d/numam-spdk
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

examples/accel_perf: don't allocate channel just to get engine caps

Browse Source 
Getting/putting the channel races with getting real IO channels
causing issues with an upcoming design. Avoid it entirely by using
the first worker channel to get the capabilties.

Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I750a8d59e294f4cda9f71f3327afc03e15342768
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/8170
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This commit is contained in:
paul luse 2021-06-03 11:54:17 -04:00 committed by Jim Harris
parent f31ad9b6f1
commit 475fadf365
1 changed files with 10 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
examples/accel/perf/accel_perf.c
View File

@ -63,7 +63,6 @@ static enum accel_capability g_workload_selection;
static struct worker_thread *g_workers = NULL;
static int g_num_workers = 0;
static pthread_mutex_t g_workers_lock = PTHREAD_MUTEX_INITIALIZER;
uint64_t g_capabilites;
struct worker_thread;
static void accel_done(void *ref, int status);
@ -731,6 +730,7 @@ _init_thread(void *arg1)
struct accel_batch *tmp;
struct accel_batch *worker_batch = NULL;
struct display_info *display = arg1;
uint64_t capabilities;
worker = calloc(1, sizeof(*worker));
if (worker == NULL) {
@ -751,6 +751,15 @@ _init_thread(void *arg1)
pthread_mutex_unlock(&g_workers_lock);
worker->ch = spdk_accel_engine_get_io_channel();
if (g_num_workers == 1) {
capabilities = spdk_accel_get_capabilities(worker->ch);
if ((capabilities & g_workload_selection) != g_workload_selection) {
g_using_sw_engine = true;
SPDK_WARNLOG("The selected workload is not natively supported by the current engine\n");
SPDK_WARNLOG("The software engine will be used instead.\n\n");
}
}
TAILQ_INIT(&worker->tasks_pool);
if (g_ops_per_batch > 0) {
@ -905,7 +914,6 @@ accel_done(void *cb_arg, int status)
static void
accel_perf_start(void *arg1)
{
struct spdk_io_channel *accel_ch;
struct spdk_cpuset tmp_cpumask = {};
char thread_name[32];
uint32_t i;
@ -913,16 +921,6 @@ accel_perf_start(void *arg1)
struct spdk_thread *thread;
struct display_info *display;
accel_ch = spdk_accel_engine_get_io_channel();
g_capabilites = spdk_accel_get_capabilities(accel_ch);
spdk_put_io_channel(accel_ch);
if ((g_capabilites & g_workload_selection) != g_workload_selection) {
g_using_sw_engine = true;
SPDK_WARNLOG("The selected workload is not natively supported by the current engine\n");
SPDK_WARNLOG("The software engine will be used instead.\n\n");
}
g_tsc_rate = spdk_get_ticks_hz();
g_tsc_end = spdk_get_ticks() + g_time_in_sec * g_tsc_rate;