From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Eric Dumazet <edumazet@google.com>,
Jason Xing <kerneljasonxing@gmail.com>,
Kuniyuki Iwashima <kuniyu@amazon.com>,
Jakub Kicinski <kuba@kernel.org>, Sasha Levin <sashal@kernel.org>,
ncardwell@google.com, davem@davemloft.net, dsahern@kernel.org,
pabeni@redhat.com, netdev@vger.kernel.org
Subject: [PATCH AUTOSEL 6.12 180/486] tcp: bring back NUMA dispersion in inet_ehash_locks_alloc()
Date: Mon, 5 May 2025 18:34:16 -0400 [thread overview]
Message-ID: <20250505223922.2682012-180-sashal@kernel.org> (raw)
In-Reply-To: <20250505223922.2682012-1-sashal@kernel.org>
From: Eric Dumazet <edumazet@google.com>
[ Upstream commit f8ece40786c9342249aa0a1b55e148ee23b2a746 ]
We have platforms with 6 NUMA nodes and 480 cpus.
inet_ehash_locks_alloc() currently allocates a single 64KB page
to hold all ehash spinlocks. This adds more pressure on a single node.
Change inet_ehash_locks_alloc() to use vmalloc() to spread
the spinlocks on all online nodes, driven by NUMA policies.
At boot time, NUMA policy is interleave=all, meaning that
tcp_hashinfo.ehash_locks gets hash dispersion on all nodes.
Tested:
lack5:~# grep inet_ehash_locks_alloc /proc/vmallocinfo
0x00000000d9aec4d1-0x00000000a828b652 69632 inet_ehash_locks_alloc+0x90/0x100 pages=16 vmalloc N0=2 N1=3 N2=3 N3=3 N4=3 N5=2
lack5:~# echo 8192 >/proc/sys/net/ipv4/tcp_child_ehash_entries
lack5:~# numactl --interleave=all unshare -n bash -c "grep inet_ehash_locks_alloc /proc/vmallocinfo"
0x000000004e99d30c-0x00000000763f3279 36864 inet_ehash_locks_alloc+0x90/0x100 pages=8 vmalloc N0=1 N1=2 N2=2 N3=1 N4=1 N5=1
0x00000000d9aec4d1-0x00000000a828b652 69632 inet_ehash_locks_alloc+0x90/0x100 pages=16 vmalloc N0=2 N1=3 N2=3 N3=3 N4=3 N5=2
lack5:~# numactl --interleave=0,5 unshare -n bash -c "grep inet_ehash_locks_alloc /proc/vmallocinfo"
0x00000000fd73a33e-0x0000000004b9a177 36864 inet_ehash_locks_alloc+0x90/0x100 pages=8 vmalloc N0=4 N5=4
0x00000000d9aec4d1-0x00000000a828b652 69632 inet_ehash_locks_alloc+0x90/0x100 pages=16 vmalloc N0=2 N1=3 N2=3 N3=3 N4=3 N5=2
lack5:~# echo 1024 >/proc/sys/net/ipv4/tcp_child_ehash_entries
lack5:~# numactl --interleave=all unshare -n bash -c "grep inet_ehash_locks_alloc /proc/vmallocinfo"
0x00000000db07d7a2-0x00000000ad697d29 8192 inet_ehash_locks_alloc+0x90/0x100 pages=1 vmalloc N2=1
0x00000000d9aec4d1-0x00000000a828b652 69632 inet_ehash_locks_alloc+0x90/0x100 pages=16 vmalloc N0=2 N1=3 N2=3 N3=3 N4=3 N5=2
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Jason Xing <kerneljasonxing@gmail.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250305130550.1865988-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
net/ipv4/inet_hashtables.c | 37 ++++++++++++++++++++++++++-----------
1 file changed, 26 insertions(+), 11 deletions(-)
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 9bfcfd016e182..2b4a588247639 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -1230,22 +1230,37 @@ int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo)
{
unsigned int locksz = sizeof(spinlock_t);
unsigned int i, nblocks = 1;
+ spinlock_t *ptr = NULL;
- if (locksz != 0) {
- /* allocate 2 cache lines or at least one spinlock per cpu */
- nblocks = max(2U * L1_CACHE_BYTES / locksz, 1U);
- nblocks = roundup_pow_of_two(nblocks * num_possible_cpus());
+ if (locksz == 0)
+ goto set_mask;
- /* no more locks than number of hash buckets */
- nblocks = min(nblocks, hashinfo->ehash_mask + 1);
+ /* Allocate 2 cache lines or at least one spinlock per cpu. */
+ nblocks = max(2U * L1_CACHE_BYTES / locksz, 1U) * num_possible_cpus();
- hashinfo->ehash_locks = kvmalloc_array(nblocks, locksz, GFP_KERNEL);
- if (!hashinfo->ehash_locks)
- return -ENOMEM;
+ /* At least one page per NUMA node. */
+ nblocks = max(nblocks, num_online_nodes() * PAGE_SIZE / locksz);
+
+ nblocks = roundup_pow_of_two(nblocks);
+
+ /* No more locks than number of hash buckets. */
+ nblocks = min(nblocks, hashinfo->ehash_mask + 1);
- for (i = 0; i < nblocks; i++)
- spin_lock_init(&hashinfo->ehash_locks[i]);
+ if (num_online_nodes() > 1) {
+ /* Use vmalloc() to allow NUMA policy to spread pages
+ * on all available nodes if desired.
+ */
+ ptr = vmalloc_array(nblocks, locksz);
+ }
+ if (!ptr) {
+ ptr = kvmalloc_array(nblocks, locksz, GFP_KERNEL);
+ if (!ptr)
+ return -ENOMEM;
}
+ for (i = 0; i < nblocks; i++)
+ spin_lock_init(&ptr[i]);
+ hashinfo->ehash_locks = ptr;
+set_mask:
hashinfo->ehash_locks_mask = nblocks - 1;
return 0;
}
--
2.39.5
next prev parent reply other threads:[~2025-05-05 22:45 UTC|newest]
Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20250505223922.2682012-1-sashal@kernel.org>
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 013/486] SUNRPC: Don't allow waiting for exiting tasks Sasha Levin
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 029/486] SUNRPC: rpc_clnt_set_transport() must not change the autobind setting Sasha Levin
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 030/486] SUNRPC: rpcbind should never reset the port to the value '0' Sasha Levin
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 034/486] mctp: Fix incorrect tx flow invalidation condition in mctp-i2c Sasha Levin
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 035/486] net: tn40xx: add pci-id of the aqr105-based Tehuti TN4010 cards Sasha Levin
2025-05-05 22:31 ` [PATCH AUTOSEL 6.12 036/486] net: tn40xx: create swnode for mdio and aqr105 phy and add to mdiobus Sasha Levin
2025-05-05 22:32 ` [PATCH AUTOSEL 6.12 046/486] r8169: disable RTL8126 ZRX-DC timeout Sasha Levin
2025-05-05 22:32 ` [PATCH AUTOSEL 6.12 092/486] bnxt_en: Query FW parameters when the CAPS_CHANGE bit is set Sasha Levin
2025-05-05 22:32 ` [PATCH AUTOSEL 6.12 103/486] tcp: reorganize tcp_in_ack_event() and tcp_count_delivered() Sasha Levin
2025-05-05 22:33 ` [PATCH AUTOSEL 6.12 116/486] net/smc: use the correct ndev to find pnetid by pnetid table Sasha Levin
2025-05-05 22:33 ` [PATCH AUTOSEL 6.12 131/486] net: stmmac: dwmac-rk: Validate GRF and peripheral GRF during probe Sasha Levin
2025-05-05 22:33 ` [PATCH AUTOSEL 6.12 132/486] net: hsr: Fix PRP duplicate detection Sasha Levin
2025-05-05 22:33 ` [PATCH AUTOSEL 6.12 135/486] netfilter: conntrack: Bound nf_conntrack sysctl writes Sasha Levin
2025-05-05 22:33 ` [PATCH AUTOSEL 6.12 155/486] ipv6: save dontfrag in cork Sasha Levin
2025-05-05 22:34 ` Sasha Levin [this message]
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 182/486] ieee802154: ca8210: Use proper setters and getters for bitwise types Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 193/486] net: phylink: use pl->link_interface in phylink_expects_phy() Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 206/486] net: ethernet: ti: cpsw_new: populate netdev of_node Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 207/486] net: phy: nxp-c45-tja11xx: add match_phy_device to TJA1103/TJA1104 Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 208/486] dpll: Add an assertion to check freq_supported_num Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 212/486] net: pktgen: fix mpls maximum labels list parsing Sasha Levin
2025-05-05 22:34 ` [PATCH AUTOSEL 6.12 216/486] ipv4: fib: Move fib_valid_key_len() to rtm_to_fib_config() Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 238/486] net/mlx5: Avoid report two health errors on same syndrome Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 239/486] selftests/net: have `gro.sh -t` return a correct exit code Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 244/486] net: ethernet: mtk_ppe_offload: Allow QinQ, double ETH_P_8021Q only Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 245/486] net: xgene-v2: remove incorrect ACPI_PTR annotation Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 246/486] bonding: report duplicate MAC address in all situations Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 250/486] Octeontx2-af: RPM: Register driver with PCI subsys IDs Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 258/486] vhost-scsi: Return queue full for page alloc failures during copy Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 263/486] net/mlx5e: Add correct match to check IPSec syndromes for switchdev mode Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 270/486] net/mlx5: Change POOL_NEXT_SIZE define value and make it global Sasha Levin
2025-05-05 22:35 ` [PATCH AUTOSEL 6.12 274/486] net: ipv6: Init tunnel link-netns before registering dev Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 291/486] net: pktgen: fix access outside of user given buffer in pktgen_thread_write() Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 294/486] bpf: Prevent unsafe access to the sock fields in the BPF timestamping callback Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 312/486] eth: mlx4: don't try to complete XDP frames in netpoll Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 315/486] vxlan: Join / leave MC group after remote changes Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 321/486] net/mlx5: Modify LSB bitmask in temperature event to include only the first bit Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 322/486] net/mlx5: Apply rate-limiting to high temperature warning Sasha Levin
2025-05-05 22:36 ` [PATCH AUTOSEL 6.12 342/486] net/mlx4_core: Avoid impossible mlx4_db_alloc() order value Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 359/486] net: stmmac: dwmac-loongson: Set correct {tx,rx}_fifo_size Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 379/486] net/mlx5: XDP, Enable TX side XDP multi-buffer support Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 380/486] net/mlx5: Extend Ethtool loopback selftest to support non-linear SKB Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 381/486] net/mlx5e: set the tx_queue_len for pfifo_fast Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 382/486] net/mlx5e: reduce rep rxq depth to 256 for ECPF Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 383/486] net/mlx5e: reduce the max log mpwrq sz for ECPF and reps Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 385/486] xfrm: prevent high SEQ input in non-ESN mode Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 387/486] mptcp: pm: userspace: flags: clearer msg if no remote addr Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 393/486] net: fec: Refactor MAC reset to function Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 397/486] ip: fib_rules: Fetch net from fib_rule in fib[46]_rule_configure() Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 398/486] r8152: add vendor/device ID pair for Dell Alienware AW1022z Sasha Levin
2025-05-05 22:37 ` [PATCH AUTOSEL 6.12 402/486] net: ethtool: prevent flow steering to RSS contexts which don't exist Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 412/486] net: page_pool: avoid false positive warning if NAPI was never added Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 420/486] eth: fbnic: set IFF_UNICAST_FLT to avoid enabling promiscuous mode when adding unicast addrs Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 421/486] tools: ynl-gen: don't output external constants Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 422/486] net/mlx5e: Avoid WARN_ON when configuring MQPRIO with HTB offload enabled Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 424/486] vxlan: Annotate FDB data races Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 425/486] ipv4: ip_gre: Fix set but not used warning in ipgre_err() if IPv4-only Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 426/486] r8169: don't scan PHY addresses > 0 Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 427/486] net: flush_backlog() small changes Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 428/486] bridge: mdb: Allow replace of a host-joined group Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 429/486] net-sysfs: remove rtnl_trylock from queue attributes Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 430/486] net-sysfs: prevent uncleared queues from being re-added Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 431/486] net-sysfs: remove rtnl_trylock from device attributes Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 432/486] ice: init flow director before RDMA Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 433/486] ice: treat dyn_allowed only as suggestion Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 438/486] ice: count combined queues using Rx/Tx count Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 440/486] net/mana: fix warning in the writer of client oob Sasha Levin
2025-05-05 22:38 ` [PATCH AUTOSEL 6.12 456/486] bpf: Use kallsyms to find the function name of a struct_ops's stub function Sasha Levin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250505223922.2682012-180-sashal@kernel.org \
--to=sashal@kernel.org \
--cc=davem@davemloft.net \
--cc=dsahern@kernel.org \
--cc=edumazet@google.com \
--cc=kerneljasonxing@gmail.com \
--cc=kuba@kernel.org \
--cc=kuniyu@amazon.com \
--cc=linux-kernel@vger.kernel.org \
--cc=ncardwell@google.com \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).