Skip to contentSkip to conversation

block/accounting: fix unlocked latency histogram reconfiguration races

PATCH Superseded Denis V. Lunev 2 participants 6 messages Jul 23

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#2 lore raw

PATCH 2/3 block/qapi: take stats->lock when reading BlockAcctStats for query-blockstats

In reply to Denis V. Lunev

bdrv_query_blk_stats() reads nr_bytes[], nr_ops[], failed_ops[], invalid_ops[], merged[], total_time_ns[], last_access_time_ns, and the latency histogram bins directly off BlockAcctStats, without taking stats->lock. block_account_one_io() updates the same fields under that lock from whatever iothread completes the I/O, so this is an unsynchronized read of concurrently-updated state.

This is a narrower issue than the one fixed for the histogram setters. query-blockstats and block-latency-histogram-set are both plain (non-allow-oob) QMP commands, so they always run serialized on the monitor thread and can never execute concurrently with each other.

Take stats->lock around the read for consistency.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>
---
 block/qapi.c | 102 ++++++++++++++++++++++++++++++---------------------
 1 file changed, 60 insertions(+), 42 deletions(-)
block/qapi.c +60 -42
index eabfbfc258..c3f20832b7 100644
@@ -535,49 +535,67 @@ static void bdrv_query_blk_stats(BlockDeviceStats *ds, BlockBackend *blk)
     BlockAcctTimedStats *ts = NULL;
     BlockLatencyHistogram *hgram;
 
-    ds->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
-    ds->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
-    ds->zone_append_bytes = stats->nr_bytes[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_bytes = stats->nr_bytes[BLOCK_ACCT_UNMAP];
-    ds->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
-    ds->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
-    ds->zone_append_operations = stats->nr_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_operations = stats->nr_ops[BLOCK_ACCT_UNMAP];
-
-    ds->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
-    ds->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
-    ds->failed_zone_append_operations =
-        stats->failed_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
-    ds->failed_unmap_operations = stats->failed_ops[BLOCK_ACCT_UNMAP];
-
-    ds->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
-    ds->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
-    ds->invalid_zone_append_operations =
-        stats->invalid_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->invalid_flush_operations =
-        stats->invalid_ops[BLOCK_ACCT_FLUSH];
-    ds->invalid_unmap_operations = stats->invalid_ops[BLOCK_ACCT_UNMAP];
-
-    ds->rd_merged = stats->merged[BLOCK_ACCT_READ];
-    ds->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
-    ds->zone_append_merged = stats->merged[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_merged = stats->merged[BLOCK_ACCT_UNMAP];
-    ds->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];
-    ds->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
-    ds->zone_append_total_time_ns =
-        stats->total_time_ns[BLOCK_ACCT_ZONE_APPEND];
-    ds->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
-    ds->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
-    ds->unmap_total_time_ns = stats->total_time_ns[BLOCK_ACCT_UNMAP];
-
-    ds->has_idle_time_ns = stats->last_access_time_ns > 0;
-    if (ds->has_idle_time_ns) {
-        ds->idle_time_ns = block_acct_idle_time_ns(stats);
-    }
+    /*
+     * nr_bytes[] etc. are mutated under stats->lock by block_account_one_io()
+     * from an iothread, and latency_histogram[].{nbins,boundaries,bins} can
+     * be freed and reallocated by a concurrent block_latency_histogram_set()
+     * on the monitor thread, so every read here needs the same lock.
+     */
+    WITH_QEMU_LOCK_GUARD(&stats->lock) {
+        ds->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
+        ds->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
+        ds->zone_append_bytes = stats->nr_bytes[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_bytes = stats->nr_bytes[BLOCK_ACCT_UNMAP];
+        ds->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
+        ds->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
+        ds->zone_append_operations = stats->nr_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_operations = stats->nr_ops[BLOCK_ACCT_UNMAP];
+
+        ds->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
+        ds->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
+        ds->failed_zone_append_operations =
+            stats->failed_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
+        ds->failed_unmap_operations = stats->failed_ops[BLOCK_ACCT_UNMAP];
+
+        ds->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
+        ds->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
+        ds->invalid_zone_append_operations =
+            stats->invalid_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->invalid_flush_operations =
+            stats->invalid_ops[BLOCK_ACCT_FLUSH];
+        ds->invalid_unmap_operations = stats->invalid_ops[BLOCK_ACCT_UNMAP];
+
+        ds->rd_merged = stats->merged[BLOCK_ACCT_READ];
+        ds->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
+        ds->zone_append_merged = stats->merged[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_merged = stats->merged[BLOCK_ACCT_UNMAP];
+        ds->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];
+        ds->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
+        ds->zone_append_total_time_ns =
+            stats->total_time_ns[BLOCK_ACCT_ZONE_APPEND];
+        ds->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
+        ds->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
+        ds->unmap_total_time_ns = stats->total_time_ns[BLOCK_ACCT_UNMAP];
+
+        ds->has_idle_time_ns = stats->last_access_time_ns > 0;
+        if (ds->has_idle_time_ns) {
+            ds->idle_time_ns = block_acct_idle_time_ns(stats);
+        }
 
-    ds->account_invalid = stats->account_invalid;
-    ds->account_failed = stats->account_failed;
+        ds->account_invalid = stats->account_invalid;
+        ds->account_failed = stats->account_failed;
+
+        hgram = stats->latency_histogram;
+        ds->rd_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_READ]);
+        ds->wr_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_WRITE]);
+        ds->zone_append_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_ZONE_APPEND]);
+        ds->flush_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_FLUSH]);
+    }
 
     while ((ts = block_acct_interval_next(stats, ts))) {
         BlockDeviceTimedStats *dev_stats = g_malloc0(sizeof(*dev_stats));
Signature · 1 line
2.53.0
#0 lore raw

cover block/accounting: fix unlocked latency histogram reconfiguration races

block_latency_histogram_set() and block_latency_histograms_clear() replace BlockLatencyHistogram's nbins/boundaries/bins without taking stats->lock, while block_account_one_io() reads those same fields under that lock from whatever iothread completes the I/O. A histogram reconfiguration (block-latency-histogram-set QMP command, monitor thread) racing an in-flight completion can therefore observe those fields torn, hitting assert(pos != NULL) in block_latency_histogram_account(), or corrupting the heap outright. This showed up as a qemu-kvm SIGABRT on a customer's virtio-blk guest.

Regression test is added for illustrative purpose but I am unsure that it is viable long term. Feel free to drop.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>

Denis V. Lunev (3):
block/accounting: take stats->lock in latency histogram setters block/qapi: take stats->lock when reading BlockAcctStats for
query-blockstats
tests/unit: add reproducer for BlockAcctStats histogram locking race

 block/accounting.c                 |  34 +++++----
 block/qapi.c                       | 102 +++++++++++++++-----------
 tests/unit/meson.build             |   1 +
 tests/unit/test-block-accounting.c | 114 +++++++++++++++++++++++++++++
 4 files changed, 194 insertions(+), 57 deletions(-)
 create mode 100644 tests/unit/test-block-accounting.c
Signature · 1 line
2.53.0
#4 lore raw

PATCH 3/3 tests/unit: add reproducer for BlockAcctStats histogram locking race

block_latency_histogram_set() and block_latency_histograms_clear() replace BlockLatencyHistogram's nbins/boundaries/bins without taking stats->lock, while block_account_one_io() reads those same fields under that lock from whatever iothread completes the I/O.

Add a test that races two real threads against block_latency_histogram_set() and block_acct_start()/block_acct_done() on the same BlockAcctStats. Applied here it passes, since the previous two commits already take the lock; reverting them locally reproduces the abort this series fixes, in about a second.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>
---
 tests/unit/meson.build             |   1 +
 tests/unit/test-block-accounting.c | 114 +++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+)
 create mode 100644 tests/unit/test-block-accounting.c
tests/unit/meson.build +1 -0
index 5ba6b1a230..dc3fb954c0 100644
@@ -75,6 +75,7 @@ if have_block
     'test-blockjob': [testblock],
     'test-blockjob-txn': [testblock],
     'test-block-backend': [testblock],
+    'test-block-accounting': [testblock],
     'test-block-iothread': [testblock],
     'test-write-threshold': [testblock],
     'test-crypto-hash': [crypto],
tests/unit/test-block-accounting.c new file +114 -0

114 lines not on this page · raw message

Signature · 1 line
2.53.0
#1 lore raw

PATCH 1/3 block/accounting: take stats->lock in latency histogram setters

In reply to Denis V. Lunev

block_latency_histogram_set() and block_latency_histograms_clear() replace BlockLatencyHistogram's nbins/boundaries/bins without taking stats->lock, while block_account_one_io() reads those same fields under that lock from whatever iothread completes the I/O. The result is usual use-after-free and qemu crash.

Take stats->lock in both setters, matching the lock already held by the reader.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>
---
 block/accounting.c | 34 +++++++++++++++++++---------------
 1 file changed, 19 insertions(+), 15 deletions(-)
block/accounting.c +19 -15
index f00fe99740..c277b796eb 100644
@@ -194,17 +194,19 @@ int block_latency_histogram_set(BlockAcctStats *stats, enum BlockAcctType type,
         return -EINVAL;
     }
 
-    hist->nbins = new_nbins;
-    g_free(hist->boundaries);
-    hist->boundaries = g_new(uint64_t, hist->nbins - 1);
-    for (entry = boundaries, ptr = hist->boundaries; entry;
-         entry = entry->next, ptr++)
-    {
-        *ptr = entry->value;
-    }
+    WITH_QEMU_LOCK_GUARD(&stats->lock) {
+        hist->nbins = new_nbins;
+        g_free(hist->boundaries);
+        hist->boundaries = g_new(uint64_t, hist->nbins - 1);
+        for (entry = boundaries, ptr = hist->boundaries; entry;
+             entry = entry->next, ptr++)
+        {
+            *ptr = entry->value;
+        }
 
-    g_free(hist->bins);
-    hist->bins = g_new0(uint64_t, hist->nbins);
+        g_free(hist->bins);
+        hist->bins = g_new0(uint64_t, hist->nbins);
+    }
 
     return 0;
 }
@@ -213,11 +215,13 @@ void block_latency_histograms_clear(BlockAcctStats *stats)
 {
     int i;
 
-    for (i = 0; i < BLOCK_MAX_IOTYPE; i++) {
-        BlockLatencyHistogram *hist = &stats->latency_histogram[i];
-        g_free(hist->bins);
-        g_free(hist->boundaries);
-        memset(hist, 0, sizeof(*hist));
+    WITH_QEMU_LOCK_GUARD(&stats->lock) {
+        for (i = 0; i < BLOCK_MAX_IOTYPE; i++) {
+            BlockLatencyHistogram *hist = &stats->latency_histogram[i];
+            g_free(hist->bins);
+            g_free(hist->boundaries);
+            memset(hist, 0, sizeof(*hist));
+        }
     }
 }
 
Signature · 1 line
2.53.0
#3 lore raw

Re: [PATCH 2/3] block/qapi: take stats->lock when reading BlockAcctStats for query-blockstats

In reply to Denis V. Lunev
Quoted 124 lines · Denis V. Lunev

bdrv_query_blk_stats() reads nr_bytes[], nr_ops[], failed_ops[], invalid_ops[], merged[], total_time_ns[], last_access_time_ns, and the latency histogram bins directly off BlockAcctStats, without taking stats->lock. block_account_one_io() updates the same fields under that lock from whatever iothread completes the I/O, so this is an unsynchronized read of concurrently-updated state.

This is a narrower issue than the one fixed for the histogram setters. query-blockstats and block-latency-histogram-set are both plain (non-allow-oob) QMP commands, so they always run serialized on the monitor thread and can never execute concurrently with each other.

Take stats->lock around the read for consistency.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>
---
 block/qapi.c | 102 ++++++++++++++++++++++++++++++---------------------
 1 file changed, 60 insertions(+), 42 deletions(-)
block/qapi.c +50 -42
index eabfbfc258..c3f20832b7 100644
@@ -535,49 +535,67 @@ static void bdrv_query_blk_stats(BlockDeviceStats *ds, BlockBackend *blk)
     BlockAcctTimedStats *ts = NULL;
     BlockLatencyHistogram *hgram;
 
-    ds->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
-    ds->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
-    ds->zone_append_bytes = stats->nr_bytes[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_bytes = stats->nr_bytes[BLOCK_ACCT_UNMAP];
-    ds->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
-    ds->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
-    ds->zone_append_operations = stats->nr_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_operations = stats->nr_ops[BLOCK_ACCT_UNMAP];
-
-    ds->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
-    ds->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
-    ds->failed_zone_append_operations =
-        stats->failed_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
-    ds->failed_unmap_operations = stats->failed_ops[BLOCK_ACCT_UNMAP];
-
-    ds->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
-    ds->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
-    ds->invalid_zone_append_operations =
-        stats->invalid_ops[BLOCK_ACCT_ZONE_APPEND];
-    ds->invalid_flush_operations =
-        stats->invalid_ops[BLOCK_ACCT_FLUSH];
-    ds->invalid_unmap_operations = stats->invalid_ops[BLOCK_ACCT_UNMAP];
-
-    ds->rd_merged = stats->merged[BLOCK_ACCT_READ];
-    ds->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
-    ds->zone_append_merged = stats->merged[BLOCK_ACCT_ZONE_APPEND];
-    ds->unmap_merged = stats->merged[BLOCK_ACCT_UNMAP];
-    ds->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];
-    ds->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
-    ds->zone_append_total_time_ns =
-        stats->total_time_ns[BLOCK_ACCT_ZONE_APPEND];
-    ds->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
-    ds->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
-    ds->unmap_total_time_ns = stats->total_time_ns[BLOCK_ACCT_UNMAP];
-
-    ds->has_idle_time_ns = stats->last_access_time_ns > 0;
-    if (ds->has_idle_time_ns) {
-        ds->idle_time_ns = block_acct_idle_time_ns(stats);
-    }
+    /*
+     * nr_bytes[] etc. are mutated under stats->lock by block_account_one_io()
+     * from an iothread, and latency_histogram[].{nbins,boundaries,bins} can
+     * be freed and reallocated by a concurrent block_latency_histogram_set()
+     * on the monitor thread, so every read here needs the same lock.
+     */
+    WITH_QEMU_LOCK_GUARD(&stats->lock) {
+        ds->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
+        ds->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
+        ds->zone_append_bytes = stats->nr_bytes[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_bytes = stats->nr_bytes[BLOCK_ACCT_UNMAP];
+        ds->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
+        ds->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
+        ds->zone_append_operations = stats->nr_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_operations = stats->nr_ops[BLOCK_ACCT_UNMAP];
+
+        ds->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
+        ds->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
+        ds->failed_zone_append_operations =
+            stats->failed_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
+        ds->failed_unmap_operations = stats->failed_ops[BLOCK_ACCT_UNMAP];
+
+        ds->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
+        ds->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
+        ds->invalid_zone_append_operations =
+            stats->invalid_ops[BLOCK_ACCT_ZONE_APPEND];
+        ds->invalid_flush_operations =
+            stats->invalid_ops[BLOCK_ACCT_FLUSH];
+        ds->invalid_unmap_operations = stats->invalid_ops[BLOCK_ACCT_UNMAP];
+
+        ds->rd_merged = stats->merged[BLOCK_ACCT_READ];
+        ds->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
+        ds->zone_append_merged = stats->merged[BLOCK_ACCT_ZONE_APPEND];
+        ds->unmap_merged = stats->merged[BLOCK_ACCT_UNMAP];
+        ds->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];
+        ds->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
+        ds->zone_append_total_time_ns =
+            stats->total_time_ns[BLOCK_ACCT_ZONE_APPEND];
+        ds->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
+        ds->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
+        ds->unmap_total_time_ns = stats->total_time_ns[BLOCK_ACCT_UNMAP];
+
+        ds->has_idle_time_ns = stats->last_access_time_ns > 0;
+        if (ds->has_idle_time_ns) {
+            ds->idle_time_ns = block_acct_idle_time_ns(stats);
+        }
 
-    ds->account_invalid = stats->account_invalid;
-    ds->account_failed = stats->account_failed;
+        ds->account_invalid = stats->account_invalid;
+        ds->account_failed = stats->account_failed;
+

block/qapi.c static void bdrv_query_blk_stats(BlockDeviceStats *ds, BlockBackend *blk) · line 598

Quoted 10 lines
block/qapi.c +10 -0
+        hgram = stats->latency_histogram;
+        ds->rd_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_READ]);
+        ds->wr_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_WRITE]);
+        ds->zone_append_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_ZONE_APPEND]);
+        ds->flush_latency_histogram
+            = bdrv_latency_histogram_stats(&hgram[BLOCK_ACCT_FLUSH]);
+    }

Smth is off with this hunk. These calls to bdrv_latency_histogram_stats() pre-exist below (after the while loop). If they need to be protected by the lock as well, we should remove those duplicating calls below.

block/qapi.c static void bdrv_query_blk_stats(BlockDeviceStats *ds, BlockBackend *blk) · line 601

Quoted 3 lines
block/qapi.c +0 -0
     while ((ts = block_acct_interval_next(stats, ts))) {
         BlockDeviceTimedStats *dev_stats = g_malloc0(sizeof(*dev_stats));

Also in the while loop we call timed_average_{min,max,avg} with no locks, whereas block_account_one_io() does update the intervals under the lock. Shouldn't we move the while loop itself under the lock guard?

Andrey

#5 lore raw

Re: [PATCH 3/3] tests/unit: add reproducer for BlockAcctStats histogram locking race

In reply to Denis V. Lunev
Quoted 82 lines · Denis V. Lunev

block_latency_histogram_set() and block_latency_histograms_clear() replace BlockLatencyHistogram's nbins/boundaries/bins without taking stats->lock, while block_account_one_io() reads those same fields under that lock from whatever iothread completes the I/O.

Add a test that races two real threads against block_latency_histogram_set() and block_acct_start()/block_acct_done() on the same BlockAcctStats. Applied here it passes, since the previous two commits already take the lock; reverting them locally reproduces the abort this series fixes, in about a second.

CC Kevin Wolf <[email protected]>
CC Hanna Reitz <[email protected]>
CC Vladimir Sementsov-Ogievskiy <[email protected]>
---
 tests/unit/meson.build             |   1 +
 tests/unit/test-block-accounting.c | 114 +++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+)
 create mode 100644 tests/unit/test-block-accounting.c
tests/unit/meson.build +1 -0
index 5ba6b1a230..dc3fb954c0 100644
@@ -75,6 +75,7 @@ if have_block
     'test-blockjob': [testblock],
     'test-blockjob-txn': [testblock],
     'test-block-backend': [testblock],
+    'test-block-accounting': [testblock],
     'test-block-iothread': [testblock],
     'test-write-threshold': [testblock],
     'test-crypto-hash': [crypto],
tests/unit/test-block-accounting.c new file +42 -0
new file mode 100644
index 0000000000..e749b73749
@@ -0,0 +1,114 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ *
+ * BlockAcctStats latency histogram locking regression test
+ *
+ * Copyright (c) 2026 Virtuozzo International GmbH.
+ *
+ * Regression test for missing stats->lock in
+ * block_latency_histogram_set()/block_latency_histograms_clear(),
+ * racing block_account_one_io() reading the same fields from an
+ * iothread. Aborts reliably before the fix, passes after it.
+ */
+
+#include "qemu/osdep.h"
+#include "block/block.h"
+#include "block/accounting.h"
+#include "system/block-backend.h"
+#include "system/block-backend-io.h"
+#include "qapi/error.h"
+#include "qemu/main-loop.h"
+#include "qemu/thread.h"
+
+#define RACE_DURATION_MS 2000
+#define NUM_READER_THREADS 8
+
+static bool stop_workers;
+
+/*
+ * Different bin counts, so the writer's g_free()/g_new() churn can be
+ * caught mid-update. Values are small enough (nanoseconds) that plain
+ * back-to-back start/done calls exercise every bin without sleeping.
+ */
+static uint64List boundaries_a[] = {
+    { .next = &boundaries_a[1], .value = 1000 },
+    { .next = &boundaries_a[2], .value = 5000 },
+    { .next = NULL,             .value = 50000 },
+};
+
+static uint64List boundaries_b[] = {
+    { .next = &boundaries_b[1], .value = 800 },
+    { .next = &boundaries_b[2], .value = 3000 },
+    { .next = &boundaries_b[3], .value = 20000 },

tests/unit/test-block-accounting.c line 52

Quoted 10 lines
tests/unit/test-block-accounting.c new file +10 -0
+    { .next = NULL,             .value = 200000 },
+};
+
+static void *writer_thread(void *opaque)
+{
+    BlockAcctStats *stats = opaque;
+
+    while (!qatomic_read(&stop_workers)) {
+        block_latency_histogram_set(stats, BLOCK_ACCT_READ, boundaries_a);
+        block_latency_histogram_set(stats, BLOCK_ACCT_READ, boundaries_b);

We fixed block_latency_histograms_clear() in the 1st patch but don't cover it here in the test. Should we add a call to block_latency_histograms_clear(stats) here in the end of while loop to get it covered?

Andrey

Quoted 62 lines
tests/unit/test-block-accounting.c new file +62 -0
+    }
+
+    return NULL;
+}
+
+static void *reader_thread(void *opaque)
+{
+    BlockAcctStats *stats = opaque;
+
+    while (!qatomic_read(&stop_workers)) {
+        BlockAcctCookie cookie;
+
+        block_acct_start(stats, &cookie, 4096, BLOCK_ACCT_READ);
+        block_acct_done(stats, &cookie);
+    }
+
+    return NULL;
+}
+
+static void test_latency_histogram_race(void)
+{
+    BlockBackend *blk = blk_new(qemu_get_aio_context(),
+                                BLK_PERM_ALL, BLK_PERM_ALL);
+    BlockAcctStats *stats = blk_get_stats(blk);
+    QemuThread writer, readers[NUM_READER_THREADS];
+    int i;
+
+    /* Histogram has to be enabled (bins != NULL) before racing it. */
+    g_assert(block_latency_histogram_set(stats, BLOCK_ACCT_READ,
+                                         boundaries_a) == 0);
+
+    stop_workers = false;
+    qemu_thread_create(&writer, "hist-writer", writer_thread, stats,
+                       QEMU_THREAD_JOINABLE);
+    for (i = 0; i < NUM_READER_THREADS; i++) {
+        qemu_thread_create(&readers[i], "hist-reader", reader_thread, stats,
+                           QEMU_THREAD_JOINABLE);
+    }
+
+    g_usleep(RACE_DURATION_MS * 1000);
+    qatomic_set(&stop_workers, true);
+
+    qemu_thread_join(&writer);
+    for (i = 0; i < NUM_READER_THREADS; i++) {
+        qemu_thread_join(&readers[i]);
+    }
+
+    blk_unref(blk);
+}
+
+int main(int argc, char **argv)
+{
+    bdrv_init();
+    qemu_init_main_loop(&error_abort);
+
+    g_test_init(&argc, &argv, NULL);
+
+    g_test_add_func("/block-accounting/latency_histogram_race",
+                    test_latency_histogram_race);
+
+    return g_test_run();
+}