003183455.3410550-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-10-03 18:34:55 +00:00
[NETNS]: Fix race between put_net() and netlink_kernel_create(). The comment about "race free view of the set of network namespaces" was a bit hasty. Look (there even can be only one CPU, as discovered by Alexey Dobriyan and Denis Lunev): put_net() if (atomic_dec_and_test(&net->refcnt)) /* true */ __put_net(net); queue_work(...); /* * note: the net now has refcnt 0, but still in * the global list of net namespaces */ == re-schedule == register_pernet_subsys(&some_ops); register_pernet_operations(&some_ops); (*some_ops)->init(net); /* * we call netlink_kernel_create() here * in some places */ netlink_kernel_create(); sk_alloc(); get_net(net); /* refcnt = 1 */ /* * now we drop the net refcount not to * block the net namespace exit in the * future (or this can be done on the * error path) */ put_net(sk->sk_net); if (atomic_dec_and_test(&...)) /* * true. BOOOM! The net is * scheduled for release twice */ When thinking on this problem, I decided, that getting and putting the net in init callback is wrong. If some init callback needs to have a refcount-less reference on the struct net, _it_ has to be careful himself, rather than relying on the infrastructure to handle this correctly. In case of netlink_kernel_create(), the problem is that the sk_alloc() gets the given namespace, but passing the info that we don't want to get it inside this call is too heavy. Instead, I propose to crate the socket inside an init_net namespace and then re-attach it to the desired one right after the socket is created. After doing this, we also have to be careful on error paths not to drop the reference on the namespace, we didn't get the one on. Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Denis Lunev <den@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-30 19:31:06 -08:00
[NETNS]: Fix race between put_net() and netlink_kernel_create(). The comment about "race free view of the set of network namespaces" was a bit hasty. Look (there even can be only one CPU, as discovered by Alexey Dobriyan and Denis Lunev): put_net() if (atomic_dec_and_test(&net->refcnt)) /* true */ __put_net(net); queue_work(...); /* * note: the net now has refcnt 0, but still in * the global list of net namespaces */ == re-schedule == register_pernet_subsys(&some_ops); register_pernet_operations(&some_ops); (*some_ops)->init(net); /* * we call netlink_kernel_create() here * in some places */ netlink_kernel_create(); sk_alloc(); get_net(net); /* refcnt = 1 */ /* * now we drop the net refcount not to * block the net namespace exit in the * future (or this can be done on the * error path) */ put_net(sk->sk_net); if (atomic_dec_and_test(&...)) /* * true. BOOOM! The net is * scheduled for release twice */ When thinking on this problem, I decided, that getting and putting the net in init callback is wrong. If some init callback needs to have a refcount-less reference on the struct net, _it_ has to be careful himself, rather than relying on the infrastructure to handle this correctly. In case of netlink_kernel_create(), the problem is that the sk_alloc() gets the given namespace, but passing the info that we don't want to get it inside this call is too heavy. Instead, I propose to crate the socket inside an init_net namespace and then re-attach it to the desired one right after the socket is created. After doing this, we also have to be careful on error paths not to drop the reference on the namespace, we didn't get the one on. Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Denis Lunev <den@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-30 19:31:06 -08:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: annotate lockless accesses to nlk->max_recvmsg_len syzbot reported a data-race in data-race in netlink_recvmsg() [1] Indeed, netlink_recvmsg() can be run concurrently, and netlink_dump() also needs protection. [1] BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg read to 0xffff888141840b38 of 8 bytes by task 23057 on cpu 0: netlink_recvmsg+0xea/0x730 net/netlink/af_netlink.c:1988 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] __sys_recvfrom+0x1ee/0x2e0 net/socket.c:2194 __do_sys_recvfrom net/socket.c:2212 [inline] __se_sys_recvfrom net/socket.c:2208 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2208 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff888141840b38 of 8 bytes by task 23037 on cpu 1: netlink_recvmsg+0x114/0x730 net/netlink/af_netlink.c:1989 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] ____sys_recvmsg+0x156/0x310 net/socket.c:2720 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x0000000000000000 -> 0x0000000000001000 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 23037 Comm: syz-executor.2 Not tainted 6.3.0-rc4-syzkaller-00195-g5a57b48fdfcb #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023 Fixes: 9063e21fb026 ("netlink: autosize skb lengthes") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/20230403214643.768555-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 21:46:43 +00:00
af_netlink: ensure that NLMSG_DONE never fails in dumps The way people generally use netlink_dump is that they fill in the skb as much as possible, breaking when nla_put returns an error. Then, they get called again and start filling out the next skb, and again, and so forth. The mechanism at work here is the ability for the iterative dumping function to detect when the skb is filled up and not fill it past the brim, waiting for a fresh skb for the rest of the data. However, if the attributes are small and nicely packed, it is possible that a dump callback function successfully fills in attributes until the skb is of size 4080 (libmnl's default page-sized receive buffer size). The dump function completes, satisfied, and then, if it happens to be that this is actually the last skb, and no further ones are to be sent, then netlink_dump will add on the NLMSG_DONE part: nlh = nlmsg_put_answer(skb, cb, NLMSG_DONE, sizeof(len), NLM_F_MULTI); It is very important that netlink_dump does this, of course. However, in this example, that call to nlmsg_put_answer will fail, because the previous filling by the dump function did not leave it enough room. And how could it possibly have done so? All of the nla_put variety of functions simply check to see if the skb has enough tailroom, independent of the context it is in. In order to keep the important assumptions of all netlink dump users, it is therefore important to give them an skb that has this end part of the tail already reserved, so that the call to nlmsg_put_answer does not fail. Otherwise, library authors are forced to find some bizarre sized receive buffer that has a large modulo relative to the common sizes of messages received, which is ugly and buggy. This patch thus saves the NLMSG_DONE for an additional message, for the case that things are dangerously close to the brim. This requires keeping track of the errno from ->dump() across calls. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-09 13:04:44 +09:00
netlink: annotate lockless accesses to nlk->max_recvmsg_len syzbot reported a data-race in data-race in netlink_recvmsg() [1] Indeed, netlink_recvmsg() can be run concurrently, and netlink_dump() also needs protection. [1] BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg read to 0xffff888141840b38 of 8 bytes by task 23057 on cpu 0: netlink_recvmsg+0xea/0x730 net/netlink/af_netlink.c:1988 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] __sys_recvfrom+0x1ee/0x2e0 net/socket.c:2194 __do_sys_recvfrom net/socket.c:2212 [inline] __se_sys_recvfrom net/socket.c:2208 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2208 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff888141840b38 of 8 bytes by task 23037 on cpu 1: netlink_recvmsg+0x114/0x730 net/netlink/af_netlink.c:1989 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg net/socket.c:1038 [inline] ____sys_recvmsg+0x156/0x310 net/socket.c:2720 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x0000000000000000 -> 0x0000000000001000 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 23037 Comm: syz-executor.2 Not tainted 6.3.0-rc4-syzkaller-00195-g5a57b48fdfcb #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023 Fixes: 9063e21fb026 ("netlink: autosize skb lengthes") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Simon Horman <simon.horman@corigine.com> Link: https://lore.kernel.org/r/20230403214643.768555-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-04-03 21:46:43 +00:00
netlink: remove mmapped netlink support mmapped netlink has a number of unresolved issues: - TX zerocopy support had to be disabled more than a year ago via commit 4682a0358639b29cf ("netlink: Always copy on mmap TX.") because the content of the mmapped area can change after netlink attribute validation but before message processing. - RX support was implemented mainly to speed up nfqueue dumping packet payload to userspace. However, since commit ae08ce0021087a5d812d2 ("netfilter: nfnetlink_queue: zero copy support") we avoid one copy with the socket-based interface too (via the skb_zerocopy helper). The other problem is that skbs attached to mmaped netlink socket behave different from normal skbs: - they don't have a shinfo area, so all functions that use skb_shinfo() (e.g. skb_clone) cannot be used. - reserving headroom prevents userspace from seeing the content as it expects message to start at skb->head. See for instance commit aa3a022094fa ("netlink: not trim skb for mmaped socket when dump"). - skbs handed e.g. to netlink_ack must have non-NULL skb->sk, else we crash because it needs the sk to check if a tx ring is attached. Also not obvious, leads to non-intuitive bug fixes such as 7c7bdf359 ("netfilter: nfnetlink: use original skbuff when acking batches"). mmaped netlink also didn't play nicely with the skb_zerocopy helper used by nfqueue and openvswitch. Daniel Borkmann fixed this via commit 6bb0fef489f6 ("netlink, mmap: fix edge-case leakages in nf queue zero-copy")' but at the cost of also needing to provide remaining length to the allocation function. nfqueue also has problems when used with mmaped rx netlink: - mmaped netlink doesn't allow use of nfqueue batch verdict messages. Problem is that in the mmap case, the allocation time also determines the ordering in which the frame will be seen by userspace (A allocating before B means that A is located in earlier ring slot, but this also means that B might get a lower sequence number then A since seqno is decided later. To fix this we would need to extend the spinlocked region to also cover the allocation and message setup which isn't desirable. - nfqueue can now be configured to queue large (GSO) skbs to userspace. Queing GSO packets is faster than having to force a software segmentation in the kernel, so this is a desirable option. However, with a mmap based ring one has to use 64kb per ring slot element, else mmap has to fall back to the socket path (NL_MMAP_STATUS_COPY) for all large packets. To use the mmap interface, userspace not only has to probe for mmap netlink support, it also has to implement a recv/socket receive path in order to handle messages that exceed the size of an rx ring element. Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Ken-ichirou MATSUZAWA <chamaken@gmail.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Cc: Patrick McHardy <kaber@trash.net> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-02-18 15:03:24 +01:00
netlink: reset network and mac headers in netlink_dump() netlink_dump() is allocating an skb, reserves space in it but forgets to reset network header. This allows a BPF program, invoked later from sk_filter() to access uninitialized kernel memory from the reserved space. Theorically mac header reset could be omitted, because it is set to a special initial value. bpf_internal_load_pointer_neg_helper calls skb_mac_header() without checking skb_mac_header_was_set(). Relying on skb->len not being too big seems fragile. We also could add a sanity check in bpf_internal_load_pointer_neg_helper() to avoid surprises in the future. syzbot report was: BUG: KMSAN: uninit-value in ___bpf_prog_run+0xa22b/0xb420 kernel/bpf/core.c:1637 ___bpf_prog_run+0xa22b/0xb420 kernel/bpf/core.c:1637 __bpf_prog_run32+0x121/0x180 kernel/bpf/core.c:1796 bpf_dispatcher_nop_func include/linux/bpf.h:784 [inline] __bpf_prog_run include/linux/filter.h:626 [inline] bpf_prog_run include/linux/filter.h:633 [inline] __bpf_prog_run_save_cb+0x168/0x580 include/linux/filter.h:756 bpf_prog_run_save_cb include/linux/filter.h:770 [inline] sk_filter_trim_cap+0x3bc/0x8c0 net/core/filter.c:150 sk_filter include/linux/filter.h:905 [inline] netlink_dump+0xe0c/0x16c0 net/netlink/af_netlink.c:2276 netlink_recvmsg+0x1129/0x1c80 net/netlink/af_netlink.c:2002 sock_recvmsg_nosec net/socket.c:948 [inline] sock_recvmsg net/socket.c:966 [inline] sock_read_iter+0x5a9/0x630 net/socket.c:1039 do_iter_readv_writev+0xa7f/0xc70 do_iter_read+0x52c/0x14c0 fs/read_write.c:786 vfs_readv fs/read_write.c:906 [inline] do_readv+0x432/0x800 fs/read_write.c:943 __do_sys_readv fs/read_write.c:1034 [inline] __se_sys_readv fs/read_write.c:1031 [inline] __x64_sys_readv+0xe5/0x120 fs/read_write.c:1031 do_syscall_x64 arch/x86/entry/common.c:51 [inline] do_syscall_64+0x54/0xd0 arch/x86/entry/common.c:81 entry_SYSCALL_64_after_hwframe+0x44/0xae Uninit was stored to memory at: ___bpf_prog_run+0x96c/0xb420 kernel/bpf/core.c:1558 __bpf_prog_run32+0x121/0x180 kernel/bpf/core.c:1796 bpf_dispatcher_nop_func include/linux/bpf.h:784 [inline] __bpf_prog_run include/linux/filter.h:626 [inline] bpf_prog_run include/linux/filter.h:633 [inline] __bpf_prog_run_save_cb+0x168/0x580 include/linux/filter.h:756 bpf_prog_run_save_cb include/linux/filter.h:770 [inline] sk_filter_trim_cap+0x3bc/0x8c0 net/core/filter.c:150 sk_filter include/linux/filter.h:905 [inline] netlink_dump+0xe0c/0x16c0 net/netlink/af_netlink.c:2276 netlink_recvmsg+0x1129/0x1c80 net/netlink/af_netlink.c:2002 sock_recvmsg_nosec net/socket.c:948 [inline] sock_recvmsg net/socket.c:966 [inline] sock_read_iter+0x5a9/0x630 net/socket.c:1039 do_iter_readv_writev+0xa7f/0xc70 do_iter_read+0x52c/0x14c0 fs/read_write.c:786 vfs_readv fs/read_write.c:906 [inline] do_readv+0x432/0x800 fs/read_write.c:943 __do_sys_readv fs/read_write.c:1034 [inline] __se_sys_readv fs/read_write.c:1031 [inline] __x64_sys_readv+0xe5/0x120 fs/read_write.c:1031 do_syscall_x64 arch/x86/entry/common.c:51 [inline] do_syscall_64+0x54/0xd0 arch/x86/entry/common.c:81 entry_SYSCALL_64_after_hwframe+0x44/0xae Uninit was created at: slab_post_alloc_hook mm/slab.h:737 [inline] slab_alloc_node mm/slub.c:3244 [inline] __kmalloc_node_track_caller+0xde3/0x14f0 mm/slub.c:4972 kmalloc_reserve net/core/skbuff.c:354 [inline] __alloc_skb+0x545/0xf90 net/core/skbuff.c:426 alloc_skb include/linux/skbuff.h:1158 [inline] netlink_dump+0x30f/0x16c0 net/netlink/af_netlink.c:2242 netlink_recvmsg+0x1129/0x1c80 net/netlink/af_netlink.c:2002 sock_recvmsg_nosec net/socket.c:948 [inline] sock_recvmsg net/socket.c:966 [inline] sock_read_iter+0x5a9/0x630 net/socket.c:1039 do_iter_readv_writev+0xa7f/0xc70 do_iter_read+0x52c/0x14c0 fs/read_write.c:786 vfs_readv fs/read_write.c:906 [inline] do_readv+0x432/0x800 fs/read_write.c:943 __do_sys_readv fs/read_write.c:1034 [inline] __se_sys_readv fs/read_write.c:1031 [inline] __x64_sys_readv+0xe5/0x120 fs/read_write.c:1031 do_syscall_x64 arch/x86/entry/common.c:51 [inline] do_syscall_64+0x54/0xd0 arch/x86/entry/common.c:81 entry_SYSCALL_64_after_hwframe+0x44/0xae CPU: 0 PID: 3470 Comm: syz-executor751 Not tainted 5.17.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: db65a3aaf29e ("netlink: Trim skb to alloc size to avoid MSG_TRUNC") Fixes: 9063e21fb026 ("netlink: autosize skb lengthes") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Link: https://lore.kernel.org/r/20220415181442.551228-1-eric.dumazet@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2022-04-15 11:14:42 -07:00
af_netlink: ensure that NLMSG_DONE never fails in dumps The way people generally use netlink_dump is that they fill in the skb as much as possible, breaking when nla_put returns an error. Then, they get called again and start filling out the next skb, and again, and so forth. The mechanism at work here is the ability for the iterative dumping function to detect when the skb is filled up and not fill it past the brim, waiting for a fresh skb for the rest of the data. However, if the attributes are small and nicely packed, it is possible that a dump callback function successfully fills in attributes until the skb is of size 4080 (libmnl's default page-sized receive buffer size). The dump function completes, satisfied, and then, if it happens to be that this is actually the last skb, and no further ones are to be sent, then netlink_dump will add on the NLMSG_DONE part: nlh = nlmsg_put_answer(skb, cb, NLMSG_DONE, sizeof(len), NLM_F_MULTI); It is very important that netlink_dump does this, of course. However, in this example, that call to nlmsg_put_answer will fail, because the previous filling by the dump function did not leave it enough room. And how could it possibly have done so? All of the nla_put variety of functions simply check to see if the skb has enough tailroom, independent of the context it is in. In order to keep the important assumptions of all netlink dump users, it is therefore important to give them an skb that has this end part of the tail already reserved, so that the call to nlmsg_put_answer does not fail. Otherwise, library authors are forced to find some bizarre sized receive buffer that has a large modulo relative to the common sizes of messages received, which is ugly and buggy. This patch thus saves the NLMSG_DONE for an additional message, for the case that things are dangerously close to the brim. This requires keeping track of the errno from ->dump() across calls. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-09 13:04:44 +09:00
af_netlink: ensure that NLMSG_DONE never fails in dumps The way people generally use netlink_dump is that they fill in the skb as much as possible, breaking when nla_put returns an error. Then, they get called again and start filling out the next skb, and again, and so forth. The mechanism at work here is the ability for the iterative dumping function to detect when the skb is filled up and not fill it past the brim, waiting for a fresh skb for the rest of the data. However, if the attributes are small and nicely packed, it is possible that a dump callback function successfully fills in attributes until the skb is of size 4080 (libmnl's default page-sized receive buffer size). The dump function completes, satisfied, and then, if it happens to be that this is actually the last skb, and no further ones are to be sent, then netlink_dump will add on the NLMSG_DONE part: nlh = nlmsg_put_answer(skb, cb, NLMSG_DONE, sizeof(len), NLM_F_MULTI); It is very important that netlink_dump does this, of course. However, in this example, that call to nlmsg_put_answer will fail, because the previous filling by the dump function did not leave it enough room. And how could it possibly have done so? All of the nla_put variety of functions simply check to see if the skb has enough tailroom, independent of the context it is in. In order to keep the important assumptions of all netlink dump users, it is therefore important to give them an skb that has this end part of the tail already reserved, so that the call to nlmsg_put_answer does not fail. Otherwise, library authors are forced to find some bizarre sized receive buffer that has a large modulo relative to the common sizes of messages received, which is ugly and buggy. This patch thus saves the NLMSG_DONE for an additional message, for the case that things are dangerously close to the brim. This requires keeping track of the errno from ->dump() across calls. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-09 13:04:44 +09:00
netlink: annotate accesses to nlk->cb_running Both netlink_recvmsg() and netlink_native_seq_show() read nlk->cb_running locklessly. Use READ_ONCE() there. Add corresponding WRITE_ONCE() to netlink_dump() and __netlink_dump_start() syzbot reported: BUG: KCSAN: data-race in __netlink_dump_start / netlink_recvmsg write to 0xffff88813ea4db59 of 1 bytes by task 28219 on cpu 0: __netlink_dump_start+0x3af/0x4d0 net/netlink/af_netlink.c:2399 netlink_dump_start include/linux/netlink.h:308 [inline] rtnetlink_rcv_msg+0x70f/0x8c0 net/core/rtnetlink.c:6130 netlink_rcv_skb+0x126/0x220 net/netlink/af_netlink.c:2577 rtnetlink_rcv+0x1c/0x20 net/core/rtnetlink.c:6192 netlink_unicast_kernel net/netlink/af_netlink.c:1339 [inline] netlink_unicast+0x56f/0x640 net/netlink/af_netlink.c:1365 netlink_sendmsg+0x665/0x770 net/netlink/af_netlink.c:1942 sock_sendmsg_nosec net/socket.c:724 [inline] sock_sendmsg net/socket.c:747 [inline] sock_write_iter+0x1aa/0x230 net/socket.c:1138 call_write_iter include/linux/fs.h:1851 [inline] new_sync_write fs/read_write.c:491 [inline] vfs_write+0x463/0x760 fs/read_write.c:584 ksys_write+0xeb/0x1a0 fs/read_write.c:637 __do_sys_write fs/read_write.c:649 [inline] __se_sys_write fs/read_write.c:646 [inline] __x64_sys_write+0x42/0x50 fs/read_write.c:646 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88813ea4db59 of 1 bytes by task 28222 on cpu 1: netlink_recvmsg+0x3b4/0x730 net/netlink/af_netlink.c:2022 sock_recvmsg_nosec+0x4c/0x80 net/socket.c:1017 ____sys_recvmsg+0x2db/0x310 net/socket.c:2718 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00 -> 0x01 Fixes: 16b304f3404f ("netlink: Eliminate kmalloc in netlink dump operation.") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-05-09 16:56:34 +00:00
netlink: Add new socket option to enable strict checking on dumps Add a new socket option, NETLINK_DUMP_STRICT_CHK, that userspace can use via setsockopt to request strict checking of headers and attributes on dump requests. To get dump features such as kernel side filtering based on data in the header or attributes appended to the dump request, userspace must call setsockopt() for NETLINK_DUMP_STRICT_CHK and a non-zero value. Since the netlink sock and its flags are private to the af_netlink code, the strict checking flag is passed to dump handlers via a flag in the netlink_callback struct. For old userspace on new kernel there is no impact as all of the data checks in later patches are wrapped in a check on the new strict flag. For new userspace on old kernel, the setsockopt will fail and even if new userspace sets data in the headers and appended attributes the kernel will silently ignore it. Moving forward when the setsockopt succeeds, the new userspace on old kernel means the dump request can pass an attribute the kernel does not understand. The dump will then fail as the older kernel does not understand it. New userspace on new kernel setting the socket option gets the benefit of the improved data dump. Kernel side the NETLINK_DUMP_STRICT_CHK uapi is converted to a generic NETLINK_F_STRICT_CHK flag which can potentially be leveraged for tighter checking on the NEW, DEL, and SET commands. Signed-off-by: David Ahern <dsahern@gmail.com> Acked-by: Christian Brauner <christian@brauner.io> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-10-07 20:16:27 -07:00
netlink: annotate accesses to nlk->cb_running Both netlink_recvmsg() and netlink_native_seq_show() read nlk->cb_running locklessly. Use READ_ONCE() there. Add corresponding WRITE_ONCE() to netlink_dump() and __netlink_dump_start() syzbot reported: BUG: KCSAN: data-race in __netlink_dump_start / netlink_recvmsg write to 0xffff88813ea4db59 of 1 bytes by task 28219 on cpu 0: __netlink_dump_start+0x3af/0x4d0 net/netlink/af_netlink.c:2399 netlink_dump_start include/linux/netlink.h:308 [inline] rtnetlink_rcv_msg+0x70f/0x8c0 net/core/rtnetlink.c:6130 netlink_rcv_skb+0x126/0x220 net/netlink/af_netlink.c:2577 rtnetlink_rcv+0x1c/0x20 net/core/rtnetlink.c:6192 netlink_unicast_kernel net/netlink/af_netlink.c:1339 [inline] netlink_unicast+0x56f/0x640 net/netlink/af_netlink.c:1365 netlink_sendmsg+0x665/0x770 net/netlink/af_netlink.c:1942 sock_sendmsg_nosec net/socket.c:724 [inline] sock_sendmsg net/socket.c:747 [inline] sock_write_iter+0x1aa/0x230 net/socket.c:1138 call_write_iter include/linux/fs.h:1851 [inline] new_sync_write fs/read_write.c:491 [inline] vfs_write+0x463/0x760 fs/read_write.c:584 ksys_write+0xeb/0x1a0 fs/read_write.c:637 __do_sys_write fs/read_write.c:649 [inline] __se_sys_write fs/read_write.c:646 [inline] __x64_sys_write+0x42/0x50 fs/read_write.c:646 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88813ea4db59 of 1 bytes by task 28222 on cpu 1: netlink_recvmsg+0x3b4/0x730 net/netlink/af_netlink.c:2022 sock_recvmsg_nosec+0x4c/0x80 net/socket.c:1017 ____sys_recvmsg+0x2db/0x310 net/socket.c:2718 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00 -> 0x01 Fixes: 16b304f3404f ("netlink: Eliminate kmalloc in netlink dump operation.") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-05-09 16:56:34 +00:00
af_netlink: ensure that NLMSG_DONE never fails in dumps The way people generally use netlink_dump is that they fill in the skb as much as possible, breaking when nla_put returns an error. Then, they get called again and start filling out the next skb, and again, and so forth. The mechanism at work here is the ability for the iterative dumping function to detect when the skb is filled up and not fill it past the brim, waiting for a fresh skb for the rest of the data. However, if the attributes are small and nicely packed, it is possible that a dump callback function successfully fills in attributes until the skb is of size 4080 (libmnl's default page-sized receive buffer size). The dump function completes, satisfied, and then, if it happens to be that this is actually the last skb, and no further ones are to be sent, then netlink_dump will add on the NLMSG_DONE part: nlh = nlmsg_put_answer(skb, cb, NLMSG_DONE, sizeof(len), NLM_F_MULTI); It is very important that netlink_dump does this, of course. However, in this example, that call to nlmsg_put_answer will fail, because the previous filling by the dump function did not leave it enough room. And how could it possibly have done so? All of the nla_put variety of functions simply check to see if the skb has enough tailroom, independent of the context it is in. In order to keep the important assumptions of all netlink dump users, it is therefore important to give them an skb that has this end part of the tail already reserved, so that the call to nlmsg_put_answer does not fail. Otherwise, library authors are forced to find some bizarre sized receive buffer that has a large modulo relative to the common sizes of messages received, which is ugly and buggy. This patch thus saves the NLMSG_DONE for an additional message, for the case that things are dangerously close to the brim. This requires keeping track of the errno from ->dump() across calls. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-09 13:04:44 +09:00
netlink: fix false positive warning in extack during dumps Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. WARNING: CPU: 1 PID: 5845 at net/netlink/af_netlink.c:2210 netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-11-19 14:44:31 -08:00
netlink: annotate data-races around sk->sk_err syzbot caught another data-race in netlink when setting sk->sk_err. Annotate all of them for good measure. BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg write to 0xffff8881613bb220 of 4 bytes by task 28147 on cpu 0: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd write to 0xffff8881613bb220 of 4 bytes by task 28146 on cpu 1: netlink_recvmsg+0x448/0x780 net/netlink/af_netlink.c:1994 sock_recvmsg_nosec net/socket.c:1027 [inline] sock_recvmsg net/socket.c:1049 [inline] __sys_recvfrom+0x1f4/0x2e0 net/socket.c:2229 __do_sys_recvfrom net/socket.c:2247 [inline] __se_sys_recvfrom net/socket.c:2243 [inline] __x64_sys_recvfrom+0x78/0x90 net/socket.c:2243 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00000000 -> 0x00000016 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 28146 Comm: syz-executor.0 Not tainted 6.6.0-rc3-syzkaller-00055-g9ed22ae6be81 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/06/2023 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20231003183455.3410550-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-10-03 18:34:55 +00:00
2009-02-24 23:18:28 -08:00
2021-07-19 13:18:16 +08:00
2009-02-24 23:18:28 -08:00
2021-07-19 13:18:16 +08:00
2009-02-24 23:18:28 -08:00
netlink: annotate accesses to nlk->cb_running Both netlink_recvmsg() and netlink_native_seq_show() read nlk->cb_running locklessly. Use READ_ONCE() there. Add corresponding WRITE_ONCE() to netlink_dump() and __netlink_dump_start() syzbot reported: BUG: KCSAN: data-race in __netlink_dump_start / netlink_recvmsg write to 0xffff88813ea4db59 of 1 bytes by task 28219 on cpu 0: __netlink_dump_start+0x3af/0x4d0 net/netlink/af_netlink.c:2399 netlink_dump_start include/linux/netlink.h:308 [inline] rtnetlink_rcv_msg+0x70f/0x8c0 net/core/rtnetlink.c:6130 netlink_rcv_skb+0x126/0x220 net/netlink/af_netlink.c:2577 rtnetlink_rcv+0x1c/0x20 net/core/rtnetlink.c:6192 netlink_unicast_kernel net/netlink/af_netlink.c:1339 [inline] netlink_unicast+0x56f/0x640 net/netlink/af_netlink.c:1365 netlink_sendmsg+0x665/0x770 net/netlink/af_netlink.c:1942 sock_sendmsg_nosec net/socket.c:724 [inline] sock_sendmsg net/socket.c:747 [inline] sock_write_iter+0x1aa/0x230 net/socket.c:1138 call_write_iter include/linux/fs.h:1851 [inline] new_sync_write fs/read_write.c:491 [inline] vfs_write+0x463/0x760 fs/read_write.c:584 ksys_write+0xeb/0x1a0 fs/read_write.c:637 __do_sys_write fs/read_write.c:649 [inline] __se_sys_write fs/read_write.c:646 [inline] __x64_sys_write+0x42/0x50 fs/read_write.c:646 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd read to 0xffff88813ea4db59 of 1 bytes by task 28222 on cpu 1: netlink_recvmsg+0x3b4/0x730 net/netlink/af_netlink.c:2022 sock_recvmsg_nosec+0x4c/0x80 net/socket.c:1017 ____sys_recvmsg+0x2db/0x310 net/socket.c:2718 ___sys_recvmsg net/socket.c:2762 [inline] do_recvmmsg+0x2e5/0x710 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd value changed: 0x00 -> 0x01 Fixes: 16b304f3404f ("netlink: Eliminate kmalloc in netlink dump operation.") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-05-09 16:56:34 +00:00
netlink: remove mmapped netlink support mmapped netlink has a number of unresolved issues: - TX zerocopy support had to be disabled more than a year ago via commit 4682a0358639b29cf ("netlink: Always copy on mmap TX.") because the content of the mmapped area can change after netlink attribute validation but before message processing. - RX support was implemented mainly to speed up nfqueue dumping packet payload to userspace. However, since commit ae08ce0021087a5d812d2 ("netfilter: nfnetlink_queue: zero copy support") we avoid one copy with the socket-based interface too (via the skb_zerocopy helper). The other problem is that skbs attached to mmaped netlink socket behave different from normal skbs: - they don't have a shinfo area, so all functions that use skb_shinfo() (e.g. skb_clone) cannot be used. - reserving headroom prevents userspace from seeing the content as it expects message to start at skb->head. See for instance commit aa3a022094fa ("netlink: not trim skb for mmaped socket when dump"). - skbs handed e.g. to netlink_ack must have non-NULL skb->sk, else we crash because it needs the sk to check if a tx ring is attached. Also not obvious, leads to non-intuitive bug fixes such as 7c7bdf359 ("netfilter: nfnetlink: use original skbuff when acking batches"). mmaped netlink also didn't play nicely with the skb_zerocopy helper used by nfqueue and openvswitch. Daniel Borkmann fixed this via commit 6bb0fef489f6 ("netlink, mmap: fix edge-case leakages in nf queue zero-copy")' but at the cost of also needing to provide remaining length to the allocation function. nfqueue also has problems when used with mmaped rx netlink: - mmaped netlink doesn't allow use of nfqueue batch verdict messages. Problem is that in the mmap case, the allocation time also determines the ordering in which the frame will be seen by userspace (A allocating before B means that A is located in earlier ring slot, but this also means that B might get a lower sequence number then A since seqno is decided later. To fix this we would need to extend the spinlocked region to also cover the allocation and message setup which isn't desirable. - nfqueue can now be configured to queue large (GSO) skbs to userspace. Queing GSO packets is faster than having to force a software segmentation in the kernel, so this is a desirable option. However, with a mmap based ring one has to use 64kb per ring slot element, else mmap has to fall back to the socket path (NL_MMAP_STATUS_COPY) for all large packets. To use the mmap interface, userspace not only has to probe for mmap netlink support, it also has to implement a recv/socket receive path in order to handle messages that exceed the size of an rx ring element. Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Ken-ichirou MATSUZAWA <chamaken@gmail.com> Cc: Pablo Neira Ayuso <pablo@netfilter.org> Cc: Patrick McHardy <kaber@trash.net> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-02-18 15:03:24 +01:00
netlink: Replace rhash_portid with bound On Mon, Sep 21, 2015 at 02:20:22PM -0400, Tejun Heo wrote: > > store_release and load_acquire are different from the usual memory > barriers and can't be paired this way. You have to pair store_release > and load_acquire. Besides, it isn't a particularly good idea to OK I've decided to drop the acquire/release helpers as they don't help us at all and simply pessimises the code by using full memory barriers (on some architectures) where only a write or read barrier is needed. > depend on memory barriers embedded in other data structures like the > above. Here, especially, rhashtable_insert() would have write barrier > *before* the entry is hashed not necessarily *after*, which means that > in the above case, a socket which appears to have set bound to a > reader might not visible when the reader tries to look up the socket > on the hashtable. But you are right we do need an explicit write barrier here to ensure that the hashing is visible. > There's no reason to be overly smart here. This isn't a crazy hot > path, write barriers tend to be very cheap, store_release more so. > Please just do smp_store_release() and note what it's paired with. It's not about being overly smart. It's about actually understanding what's going on with the code. I've seen too many instances of people simply sprinkling synchronisation primitives around without any knowledge of what is happening underneath, which is just a recipe for creating hard-to-debug races. > > @@ -1539,7 +1546,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr, > > } > > } > > > > - if (!nlk->portid) { > > + if (!nlk->bound) { > > I don't think you can skip load_acquire here just because this is the > second deref of the variable. That doesn't change anything. Race > condition could still happen between the first and second tests and > skipping the second would lead to the same kind of bug. The reason this one is OK is because we do not use nlk->portid or try to get nlk from the hash table before we return to user-space. However, there is a real bug here that none of these acquire/release helpers discovered. The two bound tests here used to be a single one. Now that they are separate it is entirely possible for another thread to come in the middle and bind the socket. So we need to repeat the portid check in order to maintain consistency. > > @@ -1587,7 +1594,7 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr, > > !netlink_allowed(sock, NL_CFG_F_NONROOT_SEND)) > > return -EPERM; > > > > - if (!nlk->portid) > > + if (!nlk->bound) > > Don't we need load_acquire here too? Is this path holding a lock > which makes that unnecessary? Ditto. ---8<--- The commit 1f770c0a09da855a2b51af6d19de97fb955eca85 ("netlink: Fix autobind race condition that leads to zero port ID") created some new races that can occur due to inconcsistencies between the two port IDs. Tejun is right that a barrier is unavoidable. Therefore I am reverting to the original patch that used a boolean to indicate that a user netlink socket has been bound. Barriers have been added where necessary to ensure that a valid portid and the hashed socket is visible. I have also changed netlink_insert to only return EBUSY if the socket is bound to a portid different to the requested one. This combined with only reading nlk->bound once in netlink_bind fixes a race where two threads that bind the socket at the same time with different port IDs may both succeed. Fixes: 1f770c0a09da ("netlink: Fix autobind race condition that leads to zero port ID") Reported-by: Tejun Heo <tj@kernel.org> Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Nacked-by: Tejun Heo <tj@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-22 11:38:56 +08:00