X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=fio.h;h=446482354fd909125c911defe6f27b9ceb83fee9;hp=db594ab8e4d01250fe7360a48b23d3a3211f350a;hb=55312f9f5572611037b6254912460f64173df574;hpb=de98bd30b02bd89a78059d162b2c8426e889703d diff --git a/fio.h b/fio.h index db594ab8..44648235 100644 --- a/fio.h +++ b/fio.h @@ -14,12 +14,10 @@ #include #include -struct thread_data; - #include "compiler/compiler.h" +#include "thread_options.h" #include "flist.h" #include "fifo.h" -#include "rbtree.h" #include "arch/arch.h" #include "os/os.h" #include "mutex.h" @@ -32,13 +30,16 @@ struct thread_data; #include "helpers.h" #include "options.h" #include "profile.h" -#include "time.h" +#include "fio_time.h" #include "gettime.h" #include "lib/getopt.h" #include "lib/rand.h" +#include "lib/rbtree.h" +#include "client.h" #include "server.h" #include "stat.h" #include "flow.h" +#include "io_u_queue.h" #ifdef CONFIG_SOLARISAIO #include @@ -54,17 +55,6 @@ struct thread_data; #define MPOL_LOCAL MPOL_MAX #endif -/* - * What type of allocation to use for io buffers - */ -enum fio_memtype { - MEM_MALLOC = 0, /* ordinary malloc */ - MEM_SHM, /* use shared memory segments */ - MEM_SHMHUGE, /* use shared memory segments with huge pages */ - MEM_MMAP, /* use anonynomous mmap */ - MEM_MMAPHUGE, /* memory mapped huge file */ -}; - /* * offset generator types */ @@ -73,241 +63,6 @@ enum { RW_SEQ_IDENT, }; -/* - * What type of errors to continue on when continue_on_error is used - */ -enum error_type_bit { - ERROR_TYPE_READ_BIT = 0, - ERROR_TYPE_WRITE_BIT = 1, - ERROR_TYPE_VERIFY_BIT = 2, - ERROR_TYPE_CNT = 3, -}; - -enum error_type { - ERROR_TYPE_NONE = 0, - ERROR_TYPE_READ = 1 << ERROR_TYPE_READ_BIT, - ERROR_TYPE_WRITE = 1 << ERROR_TYPE_WRITE_BIT, - ERROR_TYPE_VERIFY = 1 << ERROR_TYPE_VERIFY_BIT, - ERROR_TYPE_ANY = 0xffff, -}; - -struct bssplit { - unsigned int bs; - unsigned char perc; -}; - -struct thread_options { - int pad; - char *description; - char *name; - char *directory; - char *filename; - char *filename_format; - char *opendir; - char *ioengine; - enum td_ddir td_ddir; - unsigned int rw_seq; - unsigned int kb_base; - unsigned int ddir_seq_nr; - long ddir_seq_add; - unsigned int iodepth; - unsigned int iodepth_low; - unsigned int iodepth_batch; - unsigned int iodepth_batch_complete; - - unsigned long long size; - unsigned int size_percent; - unsigned int fill_device; - unsigned long long file_size_low; - unsigned long long file_size_high; - unsigned long long start_offset; - - unsigned int bs[DDIR_RWDIR_CNT]; - unsigned int ba[DDIR_RWDIR_CNT]; - unsigned int min_bs[DDIR_RWDIR_CNT]; - unsigned int max_bs[DDIR_RWDIR_CNT]; - struct bssplit *bssplit[DDIR_RWDIR_CNT]; - unsigned int bssplit_nr[DDIR_RWDIR_CNT]; - - int *ignore_error[ERROR_TYPE_CNT]; - unsigned int ignore_error_nr[ERROR_TYPE_CNT]; - unsigned int error_dump; - - unsigned int nr_files; - unsigned int open_files; - enum file_lock_mode file_lock_mode; - - unsigned int odirect; - unsigned int invalidate_cache; - unsigned int create_serialize; - unsigned int create_fsync; - unsigned int create_on_open; - unsigned int create_only; - unsigned int end_fsync; - unsigned int pre_read; - unsigned int sync_io; - unsigned int verify; - unsigned int do_verify; - unsigned int verifysort; - unsigned int verifysort_nr; - unsigned int verify_interval; - unsigned int verify_offset; - char verify_pattern[MAX_PATTERN_SIZE]; - unsigned int verify_pattern_bytes; - unsigned int verify_fatal; - unsigned int verify_dump; - unsigned int verify_async; - unsigned long long verify_backlog; - unsigned int verify_batch; - unsigned int experimental_verify; - unsigned int use_thread; - unsigned int unlink; - unsigned int do_disk_util; - unsigned int override_sync; - unsigned int rand_repeatable; - unsigned int use_os_rand; - unsigned int write_lat_log; - unsigned int write_bw_log; - unsigned int write_iops_log; - unsigned int log_avg_msec; - unsigned int norandommap; - unsigned int softrandommap; - unsigned int bs_unaligned; - unsigned int fsync_on_close; - - unsigned int random_distribution; - double zipf_theta; - double pareto_h; - - unsigned int random_generator; - - unsigned int hugepage_size; - unsigned int rw_min_bs; - unsigned int thinktime; - unsigned int thinktime_spin; - unsigned int thinktime_blocks; - unsigned int fsync_blocks; - unsigned int fdatasync_blocks; - unsigned int barrier_blocks; - unsigned long long start_delay; - unsigned long long timeout; - unsigned long long ramp_time; - unsigned int overwrite; - unsigned int bw_avg_time; - unsigned int iops_avg_time; - unsigned int loops; - unsigned long long zone_range; - unsigned long long zone_size; - unsigned long long zone_skip; - enum fio_memtype mem_type; - unsigned int mem_align; - - unsigned int max_latency; - - unsigned int stonewall; - unsigned int new_group; - unsigned int numjobs; - os_cpu_mask_t cpumask; - unsigned int cpumask_set; - os_cpu_mask_t verify_cpumask; - unsigned int verify_cpumask_set; -#ifdef CONFIG_LIBNUMA - struct bitmask *numa_cpunodesmask; - unsigned int numa_cpumask_set; - unsigned short numa_mem_mode; - unsigned int numa_mem_prefer_node; - struct bitmask *numa_memnodesmask; - unsigned int numa_memmask_set; -#endif - unsigned int iolog; - unsigned int rwmixcycle; - unsigned int rwmix[2]; - unsigned int nice; - unsigned int file_service_type; - unsigned int group_reporting; - unsigned int fadvise_hint; - enum fio_fallocate_mode fallocate_mode; - unsigned int zero_buffers; - unsigned int refill_buffers; - unsigned int scramble_buffers; - unsigned int compress_percentage; - unsigned int compress_chunk; - unsigned int time_based; - unsigned int disable_lat; - unsigned int disable_clat; - unsigned int disable_slat; - unsigned int disable_bw; - unsigned int unified_rw_rep; - unsigned int gtod_reduce; - unsigned int gtod_cpu; - unsigned int gtod_offload; - enum fio_cs clocksource; - unsigned int no_stall; - unsigned int trim_percentage; - unsigned int trim_batch; - unsigned int trim_zero; - unsigned long long trim_backlog; - unsigned int clat_percentiles; - unsigned int percentile_precision; /* digits after decimal for percentiles */ - fio_fp64_t percentile_list[FIO_IO_U_LIST_MAX_LEN]; - - char *read_iolog_file; - char *write_iolog_file; - char *bw_log_file; - char *lat_log_file; - char *iops_log_file; - char *replay_redirect; - - /* - * Pre-run and post-run shell - */ - char *exec_prerun; - char *exec_postrun; - - unsigned int rate[DDIR_RWDIR_CNT]; - unsigned int ratemin[DDIR_RWDIR_CNT]; - unsigned int ratecycle; - unsigned int rate_iops[DDIR_RWDIR_CNT]; - unsigned int rate_iops_min[DDIR_RWDIR_CNT]; - - char *ioscheduler; - - /* - * CPU "io" cycle burner - */ - unsigned int cpuload; - unsigned int cpucycle; - - /* - * I/O Error handling - */ - enum error_type continue_on_error; - - /* - * Benchmark profile type - */ - char *profile; - - /* - * blkio cgroup support - */ - char *cgroup; - unsigned int cgroup_weight; - unsigned int cgroup_nodelete; - - unsigned int uid; - unsigned int gid; - - int flow_id; - int flow; - int flow_watermark; - unsigned int flow_sleep; - - unsigned long long offset_increment; - - unsigned int sync_file_range; -}; - enum { TD_F_VER_BACKLOG = 1, TD_F_TRIM_BACKLOG = 2, @@ -316,6 +71,10 @@ enum { TD_F_SCRAMBLE_BUFFERS = 16, TD_F_VER_NONE = 32, TD_F_PROFILE_OPS = 64, + TD_F_COMPRESS = 128, + TD_F_NOIO = 256, + TD_F_COMPRESS_LOG = 512, + TD_F_VSTATE_SAVED = 1024, }; enum { @@ -327,6 +86,11 @@ enum { FIO_RAND_FILE_SIZE_OFF, FIO_RAND_TRIM_OFF, FIO_RAND_BUF_OFF, + FIO_RAND_SEQ_RAND_READ_OFF, + FIO_RAND_SEQ_RAND_WRITE_OFF, + FIO_RAND_SEQ_RAND_TRIM_OFF, + FIO_RAND_START_DELAY, + FIO_DEDUPE_OFF, FIO_RAND_NR_OFFS, }; @@ -339,22 +103,34 @@ struct thread_data { void *eo; char verror[FIO_VERROR_SIZE]; pthread_t thread; - int thread_number; - int groupid; + unsigned int thread_number; + unsigned int subjob_number; + unsigned int groupid; struct thread_stat ts; + int client_type; + struct io_log *slat_log; struct io_log *clat_log; struct io_log *lat_log; struct io_log *bw_log; struct io_log *iops_log; + struct tp_data *tp_data; + uint64_t stat_io_bytes[DDIR_RWDIR_CNT]; struct timeval bw_sample_time; uint64_t stat_io_blocks[DDIR_RWDIR_CNT]; struct timeval iops_sample_time; + /* + * Tracks the last iodepth number of completed writes, if data + * verification is enabled + */ + uint64_t *last_write_comp; + unsigned int last_write_idx; + volatile int update_rusage; struct fio_mutex *rusage_sem; struct rusage ru_start; @@ -369,23 +145,20 @@ struct thread_data { unsigned int nr_normal_files; union { unsigned int next_file; - os_random_state_t next_file_state; - struct frand_state __next_file_state; + struct frand_state next_file_state; }; int error; int sig; int done; + int stop_io; pid_t pid; char *orig_buffer; size_t orig_buffer_size; volatile int terminate; volatile int runstate; - unsigned int ioprio; - unsigned int ioprio_set; unsigned int last_was_sync; enum fio_ddir last_ddir; - char *mmapfile; int mmapfd; void *iolog_buf; @@ -395,24 +168,20 @@ struct thread_data { unsigned long rand_seeds[FIO_RAND_NR_OFFS]; - union { - os_random_state_t bsrange_state; - struct frand_state __bsrange_state; - }; - union { - os_random_state_t verify_state; - struct frand_state __verify_state; - }; - union { - os_random_state_t trim_state; - struct frand_state __trim_state; - }; + struct frand_state bsrange_state; + struct frand_state verify_state; + struct frand_state trim_state; + struct frand_state delay_state; struct frand_state buf_state; + struct frand_state buf_state_prev; + struct frand_state dedupe_state; unsigned int verify_batch; unsigned int trim_batch; + struct thread_io_list *vstate; + int shm_id; /* @@ -439,9 +208,9 @@ struct thread_data { /* * List of free and busy io_u's */ - struct flist_head io_u_freelist; - struct flist_head io_u_busylist; - struct flist_head io_u_requeues; + struct io_u_ring io_u_requeues; + struct io_u_queue io_u_freelist; + struct io_u_queue io_u_all; pthread_mutex_t io_u_lock; pthread_cond_t free_cond; @@ -466,42 +235,64 @@ struct thread_data { uint64_t total_io_size; uint64_t fill_device_size; - unsigned long io_issues[DDIR_RWDIR_CNT]; + /* + * Issue side + */ + uint64_t io_issues[DDIR_RWDIR_CNT]; + uint64_t io_issue_bytes[DDIR_RWDIR_CNT]; + uint64_t loops; + + /* + * Completions + */ uint64_t io_blocks[DDIR_RWDIR_CNT]; uint64_t this_io_blocks[DDIR_RWDIR_CNT]; uint64_t io_bytes[DDIR_RWDIR_CNT]; - uint64_t io_skip_bytes; uint64_t this_io_bytes[DDIR_RWDIR_CNT]; + uint64_t io_skip_bytes; uint64_t zone_bytes; struct fio_mutex *mutex; + uint64_t bytes_done[DDIR_RWDIR_CNT]; /* * State for random io, a bitmap of blocks done vs not done */ - union { - os_random_state_t random_state; - struct frand_state __random_state; - }; + struct frand_state random_state; struct timeval start; /* start of this loop */ struct timeval epoch; /* time job was started */ struct timeval last_issue; + long time_offset; struct timeval tv_cache; + struct timeval terminate_time; unsigned int tv_cache_nr; unsigned int tv_cache_mask; unsigned int ramp_time_over; + /* + * Time since last latency_window was started + */ + struct timeval latency_ts; + unsigned int latency_qd; + unsigned int latency_qd_high; + unsigned int latency_qd_low; + unsigned int latency_failed; + uint64_t latency_ios; + int latency_end_run; + /* * read/write mixed workload state */ - union { - os_random_state_t rwmix_state; - struct frand_state __rwmix_state; - }; + struct frand_state rwmix_state; unsigned long rwmix_issues; enum fio_ddir rwmix_ddir; unsigned int ddir_seq_nr; + /* + * rand/seq mixed workload state + */ + struct frand_state seq_rand_state[DDIR_RWDIR_CNT]; + /* * IO history logs for verification. We use a tree for sorting, * if we are overwriting. Otherwise just use a fifo. @@ -535,10 +326,7 @@ struct thread_data { /* * For generating file sizes */ - union { - os_random_state_t file_size_state; - struct frand_state __file_size_state; - }; + struct frand_state file_size_state; /* * Error counts @@ -553,6 +341,8 @@ struct thread_data { */ struct prof_io_ops prof_io_ops; void *prof_data; + + void *pinned_mem; }; /* @@ -566,10 +356,10 @@ enum { #define __td_verror(td, err, msg, func) \ do { \ + unsigned int ____e = (err); \ if ((td)->error) \ break; \ - int e = (err); \ - (td)->error = e; \ + (td)->error = ____e; \ if (!(td)->first_error) \ snprintf(td->verror, sizeof(td->verror), "file:%s:%d, func=%s, error=%s", __FILE__, __LINE__, (func), (msg)); \ } while (0) @@ -588,12 +378,11 @@ enum { extern int exitall_on_terminate; extern unsigned int thread_number; extern unsigned int stat_number; -extern unsigned int nr_process, nr_thread; extern int shm_id; extern int groupid; extern int output_format; +extern int append_terse_output; extern int temp_stall_ts; -extern unsigned long long mlock_size; extern uintptr_t page_mask, page_size; extern int read_only; extern int eta_print; @@ -609,52 +398,24 @@ extern int terse_version; extern int is_backend; extern int nr_clients; extern int log_syslog; +extern int status_interval; extern const char fio_version_string[]; +extern int helper_do_stat; +extern pthread_cond_t helper_cond; +extern char *trigger_file; +extern char *trigger_cmd; +extern char *trigger_remote_cmd; +extern long long trigger_timeout; extern struct thread_data *threads; -static inline void fio_ro_check(struct thread_data *td, struct io_u *io_u) +static inline void fio_ro_check(const struct thread_data *td, struct io_u *io_u) { assert(!(io_u->ddir == DDIR_WRITE && !td_write(td))); } #define REAL_MAX_JOBS 2048 -static inline enum error_type_bit td_error_type(enum fio_ddir ddir, int err) -{ - if (err == EILSEQ) - return ERROR_TYPE_VERIFY_BIT; - if (ddir == DDIR_READ) - return ERROR_TYPE_READ_BIT; - return ERROR_TYPE_WRITE_BIT; -} - -static int __NON_FATAL_ERR[] = {EIO, EILSEQ}; -static inline int td_non_fatal_error(struct thread_data *td, - enum error_type_bit etype, int err) -{ - int i; - if (!td->o.ignore_error[etype]) { - td->o.ignore_error[etype] = __NON_FATAL_ERR; - td->o.ignore_error_nr[etype] = sizeof(__NON_FATAL_ERR) - / sizeof(int); - } - - if (!(td->o.continue_on_error & (1 << etype))) - return 0; - for (i = 0; i < td->o.ignore_error_nr[etype]; i++) - if (td->o.ignore_error[etype][i] == err) - return 1; - return 0; -} - -static inline void update_error_count(struct thread_data *td, int err) -{ - td->total_err_count++; - if (td->total_err_count == 1) - td->first_error = err; -} - static inline int should_fsync(struct thread_data *td) { if (td->last_was_sync) @@ -668,13 +429,14 @@ static inline int should_fsync(struct thread_data *td) /* * Init/option functions */ +extern int __must_check fio_init_options(void); 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 parse_jobs_ini(char *, int, int, int); +extern int parse_cmd_line(int, char **, int); 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 int fio_options_parse(struct thread_data *, char **, int, int); extern void fio_keywords_init(void); extern int fio_cmd_option_parse(struct thread_data *, const char *, char *); extern int fio_cmd_ioengine_option_parse(struct thread_data *, const char *, char *); @@ -685,9 +447,16 @@ extern void fio_options_dup_and_init(struct option *); extern void fio_options_mem_dupe(struct thread_data *); extern void options_mem_dupe(void *data, struct fio_option *options); extern void td_fill_rand_seeds(struct thread_data *); -extern void add_job_opts(const char **); -extern char *num2str(unsigned long, int, int, int); +extern void add_job_opts(const char **, int); +extern char *num2str(uint64_t, int, int, int, int); extern int ioengine_load(struct thread_data *); +extern int parse_dryrun(void); +extern int fio_running_or_pending_io_threads(void); +extern int fio_set_fd_nonblocking(int, const char *); + +extern uintptr_t page_mask; +extern uintptr_t page_size; +extern int initialize_fio(char *envp[]); #define FIO_GETOPT_JOB 0x89000000 #define FIO_GETOPT_IOENGINE 0x98000000 @@ -698,6 +467,7 @@ extern int ioengine_load(struct thread_data *); */ extern void print_thread_status(void); extern void print_status_init(int); +extern char *fio_uint_to_kmg(unsigned int val); /* * Thread life cycle. Once a thread has a runstate beyond TD_INITIALIZED, it @@ -715,21 +485,33 @@ enum { TD_PRE_READING, TD_VERIFYING, TD_FSYNCING, + TD_FINISHING, TD_EXITED, TD_REAPED, }; extern void td_set_runstate(struct thread_data *, int); +extern int td_bump_runstate(struct thread_data *, int); +extern void td_restore_runstate(struct thread_data *, int); + +/* + * Allow 60 seconds for a job to quit on its own, otherwise reap with + * a vengeance. + */ +#define FIO_REAP_TIMEOUT 60 + #define TERMINATE_ALL (-1) extern void fio_terminate_threads(int); +extern void fio_mark_td_terminate(struct thread_data *); /* * Memory helpers */ -extern int __must_check fio_pin_memory(void); -extern void fio_unpin_memory(void); +extern int __must_check fio_pin_memory(struct thread_data *); +extern void fio_unpin_memory(struct thread_data *); extern int __must_check allocate_io_mem(struct thread_data *); extern void free_io_mem(struct thread_data *); +extern void free_threads_shm(void); /* * Reset stats after ramp time completes @@ -740,16 +522,16 @@ extern void reset_all_stats(struct thread_data *); * blktrace support */ #ifdef FIO_HAVE_BLKTRACE -extern int is_blktrace(const char *); -extern int load_blktrace(struct thread_data *, const char *); +extern int is_blktrace(const char *, int *); +extern int load_blktrace(struct thread_data *, const char *, int); #endif /* - * Mark unused variables passed to ops functions as unused, to silence gcc + * Latency target helpers */ -#define fio_unused __attribute((__unused__)) -#define fio_init __attribute__((constructor)) -#define fio_exit __attribute__((destructor)) +extern void lat_target_check(struct thread_data *); +extern void lat_target_init(struct thread_data *); +extern void lat_target_reset(struct thread_data *); #define for_each_td(td, i) \ for ((i) = 0, (td) = &threads[0]; (i) < (int) thread_number; (i)++, (td)++) @@ -793,16 +575,15 @@ static inline int __should_check_rate(struct thread_data *td, return 0; } -static inline int should_check_rate(struct thread_data *td, - uint64_t *bytes_done) +static inline int should_check_rate(struct thread_data *td) { int ret = 0; - if (bytes_done[DDIR_READ]) + if (td->bytes_done[DDIR_READ]) ret |= __should_check_rate(td, DDIR_READ); - if (bytes_done[DDIR_WRITE]) + if (td->bytes_done[DDIR_WRITE]) ret |= __should_check_rate(td, DDIR_WRITE); - if (bytes_done[DDIR_TRIM]) + if (td->bytes_done[DDIR_TRIM]) ret |= __should_check_rate(td, DDIR_TRIM); return ret; @@ -816,7 +597,15 @@ static inline unsigned int td_max_bs(struct thread_data *td) return max(td->o.max_bs[DDIR_TRIM], max_bs); } -static inline int is_power_of_2(unsigned int val) +static inline unsigned int td_min_bs(struct thread_data *td) +{ + unsigned int min_bs; + + min_bs = min(td->o.min_bs[DDIR_READ], td->o.min_bs[DDIR_WRITE]); + return min(td->o.min_bs[DDIR_TRIM], min_bs); +} + +static inline int is_power_of_2(uint64_t val) { return (val != 0 && ((val & (val - 1)) == 0)); } @@ -846,6 +635,10 @@ static inline void td_io_u_free_notify(struct thread_data *td) extern const char *fio_get_arch_string(int); extern const char *fio_get_os_string(int); +#ifdef FIO_INTERNAL +#define ARRAY_SIZE(x) (sizeof((x)) / (sizeof((x)[0]))) +#endif + enum { FIO_OUTPUT_TERSE = 0, FIO_OUTPUT_JSON, @@ -856,11 +649,23 @@ enum { FIO_RAND_DIST_RANDOM = 0, FIO_RAND_DIST_ZIPF, FIO_RAND_DIST_PARETO, + FIO_RAND_DIST_GAUSS, }; +#define FIO_DEF_ZIPF 1.1 +#define FIO_DEF_PARETO 0.2 + enum { FIO_RAND_GEN_TAUSWORTHE = 0, FIO_RAND_GEN_LFSR, }; +enum { + FIO_CPUS_SHARED = 0, + FIO_CPUS_SPLIT, +}; + +extern void exec_trigger(const char *); +extern void check_trigger_file(void); + #endif