- 14 Mar, 2019 40 commits
-
-
Bob Peterson authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit 216f0efd ] Before this patch, recovery would cause all callbacks to be delayed, put on a queue, and afterward they were all queued to the callback work queue. This patch does the same thing, but occasionally takes a break after 25 of them so it won't swamp the CPU at the expense of other RT processes like corosync. Signed-off-by:
Bob Peterson <rpeterso@redhat.com> Signed-off-by:
David Teigland <teigland@redhat.com> Signed-off-by:
Sasha Levin <sashal@kernel.org> Signed-off-by:
Juerg Haefliger <juergh@canonical.com> Signed-off-by:
Khalid Elmously <khalid.elmously@canonical.com>
-
Yufen Wang authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit 82c08c3e ] In case panic() and panic() called at the same time on different CPUS. For example: CPU 0: panic() __crash_kexec machine_crash_shutdown crash_smp_send_stop machine_kexec BUG_ON(num_online_cpus() > 1); CPU 1: panic() local_irq_disable panic_smp_self_stop If CPU 1 calls panic_smp_self_stop() before crash_smp_send_stop(), kdump fails. CPU1 can't receive the ipi irq, CPU1 will be always online. To fix this problem, this patch split out the panic_smp_self_stop() and add set_cpu_online(smp_processor_id(), false). Signed-off-by:
Yufen Wang <wangyufen@huawei.com> Signed-off-by:
Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by:
-
James Smart authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit b114d900 ] When LCB's are rejected, if beaconing was already in progress, the Reason Code Explanation was not being set. Should have been set to command in progress. Signed-off-by:
Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by:
James Smart <jsmart2021@gmail.com> Reviewed-by:
Hannes Reinecke <hare@suse.com> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by:
-
Arnd Bergmann authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit 1539c7f2 ] Randconfig testing revealed a very old bug, with gcc-8: sound/soc/intel/atom/sst/sst_loader.c: In function 'sst_load_fw': sound/soc/intel/atom/sst/sst_loader.c:357:5: error: 'fw' may be used uninitialized in this function [-Werror=maybe-uninitialized] if (fw == NULL) { ^ sound/soc/intel/atom/sst/sst_loader.c:354:25: note: 'fw' was declared here const struct firmware *fw; We must check the return code of request_firmware() before we look at the pointer result that may be uninitialized when the function fails. Fixes: 9012c954 ("ASoC: Intel: mrfld - Add DSP load and management") Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Acked-by:
Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Signed-off-by:
Mark Brown <broonie@kernel.org> Signed-off-by:
-
Slawomir Stepien authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit 0559ef7f ] Inside __ad7280_read32(), the spi_sync_transfer() can fail with negative error code. This change will ensure that this error is being passed up in the call stack, so it can be handled. Signed-off-by:
Slawomir Stepien <sst@poczta.fm> Signed-off-by:
Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by:
-
Gustavo A. R. Silva authored
BugLink: https://bugs.launchpad.net/bugs/1818813 [ Upstream commit a3780509 ] idx can be indirectly controlled by user-space, hence leading to a potential exploitation of the Spectre variant 1 vulnerability. This issue was detected with the help of Smatch: drivers/gpu/drm/drm_bufs.c:1420 drm_legacy_freebufs() warn: potential spectre issue 'dma->buflist' [r] (local cap) Fix this by sanitizing idx before using it to index dma->buflist Notice that given that speculation windows are large, the policy is to kill the speculation on the first load and not worry if it can be completed with a dependent load/store [1]. [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2Signed-off-by:
Gustavo A. R. Silva <gustavo@embeddedor.com> Signed-off-by:
Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20181016095549.GA23586@embeddedor.comSigned-off-by:
-
Greg Kroah-Hartman authored
BugLink: https://bugs.launchpad.net/bugs/1818806Signed-off-by:
-
Paul E. McKenney authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit ec3833ed upstream. The cpu_online() function can return values other than 0 and 1, which can result in subscript overflow when applied to a two-element array. This commit allows for this behavior by using "!!" on the return value from cpu_online() when used as a subscript. Signed-off-by:
Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: "Rantala, Tommi" <tommi.t.rantala@nokia.com> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by:
-
Michal Kubecek authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit ade44640 upstream. Since commit 7969e5c4 ("ip: discard IPv4 datagrams with overlapping segments.") IPv4 reassembly code drops the whole queue whenever an overlapping fragment is received. However, the test is written in a way which detects duplicate fragments as overlapping so that in environments with many duplicate packets, fragmented packets may be undeliverable. Add an extra test and for (potentially) duplicate fragment, only drop the new fragment rather than the whole queue. Only starting offset and length are checked, not the contents of the fragments as that would be too expensive. For similar reason, linear list ("run") of a rbtree node is not iterated, we only check if the new fragment is a subset of the interval covered by existing consecutive fragments. v2: instead of an exact check iterating through linear list of an rbtree node, only check if the new fragment is subset of the "run" (suggested by Eric Dumazet) Fixes: 7969e5c4 ("ip: discard IPv4 datagrams with overlapping segments.") Signed-off-by:
Michal Kubecek <mkubecek@suse.cz> Signed-off-by:
David S. Miller <davem@davemloft.net> Signed-off-by:
Mao Wenan <maowenan@huawei.com> [bwh: Backported to 4.4: - goto discard_qp, not err, in case of overlap - Set err earlier variable, as done upstream in commit 0ff89efb "ip: fail fast on IP defrag errors"] Signed-off-by:
Ben Hutchings <ben.hutchings@codethink.co.uk> Signed-off-by:
-
Dimitris Michailidis authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit d55bef50 upstream. We've been getting checksum errors involving small UDP packets, usually 59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault() has been complaining that HW is providing bad checksums. Turns out the problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98 ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends"). The source of the problem is that when the bytes we are trimming start at an odd address, as in the case of the 1 padding byte above, skb_checksum() returns a byte-swapped value. We cannot just combine this with skb->csum using csum_sub(). We need to use csum_block_sub() here that takes into account the parity of the start address and handles the swapping. Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer(). Fixes: 88078d98 ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends") Signed-off-by:
Dimitris Michailidis <dmichail@google.com> Reviewed-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 0d5b9311 upstream. Multiple cpus might attempt to insert a new fragment in rhashtable, if for example RPS is buggy, as reported by 배석진 in https://patchwork.ozlabs.org/patch/994601/ We use rhashtable_lookup_get_insert_key() instead of rhashtable_insert_fast() to let cpus losing the race free their own inet_frag_queue and use the one that was inserted by another cpu. Fixes: 648700f7 ("inet: frags: use rhashtables for reassembly units") Signed-off-by:
배석진 <soukjin.bae@samsung.com> Signed-off-by:
-
Dan Carpenter authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 70837ffe upstream. We accidentally removed the parentheses here, but they are required because '!' has higher precedence than '&'. Fixes: fa0f5273 ("ip: use rb trees for IP frag queue.") Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Taehee Yoo authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 5d407b07 upstream. A kernel crash occurrs when defragmented packet is fragmented in ip_do_fragment(). In defragment routine, skb_orphan() is called and skb->ip_defrag_offset is set. but skb->sk and skb->ip_defrag_offset are same union member. so that frag->sk is not NULL. Hence crash occurrs in skb->sk check routine in ip_do_fragment() when defragmented packet is fragmented. test commands: %iptables -t nat -I POSTROUTING -j MASQUERADE %hping3 192.168.4.2 -s 1000 -p 2000 -d 60000 splat looks like: [ 261.069429] kernel BUG at net/ipv4/ip_output.c:636! [ 261.075753] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI [ 261.083854] CPU: 1 PID: 1349 Comm: hping3 Not tainted 4.19.0-rc2+ #3 [ 261.100977] RIP: 0010:ip_do_fragment+0x1613/0x2600 [ 261.106945] Code: e8 e2 38 e3 fe 4c 8b 44 24 18 48 8b 74 24 08 e9 92 f6 ff ff 80 3c 02 00 0f 85 da 07 00 00 48 8b b5 d0 00 00 00 e9 25 f6 ff ff <0f> 0b 0f 0b 44 8b 54 24 58 4c 8b 4c 24 18 4c 8b 5c 24 60 4c 8b 6c [ 261.127015] RSP: 0018:ffff8801031cf2c0 EFLAGS: 00010202 [ 261.134156] RAX: 1ffff1002297537b RBX: ffffed0020639e6e RCX: 0000000000000004 [ 261.142156] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff880114ba9bd8 [ 261.150157] RBP: ffff880114ba8a40 R08: ffffed0022975395 R09: ffffed0022975395 [ 261.158157] R10: 0000000000000001 R11: ffffed0022975394 R12: ffff880114ba9ca4 [ 261.166159] R13: 0000000000000010 R14: ffff880114ba9bc0 R15: dffffc0000000000 [ 261.174169] FS: 00007fbae2199700(0000) GS:ffff88011b400000(0000) knlGS:0000000000000000 [ 261.183012] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 261.189013] CR2: 00005579244fe000 CR3: 0000000119bf4000 CR4: 00000000001006e0 [ 261.198158] Call Trace: [ 261.199018] ? dst_output+0x180/0x180 [ 261.205011] ? save_trace+0x300/0x300 [ 261.209018] ? ip_copy_metadata+0xb00/0xb00 [ 261.213034] ? sched_clock_local+0xd4/0x140 [ 261.218158] ? kill_l4proto+0x120/0x120 [nf_conntrack] [ 261.223014] ? rt_cpu_seq_stop+0x10/0x10 [ 261.227014] ? find_held_lock+0x39/0x1c0 [ 261.233008] ip_finish_output+0x51d/0xb50 [ 261.237006] ? ip_fragment.constprop.56+0x220/0x220 [ 261.243011] ? nf_ct_l4proto_register_one+0x5b0/0x5b0 [nf_conntrack] [ 261.250152] ? rcu_is_watching+0x77/0x120 [ 261.255010] ? nf_nat_ipv4_out+0x1e/0x2b0 [nf_nat_ipv4] [ 261.261033] ? nf_hook_slow+0xb1/0x160 [ 261.265007] ip_output+0x1c7/0x710 [ 261.269005] ? ip_mc_output+0x13f0/0x13f0 [ 261.273002] ? __local_bh_enable_ip+0xe9/0x1b0 [ 261.278152] ? ip_fragment.constprop.56+0x220/0x220 [ 261.282996] ? nf_hook_slow+0xb1/0x160 [ 261.287007] raw_sendmsg+0x21f9/0x4420 [ 261.291008] ? dst_output+0x180/0x180 [ 261.297003] ? sched_clock_cpu+0x126/0x170 [ 261.301003] ? find_held_lock+0x39/0x1c0 [ 261.306155] ? stop_critical_timings+0x420/0x420 [ 261.311004] ? check_flags.part.36+0x450/0x450 [ 261.315005] ? _raw_spin_unlock_irq+0x29/0x40 [ 261.320995] ? _raw_spin_unlock_irq+0x29/0x40 [ 261.326142] ? cyc2ns_read_end+0x10/0x10 [ 261.330139] ? raw_bind+0x280/0x280 [ 261.334138] ? sched_clock_cpu+0x126/0x170 [ 261.338995] ? check_flags.part.36+0x450/0x450 [ 261.342991] ? __lock_acquire+0x4500/0x4500 [ 261.348994] ? inet_sendmsg+0x11c/0x500 [ 261.352989] ? dst_output+0x180/0x180 [ 261.357012] inet_sendmsg+0x11c/0x500 [ ... ] v2: - clear skb->sk at reassembly routine.(Eric Dumarzet) Fixes: fa0f5273 ("ip: use rb trees for IP frag queue.") Suggested-by:
Taehee Yoo <ap420073@gmail.com> Reviewed-by:
-
Peter Oskolkov authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit a4fd284a upstream. This patch changes the runtime behavior of IP defrag queue: incoming in-order fragments are added to the end of the current list/"run" of in-order fragments at the tail. On some workloads, UDP stream performance is substantially improved: RX: ./udp_stream -F 10 -T 2 -l 60 TX: ./udp_stream -c -H <host> -F 10 -T 5 -l 60 with this patchset applied on a 10Gbps receiver: throughput=9524.18 throughput_units=Mbit/s upstream (net-next): throughput=4608.93 throughput_units=Mbit/s Reported-by:
Willem de Bruijn <willemb@google.com> Signed-off-by:
Peter Oskolkov <posk@google.com> Cc: Eric Dumazet <edumazet@google.com> Cc: Florian Westphal <fw@strlen.de> Signed-off-by:
-
Peter Oskolkov authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 353c9cb3 upstream. This patch introduces several helper functions/macros that will be used in the follow-up patch. No runtime changes yet. The new logic (fully implemented in the second patch) is as follows: * Nodes in the rb-tree will now contain not single fragments, but lists of consecutive fragments ("runs"). * At each point in time, the current "active" run at the tail is maintained/tracked. Fragments that arrive in-order, adjacent to the previous tail fragment, are added to this tail run without triggering the re-balancing of the rb-tree. * If a fragment arrives out of order with the offset _before_ the tail run, it is inserted into the rb-tree as a single fragment. * If a fragment arrives after the current tail fragment (with a gap), it starts a new "tail" run, as is inserted into the rb-tree at the end as the head of the new run. skb->cb is used to store additional information needed here (suggested by Eric Dumazet). Reported-by:
-
Peter Oskolkov authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit fa0f5273 upstream. Similar to TCP OOO RX queue, it makes sense to use rb trees to store IP fragments, so that OOO fragments are inserted faster. Tested: - a follow-up patch contains a rather comprehensive ip defrag self-test (functional) - ran neper `udp_stream -c -H <host> -F 100 -l 300 -T 20`: netstat --statistics Ip: 282078937 total packets received 0 forwarded 0 incoming packets discarded 946760 incoming packets delivered 18743456 requests sent out 101 fragments dropped after timeout 282077129 reassemblies required 944952 packets reassembled ok 262734239 packet reassembles failed (The numbers/stats above are somewhat better re: reassemblies vs a kernel without this patchset. More comprehensive performance testing TBD). Reported-by:
Jann Horn <jannh@google.com> Reported-by:
Juha-Matti Tilli <juha-matti.tilli@iki.fi> Suggested-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 88078d98 upstream. After working on IP defragmentation lately, I found that some large packets defeat CHECKSUM_COMPLETE optimization because of NIC adding zero paddings on the last (small) fragment. While removing the padding with pskb_trim_rcsum(), we set skb->ip_summed to CHECKSUM_NONE, forcing a full csum validation, even if all prior fragments had CHECKSUM_COMPLETE set. We can instead compute the checksum of the part we are trimming, usually smaller than the part we keep. Signed-off-by:
-
Florian Westphal authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 0ed4229b upstream. don't bother with pathological cases, they only waste cycles. IPv6 requires a minimum MTU of 1280 so we should never see fragments smaller than this (except last frag). v3: don't use awkward "-offset + len" v2: drop IPv4 part, which added same check w. IPV4_MIN_MTU (68). There were concerns that there could be even smaller frags generated by intermediate nodes, e.g. on radio networks. Cc: Peter Oskolkov <posk@google.com> Cc: Eric Dumazet <edumazet@google.com> Signed-off-by:
Florian Westphal <fw@strlen.de> Signed-off-by:
-
Peter Oskolkov authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 385114de upstream. Tested: see the next patch is the series. Suggested-by:
-
Peter Oskolkov authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 7969e5c4 upstream. This behavior is required in IPv6, and there is little need to tolerate overlapping fragments in IPv4. This change simplifies the code and eliminates potential DDoS attack vectors. Tested: ran ip_defrag selftest (not yet available uptream). Suggested-by:
Stephen Hemminger <stephen@networkplumber.org> Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 3d234012 upstream. Giving an integer to proc_doulongvec_minmax() is dangerous on 64bit arches, since linker might place next to it a non zero value preventing a change to ip6frag_low_thresh. ip6frag_low_thresh is not used anymore in the kernel, but we do not want to prematuraly break user scripts wanting to change it. Since specifying a minimal value of 0 for proc_doulongvec_minmax() is moot, let's remove these zero values in all defrag units. Fixes: 6e00f7dd ("ipv6: frags: fix /proc/sys/net/ipv6/ip6frag_low_thresh") Signed-off-by:
Maciej Żenczykowski <maze@google.com> Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit bf663371 upstream. ip_defrag uses skb->cb[] to store the fragment offset, and unfortunately this integer is currently in a different cache line than skb->next, meaning that we use two cache lines per skb when finding the insertion point. By aliasing skb->ip_defrag_offset and skb->dev, we pack all the fields in a single cache line and save precious memory bandwidth. Note that after the fast path added by Changli Gao in commit d6bebca9 ("fragment: add fast path for in-order fragments") this change wont help the fast path, since we still need to access prev->len (2nd cache line), but will show great benefits when slow path is entered, since we perform a linear scan of a potentially long list. Also, note that this potential long list is an attack vector, we might consider also using an rb-tree there eventually. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit c2615cf5 upstream. Put the read-mostly fields in a separate cache line at the beginning of struct netns_frags, to reduce false sharing noticed in inet_frag_kill() Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit e5d672a0 upstream. While under frags DDOS I noticed unfortunate false sharing between @nelems and @params.automatic_shrinking Move @nelems at the end of struct rhashtable so that first cache line is shared between all cpus, because almost never dirtied. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 05c0b86b upstream. Make it similar to IPv4 ip_expire(), and release the lock before calling icmp functions. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 1eec5d56 upstream. An skb_clone() was added in commit ec4fbd64 ("inet: frag: release spinlock before calling icmp_send()") While fixing the bug at that time, it also added a very high cost for DDOS frags, as the ICMP rate limit is applied after this expensive operation (skb_clone() + consume_skb(), implying memory allocations, copy, and freeing) We can use skb_get(head) here, all we want is to make sure skb wont be freed by another cpu. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 3e67f106 upstream. Some users are willing to provision huge amounts of memory to be able to perform reassembly reasonnably well under pressure. Current memory tracking is using one atomic_t and integers. Switch to atomic_long_t so that 64bit arches can use more than 2GB, without any cost for 32bit arches. Note that this patch avoids an overflow error, if high_thresh was set to ~2GB, since this test in inet_frag_alloc() was never true : if (... || frag_mem_limit(nf) > nf->high_thresh) Tested: $ echo 16000000000 >/proc/sys/net/ipv4/ipfrag_high_thresh <frag DDOS> $ grep FRAG /proc/net/sockstat FRAG: inuse 14705885 memory 16000002880 $ nstat -n ; sleep 1 ; nstat | grep Reas IpReasmReqds 3317150 0.0 IpReasmFails 3317112 0.0 Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 2d44ed22 upstream. This function is obsolete, after rhashtable addition to inet defrag. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 399d1404 upstream. This refactors ip_expire() since one indentation level is removed. Note: in the future, we should try hard to avoid the skb_clone() since this is a serious performance cost. Under DDOS, the ICMP message wont be sent because of rate limits. Fact that ip6_expire_frag_queue() does not use skb_clone() is disturbing too. Presumably IPv6 should have the same issue than the one we fixed in commit ec4fbd64 ("inet: frag: release spinlock before calling icmp_send()") Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 6befe4a7 upstream. Remove sum_frag_mem_limit(), ip_frag_mem() & ip6_frag_mem() Also since we use rhashtable we can bring back the number of fragments in "grep FRAG /proc/net/sockstat /proc/net/sockstat6" that was removed in commit 434d3054 ("inet: frag: don't account number of fragment queues") Signed-off-by:
-
Paolo Abeni authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit f6f2a4a2 upstream. Setting the low threshold to 0 has no effect on frags allocation, we need to clear high_thresh instead. The code was pre-existent to commit 648700f7 ("inet: frags: use rhashtables for reassembly units"), but before the above, such assignment had a different role: prevent concurrent eviction from the worker and the netns cleanup helper. Fixes: 648700f7 ("inet: frags: use rhashtables for reassembly units") Signed-off-by:
Paolo Abeni <pabeni@redhat.com> Signed-off-by:
-
Alexander Aring authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit f18fa5de upstream. This patch initialize stack variables which are used in frag_lowpan_compare_key to zero. In my case there are padding bytes in the structures ieee802154_addr as well in frag_lowpan_compare_key. Otherwise the key variable contains random bytes. The result is that a compare of two keys by memcmp works incorrect. Fixes: 648700f7 ("inet: frags: use rhashtables for reassembly units") Signed-off-by:
Alexander Aring <aring@mojatatu.com> Reported-by:
Stefan Schmidt <stefan@osg.samsung.com> Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 648700f7 upstream. Some applications still rely on IP fragmentation, and to be fair linux reassembly unit is not working under any serious load. It uses static hash tables of 1024 buckets, and up to 128 items per bucket (!!!) A work queue is supposed to garbage collect items when host is under memory pressure, and doing a hash rebuild, changing seed used in hash computations. This work queue blocks softirqs for up to 25 ms when doing a hash rebuild, occurring every 5 seconds if host is under fire. Then there is the problem of sharing this hash table for all netns. It is time to switch to rhashtables, and allocate one of them per netns to speedup netns dismantle, since this is a critical metric these days. Lookup is now using RCU. A followup patch will even remove the refcount hold/release left from prior implementation and save a couple of atomic operations. Before this patch, 16 cpus (16 RX queue NIC) could not handle more than 1 Mpps frags DDOS. After the patch, I reach 9 Mpps without any tuning, and can use up to 2GB of storage for the fragments (exact number depends on frags being evicted after timeout) $ grep FRAG /proc/net/sockstat FRAG: inuse 1966916 memory 2140004608 A followup patch will change the limits for 64bit arches. Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit ae6da1f5 upstream. Rehashing and destroying large hash table takes a lot of time, and happens in process context. It is safe to add cond_resched() in rhashtable_rehash_table() and rhashtable_free_and_destroy() Signed-off-by:
Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Ben Hutchings authored
BugLink: https://bugs.launchpad.net/bugs/1818806 Extracted from commit ca26893f "rhashtable: Add rhlist interface". Cc: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Pablo Neira Ayuso authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 5ca8cc5b upstream. This patch modifies __rhashtable_insert_fast() so it returns the existing object that clashes with the one that you want to insert. In case the object is successfully inserted, NULL is returned. Otherwise, you get an error via ERR_PTR(). This patch adapts the existing callers of __rhashtable_insert_fast() so they handle this new logic, and it adds a new rhashtable_lookup_get_insert_key() interface to fetch this existing object. nf_tables needs this change to improve handling of EEXIST cases via honoring the NLM_F_EXCL flag and by checking if the data part of the mapping matches what we have. Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Thomas Graf <tgraf@suug.ch> Signed-off-by:
Pablo Neira Ayuso <pablo@netfilter.org> Acked-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 807f1844 upstream. We want to call lowpan_net_frag_init() earlier. Similar to commit "inet: frags: refactor ipv6_frag_init()" This is a prereq to "inet: frags: use rhashtables for reassembly units" Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 5b975bab upstream. We want to call inet_frags_init() earlier. This is a prereq to "inet: frags: use rhashtables for reassembly units" Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 483a6e4f upstream. We need to call inet_frags_init() before register_pernet_subsys(), as a prereq for following patch ("inet: frags: use rhashtables for reassembly units") Signed-off-by:
-
Eric Dumazet authored
BugLink: https://bugs.launchpad.net/bugs/1818806 commit 093ba729 upstream. In order to simplify the API, add a pointer to struct inet_frags. This will allow us to make things less complex. These functions no longer have a struct inet_frags parameter : inet_frag_destroy(struct inet_frag_queue *q /*, struct inet_frags *f */) inet_frag_put(struct inet_frag_queue *q /*, struct inet_frags *f */) inet_frag_kill(struct inet_frag_queue *q /*, struct inet_frags *f */) inet_frags_exit_net(struct netns_frags *nf /*, struct inet_frags *f */) ip6_expire_frag_queue(struct net *net, struct frag_queue *fq) Signed-off-by:
-
