X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=iolog.c;h=d51e49c33cfba215c359a78d30ef3b6023e64041;hp=05cabd0c05abb7ff69a75dbb39b1dd4a0e01f036;hb=5fff95436922873dd9ded8b21af7222689e2ec5d;hpb=dab41b1c4322d968da5d69ceddac85a227675614

diff --git a/iolog.c b/iolog.c
index 05cabd0c..d51e49c3 100644
--- a/iolog.c
+++ b/iolog.c
@@ -4,7 +4,6 @@
  */
 #include <stdio.h>
 #include <stdlib.h>
-#include <libgen.h>
 #include <assert.h>
 #include <sys/types.h>
 #include <sys/stat.h>
@@ -15,11 +14,11 @@
 
 #include "flist.h"
 #include "fio.h"
-#include "verify.h"
 #include "trim.h"
 #include "filelock.h"
 #include "smalloc.h"
 #include "blktrace.h"
+#include "pshared.h"
 
 static int iolog_flush(struct io_log *log);
 
@@ -36,7 +35,7 @@ void log_io_u(const struct thread_data *td, const struct io_u *io_u)
 	if (!td->o.write_iolog_file)
 		return;
 
-	fprintf(td->iolog_f, "%s %s %llu %lu\n", io_u->file->file_name,
+	fprintf(td->iolog_f, "%s %s %llu %llu\n", io_u->file->file_name,
 						io_ddir_name(io_u->ddir),
 						io_u->offset, io_u->buflen);
 }
@@ -64,6 +63,7 @@ void log_file(struct thread_data *td, struct fio_file *f,
 static void iolog_delay(struct thread_data *td, unsigned long delay)
 {
 	uint64_t usec = utime_since_now(&td->last_issue);
+	unsigned long orig_delay = delay;
 	uint64_t this_delay;
 	struct timespec ts;
 
@@ -89,8 +89,8 @@ static void iolog_delay(struct thread_data *td, unsigned long delay)
 	}
 
 	usec = utime_since_now(&ts);
-	if (usec > delay)
-		td->time_offset = usec - delay;
+	if (usec > orig_delay)
+		td->time_offset = usec - orig_delay;
 	else
 		td->time_offset = 0;
 }
@@ -161,7 +161,7 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 			io_u->buflen = ipo->len;
 			io_u->file = td->files[ipo->fileno];
 			get_file(io_u->file);
-			dprint(FD_IO, "iolog: get %llu/%lu/%s\n", io_u->offset,
+			dprint(FD_IO, "iolog: get %llu/%llu/%s\n", io_u->offset,
 						io_u->buflen, io_u->file->file_name);
 			if (ipo->delay)
 				iolog_delay(td, ipo->delay);
@@ -184,7 +184,7 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 void prune_io_piece_log(struct thread_data *td)
 {
 	struct io_piece *ipo;
-	struct rb_node *n;
+	struct fio_rb_node *n;
 
 	while ((n = rb_first(&td->io_hist_tree)) != NULL) {
 		ipo = rb_entry(n, struct io_piece, rb_node);
@@ -208,10 +208,10 @@ void prune_io_piece_log(struct thread_data *td)
  */
 void log_io_piece(struct thread_data *td, struct io_u *io_u)
 {
-	struct rb_node **p, *parent;
+	struct fio_rb_node **p, *parent;
 	struct io_piece *ipo, *__ipo;
 
-	ipo = malloc(sizeof(struct io_piece));
+	ipo = calloc(1, sizeof(struct io_piece));
 	init_ipo(ipo);
 	ipo->file = io_u->file;
 	ipo->offset = io_u->offset;
@@ -227,21 +227,11 @@ void log_io_piece(struct thread_data *td, struct io_u *io_u)
 	}
 
 	/*
-	 * We don't need to sort the entries, if:
-	 *
-	 *	Sequential writes, or
-	 *	Random writes that lay out the file as it goes along
-	 *
-	 * For both these cases, just reading back data in the order we
-	 * wrote it out is the fastest.
-	 *
-	 * One exception is if we don't have a random map AND we are doing
-	 * verifies, in that case we need to check for duplicate blocks and
-	 * drop the old one, which we rely on the rb insert/lookup for
-	 * handling.
+	 * Only sort writes if we don't have a random map in which case we need
+	 * to check for duplicate blocks and drop the old one, which we rely on
+	 * the rb insert/lookup for handling.
 	 */
-	if (((!td->o.verifysort) || !td_random(td) || !td->o.overwrite) &&
-	      (file_randommap(td, ipo->file) || td->o.verify == VERIFY_NONE)) {
+	if (file_randommap(td, ipo->file)) {
 		INIT_FLIST_HEAD(&ipo->list);
 		flist_add_tail(&ipo->list, &td->io_hist_list);
 		ipo->flags |= IP_F_ONLIST;
@@ -284,7 +274,8 @@ restart:
 			td->io_hist_len--;
 			rb_erase(parent, &td->io_hist_tree);
 			remove_trim_entry(td, __ipo);
-			free(__ipo);
+			if (!(__ipo->flags & IP_F_IN_FLIGHT))
+				free(__ipo);
 			goto restart;
 		}
 	}
@@ -324,7 +315,7 @@ void unlog_io_piece(struct thread_data *td, struct io_u *io_u)
 	td->io_hist_len--;
 }
 
-void trim_io_piece(struct thread_data *td, const struct io_u *io_u)
+void trim_io_piece(const struct io_u *io_u)
 {
 	struct io_piece *ipo = io_u->ipo;
 
@@ -347,7 +338,7 @@ void write_iolog_close(struct thread_data *td)
  * Read version 2 iolog data. It is enhanced to include per-file logging,
  * syncs, etc.
  */
-static int read_iolog2(struct thread_data *td, FILE *f)
+static bool read_iolog2(struct thread_data *td, FILE *f)
 {
 	unsigned long long offset;
 	unsigned int bytes;
@@ -449,7 +440,7 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 		/*
 		 * Make note of file
 		 */
-		ipo = malloc(sizeof(*ipo));
+		ipo = calloc(1, sizeof(*ipo));
 		init_ipo(ipo);
 		ipo->ddir = rw;
 		if (rw == DDIR_WAIT) {
@@ -483,7 +474,7 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 	}
 
 	if (!reads && !writes && !waits)
-		return 1;
+		return false;
 	else if (reads && !writes)
 		td->o.td_ddir = TD_DDIR_READ;
 	else if (!reads && writes)
@@ -491,22 +482,22 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 	else
 		td->o.td_ddir = TD_DDIR_RW;
 
-	return 0;
+	return true;
 }
 
 /*
  * open iolog, check version, and call appropriate parser
  */
-static int init_iolog_read(struct thread_data *td)
+static bool init_iolog_read(struct thread_data *td)
 {
 	char buffer[256], *p;
 	FILE *f;
-	int ret;
+	bool ret;
 
 	f = fopen(td->o.read_iolog_file, "r");
 	if (!f) {
 		perror("fopen read iolog");
-		return 1;
+		return false;
 	}
 
 	p = fgets(buffer, sizeof(buffer), f);
@@ -514,7 +505,7 @@ static int init_iolog_read(struct thread_data *td)
 		td_verror(td, errno, "iolog read");
 		log_err("fio: unable to read iolog\n");
 		fclose(f);
-		return 1;
+		return false;
 	}
 
 	/*
@@ -525,7 +516,7 @@ static int init_iolog_read(struct thread_data *td)
 		ret = read_iolog2(td, f);
 	else {
 		log_err("fio: iolog version 1 is no longer supported\n");
-		ret = 1;
+		ret = false;
 	}
 
 	fclose(f);
@@ -535,7 +526,7 @@ static int init_iolog_read(struct thread_data *td)
 /*
  * Set up a log for storing io patterns.
  */
-static int init_iolog_write(struct thread_data *td)
+static bool init_iolog_write(struct thread_data *td)
 {
 	struct fio_file *ff;
 	FILE *f;
@@ -544,7 +535,7 @@ static int init_iolog_write(struct thread_data *td)
 	f = fopen(td->o.write_iolog_file, "a");
 	if (!f) {
 		perror("fopen write iolog");
-		return 1;
+		return false;
 	}
 
 	/*
@@ -559,7 +550,7 @@ static int init_iolog_write(struct thread_data *td)
 	 */
 	if (fprintf(f, "%s\n", iolog_ver2) < 0) {
 		perror("iolog init\n");
-		return 1;
+		return false;
 	}
 
 	/*
@@ -568,12 +559,12 @@ static int init_iolog_write(struct thread_data *td)
 	for_each_file(td, ff, i)
 		log_file(td, ff, FIO_LOG_ADD_FILE);
 
-	return 0;
+	return true;
 }
 
-int init_iolog(struct thread_data *td)
+bool init_iolog(struct thread_data *td)
 {
-	int ret = 0;
+	bool ret;
 
 	if (td->o.read_iolog_file) {
 		int need_swap;
@@ -588,8 +579,10 @@ int init_iolog(struct thread_data *td)
 			ret = init_iolog_read(td);
 	} else if (td->o.write_iolog_file)
 		ret = init_iolog_write(td);
+	else
+		ret = true;
 
-	if (ret)
+	if (!ret)
 		td_verror(td, EINVAL, "failed initializing iolog");
 
 	return ret;
@@ -626,12 +619,12 @@ void setup_log(struct io_log **log, struct log_params *p,
 	}
 
 	if (l->td && l->td->o.io_submit_mode != IO_MODE_OFFLOAD) {
-		struct io_logs *p;
+		struct io_logs *__p;
 
-		p = calloc(1, sizeof(*l->pending));
-		p->max_samples = DEF_LOG_ENTRIES;
-		p->log = calloc(p->max_samples, log_entry_sz(l));
-		l->pending = p;
+		__p = calloc(1, sizeof(*l->pending));
+		__p->max_samples = DEF_LOG_ENTRIES;
+		__p->log = calloc(__p->max_samples, log_entry_sz(l));
+		l->pending = __p;
 	}
 
 	if (l->log_offset)
@@ -643,6 +636,7 @@ void setup_log(struct io_log **log, struct log_params *p,
 		l->log_gz = 0;
 	else if (l->log_gz || l->log_gz_store) {
 		mutex_init_pshared(&l->chunk_lock);
+		mutex_init_pshared(&l->deferred_free_lock);
 		p->td->flags |= TD_F_COMPRESS_LOG;
 	}
 
@@ -697,10 +691,10 @@ void free_log(struct io_log *log)
 	sfree(log);
 }
 
-unsigned long hist_sum(int j, int stride, unsigned int *io_u_plat,
-		unsigned int *io_u_plat_last)
+uint64_t hist_sum(int j, int stride, uint64_t *io_u_plat,
+		uint64_t *io_u_plat_last)
 {
-	unsigned long sum;
+	uint64_t sum;
 	int k;
 
 	if (io_u_plat_last) {
@@ -721,8 +715,8 @@ static void flush_hist_samples(FILE *f, int hist_coarseness, void *samples,
 	int log_offset;
 	uint64_t i, j, nr_samples;
 	struct io_u_plat_entry *entry, *entry_before;
-	unsigned int *io_u_plat;
-	unsigned int *io_u_plat_before;
+	uint64_t *io_u_plat;
+	uint64_t *io_u_plat_before;
 
 	int stride = 1 << hist_coarseness;
 	
@@ -743,13 +737,13 @@ static void flush_hist_samples(FILE *f, int hist_coarseness, void *samples,
 		entry_before = flist_first_entry(&entry->list, struct io_u_plat_entry, list);
 		io_u_plat_before = entry_before->io_u_plat;
 
-		fprintf(f, "%lu, %u, %u, ", (unsigned long) s->time,
-						io_sample_ddir(s), s->bs);
+		fprintf(f, "%lu, %u, %llu, ", (unsigned long) s->time,
+						io_sample_ddir(s), (unsigned long long) s->bs);
 		for (j = 0; j < FIO_IO_U_PLAT_NR - stride; j += stride) {
-			fprintf(f, "%lu, ", hist_sum(j, stride, io_u_plat,
-						io_u_plat_before));
+			fprintf(f, "%llu, ", (unsigned long long)
+			        hist_sum(j, stride, io_u_plat, io_u_plat_before));
 		}
-		fprintf(f, "%lu\n", (unsigned long)
+		fprintf(f, "%llu\n", (unsigned long long)
 		        hist_sum(FIO_IO_U_PLAT_NR - stride, stride, io_u_plat,
 					io_u_plat_before));
 
@@ -776,17 +770,17 @@ void flush_samples(FILE *f, void *samples, uint64_t sample_size)
 		s = __get_sample(samples, log_offset, i);
 
 		if (!log_offset) {
-			fprintf(f, "%lu, %" PRId64 ", %u, %u\n",
+			fprintf(f, "%lu, %" PRId64 ", %u, %llu\n",
 					(unsigned long) s->time,
 					s->data.val,
-					io_sample_ddir(s), s->bs);
+					io_sample_ddir(s), (unsigned long long) s->bs);
 		} else {
 			struct io_sample_offset *so = (void *) s;
 
-			fprintf(f, "%lu, %" PRId64 ", %u, %u, %llu\n",
+			fprintf(f, "%lu, %" PRId64 ", %u, %llu, %llu\n",
 					(unsigned long) s->time,
 					s->data.val,
-					io_sample_ddir(s), s->bs,
+					io_sample_ddir(s), (unsigned long long) s->bs,
 					(unsigned long long) so->offset);
 		}
 	}
@@ -982,7 +976,7 @@ int iolog_file_inflate(const char *file)
 	struct iolog_compress ic;
 	z_stream stream;
 	struct stat sb;
-	ssize_t ret;
+	size_t ret;
 	size_t total;
 	void *buf;
 	FILE *f;
@@ -1004,12 +998,12 @@ int iolog_file_inflate(const char *file)
 	ic.seq = 1;
 
 	ret = fread(ic.buf, ic.len, 1, f);
-	if (ret < 0) {
+	if (ret == 0 && ferror(f)) {
 		perror("fread");
 		fclose(f);
 		free(buf);
 		return 1;
-	} else if (ret != 1) {
+	} else if (ferror(f) || (!feof(f) && ret != 1)) {
 		log_err("fio: short read on reading log\n");
 		fclose(f);
 		free(buf);
@@ -1144,6 +1138,38 @@ size_t log_chunk_sizes(struct io_log *log)
 
 #ifdef CONFIG_ZLIB
 
+static void iolog_put_deferred(struct io_log *log, void *ptr)
+{
+	if (!ptr)
+		return;
+
+	pthread_mutex_lock(&log->deferred_free_lock);
+	if (log->deferred < IOLOG_MAX_DEFER) {
+		log->deferred_items[log->deferred] = ptr;
+		log->deferred++;
+	} else if (!fio_did_warn(FIO_WARN_IOLOG_DROP))
+		log_err("fio: had to drop log entry free\n");
+	pthread_mutex_unlock(&log->deferred_free_lock);
+}
+
+static void iolog_free_deferred(struct io_log *log)
+{
+	int i;
+
+	if (!log->deferred)
+		return;
+
+	pthread_mutex_lock(&log->deferred_free_lock);
+
+	for (i = 0; i < log->deferred; i++) {
+		free(log->deferred_items[i]);
+		log->deferred_items[i] = NULL;
+	}
+
+	log->deferred = 0;
+	pthread_mutex_unlock(&log->deferred_free_lock);
+}
+
 static int gz_work(struct iolog_flush_data *data)
 {
 	struct iolog_compress *c = NULL;
@@ -1236,7 +1262,7 @@ static int gz_work(struct iolog_flush_data *data)
 	if (ret != Z_OK)
 		log_err("fio: deflateEnd %d\n", ret);
 
-	free(data->samples);
+	iolog_put_deferred(data->log, data->samples);
 
 	if (!flist_empty(&list)) {
 		pthread_mutex_lock(&data->log->chunk_lock);
@@ -1362,6 +1388,9 @@ int iolog_cur_flush(struct io_log *log, struct io_logs *cur_log)
 	cur_log->log = NULL;
 
 	workqueue_enqueue(&log->td->log_compress_wq, &data->work);
+
+	iolog_free_deferred(log);
+
 	return 0;
 }
 #else