#include <errno.h>
#include <string.h>
#include <stdlib.h>
-#include <dirent.h>
#include <fcntl.h>
-#include <sys/ioctl.h>
#include <sys/stat.h>
#include <unistd.h>
-#include <linux/blkzoned.h>
+#include "os/os.h"
#include "file.h"
#include "fio.h"
#include "lib/pow2.h"
#include "oslib/asprintf.h"
#include "smalloc.h"
#include "verify.h"
+#include "pshared.h"
#include "zbd.h"
+/**
+ * zbd_get_zoned_model - Get a device zoned model
+ * @td: FIO thread data
+ * @f: FIO file for which to get model information
+ */
+int zbd_get_zoned_model(struct thread_data *td, struct fio_file *f,
+ enum zbd_zoned_model *model)
+{
+ int ret;
+
+ if (td->io_ops && td->io_ops->get_zoned_model)
+ ret = td->io_ops->get_zoned_model(td, f, model);
+ else
+ ret = blkzoned_get_zoned_model(td, f, model);
+ if (ret < 0) {
+ td_verror(td, errno, "get zoned model failed");
+ log_err("%s: get zoned model failed (%d).\n",
+ f->file_name, errno);
+ }
+
+ return ret;
+}
+
+/**
+ * zbd_report_zones - Get zone information
+ * @td: FIO thread data.
+ * @f: FIO file for which to get zone information
+ * @offset: offset from which to report zones
+ * @zones: Array of struct zbd_zone
+ * @nr_zones: Size of @zones array
+ *
+ * Get zone information into @zones starting from the zone at offset @offset
+ * for the device specified by @f.
+ *
+ * Returns the number of zones reported upon success and a negative error code
+ * upon failure. If the zone report is empty, always assume an error (device
+ * problem) and return -EIO.
+ */
+int zbd_report_zones(struct thread_data *td, struct fio_file *f,
+ uint64_t offset, struct zbd_zone *zones,
+ unsigned int nr_zones)
+{
+ int ret;
+
+ if (td->io_ops && td->io_ops->report_zones)
+ ret = td->io_ops->report_zones(td, f, offset, zones, nr_zones);
+ else
+ ret = blkzoned_report_zones(td, f, offset, zones, nr_zones);
+ if (ret < 0) {
+ td_verror(td, errno, "report zones failed");
+ log_err("%s: report zones from sector %llu failed (%d).\n",
+ f->file_name, (unsigned long long)offset >> 9, errno);
+ } else if (ret == 0) {
+ td_verror(td, errno, "Empty zone report");
+ log_err("%s: report zones from sector %llu is empty.\n",
+ f->file_name, (unsigned long long)offset >> 9);
+ ret = -EIO;
+ }
+
+ return ret;
+}
+
+/**
+ * zbd_reset_wp - reset the write pointer of a range of zones
+ * @td: FIO thread data.
+ * @f: FIO file for which to reset zones
+ * @offset: Starting offset of the first zone to reset
+ * @length: Length of the range of zones to reset
+ *
+ * Reset the write pointer of all zones in the range @offset...@offset+@length.
+ * Returns 0 upon success and a negative error code upon failure.
+ */
+int zbd_reset_wp(struct thread_data *td, struct fio_file *f,
+ uint64_t offset, uint64_t length)
+{
+ int ret;
+
+ if (td->io_ops && td->io_ops->reset_wp)
+ ret = td->io_ops->reset_wp(td, f, offset, length);
+ else
+ ret = blkzoned_reset_wp(td, f, offset, length);
+ if (ret < 0) {
+ td_verror(td, errno, "resetting wp failed");
+ log_err("%s: resetting wp for %llu sectors at sector %llu failed (%d).\n",
+ f->file_name, (unsigned long long)length >> 9,
+ (unsigned long long)offset >> 9, errno);
+ }
+
+ return ret;
+}
+
/**
* zbd_zone_idx - convert an offset into a zone number
* @f: file pointer.
return min(zone_idx, f->zbd_info->nr_zones);
}
+/**
+ * zbd_zone_end - Return zone end location
+ * @z: zone info pointer.
+ */
+static inline uint64_t zbd_zone_end(const struct fio_zone_info *z)
+{
+ return (z+1)->start;
+}
+
+/**
+ * zbd_zone_capacity_end - Return zone capacity limit end location
+ * @z: zone info pointer.
+ */
+static inline uint64_t zbd_zone_capacity_end(const struct fio_zone_info *z)
+{
+ return z->start + z->capacity;
+}
+
/**
* zbd_zone_full - verify whether a minimum number of bytes remain in a zone
* @f: file pointer.
{
assert((required & 511) == 0);
- return z->type == BLK_ZONE_TYPE_SEQWRITE_REQ &&
- z->wp + required > z->start + f->zbd_info->zone_size;
+ return z->has_wp &&
+ z->wp + required > zbd_zone_capacity_end(z);
+}
+
+static void zone_lock(struct thread_data *td, const struct fio_file *f,
+ struct fio_zone_info *z)
+{
+ struct zoned_block_device_info *zbd = f->zbd_info;
+ uint32_t nz = z - zbd->zone_info;
+
+ /* A thread should never lock zones outside its working area. */
+ assert(f->min_zone <= nz && nz < f->max_zone);
+
+ assert(z->has_wp);
+
+ /*
+ * Lock the io_u target zone. The zone will be unlocked if io_u offset
+ * is changed or when io_u completes and zbd_put_io() executed.
+ * To avoid multiple jobs doing asynchronous I/Os from deadlocking each
+ * other waiting for zone locks when building an io_u batch, first
+ * only trylock the zone. If the zone is already locked by another job,
+ * process the currently queued I/Os so that I/O progress is made and
+ * zones unlocked.
+ */
+ if (pthread_mutex_trylock(&z->mutex) != 0) {
+ if (!td_ioengine_flagged(td, FIO_SYNCIO))
+ io_u_quiesce(td);
+ pthread_mutex_lock(&z->mutex);
+ }
+}
+
+static inline void zone_unlock(struct fio_zone_info *z)
+{
+ int ret;
+
+ assert(z->has_wp);
+ ret = pthread_mutex_unlock(&z->mutex);
+ assert(!ret);
}
static bool is_valid_offset(const struct fio_file *f, uint64_t offset)
return (uint64_t)(offset - f->file_offset) < f->io_size;
}
+static inline struct fio_zone_info *get_zone(const struct fio_file *f,
+ unsigned int zone_nr)
+{
+ return &f->zbd_info->zone_info[zone_nr];
+}
+
/* Verify whether direct I/O is used for all host-managed zoned drives. */
static bool zbd_using_direct_io(void)
{
continue;
for_each_file(td, f, j) {
if (f->zbd_info &&
- f->zbd_info->model == ZBD_DM_HOST_MANAGED)
+ f->zbd_info->model == ZBD_HOST_MANAGED)
return false;
}
}
zone_idx_b = zbd_zone_idx(f, f->file_offset);
zone_idx_e = zbd_zone_idx(f, f->file_offset + f->io_size - 1);
for (zone_idx = zone_idx_b; zone_idx <= zone_idx_e; zone_idx++)
- if (f->zbd_info->zone_info[zone_idx].type ==
- BLK_ZONE_TYPE_SEQWRITE_REQ)
+ if (get_zone(f, zone_idx)->has_wp)
return true;
return false;
return false;
}
- if (td->o.zone_skip &&
- (td->o.zone_skip < td->o.zone_size ||
- td->o.zone_skip % td->o.zone_size)) {
+ if (td->o.zone_skip % td->o.zone_size) {
log_err("%s: zoneskip %llu is not a multiple of the device zone size %llu.\n",
f->file_name, (unsigned long long) td->o.zone_skip,
(unsigned long long) td->o.zone_size);
}
zone_idx = zbd_zone_idx(f, f->file_offset);
- z = &f->zbd_info->zone_info[zone_idx];
- if (f->file_offset != z->start) {
- new_offset = (z+1)->start;
+ z = get_zone(f, zone_idx);
+ if ((f->file_offset != z->start) &&
+ (td->o.td_ddir != TD_DDIR_READ)) {
+ new_offset = zbd_zone_end(z);
if (new_offset >= f->file_offset + f->io_size) {
log_info("%s: io_size must be at least one zone\n",
f->file_name);
f->file_offset = new_offset;
}
zone_idx = zbd_zone_idx(f, f->file_offset + f->io_size);
- z = &f->zbd_info->zone_info[zone_idx];
+ z = get_zone(f, zone_idx);
new_end = z->start;
- if (f->file_offset + f->io_size != new_end) {
+ if ((td->o.td_ddir != TD_DDIR_READ) &&
+ (f->file_offset + f->io_size != new_end)) {
if (new_end <= f->file_offset) {
log_info("%s: io_size must be at least one zone\n",
f->file_name);
{
struct thread_data *td;
struct fio_file *f;
- uint32_t zone_size;
int i, j, k;
for_each_td(td, i) {
for_each_file(td, f, j) {
+ uint64_t zone_size;
+
if (!f->zbd_info)
continue;
zone_size = f->zbd_info->zone_size;
- for (k = 0; k < ARRAY_SIZE(td->o.bs); k++) {
+ for (k = 0; k < FIO_ARRAY_SIZE(td->o.bs); k++) {
if (td->o.verify != VERIFY_NONE &&
zone_size % td->o.bs[k] != 0) {
- log_info("%s: block size %llu is not a divisor of the zone size %d\n",
+ log_info("%s: block size %llu is not a divisor of the zone size %llu\n",
f->file_name, td->o.bs[k],
- zone_size);
+ (unsigned long long)zone_size);
return false;
}
}
return true;
}
-/*
- * Read zone information into @buf starting from sector @start_sector.
- * @fd is a file descriptor that refers to a block device and @bufsz is the
- * size of @buf.
- *
- * Returns 0 upon success and a negative error code upon failure.
- * If the zone report is empty, always assume an error (device problem) and
- * return -EIO.
- */
-static int read_zone_info(int fd, uint64_t start_sector,
- void *buf, unsigned int bufsz)
-{
- struct blk_zone_report *hdr = buf;
- int ret;
-
- if (bufsz < sizeof(*hdr))
- return -EINVAL;
-
- memset(hdr, 0, sizeof(*hdr));
-
- hdr->nr_zones = (bufsz - sizeof(*hdr)) / sizeof(struct blk_zone);
- hdr->sector = start_sector;
- ret = ioctl(fd, BLKREPORTZONE, hdr);
- if (ret)
- return -errno;
- if (!hdr->nr_zones)
- return -EIO;
- return 0;
-}
-
-/*
- * Read up to 255 characters from the first line of a file. Strip the trailing
- * newline.
- */
-static char *read_file(const char *path)
-{
- char line[256], *p = line;
- FILE *f;
-
- f = fopen(path, "rb");
- if (!f)
- return NULL;
- if (!fgets(line, sizeof(line), f))
- line[0] = '\0';
- strsep(&p, "\n");
- fclose(f);
-
- return strdup(line);
-}
-
-static enum blk_zoned_model get_zbd_model(const char *file_name)
-{
- enum blk_zoned_model model = ZBD_DM_NONE;
- char *zoned_attr_path = NULL;
- char *model_str = NULL;
- struct stat statbuf;
- char *sys_devno_path = NULL;
- char *part_attr_path = NULL;
- char *part_str = NULL;
- char sys_path[PATH_MAX];
- ssize_t sz;
- char *delim = NULL;
-
- if (stat(file_name, &statbuf) < 0)
- goto out;
-
- if (asprintf(&sys_devno_path, "/sys/dev/block/%d:%d",
- major(statbuf.st_rdev), minor(statbuf.st_rdev)) < 0)
- goto out;
-
- sz = readlink(sys_devno_path, sys_path, sizeof(sys_path) - 1);
- if (sz < 0)
- goto out;
- sys_path[sz] = '\0';
-
- /*
- * If the device is a partition device, cut the device name in the
- * canonical sysfs path to obtain the sysfs path of the holder device.
- * e.g.: /sys/devices/.../sda/sda1 -> /sys/devices/.../sda
- */
- if (asprintf(&part_attr_path, "/sys/dev/block/%s/partition",
- sys_path) < 0)
- goto out;
- part_str = read_file(part_attr_path);
- if (part_str && *part_str == '1') {
- delim = strrchr(sys_path, '/');
- if (!delim)
- goto out;
- *delim = '\0';
- }
-
- if (asprintf(&zoned_attr_path,
- "/sys/dev/block/%s/queue/zoned", sys_path) < 0)
- goto out;
-
- model_str = read_file(zoned_attr_path);
- if (!model_str)
- goto out;
- dprint(FD_ZBD, "%s: zbd model string: %s\n", file_name, model_str);
- if (strcmp(model_str, "host-aware") == 0)
- model = ZBD_DM_HOST_AWARE;
- else if (strcmp(model_str, "host-managed") == 0)
- model = ZBD_DM_HOST_MANAGED;
-
-out:
- free(model_str);
- free(zoned_attr_path);
- free(part_str);
- free(part_attr_path);
- free(sys_devno_path);
- return model;
-}
-
static int ilog2(uint64_t i)
{
int log = -1;
uint32_t nr_zones;
struct fio_zone_info *p;
uint64_t zone_size = td->o.zone_size;
+ uint64_t zone_capacity = td->o.zone_capacity;
struct zoned_block_device_info *zbd_info = NULL;
- pthread_mutexattr_t attr;
int i;
if (zone_size == 0) {
return 1;
}
+ if (zone_capacity == 0)
+ zone_capacity = zone_size;
+
+ if (zone_capacity > zone_size) {
+ log_err("%s: job parameter zonecapacity %llu is larger than zone size %llu\n",
+ f->file_name, (unsigned long long) td->o.zone_capacity,
+ (unsigned long long) td->o.zone_size);
+ return 1;
+ }
+
nr_zones = (f->real_file_size + zone_size - 1) / zone_size;
zbd_info = scalloc(1, sizeof(*zbd_info) +
(nr_zones + 1) * sizeof(zbd_info->zone_info[0]));
if (!zbd_info)
return -ENOMEM;
- pthread_mutexattr_init(&attr);
- pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_RECURSIVE);
- pthread_mutexattr_setpshared(&attr, true);
- pthread_mutex_init(&zbd_info->mutex, &attr);
+ mutex_init_pshared(&zbd_info->mutex);
zbd_info->refcount = 1;
p = &zbd_info->zone_info[0];
for (i = 0; i < nr_zones; i++, p++) {
- pthread_mutex_init(&p->mutex, &attr);
+ mutex_init_pshared_with_type(&p->mutex,
+ PTHREAD_MUTEX_RECURSIVE);
p->start = i * zone_size;
- p->wp = p->start + zone_size;
- p->type = BLK_ZONE_TYPE_SEQWRITE_REQ;
- p->cond = BLK_ZONE_COND_EMPTY;
+ p->wp = p->start;
+ p->type = ZBD_ZONE_TYPE_SWR;
+ p->cond = ZBD_ZONE_COND_EMPTY;
+ p->capacity = zone_capacity;
+ p->has_wp = 1;
}
/* a sentinel */
p->start = nr_zones * zone_size;
f->zbd_info = zbd_info;
f->zbd_info->zone_size = zone_size;
f->zbd_info->zone_size_log2 = is_power_of_2(zone_size) ?
- ilog2(zone_size) : -1;
+ ilog2(zone_size) : 0;
f->zbd_info->nr_zones = nr_zones;
- pthread_mutexattr_destroy(&attr);
return 0;
}
/*
- * Parse the BLKREPORTZONE output and store it in f->zbd_info. Must be called
- * only for devices that support this ioctl, namely zoned block devices.
+ * Maximum number of zones to report in one operation.
+ */
+#define ZBD_REPORT_MAX_ZONES 8192U
+
+/*
+ * Parse the device zone report and store it in f->zbd_info. Must be called
+ * only for devices that are zoned, namely those with a model != ZBD_NONE.
*/
static int parse_zone_info(struct thread_data *td, struct fio_file *f)
{
- const unsigned int bufsz = sizeof(struct blk_zone_report) +
- 4096 * sizeof(struct blk_zone);
- uint32_t nr_zones;
- struct blk_zone_report *hdr;
- const struct blk_zone *z;
+ int nr_zones, nrz;
+ struct zbd_zone *zones, *z;
struct fio_zone_info *p;
- uint64_t zone_size, start_sector;
+ uint64_t zone_size, offset;
struct zoned_block_device_info *zbd_info = NULL;
- pthread_mutexattr_t attr;
- void *buf;
- int fd, i, j, ret = 0;
+ int i, j, ret = -ENOMEM;
- pthread_mutexattr_init(&attr);
- pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_RECURSIVE);
- pthread_mutexattr_setpshared(&attr, true);
-
- buf = malloc(bufsz);
- if (!buf)
+ zones = calloc(ZBD_REPORT_MAX_ZONES, sizeof(struct zbd_zone));
+ if (!zones)
goto out;
- fd = open(f->file_name, O_RDONLY | O_LARGEFILE);
- if (fd < 0) {
- ret = -errno;
- goto free;
+ nrz = zbd_report_zones(td, f, 0, zones, ZBD_REPORT_MAX_ZONES);
+ if (nrz < 0) {
+ ret = nrz;
+ log_info("fio: report zones (offset 0) failed for %s (%d).\n",
+ f->file_name, -ret);
+ goto out;
}
- ret = read_zone_info(fd, 0, buf, bufsz);
- if (ret < 0) {
- log_info("fio: BLKREPORTZONE(%lu) failed for %s (%d).\n",
- 0UL, f->file_name, -ret);
- goto close;
- }
- hdr = buf;
- if (hdr->nr_zones < 1) {
- log_info("fio: %s has invalid zone information.\n",
- f->file_name);
- goto close;
- }
- z = (void *)(hdr + 1);
- zone_size = z->len << 9;
+ zone_size = zones[0].len;
nr_zones = (f->real_file_size + zone_size - 1) / zone_size;
if (td->o.zone_size == 0) {
f->file_name, (unsigned long long) td->o.zone_size,
(unsigned long long) zone_size);
ret = -EINVAL;
- goto close;
+ goto out;
}
dprint(FD_ZBD, "Device %s has %d zones of size %llu KB\n", f->file_name,
zbd_info = scalloc(1, sizeof(*zbd_info) +
(nr_zones + 1) * sizeof(zbd_info->zone_info[0]));
- ret = -ENOMEM;
if (!zbd_info)
- goto close;
- pthread_mutex_init(&zbd_info->mutex, &attr);
+ goto out;
+ mutex_init_pshared(&zbd_info->mutex);
zbd_info->refcount = 1;
p = &zbd_info->zone_info[0];
- for (start_sector = 0, j = 0; j < nr_zones;) {
- z = (void *)(hdr + 1);
- for (i = 0; i < hdr->nr_zones; i++, j++, z++, p++) {
- pthread_mutex_init(&p->mutex, &attr);
- p->start = z->start << 9;
+ for (offset = 0, j = 0; j < nr_zones;) {
+ z = &zones[0];
+ for (i = 0; i < nrz; i++, j++, z++, p++) {
+ mutex_init_pshared_with_type(&p->mutex,
+ PTHREAD_MUTEX_RECURSIVE);
+ p->start = z->start;
+ p->capacity = z->capacity;
switch (z->cond) {
- case BLK_ZONE_COND_NOT_WP:
- case BLK_ZONE_COND_FULL:
- p->wp = p->start + zone_size;
+ case ZBD_ZONE_COND_NOT_WP:
+ case ZBD_ZONE_COND_FULL:
+ p->wp = p->start + p->capacity;
break;
default:
assert(z->start <= z->wp);
- assert(z->wp <= z->start + (zone_size >> 9));
- p->wp = z->wp << 9;
+ assert(z->wp <= z->start + zone_size);
+ p->wp = z->wp;
break;
}
+
+ switch (z->type) {
+ case ZBD_ZONE_TYPE_SWR:
+ p->has_wp = 1;
+ break;
+ default:
+ p->has_wp = 0;
+ }
p->type = z->type;
p->cond = z->cond;
+
if (j > 0 && p->start != p[-1].start + zone_size) {
log_info("%s: invalid zone data\n",
f->file_name);
ret = -EINVAL;
- goto close;
+ goto out;
}
}
z--;
- start_sector = z->start + z->len;
+ offset = z->start + z->len;
if (j >= nr_zones)
break;
- ret = read_zone_info(fd, start_sector, buf, bufsz);
- if (ret < 0) {
- log_info("fio: BLKREPORTZONE(%llu) failed for %s (%d).\n",
- (unsigned long long) start_sector, f->file_name, -ret);
- goto close;
+ nrz = zbd_report_zones(td, f, offset, zones,
+ min((uint32_t)(nr_zones - j),
+ ZBD_REPORT_MAX_ZONES));
+ if (nrz < 0) {
+ ret = nrz;
+ log_info("fio: report zones (offset %llu) failed for %s (%d).\n",
+ (unsigned long long)offset,
+ f->file_name, -ret);
+ goto out;
}
}
+
/* a sentinel */
- zbd_info->zone_info[nr_zones].start = start_sector << 9;
+ zbd_info->zone_info[nr_zones].start = offset;
f->zbd_info = zbd_info;
f->zbd_info->zone_size = zone_size;
f->zbd_info->zone_size_log2 = is_power_of_2(zone_size) ?
- ilog2(zone_size) : -1;
+ ilog2(zone_size) : 0;
f->zbd_info->nr_zones = nr_zones;
zbd_info = NULL;
ret = 0;
-close:
- sfree(zbd_info);
- close(fd);
-free:
- free(buf);
out:
- pthread_mutexattr_destroy(&attr);
+ sfree(zbd_info);
+ free(zones);
return ret;
}
*/
static int zbd_create_zone_info(struct thread_data *td, struct fio_file *f)
{
- enum blk_zoned_model zbd_model;
- int ret = 0;
+ enum zbd_zoned_model zbd_model;
+ int ret;
assert(td->o.zone_mode == ZONE_MODE_ZBD);
- zbd_model = get_zbd_model(f->file_name);
+ ret = zbd_get_zoned_model(td, f, &zbd_model);
+ if (ret)
+ return ret;
+
switch (zbd_model) {
- case ZBD_DM_HOST_AWARE:
- case ZBD_DM_HOST_MANAGED:
+ case ZBD_IGNORE:
+ return 0;
+ case ZBD_HOST_AWARE:
+ case ZBD_HOST_MANAGED:
ret = parse_zone_info(td, f);
break;
- case ZBD_DM_NONE:
+ case ZBD_NONE:
ret = init_zone_info(td, f);
break;
+ default:
+ td_verror(td, EINVAL, "Unsupported zoned model");
+ log_err("Unsupported zoned model\n");
+ return -EINVAL;
}
- if (ret == 0)
+
+ if (ret == 0) {
f->zbd_info->model = zbd_model;
+ f->zbd_info->max_open_zones = td->o.max_open_zones;
+ }
return ret;
}
{
uint32_t refcount;
- if (!f->zbd_info)
- return;
+ assert(f->zbd_info);
pthread_mutex_lock(&f->zbd_info->mutex);
refcount = --f->zbd_info->refcount;
return ret;
}
-int zbd_init(struct thread_data *td)
+static bool zbd_open_zone(struct thread_data *td, const struct fio_file *f,
+ uint32_t zone_idx);
+static int zbd_reset_zone(struct thread_data *td, struct fio_file *f,
+ struct fio_zone_info *z);
+
+int zbd_init_files(struct thread_data *td)
{
struct fio_file *f;
int i;
for_each_file(td, f, i) {
- if (f->filetype != FIO_TYPE_BLOCK)
- continue;
if (zbd_init_zone_info(td, f))
return 1;
}
+ return 0;
+}
+
+void zbd_recalc_options_with_zone_granularity(struct thread_data *td)
+{
+ struct fio_file *f;
+ int i;
+
+ for_each_file(td, f, i) {
+ struct zoned_block_device_info *zbd = f->zbd_info;
+ // zonemode=strided doesn't get per-file zone size.
+ uint64_t zone_size = zbd ? zbd->zone_size : td->o.zone_size;
+
+ if (zone_size == 0)
+ continue;
+
+ if (td->o.size_nz > 0) {
+ td->o.size = td->o.size_nz * zone_size;
+ }
+ if (td->o.io_size_nz > 0) {
+ td->o.io_size = td->o.io_size_nz * zone_size;
+ }
+ if (td->o.start_offset_nz > 0) {
+ td->o.start_offset = td->o.start_offset_nz * zone_size;
+ }
+ if (td->o.offset_increment_nz > 0) {
+ td->o.offset_increment = td->o.offset_increment_nz * zone_size;
+ }
+ if (td->o.zone_skip_nz > 0) {
+ td->o.zone_skip = td->o.zone_skip_nz * zone_size;
+ }
+ }
+}
+
+int zbd_setup_files(struct thread_data *td)
+{
+ struct fio_file *f;
+ int i;
if (!zbd_using_direct_io()) {
log_err("Using direct I/O is mandatory for writing to ZBD drives\n\n");
if (!zbd_verify_bs())
return 1;
+ for_each_file(td, f, i) {
+ struct zoned_block_device_info *zbd = f->zbd_info;
+ struct fio_zone_info *z;
+ int zi;
+
+ if (!zbd)
+ continue;
+
+ f->min_zone = zbd_zone_idx(f, f->file_offset);
+ f->max_zone = zbd_zone_idx(f, f->file_offset + f->io_size);
+
+ /*
+ * When all zones in the I/O range are conventional, io_size
+ * can be smaller than zone size, making min_zone the same
+ * as max_zone. This is why the assert below needs to be made
+ * conditional.
+ */
+ if (zbd_is_seq_job(f))
+ assert(f->min_zone < f->max_zone);
+
+ zbd->max_open_zones = zbd->max_open_zones ?: ZBD_MAX_OPEN_ZONES;
+
+ if (td->o.max_open_zones > 0 &&
+ zbd->max_open_zones != td->o.max_open_zones) {
+ log_err("Different 'max_open_zones' values\n");
+ return 1;
+ }
+ if (zbd->max_open_zones > ZBD_MAX_OPEN_ZONES) {
+ log_err("'max_open_zones' value is limited by %u\n", ZBD_MAX_OPEN_ZONES);
+ return 1;
+ }
+
+ for (zi = f->min_zone; zi < f->max_zone; zi++) {
+ z = &zbd->zone_info[zi];
+ if (z->cond != ZBD_ZONE_COND_IMP_OPEN &&
+ z->cond != ZBD_ZONE_COND_EXP_OPEN)
+ continue;
+ if (zbd_open_zone(td, f, zi))
+ continue;
+ /*
+ * If the number of open zones exceeds specified limits,
+ * reset all extra open zones.
+ */
+ if (zbd_reset_zone(td, f, z) < 0) {
+ log_err("Failed to reest zone %d\n", zi);
+ return 1;
+ }
+ }
+ }
+
return 0;
}
+static inline unsigned int zbd_zone_nr(const struct fio_file *f,
+ struct fio_zone_info *zone)
+{
+ return zone - f->zbd_info->zone_info;
+}
+
/**
- * zbd_reset_range - reset zones for a range of sectors
+ * zbd_reset_zone - reset the write pointer of a single zone
* @td: FIO thread data.
- * @f: Fio file for which to reset zones
- * @sector: Starting sector in units of 512 bytes
- * @nr_sectors: Number of sectors in units of 512 bytes
+ * @f: FIO file associated with the disk for which to reset a write pointer.
+ * @z: Zone to reset.
*
* Returns 0 upon success and a negative error code upon failure.
+ *
+ * The caller must hold z->mutex.
*/
-static int zbd_reset_range(struct thread_data *td, const struct fio_file *f,
- uint64_t offset, uint64_t length)
+static int zbd_reset_zone(struct thread_data *td, struct fio_file *f,
+ struct fio_zone_info *z)
{
- struct blk_zone_range zr = {
- .sector = offset >> 9,
- .nr_sectors = length >> 9,
- };
- uint32_t zone_idx_b, zone_idx_e;
- struct fio_zone_info *zb, *ze, *z;
+ uint64_t offset = z->start;
+ uint64_t length = (z+1)->start - offset;
+ uint64_t data_in_zone = z->wp - z->start;
int ret = 0;
- assert(f->fd != -1);
+ if (!data_in_zone)
+ return 0;
+
assert(is_valid_offset(f, offset + length - 1));
+
+ dprint(FD_ZBD, "%s: resetting wp of zone %u.\n", f->file_name,
+ zbd_zone_nr(f, z));
switch (f->zbd_info->model) {
- case ZBD_DM_HOST_AWARE:
- case ZBD_DM_HOST_MANAGED:
- ret = ioctl(f->fd, BLKRESETZONE, &zr);
- if (ret < 0) {
- td_verror(td, errno, "resetting wp failed");
- log_err("%s: resetting wp for %llu sectors at sector %llu failed (%d).\n",
- f->file_name, zr.nr_sectors, zr.sector, errno);
+ case ZBD_HOST_AWARE:
+ case ZBD_HOST_MANAGED:
+ ret = zbd_reset_wp(td, f, offset, length);
+ if (ret < 0)
return ret;
- }
break;
- case ZBD_DM_NONE:
+ default:
break;
}
- zone_idx_b = zbd_zone_idx(f, offset);
- zb = &f->zbd_info->zone_info[zone_idx_b];
- zone_idx_e = zbd_zone_idx(f, offset + length);
- ze = &f->zbd_info->zone_info[zone_idx_e];
- for (z = zb; z < ze; z++) {
- pthread_mutex_lock(&z->mutex);
- pthread_mutex_lock(&f->zbd_info->mutex);
- f->zbd_info->sectors_with_data -= z->wp - z->start;
- pthread_mutex_unlock(&f->zbd_info->mutex);
- z->wp = z->start;
- z->verify_block = 0;
- pthread_mutex_unlock(&z->mutex);
- }
+ pthread_mutex_lock(&f->zbd_info->mutex);
+ f->zbd_info->sectors_with_data -= data_in_zone;
+ f->zbd_info->wp_sectors_with_data -= data_in_zone;
+ pthread_mutex_unlock(&f->zbd_info->mutex);
+ z->wp = z->start;
+ z->verify_block = 0;
- td->ts.nr_zone_resets += ze - zb;
+ td->ts.nr_zone_resets++;
return ret;
}
-static unsigned int zbd_zone_nr(struct zoned_block_device_info *zbd_info,
- struct fio_zone_info *zone)
+/* The caller must hold f->zbd_info->mutex */
+static void zbd_close_zone(struct thread_data *td, const struct fio_file *f,
+ unsigned int zone_idx)
{
- return zone - zbd_info->zone_info;
-}
+ uint32_t open_zone_idx = 0;
-/**
- * zbd_reset_zone - reset the write pointer of a single zone
- * @td: FIO thread data.
- * @f: FIO file associated with the disk for which to reset a write pointer.
- * @z: Zone to reset.
- *
- * Returns 0 upon success and a negative error code upon failure.
- */
-static int zbd_reset_zone(struct thread_data *td, const struct fio_file *f,
- struct fio_zone_info *z)
-{
- dprint(FD_ZBD, "%s: resetting wp of zone %u.\n", f->file_name,
- zbd_zone_nr(f->zbd_info, z));
+ for (; open_zone_idx < f->zbd_info->num_open_zones; open_zone_idx++) {
+ if (f->zbd_info->open_zones[open_zone_idx] == zone_idx)
+ break;
+ }
+ if (open_zone_idx == f->zbd_info->num_open_zones)
+ return;
- return zbd_reset_range(td, f, z->start, (z+1)->start - z->start);
+ dprint(FD_ZBD, "%s: closing zone %d\n", f->file_name, zone_idx);
+ memmove(f->zbd_info->open_zones + open_zone_idx,
+ f->zbd_info->open_zones + open_zone_idx + 1,
+ (ZBD_MAX_OPEN_ZONES - (open_zone_idx + 1)) *
+ sizeof(f->zbd_info->open_zones[0]));
+ f->zbd_info->num_open_zones--;
+ td->num_open_zones--;
+ get_zone(f, zone_idx)->open = 0;
}
/*
struct fio_zone_info *const zb,
struct fio_zone_info *const ze, bool all_zones)
{
- struct fio_zone_info *z, *start_z = ze;
+ struct fio_zone_info *z;
const uint32_t min_bs = td->o.min_bs[DDIR_WRITE];
bool reset_wp;
int res = 0;
+ assert(min_bs);
+
dprint(FD_ZBD, "%s: examining zones %u .. %u\n", f->file_name,
- zbd_zone_nr(f->zbd_info, zb), zbd_zone_nr(f->zbd_info, ze));
- assert(f->fd != -1);
+ zbd_zone_nr(f, zb), zbd_zone_nr(f, ze));
for (z = zb; z < ze; z++) {
- pthread_mutex_lock(&z->mutex);
- switch (z->type) {
- case BLK_ZONE_TYPE_SEQWRITE_REQ:
- reset_wp = all_zones ? z->wp != z->start :
- (td->o.td_ddir & TD_DDIR_WRITE) &&
- z->wp % min_bs != 0;
- if (start_z == ze && reset_wp) {
- start_z = z;
- } else if (start_z < ze && !reset_wp) {
- dprint(FD_ZBD,
- "%s: resetting zones %u .. %u\n",
- f->file_name,
- zbd_zone_nr(f->zbd_info, start_z),
- zbd_zone_nr(f->zbd_info, z));
- if (zbd_reset_range(td, f, start_z->start,
- z->start - start_z->start) < 0)
- res = 1;
- start_z = ze;
- }
- break;
- default:
- if (start_z == ze)
- break;
- dprint(FD_ZBD, "%s: resetting zones %u .. %u\n",
- f->file_name, zbd_zone_nr(f->zbd_info, start_z),
- zbd_zone_nr(f->zbd_info, z));
- if (zbd_reset_range(td, f, start_z->start,
- z->start - start_z->start) < 0)
+ uint32_t nz = zbd_zone_nr(f, z);
+
+ if (!z->has_wp)
+ continue;
+ zone_lock(td, f, z);
+ if (all_zones) {
+ pthread_mutex_lock(&f->zbd_info->mutex);
+ zbd_close_zone(td, f, nz);
+ pthread_mutex_unlock(&f->zbd_info->mutex);
+
+ reset_wp = z->wp != z->start;
+ } else {
+ reset_wp = z->wp % min_bs != 0;
+ }
+ if (reset_wp) {
+ dprint(FD_ZBD, "%s: resetting zone %u\n",
+ f->file_name, zbd_zone_nr(f, z));
+ if (zbd_reset_zone(td, f, z) < 0)
res = 1;
- start_z = ze;
- break;
}
+ zone_unlock(z);
}
- if (start_z < ze) {
- dprint(FD_ZBD, "%s: resetting zones %u .. %u\n", f->file_name,
- zbd_zone_nr(f->zbd_info, start_z),
- zbd_zone_nr(f->zbd_info, z));
- if (zbd_reset_range(td, f, start_z->start,
- z->start - start_z->start) < 0)
- res = 1;
- }
- for (z = zb; z < ze; z++)
- pthread_mutex_unlock(&z->mutex);
return res;
}
* Reset zbd_info.write_cnt, the counter that counts down towards the next
* zone reset.
*/
-static void zbd_reset_write_cnt(const struct thread_data *td,
- const struct fio_file *f)
+static void _zbd_reset_write_cnt(const struct thread_data *td,
+ const struct fio_file *f)
{
assert(0 <= td->o.zrf.u.f && td->o.zrf.u.f <= 1);
- pthread_mutex_lock(&f->zbd_info->mutex);
f->zbd_info->write_cnt = td->o.zrf.u.f ?
min(1.0 / td->o.zrf.u.f, 0.0 + UINT_MAX) : UINT_MAX;
+}
+
+static void zbd_reset_write_cnt(const struct thread_data *td,
+ const struct fio_file *f)
+{
+ pthread_mutex_lock(&f->zbd_info->mutex);
+ _zbd_reset_write_cnt(td, f);
pthread_mutex_unlock(&f->zbd_info->mutex);
}
if (f->zbd_info->write_cnt)
write_cnt = --f->zbd_info->write_cnt;
if (write_cnt == 0)
- zbd_reset_write_cnt(td, f);
+ _zbd_reset_write_cnt(td, f);
pthread_mutex_unlock(&f->zbd_info->mutex);
return write_cnt == 0;
};
/* Calculate the number of sectors with data (swd) and perform action 'a' */
-static uint64_t zbd_process_swd(const struct fio_file *f, enum swd_action a)
+static uint64_t zbd_process_swd(struct thread_data *td,
+ const struct fio_file *f, enum swd_action a)
{
struct fio_zone_info *zb, *ze, *z;
uint64_t swd = 0;
+ uint64_t wp_swd = 0;
- zb = &f->zbd_info->zone_info[zbd_zone_idx(f, f->file_offset)];
- ze = &f->zbd_info->zone_info[zbd_zone_idx(f, f->file_offset +
- f->io_size)];
+ zb = get_zone(f, f->min_zone);
+ ze = get_zone(f, f->max_zone);
for (z = zb; z < ze; z++) {
- pthread_mutex_lock(&z->mutex);
+ if (z->has_wp) {
+ zone_lock(td, f, z);
+ wp_swd += z->wp - z->start;
+ }
swd += z->wp - z->start;
}
pthread_mutex_lock(&f->zbd_info->mutex);
switch (a) {
case CHECK_SWD:
assert(f->zbd_info->sectors_with_data == swd);
+ assert(f->zbd_info->wp_sectors_with_data == wp_swd);
break;
case SET_SWD:
f->zbd_info->sectors_with_data = swd;
+ f->zbd_info->wp_sectors_with_data = wp_swd;
break;
}
pthread_mutex_unlock(&f->zbd_info->mutex);
for (z = zb; z < ze; z++)
- pthread_mutex_unlock(&z->mutex);
+ if (z->has_wp)
+ zone_unlock(z);
return swd;
}
*/
static const bool enable_check_swd = false;
-/* Check whether the value of zbd_info.sectors_with_data is correct. */
-static void zbd_check_swd(const struct fio_file *f)
+/* Check whether the values of zbd_info.*sectors_with_data are correct. */
+static void zbd_check_swd(struct thread_data *td, const struct fio_file *f)
{
if (!enable_check_swd)
return;
- zbd_process_swd(f, CHECK_SWD);
+ zbd_process_swd(td, f, CHECK_SWD);
}
-static void zbd_init_swd(struct fio_file *f)
+void zbd_file_reset(struct thread_data *td, struct fio_file *f)
{
+ struct fio_zone_info *zb, *ze;
uint64_t swd;
- if (!enable_check_swd)
+ if (!f->zbd_info || !td_write(td))
return;
- swd = zbd_process_swd(f, SET_SWD);
+ zb = get_zone(f, f->min_zone);
+ ze = get_zone(f, f->max_zone);
+ swd = zbd_process_swd(td, f, SET_SWD);
dprint(FD_ZBD, "%s(%s): swd = %" PRIu64 "\n", __func__, f->file_name,
swd);
-}
-
-void zbd_file_reset(struct thread_data *td, struct fio_file *f)
-{
- struct fio_zone_info *zb, *ze;
- uint32_t zone_idx_e;
-
- if (!f->zbd_info)
- return;
-
- zb = &f->zbd_info->zone_info[zbd_zone_idx(f, f->file_offset)];
- zone_idx_e = zbd_zone_idx(f, f->file_offset + f->io_size);
- ze = &f->zbd_info->zone_info[zone_idx_e];
- zbd_init_swd(f);
/*
* If data verification is enabled reset the affected zones before
* writing any data to avoid that a zone reset has to be issued while
* writing data, which causes data loss.
*/
zbd_reset_zones(td, f, zb, ze, td->o.verify != VERIFY_NONE &&
- (td->o.td_ddir & TD_DDIR_WRITE) &&
td->runstate != TD_VERIFYING);
zbd_reset_write_cnt(td, f);
}
struct zoned_block_device_info *zbdi = f->zbd_info;
int i;
- assert(td->o.max_open_zones <= ARRAY_SIZE(zbdi->open_zones));
- assert(zbdi->num_open_zones <= td->o.max_open_zones);
+ assert(td->o.job_max_open_zones == 0 || td->num_open_zones <= td->o.job_max_open_zones);
+ assert(td->o.job_max_open_zones <= zbdi->max_open_zones);
+ assert(zbdi->num_open_zones <= zbdi->max_open_zones);
for (i = 0; i < zbdi->num_open_zones; i++)
if (zbdi->open_zones[i] == zone_idx)
* was not yet open and opening a new zone would cause the zone limit to be
* exceeded.
*/
-static bool zbd_open_zone(struct thread_data *td, const struct io_u *io_u,
+static bool zbd_open_zone(struct thread_data *td, const struct fio_file *f,
uint32_t zone_idx)
{
const uint32_t min_bs = td->o.min_bs[DDIR_WRITE];
- const struct fio_file *f = io_u->file;
- struct fio_zone_info *z = &f->zbd_info->zone_info[zone_idx];
+ struct fio_zone_info *z = get_zone(f, zone_idx);
bool res = true;
- if (z->cond == BLK_ZONE_COND_OFFLINE)
+ if (z->cond == ZBD_ZONE_COND_OFFLINE)
return false;
/*
if (td->o.verify != VERIFY_NONE && zbd_zone_full(f, z, min_bs))
return false;
- /* Zero means no limit */
- if (!td->o.max_open_zones)
- return true;
-
pthread_mutex_lock(&f->zbd_info->mutex);
- if (is_zone_open(td, f, zone_idx))
+ if (is_zone_open(td, f, zone_idx)) {
+ /*
+ * If the zone is already open and going to be full by writes
+ * in-flight, handle it as a full zone instead of an open zone.
+ */
+ if (z->wp >= zbd_zone_capacity_end(z))
+ res = false;
goto out;
+ }
res = false;
- if (f->zbd_info->num_open_zones >= td->o.max_open_zones)
+ /* Zero means no limit */
+ if (td->o.job_max_open_zones > 0 &&
+ td->num_open_zones >= td->o.job_max_open_zones)
+ goto out;
+ if (f->zbd_info->num_open_zones >= f->zbd_info->max_open_zones)
goto out;
dprint(FD_ZBD, "%s: opening zone %d\n", f->file_name, zone_idx);
f->zbd_info->open_zones[f->zbd_info->num_open_zones++] = zone_idx;
+ td->num_open_zones++;
z->open = 1;
res = true;
return res;
}
-/* The caller must hold f->zbd_info->mutex */
-static void zbd_close_zone(struct thread_data *td, const struct fio_file *f,
- unsigned int open_zone_idx)
+/* Anything goes as long as it is not a constant. */
+static uint32_t pick_random_zone_idx(const struct fio_file *f,
+ const struct io_u *io_u)
{
- uint32_t zone_idx;
-
- assert(open_zone_idx < f->zbd_info->num_open_zones);
- zone_idx = f->zbd_info->open_zones[open_zone_idx];
- memmove(f->zbd_info->open_zones + open_zone_idx,
- f->zbd_info->open_zones + open_zone_idx + 1,
- (FIO_MAX_OPEN_ZBD_ZONES - (open_zone_idx + 1)) *
- sizeof(f->zbd_info->open_zones[0]));
- f->zbd_info->num_open_zones--;
- f->zbd_info->zone_info[zone_idx].open = 0;
+ return io_u->offset * f->zbd_info->num_open_zones / f->real_file_size;
}
/*
* Modify the offset of an I/O unit that does not refer to an open zone such
* that it refers to an open zone. Close an open zone and open a new zone if
- * necessary. This algorithm can only work correctly if all write pointers are
+ * necessary. The open zone is searched across sequential zones.
+ * This algorithm can only work correctly if all write pointers are
* a multiple of the fio block size. The caller must neither hold z->mutex
* nor f->zbd_info->mutex. Returns with z->mutex held upon success.
*/
struct io_u *io_u)
{
const uint32_t min_bs = td->o.min_bs[io_u->ddir];
- const struct fio_file *f = io_u->file;
+ struct fio_file *f = io_u->file;
struct fio_zone_info *z;
unsigned int open_zone_idx = -1;
uint32_t zone_idx, new_zone_idx;
int i;
+ bool wait_zone_close;
assert(is_valid_offset(f, io_u->offset));
- if (td->o.max_open_zones) {
+ if (td->o.max_open_zones || td->o.job_max_open_zones) {
/*
* This statement accesses f->zbd_info->open_zones[] on purpose
* without locking.
*/
- zone_idx = f->zbd_info->open_zones[(io_u->offset -
- f->file_offset) *
- f->zbd_info->num_open_zones / f->io_size];
+ zone_idx = f->zbd_info->open_zones[pick_random_zone_idx(f, io_u)];
} else {
zone_idx = zbd_zone_idx(f, io_u->offset);
}
+ if (zone_idx < f->min_zone)
+ zone_idx = f->min_zone;
+ else if (zone_idx >= f->max_zone)
+ zone_idx = f->max_zone - 1;
dprint(FD_ZBD, "%s(%s): starting from zone %d (offset %lld, buflen %lld)\n",
__func__, f->file_name, zone_idx, io_u->offset, io_u->buflen);
* has been obtained. Hence the loop.
*/
for (;;) {
- z = &f->zbd_info->zone_info[zone_idx];
+ uint32_t tmp_idx;
- pthread_mutex_lock(&z->mutex);
+ z = get_zone(f, zone_idx);
+ if (z->has_wp)
+ zone_lock(td, f, z);
pthread_mutex_lock(&f->zbd_info->mutex);
- if (td->o.max_open_zones == 0)
- goto examine_zone;
- if (f->zbd_info->num_open_zones == 0) {
- pthread_mutex_unlock(&f->zbd_info->mutex);
- pthread_mutex_unlock(&z->mutex);
- dprint(FD_ZBD, "%s(%s): no zones are open\n",
- __func__, f->file_name);
- return NULL;
+ if (z->has_wp) {
+ if (z->cond != ZBD_ZONE_COND_OFFLINE &&
+ td->o.max_open_zones == 0 && td->o.job_max_open_zones == 0)
+ goto examine_zone;
+ if (f->zbd_info->num_open_zones == 0) {
+ dprint(FD_ZBD, "%s(%s): no zones are open\n",
+ __func__, f->file_name);
+ goto open_other_zone;
+ }
}
- open_zone_idx = (io_u->offset - f->file_offset) *
- f->zbd_info->num_open_zones / f->io_size;
- assert(open_zone_idx < f->zbd_info->num_open_zones);
+
+ /*
+ * List of opened zones is per-device, shared across all threads.
+ * Start with quasi-random candidate zone.
+ * Ignore zones which don't belong to thread's offset/size area.
+ */
+ open_zone_idx = pick_random_zone_idx(f, io_u);
+ assert(!open_zone_idx ||
+ open_zone_idx < f->zbd_info->num_open_zones);
+ tmp_idx = open_zone_idx;
+ for (i = 0; i < f->zbd_info->num_open_zones; i++) {
+ uint32_t tmpz;
+
+ if (tmp_idx >= f->zbd_info->num_open_zones)
+ tmp_idx = 0;
+ tmpz = f->zbd_info->open_zones[tmp_idx];
+ if (f->min_zone <= tmpz && tmpz < f->max_zone) {
+ open_zone_idx = tmp_idx;
+ goto found_candidate_zone;
+ }
+
+ tmp_idx++;
+ }
+
+ dprint(FD_ZBD, "%s(%s): no candidate zone\n",
+ __func__, f->file_name);
+ pthread_mutex_unlock(&f->zbd_info->mutex);
+ if (z->has_wp)
+ zone_unlock(z);
+ return NULL;
+
+found_candidate_zone:
new_zone_idx = f->zbd_info->open_zones[open_zone_idx];
if (new_zone_idx == zone_idx)
break;
zone_idx = new_zone_idx;
pthread_mutex_unlock(&f->zbd_info->mutex);
- pthread_mutex_unlock(&z->mutex);
+ if (z->has_wp)
+ zone_unlock(z);
}
/* Both z->mutex and f->zbd_info->mutex are held. */
examine_zone:
- if (z->wp + min_bs <= (z+1)->start) {
+ if (z->wp + min_bs <= zbd_zone_capacity_end(z)) {
pthread_mutex_unlock(&f->zbd_info->mutex);
goto out;
}
- dprint(FD_ZBD, "%s(%s): closing zone %d\n", __func__, f->file_name,
- zone_idx);
- if (td->o.max_open_zones)
- zbd_close_zone(td, f, open_zone_idx);
+
+open_other_zone:
+ /* Check if number of open zones reaches one of limits. */
+ wait_zone_close =
+ f->zbd_info->num_open_zones == f->max_zone - f->min_zone ||
+ (td->o.max_open_zones &&
+ f->zbd_info->num_open_zones == td->o.max_open_zones) ||
+ (td->o.job_max_open_zones &&
+ td->num_open_zones == td->o.job_max_open_zones);
+
pthread_mutex_unlock(&f->zbd_info->mutex);
/* Only z->mutex is held. */
+ /*
+ * When number of open zones reaches to one of limits, wait for
+ * zone close before opening a new zone.
+ */
+ if (wait_zone_close) {
+ dprint(FD_ZBD, "%s(%s): quiesce to allow open zones to close\n",
+ __func__, f->file_name);
+ io_u_quiesce(td);
+ }
+
/* Zone 'z' is full, so try to open a new zone. */
for (i = f->io_size / f->zbd_info->zone_size; i > 0; i--) {
zone_idx++;
- pthread_mutex_unlock(&z->mutex);
+ if (z->has_wp)
+ zone_unlock(z);
z++;
if (!is_valid_offset(f, z->start)) {
/* Wrap-around. */
- zone_idx = zbd_zone_idx(f, f->file_offset);
- z = &f->zbd_info->zone_info[zone_idx];
+ zone_idx = f->min_zone;
+ z = get_zone(f, zone_idx);
}
assert(is_valid_offset(f, z->start));
- pthread_mutex_lock(&z->mutex);
+ if (!z->has_wp)
+ continue;
+ zone_lock(td, f, z);
if (z->open)
continue;
- if (zbd_open_zone(td, io_u, zone_idx))
+ if (zbd_open_zone(td, f, zone_idx))
goto out;
}
pthread_mutex_lock(&f->zbd_info->mutex);
for (i = 0; i < f->zbd_info->num_open_zones; i++) {
zone_idx = f->zbd_info->open_zones[i];
+ if (zone_idx < f->min_zone || zone_idx >= f->max_zone)
+ continue;
pthread_mutex_unlock(&f->zbd_info->mutex);
- pthread_mutex_unlock(&z->mutex);
+ zone_unlock(z);
- z = &f->zbd_info->zone_info[zone_idx];
+ z = get_zone(f, zone_idx);
- pthread_mutex_lock(&z->mutex);
- if (z->wp + min_bs <= (z+1)->start)
+ zone_lock(td, f, z);
+ if (z->wp + min_bs <= zbd_zone_capacity_end(z))
goto out;
pthread_mutex_lock(&f->zbd_info->mutex);
}
pthread_mutex_unlock(&f->zbd_info->mutex);
- pthread_mutex_unlock(&z->mutex);
+ zone_unlock(z);
dprint(FD_ZBD, "%s(%s): did not open another zone\n", __func__,
f->file_name);
return NULL;
dprint(FD_ZBD, "%s(%s): returning zone %d\n", __func__, f->file_name,
zone_idx);
io_u->offset = z->start;
+ assert(z->has_wp);
+ assert(z->cond != ZBD_ZONE_COND_OFFLINE);
return z;
}
const struct fio_file *f = io_u->file;
const uint32_t min_bs = td->o.min_bs[DDIR_WRITE];
- if (!zbd_open_zone(td, io_u, z - f->zbd_info->zone_info)) {
- pthread_mutex_unlock(&z->mutex);
+ if (!zbd_open_zone(td, f, zbd_zone_nr(f, z))) {
+ zone_unlock(z);
z = zbd_convert_to_open_zone(td, io_u);
assert(z);
}
- if (z->verify_block * min_bs >= f->zbd_info->zone_size)
+ if (z->verify_block * min_bs >= z->capacity) {
log_err("%s: %d * %d >= %llu\n", f->file_name, z->verify_block,
- min_bs, (unsigned long long) f->zbd_info->zone_size);
- io_u->offset = z->start + z->verify_block++ * min_bs;
+ min_bs, (unsigned long long)z->capacity);
+ /*
+ * If the assertion below fails during a test run, adding
+ * "--experimental_verify=1" to the command line may help.
+ */
+ assert(false);
+ }
+ io_u->offset = z->start + z->verify_block * min_bs;
+ if (io_u->offset + io_u->buflen >= zbd_zone_capacity_end(z)) {
+ log_err("%s: %llu + %llu >= %llu\n", f->file_name, io_u->offset,
+ io_u->buflen, (unsigned long long) zbd_zone_capacity_end(z));
+ assert(false);
+ }
+ z->verify_block += io_u->buflen / min_bs;
+
return z;
}
/*
- * Find another zone for which @io_u fits below the write pointer. Start
- * searching in zones @zb + 1 .. @zl and continue searching in zones
- * @zf .. @zb - 1.
+ * Find another zone for which @io_u fits in the readable data in the zone.
+ * Search in zones @zb + 1 .. @zl. For random workload, also search in zones
+ * @zb - 1 .. @zf.
*
- * Either returns NULL or returns a zone pointer and holds the mutex for that
- * zone.
+ * Either returns NULL or returns a zone pointer. When the zone has write
+ * pointer, hold the mutex for the zone.
*/
static struct fio_zone_info *
zbd_find_zone(struct thread_data *td, struct io_u *io_u,
struct fio_zone_info *zb, struct fio_zone_info *zl)
{
const uint32_t min_bs = td->o.min_bs[io_u->ddir];
- const struct fio_file *f = io_u->file;
+ struct fio_file *f = io_u->file;
struct fio_zone_info *z1, *z2;
- const struct fio_zone_info *const zf =
- &f->zbd_info->zone_info[zbd_zone_idx(f, f->file_offset)];
+ const struct fio_zone_info *const zf = get_zone(f, f->min_zone);
/*
* Skip to the next non-empty zone in case of sequential I/O and to
* the nearest non-empty zone in case of random I/O.
*/
for (z1 = zb + 1, z2 = zb - 1; z1 < zl || z2 >= zf; z1++, z2--) {
- if (z1 < zl && z1->cond != BLK_ZONE_COND_OFFLINE) {
- pthread_mutex_lock(&z1->mutex);
+ if (z1 < zl && z1->cond != ZBD_ZONE_COND_OFFLINE) {
+ if (z1->has_wp)
+ zone_lock(td, f, z1);
if (z1->start + min_bs <= z1->wp)
return z1;
- pthread_mutex_unlock(&z1->mutex);
+ if (z1->has_wp)
+ zone_unlock(z1);
} else if (!td_random(td)) {
break;
}
if (td_random(td) && z2 >= zf &&
- z2->cond != BLK_ZONE_COND_OFFLINE) {
- pthread_mutex_lock(&z2->mutex);
+ z2->cond != ZBD_ZONE_COND_OFFLINE) {
+ if (z2->has_wp)
+ zone_lock(td, f, z2);
if (z2->start + min_bs <= z2->wp)
return z2;
- pthread_mutex_unlock(&z2->mutex);
+ if (z2->has_wp)
+ zone_unlock(z2);
}
}
dprint(FD_ZBD, "%s: adjusting random read offset failed\n",
return NULL;
}
+/**
+ * zbd_end_zone_io - update zone status at command completion
+ * @io_u: I/O unit
+ * @z: zone info pointer
+ *
+ * If the write command made the zone full, close it.
+ *
+ * The caller must hold z->mutex.
+ */
+static void zbd_end_zone_io(struct thread_data *td, const struct io_u *io_u,
+ struct fio_zone_info *z)
+{
+ const struct fio_file *f = io_u->file;
+
+ if (io_u->ddir == DDIR_WRITE &&
+ io_u->offset + io_u->buflen >= zbd_zone_capacity_end(z)) {
+ pthread_mutex_lock(&f->zbd_info->mutex);
+ zbd_close_zone(td, f, zbd_zone_nr(f, z));
+ pthread_mutex_unlock(&f->zbd_info->mutex);
+ }
+}
+
/**
* zbd_queue_io - update the write pointer of a sequential zone
* @io_u: I/O unit
* For write and trim operations, update the write pointer of the I/O unit
* target zone.
*/
-static void zbd_queue_io(struct io_u *io_u, int q, bool success)
+static void zbd_queue_io(struct thread_data *td, struct io_u *io_u, int q,
+ bool success)
{
const struct fio_file *f = io_u->file;
struct zoned_block_device_info *zbd_info = f->zbd_info;
zone_idx = zbd_zone_idx(f, io_u->offset);
assert(zone_idx < zbd_info->nr_zones);
- z = &zbd_info->zone_info[zone_idx];
+ z = get_zone(f, zone_idx);
- if (z->type != BLK_ZONE_TYPE_SEQWRITE_REQ)
- return;
+ assert(z->has_wp);
if (!success)
goto unlock;
switch (io_u->ddir) {
case DDIR_WRITE:
zone_end = min((uint64_t)(io_u->offset + io_u->buflen),
- (z + 1)->start);
+ zbd_zone_capacity_end(z));
pthread_mutex_lock(&zbd_info->mutex);
/*
* z->wp > zone_end means that one or more I/O errors
* have occurred.
*/
- if (z->wp <= zone_end)
+ if (z->wp <= zone_end) {
zbd_info->sectors_with_data += zone_end - z->wp;
+ zbd_info->wp_sectors_with_data += zone_end - z->wp;
+ }
pthread_mutex_unlock(&zbd_info->mutex);
z->wp = zone_end;
break;
break;
}
+ if (q == FIO_Q_COMPLETED && !io_u->error)
+ zbd_end_zone_io(td, io_u, z);
+
unlock:
if (!success || q != FIO_Q_QUEUED) {
/* BUSY or COMPLETED: unlock the zone */
- pthread_mutex_unlock(&z->mutex);
+ zone_unlock(z);
io_u->zbd_put_io = NULL;
}
}
* zbd_put_io - Unlock an I/O unit target zone lock
* @io_u: I/O unit
*/
-static void zbd_put_io(const struct io_u *io_u)
+static void zbd_put_io(struct thread_data *td, const struct io_u *io_u)
{
const struct fio_file *f = io_u->file;
struct zoned_block_device_info *zbd_info = f->zbd_info;
zone_idx = zbd_zone_idx(f, io_u->offset);
assert(zone_idx < zbd_info->nr_zones);
- z = &zbd_info->zone_info[zone_idx];
+ z = get_zone(f, zone_idx);
- if (z->type != BLK_ZONE_TYPE_SEQWRITE_REQ)
- return;
+ assert(z->has_wp);
dprint(FD_ZBD,
"%s: terminate I/O (%lld, %llu) for zone %u\n",
f->file_name, io_u->offset, io_u->buflen, zone_idx);
- assert(pthread_mutex_unlock(&z->mutex) == 0);
- zbd_check_swd(f);
+ zbd_end_zone_io(td, io_u, z);
+
+ zone_unlock(z);
+ zbd_check_swd(td, f);
}
+/*
+ * Windows and MacOS do not define this.
+ */
+#ifndef EREMOTEIO
+#define EREMOTEIO 121 /* POSIX value */
+#endif
+
bool zbd_unaligned_write(int error_code)
{
switch (error_code) {
assert(td->o.zone_mode == ZONE_MODE_ZBD);
assert(td->o.zone_size);
+ zone_idx = zbd_zone_idx(f, f->last_pos[ddir]);
+ z = get_zone(f, zone_idx);
+
+ /*
+ * When the zone capacity is smaller than the zone size and the I/O is
+ * sequential write, skip to zone end if the latest position is at the
+ * zone capacity limit.
+ */
+ if (z->capacity < f->zbd_info->zone_size && !td_random(td) &&
+ ddir == DDIR_WRITE &&
+ f->last_pos[ddir] >= zbd_zone_capacity_end(z)) {
+ dprint(FD_ZBD,
+ "%s: Jump from zone capacity limit to zone end:"
+ " (%llu -> %llu) for zone %u (%llu)\n",
+ f->file_name, (unsigned long long) f->last_pos[ddir],
+ (unsigned long long) zbd_zone_end(z), zone_idx,
+ (unsigned long long) z->capacity);
+ td->io_skip_bytes += zbd_zone_end(z) - f->last_pos[ddir];
+ f->last_pos[ddir] = zbd_zone_end(z);
+ }
+
/*
* zone_skip is valid only for sequential workloads.
*/
* - For reads with td->o.read_beyond_wp == false, the last position
* reached the zone write pointer.
*/
- zone_idx = zbd_zone_idx(f, f->last_pos[ddir]);
- z = &f->zbd_info->zone_info[zone_idx];
-
if (td->zone_bytes >= td->o.zone_size ||
- f->last_pos[ddir] >= (z+1)->start ||
+ f->last_pos[ddir] >= zbd_zone_end(z) ||
(ddir == DDIR_READ &&
(!td->o.read_beyond_wp) && f->last_pos[ddir] >= z->wp)) {
/*
}
}
+/**
+ * zbd_adjust_ddir - Adjust an I/O direction for zonemode=zbd.
+ *
+ * @td: FIO thread data.
+ * @io_u: FIO I/O unit.
+ * @ddir: I/O direction before adjustment.
+ *
+ * Return adjusted I/O direction.
+ */
+enum fio_ddir zbd_adjust_ddir(struct thread_data *td, struct io_u *io_u,
+ enum fio_ddir ddir)
+{
+ /*
+ * In case read direction is chosen for the first random I/O, fio with
+ * zonemode=zbd stops because no data can be read from zoned block
+ * devices with all empty zones. Overwrite the first I/O direction as
+ * write to make sure data to read exists.
+ */
+ if (ddir != DDIR_READ || !td_rw(td))
+ return ddir;
+
+ if (io_u->file->zbd_info->sectors_with_data ||
+ td->o.read_beyond_wp)
+ return DDIR_READ;
+
+ return DDIR_WRITE;
+}
+
/**
* zbd_adjust_block - adjust the offset and length as necessary for ZBD drives
* @td: FIO thread data.
*/
enum io_u_action zbd_adjust_block(struct thread_data *td, struct io_u *io_u)
{
- const struct fio_file *f = io_u->file;
+ struct fio_file *f = io_u->file;
uint32_t zone_idx_b;
struct fio_zone_info *zb, *zl, *orig_zb;
uint32_t orig_len = io_u->buflen;
if (!f->zbd_info)
return io_u_accept;
+ assert(min_bs);
assert(is_valid_offset(f, io_u->offset));
assert(io_u->buflen);
zone_idx_b = zbd_zone_idx(f, io_u->offset);
- zb = &f->zbd_info->zone_info[zone_idx_b];
+ zb = get_zone(f, zone_idx_b);
orig_zb = zb;
- /* Accept the I/O offset for conventional zones. */
- if (zb->type == BLK_ZONE_TYPE_CONVENTIONAL)
+ if (!zb->has_wp) {
+ /* Accept non-write I/Os for conventional zones. */
+ if (io_u->ddir != DDIR_WRITE)
+ return io_u_accept;
+ /*
+ * Make sure that writes to conventional zones
+ * don't cross over to any sequential zones.
+ */
+ if (!(zb + 1)->has_wp ||
+ io_u->offset + io_u->buflen <= (zb + 1)->start)
+ return io_u_accept;
+
+ if (io_u->offset + min_bs > (zb + 1)->start) {
+ dprint(FD_IO,
+ "%s: off=%llu + min_bs=%u > next zone %llu\n",
+ f->file_name, io_u->offset,
+ min_bs, (unsigned long long) (zb + 1)->start);
+ io_u->offset = zb->start + (zb + 1)->start - io_u->offset;
+ new_len = min(io_u->buflen, (zb + 1)->start - io_u->offset);
+ } else {
+ new_len = (zb + 1)->start - io_u->offset;
+ }
+ io_u->buflen = new_len / min_bs * min_bs;
return io_u_accept;
+ }
/*
* Accept the I/O offset for reads if reading beyond the write pointer
* is enabled.
*/
- if (zb->cond != BLK_ZONE_COND_OFFLINE &&
+ if (zb->cond != ZBD_ZONE_COND_OFFLINE &&
io_u->ddir == DDIR_READ && td->o.read_beyond_wp)
return io_u_accept;
- zbd_check_swd(f);
+ zbd_check_swd(td, f);
- /*
- * Lock the io_u target zone. The zone will be unlocked if io_u offset
- * is changed or when io_u completes and zbd_put_io() executed.
- * To avoid multiple jobs doing asynchronous I/Os from deadlocking each
- * other waiting for zone locks when building an io_u batch, first
- * only trylock the zone. If the zone is already locked by another job,
- * process the currently queued I/Os so that I/O progress is made and
- * zones unlocked.
- */
- if (pthread_mutex_trylock(&zb->mutex) != 0) {
- if (!td_ioengine_flagged(td, FIO_SYNCIO))
- io_u_quiesce(td);
- pthread_mutex_lock(&zb->mutex);
- }
+ zone_lock(td, f, zb);
switch (io_u->ddir) {
case DDIR_READ:
- if (td->runstate == TD_VERIFYING) {
+ if (td->runstate == TD_VERIFYING && td_write(td)) {
zb = zbd_replay_write_order(td, io_u, zb);
goto accept;
}
* I/O of at least min_bs B. If there isn't, find a new zone for
* the I/O.
*/
- range = zb->cond != BLK_ZONE_COND_OFFLINE ?
+ range = zb->cond != ZBD_ZONE_COND_OFFLINE ?
zb->wp - zb->start : 0;
if (range < min_bs ||
((!td_random(td)) && (io_u->offset + min_bs > zb->wp))) {
- pthread_mutex_unlock(&zb->mutex);
- zl = &f->zbd_info->zone_info[zbd_zone_idx(f,
- f->file_offset + f->io_size)];
+ zone_unlock(zb);
+ zl = get_zone(f, f->max_zone);
zb = zbd_find_zone(td, io_u, zb, zl);
if (!zb) {
dprint(FD_ZBD,
io_u->offset = zb->start +
((io_u->offset - orig_zb->start) %
(range - io_u->buflen)) / min_bs * min_bs;
+ /*
+ * When zbd_find_zone() returns a conventional zone,
+ * we can simply accept the new i/o offset here.
+ */
+ if (!zb->has_wp)
+ return io_u_accept;
/*
* Make sure the I/O does not cross over the zone wp position.
*/
assert(io_u->offset + io_u->buflen <= zb->wp);
goto accept;
case DDIR_WRITE:
- if (io_u->buflen > f->zbd_info->zone_size)
+ if (io_u->buflen > f->zbd_info->zone_size) {
+ td_verror(td, EINVAL, "I/O buflen exceeds zone size");
+ dprint(FD_IO,
+ "%s: I/O buflen %llu exceeds zone size %llu\n",
+ f->file_name, io_u->buflen,
+ (unsigned long long) f->zbd_info->zone_size);
goto eof;
- if (!zbd_open_zone(td, io_u, zone_idx_b)) {
- pthread_mutex_unlock(&zb->mutex);
+ }
+ if (!zbd_open_zone(td, f, zone_idx_b)) {
+ zone_unlock(zb);
zb = zbd_convert_to_open_zone(td, io_u);
- if (!zb)
+ if (!zb) {
+ dprint(FD_IO, "%s: can't convert to open zone",
+ f->file_name);
goto eof;
- zone_idx_b = zb - f->zbd_info->zone_info;
+ }
+ zone_idx_b = zbd_zone_nr(f, zb);
}
/* Check whether the zone reset threshold has been exceeded */
if (td->o.zrf.u.f) {
- if (f->zbd_info->sectors_with_data >=
+ if (f->zbd_info->wp_sectors_with_data >=
f->io_size * td->o.zrt.u.f &&
zbd_dec_and_reset_write_cnt(td, f)) {
zb->reset_zone = 1;
zb->reset_zone = 0;
if (zbd_reset_zone(td, f, zb) < 0)
goto eof;
+
+ if (zb->capacity < min_bs) {
+ td_verror(td, EINVAL, "ZCAP is less min_bs");
+ log_err("zone capacity %llu smaller than minimum block size %d\n",
+ (unsigned long long)zb->capacity,
+ min_bs);
+ goto eof;
+ }
}
/* Make writes occur at the write pointer */
assert(!zbd_zone_full(f, zb, min_bs));
io_u->offset = zb->wp;
if (!is_valid_offset(f, io_u->offset)) {
- dprint(FD_ZBD, "Dropped request with offset %llu\n",
- io_u->offset);
+ td_verror(td, EINVAL, "invalid WP value");
+ dprint(FD_ZBD, "%s: dropped request with offset %llu\n",
+ f->file_name, io_u->offset);
goto eof;
}
/*
* small.
*/
new_len = min((unsigned long long)io_u->buflen,
- (zb + 1)->start - io_u->offset);
+ zbd_zone_capacity_end(zb) - io_u->offset);
new_len = new_len / min_bs * min_bs;
if (new_len == io_u->buflen)
goto accept;
orig_len, io_u->buflen);
goto accept;
}
- log_err("Zone remainder %lld smaller than minimum block size %d\n",
- ((zb + 1)->start - io_u->offset),
- min_bs);
+ td_verror(td, EIO, "zone remainder too small");
+ log_err("zone remainder %lld smaller than min block size %d\n",
+ (zbd_zone_capacity_end(zb) - io_u->offset), min_bs);
goto eof;
case DDIR_TRIM:
/* fall-through */
assert(false);
accept:
- assert(zb);
- assert(zb->cond != BLK_ZONE_COND_OFFLINE);
+ assert(zb->has_wp);
+ assert(zb->cond != ZBD_ZONE_COND_OFFLINE);
assert(!io_u->zbd_queue_io);
assert(!io_u->zbd_put_io);
io_u->zbd_queue_io = zbd_queue_io;
io_u->zbd_put_io = zbd_put_io;
+ /*
+ * Since we return with the zone lock still held,
+ * add an annotation to let Coverity know that it
+ * is intentional.
+ */
+ /* coverity[missing_unlock] */
return io_u_accept;
eof:
- if (zb)
- pthread_mutex_unlock(&zb->mutex);
+ if (zb && zb->has_wp)
+ zone_unlock(zb);
return io_u_eof;
}