iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

s390/pai_ext: Enable per-task counting event

Browse Source 
The PMU for PAI NNPA counters enforces the following restriction:

 - No per-task context for PAI NNPA counters.

This restriction is removed. One or more per-task/system-wide counting
events can now be active at the same time while one system wide
sampling event is active.

Acked-by: Sumanth Korikkar <sumanthk@linux.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
This commit is contained in:
Thomas Richter 2024-04-15 14:50:52 +02:00 committed by Alexander Gordeev
parent 14e3768435
commit 3f9ff4c5a0
1 changed files with 63 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

79
arch/s390/kernel/perf_pai_ext.c
View File

@ -115,22 +115,36 @@ static void paiext_free(struct paiext_mapptr *mp)
}
/* Release the PMU if event is the last perf event */
static void paiext_event_destroy(struct perf_event *event)
static void paiext_event_destroy_cpu(struct perf_event *event, int cpu)
{
struct paiext_mapptr *mp = per_cpu_ptr(paiext_root.mapptr, event->cpu);
struct paiext_mapptr *mp = per_cpu_ptr(paiext_root.mapptr, cpu);
struct paiext_map *cpump = mp->mapptr;
mutex_lock(&paiext_reserve_mutex);
if (event->attr.sample_period)
cpump->mode &= ~PAI_MODE_SAMPLING;
free_page(PAI_SAVE_AREA(event));
if (refcount_dec_and_test(&cpump->refcnt)) /* Last reference gone */
paiext_free(mp);
paiext_root_free();
mutex_unlock(&paiext_reserve_mutex);
debug_sprintf_event(paiext_dbg, 4, "%s cpu %d mapptr %p\n", __func__,
event->cpu, mp->mapptr);
}
static void paiext_event_destroy(struct perf_event *event)
{
int cpu;
free_page(PAI_SAVE_AREA(event));
if (event->cpu == -1) {
struct cpumask *mask = PAI_CPU_MASK(event);
for_each_cpu(cpu, mask)
paiext_event_destroy_cpu(event, cpu);
kfree(mask);
} else {
paiext_event_destroy_cpu(event, event->cpu);
}
debug_sprintf_event(paiext_dbg, 4, "%s cpu %d\n", __func__,
event->cpu);
}
/* Used to avoid races in checking concurrent access of counting and
@ -147,19 +161,18 @@ static void paiext_event_destroy(struct perf_event *event)
*
* Allocate the memory for the event.
*/
static int paiext_alloc(struct perf_event_attr *a, struct perf_event *event)
static int paiext_alloc_cpu(struct perf_event_attr *a, int cpu)
{
struct paiext_mapptr *mp;
struct paiext_map *cpump;
int rc;
mutex_lock(&paiext_reserve_mutex);
rc = paiext_root_alloc();
if (rc)
goto unlock;
mp = per_cpu_ptr(paiext_root.mapptr, event->cpu);
mp = per_cpu_ptr(paiext_root.mapptr, cpu);
cpump = mp->mapptr;
if (!cpump) { /* Paiext_map allocated? */
rc = -ENOMEM;
@ -217,6 +230,40 @@ unlock:
return rc;
}
static int paiext_alloc(struct perf_event *event)
{
struct cpumask *maskptr;
int cpu, rc = -ENOMEM;
maskptr = kzalloc(sizeof(*maskptr), GFP_KERNEL);
if (!maskptr)
goto out;
for_each_online_cpu(cpu) {
rc = paiext_alloc_cpu(&event->attr, cpu);
if (rc) {
for_each_cpu(cpu, maskptr)
paiext_event_destroy_cpu(event, cpu);
kfree(maskptr);
goto out;
}
cpumask_set_cpu(cpu, maskptr);
}
/*
* On error all cpumask are freed and all events have been destroyed.
* Save of which CPUs data structures have been allocated for.
* Release them in paicrypt_event_destroy call back function
* for this event.
*/
PAI_CPU_MASK(event) = maskptr;
rc = 0;
out:
debug_sprintf_event(paiext_dbg, 5, "%s cpu %u rc %d\n", __func__,
cpu, rc);
return rc;
}
/* The PAI extension 1 control block supports up to 128 entries. Return
* the index within PAIE1_CB given the event number. Also validate event
* number.
@ -246,8 +293,9 @@ static int paiext_event_init(struct perf_event *event)
rc = paiext_event_valid(event);
if (rc)
return rc;
/* Allow only CPU wide operation, no process context for now. */
if ((event->attach_state & PERF_ATTACH_TASK) || event->cpu == -1)
/* Allow only CPU wide operation for sampling */
if (a->sample_period &&
((event->attach_state & PERF_ATTACH_TASK) || event->cpu == -1))
return -ENOENT;
/* Allow only event NNPA_ALL for sampling. */
if (a->sample_period && a->config != PAI_NNPA_BASE)
@ -262,7 +310,10 @@ static int paiext_event_init(struct perf_event *event)
return -ENOMEM;
}
rc = paiext_alloc(a, event);
if (event->cpu >= 0)
rc = paiext_alloc_cpu(a, event->cpu);
else
rc = paiext_alloc(event);
if (rc) {
free_page(PAI_SAVE_AREA(event));
return rc;
@ -352,8 +403,6 @@ static int paiext_add(struct perf_event *event, int flags)
pcb->acc = virt_to_phys(cpump->area) | 0x1;
/* Enable CPU instruction lookup for PAIE1 control block */
local_ctl_set_bit(0, CR0_PAI_EXTENSION_BIT);
debug_sprintf_event(paiext_dbg, 4, "%s 1508 %llx acc %llx\n",
__func__, S390_lowcore.aicd, pcb->acc);
}
if (flags & PERF_EF_START)
paiext_start(event, PERF_EF_RELOAD);
@ -387,8 +436,6 @@ static void paiext_del(struct perf_event *event, int flags)
local_ctl_clear_bit(0, CR0_PAI_EXTENSION_BIT);
pcb->acc = 0;
S390_lowcore.aicd = 0;
debug_sprintf_event(paiext_dbg, 4, "%s 1508 %llx acc %llx\n",
__func__, S390_lowcore.aicd, pcb->acc);
}
}
@ -544,7 +591,7 @@ static const struct attribute_group *paiext_attr_groups[] = {
/* Performance monitoring unit for mapped counters */
static struct pmu paiext = {
.task_ctx_nr = perf_invalid_context,
.task_ctx_nr = perf_hw_context,
.event_init = paiext_event_init,
.add = paiext_add,
.del = paiext_del,