Extended I/O accounting

index 1bbc111..4869518 100644
@@ -576,7 +576,9 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq)
         }
 
         block_acct_start(blk_get_stats(blkdev->blk), &ioreq->acct,
-                         ioreq->v.size, BLOCK_ACCT_WRITE);
+                         ioreq->v.size,
+                         ioreq->req.operation == BLKIF_OP_WRITE ?
+                         BLOCK_ACCT_WRITE : BLOCK_ACCT_FLUSH);
         ioreq->aio_inflight++;
         blk_aio_writev(blkdev->blk, ioreq->start / BLOCK_SIZE,
                        &ioreq->v, ioreq->v.size / BLOCK_SIZE,

index 317406d..b559f1b 100644
@@ -895,7 +895,7 @@ static void ide_sector_write(IDEState *s)
     qemu_iovec_init_external(&s->qiov, &s->iov, 1);
 
     block_acct_start(blk_get_stats(s->blk), &s->acct,
-                     n * BDRV_SECTOR_SIZE, BLOCK_ACCT_READ);
+                     n * BDRV_SECTOR_SIZE, BLOCK_ACCT_WRITE);
     s->pio_aiocb = blk_aio_writev(s->blk, sector_num, &s->qiov, n,
                                   ide_sector_write_cb, s);
 }

index a423560..6f4c0f1 100644
@@ -26,13 +26,15 @@
 #include "block/block_int.h"
 #include "qemu/timer.h"
 
+static QEMUClockType clock_type = QEMU_CLOCK_REALTIME;
+
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type)
 {
     assert(type < BLOCK_MAX_IOTYPE);
 
     cookie->bytes = bytes;
-    cookie->start_time_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
+    cookie->start_time_ns = qemu_clock_get_ns(clock_type);
     cookie->type = type;
 }
 

@@ -43,7 +45,7 @@ void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
     stats->nr_bytes[cookie->type] += cookie->bytes;
     stats->nr_ops[cookie->type]++;
     stats->total_time_ns[cookie->type] +=
-        qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - cookie->start_time_ns;
+        qemu_clock_get_ns(clock_type) - cookie->start_time_ns;
 }
 
 

index 6f4c0f1..d427fa8 100644
@@ -40,12 +40,15 @@ void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
 
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
 {
+    int64_t time_ns = qemu_clock_get_ns(clock_type);
+    int64_t latency_ns = time_ns - cookie->start_time_ns;
+
     assert(cookie->type < BLOCK_MAX_IOTYPE);
 
     stats->nr_bytes[cookie->type] += cookie->bytes;
     stats->nr_ops[cookie->type]++;
-    stats->total_time_ns[cookie->type] +=
-        qemu_clock_get_ns(clock_type) - cookie->start_time_ns;
+    stats->total_time_ns[cookie->type] += latency_ns;
+    stats->last_access_time_ns = time_ns;
 }
 
 

@@ -55,3 +58,8 @@ void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
     assert(type < BLOCK_MAX_IOTYPE);
     stats->merged[type] += num_requests;
 }
+
+int64_t block_acct_idle_time_ns(BlockAcctStats *stats)
+{
+    return qemu_clock_get_ns(clock_type) - stats->last_access_time_ns;
+}

index ec0f513..539c2e3 100644
@@ -357,6 +357,11 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
         s->stats->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
         s->stats->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
         s->stats->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
+
+        s->stats->has_idle_time_ns = stats->last_access_time_ns > 0;
+        if (s->stats->has_idle_time_ns) {
+            s->stats->idle_time_ns = block_acct_idle_time_ns(stats);
+        }
     }
 
     s->stats->wr_highest_offset = bs->wr_highest_offset;

index 28caa7d..8ee473e 100644
@@ -522,6 +522,7 @@ void hmp_info_blockstats(Monitor *mon, const QDict *qdict)
                        " flush_total_time_ns=%" PRId64
                        " rd_merged=%" PRId64
                        " wr_merged=%" PRId64
+                       " idle_time_ns=%" PRId64
                        "\n",
                        stats->value->stats->rd_bytes,
                        stats->value->stats->wr_bytes,

@@ -532,7 +533,8 @@ void hmp_info_blockstats(Monitor *mon, const QDict *qdict)
                        stats->value->stats->rd_total_time_ns,
                        stats->value->stats->flush_total_time_ns,
                        stats->value->stats->rd_merged,
-                       stats->value->stats->wr_merged);
+                       stats->value->stats->wr_merged,
+                       stats->value->stats->idle_time_ns);
     }
 
     qapi_free_BlockStatsList(stats_list);

index 66637cd..4b2b999 100644
@@ -40,6 +40,7 @@ typedef struct BlockAcctStats {
     uint64_t nr_ops[BLOCK_MAX_IOTYPE];
     uint64_t total_time_ns[BLOCK_MAX_IOTYPE];
     uint64_t merged[BLOCK_MAX_IOTYPE];
+    int64_t last_access_time_ns;
 } BlockAcctStats;
 
 typedef struct BlockAcctCookie {

@@ -53,5 +54,6 @@ void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie);
 void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
                            int num_requests);
+int64_t block_acct_idle_time_ns(BlockAcctStats *stats);
 
 #endif

index d427fa8..49a9444 100644
@@ -51,6 +51,29 @@ void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
     stats->last_access_time_ns = time_ns;
 }
 
+void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)
+{
+    int64_t time_ns = qemu_clock_get_ns(clock_type);
+
+    assert(cookie->type < BLOCK_MAX_IOTYPE);
+
+    stats->failed_ops[cookie->type]++;
+    stats->total_time_ns[cookie->type] += time_ns - cookie->start_time_ns;
+    stats->last_access_time_ns = time_ns;
+}
+
+void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type)
+{
+    assert(type < BLOCK_MAX_IOTYPE);
+
+    /* block_acct_done() and block_acct_failed() update
+     * total_time_ns[], but this one does not. The reason is that
+     * invalid requests are accounted during their submission,
+     * therefore there's no actual I/O involved. */
+
+    stats->invalid_ops[type]++;
+    stats->last_access_time_ns = qemu_clock_get_ns(clock_type);
+}
 
 void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
                       int num_requests)

index 539c2e3..84d8412 100644
@@ -351,6 +351,16 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
         s->stats->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
         s->stats->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
         s->stats->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
+
+        s->stats->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
+        s->stats->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
+        s->stats->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
+
+        s->stats->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
+        s->stats->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
+        s->stats->invalid_flush_operations =
+            stats->invalid_ops[BLOCK_ACCT_FLUSH];
+
         s->stats->rd_merged = stats->merged[BLOCK_ACCT_READ];
         s->stats->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
         s->stats->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];

index 4b2b999..b50e3cc 100644
@@ -38,6 +38,8 @@ enum BlockAcctType {
 typedef struct BlockAcctStats {
     uint64_t nr_bytes[BLOCK_MAX_IOTYPE];
     uint64_t nr_ops[BLOCK_MAX_IOTYPE];
+    uint64_t invalid_ops[BLOCK_MAX_IOTYPE];
+    uint64_t failed_ops[BLOCK_MAX_IOTYPE];
     uint64_t total_time_ns[BLOCK_MAX_IOTYPE];
     uint64_t merged[BLOCK_MAX_IOTYPE];
     int64_t last_access_time_ns;

@@ -52,6 +54,8 @@ typedef struct BlockAcctCookie {
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type);
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie);
+void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie);
+void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type);
 void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
                            int num_requests);
 int64_t block_acct_idle_time_ns(BlockAcctStats *stats);

index 49a9444..923aeaf 100644
@@ -28,6 +28,13 @@
 
 static QEMUClockType clock_type = QEMU_CLOCK_REALTIME;
 
+void block_acct_init(BlockAcctStats *stats, bool account_invalid,
+                     bool account_failed)
+{
+    stats->account_invalid = account_invalid;
+    stats->account_failed = account_failed;
+}
+
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type)
 {

@@ -53,13 +60,17 @@ void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
 
 void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)
 {
-    int64_t time_ns = qemu_clock_get_ns(clock_type);
-
     assert(cookie->type < BLOCK_MAX_IOTYPE);
 
     stats->failed_ops[cookie->type]++;
-    stats->total_time_ns[cookie->type] += time_ns - cookie->start_time_ns;
-    stats->last_access_time_ns = time_ns;
+
+    if (stats->account_failed) {
+        int64_t time_ns = qemu_clock_get_ns(clock_type);
+        int64_t latency_ns = time_ns - cookie->start_time_ns;
+
+        stats->total_time_ns[cookie->type] += latency_ns;
+        stats->last_access_time_ns = time_ns;
+    }
 }
 
 void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type)

@@ -72,7 +83,10 @@ void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type)
      * therefore there's no actual I/O involved. */
 
     stats->invalid_ops[type]++;
-    stats->last_access_time_ns = qemu_clock_get_ns(clock_type);
+
+    if (stats->account_invalid) {
+        stats->last_access_time_ns = qemu_clock_get_ns(clock_type);
+    }
 }
 
 void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,

index 84d8412..56c8139 100644
@@ -372,6 +372,9 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
         if (s->stats->has_idle_time_ns) {
             s->stats->idle_time_ns = block_acct_idle_time_ns(stats);
         }
+
+        s->stats->account_invalid = stats->account_invalid;
+        s->stats->account_failed = stats->account_failed;
     }
 
     s->stats->wr_highest_offset = bs->wr_highest_offset;

index b79b0a6..94635b5 100644
@@ -467,6 +467,7 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
     const char *buf;
     int bdrv_flags = 0;
     int on_read_error, on_write_error;
+    bool account_invalid, account_failed;
     BlockBackend *blk;
     BlockDriverState *bs;
     ThrottleConfig cfg;

@@ -503,6 +504,9 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
     /* extract parameters */
     snapshot = qemu_opt_get_bool(opts, "snapshot", 0);
 
+    account_invalid = qemu_opt_get_bool(opts, "stats-account-invalid", true);
+    account_failed = qemu_opt_get_bool(opts, "stats-account-failed", true);
+
     extract_common_blockdev_options(opts, &bdrv_flags, &throttling_group, &cfg,
                                     &detect_zeroes, &error);
     if (error) {

@@ -599,6 +603,8 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
         if (bdrv_key_required(bs)) {
             autostart = 0;
         }
+
+        block_acct_init(blk_get_stats(blk), account_invalid, account_failed);
     }
 
     blk_set_on_error(blk, on_read_error, on_write_error);

@@ -3519,6 +3525,16 @@ QemuOptsList qemu_common_drive_opts = {
             .name = "detect-zeroes",
             .type = QEMU_OPT_STRING,
             .help = "try to optimize zero writes (off, on, unmap)",
+        },{
+            .name = "stats-account-invalid",
+            .type = QEMU_OPT_BOOL,
+            .help = "whether to account for invalid I/O operations "
+                    "in the statistics",
+        },{
+            .name = "stats-account-failed",
+            .type = QEMU_OPT_BOOL,
+            .help = "whether to account for failed I/O operations "
+                    "in the statistics",
         },
         { /* end of list */ }
     },

index b50e3cc..0d9b076 100644
@@ -25,6 +25,7 @@
 #define BLOCK_ACCOUNTING_H
 
 #include <stdint.h>
+#include <stdbool.h>
 
 #include "qemu/typedefs.h"
 

@@ -43,6 +44,8 @@ typedef struct BlockAcctStats {
     uint64_t total_time_ns[BLOCK_MAX_IOTYPE];
     uint64_t merged[BLOCK_MAX_IOTYPE];
     int64_t last_access_time_ns;
+    bool account_invalid;
+    bool account_failed;
 } BlockAcctStats;
 
 typedef struct BlockAcctCookie {

@@ -51,6 +54,8 @@ typedef struct BlockAcctCookie {
     enum BlockAcctType type;
 } BlockAcctCookie;
 
+void block_acct_init(BlockAcctStats *stats, bool account_invalid,
+                     bool account_failed);
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type);
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie);

index 61de8ce..a941931 100644
@@ -143,3 +143,15 @@ int64_t block_acct_idle_time_ns(BlockAcctStats *stats)
 {
     return qemu_clock_get_ns(clock_type) - stats->last_access_time_ns;
 }
+
+double block_acct_queue_depth(BlockAcctTimedStats *stats,
+                              enum BlockAcctType type)
+{
+    uint64_t sum, elapsed;
+
+    assert(type < BLOCK_MAX_IOTYPE);
+
+    sum = timed_average_sum(&stats->latency[type], &elapsed);
+
+    return (double) sum / elapsed;
+}

index 4baf6e1..99d5303 100644
@@ -402,6 +402,11 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
             dev_stats->min_flush_latency_ns = timed_average_min(fl);
             dev_stats->max_flush_latency_ns = timed_average_max(fl);
             dev_stats->avg_flush_latency_ns = timed_average_avg(fl);
+
+            dev_stats->avg_rd_queue_depth =
+                block_acct_queue_depth(ts, BLOCK_ACCT_READ);
+            dev_stats->avg_wr_queue_depth =
+                block_acct_queue_depth(ts, BLOCK_ACCT_WRITE);
         }
     }
 

index 09605bb..f41ddde 100644
@@ -78,5 +78,7 @@ void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type);
 void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
                            int num_requests);
 int64_t block_acct_idle_time_ns(BlockAcctStats *stats);
+double block_acct_queue_depth(BlockAcctTimedStats *stats,
+                              enum BlockAcctType type);
 
 #endif

index f1cdddc..364bf88 100644
@@ -59,5 +59,6 @@ void timed_average_account(TimedAverage *ta, uint64_t value);
 uint64_t timed_average_min(TimedAverage *ta);
 uint64_t timed_average_avg(TimedAverage *ta);
 uint64_t timed_average_max(TimedAverage *ta);
+uint64_t timed_average_sum(TimedAverage *ta, uint64_t *elapsed);
 
 #endif

index 98a1170..70926ef 100644
@@ -208,3 +208,20 @@ uint64_t timed_average_max(TimedAverage *ta)
     check_expirations(ta);
     return current_window(ta)->max;
 }
+
+/* Get the sum of all accounted values
+ * @ta:      the TimedAverage structure
+ * @elapsed: if non-NULL, the elapsed time (in ns) will be stored here
+ * @ret:     the sum of all accounted values
+ */
+uint64_t timed_average_sum(TimedAverage *ta, uint64_t *elapsed)
+{
+    TimedAverageWindow *w;
+    check_expirations(ta);
+    w = current_window(ta);
+    if (elapsed != NULL) {
+        int64_t remaining = w->expiration - qemu_clock_get_ns(ta->clock_type);
+        *elapsed = ta->period - remaining;
+    }
+    return w->sum;
+}

index 923aeaf..61de8ce 100644
@@ -35,6 +35,39 @@ void block_acct_init(BlockAcctStats *stats, bool account_invalid,
     stats->account_failed = account_failed;
 }
 
+void block_acct_cleanup(BlockAcctStats *stats)
+{
+    BlockAcctTimedStats *s, *next;
+    QSLIST_FOREACH_SAFE(s, &stats->intervals, entries, next) {
+        g_free(s);
+    }
+}
+
+void block_acct_add_interval(BlockAcctStats *stats, unsigned interval_length)
+{
+    BlockAcctTimedStats *s;
+    unsigned i;
+
+    s = g_new0(BlockAcctTimedStats, 1);
+    s->interval_length = interval_length;
+    QSLIST_INSERT_HEAD(&stats->intervals, s, entries);
+
+    for (i = 0; i < BLOCK_MAX_IOTYPE; i++) {
+        timed_average_init(&s->latency[i], clock_type,
+                           (uint64_t) interval_length * NANOSECONDS_PER_SECOND);
+    }
+}
+
+BlockAcctTimedStats *block_acct_interval_next(BlockAcctStats *stats,
+                                              BlockAcctTimedStats *s)
+{
+    if (s == NULL) {
+        return QSLIST_FIRST(&stats->intervals);
+    } else {
+        return QSLIST_NEXT(s, entries);
+    }
+}
+
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type)
 {

@@ -47,6 +80,7 @@ void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
 
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
 {
+    BlockAcctTimedStats *s;
     int64_t time_ns = qemu_clock_get_ns(clock_type);
     int64_t latency_ns = time_ns - cookie->start_time_ns;
 

@@ -56,6 +90,10 @@ void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
     stats->nr_ops[cookie->type]++;
     stats->total_time_ns[cookie->type] += latency_ns;
     stats->last_access_time_ns = time_ns;
+
+    QSLIST_FOREACH(s, &stats->intervals, entries) {
+        timed_average_account(&s->latency[cookie->type], latency_ns);
+    }
 }
 
 void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)

@@ -65,11 +103,16 @@ void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)
     stats->failed_ops[cookie->type]++;
 
     if (stats->account_failed) {
+        BlockAcctTimedStats *s;
         int64_t time_ns = qemu_clock_get_ns(clock_type);
         int64_t latency_ns = time_ns - cookie->start_time_ns;
 
         stats->total_time_ns[cookie->type] += latency_ns;
         stats->last_access_time_ns = time_ns;
+
+        QSLIST_FOREACH(s, &stats->intervals, entries) {
+            timed_average_account(&s->latency[cookie->type], latency_ns);
+        }
     }
 }
 

index 10e4d71..c3490fb 100644
@@ -176,6 +176,7 @@ static void blk_delete(BlockBackend *blk)
     }
     g_free(blk->name);
     drive_info_del(blk->legacy_dinfo);
+    block_acct_cleanup(&blk->stats);
     g_free(blk);
 }
 

index 56c8139..4baf6e1 100644
@@ -346,6 +346,7 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
     s->stats = g_malloc0(sizeof(*s->stats));
     if (bs->blk) {
         BlockAcctStats *stats = blk_get_stats(bs->blk);
+        BlockAcctTimedStats *ts = NULL;
 
         s->stats->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
         s->stats->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];

@@ -375,6 +376,33 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
 
         s->stats->account_invalid = stats->account_invalid;
         s->stats->account_failed = stats->account_failed;
+
+        while ((ts = block_acct_interval_next(stats, ts))) {
+            BlockDeviceTimedStatsList *timed_stats =
+                g_malloc0(sizeof(*timed_stats));
+            BlockDeviceTimedStats *dev_stats = g_malloc0(sizeof(*dev_stats));
+            timed_stats->next = s->stats->timed_stats;
+            timed_stats->value = dev_stats;
+            s->stats->timed_stats = timed_stats;
+
+            TimedAverage *rd = &ts->latency[BLOCK_ACCT_READ];
+            TimedAverage *wr = &ts->latency[BLOCK_ACCT_WRITE];
+            TimedAverage *fl = &ts->latency[BLOCK_ACCT_FLUSH];
+
+            dev_stats->interval_length = ts->interval_length;
+
+            dev_stats->min_rd_latency_ns = timed_average_min(rd);
+            dev_stats->max_rd_latency_ns = timed_average_max(rd);
+            dev_stats->avg_rd_latency_ns = timed_average_avg(rd);
+
+            dev_stats->min_wr_latency_ns = timed_average_min(wr);
+            dev_stats->max_wr_latency_ns = timed_average_max(wr);
+            dev_stats->avg_wr_latency_ns = timed_average_avg(wr);
+
+            dev_stats->min_flush_latency_ns = timed_average_min(fl);
+            dev_stats->max_flush_latency_ns = timed_average_max(fl);
+            dev_stats->avg_flush_latency_ns = timed_average_avg(fl);
+        }
     }
 
     s->stats->wr_highest_offset = bs->wr_highest_offset;

index 0d9b076..09605bb 100644
@@ -28,6 +28,9 @@
 #include <stdbool.h>
 
 #include "qemu/typedefs.h"
+#include "qemu/timed-average.h"
+
+typedef struct BlockAcctTimedStats BlockAcctTimedStats;
 
 enum BlockAcctType {
     BLOCK_ACCT_READ,

@@ -36,6 +39,12 @@ enum BlockAcctType {
     BLOCK_MAX_IOTYPE,
 };
 
+struct BlockAcctTimedStats {
+    TimedAverage latency[BLOCK_MAX_IOTYPE];
+    unsigned interval_length;
+    QSLIST_ENTRY(BlockAcctTimedStats) entries;
+};
+
 typedef struct BlockAcctStats {
     uint64_t nr_bytes[BLOCK_MAX_IOTYPE];
     uint64_t nr_ops[BLOCK_MAX_IOTYPE];

@@ -44,6 +53,7 @@ typedef struct BlockAcctStats {
     uint64_t total_time_ns[BLOCK_MAX_IOTYPE];
     uint64_t merged[BLOCK_MAX_IOTYPE];
     int64_t last_access_time_ns;
+    QSLIST_HEAD(, BlockAcctTimedStats) intervals;
     bool account_invalid;
     bool account_failed;
 } BlockAcctStats;

@@ -56,6 +66,10 @@ typedef struct BlockAcctCookie {
 
 void block_acct_init(BlockAcctStats *stats, bool account_invalid,
                      bool account_failed);
+void block_acct_cleanup(BlockAcctStats *stats);
+void block_acct_add_interval(BlockAcctStats *stats, unsigned interval_length);
+BlockAcctTimedStats *block_acct_interval_next(BlockAcctStats *stats,
+                                              BlockAcctTimedStats *s);
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type);
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie);

new file mode 100644
index 0000000..f1cdddc
@@ -0,0 +1,63 @@
+/*
+ * QEMU timed average computation
+ *
+ * Copyright (C) Nodalink, EURL. 2014
+ * Copyright (C) Igalia, S.L. 2015
+ *
+ * Authors:
+ *   Benoît Canet <[email protected]>
+ *   Alberto Garcia <[email protected]>
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 2 of the License, or
+ * (at your option) version 3 or any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef TIMED_AVERAGE_H
+#define TIMED_AVERAGE_H
+
+#include <stdint.h>
+
+#include "qemu/timer.h"
+
+typedef struct TimedAverageWindow TimedAverageWindow;
+typedef struct TimedAverage TimedAverage;
+
+/* All fields of both structures are private */
+
+struct TimedAverageWindow {
+    uint64_t      min;             /* minimum value accounted in the window */
+    uint64_t      max;             /* maximum value accounted in the window */
+    uint64_t      sum;             /* sum of all values */
+    uint64_t      count;           /* number of values */
+    int64_t       expiration;      /* the end of the current window in ns */
+};
+
+struct TimedAverage {
+    uint64_t           period;     /* period in nanoseconds */
+    TimedAverageWindow windows[2]; /* two overlapping windows of with
+                                    * an offset of period / 2 between them */
+    unsigned           current;    /* the current window index: it's also the
+                                    * oldest window index */
+    QEMUClockType      clock_type; /* the clock used */
+};
+
+void timed_average_init(TimedAverage *ta, QEMUClockType clock_type,
+                        uint64_t period);
+
+void timed_average_account(TimedAverage *ta, uint64_t value);
+
+uint64_t timed_average_min(TimedAverage *ta);
+uint64_t timed_average_avg(TimedAverage *ta);
+uint64_t timed_average_max(TimedAverage *ta);
+
+#endif

90 lines not on this page · raw message

210 lines not on this page · raw message

index 05a5c5f..185025e 100644
@@ -2,6 +2,7 @@
  * QEMU System Emulator block accounting
  *
  * Copyright (c) 2011 Christoph Hellwig
+ * Copyright (c) 2015 Igalia, S.L.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal

index f41ddde..0215a4a 100644
@@ -2,6 +2,7 @@
  * QEMU System Emulator block accounting
  *
  * Copyright (c) 2011 Christoph Hellwig
+ * Copyright (c) 2015 Igalia, S.L.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal

index 8beb26b..aea9728 100644
@@ -76,7 +76,7 @@ static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
         s->rq = req;
     } else if (action == BLOCK_ERROR_ACTION_REPORT) {
         virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
-        block_acct_done(blk_get_stats(s->blk), &req->acct);
+        block_acct_failed(blk_get_stats(s->blk), &req->acct);
         virtio_blk_free_request(req);
     }
 

@@ -536,6 +536,8 @@ void virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
         if (!virtio_blk_sect_range_ok(req->dev, req->sector_num,
                                       req->qiov.size)) {
             virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
+            block_acct_invalid(blk_get_stats(req->dev->blk),
+                               is_write ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ);
             virtio_blk_free_request(req);
             return;
         }

index a941931..05a5c5f 100644
@@ -25,14 +25,20 @@
 #include "block/accounting.h"
 #include "block/block_int.h"
 #include "qemu/timer.h"
+#include "sysemu/qtest.h"
 
 static QEMUClockType clock_type = QEMU_CLOCK_REALTIME;
+static const int qtest_latency_ns = NANOSECONDS_PER_SECOND / 1000;
 
 void block_acct_init(BlockAcctStats *stats, bool account_invalid,
                      bool account_failed)
 {
     stats->account_invalid = account_invalid;
     stats->account_failed = account_failed;
+
+    if (qtest_enabled()) {
+        clock_type = QEMU_CLOCK_VIRTUAL;
+    }
 }
 
 void block_acct_cleanup(BlockAcctStats *stats)

@@ -84,6 +90,10 @@ void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
     int64_t time_ns = qemu_clock_get_ns(clock_type);
     int64_t latency_ns = time_ns - cookie->start_time_ns;
 
+    if (qtest_enabled()) {
+        latency_ns = qtest_latency_ns;
+    }
+
     assert(cookie->type < BLOCK_MAX_IOTYPE);
 
     stats->nr_bytes[cookie->type] += cookie->bytes;

@@ -107,6 +117,10 @@ void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)
         int64_t time_ns = qemu_clock_get_ns(clock_type);
         int64_t latency_ns = time_ns - cookie->start_time_ns;
 
+        if (qtest_enabled()) {
+            latency_ns = qtest_latency_ns;
+        }
+
         stats->total_time_ns[cookie->type] += latency_ns;
         stats->last_access_time_ns = time_ns;
 

index 5da41b2..169e4fa 100644
@@ -201,10 +201,11 @@ static void nvme_rw_cb(void *opaque, int ret)
     NvmeCtrl *n = sq->ctrl;
     NvmeCQueue *cq = n->cq[sq->cqid];
 
-    block_acct_done(blk_get_stats(n->conf.blk), &req->acct);
     if (!ret) {
+        block_acct_done(blk_get_stats(n->conf.blk), &req->acct);
         req->status = NVME_SUCCESS;
     } else {
+        block_acct_failed(blk_get_stats(n->conf.blk), &req->acct);
         req->status = NVME_INTERNAL_DEV_ERROR;
     }
     if (req->has_sg) {

@@ -238,18 +239,22 @@ static uint16_t nvme_rw(NvmeCtrl *n, NvmeNamespace *ns, NvmeCmd *cmd,
     uint64_t data_size = (uint64_t)nlb << data_shift;
     uint64_t aio_slba  = slba << (data_shift - BDRV_SECTOR_BITS);
     int is_write = rw->opcode == NVME_CMD_WRITE ? 1 : 0;
+    enum BlockAcctType acct = is_write ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ;
 
     if ((slba + nlb) > ns->id_ns.nsze) {
+        block_acct_invalid(blk_get_stats(n->conf.blk), acct);
         return NVME_LBA_RANGE | NVME_DNR;
     }
+
     if (nvme_map_prp(&req->qsg, prp1, prp2, data_size, n)) {
+        block_acct_invalid(blk_get_stats(n->conf.blk), acct);
         return NVME_INVALID_FIELD | NVME_DNR;
     }
+
     assert((nlb << data_shift) == req->qsg.size);
 
     req->has_sg = true;
-    dma_acct_start(n->conf.blk, &req->acct, &req->qsg,
-                   is_write ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ);
+    dma_acct_start(n->conf.blk, &req->acct, &req->qsg, acct);
     req->aiocb = is_write ?
         dma_blk_write(n->conf.blk, &req->qsg, aio_slba, nvme_rw_cb, req) :
         dma_blk_read(n->conf.blk, &req->qsg, aio_slba, nvme_rw_cb, req);

index 747f466..cf0b78e 100644
@@ -108,27 +108,30 @@ static void cd_data_to_raw(uint8_t *buf, int lba)
 static int cd_read_sector(IDEState *s, int lba, uint8_t *buf, int sector_size)
 {
     int ret;
+    block_acct_start(blk_get_stats(s->blk), &s->acct,
+                     4 * BDRV_SECTOR_SIZE, BLOCK_ACCT_READ);
 
     switch(sector_size) {
     case 2048:
-        block_acct_start(blk_get_stats(s->blk), &s->acct,
-                         4 * BDRV_SECTOR_SIZE, BLOCK_ACCT_READ);
         ret = blk_read(s->blk, (int64_t)lba << 2, buf, 4);
-        block_acct_done(blk_get_stats(s->blk), &s->acct);
         break;
     case 2352:
-        block_acct_start(blk_get_stats(s->blk), &s->acct,
-                         4 * BDRV_SECTOR_SIZE, BLOCK_ACCT_READ);
         ret = blk_read(s->blk, (int64_t)lba << 2, buf + 16, 4);
-        block_acct_done(blk_get_stats(s->blk), &s->acct);
-        if (ret < 0)
-            return ret;
-        cd_data_to_raw(buf, lba);
+        if (ret >= 0) {
+            cd_data_to_raw(buf, lba);
+        }
         break;
     default:
-        ret = -EIO;
-        break;
+        block_acct_invalid(blk_get_stats(s->blk), BLOCK_ACCT_READ);
+        return -EIO;
     }
+
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->blk), &s->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->blk), &s->acct);
+    }
+
     return ret;
 }
 

@@ -357,7 +360,11 @@ static void ide_atapi_cmd_read_dma_cb(void *opaque, int ret)
     return;
 
 eot:
-    block_acct_done(blk_get_stats(s->blk), &s->acct);
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->blk), &s->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->blk), &s->acct);
+    }
     ide_set_inactive(s, false);
 }
 

index b559f1b..dd7e9c5 100644
@@ -574,7 +574,6 @@ static void ide_sector_read_cb(void *opaque, int ret)
     if (ret == -ECANCELED) {
         return;
     }
-    block_acct_done(blk_get_stats(s->blk), &s->acct);
     if (ret != 0) {
         if (ide_handle_rw_error(s, -ret, IDE_RETRY_PIO |
                                 IDE_RETRY_READ)) {

@@ -582,6 +581,8 @@ static void ide_sector_read_cb(void *opaque, int ret)
         }
     }
 
+    block_acct_done(blk_get_stats(s->blk), &s->acct);
+
     n = s->nsector;
     if (n > s->req_nb_sectors) {
         n = s->req_nb_sectors;

@@ -621,6 +622,7 @@ static void ide_sector_read(IDEState *s)
 
     if (!ide_sect_range_ok(s, sector_num, n)) {
         ide_rw_error(s);
+        block_acct_invalid(blk_get_stats(s->blk), BLOCK_ACCT_READ);
         return;
     }
 

@@ -672,6 +674,7 @@ static int ide_handle_rw_error(IDEState *s, int error, int op)
         assert(s->bus->retry_unit == s->unit);
         s->bus->error_status = op;
     } else if (action == BLOCK_ERROR_ACTION_REPORT) {
+        block_acct_failed(blk_get_stats(s->blk), &s->acct);
         if (op & IDE_RETRY_DMA) {
             ide_dma_error(s);
         } else {

@@ -750,6 +753,7 @@ static void ide_dma_cb(void *opaque, int ret)
     if ((s->dma_cmd == IDE_DMA_READ || s->dma_cmd == IDE_DMA_WRITE) &&
         !ide_sect_range_ok(s, sector_num, n)) {
         ide_dma_error(s);
+        block_acct_invalid(blk_get_stats(s->blk), s->acct.type);
         return;
     }
 

@@ -826,7 +830,6 @@ static void ide_sector_write_cb(void *opaque, int ret)
     if (ret == -ECANCELED) {
         return;
     }
-    block_acct_done(blk_get_stats(s->blk), &s->acct);
 
     s->pio_aiocb = NULL;
     s->status &= ~BUSY_STAT;

@@ -837,6 +840,8 @@ static void ide_sector_write_cb(void *opaque, int ret)
         }
     }
 
+    block_acct_done(blk_get_stats(s->blk), &s->acct);
+
     n = s->nsector;
     if (n > s->req_nb_sectors) {
         n = s->req_nb_sectors;

@@ -887,6 +892,7 @@ static void ide_sector_write(IDEState *s)
 
     if (!ide_sect_range_ok(s, sector_num, n)) {
         ide_rw_error(s);
+        block_acct_invalid(blk_get_stats(s->blk), BLOCK_ACCT_WRITE);
         return;
     }
 

index 66ac2ba..176e331 100644
@@ -286,7 +286,11 @@ static void pmac_ide_atapi_transfer_cb(void *opaque, int ret)
     return;
 
 done:
-    block_acct_done(blk_get_stats(s->blk), &s->acct);
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->blk), &s->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->blk), &s->acct);
+    }
     io->dma_end(opaque);
 
     return;

@@ -348,7 +352,11 @@ static void pmac_ide_transfer_cb(void *opaque, int ret)
 
 done:
     if (s->dma_cmd == IDE_DMA_READ || s->dma_cmd == IDE_DMA_WRITE) {
-        block_acct_done(blk_get_stats(s->blk), &s->acct);
+        if (ret < 0) {
+            block_acct_failed(blk_get_stats(s->blk), &s->acct);
+        } else {
+            block_acct_done(blk_get_stats(s->blk), &s->acct);
+        }
     }
     io->dma_end(opaque);
 }

index bada9a7..20a31a7 100644
@@ -90,7 +90,7 @@ struct SCSIDiskState
     bool tray_locked;
 };
 
-static int scsi_handle_rw_error(SCSIDiskReq *r, int error);
+static int scsi_handle_rw_error(SCSIDiskReq *r, int error, bool acct_failed);
 
 static void scsi_free_request(SCSIRequest *req)
 {

@@ -169,18 +169,18 @@ static void scsi_aio_complete(void *opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     if (r->req.io_canceled) {
         scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, true)) {
             goto done;
         }
     }
 
+    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     scsi_req_complete(&r->req, GOOD);
 
 done:

@@ -247,7 +247,7 @@ static void scsi_dma_complete_noio(SCSIDiskReq *r, int ret)
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, false)) {
             goto done;
         }
     }

@@ -273,7 +273,11 @@ static void scsi_dma_complete(void *opaque, int ret)
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    }
     scsi_dma_complete_noio(r, ret);
 }
 

@@ -285,18 +289,18 @@ static void scsi_read_complete(void * opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     if (r->req.io_canceled) {
         scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, true)) {
             goto done;
         }
     }
 
+    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     DPRINTF("Data ready tag=0x%x len=%zd\n", r->req.tag, r->qiov.size);
 
     n = r->qiov.size / 512;

@@ -322,7 +326,7 @@ static void scsi_do_read(SCSIDiskReq *r, int ret)
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, false)) {
             goto done;
         }
     }

@@ -355,7 +359,11 @@ static void scsi_do_read_cb(void *opaque, int ret)
     assert (r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    }
     scsi_do_read(opaque, ret);
 }
 

@@ -407,7 +415,7 @@ static void scsi_read_data(SCSIRequest *req)
  * scsi_handle_rw_error always manages its reference counts, independent
  * of the return value.
  */
-static int scsi_handle_rw_error(SCSIDiskReq *r, int error)
+static int scsi_handle_rw_error(SCSIDiskReq *r, int error, bool acct_failed)
 {
     bool is_read = (r->req.cmd.mode == SCSI_XFER_FROM_DEV);
     SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);

@@ -415,6 +423,9 @@ static int scsi_handle_rw_error(SCSIDiskReq *r, int error)
                                                    is_read, error);
 
     if (action == BLOCK_ERROR_ACTION_REPORT) {
+        if (acct_failed) {
+            block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
+        }
         switch (error) {
         case ENOMEDIUM:
             scsi_check_condition(r, SENSE_CODE(NO_MEDIUM));

@@ -452,7 +463,7 @@ static void scsi_write_complete_noio(SCSIDiskReq *r, int ret)
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, false)) {
             goto done;
         }
     }

@@ -481,7 +492,11 @@ static void scsi_write_complete(void * opaque, int ret)
     assert (r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    if (ret < 0) {
+        block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    } else {
+        block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+    }
     scsi_write_complete_noio(r, ret);
 }
 

@@ -1592,7 +1607,7 @@ static void scsi_unmap_complete_noio(UnmapCBData *data, int ret)
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, false)) {
             goto done;
         }
     }

@@ -1696,18 +1711,19 @@ static void scsi_write_same_complete(void *opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
-    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     if (r->req.io_canceled) {
         scsi_req_cancel_complete(&r->req);
         goto done;
     }
 
     if (ret < 0) {
-        if (scsi_handle_rw_error(r, -ret)) {
+        if (scsi_handle_rw_error(r, -ret, true)) {
             goto done;
         }
     }
 
+    block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
+
     data->nb_sectors -= data->iov.iov_len / 512;
     data->sector += data->iov.iov_len / 512;
     data->iov.iov_len = MIN(data->nb_sectors * 512, data->iov.iov_len);

index 6e5d1e4..0cac623 100644
@@ -1364,6 +1364,7 @@ static void aio_write_done(void *opaque, int ret)
 
     if (ret < 0) {
         printf("aio_write failed: %s\n", strerror(-ret));
+        block_acct_failed(blk_get_stats(ctx->blk), &ctx->acct);
         goto out;
     }
 

@@ -1392,6 +1393,7 @@ static void aio_read_done(void *opaque, int ret)
 
     if (ret < 0) {
         printf("readv failed: %s\n", strerror(-ret));
+        block_acct_failed(blk_get_stats(ctx->blk), &ctx->acct);
         goto out;
     }
 

@@ -1505,6 +1507,7 @@ static int aio_read_f(BlockBackend *blk, int argc, char **argv)
     if (ctx->offset & 0x1ff) {
         printf("offset %" PRId64 " is not sector aligned\n",
                ctx->offset);
+        block_acct_invalid(blk_get_stats(blk), BLOCK_ACCT_READ);
         g_free(ctx);
         return 0;
     }

@@ -1512,6 +1515,7 @@ static int aio_read_f(BlockBackend *blk, int argc, char **argv)
     nr_iov = argc - optind;
     ctx->buf = create_iovec(blk, &ctx->qiov, &argv[optind], nr_iov, 0xab);
     if (ctx->buf == NULL) {
+        block_acct_invalid(blk_get_stats(blk), BLOCK_ACCT_READ);
         g_free(ctx);
         return 0;
     }

@@ -1600,6 +1604,7 @@ static int aio_write_f(BlockBackend *blk, int argc, char **argv)
     if (ctx->offset & 0x1ff) {
         printf("offset %" PRId64 " is not sector aligned\n",
                ctx->offset);
+        block_acct_invalid(blk_get_stats(blk), BLOCK_ACCT_WRITE);
         g_free(ctx);
         return 0;
     }

@@ -1607,6 +1612,7 @@ static int aio_write_f(BlockBackend *blk, int argc, char **argv)
     nr_iov = argc - optind;
     ctx->buf = create_iovec(blk, &ctx->qiov, &argv[optind], nr_iov, pattern);
     if (ctx->buf == NULL) {
+        block_acct_invalid(blk_get_stats(blk), BLOCK_ACCT_WRITE);
         g_free(ctx);
         return 0;
     }

@@ -1621,7 +1627,10 @@ static int aio_write_f(BlockBackend *blk, int argc, char **argv)
 
 static int aio_flush_f(BlockBackend *blk, int argc, char **argv)
 {
+    BlockAcctCookie cookie;
+    block_acct_start(blk_get_stats(blk), &cookie, 0, BLOCK_ACCT_FLUSH);
     blk_drain_all();
+    block_acct_done(blk_get_stats(blk), &cookie);
     return 0;
 }
 

index 4869518..02eda6e 100644
@@ -537,7 +537,11 @@ static void qemu_aio_complete(void *opaque, int ret)
             break;
         }
     case BLKIF_OP_READ:
-        block_acct_done(blk_get_stats(ioreq->blkdev->blk), &ioreq->acct);
+        if (ioreq->status == BLKIF_RSP_OKAY) {
+            block_acct_done(blk_get_stats(ioreq->blkdev->blk), &ioreq->acct);
+        } else {
+            block_acct_failed(blk_get_stats(ioreq->blkdev->blk), &ioreq->acct);
+        }
         break;
     case BLKIF_OP_DISCARD:
     default:

@@ -722,6 +726,23 @@ static void blk_handle_requests(struct XenBlkDev *blkdev)
 
         /* parse them */
         if (ioreq_parse(ioreq) != 0) {
+
+            switch (ioreq->req.operation) {
+            case BLKIF_OP_READ:
+                block_acct_invalid(blk_get_stats(blkdev->blk),
+                                   BLOCK_ACCT_READ);
+                break;
+            case BLKIF_OP_WRITE:
+                block_acct_invalid(blk_get_stats(blkdev->blk),
+                                   BLOCK_ACCT_WRITE);
+                break;
+            case BLKIF_OP_FLUSH_DISKCACHE:
+                block_acct_invalid(blk_get_stats(blkdev->blk),
+                                   BLOCK_ACCT_FLUSH);
+            default:
+                break;
+            };
+
             if (blk_send_response_one(ioreq)) {
                 xen_be_send_notify(&blkdev->xendev);
             }

index 94635b5..c316c0a 100644
@@ -468,6 +468,7 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
     int bdrv_flags = 0;
     int on_read_error, on_write_error;
     bool account_invalid, account_failed;
+    const char *stats_intervals;
     BlockBackend *blk;
     BlockDriverState *bs;
     ThrottleConfig cfg;

@@ -507,6 +508,8 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
     account_invalid = qemu_opt_get_bool(opts, "stats-account-invalid", true);
     account_failed = qemu_opt_get_bool(opts, "stats-account-failed", true);
 
+    stats_intervals = qemu_opt_get(opts, "stats-intervals");
+
     extract_common_blockdev_options(opts, &bdrv_flags, &throttling_group, &cfg,
                                     &detect_zeroes, &error);
     if (error) {

@@ -605,6 +608,35 @@ static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
         }
 
         block_acct_init(blk_get_stats(blk), account_invalid, account_failed);
+
+        if (stats_intervals) {
+            char **intervals = g_strsplit(stats_intervals, ":", 0);
+            unsigned i;
+
+            if (*stats_intervals == '\0') {
+                error_setg(&error, "stats-intervals can't have an empty value");
+            }
+
+            for (i = 0; !error && intervals[i] != NULL; i++) {
+                unsigned long long val;
+                if (parse_uint_full(intervals[i], &val, 10) == 0 &&
+                    val > 0 && val <= UINT_MAX) {
+                    block_acct_add_interval(blk_get_stats(blk), val);
+                } else {
+                    error_setg(&error, "Invalid interval length: '%s'",
+                               intervals[i]);
+                }
+            }
+
+            g_strfreev(intervals);
+
+            if (error) {
+                error_propagate(errp, error);
+                blk_unref(blk);
+                blk = NULL;
+                goto err_no_bs_opts;
+            }
+        }
     }
 
     blk_set_on_error(blk, on_read_error, on_write_error);

@@ -3535,6 +3567,11 @@ QemuOptsList qemu_common_drive_opts = {
             .type = QEMU_OPT_BOOL,
             .help = "whether to account for failed I/O operations "
                     "in the statistics",
+        },{
+            .name = "stats-intervals",
+            .type = QEMU_OPT_STRING,
+            .help = "colon-separated list of intervals "
+                    "for collecting I/O statistics, in seconds",
         },
         { /* end of list */ }
     },