#include <unistd.h>
#include <errno.h>
#include <libaio.h>
+#include <sys/time.h>
+#include <sys/resource.h>
#include "../fio.h"
#include "../lib/pow2.h"
#include "../optgroup.h"
+#include "../lib/memalign.h"
+#include "cmdprio.h"
-#ifndef IOCB_FLAG_HIPRI
-#define IOCB_FLAG_HIPRI (1 << 2)
-#endif
-#ifndef IOCTX_FLAG_IOPOLL
-#define IOCTX_FLAG_IOPOLL (1 << 0)
-#endif
-#ifndef IOCTX_FLAG_USERIOCB
-#define IOCTX_FLAG_USERIOCB (1 << 1)
+/* Should be defined in newest aio_abi.h */
+#ifndef IOCB_FLAG_IOPRIO
+#define IOCB_FLAG_IOPRIO (1 << 1)
#endif
+/* Hack for libaio < 0.3.111 */
+#ifndef CONFIG_LIBAIO_RW_FLAGS
+#define aio_rw_flags __pad2
+#endif
static int fio_libaio_commit(struct thread_data *td);
+static int fio_libaio_init(struct thread_data *td);
struct libaio_data {
io_context_t aio_ctx;
struct iocb **iocbs;
struct io_u **io_us;
- struct iocb *user_iocbs;
struct io_u **io_u_index;
/*
unsigned int queued;
unsigned int head;
unsigned int tail;
+
+ struct cmdprio cmdprio;
};
struct libaio_options {
- void *pad;
+ struct thread_data *td;
unsigned int userspace_reap;
- unsigned int hipri;
- unsigned int useriocb;
+ struct cmdprio_options cmdprio_options;
+ unsigned int nowait;
};
static struct fio_option options[] = {
.category = FIO_OPT_C_ENGINE,
.group = FIO_OPT_G_LIBAIO,
},
+#ifdef FIO_HAVE_IOPRIO_CLASS
{
- .name = "hipri",
- .lname = "High Priority",
- .type = FIO_OPT_STR_SET,
- .off1 = offsetof(struct libaio_options, hipri),
- .help = "Use polled IO completions",
+ .name = "cmdprio_percentage",
+ .lname = "high priority percentage",
+ .type = FIO_OPT_INT,
+ .off1 = offsetof(struct libaio_options,
+ cmdprio_options.percentage[DDIR_READ]),
+ .off2 = offsetof(struct libaio_options,
+ cmdprio_options.percentage[DDIR_WRITE]),
+ .minval = 0,
+ .maxval = 100,
+ .help = "Send high priority I/O this percentage of the time",
.category = FIO_OPT_C_ENGINE,
.group = FIO_OPT_G_LIBAIO,
},
{
- .name = "useriocb",
- .lname = "User IOCBs",
- .type = FIO_OPT_STR_SET,
- .off1 = offsetof(struct libaio_options, useriocb),
- .help = "Use user mapped IOCBs",
+ .name = "cmdprio_class",
+ .lname = "Asynchronous I/O priority class",
+ .type = FIO_OPT_INT,
+ .off1 = offsetof(struct libaio_options,
+ cmdprio_options.class[DDIR_READ]),
+ .off2 = offsetof(struct libaio_options,
+ cmdprio_options.class[DDIR_WRITE]),
+ .help = "Set asynchronous IO priority class",
+ .minval = IOPRIO_MIN_PRIO_CLASS + 1,
+ .maxval = IOPRIO_MAX_PRIO_CLASS,
+ .interval = 1,
+ .category = FIO_OPT_C_ENGINE,
+ .group = FIO_OPT_G_LIBAIO,
+ },
+ {
+ .name = "cmdprio",
+ .lname = "Asynchronous I/O priority level",
+ .type = FIO_OPT_INT,
+ .off1 = offsetof(struct libaio_options,
+ cmdprio_options.level[DDIR_READ]),
+ .off2 = offsetof(struct libaio_options,
+ cmdprio_options.level[DDIR_WRITE]),
+ .help = "Set asynchronous IO priority level",
+ .minval = IOPRIO_MIN_PRIO,
+ .maxval = IOPRIO_MAX_PRIO,
+ .interval = 1,
+ .category = FIO_OPT_C_ENGINE,
+ .group = FIO_OPT_G_LIBAIO,
+ },
+ {
+ .name = "cmdprio_bssplit",
+ .lname = "Priority percentage block size split",
+ .type = FIO_OPT_STR_STORE,
+ .off1 = offsetof(struct libaio_options,
+ cmdprio_options.bssplit_str),
+ .help = "Set priority percentages for different block sizes",
+ .category = FIO_OPT_C_ENGINE,
+ .group = FIO_OPT_G_LIBAIO,
+ },
+#else
+ {
+ .name = "cmdprio_percentage",
+ .lname = "high priority percentage",
+ .type = FIO_OPT_UNSUPPORTED,
+ .help = "Your platform does not support I/O priority classes",
+ },
+ {
+ .name = "cmdprio_class",
+ .lname = "Asynchronous I/O priority class",
+ .type = FIO_OPT_UNSUPPORTED,
+ .help = "Your platform does not support I/O priority classes",
+ },
+ {
+ .name = "cmdprio",
+ .lname = "Asynchronous I/O priority level",
+ .type = FIO_OPT_UNSUPPORTED,
+ .help = "Your platform does not support I/O priority classes",
+ },
+ {
+ .name = "cmdprio_bssplit",
+ .lname = "Priority percentage block size split",
+ .type = FIO_OPT_UNSUPPORTED,
+ .help = "Your platform does not support I/O priority classes",
+ },
+#endif
+ {
+ .name = "nowait",
+ .lname = "RWF_NOWAIT",
+ .type = FIO_OPT_BOOL,
+ .off1 = offsetof(struct libaio_options, nowait),
+ .help = "Set RWF_NOWAIT for reads/writes",
.category = FIO_OPT_C_ENGINE,
.group = FIO_OPT_G_LIBAIO,
},
*val = (*val + add) % ld->entries;
}
-static int fio_libaio_prep(struct thread_data fio_unused *td, struct io_u *io_u)
+static int fio_libaio_prep(struct thread_data *td, struct io_u *io_u)
{
- struct libaio_data *ld = td->io_ops_data;
- struct fio_file *f = io_u->file;
struct libaio_options *o = td->eo;
- struct iocb *iocb;
-
- if (o->useriocb)
- iocb = &ld->user_iocbs[io_u->index];
- else
- iocb = &io_u->iocb;
+ struct fio_file *f = io_u->file;
+ struct iocb *iocb = &io_u->iocb;
if (io_u->ddir == DDIR_READ) {
io_prep_pread(iocb, f->fd, io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
- if (o->hipri)
- iocb->u.c.flags |= IOCB_FLAG_HIPRI;
+ if (o->nowait)
+ iocb->aio_rw_flags |= RWF_NOWAIT;
} else if (io_u->ddir == DDIR_WRITE) {
io_prep_pwrite(iocb, f->fd, io_u->xfer_buf, io_u->xfer_buflen, io_u->offset);
- if (o->hipri)
- iocb->u.c.flags |= IOCB_FLAG_HIPRI;
+ if (o->nowait)
+ iocb->aio_rw_flags |= RWF_NOWAIT;
} else if (ddir_sync(io_u->ddir))
io_prep_fsync(iocb, f->fd);
return 0;
}
+static inline void fio_libaio_cmdprio_prep(struct thread_data *td,
+ struct io_u *io_u)
+{
+ struct libaio_data *ld = td->io_ops_data;
+ struct cmdprio *cmdprio = &ld->cmdprio;
+
+ if (fio_cmdprio_set_ioprio(td, cmdprio, io_u)) {
+ io_u->iocb.aio_reqprio = io_u->ioprio;
+ io_u->iocb.u.c.flags |= IOCB_FLAG_IOPRIO;
+ }
+}
+
static struct io_u *fio_libaio_event(struct thread_data *td, int event)
{
struct libaio_data *ld = td->io_ops_data;
- struct libaio_options *o = td->eo;
struct io_event *ev;
struct io_u *io_u;
ev = ld->aio_events + event;
- if (o->useriocb) {
- int index = (int) (uintptr_t) ev->obj;
- io_u = ld->io_u_index[index];
- } else
- io_u = container_of(ev->obj, struct io_u, iocb);
+ io_u = container_of(ev->obj, struct io_u, iocb);
if (ev->res != io_u->xfer_buflen) {
if (ev->res > io_u->xfer_buflen)
} else {
/* There is another completion to reap */
events[i] = ring->events[head];
- read_barrier();
- ring->head = (head + 1) % ring->nr;
+ atomic_store_release(&ring->head,
+ (head + 1) % ring->nr);
i++;
}
}
struct io_u *io_u)
{
struct libaio_data *ld = td->io_ops_data;
- struct libaio_options *o = td->eo;
fio_ro_check(td, io_u);
return FIO_Q_COMPLETED;
}
- if (o->useriocb)
- ld->iocbs[ld->head] = (struct iocb *) (uintptr_t) io_u->index;
- else
- ld->iocbs[ld->head] = &io_u->iocb;
+ if (ld->cmdprio.mode != CMDPRIO_MODE_NONE)
+ fio_libaio_cmdprio_prep(td, io_u);
+ ld->iocbs[ld->head] = &io_u->iocb;
ld->io_us[ld->head] = io_u;
ring_inc(ld, &ld->head, 1);
ld->queued++;
memcpy(&io_u->issue_time, &now, sizeof(now));
io_u_queued(td, io_u);
}
+
+ /*
+ * only used for iolog
+ */
+ if (td->o.read_iolog_file)
+ memcpy(&td->last_issue, &now, sizeof(now));
}
static int fio_libaio_commit(struct thread_data *td)
*/
if (!(td->flags & TD_F_CHILD))
io_destroy(ld->aio_ctx);
+
+ fio_cmdprio_cleanup(&ld->cmdprio);
free(ld->aio_events);
free(ld->iocbs);
free(ld->io_us);
- if (ld->user_iocbs)
- free(ld->user_iocbs);
free(ld);
}
}
-static int fio_libaio_old_queue_init(struct libaio_data *ld, unsigned int depth,
- bool hipri, bool useriocb)
+static int fio_libaio_post_init(struct thread_data *td)
{
- if (hipri) {
- log_err("fio: polled aio not available on your platform\n");
- return 1;
- }
- if (useriocb) {
- log_err("fio: user mapped iocbs not available on your platform\n");
+ struct libaio_data *ld = td->io_ops_data;
+ int err;
+
+ err = io_queue_init(td->o.iodepth, &ld->aio_ctx);
+ if (err) {
+ td_verror(td, -err, "io_queue_init");
return 1;
}
- return io_queue_init(depth, &ld->aio_ctx);
-}
-
-static int fio_libaio_queue_init(struct libaio_data *ld, unsigned int depth,
- bool hipri, bool useriocb)
-{
-#ifdef __NR_sys_io_setup2
- int ret, flags = 0;
-
- if (hipri)
- flags |= IOCTX_FLAG_IOPOLL;
- if (useriocb)
- flags |= IOCTX_FLAG_USERIOCB;
-
- ret = syscall(__NR_sys_io_setup2, depth, flags, ld->user_iocbs,
- &ld->aio_ctx);
- if (!ret)
- return 0;
-
- return fio_libaio_old_queue_init(ld, depth, hipri, useriocb);
-#else
- return fio_libaio_old_queue_init(ld, depth, hipri, useriocb);
-#endif
+ return 0;
}
static int fio_libaio_init(struct thread_data *td)
{
- struct libaio_options *o = td->eo;
struct libaio_data *ld;
- int err = 0;
+ struct libaio_options *o = td->eo;
+ int ret;
ld = calloc(1, sizeof(*ld));
- if (o->useriocb) {
- size_t size;
- void *p;
-
- ld->io_u_index = calloc(td->o.iodepth, sizeof(struct io_u *));
- size = td->o.iodepth * sizeof(struct iocb);
- if (posix_memalign(&p, page_size, size)) {
- log_err("fio: libaio iocb allocation failure\n");
- free(ld);
- return 1;
- }
- ld->user_iocbs = p;
- }
-
- /*
- * First try passing in 0 for queue depth, since we don't
- * care about the user ring. If that fails, the kernel is too old
- * and we need the right depth.
- */
- err = fio_libaio_queue_init(ld, td->o.iodepth, o->hipri, o->useriocb);
- if (err) {
- td_verror(td, -err, "io_queue_init");
- log_err("fio: check /proc/sys/fs/aio-max-nr\n");
- if (ld->user_iocbs)
- free(ld->user_iocbs);
- free(ld);
- return 1;
- }
-
ld->entries = td->o.iodepth;
ld->is_pow2 = is_power_of_2(ld->entries);
ld->aio_events = calloc(ld->entries, sizeof(struct io_event));
ld->io_us = calloc(ld->entries, sizeof(struct io_u *));
td->io_ops_data = ld;
- return 0;
-}
-static int fio_libaio_io_u_init(struct thread_data *td, struct io_u *io_u)
-{
- struct libaio_options *o = td->eo;
-
- if (o->useriocb) {
- struct libaio_data *ld = td->io_ops_data;
-
- ld->io_u_index[io_u->index] = io_u;
+ ret = fio_cmdprio_init(td, &ld->cmdprio, &o->cmdprio_options);
+ if (ret) {
+ td_verror(td, EINVAL, "fio_libaio_init");
+ return 1;
}
return 0;
}
-static struct ioengine_ops ioengine = {
+FIO_STATIC struct ioengine_ops ioengine = {
.name = "libaio",
.version = FIO_IOOPS_VERSION,
+ .flags = FIO_ASYNCIO_SYNC_TRIM |
+ FIO_ASYNCIO_SETS_ISSUE_TIME,
.init = fio_libaio_init,
- .io_u_init = fio_libaio_io_u_init,
+ .post_init = fio_libaio_post_init,
.prep = fio_libaio_prep,
.queue = fio_libaio_queue,
.commit = fio_libaio_commit,