Move some code around to better separate front/backend
authorJens Axboe <axboe@kernel.dk>
Thu, 9 Feb 2012 10:15:02 +0000 (11:15 +0100)
committerJens Axboe <axboe@kernel.dk>
Thu, 9 Feb 2012 10:15:02 +0000 (11:15 +0100)
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Makefile
backend.c [new file with mode: 0644]
fio.c
fio.h
libfio.c [new file with mode: 0644]
server.c

index 8d8802d..dc468b5 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -14,7 +14,7 @@ SOURCE = gettime.c fio.c ioengines.c init.c stat.c log.c time.c filesetup.c \
                rbtree.c smalloc.c filehash.c profile.c debug.c lib/rand.c \
                lib/num2str.c lib/ieee754.c $(wildcard crc/*.c) engines/cpu.c \
                engines/mmap.c engines/sync.c engines/null.c engines/net.c \
-               memalign.c server.c client.c iolog.c
+               memalign.c server.c client.c iolog.c backend.c libfio.c
 
 ifeq ($(UNAME), Linux)
   SOURCE += diskutil.c fifo.c blktrace.c helpers.c cgroup.c trim.c \
diff --git a/backend.c b/backend.c
new file mode 100644 (file)
index 0000000..821e978
--- /dev/null
+++ b/backend.c
@@ -0,0 +1,1638 @@
+/*
+ * fio - the flexible io tester
+ *
+ * Copyright (C) 2005 Jens Axboe <axboe@suse.de>
+ * Copyright (C) 2006-2012 Jens Axboe <axboe@kernel.dk>
+ *
+ * The license below covers all files distributed with fio unless otherwise
+ * noted in the file itself.
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2 as
+ *  published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ *
+ */
+#include <unistd.h>
+#include <fcntl.h>
+#include <string.h>
+#include <limits.h>
+#include <signal.h>
+#include <time.h>
+#include <locale.h>
+#include <assert.h>
+#include <time.h>
+#include <sys/stat.h>
+#include <sys/wait.h>
+#include <sys/ipc.h>
+#include <sys/shm.h>
+#include <sys/mman.h>
+
+#include "fio.h"
+#include "hash.h"
+#include "smalloc.h"
+#include "verify.h"
+#include "trim.h"
+#include "diskutil.h"
+#include "cgroup.h"
+#include "profile.h"
+#include "lib/rand.h"
+#include "memalign.h"
+#include "server.h"
+
+static pthread_t disk_util_thread;
+static struct fio_mutex *startup_mutex;
+static struct fio_mutex *writeout_mutex;
+static struct flist_head *cgroup_list;
+static char *cgroup_mnt;
+static int exit_value;
+static volatile int fio_abort;
+
+struct io_log *agg_io_log[2];
+
+#define PAGE_ALIGN(buf)        \
+       (char *) (((unsigned long) (buf) + page_mask) & ~page_mask)
+
+#define JOB_START_TIMEOUT      (5 * 1000)
+
+static void sig_int(int sig)
+{
+       if (threads) {
+               if (is_backend)
+                       fio_server_got_signal(sig);
+               else {
+                       log_info("\nfio: terminating on signal %d\n", sig);
+                       fflush(stdout);
+                       exit_value = 128;
+               }
+
+               fio_terminate_threads(TERMINATE_ALL);
+       }
+}
+
+static void set_sig_handlers(void)
+{
+       struct sigaction act;
+
+       memset(&act, 0, sizeof(act));
+       act.sa_handler = sig_int;
+       act.sa_flags = SA_RESTART;
+       sigaction(SIGINT, &act, NULL);
+
+       memset(&act, 0, sizeof(act));
+       act.sa_handler = sig_int;
+       act.sa_flags = SA_RESTART;
+       sigaction(SIGTERM, &act, NULL);
+
+       if (is_backend) {
+               memset(&act, 0, sizeof(act));
+               act.sa_handler = sig_int;
+               act.sa_flags = SA_RESTART;
+               sigaction(SIGPIPE, &act, NULL);
+       }
+}
+
+/*
+ * Check if we are above the minimum rate given.
+ */
+static int __check_min_rate(struct thread_data *td, struct timeval *now,
+                           enum fio_ddir ddir)
+{
+       unsigned long long bytes = 0;
+       unsigned long iops = 0;
+       unsigned long spent;
+       unsigned long rate;
+       unsigned int ratemin = 0;
+       unsigned int rate_iops = 0;
+       unsigned int rate_iops_min = 0;
+
+       assert(ddir_rw(ddir));
+
+       if (!td->o.ratemin[ddir] && !td->o.rate_iops_min[ddir])
+               return 0;
+
+       /*
+        * allow a 2 second settle period in the beginning
+        */
+       if (mtime_since(&td->start, now) < 2000)
+               return 0;
+
+       iops += td->this_io_blocks[ddir];
+       bytes += td->this_io_bytes[ddir];
+       ratemin += td->o.ratemin[ddir];
+       rate_iops += td->o.rate_iops[ddir];
+       rate_iops_min += td->o.rate_iops_min[ddir];
+
+       /*
+        * if rate blocks is set, sample is running
+        */
+       if (td->rate_bytes[ddir] || td->rate_blocks[ddir]) {
+               spent = mtime_since(&td->lastrate[ddir], now);
+               if (spent < td->o.ratecycle)
+                       return 0;
+
+               if (td->o.rate[ddir]) {
+                       /*
+                        * check bandwidth specified rate
+                        */
+                       if (bytes < td->rate_bytes[ddir]) {
+                               log_err("%s: min rate %u not met\n", td->o.name,
+                                                               ratemin);
+                               return 1;
+                       } else {
+                               rate = ((bytes - td->rate_bytes[ddir]) * 1000) / spent;
+                               if (rate < ratemin ||
+                                   bytes < td->rate_bytes[ddir]) {
+                                       log_err("%s: min rate %u not met, got"
+                                               " %luKB/sec\n", td->o.name,
+                                                       ratemin, rate);
+                                       return 1;
+                               }
+                       }
+               } else {
+                       /*
+                        * checks iops specified rate
+                        */
+                       if (iops < rate_iops) {
+                               log_err("%s: min iops rate %u not met\n",
+                                               td->o.name, rate_iops);
+                               return 1;
+                       } else {
+                               rate = ((iops - td->rate_blocks[ddir]) * 1000) / spent;
+                               if (rate < rate_iops_min ||
+                                   iops < td->rate_blocks[ddir]) {
+                                       log_err("%s: min iops rate %u not met,"
+                                               " got %lu\n", td->o.name,
+                                                       rate_iops_min, rate);
+                               }
+                       }
+               }
+       }
+
+       td->rate_bytes[ddir] = bytes;
+       td->rate_blocks[ddir] = iops;
+       memcpy(&td->lastrate[ddir], now, sizeof(*now));
+       return 0;
+}
+
+static int check_min_rate(struct thread_data *td, struct timeval *now,
+                         unsigned long *bytes_done)
+{
+       int ret = 0;
+
+       if (bytes_done[0])
+               ret |= __check_min_rate(td, now, 0);
+       if (bytes_done[1])
+               ret |= __check_min_rate(td, now, 1);
+
+       return ret;
+}
+
+/*
+ * When job exits, we can cancel the in-flight IO if we are using async
+ * io. Attempt to do so.
+ */
+static void cleanup_pending_aio(struct thread_data *td)
+{
+       struct flist_head *entry, *n;
+       struct io_u *io_u;
+       int r;
+
+       /*
+        * get immediately available events, if any
+        */
+       r = io_u_queued_complete(td, 0, NULL);
+       if (r < 0)
+               return;
+
+       /*
+        * now cancel remaining active events
+        */
+       if (td->io_ops->cancel) {
+               flist_for_each_safe(entry, n, &td->io_u_busylist) {
+                       io_u = flist_entry(entry, struct io_u, list);
+
+                       /*
+                        * if the io_u isn't in flight, then that generally
+                        * means someone leaked an io_u. complain but fix
+                        * it up, so we don't stall here.
+                        */
+                       if ((io_u->flags & IO_U_F_FLIGHT) == 0) {
+                               log_err("fio: non-busy IO on busy list\n");
+                               put_io_u(td, io_u);
+                       } else {
+                               r = td->io_ops->cancel(td, io_u);
+                               if (!r)
+                                       put_io_u(td, io_u);
+                       }
+               }
+       }
+
+       if (td->cur_depth)
+               r = io_u_queued_complete(td, td->cur_depth, NULL);
+}
+
+/*
+ * Helper to handle the final sync of a file. Works just like the normal
+ * io path, just does everything sync.
+ */
+static int fio_io_sync(struct thread_data *td, struct fio_file *f)
+{
+       struct io_u *io_u = __get_io_u(td);
+       int ret;
+
+       if (!io_u)
+               return 1;
+
+       io_u->ddir = DDIR_SYNC;
+       io_u->file = f;
+
+       if (td_io_prep(td, io_u)) {
+               put_io_u(td, io_u);
+               return 1;
+       }
+
+requeue:
+       ret = td_io_queue(td, io_u);
+       if (ret < 0) {
+               td_verror(td, io_u->error, "td_io_queue");
+               put_io_u(td, io_u);
+               return 1;
+       } else if (ret == FIO_Q_QUEUED) {
+               if (io_u_queued_complete(td, 1, NULL) < 0)
+                       return 1;
+       } else if (ret == FIO_Q_COMPLETED) {
+               if (io_u->error) {
+                       td_verror(td, io_u->error, "td_io_queue");
+                       return 1;
+               }
+
+               if (io_u_sync_complete(td, io_u, NULL) < 0)
+                       return 1;
+       } else if (ret == FIO_Q_BUSY) {
+               if (td_io_commit(td))
+                       return 1;
+               goto requeue;
+       }
+
+       return 0;
+}
+static inline void __update_tv_cache(struct thread_data *td)
+{
+       fio_gettime(&td->tv_cache, NULL);
+}
+
+static inline void update_tv_cache(struct thread_data *td)
+{
+       if ((++td->tv_cache_nr & td->tv_cache_mask) == td->tv_cache_mask)
+               __update_tv_cache(td);
+}
+
+static inline int runtime_exceeded(struct thread_data *td, struct timeval *t)
+{
+       if (in_ramp_time(td))
+               return 0;
+       if (!td->o.timeout)
+               return 0;
+       if (mtime_since(&td->epoch, t) >= td->o.timeout * 1000)
+               return 1;
+
+       return 0;
+}
+
+static int break_on_this_error(struct thread_data *td, enum fio_ddir ddir,
+                              int *retptr)
+{
+       int ret = *retptr;
+
+       if (ret < 0 || td->error) {
+               int err;
+
+               if (ret < 0)
+                       err = -ret;
+               else
+                       err = td->error;
+
+               if (!(td->o.continue_on_error & td_error_type(ddir, err)))
+                       return 1;
+
+               if (td_non_fatal_error(err)) {
+                       /*
+                        * Continue with the I/Os in case of
+                        * a non fatal error.
+                        */
+                       update_error_count(td, err);
+                       td_clear_error(td);
+                       *retptr = 0;
+                       return 0;
+               } else if (td->o.fill_device && err == ENOSPC) {
+                       /*
+                        * We expect to hit this error if
+                        * fill_device option is set.
+                        */
+                       td_clear_error(td);
+                       td->terminate = 1;
+                       return 1;
+               } else {
+                       /*
+                        * Stop the I/O in case of a fatal
+                        * error.
+                        */
+                       update_error_count(td, err);
+                       return 1;
+               }
+       }
+
+       return 0;
+}
+
+
+
+/*
+ * The main verify engine. Runs over the writes we previously submitted,
+ * reads the blocks back in, and checks the crc/md5 of the data.
+ */
+static void do_verify(struct thread_data *td)
+{
+       struct fio_file *f;
+       struct io_u *io_u;
+       int ret, min_events;
+       unsigned int i;
+
+       dprint(FD_VERIFY, "starting loop\n");
+
+       /*
+        * sync io first and invalidate cache, to make sure we really
+        * read from disk.
+        */
+       for_each_file(td, f, i) {
+               if (!fio_file_open(f))
+                       continue;
+               if (fio_io_sync(td, f))
+                       break;
+               if (file_invalidate_cache(td, f))
+                       break;
+       }
+
+       if (td->error)
+               return;
+
+       td_set_runstate(td, TD_VERIFYING);
+
+       io_u = NULL;
+       while (!td->terminate) {
+               int ret2, full;
+
+               update_tv_cache(td);
+
+               if (runtime_exceeded(td, &td->tv_cache)) {
+                       __update_tv_cache(td);
+                       if (runtime_exceeded(td, &td->tv_cache)) {
+                               td->terminate = 1;
+                               break;
+                       }
+               }
+
+               io_u = __get_io_u(td);
+               if (!io_u)
+                       break;
+
+               if (get_next_verify(td, io_u)) {
+                       put_io_u(td, io_u);
+                       break;
+               }
+
+               if (td_io_prep(td, io_u)) {
+                       put_io_u(td, io_u);
+                       break;
+               }
+
+               if (td->o.verify_async)
+                       io_u->end_io = verify_io_u_async;
+               else
+                       io_u->end_io = verify_io_u;
+
+               ret = td_io_queue(td, io_u);
+               switch (ret) {
+               case FIO_Q_COMPLETED:
+                       if (io_u->error) {
+                               ret = -io_u->error;
+                               clear_io_u(td, io_u);
+                       } else if (io_u->resid) {
+                               int bytes = io_u->xfer_buflen - io_u->resid;
+
+                               /*
+                                * zero read, fail
+                                */
+                               if (!bytes) {
+                                       td_verror(td, EIO, "full resid");
+                                       put_io_u(td, io_u);
+                                       break;
+                               }
+
+                               io_u->xfer_buflen = io_u->resid;
+                               io_u->xfer_buf += bytes;
+                               io_u->offset += bytes;
+
+                               if (ddir_rw(io_u->ddir))
+                                       td->ts.short_io_u[io_u->ddir]++;
+
+                               f = io_u->file;
+                               if (io_u->offset == f->real_file_size)
+                                       goto sync_done;
+
+                               requeue_io_u(td, &io_u);
+                       } else {
+sync_done:
+                               ret = io_u_sync_complete(td, io_u, NULL);
+                               if (ret < 0)
+                                       break;
+                       }
+                       continue;
+               case FIO_Q_QUEUED:
+                       break;
+               case FIO_Q_BUSY:
+                       requeue_io_u(td, &io_u);
+                       ret2 = td_io_commit(td);
+                       if (ret2 < 0)
+                               ret = ret2;
+                       break;
+               default:
+                       assert(ret < 0);
+                       td_verror(td, -ret, "td_io_queue");
+                       break;
+               }
+
+               if (break_on_this_error(td, io_u->ddir, &ret))
+                       break;
+
+               /*
+                * if we can queue more, do so. but check if there are
+                * completed io_u's first. Note that we can get BUSY even
+                * without IO queued, if the system is resource starved.
+                */
+               full = queue_full(td) || (ret == FIO_Q_BUSY && td->cur_depth);
+               if (full || !td->o.iodepth_batch_complete) {
+                       min_events = min(td->o.iodepth_batch_complete,
+                                        td->cur_depth);
+                       if (full && !min_events && td->o.iodepth_batch_complete != 0)
+                               min_events = 1;
+
+                       do {
+                               /*
+                                * Reap required number of io units, if any,
+                                * and do the verification on them through
+                                * the callback handler
+                                */
+                               if (io_u_queued_complete(td, min_events, NULL) < 0) {
+                                       ret = -1;
+                                       break;
+                               }
+                       } while (full && (td->cur_depth > td->o.iodepth_low));
+               }
+               if (ret < 0)
+                       break;
+       }
+
+       if (!td->error) {
+               min_events = td->cur_depth;
+
+               if (min_events)
+                       ret = io_u_queued_complete(td, min_events, NULL);
+       } else
+               cleanup_pending_aio(td);
+
+       td_set_runstate(td, TD_RUNNING);
+
+       dprint(FD_VERIFY, "exiting loop\n");
+}
+
+/*
+ * Main IO worker function. It retrieves io_u's to process and queues
+ * and reaps them, checking for rate and errors along the way.
+ */
+static void do_io(struct thread_data *td)
+{
+       unsigned int i;
+       int ret = 0;
+
+       if (in_ramp_time(td))
+               td_set_runstate(td, TD_RAMP);
+       else
+               td_set_runstate(td, TD_RUNNING);
+
+       while ( (td->o.read_iolog_file && !flist_empty(&td->io_log_list)) ||
+               (!flist_empty(&td->trim_list)) ||
+               ((td->this_io_bytes[0] + td->this_io_bytes[1]) < td->o.size) ) {
+               struct timeval comp_time;
+               unsigned long bytes_done[2] = { 0, 0 };
+               int min_evts = 0;
+               struct io_u *io_u;
+               int ret2, full;
+               enum fio_ddir ddir;
+
+               if (td->terminate)
+                       break;
+
+               update_tv_cache(td);
+
+               if (runtime_exceeded(td, &td->tv_cache)) {
+                       __update_tv_cache(td);
+                       if (runtime_exceeded(td, &td->tv_cache)) {
+                               td->terminate = 1;
+                               break;
+                       }
+               }
+
+               io_u = get_io_u(td);
+               if (!io_u)
+                       break;
+
+               ddir = io_u->ddir;
+
+               /*
+                * Add verification end_io handler, if asked to verify
+                * a previously written file.
+                */
+               if (td->o.verify != VERIFY_NONE && io_u->ddir == DDIR_READ &&
+                   !td_rw(td)) {
+                       if (td->o.verify_async)
+                               io_u->end_io = verify_io_u_async;
+                       else
+                               io_u->end_io = verify_io_u;
+                       td_set_runstate(td, TD_VERIFYING);
+               } else if (in_ramp_time(td))
+                       td_set_runstate(td, TD_RAMP);
+               else
+                       td_set_runstate(td, TD_RUNNING);
+
+               ret = td_io_queue(td, io_u);
+               switch (ret) {
+               case FIO_Q_COMPLETED:
+                       if (io_u->error) {
+                               ret = -io_u->error;
+                               clear_io_u(td, io_u);
+                       } else if (io_u->resid) {
+                               int bytes = io_u->xfer_buflen - io_u->resid;
+                               struct fio_file *f = io_u->file;
+
+                               /*
+                                * zero read, fail
+                                */
+                               if (!bytes) {
+                                       td_verror(td, EIO, "full resid");
+                                       put_io_u(td, io_u);
+                                       break;
+                               }
+
+                               io_u->xfer_buflen = io_u->resid;
+                               io_u->xfer_buf += bytes;
+                               io_u->offset += bytes;
+
+                               if (ddir_rw(io_u->ddir))
+                                       td->ts.short_io_u[io_u->ddir]++;
+
+                               if (io_u->offset == f->real_file_size)
+                                       goto sync_done;
+
+                               requeue_io_u(td, &io_u);
+                       } else {
+sync_done:
+                               if (__should_check_rate(td, 0) ||
+                                   __should_check_rate(td, 1))
+                                       fio_gettime(&comp_time, NULL);
+
+                               ret = io_u_sync_complete(td, io_u, bytes_done);
+                               if (ret < 0)
+                                       break;
+                       }
+                       break;
+               case FIO_Q_QUEUED:
+                       /*
+                        * if the engine doesn't have a commit hook,
+                        * the io_u is really queued. if it does have such
+                        * a hook, it has to call io_u_queued() itself.
+                        */
+                       if (td->io_ops->commit == NULL)
+                               io_u_queued(td, io_u);
+                       break;
+               case FIO_Q_BUSY:
+                       requeue_io_u(td, &io_u);
+                       ret2 = td_io_commit(td);
+                       if (ret2 < 0)
+                               ret = ret2;
+                       break;
+               default:
+                       assert(ret < 0);
+                       put_io_u(td, io_u);
+                       break;
+               }
+
+               if (break_on_this_error(td, ddir, &ret))
+                       break;
+
+               /*
+                * See if we need to complete some commands. Note that we
+                * can get BUSY even without IO queued, if the system is
+                * resource starved.
+                */
+               full = queue_full(td) || (ret == FIO_Q_BUSY && td->cur_depth);
+               if (full || !td->o.iodepth_batch_complete) {
+                       min_evts = min(td->o.iodepth_batch_complete,
+                                       td->cur_depth);
+                       if (full && !min_evts && td->o.iodepth_batch_complete != 0)
+                               min_evts = 1;
+
+                       if (__should_check_rate(td, 0) ||
+                           __should_check_rate(td, 1))
+                               fio_gettime(&comp_time, NULL);
+
+                       do {
+                               ret = io_u_queued_complete(td, min_evts, bytes_done);
+                               if (ret < 0)
+                                       break;
+
+                       } while (full && (td->cur_depth > td->o.iodepth_low));
+               }
+
+               if (ret < 0)
+                       break;
+               if (!(bytes_done[0] + bytes_done[1]))
+                       continue;
+
+               if (!in_ramp_time(td) && should_check_rate(td, bytes_done)) {
+                       if (check_min_rate(td, &comp_time, bytes_done)) {
+                               if (exitall_on_terminate)
+                                       fio_terminate_threads(td->groupid);
+                               td_verror(td, EIO, "check_min_rate");
+                               break;
+                       }
+               }
+
+               if (td->o.thinktime) {
+                       unsigned long long b;
+
+                       b = td->io_blocks[0] + td->io_blocks[1];
+                       if (!(b % td->o.thinktime_blocks)) {
+                               int left;
+
+                               if (td->o.thinktime_spin)
+                                       usec_spin(td->o.thinktime_spin);
+
+                               left = td->o.thinktime - td->o.thinktime_spin;
+                               if (left)
+                                       usec_sleep(td, left);
+                       }
+               }
+       }
+
+       if (td->trim_entries)
+               log_err("fio: %d trim entries leaked?\n", td->trim_entries);
+
+       if (td->o.fill_device && td->error == ENOSPC) {
+               td->error = 0;
+               td->terminate = 1;
+       }
+       if (!td->error) {
+               struct fio_file *f;
+
+               i = td->cur_depth;
+               if (i) {
+                       ret = io_u_queued_complete(td, i, NULL);
+                       if (td->o.fill_device && td->error == ENOSPC)
+                               td->error = 0;
+               }
+
+               if (should_fsync(td) && td->o.end_fsync) {
+                       td_set_runstate(td, TD_FSYNCING);
+
+                       for_each_file(td, f, i) {
+                               if (!fio_file_open(f))
+                                       continue;
+                               fio_io_sync(td, f);
+                       }
+               }
+       } else
+               cleanup_pending_aio(td);
+
+       /*
+        * stop job if we failed doing any IO
+        */
+       if ((td->this_io_bytes[0] + td->this_io_bytes[1]) == 0)
+               td->done = 1;
+}
+
+static void cleanup_io_u(struct thread_data *td)
+{
+       struct flist_head *entry, *n;
+       struct io_u *io_u;
+
+       flist_for_each_safe(entry, n, &td->io_u_freelist) {
+               io_u = flist_entry(entry, struct io_u, list);
+
+               flist_del(&io_u->list);
+               fio_memfree(io_u, sizeof(*io_u));
+       }
+
+       free_io_mem(td);
+}
+
+static int init_io_u(struct thread_data *td)
+{
+       struct io_u *io_u;
+       unsigned int max_bs;
+       int cl_align, i, max_units;
+       char *p;
+
+       max_units = td->o.iodepth;
+       max_bs = max(td->o.max_bs[DDIR_READ], td->o.max_bs[DDIR_WRITE]);
+       td->orig_buffer_size = (unsigned long long) max_bs
+                                       * (unsigned long long) max_units;
+
+       if (td->o.mem_type == MEM_SHMHUGE || td->o.mem_type == MEM_MMAPHUGE) {
+               unsigned long bs;
+
+               bs = td->orig_buffer_size + td->o.hugepage_size - 1;
+               td->orig_buffer_size = bs & ~(td->o.hugepage_size - 1);
+       }
+
+       if (td->orig_buffer_size != (size_t) td->orig_buffer_size) {
+               log_err("fio: IO memory too large. Reduce max_bs or iodepth\n");
+               return 1;
+       }
+
+       if (allocate_io_mem(td))
+               return 1;
+
+       if (td->o.odirect || td->o.mem_align ||
+           (td->io_ops->flags & FIO_RAWIO))
+               p = PAGE_ALIGN(td->orig_buffer) + td->o.mem_align;
+       else
+               p = td->orig_buffer;
+
+       cl_align = os_cache_line_size();
+
+       for (i = 0; i < max_units; i++) {
+               void *ptr;
+
+               if (td->terminate)
+                       return 1;
+
+               ptr = fio_memalign(cl_align, sizeof(*io_u));
+               if (!ptr) {
+                       log_err("fio: unable to allocate aligned memory\n");
+                       break;
+               }
+
+               io_u = ptr;
+               memset(io_u, 0, sizeof(*io_u));
+               INIT_FLIST_HEAD(&io_u->list);
+               dprint(FD_MEM, "io_u alloc %p, index %u\n", io_u, i);
+
+               if (!(td->io_ops->flags & FIO_NOIO)) {
+                       io_u->buf = p;
+                       dprint(FD_MEM, "io_u %p, mem %p\n", io_u, io_u->buf);
+
+                       if (td_write(td))
+                               io_u_fill_buffer(td, io_u, max_bs);
+                       if (td_write(td) && td->o.verify_pattern_bytes) {
+                               /*
+                                * Fill the buffer with the pattern if we are
+                                * going to be doing writes.
+                                */
+                               fill_pattern(td, io_u->buf, max_bs, io_u, 0, 0);
+                       }
+               }
+
+               io_u->index = i;
+               io_u->flags = IO_U_F_FREE;
+               flist_add(&io_u->list, &td->io_u_freelist);
+               p += max_bs;
+       }
+
+       return 0;
+}
+
+static int switch_ioscheduler(struct thread_data *td)
+{
+       char tmp[256], tmp2[128];
+       FILE *f;
+       int ret;
+
+       if (td->io_ops->flags & FIO_DISKLESSIO)
+               return 0;
+
+       sprintf(tmp, "%s/queue/scheduler", td->sysfs_root);
+
+       f = fopen(tmp, "r+");
+       if (!f) {
+               if (errno == ENOENT) {
+                       log_err("fio: os or kernel doesn't support IO scheduler"
+                               " switching\n");
+                       return 0;
+               }
+               td_verror(td, errno, "fopen iosched");
+               return 1;
+       }
+
+       /*
+        * Set io scheduler.
+        */
+       ret = fwrite(td->o.ioscheduler, strlen(td->o.ioscheduler), 1, f);
+       if (ferror(f) || ret != 1) {
+               td_verror(td, errno, "fwrite");
+               fclose(f);
+               return 1;
+       }
+
+       rewind(f);
+
+       /*
+        * Read back and check that the selected scheduler is now the default.
+        */
+       ret = fread(tmp, 1, sizeof(tmp), f);
+       if (ferror(f) || ret < 0) {
+               td_verror(td, errno, "fread");
+               fclose(f);
+               return 1;
+       }
+
+       sprintf(tmp2, "[%s]", td->o.ioscheduler);
+       if (!strstr(tmp, tmp2)) {
+               log_err("fio: io scheduler %s not found\n", td->o.ioscheduler);
+               td_verror(td, EINVAL, "iosched_switch");
+               fclose(f);
+               return 1;
+       }
+
+       fclose(f);
+       return 0;
+}
+
+static int keep_running(struct thread_data *td)
+{
+       unsigned long long io_done;
+
+       if (td->done)
+               return 0;
+       if (td->o.time_based)
+               return 1;
+       if (td->o.loops) {
+               td->o.loops--;
+               return 1;
+       }
+
+       io_done = td->io_bytes[DDIR_READ] + td->io_bytes[DDIR_WRITE]
+                       + td->io_skip_bytes;
+       if (io_done < td->o.size)
+               return 1;
+
+       return 0;
+}
+
+static int exec_string(const char *string)
+{
+       int ret, newlen = strlen(string) + 1 + 8;
+       char *str;
+
+       str = malloc(newlen);
+       sprintf(str, "sh -c %s", string);
+
+       ret = system(str);
+       if (ret == -1)
+               log_err("fio: exec of cmd <%s> failed\n", str);
+
+       free(str);
+       return ret;
+}
+
+/*
+ * Entry point for the thread based jobs. The process based jobs end up
+ * here as well, after a little setup.
+ */
+static void *thread_main(void *data)
+{
+       unsigned long long elapsed;
+       struct thread_data *td = data;
+       pthread_condattr_t attr;
+       int clear_state;
+
+       if (!td->o.use_thread) {
+               setsid();
+               td->pid = getpid();
+       } else
+               td->pid = gettid();
+
+       dprint(FD_PROCESS, "jobs pid=%d started\n", (int) td->pid);
+
+       INIT_FLIST_HEAD(&td->io_u_freelist);
+       INIT_FLIST_HEAD(&td->io_u_busylist);
+       INIT_FLIST_HEAD(&td->io_u_requeues);
+       INIT_FLIST_HEAD(&td->io_log_list);
+       INIT_FLIST_HEAD(&td->io_hist_list);
+       INIT_FLIST_HEAD(&td->verify_list);
+       INIT_FLIST_HEAD(&td->trim_list);
+       pthread_mutex_init(&td->io_u_lock, NULL);
+       td->io_hist_tree = RB_ROOT;
+
+       pthread_condattr_init(&attr);
+       pthread_cond_init(&td->verify_cond, &attr);
+       pthread_cond_init(&td->free_cond, &attr);
+
+       td_set_runstate(td, TD_INITIALIZED);
+       dprint(FD_MUTEX, "up startup_mutex\n");
+       fio_mutex_up(startup_mutex);
+       dprint(FD_MUTEX, "wait on td->mutex\n");
+       fio_mutex_down(td->mutex);
+       dprint(FD_MUTEX, "done waiting on td->mutex\n");
+
+       /*
+        * the ->mutex mutex is now no longer used, close it to avoid
+        * eating a file descriptor
+        */
+       fio_mutex_remove(td->mutex);
+
+       /*
+        * A new gid requires privilege, so we need to do this before setting
+        * the uid.
+        */
+       if (td->o.gid != -1U && setgid(td->o.gid)) {
+               td_verror(td, errno, "setgid");
+               goto err;
+       }
+       if (td->o.uid != -1U && setuid(td->o.uid)) {
+               td_verror(td, errno, "setuid");
+               goto err;
+       }
+
+       /*
+        * If we have a gettimeofday() thread, make sure we exclude that
+        * thread from this job
+        */
+       if (td->o.gtod_cpu)
+               fio_cpu_clear(&td->o.cpumask, td->o.gtod_cpu);
+
+       /*
+        * Set affinity first, in case it has an impact on the memory
+        * allocations.
+        */
+       if (td->o.cpumask_set && fio_setaffinity(td->pid, td->o.cpumask) == -1) {
+               td_verror(td, errno, "cpu_set_affinity");
+               goto err;
+       }
+
+       /*
+        * May alter parameters that init_io_u() will use, so we need to
+        * do this first.
+        */
+       if (init_iolog(td))
+               goto err;
+
+       if (init_io_u(td))
+               goto err;
+
+       if (td->o.verify_async && verify_async_init(td))
+               goto err;
+
+       if (td->ioprio_set) {
+               if (ioprio_set(IOPRIO_WHO_PROCESS, 0, td->ioprio) == -1) {
+                       td_verror(td, errno, "ioprio_set");
+                       goto err;
+               }
+       }
+
+       if (td->o.cgroup_weight && cgroup_setup(td, cgroup_list, &cgroup_mnt))
+               goto err;
+
+       if (nice(td->o.nice) == -1) {
+               td_verror(td, errno, "nice");
+               goto err;
+       }
+
+       if (td->o.ioscheduler && switch_ioscheduler(td))
+               goto err;
+
+       if (!td->o.create_serialize && setup_files(td))
+               goto err;
+
+       if (td_io_init(td))
+               goto err;
+
+       if (init_random_map(td))
+               goto err;
+
+       if (td->o.exec_prerun) {
+               if (exec_string(td->o.exec_prerun))
+                       goto err;
+       }
+
+       if (td->o.pre_read) {
+               if (pre_read_files(td) < 0)
+                       goto err;
+       }
+
+       fio_gettime(&td->epoch, NULL);
+       getrusage(RUSAGE_SELF, &td->ru_start);
+
+       clear_state = 0;
+       while (keep_running(td)) {
+               fio_gettime(&td->start, NULL);
+               memcpy(&td->bw_sample_time, &td->start, sizeof(td->start));
+               memcpy(&td->iops_sample_time, &td->start, sizeof(td->start));
+               memcpy(&td->tv_cache, &td->start, sizeof(td->start));
+
+               if (td->o.ratemin[0] || td->o.ratemin[1]) {
+                       memcpy(&td->lastrate[0], &td->bw_sample_time,
+                                               sizeof(td->bw_sample_time));
+                       memcpy(&td->lastrate[1], &td->bw_sample_time,
+                                               sizeof(td->bw_sample_time));
+               }
+
+               if (clear_state)
+                       clear_io_state(td);
+
+               prune_io_piece_log(td);
+
+               do_io(td);
+
+               clear_state = 1;
+
+               if (td_read(td) && td->io_bytes[DDIR_READ]) {
+                       elapsed = utime_since_now(&td->start);
+                       td->ts.runtime[DDIR_READ] += elapsed;
+               }
+               if (td_write(td) && td->io_bytes[DDIR_WRITE]) {
+                       elapsed = utime_since_now(&td->start);
+                       td->ts.runtime[DDIR_WRITE] += elapsed;
+               }
+
+               if (td->error || td->terminate)
+                       break;
+
+               if (!td->o.do_verify ||
+                   td->o.verify == VERIFY_NONE ||
+                   (td->io_ops->flags & FIO_UNIDIR))
+                       continue;
+
+               clear_io_state(td);
+
+               fio_gettime(&td->start, NULL);
+
+               do_verify(td);
+
+               td->ts.runtime[DDIR_READ] += utime_since_now(&td->start);
+
+               if (td->error || td->terminate)
+                       break;
+       }
+
+       update_rusage_stat(td);
+       td->ts.runtime[0] = (td->ts.runtime[0] + 999) / 1000;
+       td->ts.runtime[1] = (td->ts.runtime[1] + 999) / 1000;
+       td->ts.total_run_time = mtime_since_now(&td->epoch);
+       td->ts.io_bytes[0] = td->io_bytes[0];
+       td->ts.io_bytes[1] = td->io_bytes[1];
+
+       fio_mutex_down(writeout_mutex);
+       if (td->bw_log) {
+               if (td->o.bw_log_file) {
+                       finish_log_named(td, td->bw_log,
+                                               td->o.bw_log_file, "bw");
+               } else
+                       finish_log(td, td->bw_log, "bw");
+       }
+       if (td->lat_log) {
+               if (td->o.lat_log_file) {
+                       finish_log_named(td, td->lat_log,
+                                               td->o.lat_log_file, "lat");
+               } else
+                       finish_log(td, td->lat_log, "lat");
+       }
+       if (td->slat_log) {
+               if (td->o.lat_log_file) {
+                       finish_log_named(td, td->slat_log,
+                                               td->o.lat_log_file, "slat");
+               } else
+                       finish_log(td, td->slat_log, "slat");
+       }
+       if (td->clat_log) {
+               if (td->o.lat_log_file) {
+                       finish_log_named(td, td->clat_log,
+                                               td->o.lat_log_file, "clat");
+               } else
+                       finish_log(td, td->clat_log, "clat");
+       }
+       if (td->iops_log) {
+               if (td->o.iops_log_file) {
+                       finish_log_named(td, td->iops_log,
+                                               td->o.iops_log_file, "iops");
+               } else
+                       finish_log(td, td->iops_log, "iops");
+       }
+
+       fio_mutex_up(writeout_mutex);
+       if (td->o.exec_postrun)
+               exec_string(td->o.exec_postrun);
+
+       if (exitall_on_terminate)
+               fio_terminate_threads(td->groupid);
+
+err:
+       if (td->error)
+               log_info("fio: pid=%d, err=%d/%s\n", (int) td->pid, td->error,
+                                                       td->verror);
+
+       if (td->o.verify_async)
+               verify_async_exit(td);
+
+       close_and_free_files(td);
+       close_ioengine(td);
+       cleanup_io_u(td);
+       cgroup_shutdown(td, &cgroup_mnt);
+
+       if (td->o.cpumask_set) {
+               int ret = fio_cpuset_exit(&td->o.cpumask);
+
+               td_verror(td, ret, "fio_cpuset_exit");
+       }
+
+       /*
+        * do this very late, it will log file closing as well
+        */
+       if (td->o.write_iolog_file)
+               write_iolog_close(td);
+
+       td_set_runstate(td, TD_EXITED);
+       return (void *) (unsigned long) td->error;
+}
+
+
+/*
+ * We cannot pass the td data into a forked process, so attach the td and
+ * pass it to the thread worker.
+ */
+static int fork_main(int shmid, int offset)
+{
+       struct thread_data *td;
+       void *data, *ret;
+
+#ifndef __hpux
+       data = shmat(shmid, NULL, 0);
+       if (data == (void *) -1) {
+               int __err = errno;
+
+               perror("shmat");
+               return __err;
+       }
+#else
+       /*
+        * HP-UX inherits shm mappings?
+        */
+       data = threads;
+#endif
+
+       td = data + offset * sizeof(struct thread_data);
+       ret = thread_main(td);
+       shmdt(data);
+       return (int) (unsigned long) ret;
+}
+
+/*
+ * Run over the job map and reap the threads that have exited, if any.
+ */
+static void reap_threads(unsigned int *nr_running, unsigned int *t_rate,
+                        unsigned int *m_rate)
+{
+       struct thread_data *td;
+       unsigned int cputhreads, realthreads, pending;
+       int i, status, ret;
+
+       /*
+        * reap exited threads (TD_EXITED -> TD_REAPED)
+        */
+       realthreads = pending = cputhreads = 0;
+       for_each_td(td, i) {
+               int flags = 0;
+
+               /*
+                * ->io_ops is NULL for a thread that has closed its
+                * io engine
+                */
+               if (td->io_ops && !strcmp(td->io_ops->name, "cpuio"))
+                       cputhreads++;
+               else
+                       realthreads++;
+
+               if (!td->pid) {
+                       pending++;
+                       continue;
+               }
+               if (td->runstate == TD_REAPED)
+                       continue;
+               if (td->o.use_thread) {
+                       if (td->runstate == TD_EXITED) {
+                               td_set_runstate(td, TD_REAPED);
+                               goto reaped;
+                       }
+                       continue;
+               }
+
+               flags = WNOHANG;
+               if (td->runstate == TD_EXITED)
+                       flags = 0;
+
+               /*
+                * check if someone quit or got killed in an unusual way
+                */
+               ret = waitpid(td->pid, &status, flags);
+               if (ret < 0) {
+                       if (errno == ECHILD) {
+                               log_err("fio: pid=%d disappeared %d\n",
+                                               (int) td->pid, td->runstate);
+                               td_set_runstate(td, TD_REAPED);
+                               goto reaped;
+                       }
+                       perror("waitpid");
+               } else if (ret == td->pid) {
+                       if (WIFSIGNALED(status)) {
+                               int sig = WTERMSIG(status);
+
+                               if (sig != SIGTERM)
+                                       log_err("fio: pid=%d, got signal=%d\n",
+                                                       (int) td->pid, sig);
+                               td_set_runstate(td, TD_REAPED);
+                               goto reaped;
+                       }
+                       if (WIFEXITED(status)) {
+                               if (WEXITSTATUS(status) && !td->error)
+                                       td->error = WEXITSTATUS(status);
+
+                               td_set_runstate(td, TD_REAPED);
+                               goto reaped;
+                       }
+               }
+
+               /*
+                * thread is not dead, continue
+                */
+               pending++;
+               continue;
+reaped:
+               (*nr_running)--;
+               (*m_rate) -= (td->o.ratemin[0] + td->o.ratemin[1]);
+               (*t_rate) -= (td->o.rate[0] + td->o.rate[1]);
+               if (!td->pid)
+                       pending--;
+
+               if (td->error)
+                       exit_value++;
+
+               done_secs += mtime_since_now(&td->epoch) / 1000;
+       }
+
+       if (*nr_running == cputhreads && !pending && realthreads)
+               fio_terminate_threads(TERMINATE_ALL);
+}
+
+
+
+/*
+ * Main function for kicking off and reaping jobs, as needed.
+ */
+static void run_threads(void)
+{
+       struct thread_data *td;
+       unsigned long spent;
+       unsigned int i, todo, nr_running, m_rate, t_rate, nr_started;
+
+       if (fio_pin_memory())
+               return;
+
+       if (fio_gtod_offload && fio_start_gtod_thread())
+               return;
+
+       set_sig_handlers();
+
+       if (!terse_output) {
+               log_info("Starting ");
+               if (nr_thread)
+                       log_info("%d thread%s", nr_thread,
+                                               nr_thread > 1 ? "s" : "");
+               if (nr_process) {
+                       if (nr_thread)
+                               log_info(" and ");
+                       log_info("%d process%s", nr_process,
+                                               nr_process > 1 ? "es" : "");
+               }
+               log_info("\n");
+               fflush(stdout);
+       }
+
+       todo = thread_number;
+       nr_running = 0;
+       nr_started = 0;
+       m_rate = t_rate = 0;
+
+       for_each_td(td, i) {
+               print_status_init(td->thread_number - 1);
+
+               if (!td->o.create_serialize)
+                       continue;
+
+               /*
+                * do file setup here so it happens sequentially,
+                * we don't want X number of threads getting their
+                * client data interspersed on disk
+                */
+               if (setup_files(td)) {
+                       exit_value++;
+                       if (td->error)
+                               log_err("fio: pid=%d, err=%d/%s\n",
+                                       (int) td->pid, td->error, td->verror);
+                       td_set_runstate(td, TD_REAPED);
+                       todo--;
+               } else {
+                       struct fio_file *f;
+                       unsigned int j;
+
+                       /*
+                        * for sharing to work, each job must always open
+                        * its own files. so close them, if we opened them
+                        * for creation
+                        */
+                       for_each_file(td, f, j) {
+                               if (fio_file_open(f))
+                                       td_io_close_file(td, f);
+                       }
+               }
+       }
+
+       set_genesis_time();
+
+       while (todo) {
+               struct thread_data *map[REAL_MAX_JOBS];
+               struct timeval this_start;
+               int this_jobs = 0, left;
+
+               /*
+                * create threads (TD_NOT_CREATED -> TD_CREATED)
+                */
+               for_each_td(td, i) {
+                       if (td->runstate != TD_NOT_CREATED)
+                               continue;
+
+                       /*
+                        * never got a chance to start, killed by other
+                        * thread for some reason
+                        */
+                       if (td->terminate) {
+                               todo--;
+                               continue;
+                       }
+
+                       if (td->o.start_delay) {
+                               spent = mtime_since_genesis();
+
+                               if (td->o.start_delay * 1000 > spent)
+                                       continue;
+                       }
+
+                       if (td->o.stonewall && (nr_started || nr_running)) {
+                               dprint(FD_PROCESS, "%s: stonewall wait\n",
+                                                       td->o.name);
+                               break;
+                       }
+
+                       init_disk_util(td);
+
+                       /*
+                        * Set state to created. Thread will transition
+                        * to TD_INITIALIZED when it's done setting up.
+                        */
+                       td_set_runstate(td, TD_CREATED);
+                       map[this_jobs++] = td;
+                       nr_started++;
+
+                       if (td->o.use_thread) {
+                               int ret;
+
+                               dprint(FD_PROCESS, "will pthread_create\n");
+                               ret = pthread_create(&td->thread, NULL,
+                                                       thread_main, td);
+                               if (ret) {
+                                       log_err("pthread_create: %s\n",
+                                                       strerror(ret));
+                                       nr_started--;
+                                       break;
+                               }
+                               ret = pthread_detach(td->thread);
+                               if (ret)
+                                       log_err("pthread_detach: %s",
+                                                       strerror(ret));
+                       } else {
+                               pid_t pid;
+                               dprint(FD_PROCESS, "will fork\n");
+                               pid = fork();
+                               if (!pid) {
+                                       int ret = fork_main(shm_id, i);
+
+                                       _exit(ret);
+                               } else if (i == fio_debug_jobno)
+                                       *fio_debug_jobp = pid;
+                       }
+                       dprint(FD_MUTEX, "wait on startup_mutex\n");
+                       if (fio_mutex_down_timeout(startup_mutex, 10)) {
+                               log_err("fio: job startup hung? exiting.\n");
+                               fio_terminate_threads(TERMINATE_ALL);
+                               fio_abort = 1;
+                               nr_started--;
+                               break;
+                       }
+                       dprint(FD_MUTEX, "done waiting on startup_mutex\n");
+               }
+
+               /*
+                * Wait for the started threads to transition to
+                * TD_INITIALIZED.
+                */
+               fio_gettime(&this_start, NULL);
+               left = this_jobs;
+               while (left && !fio_abort) {
+                       if (mtime_since_now(&this_start) > JOB_START_TIMEOUT)
+                               break;
+
+                       usleep(100000);
+
+                       for (i = 0; i < this_jobs; i++) {
+                               td = map[i];
+                               if (!td)
+                                       continue;
+                               if (td->runstate == TD_INITIALIZED) {
+                                       map[i] = NULL;
+                                       left--;
+                               } else if (td->runstate >= TD_EXITED) {
+                                       map[i] = NULL;
+                                       left--;
+                                       todo--;
+                                       nr_running++; /* work-around... */
+                               }
+                       }
+               }
+
+               if (left) {
+                       log_err("fio: %d jobs failed to start\n", left);
+                       for (i = 0; i < this_jobs; i++) {
+                               td = map[i];
+                               if (!td)
+                                       continue;
+                               kill(td->pid, SIGTERM);
+                       }
+                       break;
+               }
+
+               /*
+                * start created threads (TD_INITIALIZED -> TD_RUNNING).
+                */
+               for_each_td(td, i) {
+                       if (td->runstate != TD_INITIALIZED)
+                               continue;
+
+                       if (in_ramp_time(td))
+                               td_set_runstate(td, TD_RAMP);
+                       else
+                               td_set_runstate(td, TD_RUNNING);
+                       nr_running++;
+                       nr_started--;
+                       m_rate += td->o.ratemin[0] + td->o.ratemin[1];
+                       t_rate += td->o.rate[0] + td->o.rate[1];
+                       todo--;
+                       fio_mutex_up(td->mutex);
+               }
+
+               reap_threads(&nr_running, &t_rate, &m_rate);
+
+               if (todo) {
+                       if (is_backend)
+                               fio_server_idle_loop();
+                       else
+                               usleep(100000);
+               }
+       }
+
+       while (nr_running) {
+               reap_threads(&nr_running, &t_rate, &m_rate);
+
+               if (is_backend)
+                       fio_server_idle_loop();
+               else
+                       usleep(10000);
+       }
+
+       update_io_ticks();
+       fio_unpin_memory();
+}
+
+static void *disk_thread_main(void *data)
+{
+       fio_mutex_up(startup_mutex);
+
+       while (threads) {
+               usleep(DISK_UTIL_MSEC * 1000);
+               if (!threads)
+                       break;
+               update_io_ticks();
+
+               if (!is_backend)
+                       print_thread_status();
+       }
+
+       return NULL;
+}
+
+static int create_disk_util_thread(void)
+{
+       int ret;
+
+       ret = pthread_create(&disk_util_thread, NULL, disk_thread_main, NULL);
+       if (ret) {
+               log_err("Can't create disk util thread: %s\n", strerror(ret));
+               return 1;
+       }
+
+       ret = pthread_detach(disk_util_thread);
+       if (ret) {
+               log_err("Can't detatch disk util thread: %s\n", strerror(ret));
+               return 1;
+       }
+
+       dprint(FD_MUTEX, "wait on startup_mutex\n");
+       fio_mutex_down(startup_mutex);
+       dprint(FD_MUTEX, "done waiting on startup_mutex\n");
+       return 0;
+}
+
+
+int fio_backend(void)
+{
+       struct thread_data *td;
+       int i;
+
+       if (exec_profile) {
+               if (load_profile(exec_profile))
+                       return 1;
+               free(exec_profile);
+               exec_profile = NULL;
+       }
+       if (!thread_number)
+               return 0;
+
+       if (write_bw_log) {
+               setup_log(&agg_io_log[DDIR_READ], 0);
+               setup_log(&agg_io_log[DDIR_WRITE], 0);
+       }
+
+       startup_mutex = fio_mutex_init(0);
+       if (startup_mutex == NULL)
+               return 1;
+       writeout_mutex = fio_mutex_init(1);
+       if (writeout_mutex == NULL)
+               return 1;
+
+       set_genesis_time();
+       create_disk_util_thread();
+
+       cgroup_list = smalloc(sizeof(*cgroup_list));
+       INIT_FLIST_HEAD(cgroup_list);
+
+       run_threads();
+
+       if (!fio_abort) {
+               show_run_stats();
+               if (write_bw_log) {
+                       __finish_log(agg_io_log[DDIR_READ], "agg-read_bw.log");
+                       __finish_log(agg_io_log[DDIR_WRITE],
+                                       "agg-write_bw.log");
+               }
+       }
+
+       for_each_td(td, i)
+               fio_options_free(td);
+
+       cgroup_kill(cgroup_list);
+       sfree(cgroup_list);
+       sfree(cgroup_mnt);
+
+       fio_mutex_remove(startup_mutex);
+       fio_mutex_remove(writeout_mutex);
+       return exit_value;
+}
+
+
diff --git a/fio.c b/fio.c
index 4f07214..f0a8ee2 100644 (file)
--- a/fio.c
+++ b/fio.c
@@ -2,7 +2,7 @@
  * fio - the flexible io tester
  *
  * Copyright (C) 2005 Jens Axboe <axboe@suse.de>
- * Copyright (C) 2006 Jens Axboe <axboe@kernel.dk>
+ * Copyright (C) 2006-2012 Jens Axboe <axboe@kernel.dk>
  *
  * The license below covers all files distributed with fio unless otherwise
  * noted in the file itself.
@@ -51,9 +51,6 @@
 unsigned long page_mask;
 unsigned long page_size;
 
-#define PAGE_ALIGN(buf)        \
-       (char *) (((unsigned long) (buf) + page_mask) & ~page_mask)
-
 int groupid = 0;
 unsigned int thread_number = 0;
 unsigned int nr_process = 0;
@@ -69,1740 +66,8 @@ unsigned long done_secs = 0;
 FLIST_HEAD(disk_list);
 #endif
 
-static struct fio_mutex *startup_mutex;
-static struct fio_mutex *writeout_mutex;
-static volatile int fio_abort;
-static int exit_value;
-static pthread_t disk_util_thread;
-static struct flist_head *cgroup_list;
-static char *cgroup_mnt;
-
 unsigned long arch_flags = 0;
 
-struct io_log *agg_io_log[2];
-
-#define JOB_START_TIMEOUT      (5 * 1000)
-
-static const char *fio_os_strings[os_nr] = {
-       "Invalid",
-       "Linux",
-       "AIX",
-       "FreeBSD",
-       "HP-UX",
-       "OSX",
-       "NetBSD",
-       "Solaris",
-       "Windows"
-};
-
-static const char *fio_arch_strings[arch_nr] = {
-       "Invalid",
-       "x86-64",
-       "x86",
-       "ppc",
-       "ia64",
-       "s390",
-       "alpha",
-       "sparc",
-       "sparc64",
-       "arm",
-       "sh",
-       "hppa",
-       "generic"
-};
-
-const char *fio_get_os_string(int nr)
-{
-       if (nr < os_nr)
-               return fio_os_strings[nr];
-
-       return NULL;
-}
-
-const char *fio_get_arch_string(int nr)
-{
-       if (nr < arch_nr)
-               return fio_arch_strings[nr];
-
-       return NULL;
-}
-
-void td_set_runstate(struct thread_data *td, int runstate)
-{
-       if (td->runstate == runstate)
-               return;
-
-       dprint(FD_PROCESS, "pid=%d: runstate %d -> %d\n", (int) td->pid,
-                                               td->runstate, runstate);
-       td->runstate = runstate;
-}
-
-void fio_terminate_threads(int group_id)
-{
-       struct thread_data *td;
-       int i;
-
-       dprint(FD_PROCESS, "terminate group_id=%d\n", group_id);
-
-       for_each_td(td, i) {
-               if (group_id == TERMINATE_ALL || groupid == td->groupid) {
-                       dprint(FD_PROCESS, "setting terminate on %s/%d\n",
-                                               td->o.name, (int) td->pid);
-                       td->terminate = 1;
-                       td->o.start_delay = 0;
-
-                       /*
-                        * if the thread is running, just let it exit
-                        */
-                       if (!td->pid)
-                               continue;
-                       else if (td->runstate < TD_RAMP)
-                               kill(td->pid, SIGTERM);
-                       else {
-                               struct ioengine_ops *ops = td->io_ops;
-
-                               if (ops && (ops->flags & FIO_SIGTERM))
-                                       kill(td->pid, SIGTERM);
-                       }
-               }
-       }
-}
-
-static void sig_int(int sig)
-{
-       if (threads) {
-               if (is_backend)
-                       fio_server_got_signal(sig);
-               else {
-                       log_info("\nfio: terminating on signal %d\n", sig);
-                       fflush(stdout);
-                       exit_value = 128;
-               }
-
-               fio_terminate_threads(TERMINATE_ALL);
-       }
-}
-
-static void *disk_thread_main(void *data)
-{
-       fio_mutex_up(startup_mutex);
-
-       while (threads) {
-               usleep(DISK_UTIL_MSEC * 1000);
-               if (!threads)
-                       break;
-               update_io_ticks();
-
-               if (!is_backend)
-                       print_thread_status();
-       }
-
-       return NULL;
-}
-
-static int create_disk_util_thread(void)
-{
-       int ret;
-
-       ret = pthread_create(&disk_util_thread, NULL, disk_thread_main, NULL);
-       if (ret) {
-               log_err("Can't create disk util thread: %s\n", strerror(ret));
-               return 1;
-       }
-
-       ret = pthread_detach(disk_util_thread);
-       if (ret) {
-               log_err("Can't detatch disk util thread: %s\n", strerror(ret));
-               return 1;
-       }
-
-       dprint(FD_MUTEX, "wait on startup_mutex\n");
-       fio_mutex_down(startup_mutex);
-       dprint(FD_MUTEX, "done waiting on startup_mutex\n");
-       return 0;
-}
-
-static void set_sig_handlers(void)
-{
-       struct sigaction act;
-
-       memset(&act, 0, sizeof(act));
-       act.sa_handler = sig_int;
-       act.sa_flags = SA_RESTART;
-       sigaction(SIGINT, &act, NULL);
-
-       memset(&act, 0, sizeof(act));
-       act.sa_handler = sig_int;
-       act.sa_flags = SA_RESTART;
-       sigaction(SIGTERM, &act, NULL);
-
-       if (is_backend) {
-               memset(&act, 0, sizeof(act));
-               act.sa_handler = sig_int;
-               act.sa_flags = SA_RESTART;
-               sigaction(SIGPIPE, &act, NULL);
-       }
-}
-
-/*
- * Check if we are above the minimum rate given.
- */
-static int __check_min_rate(struct thread_data *td, struct timeval *now,
-                           enum fio_ddir ddir)
-{
-       unsigned long long bytes = 0;
-       unsigned long iops = 0;
-       unsigned long spent;
-       unsigned long rate;
-       unsigned int ratemin = 0;
-       unsigned int rate_iops = 0;
-       unsigned int rate_iops_min = 0;
-
-       assert(ddir_rw(ddir));
-
-       if (!td->o.ratemin[ddir] && !td->o.rate_iops_min[ddir])
-               return 0;
-
-       /*
-        * allow a 2 second settle period in the beginning
-        */
-       if (mtime_since(&td->start, now) < 2000)
-               return 0;
-
-       iops += td->this_io_blocks[ddir];
-       bytes += td->this_io_bytes[ddir];
-       ratemin += td->o.ratemin[ddir];
-       rate_iops += td->o.rate_iops[ddir];
-       rate_iops_min += td->o.rate_iops_min[ddir];
-
-       /*
-        * if rate blocks is set, sample is running
-        */
-       if (td->rate_bytes[ddir] || td->rate_blocks[ddir]) {
-               spent = mtime_since(&td->lastrate[ddir], now);
-               if (spent < td->o.ratecycle)
-                       return 0;
-
-               if (td->o.rate[ddir]) {
-                       /*
-                        * check bandwidth specified rate
-                        */
-                       if (bytes < td->rate_bytes[ddir]) {
-                               log_err("%s: min rate %u not met\n", td->o.name,
-                                                               ratemin);
-                               return 1;
-                       } else {
-                               rate = ((bytes - td->rate_bytes[ddir]) * 1000) / spent;
-                               if (rate < ratemin ||
-                                   bytes < td->rate_bytes[ddir]) {
-                                       log_err("%s: min rate %u not met, got"
-                                               " %luKB/sec\n", td->o.name,
-                                                       ratemin, rate);
-                                       return 1;
-                               }
-                       }
-               } else {
-                       /*
-                        * checks iops specified rate
-                        */
-                       if (iops < rate_iops) {
-                               log_err("%s: min iops rate %u not met\n",
-                                               td->o.name, rate_iops);
-                               return 1;
-                       } else {
-                               rate = ((iops - td->rate_blocks[ddir]) * 1000) / spent;
-                               if (rate < rate_iops_min ||
-                                   iops < td->rate_blocks[ddir]) {
-                                       log_err("%s: min iops rate %u not met,"
-                                               " got %lu\n", td->o.name,
-                                                       rate_iops_min, rate);
-                               }
-                       }
-               }
-       }
-
-       td->rate_bytes[ddir] = bytes;
-       td->rate_blocks[ddir] = iops;
-       memcpy(&td->lastrate[ddir], now, sizeof(*now));
-       return 0;
-}
-
-static int check_min_rate(struct thread_data *td, struct timeval *now,
-                         unsigned long *bytes_done)
-{
-       int ret = 0;
-
-       if (bytes_done[0])
-               ret |= __check_min_rate(td, now, 0);
-       if (bytes_done[1])
-               ret |= __check_min_rate(td, now, 1);
-
-       return ret;
-}
-
-static inline int runtime_exceeded(struct thread_data *td, struct timeval *t)
-{
-       if (in_ramp_time(td))
-               return 0;
-       if (!td->o.timeout)
-               return 0;
-       if (mtime_since(&td->epoch, t) >= td->o.timeout * 1000)
-               return 1;
-
-       return 0;
-}
-
-/*
- * When job exits, we can cancel the in-flight IO if we are using async
- * io. Attempt to do so.
- */
-static void cleanup_pending_aio(struct thread_data *td)
-{
-       struct flist_head *entry, *n;
-       struct io_u *io_u;
-       int r;
-
-       /*
-        * get immediately available events, if any
-        */
-       r = io_u_queued_complete(td, 0, NULL);
-       if (r < 0)
-               return;
-
-       /*
-        * now cancel remaining active events
-        */
-       if (td->io_ops->cancel) {
-               flist_for_each_safe(entry, n, &td->io_u_busylist) {
-                       io_u = flist_entry(entry, struct io_u, list);
-
-                       /*
-                        * if the io_u isn't in flight, then that generally
-                        * means someone leaked an io_u. complain but fix
-                        * it up, so we don't stall here.
-                        */
-                       if ((io_u->flags & IO_U_F_FLIGHT) == 0) {
-                               log_err("fio: non-busy IO on busy list\n");
-                               put_io_u(td, io_u);
-                       } else {
-                               r = td->io_ops->cancel(td, io_u);
-                               if (!r)
-                                       put_io_u(td, io_u);
-                       }
-               }
-       }
-
-       if (td->cur_depth)
-               r = io_u_queued_complete(td, td->cur_depth, NULL);
-}
-
-/*
- * Helper to handle the final sync of a file. Works just like the normal
- * io path, just does everything sync.
- */
-static int fio_io_sync(struct thread_data *td, struct fio_file *f)
-{
-       struct io_u *io_u = __get_io_u(td);
-       int ret;
-
-       if (!io_u)
-               return 1;
-
-       io_u->ddir = DDIR_SYNC;
-       io_u->file = f;
-
-       if (td_io_prep(td, io_u)) {
-               put_io_u(td, io_u);
-               return 1;
-       }
-
-requeue:
-       ret = td_io_queue(td, io_u);
-       if (ret < 0) {
-               td_verror(td, io_u->error, "td_io_queue");
-               put_io_u(td, io_u);
-               return 1;
-       } else if (ret == FIO_Q_QUEUED) {
-               if (io_u_queued_complete(td, 1, NULL) < 0)
-                       return 1;
-       } else if (ret == FIO_Q_COMPLETED) {
-               if (io_u->error) {
-                       td_verror(td, io_u->error, "td_io_queue");
-                       return 1;
-               }
-
-               if (io_u_sync_complete(td, io_u, NULL) < 0)
-                       return 1;
-       } else if (ret == FIO_Q_BUSY) {
-               if (td_io_commit(td))
-                       return 1;
-               goto requeue;
-       }
-
-       return 0;
-}
-
-static inline void __update_tv_cache(struct thread_data *td)
-{
-       fio_gettime(&td->tv_cache, NULL);
-}
-
-static inline void update_tv_cache(struct thread_data *td)
-{
-       if ((++td->tv_cache_nr & td->tv_cache_mask) == td->tv_cache_mask)
-               __update_tv_cache(td);
-}
-
-static int break_on_this_error(struct thread_data *td, enum fio_ddir ddir,
-                              int *retptr)
-{
-       int ret = *retptr;
-
-       if (ret < 0 || td->error) {
-               int err;
-
-               if (ret < 0)
-                       err = -ret;
-               else
-                       err = td->error;
-
-               if (!(td->o.continue_on_error & td_error_type(ddir, err)))
-                       return 1;
-
-               if (td_non_fatal_error(err)) {
-                       /*
-                        * Continue with the I/Os in case of
-                        * a non fatal error.
-                        */
-                       update_error_count(td, err);
-                       td_clear_error(td);
-                       *retptr = 0;
-                       return 0;
-               } else if (td->o.fill_device && err == ENOSPC) {
-                       /*
-                        * We expect to hit this error if
-                        * fill_device option is set.
-                        */
-                       td_clear_error(td);
-                       td->terminate = 1;
-                       return 1;
-               } else {
-                       /*
-                        * Stop the I/O in case of a fatal
-                        * error.
-                        */
-                       update_error_count(td, err);
-                       return 1;
-               }
-       }
-
-       return 0;
-}
-
-/*
- * The main verify engine. Runs over the writes we previously submitted,
- * reads the blocks back in, and checks the crc/md5 of the data.
- */
-static void do_verify(struct thread_data *td)
-{
-       struct fio_file *f;
-       struct io_u *io_u;
-       int ret, min_events;
-       unsigned int i;
-
-       dprint(FD_VERIFY, "starting loop\n");
-
-       /*
-        * sync io first and invalidate cache, to make sure we really
-        * read from disk.
-        */
-       for_each_file(td, f, i) {
-               if (!fio_file_open(f))
-                       continue;
-               if (fio_io_sync(td, f))
-                       break;
-               if (file_invalidate_cache(td, f))
-                       break;
-       }
-
-       if (td->error)
-               return;
-
-       td_set_runstate(td, TD_VERIFYING);
-
-       io_u = NULL;
-       while (!td->terminate) {
-               int ret2, full;
-
-               update_tv_cache(td);
-
-               if (runtime_exceeded(td, &td->tv_cache)) {
-                       __update_tv_cache(td);
-                       if (runtime_exceeded(td, &td->tv_cache)) {
-                               td->terminate = 1;
-                               break;
-                       }
-               }
-
-               io_u = __get_io_u(td);
-               if (!io_u)
-                       break;
-
-               if (get_next_verify(td, io_u)) {
-                       put_io_u(td, io_u);
-                       break;
-               }
-
-               if (td_io_prep(td, io_u)) {
-                       put_io_u(td, io_u);
-                       break;
-               }
-
-               if (td->o.verify_async)
-                       io_u->end_io = verify_io_u_async;
-               else
-                       io_u->end_io = verify_io_u;
-
-               ret = td_io_queue(td, io_u);
-               switch (ret) {
-               case FIO_Q_COMPLETED:
-                       if (io_u->error) {
-                               ret = -io_u->error;
-                               clear_io_u(td, io_u);
-                       } else if (io_u->resid) {
-                               int bytes = io_u->xfer_buflen - io_u->resid;
-
-                               /*
-                                * zero read, fail
-                                */
-                               if (!bytes) {
-                                       td_verror(td, EIO, "full resid");
-                                       put_io_u(td, io_u);
-                                       break;
-                               }
-
-                               io_u->xfer_buflen = io_u->resid;
-                               io_u->xfer_buf += bytes;
-                               io_u->offset += bytes;
-
-                               if (ddir_rw(io_u->ddir))
-                                       td->ts.short_io_u[io_u->ddir]++;
-
-                               f = io_u->file;
-                               if (io_u->offset == f->real_file_size)
-                                       goto sync_done;
-
-                               requeue_io_u(td, &io_u);
-                       } else {
-sync_done:
-                               ret = io_u_sync_complete(td, io_u, NULL);
-                               if (ret < 0)
-                                       break;
-                       }
-                       continue;
-               case FIO_Q_QUEUED:
-                       break;
-               case FIO_Q_BUSY:
-                       requeue_io_u(td, &io_u);
-                       ret2 = td_io_commit(td);
-                       if (ret2 < 0)
-                               ret = ret2;
-                       break;
-               default:
-                       assert(ret < 0);
-                       td_verror(td, -ret, "td_io_queue");
-                       break;
-               }
-
-               if (break_on_this_error(td, io_u->ddir, &ret))
-                       break;
-
-               /*
-                * if we can queue more, do so. but check if there are
-                * completed io_u's first. Note that we can get BUSY even
-                * without IO queued, if the system is resource starved.
-                */
-               full = queue_full(td) || (ret == FIO_Q_BUSY && td->cur_depth);
-               if (full || !td->o.iodepth_batch_complete) {
-                       min_events = min(td->o.iodepth_batch_complete,
-                                        td->cur_depth);
-                       if (full && !min_events && td->o.iodepth_batch_complete != 0)
-                               min_events = 1;
-
-                       do {
-                               /*
-                                * Reap required number of io units, if any,
-                                * and do the verification on them through
-                                * the callback handler
-                                */
-                               if (io_u_queued_complete(td, min_events, NULL) < 0) {
-                                       ret = -1;
-                                       break;
-                               }
-                       } while (full && (td->cur_depth > td->o.iodepth_low));
-               }
-               if (ret < 0)
-                       break;
-       }
-
-       if (!td->error) {
-               min_events = td->cur_depth;
-
-               if (min_events)
-                       ret = io_u_queued_complete(td, min_events, NULL);
-       } else
-               cleanup_pending_aio(td);
-
-       td_set_runstate(td, TD_RUNNING);
-
-       dprint(FD_VERIFY, "exiting loop\n");
-}
-
-/*
- * Main IO worker function. It retrieves io_u's to process and queues
- * and reaps them, checking for rate and errors along the way.
- */
-static void do_io(struct thread_data *td)
-{
-       unsigned int i;
-       int ret = 0;
-
-       if (in_ramp_time(td))
-               td_set_runstate(td, TD_RAMP);
-       else
-               td_set_runstate(td, TD_RUNNING);
-
-       while ( (td->o.read_iolog_file && !flist_empty(&td->io_log_list)) ||
-               (!flist_empty(&td->trim_list)) ||
-               ((td->this_io_bytes[0] + td->this_io_bytes[1]) < td->o.size) ) {
-               struct timeval comp_time;
-               unsigned long bytes_done[2] = { 0, 0 };
-               int min_evts = 0;
-               struct io_u *io_u;
-               int ret2, full;
-               enum fio_ddir ddir;
-
-               if (td->terminate)
-                       break;
-
-               update_tv_cache(td);
-
-               if (runtime_exceeded(td, &td->tv_cache)) {
-                       __update_tv_cache(td);
-                       if (runtime_exceeded(td, &td->tv_cache)) {
-                               td->terminate = 1;
-                               break;
-                       }
-               }
-
-               io_u = get_io_u(td);
-               if (!io_u)
-                       break;
-
-               ddir = io_u->ddir;
-
-               /*
-                * Add verification end_io handler, if asked to verify
-                * a previously written file.
-                */
-               if (td->o.verify != VERIFY_NONE && io_u->ddir == DDIR_READ &&
-                   !td_rw(td)) {
-                       if (td->o.verify_async)
-                               io_u->end_io = verify_io_u_async;
-                       else
-                               io_u->end_io = verify_io_u;
-                       td_set_runstate(td, TD_VERIFYING);
-               } else if (in_ramp_time(td))
-                       td_set_runstate(td, TD_RAMP);
-               else
-                       td_set_runstate(td, TD_RUNNING);
-
-               ret = td_io_queue(td, io_u);
-               switch (ret) {
-               case FIO_Q_COMPLETED:
-                       if (io_u->error) {
-                               ret = -io_u->error;
-                               clear_io_u(td, io_u);
-                       } else if (io_u->resid) {
-                               int bytes = io_u->xfer_buflen - io_u->resid;
-                               struct fio_file *f = io_u->file;
-
-                               /*
-                                * zero read, fail
-                                */
-                               if (!bytes) {
-                                       td_verror(td, EIO, "full resid");
-                                       put_io_u(td, io_u);
-                                       break;
-                               }
-
-                               io_u->xfer_buflen = io_u->resid;
-                               io_u->xfer_buf += bytes;
-                               io_u->offset += bytes;
-
-                               if (ddir_rw(io_u->ddir))
-                                       td->ts.short_io_u[io_u->ddir]++;
-
-                               if (io_u->offset == f->real_file_size)
-                                       goto sync_done;
-
-                               requeue_io_u(td, &io_u);
-                       } else {
-sync_done:
-                               if (__should_check_rate(td, 0) ||
-                                   __should_check_rate(td, 1))
-                                       fio_gettime(&comp_time, NULL);
-
-                               ret = io_u_sync_complete(td, io_u, bytes_done);
-                               if (ret < 0)
-                                       break;
-                       }
-                       break;
-               case FIO_Q_QUEUED:
-                       /*
-                        * if the engine doesn't have a commit hook,
-                        * the io_u is really queued. if it does have such
-                        * a hook, it has to call io_u_queued() itself.
-                        */
-                       if (td->io_ops->commit == NULL)
-                               io_u_queued(td, io_u);
-                       break;
-               case FIO_Q_BUSY:
-                       requeue_io_u(td, &io_u);
-                       ret2 = td_io_commit(td);
-                       if (ret2 < 0)
-                               ret = ret2;
-                       break;
-               default:
-                       assert(ret < 0);
-                       put_io_u(td, io_u);
-                       break;
-               }
-
-               if (break_on_this_error(td, ddir, &ret))
-                       break;
-
-               /*
-                * See if we need to complete some commands. Note that we
-                * can get BUSY even without IO queued, if the system is
-                * resource starved.
-                */
-               full = queue_full(td) || (ret == FIO_Q_BUSY && td->cur_depth);
-               if (full || !td->o.iodepth_batch_complete) {
-                       min_evts = min(td->o.iodepth_batch_complete,
-                                       td->cur_depth);
-                       if (full && !min_evts && td->o.iodepth_batch_complete != 0)
-                               min_evts = 1;
-
-                       if (__should_check_rate(td, 0) ||
-                           __should_check_rate(td, 1))
-                               fio_gettime(&comp_time, NULL);
-
-                       do {
-                               ret = io_u_queued_complete(td, min_evts, bytes_done);
-                               if (ret < 0)
-                                       break;
-
-                       } while (full && (td->cur_depth > td->o.iodepth_low));
-               }
-
-               if (ret < 0)
-                       break;
-               if (!(bytes_done[0] + bytes_done[1]))
-                       continue;
-
-               if (!in_ramp_time(td) && should_check_rate(td, bytes_done)) {
-                       if (check_min_rate(td, &comp_time, bytes_done)) {
-                               if (exitall_on_terminate)
-                                       fio_terminate_threads(td->groupid);
-                               td_verror(td, EIO, "check_min_rate");
-                               break;
-                       }
-               }
-
-               if (td->o.thinktime) {
-                       unsigned long long b;
-
-                       b = td->io_blocks[0] + td->io_blocks[1];
-                       if (!(b % td->o.thinktime_blocks)) {
-                               int left;
-
-                               if (td->o.thinktime_spin)
-                                       usec_spin(td->o.thinktime_spin);
-
-                               left = td->o.thinktime - td->o.thinktime_spin;
-                               if (left)
-                                       usec_sleep(td, left);
-                       }
-               }
-       }
-
-       if (td->trim_entries)
-               log_err("fio: %d trim entries leaked?\n", td->trim_entries);
-
-       if (td->o.fill_device && td->error == ENOSPC) {
-               td->error = 0;
-               td->terminate = 1;
-       }
-       if (!td->error) {
-               struct fio_file *f;
-
-               i = td->cur_depth;
-               if (i) {
-                       ret = io_u_queued_complete(td, i, NULL);
-                       if (td->o.fill_device && td->error == ENOSPC)
-                               td->error = 0;
-               }
-
-               if (should_fsync(td) && td->o.end_fsync) {
-                       td_set_runstate(td, TD_FSYNCING);
-
-                       for_each_file(td, f, i) {
-                               if (!fio_file_open(f))
-                                       continue;
-                               fio_io_sync(td, f);
-                       }
-               }
-       } else
-               cleanup_pending_aio(td);
-
-       /*
-        * stop job if we failed doing any IO
-        */
-       if ((td->this_io_bytes[0] + td->this_io_bytes[1]) == 0)
-               td->done = 1;
-}
-
-static void cleanup_io_u(struct thread_data *td)
-{
-       struct flist_head *entry, *n;
-       struct io_u *io_u;
-
-       flist_for_each_safe(entry, n, &td->io_u_freelist) {
-               io_u = flist_entry(entry, struct io_u, list);
-
-               flist_del(&io_u->list);
-               fio_memfree(io_u, sizeof(*io_u));
-       }
-
-       free_io_mem(td);
-}
-
-static int init_io_u(struct thread_data *td)
-{
-       struct io_u *io_u;
-       unsigned int max_bs;
-       int cl_align, i, max_units;
-       char *p;
-
-       max_units = td->o.iodepth;
-       max_bs = max(td->o.max_bs[DDIR_READ], td->o.max_bs[DDIR_WRITE]);
-       td->orig_buffer_size = (unsigned long long) max_bs
-                                       * (unsigned long long) max_units;
-
-       if (td->o.mem_type == MEM_SHMHUGE || td->o.mem_type == MEM_MMAPHUGE) {
-               unsigned long bs;
-
-               bs = td->orig_buffer_size + td->o.hugepage_size - 1;
-               td->orig_buffer_size = bs & ~(td->o.hugepage_size - 1);
-       }
-
-       if (td->orig_buffer_size != (size_t) td->orig_buffer_size) {
-               log_err("fio: IO memory too large. Reduce max_bs or iodepth\n");
-               return 1;
-       }
-
-       if (allocate_io_mem(td))
-               return 1;
-
-       if (td->o.odirect || td->o.mem_align ||
-           (td->io_ops->flags & FIO_RAWIO))
-               p = PAGE_ALIGN(td->orig_buffer) + td->o.mem_align;
-       else
-               p = td->orig_buffer;
-
-       cl_align = os_cache_line_size();
-
-       for (i = 0; i < max_units; i++) {
-               void *ptr;
-
-               if (td->terminate)
-                       return 1;
-
-               ptr = fio_memalign(cl_align, sizeof(*io_u));
-               if (!ptr) {
-                       log_err("fio: unable to allocate aligned memory\n");
-                       break;
-               }
-
-               io_u = ptr;
-               memset(io_u, 0, sizeof(*io_u));
-               INIT_FLIST_HEAD(&io_u->list);
-               dprint(FD_MEM, "io_u alloc %p, index %u\n", io_u, i);
-
-               if (!(td->io_ops->flags & FIO_NOIO)) {
-                       io_u->buf = p;
-                       dprint(FD_MEM, "io_u %p, mem %p\n", io_u, io_u->buf);
-
-                       if (td_write(td))
-                               io_u_fill_buffer(td, io_u, max_bs);
-                       if (td_write(td) && td->o.verify_pattern_bytes) {
-                               /*
-                                * Fill the buffer with the pattern if we are
-                                * going to be doing writes.
-                                */
-                               fill_pattern(td, io_u->buf, max_bs, io_u, 0, 0);
-                       }
-               }
-
-               io_u->index = i;
-               io_u->flags = IO_U_F_FREE;
-               flist_add(&io_u->list, &td->io_u_freelist);
-               p += max_bs;
-       }
-
-       return 0;
-}
-
-static int switch_ioscheduler(struct thread_data *td)
-{
-       char tmp[256], tmp2[128];
-       FILE *f;
-       int ret;
-
-       if (td->io_ops->flags & FIO_DISKLESSIO)
-               return 0;
-
-       sprintf(tmp, "%s/queue/scheduler", td->sysfs_root);
-
-       f = fopen(tmp, "r+");
-       if (!f) {
-               if (errno == ENOENT) {
-                       log_err("fio: os or kernel doesn't support IO scheduler"
-                               " switching\n");
-                       return 0;
-               }
-               td_verror(td, errno, "fopen iosched");
-               return 1;
-       }
-
-       /*
-        * Set io scheduler.
-        */
-       ret = fwrite(td->o.ioscheduler, strlen(td->o.ioscheduler), 1, f);
-       if (ferror(f) || ret != 1) {
-               td_verror(td, errno, "fwrite");
-               fclose(f);
-               return 1;
-       }
-
-       rewind(f);
-
-       /*
-        * Read back and check that the selected scheduler is now the default.
-        */
-       ret = fread(tmp, 1, sizeof(tmp), f);
-       if (ferror(f) || ret < 0) {
-               td_verror(td, errno, "fread");
-               fclose(f);
-               return 1;
-       }
-
-       sprintf(tmp2, "[%s]", td->o.ioscheduler);
-       if (!strstr(tmp, tmp2)) {
-               log_err("fio: io scheduler %s not found\n", td->o.ioscheduler);
-               td_verror(td, EINVAL, "iosched_switch");
-               fclose(f);
-               return 1;
-       }
-
-       fclose(f);
-       return 0;
-}
-
-static int keep_running(struct thread_data *td)
-{
-       unsigned long long io_done;
-
-       if (td->done)
-               return 0;
-       if (td->o.time_based)
-               return 1;
-       if (td->o.loops) {
-               td->o.loops--;
-               return 1;
-       }
-
-       io_done = td->io_bytes[DDIR_READ] + td->io_bytes[DDIR_WRITE]
-                       + td->io_skip_bytes;
-       if (io_done < td->o.size)
-               return 1;
-
-       return 0;
-}
-
-static void reset_io_counters(struct thread_data *td)
-{
-       td->stat_io_bytes[0] = td->stat_io_bytes[1] = 0;
-       td->this_io_bytes[0] = td->this_io_bytes[1] = 0;
-       td->stat_io_blocks[0] = td->stat_io_blocks[1] = 0;
-       td->this_io_blocks[0] = td->this_io_blocks[1] = 0;
-       td->zone_bytes = 0;
-       td->rate_bytes[0] = td->rate_bytes[1] = 0;
-       td->rate_blocks[0] = td->rate_blocks[1] = 0;
-
-       td->last_was_sync = 0;
-
-       /*
-        * reset file done count if we are to start over
-        */
-       if (td->o.time_based || td->o.loops)
-               td->nr_done_files = 0;
-}
-
-void reset_all_stats(struct thread_data *td)
-{
-       struct timeval tv;
-       int i;
-
-       reset_io_counters(td);
-
-       for (i = 0; i < 2; i++) {
-               td->io_bytes[i] = 0;
-               td->io_blocks[i] = 0;
-               td->io_issues[i] = 0;
-               td->ts.total_io_u[i] = 0;
-       }
-
-       fio_gettime(&tv, NULL);
-       td->ts.runtime[0] = 0;
-       td->ts.runtime[1] = 0;
-       memcpy(&td->epoch, &tv, sizeof(tv));
-       memcpy(&td->start, &tv, sizeof(tv));
-}
-
-static void clear_io_state(struct thread_data *td)
-{
-       struct fio_file *f;
-       unsigned int i;
-
-       reset_io_counters(td);
-
-       close_files(td);
-       for_each_file(td, f, i)
-               fio_file_clear_done(f);
-
-       /*
-        * Set the same seed to get repeatable runs
-        */
-       td_fill_rand_seeds(td);
-}
-
-static int exec_string(const char *string)
-{
-       int ret, newlen = strlen(string) + 1 + 8;
-       char *str;
-
-       str = malloc(newlen);
-       sprintf(str, "sh -c %s", string);
-
-       ret = system(str);
-       if (ret == -1)
-               log_err("fio: exec of cmd <%s> failed\n", str);
-
-       free(str);
-       return ret;
-}
-
-/*
- * Entry point for the thread based jobs. The process based jobs end up
- * here as well, after a little setup.
- */
-static void *thread_main(void *data)
-{
-       unsigned long long elapsed;
-       struct thread_data *td = data;
-       pthread_condattr_t attr;
-       int clear_state;
-
-       if (!td->o.use_thread) {
-               setsid();
-               td->pid = getpid();
-       } else
-               td->pid = gettid();
-
-       dprint(FD_PROCESS, "jobs pid=%d started\n", (int) td->pid);
-
-       INIT_FLIST_HEAD(&td->io_u_freelist);
-       INIT_FLIST_HEAD(&td->io_u_busylist);
-       INIT_FLIST_HEAD(&td->io_u_requeues);
-       INIT_FLIST_HEAD(&td->io_log_list);
-       INIT_FLIST_HEAD(&td->io_hist_list);
-       INIT_FLIST_HEAD(&td->verify_list);
-       INIT_FLIST_HEAD(&td->trim_list);
-       pthread_mutex_init(&td->io_u_lock, NULL);
-       td->io_hist_tree = RB_ROOT;
-
-       pthread_condattr_init(&attr);
-       pthread_cond_init(&td->verify_cond, &attr);
-       pthread_cond_init(&td->free_cond, &attr);
-
-       td_set_runstate(td, TD_INITIALIZED);
-       dprint(FD_MUTEX, "up startup_mutex\n");
-       fio_mutex_up(startup_mutex);
-       dprint(FD_MUTEX, "wait on td->mutex\n");
-       fio_mutex_down(td->mutex);
-       dprint(FD_MUTEX, "done waiting on td->mutex\n");
-
-       /*
-        * the ->mutex mutex is now no longer used, close it to avoid
-        * eating a file descriptor
-        */
-       fio_mutex_remove(td->mutex);
-
-       /*
-        * A new gid requires privilege, so we need to do this before setting
-        * the uid.
-        */
-       if (td->o.gid != -1U && setgid(td->o.gid)) {
-               td_verror(td, errno, "setgid");
-               goto err;
-       }
-       if (td->o.uid != -1U && setuid(td->o.uid)) {
-               td_verror(td, errno, "setuid");
-               goto err;
-       }
-
-       /*
-        * If we have a gettimeofday() thread, make sure we exclude that
-        * thread from this job
-        */
-       if (td->o.gtod_cpu)
-               fio_cpu_clear(&td->o.cpumask, td->o.gtod_cpu);
-
-       /*
-        * Set affinity first, in case it has an impact on the memory
-        * allocations.
-        */
-       if (td->o.cpumask_set && fio_setaffinity(td->pid, td->o.cpumask) == -1) {
-               td_verror(td, errno, "cpu_set_affinity");
-               goto err;
-       }
-
-       /*
-        * May alter parameters that init_io_u() will use, so we need to
-        * do this first.
-        */
-       if (init_iolog(td))
-               goto err;
-
-       if (init_io_u(td))
-               goto err;
-
-       if (td->o.verify_async && verify_async_init(td))
-               goto err;
-
-       if (td->ioprio_set) {
-               if (ioprio_set(IOPRIO_WHO_PROCESS, 0, td->ioprio) == -1) {
-                       td_verror(td, errno, "ioprio_set");
-                       goto err;
-               }
-       }
-
-       if (td->o.cgroup_weight && cgroup_setup(td, cgroup_list, &cgroup_mnt))
-               goto err;
-
-       if (nice(td->o.nice) == -1) {
-               td_verror(td, errno, "nice");
-               goto err;
-       }
-
-       if (td->o.ioscheduler && switch_ioscheduler(td))
-               goto err;
-
-       if (!td->o.create_serialize && setup_files(td))
-               goto err;
-
-       if (td_io_init(td))
-               goto err;
-
-       if (init_random_map(td))
-               goto err;
-
-       if (td->o.exec_prerun) {
-               if (exec_string(td->o.exec_prerun))
-                       goto err;
-       }
-
-       if (td->o.pre_read) {
-               if (pre_read_files(td) < 0)
-                       goto err;
-       }
-
-       fio_gettime(&td->epoch, NULL);
-       getrusage(RUSAGE_SELF, &td->ru_start);
-
-       clear_state = 0;
-       while (keep_running(td)) {
-               fio_gettime(&td->start, NULL);
-               memcpy(&td->bw_sample_time, &td->start, sizeof(td->start));
-               memcpy(&td->iops_sample_time, &td->start, sizeof(td->start));
-               memcpy(&td->tv_cache, &td->start, sizeof(td->start));
-
-               if (td->o.ratemin[0] || td->o.ratemin[1]) {
-                       memcpy(&td->lastrate[0], &td->bw_sample_time,
-                                               sizeof(td->bw_sample_time));
-                       memcpy(&td->lastrate[1], &td->bw_sample_time,
-                                               sizeof(td->bw_sample_time));
-               }
-
-               if (clear_state)
-                       clear_io_state(td);
-
-               prune_io_piece_log(td);
-
-               do_io(td);
-
-               clear_state = 1;
-
-               if (td_read(td) && td->io_bytes[DDIR_READ]) {
-                       elapsed = utime_since_now(&td->start);
-                       td->ts.runtime[DDIR_READ] += elapsed;
-               }
-               if (td_write(td) && td->io_bytes[DDIR_WRITE]) {
-                       elapsed = utime_since_now(&td->start);
-                       td->ts.runtime[DDIR_WRITE] += elapsed;
-               }
-
-               if (td->error || td->terminate)
-                       break;
-
-               if (!td->o.do_verify ||
-                   td->o.verify == VERIFY_NONE ||
-                   (td->io_ops->flags & FIO_UNIDIR))
-                       continue;
-
-               clear_io_state(td);
-
-               fio_gettime(&td->start, NULL);
-
-               do_verify(td);
-
-               td->ts.runtime[DDIR_READ] += utime_since_now(&td->start);
-
-               if (td->error || td->terminate)
-                       break;
-       }
-
-       update_rusage_stat(td);
-       td->ts.runtime[0] = (td->ts.runtime[0] + 999) / 1000;
-       td->ts.runtime[1] = (td->ts.runtime[1] + 999) / 1000;
-       td->ts.total_run_time = mtime_since_now(&td->epoch);
-       td->ts.io_bytes[0] = td->io_bytes[0];
-       td->ts.io_bytes[1] = td->io_bytes[1];
-
-       fio_mutex_down(writeout_mutex);
-       if (td->bw_log) {
-               if (td->o.bw_log_file) {
-                       finish_log_named(td, td->bw_log,
-                                               td->o.bw_log_file, "bw");
-               } else
-                       finish_log(td, td->bw_log, "bw");
-       }
-       if (td->lat_log) {
-               if (td->o.lat_log_file) {
-                       finish_log_named(td, td->lat_log,
-                                               td->o.lat_log_file, "lat");
-               } else
-                       finish_log(td, td->lat_log, "lat");
-       }
-       if (td->slat_log) {
-               if (td->o.lat_log_file) {
-                       finish_log_named(td, td->slat_log,
-                                               td->o.lat_log_file, "slat");
-               } else
-                       finish_log(td, td->slat_log, "slat");
-       }
-       if (td->clat_log) {
-               if (td->o.lat_log_file) {
-                       finish_log_named(td, td->clat_log,
-                                               td->o.lat_log_file, "clat");
-               } else
-                       finish_log(td, td->clat_log, "clat");
-       }
-       if (td->iops_log) {
-               if (td->o.iops_log_file) {
-                       finish_log_named(td, td->iops_log,
-                                               td->o.iops_log_file, "iops");
-               } else
-                       finish_log(td, td->iops_log, "iops");
-       }
-
-       fio_mutex_up(writeout_mutex);
-       if (td->o.exec_postrun)
-               exec_string(td->o.exec_postrun);
-
-       if (exitall_on_terminate)
-               fio_terminate_threads(td->groupid);
-
-err:
-       if (td->error)
-               log_info("fio: pid=%d, err=%d/%s\n", (int) td->pid, td->error,
-                                                       td->verror);
-
-       if (td->o.verify_async)
-               verify_async_exit(td);
-
-       close_and_free_files(td);
-       close_ioengine(td);
-       cleanup_io_u(td);
-       cgroup_shutdown(td, &cgroup_mnt);
-
-       if (td->o.cpumask_set) {
-               int ret = fio_cpuset_exit(&td->o.cpumask);
-
-               td_verror(td, ret, "fio_cpuset_exit");
-       }
-
-       /*
-        * do this very late, it will log file closing as well
-        */
-       if (td->o.write_iolog_file)
-               write_iolog_close(td);
-
-       td_set_runstate(td, TD_EXITED);
-       return (void *) (unsigned long) td->error;
-}
-
-/*
- * We cannot pass the td data into a forked process, so attach the td and
- * pass it to the thread worker.
- */
-static int fork_main(int shmid, int offset)
-{
-       struct thread_data *td;
-       void *data, *ret;
-
-#ifndef __hpux
-       data = shmat(shmid, NULL, 0);
-       if (data == (void *) -1) {
-               int __err = errno;
-
-               perror("shmat");
-               return __err;
-       }
-#else
-       /*
-        * HP-UX inherits shm mappings?
-        */
-       data = threads;
-#endif
-
-       td = data + offset * sizeof(struct thread_data);
-       ret = thread_main(td);
-       shmdt(data);
-       return (int) (unsigned long) ret;
-}
-
-/*
- * Run over the job map and reap the threads that have exited, if any.
- */
-static void reap_threads(unsigned int *nr_running, unsigned int *t_rate,
-                        unsigned int *m_rate)
-{
-       struct thread_data *td;
-       unsigned int cputhreads, realthreads, pending;
-       int i, status, ret;
-
-       /*
-        * reap exited threads (TD_EXITED -> TD_REAPED)
-        */
-       realthreads = pending = cputhreads = 0;
-       for_each_td(td, i) {
-               int flags = 0;
-
-               /*
-                * ->io_ops is NULL for a thread that has closed its
-                * io engine
-                */
-               if (td->io_ops && !strcmp(td->io_ops->name, "cpuio"))
-                       cputhreads++;
-               else
-                       realthreads++;
-
-               if (!td->pid) {
-                       pending++;
-                       continue;
-               }
-               if (td->runstate == TD_REAPED)
-                       continue;
-               if (td->o.use_thread) {
-                       if (td->runstate == TD_EXITED) {
-                               td_set_runstate(td, TD_REAPED);
-                               goto reaped;
-                       }
-                       continue;
-               }
-
-               flags = WNOHANG;
-               if (td->runstate == TD_EXITED)
-                       flags = 0;
-
-               /*
-                * check if someone quit or got killed in an unusual way
-                */
-               ret = waitpid(td->pid, &status, flags);
-               if (ret < 0) {
-                       if (errno == ECHILD) {
-                               log_err("fio: pid=%d disappeared %d\n",
-                                               (int) td->pid, td->runstate);
-                               td_set_runstate(td, TD_REAPED);
-                               goto reaped;
-                       }
-                       perror("waitpid");
-               } else if (ret == td->pid) {
-                       if (WIFSIGNALED(status)) {
-                               int sig = WTERMSIG(status);
-
-                               if (sig != SIGTERM)
-                                       log_err("fio: pid=%d, got signal=%d\n",
-                                                       (int) td->pid, sig);
-                               td_set_runstate(td, TD_REAPED);
-                               goto reaped;
-                       }
-                       if (WIFEXITED(status)) {
-                               if (WEXITSTATUS(status) && !td->error)
-                                       td->error = WEXITSTATUS(status);
-
-                               td_set_runstate(td, TD_REAPED);
-                               goto reaped;
-                       }
-               }
-
-               /*
-                * thread is not dead, continue
-                */
-               pending++;
-               continue;
-reaped:
-               (*nr_running)--;
-               (*m_rate) -= (td->o.ratemin[0] + td->o.ratemin[1]);
-               (*t_rate) -= (td->o.rate[0] + td->o.rate[1]);
-               if (!td->pid)
-                       pending--;
-
-               if (td->error)
-                       exit_value++;
-
-               done_secs += mtime_since_now(&td->epoch) / 1000;
-       }
-
-       if (*nr_running == cputhreads && !pending && realthreads)
-               fio_terminate_threads(TERMINATE_ALL);
-}
-
-/*
- * Main function for kicking off and reaping jobs, as needed.
- */
-static void run_threads(void)
-{
-       struct thread_data *td;
-       unsigned long spent;
-       unsigned int i, todo, nr_running, m_rate, t_rate, nr_started;
-
-       if (fio_pin_memory())
-               return;
-
-       if (fio_gtod_offload && fio_start_gtod_thread())
-               return;
-
-       set_sig_handlers();
-
-       if (!terse_output) {
-               log_info("Starting ");
-               if (nr_thread)
-                       log_info("%d thread%s", nr_thread,
-                                               nr_thread > 1 ? "s" : "");
-               if (nr_process) {
-                       if (nr_thread)
-                               log_info(" and ");
-                       log_info("%d process%s", nr_process,
-                                               nr_process > 1 ? "es" : "");
-               }
-               log_info("\n");
-               fflush(stdout);
-       }
-
-       todo = thread_number;
-       nr_running = 0;
-       nr_started = 0;
-       m_rate = t_rate = 0;
-
-       for_each_td(td, i) {
-               print_status_init(td->thread_number - 1);
-
-               if (!td->o.create_serialize)
-                       continue;
-
-               /*
-                * do file setup here so it happens sequentially,
-                * we don't want X number of threads getting their
-                * client data interspersed on disk
-                */
-               if (setup_files(td)) {
-                       exit_value++;
-                       if (td->error)
-                               log_err("fio: pid=%d, err=%d/%s\n",
-                                       (int) td->pid, td->error, td->verror);
-                       td_set_runstate(td, TD_REAPED);
-                       todo--;
-               } else {
-                       struct fio_file *f;
-                       unsigned int j;
-
-                       /*
-                        * for sharing to work, each job must always open
-                        * its own files. so close them, if we opened them
-                        * for creation
-                        */
-                       for_each_file(td, f, j) {
-                               if (fio_file_open(f))
-                                       td_io_close_file(td, f);
-                       }
-               }
-       }
-
-       set_genesis_time();
-
-       while (todo) {
-               struct thread_data *map[REAL_MAX_JOBS];
-               struct timeval this_start;
-               int this_jobs = 0, left;
-
-               /*
-                * create threads (TD_NOT_CREATED -> TD_CREATED)
-                */
-               for_each_td(td, i) {
-                       if (td->runstate != TD_NOT_CREATED)
-                               continue;
-
-                       /*
-                        * never got a chance to start, killed by other
-                        * thread for some reason
-                        */
-                       if (td->terminate) {
-                               todo--;
-                               continue;
-                       }
-
-                       if (td->o.start_delay) {
-                               spent = mtime_since_genesis();
-
-                               if (td->o.start_delay * 1000 > spent)
-                                       continue;
-                       }
-
-                       if (td->o.stonewall && (nr_started || nr_running)) {
-                               dprint(FD_PROCESS, "%s: stonewall wait\n",
-                                                       td->o.name);
-                               break;
-                       }
-
-                       init_disk_util(td);
-
-                       /*
-                        * Set state to created. Thread will transition
-                        * to TD_INITIALIZED when it's done setting up.
-                        */
-                       td_set_runstate(td, TD_CREATED);
-                       map[this_jobs++] = td;
-                       nr_started++;
-
-                       if (td->o.use_thread) {
-                               int ret;
-
-                               dprint(FD_PROCESS, "will pthread_create\n");
-                               ret = pthread_create(&td->thread, NULL,
-                                                       thread_main, td);
-                               if (ret) {
-                                       log_err("pthread_create: %s\n",
-                                                       strerror(ret));
-                                       nr_started--;
-                                       break;
-                               }
-                               ret = pthread_detach(td->thread);
-                               if (ret)
-                                       log_err("pthread_detach: %s",
-                                                       strerror(ret));
-                       } else {
-                               pid_t pid;
-                               dprint(FD_PROCESS, "will fork\n");
-                               pid = fork();
-                               if (!pid) {
-                                       int ret = fork_main(shm_id, i);
-
-                                       _exit(ret);
-                               } else if (i == fio_debug_jobno)
-                                       *fio_debug_jobp = pid;
-                       }
-                       dprint(FD_MUTEX, "wait on startup_mutex\n");
-                       if (fio_mutex_down_timeout(startup_mutex, 10)) {
-                               log_err("fio: job startup hung? exiting.\n");
-                               fio_terminate_threads(TERMINATE_ALL);
-                               fio_abort = 1;
-                               nr_started--;
-                               break;
-                       }
-                       dprint(FD_MUTEX, "done waiting on startup_mutex\n");
-               }
-
-               /*
-                * Wait for the started threads to transition to
-                * TD_INITIALIZED.
-                */
-               fio_gettime(&this_start, NULL);
-               left = this_jobs;
-               while (left && !fio_abort) {
-                       if (mtime_since_now(&this_start) > JOB_START_TIMEOUT)
-                               break;
-
-                       usleep(100000);
-
-                       for (i = 0; i < this_jobs; i++) {
-                               td = map[i];
-                               if (!td)
-                                       continue;
-                               if (td->runstate == TD_INITIALIZED) {
-                                       map[i] = NULL;
-                                       left--;
-                               } else if (td->runstate >= TD_EXITED) {
-                                       map[i] = NULL;
-                                       left--;
-                                       todo--;
-                                       nr_running++; /* work-around... */
-                               }
-                       }
-               }
-
-               if (left) {
-                       log_err("fio: %d jobs failed to start\n", left);
-                       for (i = 0; i < this_jobs; i++) {
-                               td = map[i];
-                               if (!td)
-                                       continue;
-                               kill(td->pid, SIGTERM);
-                       }
-                       break;
-               }
-
-               /*
-                * start created threads (TD_INITIALIZED -> TD_RUNNING).
-                */
-               for_each_td(td, i) {
-                       if (td->runstate != TD_INITIALIZED)
-                               continue;
-
-                       if (in_ramp_time(td))
-                               td_set_runstate(td, TD_RAMP);
-                       else
-                               td_set_runstate(td, TD_RUNNING);
-                       nr_running++;
-                       nr_started--;
-                       m_rate += td->o.ratemin[0] + td->o.ratemin[1];
-                       t_rate += td->o.rate[0] + td->o.rate[1];
-                       todo--;
-                       fio_mutex_up(td->mutex);
-               }
-
-               reap_threads(&nr_running, &t_rate, &m_rate);
-
-               if (todo) {
-                       if (is_backend)
-                               fio_server_idle_loop();
-                       else
-                               usleep(100000);
-               }
-       }
-
-       while (nr_running) {
-               reap_threads(&nr_running, &t_rate, &m_rate);
-
-               if (is_backend)
-                       fio_server_idle_loop();
-               else
-                       usleep(10000);
-       }
-
-       update_io_ticks();
-       fio_unpin_memory();
-}
-
-int exec_run(void)
-{
-       struct thread_data *td;
-       int i;
-
-       if (nr_clients)
-               return fio_handle_clients();
-       if (exec_profile) {
-               if (load_profile(exec_profile))
-                       return 1;
-               free(exec_profile);
-               exec_profile = NULL;
-       }
-       if (!thread_number)
-               return 0;
-
-       if (write_bw_log) {
-               setup_log(&agg_io_log[DDIR_READ], 0);
-               setup_log(&agg_io_log[DDIR_WRITE], 0);
-       }
-
-       startup_mutex = fio_mutex_init(0);
-       if (startup_mutex == NULL)
-               return 1;
-       writeout_mutex = fio_mutex_init(1);
-       if (writeout_mutex == NULL)
-               return 1;
-
-       set_genesis_time();
-       create_disk_util_thread();
-
-       cgroup_list = smalloc(sizeof(*cgroup_list));
-       INIT_FLIST_HEAD(cgroup_list);
-
-       run_threads();
-
-       if (!fio_abort) {
-               show_run_stats();
-               if (write_bw_log) {
-                       __finish_log(agg_io_log[DDIR_READ], "agg-read_bw.log");
-                       __finish_log(agg_io_log[DDIR_WRITE],
-                                       "agg-write_bw.log");
-               }
-       }
-
-       for_each_td(td, i)
-               fio_options_free(td);
-
-       cgroup_kill(cgroup_list);
-       sfree(cgroup_list);
-       sfree(cgroup_mnt);
-
-       fio_mutex_remove(startup_mutex);
-       fio_mutex_remove(writeout_mutex);
-       return exit_value;
-}
-
-void reset_fio_state(void)
-{
-       groupid = 0;
-       thread_number = 0;
-       nr_process = 0;
-       nr_thread = 0;
-       done_secs = 0;
-}
-
 static int endian_check(void)
 {
        union {
@@ -1868,5 +133,8 @@ int main(int argc, char *argv[], char *envp[])
        if (parse_options(argc, argv))
                return 1;
 
-       return exec_run();
+       if (nr_clients)
+               return fio_handle_clients();
+       else
+               return fio_backend();
 }
diff --git a/fio.h b/fio.h
index c8b94f6..9eb09b2 100644 (file)
--- a/fio.h
+++ b/fio.h
@@ -580,8 +580,9 @@ static inline int should_fsync(struct thread_data *td)
 extern int __must_check parse_options(int, char **);
 extern int parse_jobs_ini(char *, int, int);
 extern int parse_cmd_line(int, char **);
-extern int exec_run(void);
+extern int fio_backend(void);
 extern void reset_fio_state(void);
+extern void clear_io_state(struct thread_data *);
 extern int fio_options_parse(struct thread_data *, char **, int);
 extern void fio_keywords_init(void);
 extern int fio_cmd_option_parse(struct thread_data *, const char *, char *);
diff --git a/libfio.c b/libfio.c
new file mode 100644 (file)
index 0000000..cb917ce
--- /dev/null
+++ b/libfio.c
@@ -0,0 +1,179 @@
+/*
+ * fio - the flexible io tester
+ *
+ * Copyright (C) 2005 Jens Axboe <axboe@suse.de>
+ * Copyright (C) 2006-2012 Jens Axboe <axboe@kernel.dk>
+ *
+ * The license below covers all files distributed with fio unless otherwise
+ * noted in the file itself.
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2 as
+ *  published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ *
+ */
+
+#include <string.h>
+#include "fio.h"
+
+static const char *fio_os_strings[os_nr] = {
+       "Invalid",
+       "Linux",
+       "AIX",
+       "FreeBSD",
+       "HP-UX",
+       "OSX",
+       "NetBSD",
+       "Solaris",
+       "Windows"
+};
+
+static const char *fio_arch_strings[arch_nr] = {
+       "Invalid",
+       "x86-64",
+       "x86",
+       "ppc",
+       "ia64",
+       "s390",
+       "alpha",
+       "sparc",
+       "sparc64",
+       "arm",
+       "sh",
+       "hppa",
+       "generic"
+};
+
+static void reset_io_counters(struct thread_data *td)
+{
+       td->stat_io_bytes[0] = td->stat_io_bytes[1] = 0;
+       td->this_io_bytes[0] = td->this_io_bytes[1] = 0;
+       td->stat_io_blocks[0] = td->stat_io_blocks[1] = 0;
+       td->this_io_blocks[0] = td->this_io_blocks[1] = 0;
+       td->zone_bytes = 0;
+       td->rate_bytes[0] = td->rate_bytes[1] = 0;
+       td->rate_blocks[0] = td->rate_blocks[1] = 0;
+
+       td->last_was_sync = 0;
+
+       /*
+        * reset file done count if we are to start over
+        */
+       if (td->o.time_based || td->o.loops)
+               td->nr_done_files = 0;
+}
+
+void clear_io_state(struct thread_data *td)
+{
+       struct fio_file *f;
+       unsigned int i;
+
+       reset_io_counters(td);
+
+       close_files(td);
+       for_each_file(td, f, i)
+               fio_file_clear_done(f);
+
+       /*
+        * Set the same seed to get repeatable runs
+        */
+       td_fill_rand_seeds(td);
+}
+
+void reset_all_stats(struct thread_data *td)
+{
+       struct timeval tv;
+       int i;
+
+       reset_io_counters(td);
+
+       for (i = 0; i < 2; i++) {
+               td->io_bytes[i] = 0;
+               td->io_blocks[i] = 0;
+               td->io_issues[i] = 0;
+               td->ts.total_io_u[i] = 0;
+       }
+
+       fio_gettime(&tv, NULL);
+       td->ts.runtime[0] = 0;
+       td->ts.runtime[1] = 0;
+       memcpy(&td->epoch, &tv, sizeof(tv));
+       memcpy(&td->start, &tv, sizeof(tv));
+}
+
+void reset_fio_state(void)
+{
+       groupid = 0;
+       thread_number = 0;
+       nr_process = 0;
+       nr_thread = 0;
+       done_secs = 0;
+}
+
+const char *fio_get_os_string(int nr)
+{
+       if (nr < os_nr)
+               return fio_os_strings[nr];
+
+       return NULL;
+}
+
+const char *fio_get_arch_string(int nr)
+{
+       if (nr < arch_nr)
+               return fio_arch_strings[nr];
+
+       return NULL;
+}
+
+void td_set_runstate(struct thread_data *td, int runstate)
+{
+       if (td->runstate == runstate)
+               return;
+
+       dprint(FD_PROCESS, "pid=%d: runstate %d -> %d\n", (int) td->pid,
+                                               td->runstate, runstate);
+       td->runstate = runstate;
+}
+
+void fio_terminate_threads(int group_id)
+{
+       struct thread_data *td;
+       int i;
+
+       dprint(FD_PROCESS, "terminate group_id=%d\n", group_id);
+
+       for_each_td(td, i) {
+               if (group_id == TERMINATE_ALL || groupid == td->groupid) {
+                       dprint(FD_PROCESS, "setting terminate on %s/%d\n",
+                                               td->o.name, (int) td->pid);
+                       td->terminate = 1;
+                       td->o.start_delay = 0;
+
+                       /*
+                        * if the thread is running, just let it exit
+                        */
+                       if (!td->pid)
+                               continue;
+                       else if (td->runstate < TD_RAMP)
+                               kill(td->pid, SIGTERM);
+                       else {
+                               struct ioengine_ops *ops = td->io_ops;
+
+                               if (ops && (ops->flags & FIO_SIGTERM))
+                                       kill(td->pid, SIGTERM);
+                       }
+               }
+       }
+}
+
+
index e7107d4..9af278a 100644 (file)
--- a/server.c
+++ b/server.c
@@ -346,7 +346,7 @@ static int handle_job_cmd(struct fio_net_cmd *cmd)
        spdu.jobs = cpu_to_le32(thread_number);
        fio_net_send_cmd(server_fd, FIO_NET_CMD_START, &spdu, sizeof(spdu), 0);
 
-       ret = exec_run();
+       ret = fio_backend();
 
        epdu.error = ret;
        fio_net_send_cmd(server_fd, FIO_NET_CMD_STOP, &epdu, sizeof(epdu), 0);
@@ -390,7 +390,7 @@ static int handle_jobline_cmd(struct fio_net_cmd *cmd)
 
        fio_net_send_simple_cmd(server_fd, FIO_NET_CMD_START, 0, NULL);
 
-       ret = exec_run();
+       ret = fio_backend();
        fio_server_send_quit_cmd();
        reset_fio_state();
        return ret;