bdevperf: Allocate task for each target in the beginning

This patch could prevent the potential coredump of
accessing rte_mempool among multiple threads. We
allocate the tasks from the beginning.

BTW, this patch also refactors the bdevperf to remove
dpdk dependency

Change-Id: I841e6c8b43276923acaf6686377741c37e2e71ec
Signed-off-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-on: https://review.gerrithub.io/387871
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This commit is contained in:
Ziye Yang 2017-11-20 18:22:45 +08:00 committed by Jim Harris
parent dd01e9353d
commit 8d62bec415

View File

@ -34,10 +34,6 @@
#include "spdk/stdinc.h"
#include <rte_config.h>
#include <rte_mempool.h>
#include <rte_lcore.h>
#include "spdk/bdev.h"
#include "spdk/copy_engine.h"
#include "spdk/endian.h"
@ -53,6 +49,7 @@ struct bdevperf_task {
struct io_target *target;
void *buf;
uint64_t offset_blocks;
TAILQ_ENTRY(bdevperf_task) link;
};
static int g_io_size = 0;
@ -67,6 +64,10 @@ static int g_time_in_sec;
static int g_show_performance_real_time = 0;
static bool g_run_failed = false;
static bool g_zcopy = true;
static struct spdk_mempool *task_pool;
static int g_mem_size = 0;
static int g_task_num;
static unsigned g_master_core;
static struct spdk_poller *g_perf_timer = NULL;
@ -75,24 +76,26 @@ static void bdevperf_submit_single(struct io_target *target);
#include "../common.c"
struct io_target {
char *name;
struct spdk_bdev *bdev;
struct spdk_bdev_desc *bdev_desc;
struct spdk_io_channel *ch;
struct io_target *next;
unsigned lcore;
int io_completed;
int current_queue_depth;
uint64_t size_in_ios;
uint64_t offset_in_ios;
uint64_t io_size_blocks;
bool is_draining;
struct spdk_poller *run_timer;
struct spdk_poller *reset_timer;
char *name;
struct spdk_bdev *bdev;
struct spdk_bdev_desc *bdev_desc;
struct spdk_io_channel *ch;
struct io_target *next;
unsigned lcore;
int io_completed;
int current_queue_depth;
uint64_t size_in_ios;
uint64_t offset_in_ios;
uint64_t io_size_blocks;
bool is_draining;
struct spdk_poller *run_timer;
struct spdk_poller *reset_timer;
TAILQ_HEAD(, bdevperf_task) task_list;
};
struct io_target *head[RTE_MAX_LCORE];
uint32_t coremap[RTE_MAX_LCORE];
#define SPDK_MAX_LCORE 128
struct io_target *head[SPDK_MAX_LCORE];
uint32_t coremap[SPDK_MAX_LCORE];
static int g_target_count = 0;
/*
@ -108,7 +111,7 @@ blockdev_heads_init(void)
{
uint32_t i, idx;
for (i = 0; i < RTE_MAX_LCORE; i++) {
for (i = 0; i < SPDK_MAX_LCORE; i++) {
head[i] = NULL;
}
@ -118,21 +121,43 @@ blockdev_heads_init(void)
}
}
static void
bdevperf_free_target(struct io_target *target)
{
struct bdevperf_task *task, *tmp;
TAILQ_FOREACH_SAFE(task, &target->task_list, link, tmp) {
TAILQ_REMOVE(&target->task_list, task, link);
spdk_dma_free(task->buf);
spdk_mempool_put(task_pool, task);
}
free(target->name);
free(target);
}
static void
blockdev_heads_destroy(void)
{
uint32_t i;
struct io_target *target, *next_target;
for (i = 0; i < RTE_MAX_LCORE; i++) {
for (i = 0; i < SPDK_MAX_LCORE; i++) {
target = head[i];
while (target != NULL) {
next_target = target->next;
free(target->name);
free(target);
bdevperf_free_target(target);
target = next_target;
}
}
if (!task_pool) {
if (spdk_mempool_count(task_pool) != (size_t)g_task_num) {
SPDK_ERRLOG("task_pool count is %zu but should be %d\n",
spdk_mempool_count(task_pool), g_task_num);
}
spdk_mempool_free(task_pool);
}
}
static void
@ -195,6 +220,7 @@ bdevperf_construct_targets(void)
target->is_draining = false;
target->run_timer = NULL;
target->reset_timer = NULL;
TAILQ_INIT(&target->task_list);
head[index] = target;
g_target_count++;
@ -222,8 +248,6 @@ end_run(void *arg1, void *arg2)
}
}
struct rte_mempool *task_pool;
static void
bdevperf_complete(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
{
@ -254,7 +278,7 @@ bdevperf_complete(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
target->current_queue_depth--;
target->io_completed++;
rte_mempool_put(task_pool, task);
TAILQ_INSERT_TAIL(&target->task_list, task, link);
spdk_bdev_free_io(bdev_io);
@ -267,7 +291,7 @@ bdevperf_complete(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
if (!target->is_draining) {
bdevperf_submit_single(target);
} else if (target->current_queue_depth == 0) {
complete = spdk_event_allocate(rte_get_master_lcore(), end_run, target, NULL);
complete = spdk_event_allocate(g_master_core, end_run, target, NULL);
spdk_event_call(complete);
}
}
@ -332,14 +356,6 @@ bdevperf_verify_write_complete(struct spdk_bdev_io *bdev_io, bool success,
spdk_bdev_free_io(bdev_io);
}
static void
task_ctor(struct rte_mempool *mp, void *arg, void *__task, unsigned id)
{
struct bdevperf_task *task = __task;
task->buf = spdk_dma_zmalloc(g_io_size, g_min_alignment, NULL);
}
static __thread unsigned int seed = 0;
static void
@ -355,12 +371,13 @@ bdevperf_submit_single(struct io_target *target)
desc = target->bdev_desc;
ch = target->ch;
if (rte_mempool_get(task_pool, (void **)&task) != 0 || task == NULL) {
printf("Task pool allocation failed\n");
task = TAILQ_FIRST(&target->task_list);
if (!task) {
printf("Task allocation failed\n");
abort();
}
task->target = target;
TAILQ_REMOVE(&target->task_list, task, link);
if (g_is_random) {
offset_in_ios = rand_r(&seed) % target->size_in_ios;
@ -446,7 +463,7 @@ reset_cb(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
g_run_failed = true;
}
rte_mempool_put(task_pool, task);
TAILQ_INSERT_TAIL(&target->task_list, task, link);
spdk_bdev_free_io(bdev_io);
spdk_poller_register(&target->reset_timer, reset_target, target,
@ -463,8 +480,13 @@ reset_target(void *arg)
spdk_poller_unregister(&target->reset_timer);
/* Do reset. */
rte_mempool_get(task_pool, (void **)&task);
task->target = target;
task = TAILQ_FIRST(&target->task_list);
if (!task) {
printf("Task allocation failed\n");
abort();
}
TAILQ_REMOVE(&target->task_list, task, link);
rc = spdk_bdev_reset(target->bdev_desc, target->ch,
reset_cb, task);
if (rc) {
@ -500,6 +522,7 @@ static void usage(char *program_name)
{
printf("%s options\n", program_name);
printf("\t[-c configuration file]\n");
printf("\t[-d memory size in MB]\n");
printf("\t[-m core mask for distributing I/O submission/completion work\n");
printf("\t\t(default: 0x1 - use core 0 only)]\n");
printf("\t[-q io depth]\n");
@ -554,27 +577,79 @@ performance_statistics_thread(void *arg)
performance_dump(1);
}
static void
bdevperf_run(void *arg1, void *arg2)
static int
bdevperf_construct_targets_tasks(void)
{
uint32_t i;
struct io_target *target;
struct spdk_event *event;
blockdev_heads_init();
bdevperf_construct_targets();
struct bdevperf_task *task;
int j, task_num = g_queue_depth;
/*
* Create the task pool after we have enumerated the targets, so that we know
* the min buffer alignment. Some backends such as AIO have alignment restrictions
* that must be accounted for.
*/
task_pool = rte_mempool_create("task_pool", g_target_count * g_queue_depth,
sizeof(struct bdevperf_task),
64, 0, NULL, NULL, task_ctor, NULL,
SOCKET_ID_ANY, 0);
if (g_reset) {
task_num += 1;
}
g_task_num = g_target_count * task_num;
task_pool = spdk_mempool_create("task_pool", g_task_num, sizeof(struct bdevperf_task),
64, SPDK_ENV_SOCKET_ID_ANY);
if (!task_pool) {
SPDK_ERRLOG("Cannot allocate %d tasks\n", g_target_count * g_queue_depth);
SPDK_ERRLOG("Cannot allocate %d tasks\n", g_task_num);
goto ret;
}
/* Initialize task list for each target */
for (i = 0; i < spdk_env_get_core_count(); i++) {
target = head[i];
if (!target) {
break;
}
while (target != NULL) {
for (j = 0; j < task_num; j++) {
task = spdk_mempool_get(task_pool);
if (!task) {
fprintf(stderr, "Get task from task_pool failed\n");
goto ret;
}
task->buf = spdk_dma_zmalloc(g_io_size, g_min_alignment, NULL);
if (!task->buf) {
fprintf(stderr, "Cannot allocate task->buf\n");
spdk_mempool_put(task_pool, task);
goto ret;
}
task->target = target;
TAILQ_INSERT_TAIL(&target->task_list, task, link);
}
target = target->next;
}
}
return 0;
ret:
fprintf(stderr, "Bdevperf program exits due to memory allocation issue\n");
fprintf(stderr, "Use -d XXX to allocate more huge pages, e.g., -d 4096\n");
return -1;
}
static void
bdevperf_run(void *arg1, void *arg2)
{
uint32_t i;
struct io_target *target;
struct spdk_event *event;
int rc;
blockdev_heads_init();
bdevperf_construct_targets();
rc = bdevperf_construct_targets_tasks();
if (rc) {
spdk_app_stop(1);
return;
}
@ -588,6 +663,7 @@ bdevperf_run(void *arg1, void *arg2)
1000000);
}
g_master_core = spdk_env_get_current_core();
/* Send events to start all I/O */
for (i = 0; i < spdk_env_get_core_count(); i++) {
target = head[i];
@ -619,11 +695,14 @@ main(int argc, char **argv)
mix_specified = false;
core_mask = NULL;
while ((op = getopt(argc, argv, "c:m:q:s:t:w:M:S")) != -1) {
while ((op = getopt(argc, argv, "c:d:m:q:s:t:w:M:S")) != -1) {
switch (op) {
case 'c':
config_file = optarg;
break;
case 'd':
g_mem_size = atoi(optarg);
break;
case 'm':
core_mask = optarg;
break;
@ -763,6 +842,9 @@ main(int argc, char **argv)
bdevtest_init(config_file, core_mask, &opts);
opts.rpc_addr = NULL;
if (g_mem_size) {
opts.mem_size = g_mem_size;
}
spdk_app_start(&opts, bdevperf_run, NULL, NULL);