FEX-Emu/linux
1
0
Fork 0
mirror of https://github.com/FEX-Emu/linux.git synced 2024-12-02 20:59:30 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

srcu: Exact tracking of srcu_data structures containing callbacks

Browse Source 
The current Tree SRCU implementation schedules a workqueue for every
srcu_data covered by a given leaf srcu_node structure having callbacks,
even if only one of those srcu_data structures actually contains
callbacks.  This is clearly inefficient for workloads that don't feature
callbacks everywhere all the time.  This commit therefore adds an array
of masks that are used by the leaf srcu_node structures to track exactly
which srcu_data structures contain callbacks.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
This commit is contained in:
Paul E. McKenney 2017-04-18 16:01:46 -07:00
parent d160a727c4
commit c7e88067c1
2 changed files with 27 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
include/linux/srcutree.h
View File

@ -47,6 +47,8 @@ struct srcu_data {
struct delayed_work work; /* Context for CB invoking. */
struct rcu_head srcu_barrier_head; /* For srcu_barrier() use. */
struct srcu_node *mynode; /* Leaf srcu_node. */
unsigned long grpmask; /* Mask for leaf srcu_node */
/* ->srcu_data_have_cbs[]. */
int cpu;
struct srcu_struct *sp;
};
@ -59,6 +61,8 @@ struct srcu_node {
unsigned long srcu_have_cbs[4]; /* GP seq for children */
/* having CBs, but only */
/* is > ->srcu_gq_seq. */
unsigned long srcu_data_have_cbs[4]; /* Which srcu_data structs */
/* have CBs for given GP? */
struct srcu_node *srcu_parent; /* Next up in tree. */
int grplo; /* Least CPU for node. */
int grphi; /* Biggest CPU for node. */

29
kernel/rcu/srcutree.c
View File

@ -66,8 +66,12 @@ static void init_srcu_struct_nodes(struct srcu_struct *sp, bool is_static)
/* Each pass through this loop initializes one srcu_node structure. */
rcu_for_each_node_breadth_first(sp, snp) {
spin_lock_init(&snp->lock);
for (i = 0; i < ARRAY_SIZE(snp->srcu_have_cbs); i++)
WARN_ON_ONCE(ARRAY_SIZE(snp->srcu_have_cbs) !=
ARRAY_SIZE(snp->srcu_data_have_cbs));
for (i = 0; i < ARRAY_SIZE(snp->srcu_have_cbs); i++) {
snp->srcu_have_cbs[i] = 0;
snp->srcu_data_have_cbs[i] = 0;
}
snp->grplo = -1;
snp->grphi = -1;
if (snp == &sp->node[0]) {
@ -107,6 +111,7 @@ static void init_srcu_struct_nodes(struct srcu_struct *sp, bool is_static)
sdp->cpu = cpu;
INIT_DELAYED_WORK(&sdp->work, srcu_invoke_callbacks);
sdp->sp = sp;
sdp->grpmask = 1 << (cpu - sdp->mynode->grplo);
if (is_static)
continue;
@ -434,16 +439,21 @@ static void srcu_schedule_cbs_sdp(struct srcu_data *sdp, unsigned long delay)
/*
* Schedule callback invocation for all srcu_data structures associated
* with the specified srcu_node structure, if possible, on the corresponding
* CPUs.
* with the specified srcu_node structure that have callbacks for the
* just-completed grace period, the one corresponding to idx. If possible,
* schedule this invocation on the corresponding CPUs.
*/
static void srcu_schedule_cbs_snp(struct srcu_struct *sp, struct srcu_node *snp)
static void srcu_schedule_cbs_snp(struct srcu_struct *sp, struct srcu_node *snp,
unsigned long mask)
{
int cpu;
for (cpu = snp->grplo; cpu <= snp->grphi; cpu++)
for (cpu = snp->grplo; cpu <= snp->grphi; cpu++) {
if (!(mask & (1 << (cpu - snp->grplo))))
continue;
srcu_schedule_cbs_sdp(per_cpu_ptr(sp->sda, cpu),
atomic_read(&sp->srcu_exp_cnt) ? 0 : SRCU_INTERVAL);
}
}
/*
@ -461,6 +471,7 @@ static void srcu_gp_end(struct srcu_struct *sp)
unsigned long gpseq;
int idx;
int idxnext;
unsigned long mask;
struct srcu_node *snp;
/* Prevent more than one additional grace period. */
@ -486,10 +497,12 @@ static void srcu_gp_end(struct srcu_struct *sp)
cbs = snp->srcu_have_cbs[idx] == gpseq;
snp->srcu_have_cbs[idx] = gpseq;
rcu_seq_set_state(&snp->srcu_have_cbs[idx], 1);
mask = snp->srcu_data_have_cbs[idx];
snp->srcu_data_have_cbs[idx] = 0;
spin_unlock_irq(&snp->lock);
if (cbs) {
smp_mb(); /* GP end before CB invocation. */
srcu_schedule_cbs_snp(sp, snp);
srcu_schedule_cbs_snp(sp, snp, mask);
}
}
@ -536,6 +549,8 @@ static void srcu_funnel_gp_start(struct srcu_struct *sp,
spin_lock_irqsave(&snp->lock, flags);
if (ULONG_CMP_GE(snp->srcu_have_cbs[idx], s)) {
snp_seq = snp->srcu_have_cbs[idx];
if (snp == sdp->mynode && snp_seq == s)
snp->srcu_data_have_cbs[idx] |= sdp->grpmask;
spin_unlock_irqrestore(&snp->lock, flags);
if (snp == sdp->mynode && snp_seq != s) {
smp_mb(); /* CBs after GP! */
@ -544,6 +559,8 @@ static void srcu_funnel_gp_start(struct srcu_struct *sp,
return;
}
snp->srcu_have_cbs[idx] = s;
if (snp == sdp->mynode)
snp->srcu_data_have_cbs[idx] |= sdp->grpmask;
spin_unlock_irqrestore(&snp->lock, flags);
}