X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=iolog.c;h=760d7b0a43d97962bb7d065c6e464c7b89a7f5f5;hp=2bc3e3a1ca9961d0d2ac1f88fb40f0afcafe18b7;hb=4a06cb413586b95ebc271b0985e907ddcdb3dc4d;hpb=af2fde1961fe6847b7f3dbbc74aa46ec09d0e129

diff --git a/iolog.c b/iolog.c
index 2bc3e3a1..760d7b0a 100644
--- a/iolog.c
+++ b/iolog.c
@@ -19,6 +19,7 @@
 #include "trim.h"
 #include "filelock.h"
 #include "smalloc.h"
+#include "blktrace.h"
 
 static int iolog_flush(struct io_log *log);
 
@@ -64,7 +65,7 @@ static void iolog_delay(struct thread_data *td, unsigned long delay)
 {
 	uint64_t usec = utime_since_now(&td->last_issue);
 	uint64_t this_delay;
-	struct timeval tv;
+	struct timespec ts;
 
 	if (delay < td->time_offset) {
 		td->time_offset = 0;
@@ -77,7 +78,7 @@ static void iolog_delay(struct thread_data *td, unsigned long delay)
 
 	delay -= usec;
 
-	fio_gettime(&tv, NULL);
+	fio_gettime(&ts, NULL);
 	while (delay && !td->terminate) {
 		this_delay = delay;
 		if (this_delay > 500000)
@@ -87,7 +88,7 @@ static void iolog_delay(struct thread_data *td, unsigned long delay)
 		delay -= this_delay;
 	}
 
-	usec = utime_since_now(&tv);
+	usec = utime_since_now(&ts);
 	if (usec > delay)
 		td->time_offset = usec - delay;
 	else
@@ -109,6 +110,11 @@ static int ipo_special(struct thread_data *td, struct io_piece *ipo)
 
 	switch (ipo->file_action) {
 	case FIO_LOG_OPEN_FILE:
+		if (td->o.replay_redirect && fio_file_open(f)) {
+			dprint(FD_FILE, "iolog: ignoring re-open of file %s\n",
+					f->file_name);
+			break;
+		}
 		ret = td_io_open_file(td, f);
 		if (!ret)
 			break;
@@ -221,21 +227,16 @@ 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.
+	 * We don't need to sort the entries if we only performed sequential
+	 * writes. In this case, just reading back data in the order we wrote
+	 * it out is the faster but still safe.
 	 *
-	 * 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.
+	 * One exception is 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 (((!td->o.verifysort) || !td_random(td)) &&
+	      file_randommap(td, ipo->file)) {
 		INIT_FLIST_HEAD(&ipo->list);
 		flist_add_tail(&ipo->list, &td->io_hist_list);
 		ipo->flags |= IP_F_ONLIST;
@@ -272,13 +273,14 @@ restart:
 			overlap = 1;
 
 		if (overlap) {
-			dprint(FD_IO, "iolog: overlap %llu/%lu, %llu/%lu",
+			dprint(FD_IO, "iolog: overlap %llu/%lu, %llu/%lu\n",
 				__ipo->offset, __ipo->len,
 				ipo->offset, ipo->len);
 			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;
 		}
 	}
@@ -396,8 +398,14 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 		} else if (r == 2) {
 			rw = DDIR_INVAL;
 			if (!strcmp(act, "add")) {
-				fileno = add_file(td, fname, 0, 1);
-				file_action = FIO_LOG_ADD_FILE;
+				if (td->o.replay_redirect &&
+				    get_fileno(td, fname) != -1) {
+					dprint(FD_FILE, "iolog: ignoring"
+						" re-add of file %s\n", fname);
+				} else {
+					fileno = add_file(td, fname, 0, 1);
+					file_action = FIO_LOG_ADD_FILE;
+				}
 				continue;
 			} else if (!strcmp(act, "open")) {
 				fileno = get_fileno(td, fname);
@@ -411,7 +419,7 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 				continue;
 			}
 		} else {
-			log_err("bad iolog2: %s", p);
+			log_err("bad iolog2: %s\n", p);
 			continue;
 		}
 
@@ -443,7 +451,12 @@ static int read_iolog2(struct thread_data *td, FILE *f)
 		if (rw == DDIR_WAIT) {
 			ipo->delay = offset;
 		} else {
-			ipo->offset = offset;
+			if (td->o.replay_scale)
+				ipo->offset = offset / td->o.replay_scale;
+			else
+				ipo->offset = offset;
+			ipo_bytes_align(td->o.replay_align, ipo);
+
 			ipo->len = bytes;
 			if (rw != DDIR_INVAL && bytes > td->o.max_bs[rw])
 				td->o.max_bs[rw] = bytes;
@@ -626,6 +639,7 @@ 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;
 	}
 
@@ -680,7 +694,7 @@ void free_log(struct io_log *log)
 	sfree(log);
 }
 
-inline unsigned long hist_sum(int j, int stride, unsigned int *io_u_plat,
+unsigned long hist_sum(int j, int stride, unsigned int *io_u_plat,
 		unsigned int *io_u_plat_last)
 {
 	unsigned long sum;
@@ -1127,6 +1141,42 @@ size_t log_chunk_sizes(struct io_log *log)
 
 #ifdef CONFIG_ZLIB
 
+static bool warned_on_drop;
+
+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 (!warned_on_drop) {
+		log_err("fio: had to drop log entry free\n");
+		warned_on_drop = true;
+	}
+	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;
@@ -1219,7 +1269,7 @@ 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);
@@ -1230,7 +1280,7 @@ static int gz_work(struct iolog_flush_data *data)
 	ret = 0;
 done:
 	if (data->free)
-		free(data);
+		sfree(data);
 	return ret;
 err:
 	while (!flist_empty(&list)) {
@@ -1331,7 +1381,7 @@ int iolog_cur_flush(struct io_log *log, struct io_logs *cur_log)
 {
 	struct iolog_flush_data *data;
 
-	data = malloc(sizeof(*data));
+	data = smalloc(sizeof(*data));
 	if (!data)
 		return 1;
 
@@ -1345,6 +1395,9 @@ 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