| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Fri, 3 Apr 2020 16:59:29 -0400
From: Alex Kogan <alex.kogan@...cle.com>
To: linux@...linux.org.uk, peterz@...radead.org, mingo@...hat.com,
will.deacon@....com, arnd@...db.de, longman@...hat.com,
linux-arch@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
linux-kernel@...r.kernel.org, tglx@...utronix.de, bp@...en8.de,
hpa@...or.com, x86@...nel.org, guohanjun@...wei.com,
jglauber@...vell.com
Cc: steven.sistare@...cle.com, daniel.m.jordan@...cle.com,
alex.kogan@...cle.com, dave.dice@...cle.com
Subject: [PATCH v10 4/5] locking/qspinlock: Introduce starvation avoidance into CNA
Keep track of the number of intra-node lock handoffs, and force
inter-node handoff once this number reaches a preset threshold.
The default value for the threshold can be overridden with
the new kernel boot command-line option "numa_spinlock_threshold".
Signed-off-by: Alex Kogan <alex.kogan@...cle.com>
Reviewed-by: Steve Sistare <steven.sistare@...cle.com>
Reviewed-by: Waiman Long <longman@...hat.com>
---
.../admin-guide/kernel-parameters.txt | 8 +++
kernel/locking/qspinlock.c | 3 +
kernel/locking/qspinlock_cna.h | 55 ++++++++++++++++---
3 files changed, 59 insertions(+), 7 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index cf3ede858e01..c23bbf49024b 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3249,6 +3249,14 @@
Not specifying this option is equivalent to
numa_spinlock=auto.
+ numa_spinlock_threshold= [NUMA, PV_OPS]
+ Set the threshold for the number of intra-node
+ lock hand-offs before the NUMA-aware spinlock
+ is forced to be passed to a thread on another NUMA node.
+ Valid values are in the [0..31] range. Smaller values
+ result in a more fair, but less performant spinlock, and
+ vice versa. The default value is 16.
+
cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
Some features depend on CPU0. Known dependencies are:
diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 5b01ab0cc944..29e480235cce 100644
--- a/kernel/locking/qspinlock.c
+++ b/kernel/locking/qspinlock.c
@@ -598,6 +598,9 @@ EXPORT_SYMBOL(queued_spin_lock_slowpath);
#if !defined(_GEN_CNA_LOCK_SLOWPATH) && defined(CONFIG_NUMA_AWARE_SPINLOCKS)
#define _GEN_CNA_LOCK_SLOWPATH
+#undef pv_init_node
+#define pv_init_node cna_init_node
+
#undef pv_wait_head_or_lock
#define pv_wait_head_or_lock cna_wait_head_or_lock
diff --git a/kernel/locking/qspinlock_cna.h b/kernel/locking/qspinlock_cna.h
index 619883f3dfd3..e3180f6f5cdc 100644
--- a/kernel/locking/qspinlock_cna.h
+++ b/kernel/locking/qspinlock_cna.h
@@ -38,7 +38,9 @@
* when unlocking the MCS lock (post-scan), starting at the node where pre-scan
* stopped. If both scans fail to find such T, the MCS lock is passed to the
* first thread in the secondary queue. If the secondary queue is empty, the
- * lock is passed to the next thread in the primary queue.
+ * lock is passed to the next thread in the primary queue. To avoid starvation
+ * of threads in the secondary queue, those threads are moved back to the head
+ * of the primary queue after a certain number of intra-node lock hand-offs.
*
* For more details, see https://arxiv.org/abs/1810.05600.
*
@@ -51,13 +53,23 @@ struct cna_node {
int numa_node;
u32 encoded_tail; /* self */
u32 partial_order; /* encoded tail or enum val */
+ u32 intra_count;
};
enum {
LOCAL_WAITER_FOUND = 2, /* 0 and 1 are reserved for @locked */
+ FLUSH_SECONDARY_QUEUE = 3,
MIN_ENCODED_TAIL
};
+/*
+ * Controls the threshold for the number of intra-node lock hand-offs before
+ * the NUMA-aware variant of spinlock is forced to be passed to a thread on
+ * another NUMA node. The default setting can be changed with the
+ * "numa_spinlock_threshold" boot option.
+ */
+unsigned int intra_node_handoff_threshold __ro_after_init = 1 << 16;
+
static void __init cna_init_nodes_per_cpu(unsigned int cpu)
{
struct mcs_spinlock *base = per_cpu_ptr(&qnodes[0].mcs, cpu);
@@ -96,6 +108,11 @@ static int __init cna_init_nodes(void)
return 0;
}
+static __always_inline void cna_init_node(struct mcs_spinlock *node)
+{
+ ((struct cna_node *)node)->intra_count = 0;
+}
+
/*
* cna_splice_head -- splice the entire secondary queue onto the head of the
* primary queue.
@@ -250,11 +267,15 @@ static __always_inline u32 cna_wait_head_or_lock(struct qspinlock *lock,
{
struct cna_node *cn = (struct cna_node *)node;
- /*
- * Try and put the time otherwise spent spin waiting on
- * _Q_LOCKED_PENDING_MASK to use by sorting our lists.
- */
- cn->partial_order = cna_order_queue(node, node);
+ if (cn->intra_count < intra_node_handoff_threshold) {
+ /*
+ * Try and put the time otherwise spent spin waiting on
+ * _Q_LOCKED_PENDING_MASK to use by sorting our lists.
+ */
+ cn->partial_order = cna_order_queue(node, node);
+ } else {
+ cn->partial_order = FLUSH_SECONDARY_QUEUE;
+ }
return 0; /* we lied; we didn't wait, go do so now */
}
@@ -281,8 +302,11 @@ static inline void cna_lock_handoff(struct mcs_spinlock *node,
* cna_order_queue() above.
*/
next = node->next;
- if (node->locked > 1)
+ if (node->locked > 1) {
val = node->locked; /* preseve secondary queue */
+ ((struct cna_node *)next)->intra_count =
+ cn->intra_count + 1;
+ }
} else if (node->locked > 1) {
/*
* When there are no local waiters on the primary queue, splice
@@ -342,3 +366,20 @@ void __init cna_configure_spin_lock_slowpath(void)
pr_info("Enabling CNA spinlock\n");
}
+
+static int __init numa_spinlock_threshold_setup(char *str)
+{
+ int new_threshold_param;
+
+ if (get_option(&str, &new_threshold_param)) {
+ /* valid value is between 0 and 31 */
+ if (new_threshold_param < 0 || new_threshold_param > 31)
+ return 0;
+
+ intra_node_handoff_threshold = 1 << new_threshold_param;
+ return 1;
+ }
+
+ return 0;
+}
+__setup("numa_spinlock_threshold=", numa_spinlock_threshold_setup);
--
2.21.1 (Apple Git-122.3)
Powered by blists - more mailing lists