Staging
v0.7.0
v0.7.0
https://github.com/torvalds/linux
Tip revision: 6d796c50f84ca79f1722bb131799e5a5710c4700 authored by Linus Torvalds on 29 January 2023, 21:59:43 UTC
Linux 6.2-rc6
Linux 6.2-rc6
Tip revision: 6d796c5
perf_cpum_sf.c
// SPDX-License-Identifier: GPL-2.0
/*
* Performance event support for the System z CPU-measurement Sampling Facility
*
* Copyright IBM Corp. 2013, 2018
* Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
*/
#define KMSG_COMPONENT "cpum_sf"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
#include <linux/kernel.h>
#include <linux/kernel_stat.h>
#include <linux/perf_event.h>
#include <linux/percpu.h>
#include <linux/pid.h>
#include <linux/notifier.h>
#include <linux/export.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/moduleparam.h>
#include <asm/cpu_mf.h>
#include <asm/irq.h>
#include <asm/debug.h>
#include <asm/timex.h>
/* Minimum number of sample-data-block-tables:
* At least one table is required for the sampling buffer structure.
* A single table contains up to 511 pointers to sample-data-blocks.
*/
#define CPUM_SF_MIN_SDBT 1
/* Number of sample-data-blocks per sample-data-block-table (SDBT):
* A table contains SDB pointers (8 bytes) and one table-link entry
* that points to the origin of the next SDBT.
*/
#define CPUM_SF_SDB_PER_TABLE ((PAGE_SIZE - 8) / 8)
/* Maximum page offset for an SDBT table-link entry:
* If this page offset is reached, a table-link entry to the next SDBT
* must be added.
*/
#define CPUM_SF_SDBT_TL_OFFSET (CPUM_SF_SDB_PER_TABLE * 8)
static inline int require_table_link(const void *sdbt)
{
return ((unsigned long) sdbt & ~PAGE_MASK) == CPUM_SF_SDBT_TL_OFFSET;
}
/* Minimum and maximum sampling buffer sizes:
*
* This number represents the maximum size of the sampling buffer taking
* the number of sample-data-block-tables into account. Note that these
* numbers apply to the basic-sampling function only.
* The maximum number of SDBs is increased by CPUM_SF_SDB_DIAG_FACTOR if
* the diagnostic-sampling function is active.
*
* Sampling buffer size Buffer characteristics
* ---------------------------------------------------
* 64KB == 16 pages (4KB per page)
* 1 page for SDB-tables
* 15 pages for SDBs
*
* 32MB == 8192 pages (4KB per page)
* 16 pages for SDB-tables
* 8176 pages for SDBs
*/
static unsigned long __read_mostly CPUM_SF_MIN_SDB = 15;
static unsigned long __read_mostly CPUM_SF_MAX_SDB = 8176;
static unsigned long __read_mostly CPUM_SF_SDB_DIAG_FACTOR = 1;
struct sf_buffer {
unsigned long *sdbt; /* Sample-data-block-table origin */
/* buffer characteristics (required for buffer increments) */
unsigned long num_sdb; /* Number of sample-data-blocks */
unsigned long num_sdbt; /* Number of sample-data-block-tables */
unsigned long *tail; /* last sample-data-block-table */
};
struct aux_buffer {
struct sf_buffer sfb;
unsigned long head; /* index of SDB of buffer head */
unsigned long alert_mark; /* index of SDB of alert request position */
unsigned long empty_mark; /* mark of SDB not marked full */
unsigned long *sdb_index; /* SDB address for fast lookup */
unsigned long *sdbt_index; /* SDBT address for fast lookup */
};
struct cpu_hw_sf {
/* CPU-measurement sampling information block */
struct hws_qsi_info_block qsi;
/* CPU-measurement sampling control block */
struct hws_lsctl_request_block lsctl;
struct sf_buffer sfb; /* Sampling buffer */
unsigned int flags; /* Status flags */
struct perf_event *event; /* Scheduled perf event */
struct perf_output_handle handle; /* AUX buffer output handle */
};
static DEFINE_PER_CPU(struct cpu_hw_sf, cpu_hw_sf);
/* Debug feature */
static debug_info_t *sfdbg;
/*
* sf_disable() - Switch off sampling facility
*/
static int sf_disable(void)
{
struct hws_lsctl_request_block sreq;
memset(&sreq, 0, sizeof(sreq));
return lsctl(&sreq);
}
/*
* sf_buffer_available() - Check for an allocated sampling buffer
*/
static int sf_buffer_available(struct cpu_hw_sf *cpuhw)
{
return !!cpuhw->sfb.sdbt;
}
/*
* deallocate sampling facility buffer
*/
static void free_sampling_buffer(struct sf_buffer *sfb)
{
unsigned long *sdbt, *curr;
if (!sfb->sdbt)
return;
sdbt = sfb->sdbt;
curr = sdbt;
/* Free the SDBT after all SDBs are processed... */
while (1) {
if (!*curr || !sdbt)
break;
/* Process table-link entries */
if (is_link_entry(curr)) {
curr = get_next_sdbt(curr);
if (sdbt)
free_page((unsigned long) sdbt);
/* If the origin is reached, sampling buffer is freed */
if (curr == sfb->sdbt)
break;
else
sdbt = curr;
} else {
/* Process SDB pointer */
if (*curr) {
free_page(*curr);
curr++;
}
}
}
debug_sprintf_event(sfdbg, 5, "%s: freed sdbt %#lx\n", __func__,
(unsigned long)sfb->sdbt);
memset(sfb, 0, sizeof(*sfb));
}
static int alloc_sample_data_block(unsigned long *sdbt, gfp_t gfp_flags)
{
struct hws_trailer_entry *te;
unsigned long sdb;
/* Allocate and initialize sample-data-block */
sdb = get_zeroed_page(gfp_flags);
if (!sdb)
return -ENOMEM;
te = (struct hws_trailer_entry *)trailer_entry_ptr(sdb);
te->header.a = 1;
/* Link SDB into the sample-data-block-table */
*sdbt = sdb;
return 0;
}
/*
* realloc_sampling_buffer() - extend sampler memory
*
* Allocates new sample-data-blocks and adds them to the specified sampling
* buffer memory.
*
* Important: This modifies the sampling buffer and must be called when the
* sampling facility is disabled.
*
* Returns zero on success, non-zero otherwise.
*/
static int realloc_sampling_buffer(struct sf_buffer *sfb,
unsigned long num_sdb, gfp_t gfp_flags)
{
int i, rc;
unsigned long *new, *tail, *tail_prev = NULL;
if (!sfb->sdbt || !sfb->tail)
return -EINVAL;
if (!is_link_entry(sfb->tail))
return -EINVAL;
/* Append to the existing sampling buffer, overwriting the table-link
* register.
* The tail variables always points to the "tail" (last and table-link)
* entry in an SDB-table.
*/
tail = sfb->tail;
/* Do a sanity check whether the table-link entry points to
* the sampling buffer origin.
*/
if (sfb->sdbt != get_next_sdbt(tail)) {
debug_sprintf_event(sfdbg, 3, "%s: "
"sampling buffer is not linked: origin %#lx"
" tail %#lx\n", __func__,
(unsigned long)sfb->sdbt,
(unsigned long)tail);
return -EINVAL;
}
/* Allocate remaining SDBs */
rc = 0;
for (i = 0; i < num_sdb; i++) {
/* Allocate a new SDB-table if it is full. */
if (require_table_link(tail)) {
new = (unsigned long *) get_zeroed_page(gfp_flags);
if (!new) {
rc = -ENOMEM;
break;
}
sfb->num_sdbt++;
/* Link current page to tail of chain */
*tail = (unsigned long)(void *) new + 1;
tail_prev = tail;
tail = new;
}
/* Allocate a new sample-data-block.
* If there is not enough memory, stop the realloc process
* and simply use what was allocated. If this is a temporary
* issue, a new realloc call (if required) might succeed.
*/
rc = alloc_sample_data_block(tail, gfp_flags);
if (rc) {
/* Undo last SDBT. An SDBT with no SDB at its first
* entry but with an SDBT entry instead can not be
* handled by the interrupt handler code.
* Avoid this situation.
*/
if (tail_prev) {
sfb->num_sdbt--;
free_page((unsigned long) new);
tail = tail_prev;
}
break;
}
sfb->num_sdb++;
tail++;
tail_prev = new = NULL; /* Allocated at least one SBD */
}
/* Link sampling buffer to its origin */
*tail = (unsigned long) sfb->sdbt + 1;
sfb->tail = tail;
debug_sprintf_event(sfdbg, 4, "%s: new buffer"
" settings: sdbt %lu sdb %lu\n", __func__,
sfb->num_sdbt, sfb->num_sdb);
return rc;
}
/*
* allocate_sampling_buffer() - allocate sampler memory
*
* Allocates and initializes a sampling buffer structure using the
* specified number of sample-data-blocks (SDB). For each allocation,
* a 4K page is used. The number of sample-data-block-tables (SDBT)
* are calculated from SDBs.
* Also set the ALERT_REQ mask in each SDBs trailer.
*
* Returns zero on success, non-zero otherwise.
*/
static int alloc_sampling_buffer(struct sf_buffer *sfb, unsigned long num_sdb)
{
int rc;
if (sfb->sdbt)
return -EINVAL;
/* Allocate the sample-data-block-table origin */
sfb->sdbt = (unsigned long *) get_zeroed_page(GFP_KERNEL);
if (!sfb->sdbt)
return -ENOMEM;
sfb->num_sdb = 0;
sfb->num_sdbt = 1;
/* Link the table origin to point to itself to prepare for
* realloc_sampling_buffer() invocation.
*/
sfb->tail = sfb->sdbt;
*sfb->tail = (unsigned long)(void *) sfb->sdbt + 1;
/* Allocate requested number of sample-data-blocks */
rc = realloc_sampling_buffer(sfb, num_sdb, GFP_KERNEL);
if (rc) {
free_sampling_buffer(sfb);
debug_sprintf_event(sfdbg, 4, "%s: "
"realloc_sampling_buffer failed with rc %i\n",
__func__, rc);
} else
debug_sprintf_event(sfdbg, 4,
"%s: tear %#lx dear %#lx\n", __func__,
(unsigned long)sfb->sdbt, (unsigned long)*sfb->sdbt);
return rc;
}
static void sfb_set_limits(unsigned long min, unsigned long max)
{
struct hws_qsi_info_block si;
CPUM_SF_MIN_SDB = min;
CPUM_SF_MAX_SDB = max;
memset(&si, 0, sizeof(si));
if (!qsi(&si))
CPUM_SF_SDB_DIAG_FACTOR = DIV_ROUND_UP(si.dsdes, si.bsdes);
}
static unsigned long sfb_max_limit(struct hw_perf_event *hwc)
{
return SAMPL_DIAG_MODE(hwc) ? CPUM_SF_MAX_SDB * CPUM_SF_SDB_DIAG_FACTOR
: CPUM_SF_MAX_SDB;
}
static unsigned long sfb_pending_allocs(struct sf_buffer *sfb,
struct hw_perf_event *hwc)
{
if (!sfb->sdbt)
return SFB_ALLOC_REG(hwc);
if (SFB_ALLOC_REG(hwc) > sfb->num_sdb)
return SFB_ALLOC_REG(hwc) - sfb->num_sdb;
return 0;
}
static int sfb_has_pending_allocs(struct sf_buffer *sfb,
struct hw_perf_event *hwc)
{
return sfb_pending_allocs(sfb, hwc) > 0;
}
static void sfb_account_allocs(unsigned long num, struct hw_perf_event *hwc)
{
/* Limit the number of SDBs to not exceed the maximum */
num = min_t(unsigned long, num, sfb_max_limit(hwc) - SFB_ALLOC_REG(hwc));
if (num)
SFB_ALLOC_REG(hwc) += num;
}
static void sfb_init_allocs(unsigned long num, struct hw_perf_event *hwc)
{
SFB_ALLOC_REG(hwc) = 0;
sfb_account_allocs(num, hwc);
}
static void deallocate_buffers(struct cpu_hw_sf *cpuhw)
{
if (cpuhw->sfb.sdbt)
free_sampling_buffer(&cpuhw->sfb);
}
static int allocate_buffers(struct cpu_hw_sf *cpuhw, struct hw_perf_event *hwc)
{
unsigned long n_sdb, freq;
size_t sample_size;
/* Calculate sampling buffers using 4K pages
*
* 1. The sampling size is 32 bytes for basic sampling. This size
* is the same for all machine types. Diagnostic
* sampling uses auxlilary data buffer setup which provides the
* memory for SDBs using linux common code auxiliary trace
* setup.
*
* 2. Function alloc_sampling_buffer() sets the Alert Request
* Control indicator to trigger a measurement-alert to harvest
* sample-data-blocks (SDB). This is done per SDB. This
* measurement alert interrupt fires quick enough to handle
* one SDB, on very high frequency and work loads there might
* be 2 to 3 SBDs available for sample processing.
* Currently there is no need for setup alert request on every
* n-th page. This is counterproductive as one IRQ triggers
* a very high number of samples to be processed at one IRQ.
*
* 3. Use the sampling frequency as input.
* Compute the number of SDBs and ensure a minimum
* of CPUM_SF_MIN_SDB. Depending on frequency add some more
* SDBs to handle a higher sampling rate.
* Use a minimum of CPUM_SF_MIN_SDB and allow for 100 samples
* (one SDB) for every 10000 HZ frequency increment.
*
* 4. Compute the number of sample-data-block-tables (SDBT) and
* ensure a minimum of CPUM_SF_MIN_SDBT (one table can manage up
* to 511 SDBs).
*/
sample_size = sizeof(struct hws_basic_entry);
freq = sample_rate_to_freq(&cpuhw->qsi, SAMPL_RATE(hwc));
n_sdb = CPUM_SF_MIN_SDB + DIV_ROUND_UP(freq, 10000);
/* If there is already a sampling buffer allocated, it is very likely
* that the sampling facility is enabled too. If the event to be
* initialized requires a greater sampling buffer, the allocation must
* be postponed. Changing the sampling buffer requires the sampling
* facility to be in the disabled state. So, account the number of
* required SDBs and let cpumsf_pmu_enable() resize the buffer just
* before the event is started.
*/
sfb_init_allocs(n_sdb, hwc);
if (sf_buffer_available(cpuhw))
return 0;
debug_sprintf_event(sfdbg, 3,
"%s: rate %lu f %lu sdb %lu/%lu"
" sample_size %lu cpuhw %p\n", __func__,
SAMPL_RATE(hwc), freq, n_sdb, sfb_max_limit(hwc),
sample_size, cpuhw);
return alloc_sampling_buffer(&cpuhw->sfb,
sfb_pending_allocs(&cpuhw->sfb, hwc));
}
static unsigned long min_percent(unsigned int percent, unsigned long base,
unsigned long min)
{
return min_t(unsigned long, min, DIV_ROUND_UP(percent * base, 100));
}
static unsigned long compute_sfb_extent(unsigned long ratio, unsigned long base)
{
/* Use a percentage-based approach to extend the sampling facility
* buffer. Accept up to 5% sample data loss.
* Vary the extents between 1% to 5% of the current number of
* sample-data-blocks.
*/
if (ratio <= 5)
return 0;
if (ratio <= 25)
return min_percent(1, base, 1);
if (ratio <= 50)
return min_percent(1, base, 1);
if (ratio <= 75)
return min_percent(2, base, 2);
if (ratio <= 100)
return min_percent(3, base, 3);
if (ratio <= 250)
return min_percent(4, base, 4);
return min_percent(5, base, 8);
}
static void sfb_account_overflows(struct cpu_hw_sf *cpuhw,
struct hw_perf_event *hwc)
{
unsigned long ratio, num;
if (!OVERFLOW_REG(hwc))
return;
/* The sample_overflow contains the average number of sample data
* that has been lost because sample-data-blocks were full.
*
* Calculate the total number of sample data entries that has been
* discarded. Then calculate the ratio of lost samples to total samples
* per second in percent.
*/
ratio = DIV_ROUND_UP(100 * OVERFLOW_REG(hwc) * cpuhw->sfb.num_sdb,
sample_rate_to_freq(&cpuhw->qsi, SAMPL_RATE(hwc)));
/* Compute number of sample-data-blocks */
num = compute_sfb_extent(ratio, cpuhw->sfb.num_sdb);
if (num)
sfb_account_allocs(num, hwc);
debug_sprintf_event(sfdbg, 5, "%s: overflow %llu ratio %lu num %lu\n",
__func__, OVERFLOW_REG(hwc), ratio, num);
OVERFLOW_REG(hwc) = 0;
}
/* extend_sampling_buffer() - Extend sampling buffer
* @sfb: Sampling buffer structure (for local CPU)
* @hwc: Perf event hardware structure
*
* Use this function to extend the sampling buffer based on the overflow counter
* and postponed allocation extents stored in the specified Perf event hardware.
*
* Important: This function disables the sampling facility in order to safely
* change the sampling buffer structure. Do not call this function
* when the PMU is active.
*/
static void extend_sampling_buffer(struct sf_buffer *sfb,
struct hw_perf_event *hwc)
{
unsigned long num, num_old;
int rc;
num = sfb_pending_allocs(sfb, hwc);
if (!num)
return;
num_old = sfb->num_sdb;
/* Disable the sampling facility to reset any states and also
* clear pending measurement alerts.
*/
sf_disable();
/* Extend the sampling buffer.
* This memory allocation typically happens in an atomic context when
* called by perf. Because this is a reallocation, it is fine if the
* new SDB-request cannot be satisfied immediately.
*/
rc = realloc_sampling_buffer(sfb, num, GFP_ATOMIC);
if (rc)
debug_sprintf_event(sfdbg, 5, "%s: realloc failed with rc %i\n",
__func__, rc);
if (sfb_has_pending_allocs(sfb, hwc))
debug_sprintf_event(sfdbg, 5, "%s: "
"req %lu alloc %lu remaining %lu\n",
__func__, num, sfb->num_sdb - num_old,
sfb_pending_allocs(sfb, hwc));
}
/* Number of perf events counting hardware events */
static atomic_t num_events;
/* Used to avoid races in calling reserve/release_cpumf_hardware */
static DEFINE_MUTEX(pmc_reserve_mutex);
#define PMC_INIT 0
#define PMC_RELEASE 1
#define PMC_FAILURE 2
static void setup_pmc_cpu(void *flags)
{
int err;
struct cpu_hw_sf *cpusf = this_cpu_ptr(&cpu_hw_sf);
err = 0;
switch (*((int *) flags)) {
case PMC_INIT:
memset(cpusf, 0, sizeof(*cpusf));
err = qsi(&cpusf->qsi);
if (err)
break;
cpusf->flags |= PMU_F_RESERVED;
err = sf_disable();
if (err)
pr_err("Switching off the sampling facility failed "
"with rc %i\n", err);
debug_sprintf_event(sfdbg, 5,
"%s: initialized: cpuhw %p\n", __func__,
cpusf);
break;
case PMC_RELEASE:
cpusf->flags &= ~PMU_F_RESERVED;
err = sf_disable();
if (err) {
pr_err("Switching off the sampling facility failed "
"with rc %i\n", err);
} else
deallocate_buffers(cpusf);
debug_sprintf_event(sfdbg, 5,
"%s: released: cpuhw %p\n", __func__,
cpusf);
break;
}
if (err)
*((int *) flags) |= PMC_FAILURE;
}
static void release_pmc_hardware(void)
{
int flags = PMC_RELEASE;
irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT);
on_each_cpu(setup_pmc_cpu, &flags, 1);
}
static int reserve_pmc_hardware(void)
{
int flags = PMC_INIT;
on_each_cpu(setup_pmc_cpu, &flags, 1);
if (flags & PMC_FAILURE) {
release_pmc_hardware();
return -ENODEV;
}
irq_subclass_register(IRQ_SUBCLASS_MEASUREMENT_ALERT);
return 0;
}
static void hw_perf_event_destroy(struct perf_event *event)
{
/* Release PMC if this is the last perf event */
if (!atomic_add_unless(&num_events, -1, 1)) {
mutex_lock(&pmc_reserve_mutex);
if (atomic_dec_return(&num_events) == 0)
release_pmc_hardware();
mutex_unlock(&pmc_reserve_mutex);
}
}
static void hw_init_period(struct hw_perf_event *hwc, u64 period)
{
hwc->sample_period = period;
hwc->last_period = hwc->sample_period;
local64_set(&hwc->period_left, hwc->sample_period);
}
static unsigned long hw_limit_rate(const struct hws_qsi_info_block *si,
unsigned long rate)
{
return clamp_t(unsigned long, rate,
si->min_sampl_rate, si->max_sampl_rate);
}
static u32 cpumsf_pid_type(struct perf_event *event,
u32 pid, enum pid_type type)
{
struct task_struct *tsk;
/* Idle process */
if (!pid)
goto out;
tsk = find_task_by_pid_ns(pid, &init_pid_ns);
pid = -1;
if (tsk) {
/*
* Only top level events contain the pid namespace in which
* they are created.
*/
if (event->parent)
event = event->parent;
pid = __task_pid_nr_ns(tsk, type, event->ns);
/*
* See also 1d953111b648
* "perf/core: Don't report zero PIDs for exiting tasks".
*/
if (!pid && !pid_alive(tsk))
pid = -1;
}
out:
return pid;
}
static void cpumsf_output_event_pid(struct perf_event *event,
struct perf_sample_data *data,
struct pt_regs *regs)
{
u32 pid;
struct perf_event_header header;
struct perf_output_handle handle;
/*
* Obtain the PID from the basic-sampling data entry and
* correct the data->tid_entry.pid value.
*/
pid = data->tid_entry.pid;
/* Protect callchain buffers, tasks */
rcu_read_lock();
perf_prepare_sample(&header, data, event, regs);
if (perf_output_begin(&handle, data, event, header.size))
goto out;
/* Update the process ID (see also kernel/events/core.c) */
data->tid_entry.pid = cpumsf_pid_type(event, pid, PIDTYPE_TGID);
data->tid_entry.tid = cpumsf_pid_type(event, pid, PIDTYPE_PID);
perf_output_sample(&handle, &header, data, event);
perf_output_end(&handle);
out:
rcu_read_unlock();
}
static unsigned long getrate(bool freq, unsigned long sample,
struct hws_qsi_info_block *si)
{
unsigned long rate;
if (freq) {
rate = freq_to_sample_rate(si, sample);
rate = hw_limit_rate(si, rate);
} else {
/* The min/max sampling rates specifies the valid range
* of sample periods. If the specified sample period is
* out of range, limit the period to the range boundary.
*/
rate = hw_limit_rate(si, sample);
/* The perf core maintains a maximum sample rate that is
* configurable through the sysctl interface. Ensure the
* sampling rate does not exceed this value. This also helps
* to avoid throttling when pushing samples with
* perf_event_overflow().
*/
if (sample_rate_to_freq(si, rate) >
sysctl_perf_event_sample_rate) {
debug_sprintf_event(sfdbg, 1, "%s: "
"Sampling rate exceeds maximum "
"perf sample rate\n", __func__);
rate = 0;
}
}
return rate;
}
/* The sampling information (si) contains information about the
* min/max sampling intervals and the CPU speed. So calculate the
* correct sampling interval and avoid the whole period adjust
* feedback loop.
*
* Since the CPU Measurement sampling facility can not handle frequency
* calculate the sampling interval when frequency is specified using
* this formula:
* interval := cpu_speed * 1000000 / sample_freq
*
* Returns errno on bad input and zero on success with parameter interval
* set to the correct sampling rate.
*
* Note: This function turns off freq bit to avoid calling function
* perf_adjust_period(). This causes frequency adjustment in the common
* code part which causes tremendous variations in the counter values.
*/
static int __hw_perf_event_init_rate(struct perf_event *event,
struct hws_qsi_info_block *si)
{
struct perf_event_attr *attr = &event->attr;
struct hw_perf_event *hwc = &event->hw;
unsigned long rate;
if (attr->freq) {
if (!attr->sample_freq)
return -EINVAL;
rate = getrate(attr->freq, attr->sample_freq, si);
attr->freq = 0; /* Don't call perf_adjust_period() */
SAMPL_FLAGS(hwc) |= PERF_CPUM_SF_FREQ_MODE;
} else {
rate = getrate(attr->freq, attr->sample_period, si);
if (!rate)
return -EINVAL;
}
attr->sample_period = rate;
SAMPL_RATE(hwc) = rate;
hw_init_period(hwc, SAMPL_RATE(hwc));
debug_sprintf_event(sfdbg, 4, "%s: cpu %d period %#llx freq %d,%#lx\n",
__func__, event->cpu, event->attr.sample_period,
event->attr.freq, SAMPLE_FREQ_MODE(hwc));
return 0;
}
static int __hw_perf_event_init(struct perf_event *event)
{
struct cpu_hw_sf *cpuhw;
struct hws_qsi_info_block si;
struct perf_event_attr *attr = &event->attr;
struct hw_perf_event *hwc = &event->hw;
int cpu, err;
/* Reserve CPU-measurement sampling facility */
err = 0;
if (!atomic_inc_not_zero(&num_events)) {
mutex_lock(&pmc_reserve_mutex);
if (atomic_read(&num_events) == 0 && reserve_pmc_hardware())
err = -EBUSY;
else
atomic_inc(&num_events);
mutex_unlock(&pmc_reserve_mutex);
}
event->destroy = hw_perf_event_destroy;
if (err)
goto out;
/* Access per-CPU sampling information (query sampling info) */
/*
* The event->cpu value can be -1 to count on every CPU, for example,
* when attaching to a task. If this is specified, use the query
* sampling info from the current CPU, otherwise use event->cpu to
* retrieve the per-CPU information.
* Later, cpuhw indicates whether to allocate sampling buffers for a
* particular CPU (cpuhw!=NULL) or each online CPU (cpuw==NULL).
*/
memset(&si, 0, sizeof(si));
cpuhw = NULL;
if (event->cpu == -1)
qsi(&si);
else {
/* Event is pinned to a particular CPU, retrieve the per-CPU
* sampling structure for accessing the CPU-specific QSI.
*/
cpuhw = &per_cpu(cpu_hw_sf, event->cpu);
si = cpuhw->qsi;
}
/* Check sampling facility authorization and, if not authorized,
* fall back to other PMUs. It is safe to check any CPU because
* the authorization is identical for all configured CPUs.
*/
if (!si.as) {
err = -ENOENT;
goto out;
}
if (si.ribm & CPU_MF_SF_RIBM_NOTAV) {
pr_warn("CPU Measurement Facility sampling is temporarily not available\n");
err = -EBUSY;
goto out;
}
/* Always enable basic sampling */
SAMPL_FLAGS(hwc) = PERF_CPUM_SF_BASIC_MODE;
/* Check if diagnostic sampling is requested. Deny if the required
* sampling authorization is missing.
*/
if (attr->config == PERF_EVENT_CPUM_SF_DIAG) {
if (!si.ad) {
err = -EPERM;
goto out;
}
SAMPL_FLAGS(hwc) |= PERF_CPUM_SF_DIAG_MODE;
}
/* Check and set other sampling flags */
if (attr->config1 & PERF_CPUM_SF_FULL_BLOCKS)
SAMPL_FLAGS(hwc) |= PERF_CPUM_SF_FULL_BLOCKS;
err = __hw_perf_event_init_rate(event, &si);
if (err)
goto out;
/* Initialize sample data overflow accounting */
hwc->extra_reg.reg = REG_OVERFLOW;
OVERFLOW_REG(hwc) = 0;
/* Use AUX buffer. No need to allocate it by ourself */
if (attr->config == PERF_EVENT_CPUM_SF_DIAG)
return 0;
/* Allocate the per-CPU sampling buffer using the CPU information
* from the event. If the event is not pinned to a particular
* CPU (event->cpu == -1; or cpuhw == NULL), allocate sampling
* buffers for each online CPU.
*/
if (cpuhw)
/* Event is pinned to a particular CPU */
err = allocate_buffers(cpuhw, hwc);
else {
/* Event is not pinned, allocate sampling buffer on
* each online CPU
*/
for_each_online_cpu(cpu) {
cpuhw = &per_cpu(cpu_hw_sf, cpu);
err = allocate_buffers(cpuhw, hwc);
if (err)
break;
}
}
/* If PID/TID sampling is active, replace the default overflow
* handler to extract and resolve the PIDs from the basic-sampling
* data entries.
*/
if (event->attr.sample_type & PERF_SAMPLE_TID)
if (is_default_overflow_handler(event))
event->overflow_handler = cpumsf_output_event_pid;
out:
return err;
}
static bool is_callchain_event(struct perf_event *event)
{
u64 sample_type = event->attr.sample_type;
return sample_type & (PERF_SAMPLE_CALLCHAIN | PERF_SAMPLE_REGS_USER |
PERF_SAMPLE_STACK_USER);
}
static int cpumsf_pmu_event_init(struct perf_event *event)
{
int err;
/* No support for taken branch sampling */
/* No support for callchain, stacks and registers */
if (has_branch_stack(event) || is_callchain_event(event))
return -EOPNOTSUPP;
switch (event->attr.type) {
case PERF_TYPE_RAW:
if ((event->attr.config != PERF_EVENT_CPUM_SF) &&
(event->attr.config != PERF_EVENT_CPUM_SF_DIAG))
return -ENOENT;
break;
case PERF_TYPE_HARDWARE:
/* Support sampling of CPU cycles in addition to the
* counter facility. However, the counter facility
* is more precise and, hence, restrict this PMU to
* sampling events only.
*/
if (event->attr.config != PERF_COUNT_HW_CPU_CYCLES)
return -ENOENT;
if (!is_sampling_event(event))
return -ENOENT;
break;
default:
return -ENOENT;
}
/* Check online status of the CPU to which the event is pinned */
if (event->cpu >= 0 && !cpu_online(event->cpu))
return -ENODEV;
/* Force reset of idle/hv excludes regardless of what the
* user requested.
*/
if (event->attr.exclude_hv)
event->attr.exclude_hv = 0;
if (event->attr.exclude_idle)
event->attr.exclude_idle = 0;
err = __hw_perf_event_init(event);
if (unlikely(err))
if (event->destroy)
event->destroy(event);
return err;
}
static void cpumsf_pmu_enable(struct pmu *pmu)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
struct hw_perf_event *hwc;
int err;
if (cpuhw->flags & PMU_F_ENABLED)
return;
if (cpuhw->flags & PMU_F_ERR_MASK)
return;
/* Check whether to extent the sampling buffer.
*
* Two conditions trigger an increase of the sampling buffer for a
* perf event:
* 1. Postponed buffer allocations from the event initialization.
* 2. Sampling overflows that contribute to pending allocations.
*
* Note that the extend_sampling_buffer() function disables the sampling
* facility, but it can be fully re-enabled using sampling controls that
* have been saved in cpumsf_pmu_disable().
*/
if (cpuhw->event) {
hwc = &cpuhw->event->hw;
if (!(SAMPL_DIAG_MODE(hwc))) {
/*
* Account number of overflow-designated
* buffer extents
*/
sfb_account_overflows(cpuhw, hwc);
extend_sampling_buffer(&cpuhw->sfb, hwc);
}
/* Rate may be adjusted with ioctl() */
cpuhw->lsctl.interval = SAMPL_RATE(&cpuhw->event->hw);
}
/* (Re)enable the PMU and sampling facility */
cpuhw->flags |= PMU_F_ENABLED;
barrier();
err = lsctl(&cpuhw->lsctl);
if (err) {
cpuhw->flags &= ~PMU_F_ENABLED;
pr_err("Loading sampling controls failed: op %i err %i\n",
1, err);
return;
}
/* Load current program parameter */
lpp(&S390_lowcore.lpp);
debug_sprintf_event(sfdbg, 6, "%s: es %i cs %i ed %i cd %i "
"interval %#lx tear %#lx dear %#lx\n", __func__,
cpuhw->lsctl.es, cpuhw->lsctl.cs, cpuhw->lsctl.ed,
cpuhw->lsctl.cd, cpuhw->lsctl.interval,
cpuhw->lsctl.tear, cpuhw->lsctl.dear);
}
static void cpumsf_pmu_disable(struct pmu *pmu)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
struct hws_lsctl_request_block inactive;
struct hws_qsi_info_block si;
int err;
if (!(cpuhw->flags & PMU_F_ENABLED))
return;
if (cpuhw->flags & PMU_F_ERR_MASK)
return;
/* Switch off sampling activation control */
inactive = cpuhw->lsctl;
inactive.cs = 0;
inactive.cd = 0;
err = lsctl(&inactive);
if (err) {
pr_err("Loading sampling controls failed: op %i err %i\n",
2, err);
return;
}
/* Save state of TEAR and DEAR register contents */
err = qsi(&si);
if (!err) {
/* TEAR/DEAR values are valid only if the sampling facility is
* enabled. Note that cpumsf_pmu_disable() might be called even
* for a disabled sampling facility because cpumsf_pmu_enable()
* controls the enable/disable state.
*/
if (si.es) {
cpuhw->lsctl.tear = si.tear;
cpuhw->lsctl.dear = si.dear;
}
} else
debug_sprintf_event(sfdbg, 3, "%s: qsi() failed with err %i\n",
__func__, err);
cpuhw->flags &= ~PMU_F_ENABLED;
}
/* perf_exclude_event() - Filter event
* @event: The perf event
* @regs: pt_regs structure
* @sde_regs: Sample-data-entry (sde) regs structure
*
* Filter perf events according to their exclude specification.
*
* Return non-zero if the event shall be excluded.
*/
static int perf_exclude_event(struct perf_event *event, struct pt_regs *regs,
struct perf_sf_sde_regs *sde_regs)
{
if (event->attr.exclude_user && user_mode(regs))
return 1;
if (event->attr.exclude_kernel && !user_mode(regs))
return 1;
if (event->attr.exclude_guest && sde_regs->in_guest)
return 1;
if (event->attr.exclude_host && !sde_regs->in_guest)
return 1;
return 0;
}
/* perf_push_sample() - Push samples to perf
* @event: The perf event
* @sample: Hardware sample data
*
* Use the hardware sample data to create perf event sample. The sample
* is the pushed to the event subsystem and the function checks for
* possible event overflows. If an event overflow occurs, the PMU is
* stopped.
*
* Return non-zero if an event overflow occurred.
*/
static int perf_push_sample(struct perf_event *event,
struct hws_basic_entry *basic)
{
int overflow;
struct pt_regs regs;
struct perf_sf_sde_regs *sde_regs;
struct perf_sample_data data;
/* Setup perf sample */
perf_sample_data_init(&data, 0, event->hw.last_period);
/* Setup pt_regs to look like an CPU-measurement external interrupt
* using the Program Request Alert code. The regs.int_parm_long
* field which is unused contains additional sample-data-entry related
* indicators.
*/
memset(®s, 0, sizeof(regs));
regs.int_code = 0x1407;
regs.int_parm = CPU_MF_INT_SF_PRA;
sde_regs = (struct perf_sf_sde_regs *) ®s.int_parm_long;
psw_bits(regs.psw).ia = basic->ia;
psw_bits(regs.psw).dat = basic->T;
psw_bits(regs.psw).wait = basic->W;
psw_bits(regs.psw).pstate = basic->P;
psw_bits(regs.psw).as = basic->AS;
/*
* Use the hardware provided configuration level to decide if the
* sample belongs to a guest or host. If that is not available,
* fall back to the following heuristics:
* A non-zero guest program parameter always indicates a guest
* sample. Some early samples or samples from guests without
* lpp usage would be misaccounted to the host. We use the asn
* value as an addon heuristic to detect most of these guest samples.
* If the value differs from 0xffff (the host value), we assume to
* be a KVM guest.
*/
switch (basic->CL) {
case 1: /* logical partition */
sde_regs->in_guest = 0;
break;
case 2: /* virtual machine */
sde_regs->in_guest = 1;
break;
default: /* old machine, use heuristics */
if (basic->gpp || basic->prim_asn != 0xffff)
sde_regs->in_guest = 1;
break;
}
/*
* Store the PID value from the sample-data-entry to be
* processed and resolved by cpumsf_output_event_pid().
*/
data.tid_entry.pid = basic->hpp & LPP_PID_MASK;
overflow = 0;
if (perf_exclude_event(event, ®s, sde_regs))
goto out;
if (perf_event_overflow(event, &data, ®s)) {
overflow = 1;
event->pmu->stop(event, 0);
}
perf_event_update_userpage(event);
out:
return overflow;
}
static void perf_event_count_update(struct perf_event *event, u64 count)
{
local64_add(count, &event->count);
}
/* hw_collect_samples() - Walk through a sample-data-block and collect samples
* @event: The perf event
* @sdbt: Sample-data-block table
* @overflow: Event overflow counter
*
* Walks through a sample-data-block and collects sampling data entries that are
* then pushed to the perf event subsystem. Depending on the sampling function,
* there can be either basic-sampling or combined-sampling data entries. A
* combined-sampling data entry consists of a basic- and a diagnostic-sampling
* data entry. The sampling function is determined by the flags in the perf
* event hardware structure. The function always works with a combined-sampling
* data entry but ignores the the diagnostic portion if it is not available.
*
* Note that the implementation focuses on basic-sampling data entries and, if
* such an entry is not valid, the entire combined-sampling data entry is
* ignored.
*
* The overflow variables counts the number of samples that has been discarded
* due to a perf event overflow.
*/
static void hw_collect_samples(struct perf_event *event, unsigned long *sdbt,
unsigned long long *overflow)
{
struct hws_trailer_entry *te;
struct hws_basic_entry *sample;
te = (struct hws_trailer_entry *) trailer_entry_ptr(*sdbt);
sample = (struct hws_basic_entry *) *sdbt;
while ((unsigned long *) sample < (unsigned long *) te) {
/* Check for an empty sample */
if (!sample->def || sample->LS)
break;
/* Update perf event period */
perf_event_count_update(event, SAMPL_RATE(&event->hw));
/* Check whether sample is valid */
if (sample->def == 0x0001) {
/* If an event overflow occurred, the PMU is stopped to
* throttle event delivery. Remaining sample data is
* discarded.
*/
if (!*overflow) {
/* Check whether sample is consistent */
if (sample->I == 0 && sample->W == 0) {
/* Deliver sample data to perf */
*overflow = perf_push_sample(event,
sample);
}
} else
/* Count discarded samples */
*overflow += 1;
} else {
debug_sprintf_event(sfdbg, 4,
"%s: Found unknown"
" sampling data entry: te->f %i"
" basic.def %#4x (%p)\n", __func__,
te->header.f, sample->def, sample);
/* Sample slot is not yet written or other record.
*
* This condition can occur if the buffer was reused
* from a combined basic- and diagnostic-sampling.
* If only basic-sampling is then active, entries are
* written into the larger diagnostic entries.
* This is typically the case for sample-data-blocks
* that are not full. Stop processing if the first
* invalid format was detected.
*/
if (!te->header.f)
break;
}
/* Reset sample slot and advance to next sample */
sample->def = 0;
sample++;
}
}
static inline __uint128_t __cdsg(__uint128_t *ptr, __uint128_t old, __uint128_t new)
{
asm volatile(
" cdsg %[old],%[new],%[ptr]\n"
: [old] "+d" (old), [ptr] "+QS" (*ptr)
: [new] "d" (new)
: "memory", "cc");
return old;
}
/* hw_perf_event_update() - Process sampling buffer
* @event: The perf event
* @flush_all: Flag to also flush partially filled sample-data-blocks
*
* Processes the sampling buffer and create perf event samples.
* The sampling buffer position are retrieved and saved in the TEAR_REG
* register of the specified perf event.
*
* Only full sample-data-blocks are processed. Specify the flash_all flag
* to also walk through partially filled sample-data-blocks. It is ignored
* if PERF_CPUM_SF_FULL_BLOCKS is set. The PERF_CPUM_SF_FULL_BLOCKS flag
* enforces the processing of full sample-data-blocks only (trailer entries
* with the block-full-indicator bit set).
*/
static void hw_perf_event_update(struct perf_event *event, int flush_all)
{
unsigned long long event_overflow, sampl_overflow, num_sdb;
union hws_trailer_header old, prev, new;
struct hw_perf_event *hwc = &event->hw;
struct hws_trailer_entry *te;
unsigned long *sdbt;
int done;
/*
* AUX buffer is used when in diagnostic sampling mode.
* No perf events/samples are created.
*/
if (SAMPL_DIAG_MODE(&event->hw))
return;
if (flush_all && SDB_FULL_BLOCKS(hwc))
flush_all = 0;
sdbt = (unsigned long *) TEAR_REG(hwc);
done = event_overflow = sampl_overflow = num_sdb = 0;
while (!done) {
/* Get the trailer entry of the sample-data-block */
te = (struct hws_trailer_entry *) trailer_entry_ptr(*sdbt);
/* Leave loop if no more work to do (block full indicator) */
if (!te->header.f) {
done = 1;
if (!flush_all)
break;
}
/* Check the sample overflow count */
if (te->header.overflow)
/* Account sample overflows and, if a particular limit
* is reached, extend the sampling buffer.
* For details, see sfb_account_overflows().
*/
sampl_overflow += te->header.overflow;
/* Timestamps are valid for full sample-data-blocks only */
debug_sprintf_event(sfdbg, 6, "%s: sdbt %#lx "
"overflow %llu timestamp %#llx\n",
__func__, (unsigned long)sdbt, te->header.overflow,
(te->header.f) ? trailer_timestamp(te) : 0ULL);
/* Collect all samples from a single sample-data-block and
* flag if an (perf) event overflow happened. If so, the PMU
* is stopped and remaining samples will be discarded.
*/
hw_collect_samples(event, sdbt, &event_overflow);
num_sdb++;
/* Reset trailer (using compare-double-and-swap) */
/* READ_ONCE() 16 byte header */
prev.val = __cdsg(&te->header.val, 0, 0);
do {
old.val = prev.val;
new.val = prev.val;
new.f = 0;
new.a = 1;
new.overflow = 0;
prev.val = __cdsg(&te->header.val, old.val, new.val);
} while (prev.val != old.val);
/* Advance to next sample-data-block */
sdbt++;
if (is_link_entry(sdbt))
sdbt = get_next_sdbt(sdbt);
/* Update event hardware registers */
TEAR_REG(hwc) = (unsigned long) sdbt;
/* Stop processing sample-data if all samples of the current
* sample-data-block were flushed even if it was not full.
*/
if (flush_all && done)
break;
}
/* Account sample overflows in the event hardware structure */
if (sampl_overflow)
OVERFLOW_REG(hwc) = DIV_ROUND_UP(OVERFLOW_REG(hwc) +
sampl_overflow, 1 + num_sdb);
/* Perf_event_overflow() and perf_event_account_interrupt() limit
* the interrupt rate to an upper limit. Roughly 1000 samples per
* task tick.
* Hitting this limit results in a large number
* of throttled REF_REPORT_THROTTLE entries and the samples
* are dropped.
* Slightly increase the interval to avoid hitting this limit.
*/
if (event_overflow) {
SAMPL_RATE(hwc) += DIV_ROUND_UP(SAMPL_RATE(hwc), 10);
debug_sprintf_event(sfdbg, 1, "%s: rate adjustment %ld\n",
__func__,
DIV_ROUND_UP(SAMPL_RATE(hwc), 10));
}
if (sampl_overflow || event_overflow)
debug_sprintf_event(sfdbg, 4, "%s: "
"overflows: sample %llu event %llu"
" total %llu num_sdb %llu\n",
__func__, sampl_overflow, event_overflow,
OVERFLOW_REG(hwc), num_sdb);
}
#define AUX_SDB_INDEX(aux, i) ((i) % aux->sfb.num_sdb)
#define AUX_SDB_NUM(aux, start, end) (end >= start ? end - start + 1 : 0)
#define AUX_SDB_NUM_ALERT(aux) AUX_SDB_NUM(aux, aux->head, aux->alert_mark)
#define AUX_SDB_NUM_EMPTY(aux) AUX_SDB_NUM(aux, aux->head, aux->empty_mark)
/*
* Get trailer entry by index of SDB.
*/
static struct hws_trailer_entry *aux_sdb_trailer(struct aux_buffer *aux,
unsigned long index)
{
unsigned long sdb;
index = AUX_SDB_INDEX(aux, index);
sdb = aux->sdb_index[index];
return (struct hws_trailer_entry *)trailer_entry_ptr(sdb);
}
/*
* Finish sampling on the cpu. Called by cpumsf_pmu_del() with pmu
* disabled. Collect the full SDBs in AUX buffer which have not reached
* the point of alert indicator. And ignore the SDBs which are not
* full.
*
* 1. Scan SDBs to see how much data is there and consume them.
* 2. Remove alert indicator in the buffer.
*/
static void aux_output_end(struct perf_output_handle *handle)
{
unsigned long i, range_scan, idx;
struct aux_buffer *aux;
struct hws_trailer_entry *te;
aux = perf_get_aux(handle);
if (!aux)
return;
range_scan = AUX_SDB_NUM_ALERT(aux);
for (i = 0, idx = aux->head; i < range_scan; i++, idx++) {
te = aux_sdb_trailer(aux, idx);
if (!te->header.f)
break;
}
/* i is num of SDBs which are full */
perf_aux_output_end(handle, i << PAGE_SHIFT);
/* Remove alert indicators in the buffer */
te = aux_sdb_trailer(aux, aux->alert_mark);
te->header.a = 0;
debug_sprintf_event(sfdbg, 6, "%s: SDBs %ld range %ld head %ld\n",
__func__, i, range_scan, aux->head);
}
/*
* Start sampling on the CPU. Called by cpumsf_pmu_add() when an event
* is first added to the CPU or rescheduled again to the CPU. It is called
* with pmu disabled.
*
* 1. Reset the trailer of SDBs to get ready for new data.
* 2. Tell the hardware where to put the data by reset the SDBs buffer
* head(tear/dear).
*/
static int aux_output_begin(struct perf_output_handle *handle,
struct aux_buffer *aux,
struct cpu_hw_sf *cpuhw)
{
unsigned long range;
unsigned long i, range_scan, idx;
unsigned long head, base, offset;
struct hws_trailer_entry *te;
if (WARN_ON_ONCE(handle->head & ~PAGE_MASK))
return -EINVAL;
aux->head = handle->head >> PAGE_SHIFT;
range = (handle->size + 1) >> PAGE_SHIFT;
if (range <= 1)
return -ENOMEM;
/*
* SDBs between aux->head and aux->empty_mark are already ready
* for new data. range_scan is num of SDBs not within them.
*/
debug_sprintf_event(sfdbg, 6,
"%s: range %ld head %ld alert %ld empty %ld\n",
__func__, range, aux->head, aux->alert_mark,
aux->empty_mark);
if (range > AUX_SDB_NUM_EMPTY(aux)) {
range_scan = range - AUX_SDB_NUM_EMPTY(aux);
idx = aux->empty_mark + 1;
for (i = 0; i < range_scan; i++, idx++) {
te = aux_sdb_trailer(aux, idx);
te->header.f = 0;
te->header.a = 0;
te->header.overflow = 0;
}
/* Save the position of empty SDBs */
aux->empty_mark = aux->head + range - 1;
}
/* Set alert indicator */
aux->alert_mark = aux->head + range/2 - 1;
te = aux_sdb_trailer(aux, aux->alert_mark);
te->header.a = 1;
/* Reset hardware buffer head */
head = AUX_SDB_INDEX(aux, aux->head);
base = aux->sdbt_index[head / CPUM_SF_SDB_PER_TABLE];
offset = head % CPUM_SF_SDB_PER_TABLE;
cpuhw->lsctl.tear = base + offset * sizeof(unsigned long);
cpuhw->lsctl.dear = aux->sdb_index[head];
debug_sprintf_event(sfdbg, 6, "%s: head %ld alert %ld empty %ld "
"index %ld tear %#lx dear %#lx\n", __func__,
aux->head, aux->alert_mark, aux->empty_mark,
head / CPUM_SF_SDB_PER_TABLE,
cpuhw->lsctl.tear, cpuhw->lsctl.dear);
return 0;
}
/*
* Set alert indicator on SDB at index @alert_index while sampler is running.
*
* Return true if successfully.
* Return false if full indicator is already set by hardware sampler.
*/
static bool aux_set_alert(struct aux_buffer *aux, unsigned long alert_index,
unsigned long long *overflow)
{
union hws_trailer_header old, prev, new;
struct hws_trailer_entry *te;
te = aux_sdb_trailer(aux, alert_index);
/* READ_ONCE() 16 byte header */
prev.val = __cdsg(&te->header.val, 0, 0);
do {
old.val = prev.val;
new.val = prev.val;
*overflow = old.overflow;
if (old.f) {
/*
* SDB is already set by hardware.
* Abort and try to set somewhere
* behind.
*/
return false;
}
new.a = 1;
new.overflow = 0;
prev.val = __cdsg(&te->header.val, old.val, new.val);
} while (prev.val != old.val);
return true;
}
/*
* aux_reset_buffer() - Scan and setup SDBs for new samples
* @aux: The AUX buffer to set
* @range: The range of SDBs to scan started from aux->head
* @overflow: Set to overflow count
*
* Set alert indicator on the SDB at index of aux->alert_mark. If this SDB is
* marked as empty, check if it is already set full by the hardware sampler.
* If yes, that means new data is already there before we can set an alert
* indicator. Caller should try to set alert indicator to some position behind.
*
* Scan the SDBs in AUX buffer from behind aux->empty_mark. They are used
* previously and have already been consumed by user space. Reset these SDBs
* (clear full indicator and alert indicator) for new data.
* If aux->alert_mark fall in this area, just set it. Overflow count is
* recorded while scanning.
*
* SDBs between aux->head and aux->empty_mark are already reset at last time.
* and ready for new samples. So scanning on this area could be skipped.
*
* Return true if alert indicator is set successfully and false if not.
*/
static bool aux_reset_buffer(struct aux_buffer *aux, unsigned long range,
unsigned long long *overflow)
{
unsigned long i, range_scan, idx, idx_old;
union hws_trailer_header old, prev, new;
unsigned long long orig_overflow;
struct hws_trailer_entry *te;
debug_sprintf_event(sfdbg, 6, "%s: range %ld head %ld alert %ld "
"empty %ld\n", __func__, range, aux->head,
aux->alert_mark, aux->empty_mark);
if (range <= AUX_SDB_NUM_EMPTY(aux))
/*
* No need to scan. All SDBs in range are marked as empty.
* Just set alert indicator. Should check race with hardware
* sampler.
*/
return aux_set_alert(aux, aux->alert_mark, overflow);
if (aux->alert_mark <= aux->empty_mark)
/*
* Set alert indicator on empty SDB. Should check race
* with hardware sampler.
*/
if (!aux_set_alert(aux, aux->alert_mark, overflow))
return false;
/*
* Scan the SDBs to clear full and alert indicator used previously.
* Start scanning from one SDB behind empty_mark. If the new alert
* indicator fall into this range, set it.
*/
range_scan = range - AUX_SDB_NUM_EMPTY(aux);
idx_old = idx = aux->empty_mark + 1;
for (i = 0; i < range_scan; i++, idx++) {
te = aux_sdb_trailer(aux, idx);
/* READ_ONCE() 16 byte header */
prev.val = __cdsg(&te->header.val, 0, 0);
do {
old.val = prev.val;
new.val = prev.val;
orig_overflow = old.overflow;
new.f = 0;
new.overflow = 0;
if (idx == aux->alert_mark)
new.a = 1;
else
new.a = 0;
prev.val = __cdsg(&te->header.val, old.val, new.val);
} while (prev.val != old.val);
*overflow += orig_overflow;
}
/* Update empty_mark to new position */
aux->empty_mark = aux->head + range - 1;
debug_sprintf_event(sfdbg, 6, "%s: range_scan %ld idx %ld..%ld "
"empty %ld\n", __func__, range_scan, idx_old,
idx - 1, aux->empty_mark);
return true;
}
/*
* Measurement alert handler for diagnostic mode sampling.
*/
static void hw_collect_aux(struct cpu_hw_sf *cpuhw)
{
struct aux_buffer *aux;
int done = 0;
unsigned long range = 0, size;
unsigned long long overflow = 0;
struct perf_output_handle *handle = &cpuhw->handle;
unsigned long num_sdb;
aux = perf_get_aux(handle);
if (WARN_ON_ONCE(!aux))
return;
/* Inform user space new data arrived */
size = AUX_SDB_NUM_ALERT(aux) << PAGE_SHIFT;
debug_sprintf_event(sfdbg, 6, "%s: #alert %ld\n", __func__,
size >> PAGE_SHIFT);
perf_aux_output_end(handle, size);
num_sdb = aux->sfb.num_sdb;
while (!done) {
/* Get an output handle */
aux = perf_aux_output_begin(handle, cpuhw->event);
if (handle->size == 0) {
pr_err("The AUX buffer with %lu pages for the "
"diagnostic-sampling mode is full\n",
num_sdb);
debug_sprintf_event(sfdbg, 1,
"%s: AUX buffer used up\n",
__func__);
break;
}
if (WARN_ON_ONCE(!aux))
return;
/* Update head and alert_mark to new position */
aux->head = handle->head >> PAGE_SHIFT;
range = (handle->size + 1) >> PAGE_SHIFT;
if (range == 1)
aux->alert_mark = aux->head;
else
aux->alert_mark = aux->head + range/2 - 1;
if (aux_reset_buffer(aux, range, &overflow)) {
if (!overflow) {
done = 1;
break;
}
size = range << PAGE_SHIFT;
perf_aux_output_end(&cpuhw->handle, size);
pr_err("Sample data caused the AUX buffer with %lu "
"pages to overflow\n", aux->sfb.num_sdb);
debug_sprintf_event(sfdbg, 1, "%s: head %ld range %ld "
"overflow %lld\n", __func__,
aux->head, range, overflow);
} else {
size = AUX_SDB_NUM_ALERT(aux) << PAGE_SHIFT;
perf_aux_output_end(&cpuhw->handle, size);
debug_sprintf_event(sfdbg, 6, "%s: head %ld alert %ld "
"already full, try another\n",
__func__,
aux->head, aux->alert_mark);
}
}
if (done)
debug_sprintf_event(sfdbg, 6, "%s: head %ld alert %ld "
"empty %ld\n", __func__, aux->head,
aux->alert_mark, aux->empty_mark);
}
/*
* Callback when freeing AUX buffers.
*/
static void aux_buffer_free(void *data)
{
struct aux_buffer *aux = data;
unsigned long i, num_sdbt;
if (!aux)
return;
/* Free SDBT. SDB is freed by the caller */
num_sdbt = aux->sfb.num_sdbt;
for (i = 0; i < num_sdbt; i++)
free_page(aux->sdbt_index[i]);
kfree(aux->sdbt_index);
kfree(aux->sdb_index);
kfree(aux);
debug_sprintf_event(sfdbg, 4, "%s: SDBTs %lu\n", __func__, num_sdbt);
}
static void aux_sdb_init(unsigned long sdb)
{
struct hws_trailer_entry *te;
te = (struct hws_trailer_entry *)trailer_entry_ptr(sdb);
/* Save clock base */
te->clock_base = 1;
te->progusage2 = tod_clock_base.tod;
}
/*
* aux_buffer_setup() - Setup AUX buffer for diagnostic mode sampling
* @event: Event the buffer is setup for, event->cpu == -1 means current
* @pages: Array of pointers to buffer pages passed from perf core
* @nr_pages: Total pages
* @snapshot: Flag for snapshot mode
*
* This is the callback when setup an event using AUX buffer. Perf tool can
* trigger this by an additional mmap() call on the event. Unlike the buffer
* for basic samples, AUX buffer belongs to the event. It is scheduled with
* the task among online cpus when it is a per-thread event.
*
* Return the private AUX buffer structure if success or NULL if fails.
*/
static void *aux_buffer_setup(struct perf_event *event, void **pages,
int nr_pages, bool snapshot)
{
struct sf_buffer *sfb;
struct aux_buffer *aux;
unsigned long *new, *tail;
int i, n_sdbt;
if (!nr_pages || !pages)
return NULL;
if (nr_pages > CPUM_SF_MAX_SDB * CPUM_SF_SDB_DIAG_FACTOR) {
pr_err("AUX buffer size (%i pages) is larger than the "
"maximum sampling buffer limit\n",
nr_pages);
return NULL;
} else if (nr_pages < CPUM_SF_MIN_SDB * CPUM_SF_SDB_DIAG_FACTOR) {
pr_err("AUX buffer size (%i pages) is less than the "
"minimum sampling buffer limit\n",
nr_pages);
return NULL;
}
/* Allocate aux_buffer struct for the event */
aux = kzalloc(sizeof(struct aux_buffer), GFP_KERNEL);
if (!aux)
goto no_aux;
sfb = &aux->sfb;
/* Allocate sdbt_index for fast reference */
n_sdbt = DIV_ROUND_UP(nr_pages, CPUM_SF_SDB_PER_TABLE);
aux->sdbt_index = kmalloc_array(n_sdbt, sizeof(void *), GFP_KERNEL);
if (!aux->sdbt_index)
goto no_sdbt_index;
/* Allocate sdb_index for fast reference */
aux->sdb_index = kmalloc_array(nr_pages, sizeof(void *), GFP_KERNEL);
if (!aux->sdb_index)
goto no_sdb_index;
/* Allocate the first SDBT */
sfb->num_sdbt = 0;
sfb->sdbt = (unsigned long *) get_zeroed_page(GFP_KERNEL);
if (!sfb->sdbt)
goto no_sdbt;
aux->sdbt_index[sfb->num_sdbt++] = (unsigned long)sfb->sdbt;
tail = sfb->tail = sfb->sdbt;
/*
* Link the provided pages of AUX buffer to SDBT.
* Allocate SDBT if needed.
*/
for (i = 0; i < nr_pages; i++, tail++) {
if (require_table_link(tail)) {
new = (unsigned long *) get_zeroed_page(GFP_KERNEL);
if (!new)
goto no_sdbt;
aux->sdbt_index[sfb->num_sdbt++] = (unsigned long)new;
/* Link current page to tail of chain */
*tail = (unsigned long)(void *) new + 1;
tail = new;
}
/* Tail is the entry in a SDBT */
*tail = (unsigned long)pages[i];
aux->sdb_index[i] = (unsigned long)pages[i];
aux_sdb_init((unsigned long)pages[i]);
}
sfb->num_sdb = nr_pages;
/* Link the last entry in the SDBT to the first SDBT */
*tail = (unsigned long) sfb->sdbt + 1;
sfb->tail = tail;
/*
* Initial all SDBs are zeroed. Mark it as empty.
* So there is no need to clear the full indicator
* when this event is first added.
*/
aux->empty_mark = sfb->num_sdb - 1;
debug_sprintf_event(sfdbg, 4, "%s: SDBTs %lu SDBs %lu\n", __func__,
sfb->num_sdbt, sfb->num_sdb);
return aux;
no_sdbt:
/* SDBs (AUX buffer pages) are freed by caller */
for (i = 0; i < sfb->num_sdbt; i++)
free_page(aux->sdbt_index[i]);
kfree(aux->sdb_index);
no_sdb_index:
kfree(aux->sdbt_index);
no_sdbt_index:
kfree(aux);
no_aux:
return NULL;
}
static void cpumsf_pmu_read(struct perf_event *event)
{
/* Nothing to do ... updates are interrupt-driven */
}
/* Check if the new sampling period/freqeuncy is appropriate.
*
* Return non-zero on error and zero on passed checks.
*/
static int cpumsf_pmu_check_period(struct perf_event *event, u64 value)
{
struct hws_qsi_info_block si;
unsigned long rate;
bool do_freq;
memset(&si, 0, sizeof(si));
if (event->cpu == -1) {
if (qsi(&si))
return -ENODEV;
} else {
/* Event is pinned to a particular CPU, retrieve the per-CPU
* sampling structure for accessing the CPU-specific QSI.
*/
struct cpu_hw_sf *cpuhw = &per_cpu(cpu_hw_sf, event->cpu);
si = cpuhw->qsi;
}
do_freq = !!SAMPLE_FREQ_MODE(&event->hw);
rate = getrate(do_freq, value, &si);
if (!rate)
return -EINVAL;
event->attr.sample_period = rate;
SAMPL_RATE(&event->hw) = rate;
hw_init_period(&event->hw, SAMPL_RATE(&event->hw));
debug_sprintf_event(sfdbg, 4, "%s:"
" cpu %d value %#llx period %#llx freq %d\n",
__func__, event->cpu, value,
event->attr.sample_period, do_freq);
return 0;
}
/* Activate sampling control.
* Next call of pmu_enable() starts sampling.
*/
static void cpumsf_pmu_start(struct perf_event *event, int flags)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED)))
return;
if (flags & PERF_EF_RELOAD)
WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE));
perf_pmu_disable(event->pmu);
event->hw.state = 0;
cpuhw->lsctl.cs = 1;
if (SAMPL_DIAG_MODE(&event->hw))
cpuhw->lsctl.cd = 1;
perf_pmu_enable(event->pmu);
}
/* Deactivate sampling control.
* Next call of pmu_enable() stops sampling.
*/
static void cpumsf_pmu_stop(struct perf_event *event, int flags)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
if (event->hw.state & PERF_HES_STOPPED)
return;
perf_pmu_disable(event->pmu);
cpuhw->lsctl.cs = 0;
cpuhw->lsctl.cd = 0;
event->hw.state |= PERF_HES_STOPPED;
if ((flags & PERF_EF_UPDATE) && !(event->hw.state & PERF_HES_UPTODATE)) {
hw_perf_event_update(event, 1);
event->hw.state |= PERF_HES_UPTODATE;
}
perf_pmu_enable(event->pmu);
}
static int cpumsf_pmu_add(struct perf_event *event, int flags)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
struct aux_buffer *aux;
int err;
if (cpuhw->flags & PMU_F_IN_USE)
return -EAGAIN;
if (!SAMPL_DIAG_MODE(&event->hw) && !cpuhw->sfb.sdbt)
return -EINVAL;
err = 0;
perf_pmu_disable(event->pmu);
event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
/* Set up sampling controls. Always program the sampling register
* using the SDB-table start. Reset TEAR_REG event hardware register
* that is used by hw_perf_event_update() to store the sampling buffer
* position after samples have been flushed.
*/
cpuhw->lsctl.s = 0;
cpuhw->lsctl.h = 1;
cpuhw->lsctl.interval = SAMPL_RATE(&event->hw);
if (!SAMPL_DIAG_MODE(&event->hw)) {
cpuhw->lsctl.tear = (unsigned long) cpuhw->sfb.sdbt;
cpuhw->lsctl.dear = *(unsigned long *) cpuhw->sfb.sdbt;
TEAR_REG(&event->hw) = (unsigned long) cpuhw->sfb.sdbt;
}
/* Ensure sampling functions are in the disabled state. If disabled,
* switch on sampling enable control. */
if (WARN_ON_ONCE(cpuhw->lsctl.es == 1 || cpuhw->lsctl.ed == 1)) {
err = -EAGAIN;
goto out;
}
if (SAMPL_DIAG_MODE(&event->hw)) {
aux = perf_aux_output_begin(&cpuhw->handle, event);
if (!aux) {
err = -EINVAL;
goto out;
}
err = aux_output_begin(&cpuhw->handle, aux, cpuhw);
if (err)
goto out;
cpuhw->lsctl.ed = 1;
}
cpuhw->lsctl.es = 1;
/* Set in_use flag and store event */
cpuhw->event = event;
cpuhw->flags |= PMU_F_IN_USE;
if (flags & PERF_EF_START)
cpumsf_pmu_start(event, PERF_EF_RELOAD);
out:
perf_event_update_userpage(event);
perf_pmu_enable(event->pmu);
return err;
}
static void cpumsf_pmu_del(struct perf_event *event, int flags)
{
struct cpu_hw_sf *cpuhw = this_cpu_ptr(&cpu_hw_sf);
perf_pmu_disable(event->pmu);
cpumsf_pmu_stop(event, PERF_EF_UPDATE);
cpuhw->lsctl.es = 0;
cpuhw->lsctl.ed = 0;
cpuhw->flags &= ~PMU_F_IN_USE;
cpuhw->event = NULL;
if (SAMPL_DIAG_MODE(&event->hw))
aux_output_end(&cpuhw->handle);
perf_event_update_userpage(event);
perf_pmu_enable(event->pmu);
}
CPUMF_EVENT_ATTR(SF, SF_CYCLES_BASIC, PERF_EVENT_CPUM_SF);
CPUMF_EVENT_ATTR(SF, SF_CYCLES_BASIC_DIAG, PERF_EVENT_CPUM_SF_DIAG);
/* Attribute list for CPU_SF.
*
* The availablitiy depends on the CPU_MF sampling facility authorization
* for basic + diagnositic samples. This is determined at initialization
* time by the sampling facility device driver.
* If the authorization for basic samples is turned off, it should be
* also turned off for diagnostic sampling.
*
* During initialization of the device driver, check the authorization
* level for diagnostic sampling and installs the attribute
* file for diagnostic sampling if necessary.
*
* For now install a placeholder to reference all possible attributes:
* SF_CYCLES_BASIC and SF_CYCLES_BASIC_DIAG.
* Add another entry for the final NULL pointer.
*/
enum {
SF_CYCLES_BASIC_ATTR_IDX = 0,
SF_CYCLES_BASIC_DIAG_ATTR_IDX,
SF_CYCLES_ATTR_MAX
};
static struct attribute *cpumsf_pmu_events_attr[SF_CYCLES_ATTR_MAX + 1] = {
[SF_CYCLES_BASIC_ATTR_IDX] = CPUMF_EVENT_PTR(SF, SF_CYCLES_BASIC)
};
PMU_FORMAT_ATTR(event, "config:0-63");
static struct attribute *cpumsf_pmu_format_attr[] = {
&format_attr_event.attr,
NULL,
};
static struct attribute_group cpumsf_pmu_events_group = {
.name = "events",
.attrs = cpumsf_pmu_events_attr,
};
static struct attribute_group cpumsf_pmu_format_group = {
.name = "format",
.attrs = cpumsf_pmu_format_attr,
};
static const struct attribute_group *cpumsf_pmu_attr_groups[] = {
&cpumsf_pmu_events_group,
&cpumsf_pmu_format_group,
NULL,
};
static struct pmu cpumf_sampling = {
.pmu_enable = cpumsf_pmu_enable,
.pmu_disable = cpumsf_pmu_disable,
.event_init = cpumsf_pmu_event_init,
.add = cpumsf_pmu_add,
.del = cpumsf_pmu_del,
.start = cpumsf_pmu_start,
.stop = cpumsf_pmu_stop,
.read = cpumsf_pmu_read,
.attr_groups = cpumsf_pmu_attr_groups,
.setup_aux = aux_buffer_setup,
.free_aux = aux_buffer_free,
.check_period = cpumsf_pmu_check_period,
};
static void cpumf_measurement_alert(struct ext_code ext_code,
unsigned int alert, unsigned long unused)
{
struct cpu_hw_sf *cpuhw;
if (!(alert & CPU_MF_INT_SF_MASK))
return;
inc_irq_stat(IRQEXT_CMS);
cpuhw = this_cpu_ptr(&cpu_hw_sf);
/* Measurement alerts are shared and might happen when the PMU
* is not reserved. Ignore these alerts in this case. */
if (!(cpuhw->flags & PMU_F_RESERVED))
return;
/* The processing below must take care of multiple alert events that
* might be indicated concurrently. */
/* Program alert request */
if (alert & CPU_MF_INT_SF_PRA) {
if (cpuhw->flags & PMU_F_IN_USE)
if (SAMPL_DIAG_MODE(&cpuhw->event->hw))
hw_collect_aux(cpuhw);
else
hw_perf_event_update(cpuhw->event, 0);
else
WARN_ON_ONCE(!(cpuhw->flags & PMU_F_IN_USE));
}
/* Report measurement alerts only for non-PRA codes */
if (alert != CPU_MF_INT_SF_PRA)
debug_sprintf_event(sfdbg, 6, "%s: alert %#x\n", __func__,
alert);
/* Sampling authorization change request */
if (alert & CPU_MF_INT_SF_SACA)
qsi(&cpuhw->qsi);
/* Loss of sample data due to high-priority machine activities */
if (alert & CPU_MF_INT_SF_LSDA) {
pr_err("Sample data was lost\n");
cpuhw->flags |= PMU_F_ERR_LSDA;
sf_disable();
}
/* Invalid sampling buffer entry */
if (alert & (CPU_MF_INT_SF_IAE|CPU_MF_INT_SF_ISE)) {
pr_err("A sampling buffer entry is incorrect (alert=0x%x)\n",
alert);
cpuhw->flags |= PMU_F_ERR_IBE;
sf_disable();
}
}
static int cpusf_pmu_setup(unsigned int cpu, int flags)
{
/* Ignore the notification if no events are scheduled on the PMU.
* This might be racy...
*/
if (!atomic_read(&num_events))
return 0;
local_irq_disable();
setup_pmc_cpu(&flags);
local_irq_enable();
return 0;
}
static int s390_pmu_sf_online_cpu(unsigned int cpu)
{
return cpusf_pmu_setup(cpu, PMC_INIT);
}
static int s390_pmu_sf_offline_cpu(unsigned int cpu)
{
return cpusf_pmu_setup(cpu, PMC_RELEASE);
}
static int param_get_sfb_size(char *buffer, const struct kernel_param *kp)
{
if (!cpum_sf_avail())
return -ENODEV;
return sprintf(buffer, "%lu,%lu", CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB);
}
static int param_set_sfb_size(const char *val, const struct kernel_param *kp)
{
int rc;
unsigned long min, max;
if (!cpum_sf_avail())
return -ENODEV;
if (!val || !strlen(val))
return -EINVAL;
/* Valid parameter values: "min,max" or "max" */
min = CPUM_SF_MIN_SDB;
max = CPUM_SF_MAX_SDB;
if (strchr(val, ','))
rc = (sscanf(val, "%lu,%lu", &min, &max) == 2) ? 0 : -EINVAL;
else
rc = kstrtoul(val, 10, &max);
if (min < 2 || min >= max || max > get_num_physpages())
rc = -EINVAL;
if (rc)
return rc;
sfb_set_limits(min, max);
pr_info("The sampling buffer limits have changed to: "
"min %lu max %lu (diag %lu)\n",
CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB, CPUM_SF_SDB_DIAG_FACTOR);
return 0;
}
#define param_check_sfb_size(name, p) __param_check(name, p, void)
static const struct kernel_param_ops param_ops_sfb_size = {
.set = param_set_sfb_size,
.get = param_get_sfb_size,
};
#define RS_INIT_FAILURE_QSI 0x0001
#define RS_INIT_FAILURE_BSDES 0x0002
#define RS_INIT_FAILURE_ALRT 0x0003
#define RS_INIT_FAILURE_PERF 0x0004
static void __init pr_cpumsf_err(unsigned int reason)
{
pr_err("Sampling facility support for perf is not available: "
"reason %#x\n", reason);
}
static int __init init_cpum_sampling_pmu(void)
{
struct hws_qsi_info_block si;
int err;
if (!cpum_sf_avail())
return -ENODEV;
memset(&si, 0, sizeof(si));
if (qsi(&si)) {
pr_cpumsf_err(RS_INIT_FAILURE_QSI);
return -ENODEV;
}
if (!si.as && !si.ad)
return -ENODEV;
if (si.bsdes != sizeof(struct hws_basic_entry)) {
pr_cpumsf_err(RS_INIT_FAILURE_BSDES);
return -EINVAL;
}
if (si.ad) {
sfb_set_limits(CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB);
/* Sampling of diagnostic data authorized,
* install event into attribute list of PMU device.
*/
cpumsf_pmu_events_attr[SF_CYCLES_BASIC_DIAG_ATTR_IDX] =
CPUMF_EVENT_PTR(SF, SF_CYCLES_BASIC_DIAG);
}
sfdbg = debug_register(KMSG_COMPONENT, 2, 1, 80);
if (!sfdbg) {
pr_err("Registering for s390dbf failed\n");
return -ENOMEM;
}
debug_register_view(sfdbg, &debug_sprintf_view);
err = register_external_irq(EXT_IRQ_MEASURE_ALERT,
cpumf_measurement_alert);
if (err) {
pr_cpumsf_err(RS_INIT_FAILURE_ALRT);
debug_unregister(sfdbg);
goto out;
}
err = perf_pmu_register(&cpumf_sampling, "cpum_sf", PERF_TYPE_RAW);
if (err) {
pr_cpumsf_err(RS_INIT_FAILURE_PERF);
unregister_external_irq(EXT_IRQ_MEASURE_ALERT,
cpumf_measurement_alert);
debug_unregister(sfdbg);
goto out;
}
cpuhp_setup_state(CPUHP_AP_PERF_S390_SF_ONLINE, "perf/s390/sf:online",
s390_pmu_sf_online_cpu, s390_pmu_sf_offline_cpu);
out:
return err;
}
arch_initcall(init_cpum_sampling_pmu);
core_param(cpum_sfb_size, CPUM_SF_MAX_SDB, sfb_size, 0644);