#include <dirent.h>
#include <libgen.h>
#include <sys/stat.h>
-#include <sys/mman.h>
-#include <sys/types.h>
#include "fio.h"
#include "smalloc.h"
#include "os/os.h"
#include "hash.h"
#include "lib/axmap.h"
-#include "lib/memalign.h"
+#include "rwlock.h"
+#include "zbd.h"
#ifdef CONFIG_LINUX_FALLOCATE
#include <linux/falloc.h>
#endif
-static int root_warn;
-
static FLIST_HEAD(filename_list);
/*
td->verror[0] = '\0';
}
-static inline int native_fallocate(struct thread_data *td, struct fio_file *f)
+static int native_fallocate(struct thread_data *td, struct fio_file *f)
{
bool success;
!success ? "un": "");
if (success)
- return 0;
+ return false;
if (errno == ENOSYS)
dprint(FD_FILE, "native fallocate is not implemented\n");
- return -1;
+ return true;
}
static void fallocate_file(struct thread_data *td, struct fio_file *f)
{
- int r;
-
if (td->o.fill_device)
return;
switch (td->o.fallocate_mode) {
case FIO_FALLOCATE_NATIVE:
- r = native_fallocate(td, f);
- if (r != 0 && errno != ENOSYS)
- log_err("fio: native_fallocate call failed: %s\n",
- strerror(errno));
+ native_fallocate(td, f);
break;
case FIO_FALLOCATE_NONE:
break;
#ifdef CONFIG_POSIX_FALLOCATE
- case FIO_FALLOCATE_POSIX:
+ case FIO_FALLOCATE_POSIX: {
+ int r;
+
dprint(FD_FILE, "posix_fallocate file %s size %llu\n",
f->file_name,
(unsigned long long) f->real_file_size);
if (r > 0)
log_err("fio: posix_fallocate fails: %s\n", strerror(r));
break;
+ }
#endif /* CONFIG_POSIX_FALLOCATE */
#ifdef CONFIG_LINUX_FALLOCATE
- case FIO_FALLOCATE_KEEP_SIZE:
+ case FIO_FALLOCATE_KEEP_SIZE: {
+ int r;
+
dprint(FD_FILE, "fallocate(FALLOC_FL_KEEP_SIZE) "
"file %s size %llu\n", f->file_name,
(unsigned long long) f->real_file_size);
td_verror(td, errno, "fallocate");
break;
+ }
#endif /* CONFIG_LINUX_FALLOCATE */
default:
log_err("fio: unknown fallocate mode: %d\n", td->o.fallocate_mode);
{
int new_layout = 0, unlink_file = 0, flags;
unsigned long long left;
- unsigned int bs;
+ unsigned long long bs;
char *b = NULL;
if (read_only) {
flags |= O_CREAT;
if (new_layout)
flags |= O_TRUNC;
- if (td->o.odirect)
- flags |= OS_O_DIRECT;
#ifdef WIN32
flags |= _O_BINARY;
if (bs > left)
bs = left;
- b = fio_memalign(page_size, bs);
+ b = malloc(bs);
if (!b) {
- td_verror(td, errno, "fio_memalign");
+ td_verror(td, errno, "malloc");
goto err;
}
f->io_size = f->real_file_size;
}
- fio_memfree(b, bs);
+ free(b);
done:
return 0;
err:
close(f->fd);
f->fd = -1;
if (b)
- fio_memfree(b, bs);
+ free(b);
return 1;
}
-static int pre_read_file(struct thread_data *td, struct fio_file *f)
+static bool pre_read_file(struct thread_data *td, struct fio_file *f)
{
- int ret = 0, r, did_open = 0, old_runstate;
+ int r, did_open = 0, old_runstate;
unsigned long long left;
- unsigned int bs;
+ unsigned long long bs;
+ bool ret = true;
char *b;
if (td_ioengine_flagged(td, FIO_PIPEIO) ||
td_ioengine_flagged(td, FIO_NOIO))
- return 0;
+ return true;
if (f->filetype == FIO_TYPE_CHAR)
- return 0;
+ return true;
if (!fio_file_open(f)) {
if (td->io_ops->open_file(td, f)) {
log_err("fio: cannot pre-read, failed to open file\n");
- return 1;
+ return false;
}
did_open = 1;
}
b = malloc(bs);
if (!b) {
td_verror(td, errno, "malloc");
- ret = 1;
+ ret = false;
goto error;
}
memset(b, 0, bs);
if (lseek(f->fd, f->file_offset, SEEK_SET) < 0) {
td_verror(td, errno, "lseek");
log_err("fio: failed to lseek pre-read file\n");
- ret = 1;
+ ret = false;
goto error;
}
{
unsigned long long ret, sized;
uint64_t frand_max;
- unsigned long r;
+ uint64_t r;
frand_max = rand_max(&td->file_size_state);
r = __rand(&td->file_size_state);
ret = bdev_size(td, f);
else if (f->filetype == FIO_TYPE_CHAR)
ret = char_size(td, f);
- else
- f->real_file_size = -1ULL;
+ else {
+ f->real_file_size = -1;
+ log_info("%s: failed to get file size of %s\n", td->o.name,
+ f->file_name);
+ return 1; /* avoid offset extends end error message */
+ }
/*
* Leave ->real_file_size with 0 since it could be expectation
if (ret)
return ret;
- /*
- * If ->real_file_size is -1, a conditional for the message
- * "offset extends end" is always true, but it makes no sense,
- * so just return the same value here.
- */
- if (f->real_file_size == -1ULL) {
- log_info("%s: failed to get file size of %s\n", td->o.name,
- f->file_name);
- return 1;
- }
-
- if (td->o.start_offset && f->file_offset == 0)
- dprint(FD_FILE, "offset of file %s not initialized yet\n",
- f->file_name);
/*
* ->file_offset normally hasn't been initialized yet, so this
- * is basically always false.
+ * is basically always false unless ->real_file_size is -1, but
+ * if ->real_file_size is -1 this message doesn't make sense.
+ * As a result, this message is basically useless.
*/
if (f->file_offset > f->real_file_size) {
log_err("%s: offset extends end (%llu > %llu)\n", td->o.name,
ret = td->io_ops->invalidate(td, f);
if (ret < 0)
errval = -ret;
+ } else if (td_ioengine_flagged(td, FIO_DISKLESSIO)) {
+ dprint(FD_IO, "invalidate not supported by ioengine %s\n",
+ td->io_ops->name);
} else if (f->filetype == FIO_TYPE_FILE) {
dprint(FD_IO, "declare unneeded cache %s: %llu/%llu\n",
f->file_name, off, len);
ret = blockdev_invalidate_cache(f);
}
if (ret < 0 && errno == EACCES && geteuid()) {
- if (!root_warn) {
+ if (!fio_did_warn(FIO_WARN_ROOT_FLUSH)) {
log_err("fio: only root may flush block "
"devices. Cache flush bypassed!\n");
- root_warn = 1;
}
- ret = 0;
}
if (ret < 0)
errval = errno;
} else if (f->filetype == FIO_TYPE_CHAR ||
f->filetype == FIO_TYPE_PIPE) {
dprint(FD_IO, "invalidate not supported %s\n", f->file_name);
- ret = 0;
}
/*
from_hash = file_lookup_open(f, flags);
} else if (td_trim(td)) {
assert(!td_rw(td)); /* should have matched above */
- flags |= O_RDWR;
+ if (!read_only)
+ flags |= O_RDWR;
from_hash = file_lookup_open(f, flags);
}
} else if (f->filetype != FIO_TYPE_FILE)
continue;
- buf[255] = '\0';
- strncpy(buf, f->file_name, 255);
+ snprintf(buf, ARRAY_SIZE(buf), "%s", f->file_name);
if (stat(buf, &sb) < 0) {
if (errno != ENOENT)
continue;
fm = calloc(1, sizeof(*fm));
- strncpy(fm->__base, buf, sizeof(fm->__base) - 1);
+ snprintf(fm->__base, ARRAY_SIZE(fm->__base), "%s", buf);
fm->base = basename(fm->__base);
fm->key = sb.st_dev;
flist_add(&fm->list, &list);
uint64_t get_start_offset(struct thread_data *td, struct fio_file *f)
{
+ bool align = false;
struct thread_options *o = &td->o;
unsigned long long align_bs;
unsigned long long offset;
+ unsigned long long increment;
if (o->file_append && f->filetype == FIO_TYPE_FILE)
return f->real_file_size;
+ if (o->offset_increment_percent) {
+ assert(!o->offset_increment);
+ increment = o->offset_increment_percent * f->real_file_size / 100;
+ align = true;
+ } else
+ increment = o->offset_increment;
+
if (o->start_offset_percent > 0) {
+ /* calculate the raw offset */
+ offset = (f->real_file_size * o->start_offset_percent / 100) +
+ (td->subjob_number * increment);
+
+ align = true;
+ } else {
+ /* start_offset_percent not set */
+ offset = o->start_offset +
+ td->subjob_number * increment;
+ }
+
+ if (align) {
/*
- * if blockalign is provided, find the min across read, write,
- * and trim
+ * if offset_align is provided, use it
*/
- if (fio_option_is_set(o, ba)) {
- align_bs = (unsigned long long) min(o->ba[DDIR_READ], o->ba[DDIR_WRITE]);
- align_bs = min((unsigned long long) o->ba[DDIR_TRIM], align_bs);
+ if (fio_option_is_set(o, start_offset_align)) {
+ align_bs = o->start_offset_align;
} else {
/* else take the minimum block size */
align_bs = td_min_bs(td);
}
- /* calculate the raw offset */
- offset = (f->real_file_size * o->start_offset_percent / 100) +
- (td->subjob_number * o->offset_increment);
-
/*
* block align the offset at the next available boundary at
* ceiling(offset / align_bs) * align_bs
*/
offset = (offset / align_bs + (offset % align_bs != 0)) * align_bs;
+ }
+
+ return offset;
+}
+static bool create_work_dirs(struct thread_data *td, const char *fname)
+{
+ char path[PATH_MAX];
+ char *start, *end;
+
+ if (td->o.directory) {
+ snprintf(path, PATH_MAX, "%s%c%s", td->o.directory,
+ FIO_OS_PATH_SEPARATOR, fname);
+ start = strstr(path, fname);
} else {
- /* start_offset_percent not set */
- offset = o->start_offset +
- td->subjob_number * o->offset_increment;
+ snprintf(path, PATH_MAX, "%s", fname);
+ start = path;
}
- return offset;
+ end = start;
+ while ((end = strchr(end, FIO_OS_PATH_SEPARATOR)) != NULL) {
+ if (end == start)
+ break;
+ *end = '\0';
+ errno = 0;
+#ifdef CONFIG_HAVE_MKDIR_TWO
+ if (mkdir(path, 0600) && errno != EEXIST) {
+#else
+ if (mkdir(path) && errno != EEXIST) {
+#endif
+ log_err("fio: failed to create dir (%s): %d\n",
+ start, errno);
+ return false;
+ }
+ *end = FIO_OS_PATH_SEPARATOR;
+ end++;
+ }
+ td->flags |= TD_F_DIRS_CREATED;
+ return true;
}
/*
unsigned int i, nr_fs_extra = 0;
int err = 0, need_extend;
int old_state;
- const unsigned int bs = td_min_bs(td);
+ const unsigned long long bs = td_min_bs(td);
uint64_t fs = 0;
dprint(FD_FILE, "setup files\n");
old_state = td_bump_runstate(td, TD_SETTING_UP);
- if (o->read_iolog_file)
- goto done;
+ for_each_file(td, f, i) {
+ if (!td_ioengine_flagged(td, FIO_DISKLESSIO) &&
+ strchr(f->file_name, FIO_OS_PATH_SEPARATOR) &&
+ !(td->flags & TD_F_DIRS_CREATED) &&
+ !create_work_dirs(td, f->file_name))
+ goto err_out;
+ }
/*
* Find out physical size of files or devices for this thread,
if (err)
goto err_out;
+ if (o->read_iolog_file)
+ goto done;
+
/*
* check sizes. if the files/devices do not exist and the size
* isn't passed to fio, abort.
if (f->io_size == -1ULL)
total_size = -1ULL;
else {
- if (o->size_percent) {
+ if (o->size_percent && o->size_percent != 100) {
uint64_t file_size;
file_size = f->io_size + f->file_offset;
}
if (f->filetype == FIO_TYPE_FILE &&
- (f->io_size + f->file_offset) > f->real_file_size &&
- !td_ioengine_flagged(td, FIO_DISKLESSIO)) {
- if (!o->create_on_open) {
+ (f->io_size + f->file_offset) > f->real_file_size) {
+ if (!td_ioengine_flagged(td, FIO_DISKLESSIO) &&
+ !o->create_on_open) {
need_extend++;
extend_size += (f->io_size + f->file_offset);
fio_file_set_extend(f);
- } else
+ } else if (!td_ioengine_flagged(td, FIO_DISKLESSIO) ||
+ (td_ioengine_flagged(td, FIO_DISKLESSIO) &&
+ td_ioengine_flagged(td, FIO_FAKEIO)))
f->real_file_size = f->io_size + f->file_offset;
}
}
if (err)
goto err_out;
- if (!o->zone_size)
- o->zone_size = o->size;
-
/*
* iolog already set the total io size, if we read back
* stored entries.
td->done = 1;
td_restore_runstate(td, old_state);
+
+ if (td->o.zone_mode == ZONE_MODE_ZBD) {
+ err = zbd_init(td);
+ if (err)
+ goto err_out;
+ }
return 0;
+
err_offset:
log_err("%s: you need to specify valid offset=\n", o->name);
err_out:
return 1;
}
-int pre_read_files(struct thread_data *td)
+bool pre_read_files(struct thread_data *td)
{
struct fio_file *f;
unsigned int i;
dprint(FD_FILE, "pre_read files\n");
for_each_file(td, f, i) {
- if (pre_read_file(td, f))
- return -1;
+ if (!pre_read_file(td, f))
+ return false;
}
- return 0;
+ return true;
}
-static int __init_rand_distribution(struct thread_data *td, struct fio_file *f)
+static void __init_rand_distribution(struct thread_data *td, struct fio_file *f)
{
unsigned int range_size, seed;
- unsigned long nranges;
+ uint64_t nranges;
uint64_t fsize;
range_size = min(td->o.min_bs[DDIR_READ], td->o.min_bs[DDIR_WRITE]);
fsize = min(f->real_file_size, f->io_size);
- nranges = (fsize + range_size - 1) / range_size;
+ nranges = (fsize + range_size - 1ULL) / range_size;
seed = jhash(f->file_name, strlen(f->file_name), 0) * td->thread_number;
if (!td->o.rand_repeatable)
pareto_init(&f->zipf, nranges, td->o.pareto_h.u.f, seed);
else if (td->o.random_distribution == FIO_RAND_DIST_GAUSS)
gauss_init(&f->gauss, nranges, td->o.gauss_dev.u.f, seed);
-
- return 1;
}
-static int init_rand_distribution(struct thread_data *td)
+static bool init_rand_distribution(struct thread_data *td)
{
struct fio_file *f;
unsigned int i;
int state;
if (td->o.random_distribution == FIO_RAND_DIST_RANDOM)
- return 0;
+ return false;
state = td_bump_runstate(td, TD_SETTING_UP);
__init_rand_distribution(td, f);
td_restore_runstate(td, state);
-
- return 1;
+ return true;
}
/*
return 0;
}
-int init_random_map(struct thread_data *td)
+bool init_random_map(struct thread_data *td)
{
unsigned long long blocks;
struct fio_file *f;
unsigned int i;
if (init_rand_distribution(td))
- return 0;
+ return true;
if (!td_random(td))
- return 0;
+ return true;
for_each_file(td, f, i) {
uint64_t fsize = min(f->real_file_size, f->io_size);
+ if (td->o.zone_mode == ZONE_MODE_STRIDED)
+ fsize = td->o.zone_range;
+
blocks = fsize / (unsigned long long) td->o.rw_min_bs;
if (check_rand_gen_limits(td, f, blocks))
- return 1;
+ return false;
if (td->o.random_generator == FIO_RAND_GEN_LFSR) {
- unsigned long seed;
+ uint64_t seed;
seed = td->rand_seeds[FIO_RAND_BLOCK_OFF];
if (!lfsr_init(&f->lfsr, blocks, seed, 0)) {
fio_file_set_lfsr(f);
continue;
+ } else {
+ log_err("fio: failed initializing LFSR\n");
+ return false;
}
} else if (!td->o.norandommap) {
f->io_axmap = axmap_new(blocks);
" a large number of jobs, try the 'norandommap'"
" option or set 'softrandommap'. Or give"
" a larger --alloc-size to fio.\n");
- return 1;
+ return false;
}
log_info("fio: file %s failed allocating random map. Running "
"job without.\n", f->file_name);
}
- return 0;
+ return true;
}
void close_files(struct thread_data *td)
{
struct fio_file *f;
unsigned int i;
+ bool use_free = td_ioengine_flagged(td, FIO_NOFILEHASH);
dprint(FD_FILE, "close files\n");
td_io_unlink_file(td, f);
}
- sfree(f->file_name);
+ zbd_free_zone_info(f);
+
+ if (use_free)
+ free(f->file_name);
+ else
+ sfree(f->file_name);
f->file_name = NULL;
if (fio_file_axmap(f)) {
axmap_free(f->io_axmap);
f->io_axmap = NULL;
}
- sfree(f);
+ if (use_free)
+ free(f);
+ else
+ sfree(f);
}
td->o.filename = NULL;
{
struct fio_file *f;
- f = smalloc(sizeof(*f));
+ if (td_ioengine_flagged(td, FIO_NOFILEHASH))
+ f = calloc(1, sizeof(*f));
+ else
+ f = scalloc(1, sizeof(*f));
if (!f) {
assert(0);
return NULL;
if (td->io_ops && td_ioengine_flagged(td, FIO_DISKLESSIO))
f->real_file_size = -1ULL;
- f->file_name = smalloc_strdup(file_name);
- if (!f->file_name)
- assert(0);
+ if (td_ioengine_flagged(td, FIO_NOFILEHASH))
+ f->file_name = strdup(file_name);
+ else
+ f->file_name = smalloc_strdup(file_name);
+
+ /* can't handle smalloc failure from here */
+ assert(f->file_name);
get_file_type(f);
f->rwlock = fio_rwlock_init();
break;
case FILE_LOCK_EXCLUSIVE:
- f->lock = fio_mutex_init(FIO_MUTEX_UNLOCKED);
+ f->lock = fio_sem_init(FIO_SEM_UNLOCKED);
break;
default:
log_err("fio: unknown lock mode: %d\n", td->o.file_lock_mode);
}
td->files_index++;
- if (f->filetype == FIO_TYPE_FILE)
- td->nr_normal_files++;
- set_already_allocated(file_name);
+ if (td->o.numjobs > 1)
+ set_already_allocated(file_name);
if (inc)
td->o.nr_files++;
if (--f->references)
return 0;
+ disk_util_dec(f->du);
+
+ if (td->o.file_lock_mode != FILE_LOCK_NONE)
+ unlock_file_all(td, f);
+
if (should_fsync(td) && td->o.fsync_on_close) {
f_ret = fsync(f->fd);
if (f_ret < 0)
ret = f_ret;
td->nr_open_files--;
+ fio_file_clear_closing(f);
fio_file_clear_open(f);
assert(f->fd == -1);
return ret;
else
fio_rwlock_write(f->rwlock);
} else if (td->o.file_lock_mode == FILE_LOCK_EXCLUSIVE)
- fio_mutex_down(f->lock);
+ fio_sem_down(f->lock);
td->file_locks[f->fileno] = td->o.file_lock_mode;
}
if (td->o.file_lock_mode == FILE_LOCK_READWRITE)
fio_rwlock_unlock(f->rwlock);
else if (td->o.file_lock_mode == FILE_LOCK_EXCLUSIVE)
- fio_mutex_up(f->lock);
+ fio_sem_up(f->lock);
td->file_locks[f->fileno] = FILE_LOCK_NONE;
}
unlock_file(td, f);
}
-static int recurse_dir(struct thread_data *td, const char *dirname)
+static bool recurse_dir(struct thread_data *td, const char *dirname)
{
struct dirent *dir;
- int ret = 0;
+ bool ret = false;
DIR *D;
D = opendir(dirname);
snprintf(buf, FIO_VERROR_SIZE, "opendir(%s)", dirname);
td_verror(td, errno, buf);
- return 1;
+ return true;
}
while ((dir = readdir(D)) != NULL) {
if (!strcmp(dir->d_name, ".") || !strcmp(dir->d_name, ".."))
continue;
- sprintf(full_path, "%s%s%s", dirname, FIO_OS_PATH_SEPARATOR, dir->d_name);
+ sprintf(full_path, "%s%c%s", dirname, FIO_OS_PATH_SEPARATOR, dir->d_name);
if (lstat(full_path, &sb) == -1) {
if (errno != ENOENT) {
td_verror(td, errno, "stat");
- ret = 1;
+ ret = true;
break;
}
}
__f = alloc_new_file(td);
if (f->file_name) {
- __f->file_name = smalloc_strdup(f->file_name);
- if (!__f->file_name)
- assert(0);
+ if (td_ioengine_flagged(td, FIO_NOFILEHASH))
+ __f->file_name = strdup(f->file_name);
+ else
+ __f->file_name = smalloc_strdup(f->file_name);
+ /* can't handle smalloc failure from here */
+ assert(__f->file_name);
__f->filetype = f->filetype;
}
td->o.nr_files = 0;
td->o.open_files = 0;
td->files_index = 0;
- td->nr_normal_files = 0;
}
void fio_file_reset(struct thread_data *td, struct fio_file *f)
axmap_reset(f->io_axmap);
else if (fio_file_lfsr(f))
lfsr_reset(&f->lfsr, td->rand_seeds[FIO_RAND_BLOCK_OFF]);
+
+ zbd_file_reset(td, f);
}
bool fio_files_done(struct thread_data *td)
{
free_already_allocated();
}
+
+/*
+ * This function is for platforms which support direct I/O but not O_DIRECT.
+ */
+int fio_set_directio(struct thread_data *td, struct fio_file *f)
+{
+#ifdef FIO_OS_DIRECTIO
+ int ret = fio_set_odirect(f);
+
+ if (ret) {
+ td_verror(td, ret, "fio_set_directio");
+#if defined(__sun__)
+ if (ret == ENOTTY) { /* ENOTTY suggests RAW device or ZFS */
+ log_err("fio: doing directIO to RAW devices or ZFS not supported\n");
+ } else {
+ log_err("fio: the file system does not seem to support direct IO\n");
+ }
+#else
+ log_err("fio: the file system does not seem to support direct IO\n");
+#endif
+ return -1;
+ }
+
+ return 0;
+#else
+ log_err("fio: direct IO is not supported on this host operating system\n");
+ return -1;
+#endif
+}