diff --git a/block/blkverify.c b/block/blkverify.c
index d1facf5..53ba1c9 100644
--- a/block/blkverify.c
+++ b/block/blkverify.c
@@ -16,6 +16,7 @@
 #include "qemu/cutils.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
+#include "qemu/memalign.h"
 
 typedef struct {
     BdrvChild *test_file;
diff --git a/block/block-copy.c b/block/block-copy.c
index ce11631..619e558 100644
--- a/block/block-copy.c
+++ b/block/block-copy.c
@@ -22,6 +22,7 @@
 #include "qemu/coroutine.h"
 #include "block/aio_task.h"
 #include "qemu/error-report.h"
+#include "qemu/memalign.h"
 
 #define BLOCK_COPY_MAX_COPY_RANGE (16 * MiB)
 #define BLOCK_COPY_MAX_BUFFER (1 * MiB)
diff --git a/block/commit.c b/block/commit.c
index c76899f..851d1c5 100644
--- a/block/commit.c
+++ b/block/commit.c
@@ -20,6 +20,7 @@
 #include "qapi/error.h"
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
+#include "qemu/memalign.h"
 #include "sysemu/block-backend.h"
 
 enum {
diff --git a/block/crypto.c b/block/crypto.c
index 9d5fecb..1ba8298 100644
--- a/block/crypto.c
+++ b/block/crypto.c
@@ -31,6 +31,7 @@
 #include "qemu/module.h"
 #include "qemu/option.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include "crypto.h"
 
 typedef struct BlockCrypto BlockCrypto;
diff --git a/block/dmg.c b/block/dmg.c
index 447901f..c626587 100644
--- a/block/dmg.c
+++ b/block/dmg.c
@@ -27,6 +27,7 @@
 #include "qemu/bswap.h"
 #include "qemu/error-report.h"
 #include "qemu/module.h"
+#include "qemu/memalign.h"
 #include "dmg.h"
 
 int (*dmg_uncompress_bz2)(char *next_in, unsigned int avail_in,
diff --git a/block/export/fuse.c b/block/export/fuse.c
index 5029e70..e80b24a 100644
--- a/block/export/fuse.c
+++ b/block/export/fuse.c
@@ -19,6 +19,7 @@
 #define FUSE_USE_VERSION 31
 
 #include "qemu/osdep.h"
+#include "qemu/memalign.h"
 #include "block/aio.h"
 #include "block/block.h"
 #include "block/export.h"
diff --git a/block/file-posix.c b/block/file-posix.c
index 1f1756e..c000a61 100644
--- a/block/file-posix.c
+++ b/block/file-posix.c
@@ -31,6 +31,7 @@
 #include "qemu/module.h"
 #include "qemu/option.h"
 #include "qemu/units.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 #include "block/thread-pool.h"
 #include "qemu/iov.h"
diff --git a/block/io.c b/block/io.c
index efc011c..8e621a4 100644
--- a/block/io.c
+++ b/block/io.c
@@ -32,6 +32,7 @@
 #include "block/coroutines.h"
 #include "block/write-threshold.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include "qapi/error.h"
 #include "qemu/error-report.h"
 #include "qemu/main-loop.h"
diff --git a/block/mirror.c b/block/mirror.c
index ce6bc58..d8ecb9e 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -23,6 +23,7 @@
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
 #include "qemu/bitmap.h"
+#include "qemu/memalign.h"
 
 #define MAX_IN_FLIGHT 16
 #define MAX_IO_BYTES (1 << 20) /* 1 Mb */
diff --git a/block/nvme.c b/block/nvme.c
index dd20de3..5520299 100644
--- a/block/nvme.c
+++ b/block/nvme.c
@@ -21,6 +21,7 @@
 #include "qemu/module.h"
 #include "qemu/cutils.h"
 #include "qemu/option.h"
+#include "qemu/memalign.h"
 #include "qemu/vfio-helpers.h"
 #include "block/block_int.h"
 #include "sysemu/replay.h"
diff --git a/block/parallels-ext.c b/block/parallels-ext.c
index e0dd097..cb22a42 100644
--- a/block/parallels-ext.c
+++ b/block/parallels-ext.c
@@ -29,6 +29,7 @@
 #include "parallels.h"
 #include "crypto/hash.h"
 #include "qemu/uuid.h"
+#include "qemu/memalign.h"
 
 #define PARALLELS_FORMAT_EXTENSION_MAGIC 0xAB234CEF23DCEA87ULL
 
diff --git a/block/parallels.c b/block/parallels.c
index e58c828..cd23e02 100644
--- a/block/parallels.c
+++ b/block/parallels.c
@@ -41,6 +41,7 @@
 #include "qapi/qapi-visit-block-core.h"
 #include "qemu/bswap.h"
 #include "qemu/bitmap.h"
+#include "qemu/memalign.h"
 #include "migration/blocker.h"
 #include "parallels.h"
 
diff --git a/block/qcow.c b/block/qcow.c
index c39940f..4fba1b9 100644
--- a/block/qcow.c
+++ b/block/qcow.c
@@ -32,6 +32,7 @@
 #include "qemu/option.h"
 #include "qemu/bswap.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include <zlib.h>
 #include "qapi/qmp/qdict.h"
 #include "qapi/qmp/qstring.h"
diff --git a/block/qcow2-cache.c b/block/qcow2-cache.c
index 7444b9c..8a01059 100644
--- a/block/qcow2-cache.c
+++ b/block/qcow2-cache.c
@@ -23,6 +23,7 @@
  */
 
 #include "qemu/osdep.h"
+#include "qemu/memalign.h"
 #include "qcow2.h"
 #include "trace.h"
 
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index 21884a1..20a16ba 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -28,6 +28,7 @@
 #include "qapi/error.h"
 #include "qcow2.h"
 #include "qemu/bswap.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 
 int qcow2_shrink_l1_table(BlockDriverState *bs, uint64_t exact_size)
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 4614572..9403397 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -28,6 +28,7 @@
 #include "qemu/range.h"
 #include "qemu/bswap.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 
 static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size,
diff --git a/block/qcow2-snapshot.c b/block/qcow2-snapshot.c
index 71ddb08..075269a 100644
--- a/block/qcow2-snapshot.c
+++ b/block/qcow2-snapshot.c
@@ -29,6 +29,7 @@
 #include "qemu/bswap.h"
 #include "qemu/error-report.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 
 static void qcow2_free_single_snapshot(BlockDriverState *bs, int i)
 {
diff --git a/block/qcow2.c b/block/qcow2.c
index c8115e1..b5c4793 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -38,6 +38,7 @@
 #include "qemu/option_int.h"
 #include "qemu/cutils.h"
 #include "qemu/bswap.h"
+#include "qemu/memalign.h"
 #include "qapi/qobject-input-visitor.h"
 #include "qapi/qapi-visit-block-core.h"
 #include "crypto.h"
diff --git a/block/qed-l2-cache.c b/block/qed-l2-cache.c
index b548362..caf2c02 100644
--- a/block/qed-l2-cache.c
+++ b/block/qed-l2-cache.c
@@ -51,6 +51,7 @@
  */
 
 #include "qemu/osdep.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 #include "qed.h"
 
diff --git a/block/qed-table.c b/block/qed-table.c
index 405d446..1cc844b 100644
--- a/block/qed-table.c
+++ b/block/qed-table.c
@@ -17,6 +17,7 @@
 #include "qemu/sockets.h" /* for EINPROGRESS on Windows */
 #include "qed.h"
 #include "qemu/bswap.h"
+#include "qemu/memalign.h"
 
 /* Called with table_lock held.  */
 static int coroutine_fn qed_read_table(BDRVQEDState *s, uint64_t offset,
diff --git a/block/qed.c b/block/qed.c
index 558d364..f34d9a3 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -20,6 +20,7 @@
 #include "qemu/main-loop.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 #include "qed.h"
 #include "sysemu/block-backend.h"
diff --git a/block/quorum.c b/block/quorum.c
index c28dda7..f33f30d 100644
--- a/block/quorum.c
+++ b/block/quorum.c
@@ -17,6 +17,7 @@
 #include "qemu/cutils.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
+#include "qemu/memalign.h"
 #include "block/block_int.h"
 #include "block/coroutines.h"
 #include "block/qdict.h"
diff --git a/block/raw-format.c b/block/raw-format.c
index bda757f..69fd650 100644
--- a/block/raw-format.c
+++ b/block/raw-format.c
@@ -31,6 +31,7 @@
 #include "qapi/error.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
+#include "qemu/memalign.h"
 
 typedef struct BDRVRawState {
     uint64_t offset;
diff --git a/block/vdi.c b/block/vdi.c
index bdc58d7..cca3a3a 100644
--- a/block/vdi.c
+++ b/block/vdi.c
@@ -64,6 +64,7 @@
 #include "qemu/coroutine.h"
 #include "qemu/cutils.h"
 #include "qemu/uuid.h"
+#include "qemu/memalign.h"
 
 /* Code configuration options. */
 
diff --git a/block/vhdx-log.c b/block/vhdx-log.c
index 7672161..ff0d4e0 100644
--- a/block/vhdx-log.c
+++ b/block/vhdx-log.c
@@ -23,6 +23,7 @@
 #include "block/block_int.h"
 #include "qemu/error-report.h"
 #include "qemu/bswap.h"
+#include "qemu/memalign.h"
 #include "vhdx.h"
 
 
diff --git a/block/vhdx.c b/block/vhdx.c
index 356ec4c..410c6f9 100644
--- a/block/vhdx.c
+++ b/block/vhdx.c
@@ -25,6 +25,7 @@
 #include "qemu/crc32c.h"
 #include "qemu/bswap.h"
 #include "qemu/error-report.h"
+#include "qemu/memalign.h"
 #include "vhdx.h"
 #include "migration/blocker.h"
 #include "qemu/uuid.h"
diff --git a/block/vmdk.c b/block/vmdk.c
index 0dfab6e..37c0946 100644
--- a/block/vmdk.c
+++ b/block/vmdk.c
@@ -33,6 +33,7 @@
 #include "qemu/module.h"
 #include "qemu/option.h"
 #include "qemu/bswap.h"
+#include "qemu/memalign.h"
 #include "migration/blocker.h"
 #include "qemu/cutils.h"
 #include <zlib.h>
diff --git a/block/vpc.c b/block/vpc.c
index 297a262..4d8f16e 100644
--- a/block/vpc.c
+++ b/block/vpc.c
@@ -33,6 +33,7 @@
 #include "migration/blocker.h"
 #include "qemu/bswap.h"
 #include "qemu/uuid.h"
+#include "qemu/memalign.h"
 #include "qapi/qmp/qdict.h"
 #include "qapi/qobject-input-visitor.h"
 #include "qapi/qapi-visit-block-core.h"
diff --git a/block/win32-aio.c b/block/win32-aio.c
index c57e10c..aadc7b1 100644
--- a/block/win32-aio.c
+++ b/block/win32-aio.c
@@ -29,6 +29,7 @@
 #include "block/raw-aio.h"
 #include "qemu/event_notifier.h"
 #include "qemu/iov.h"
+#include "qemu/memalign.h"
 #include <windows.h>
 #include <winioctl.h>
 
diff --git a/hw/block/dataplane/xen-block.c b/hw/block/dataplane/xen-block.c
index 8607875..2785b9e 100644
--- a/hw/block/dataplane/xen-block.c
+++ b/hw/block/dataplane/xen-block.c
@@ -21,6 +21,7 @@
 #include "qemu/osdep.h"
 #include "qemu/error-report.h"
 #include "qemu/main-loop.h"
+#include "qemu/memalign.h"
 #include "qapi/error.h"
 #include "hw/xen/xen_common.h"
 #include "hw/block/xen_blkif.h"
diff --git a/hw/block/fdc.c b/hw/block/fdc.c
index 21d18ac..347875a 100644
--- a/hw/block/fdc.c
+++ b/hw/block/fdc.c
@@ -32,6 +32,7 @@
 #include "qapi/error.h"
 #include "qemu/error-report.h"
 #include "qemu/timer.h"
+#include "qemu/memalign.h"
 #include "hw/irq.h"
 #include "hw/isa/isa.h"
 #include "hw/qdev-properties.h"
diff --git a/hw/ide/core.c b/hw/ide/core.c
index 33463d9..a7ac4de 100644
--- a/hw/ide/core.c
+++ b/hw/ide/core.c
@@ -30,6 +30,7 @@
 #include "qemu/main-loop.h"
 #include "qemu/timer.h"
 #include "qemu/hw-version.h"
+#include "qemu/memalign.h"
 #include "sysemu/sysemu.h"
 #include "sysemu/blockdev.h"
 #include "sysemu/dma.h"
diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
index 4cc204f..953fc65 100644
--- a/hw/ppc/spapr.c
+++ b/hw/ppc/spapr.c
@@ -27,6 +27,7 @@
 #include "qemu/osdep.h"
 #include "qemu-common.h"
 #include "qemu/datadir.h"
+#include "qemu/memalign.h"
 #include "qapi/error.h"
 #include "qapi/qapi-events-machine.h"
 #include "qapi/qapi-events-qdev.h"
diff --git a/hw/ppc/spapr_softmmu.c b/hw/ppc/spapr_softmmu.c
index 4ee03c8..5170a33 100644
--- a/hw/ppc/spapr_softmmu.c
+++ b/hw/ppc/spapr_softmmu.c
@@ -1,5 +1,6 @@
 #include "qemu/osdep.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include "cpu.h"
 #include "helper_regs.h"
 #include "hw/ppc/spapr.h"
diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
index 3666b8d..072686e 100644
--- a/hw/scsi/scsi-disk.c
+++ b/hw/scsi/scsi-disk.c
@@ -26,6 +26,7 @@
 #include "qemu/main-loop.h"
 #include "qemu/module.h"
 #include "qemu/hw-version.h"
+#include "qemu/memalign.h"
 #include "hw/scsi/scsi.h"
 #include "migration/qemu-file-types.h"
 #include "migration/vmstate.h"
diff --git a/hw/tpm/tpm_ppi.c b/hw/tpm/tpm_ppi.c
index 6dbb9f4..c89ac53 100644
--- a/hw/tpm/tpm_ppi.c
+++ b/hw/tpm/tpm_ppi.c
@@ -12,7 +12,7 @@
  */
 
 #include "qemu/osdep.h"
-
+#include "qemu/memalign.h"
 #include "qapi/error.h"
 #include "sysemu/memory_mapping.h"
 #include "migration/vmstate.h"
diff --git a/include/qemu/memalign.h b/include/qemu/memalign.h
new file mode 100644
index 0000000..fa299f3
--- /dev/null
+++ b/include/qemu/memalign.h
@@ -0,0 +1,61 @@
+/*
+ * Allocation and free functions for aligned memory
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_MEMALIGN_H
+#define QEMU_MEMALIGN_H
+
+/**
+ * qemu_try_memalign: Allocate aligned memory
+ * @alignment: required alignment, in bytes
+ * @size: size of allocation, in bytes
+ *
+ * Allocate memory on an aligned boundary (i.e. the returned
+ * address will be an exact multiple of @alignment).
+ * @alignment must be a power of 2, or the function will assert().
+ * On success, returns allocated memory; on failure, returns NULL.
+ *
+ * The memory allocated through this function must be freed via
+ * qemu_vfree() (and not via free()).
+ */
+void *qemu_try_memalign(size_t alignment, size_t size);
+/**
+ * qemu_memalign: Allocate aligned memory, without failing
+ * @alignment: required alignment, in bytes
+ * @size: size of allocation, in bytes
+ *
+ * Allocate memory in the same way as qemu_try_memalign(), but
+ * abort() with an error message if the memory allocation fails.
+ *
+ * The memory allocated through this function must be freed via
+ * qemu_vfree() (and not via free()).
+ */
+void *qemu_memalign(size_t alignment, size_t size);
+/**
+ * qemu_vfree: Free memory allocated through qemu_memalign
+ * @ptr: memory to free
+ *
+ * This function must be used to free memory allocated via qemu_memalign()
+ * or qemu_try_memalign(). (Using the wrong free function will cause
+ * subtle bugs on Windows hosts.)
+ */
+void qemu_vfree(void *ptr);
+/*
+ * It's an analog of GLIB's g_autoptr_cleanup_generic_gfree(), used to define
+ * g_autofree macro.
+ */
+static inline void qemu_cleanup_generic_vfree(void *p)
+{
+  void **pp = (void **)p;
+  qemu_vfree(*pp);
+}
+
+/*
+ * Analog of g_autofree, but qemu_vfree is called on cleanup instead of g_free.
+ */
+#define QEMU_AUTO_VFREE __attribute__((cleanup(qemu_cleanup_generic_vfree)))
+
+#endif
diff --git a/include/qemu/osdep.h b/include/qemu/osdep.h
index 7bcce3b..bc3df26 100644
--- a/include/qemu/osdep.h
+++ b/include/qemu/osdep.h
@@ -379,28 +379,10 @@
 #endif
 
 int qemu_daemon(int nochdir, int noclose);
-void *qemu_try_memalign(size_t alignment, size_t size);
-void *qemu_memalign(size_t alignment, size_t size);
 void *qemu_anon_ram_alloc(size_t size, uint64_t *align, bool shared,
                           bool noreserve);
-void qemu_vfree(void *ptr);
 void qemu_anon_ram_free(void *ptr, size_t size);
 
-/*
- * It's an analog of GLIB's g_autoptr_cleanup_generic_gfree(), used to define
- * g_autofree macro.
- */
-static inline void qemu_cleanup_generic_vfree(void *p)
-{
-  void **pp = (void **)p;
-  qemu_vfree(*pp);
-}
-
-/*
- * Analog of g_autofree, but qemu_vfree is called on cleanup instead of g_free.
- */
-#define QEMU_AUTO_VFREE __attribute__((cleanup(qemu_cleanup_generic_vfree)))
-
 #ifdef _WIN32
 #define HAVE_CHARDEV_SERIAL 1
 #elif defined(__linux__) || defined(__sun__) || defined(__FreeBSD__)    \
diff --git a/nbd/server.c b/nbd/server.c
index 9fb2f26..53e68cf 100644
--- a/nbd/server.c
+++ b/nbd/server.c
@@ -25,6 +25,7 @@
 #include "trace.h"
 #include "nbd-internal.h"
 #include "qemu/units.h"
+#include "qemu/memalign.h"
 
 #define NBD_META_ID_BASE_ALLOCATION 0
 #define NBD_META_ID_ALLOCATION_DEPTH 1
diff --git a/net/l2tpv3.c b/net/l2tpv3.c
index e4d4218..b8faa87 100644
--- a/net/l2tpv3.c
+++ b/net/l2tpv3.c
@@ -34,7 +34,7 @@
 #include "qemu/sockets.h"
 #include "qemu/iov.h"
 #include "qemu/main-loop.h"
-
+#include "qemu/memalign.h"
 
 /* The buffer size needs to be investigated for optimum numbers and
  * optimum means of paging in on different systems. This size is
diff --git a/plugins/loader.c b/plugins/loader.c
index 4883b0a..88c30bd 100644
--- a/plugins/loader.c
+++ b/plugins/loader.c
@@ -27,6 +27,7 @@
 #include "qemu/cacheinfo.h"
 #include "qemu/xxhash.h"
 #include "qemu/plugin.h"
+#include "qemu/memalign.h"
 #include "hw/core/cpu.h"
 #include "exec/exec-all.h"
 #ifndef CONFIG_USER_ONLY
diff --git a/qemu-img.c b/qemu-img.c
index 6fe2466..5dffb3e 100644
--- a/qemu-img.c
+++ b/qemu-img.c
@@ -42,6 +42,7 @@
 #include "qemu/module.h"
 #include "qemu/sockets.h"
 #include "qemu/units.h"
+#include "qemu/memalign.h"
 #include "qom/object_interfaces.h"
 #include "sysemu/block-backend.h"
 #include "block/block_int.h"
diff --git a/qemu-io-cmds.c b/qemu-io-cmds.c
index 46593d6..633b46c 100644
--- a/qemu-io-cmds.c
+++ b/qemu-io-cmds.c
@@ -21,6 +21,7 @@
 #include "qemu/option.h"
 #include "qemu/timer.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 
 #define CMD_NOFILE_OK   0x01
 
diff --git a/qom/object.c b/qom/object.c
index 9f7a331..4048a2f 100644
--- a/qom/object.c
+++ b/qom/object.c
@@ -16,6 +16,7 @@
 #include "qom/object.h"
 #include "qom/object_interfaces.h"
 #include "qemu/cutils.h"
+#include "qemu/memalign.h"
 #include "qapi/visitor.h"
 #include "qapi/string-input-visitor.h"
 #include "qapi/string-output-visitor.h"
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index a13289a..59dcf13 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -42,6 +42,7 @@
 #include "qemu/config-file.h"
 #include "qemu/error-report.h"
 #include "qemu/qemu-print.h"
+#include "qemu/memalign.h"
 #include "exec/memory.h"
 #include "exec/ioport.h"
 #include "sysemu/dma.h"
diff --git a/target/i386/hvf/hvf.c b/target/i386/hvf/hvf.c
index 4ba6e82..fc12c02 100644
--- a/target/i386/hvf/hvf.c
+++ b/target/i386/hvf/hvf.c
@@ -49,6 +49,7 @@
 #include "qemu/osdep.h"
 #include "qemu-common.h"
 #include "qemu/error-report.h"
+#include "qemu/memalign.h"
 
 #include "sysemu/hvf.h"
 #include "sysemu/hvf_int.h"
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 2c8feb4..83d0988 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -37,6 +37,7 @@
 #include "qemu/main-loop.h"
 #include "qemu/config-file.h"
 #include "qemu/error-report.h"
+#include "qemu/memalign.h"
 #include "hw/i386/x86.h"
 #include "hw/i386/apic.h"
 #include "hw/i386/apic_internal.h"
diff --git a/tcg/region.c b/tcg/region.c
index 72afb35..97ca529 100644
--- a/tcg/region.c
+++ b/tcg/region.c
@@ -26,6 +26,7 @@
 #include "qemu/units.h"
 #include "qemu/madvise.h"
 #include "qemu/mprotect.h"
+#include "qemu/memalign.h"
 #include "qemu/cacheinfo.h"
 #include "qapi/error.h"
 #include "exec/exec-all.h"
diff --git a/tests/bench/atomic_add-bench.c b/tests/bench/atomic_add-bench.c
index f05471a..8a6faad 100644
--- a/tests/bench/atomic_add-bench.c
+++ b/tests/bench/atomic_add-bench.c
@@ -2,6 +2,7 @@
 #include "qemu/thread.h"
 #include "qemu/host-utils.h"
 #include "qemu/processor.h"
+#include "qemu/memalign.h"
 
 struct thread_info {
     uint64_t r;
diff --git a/tests/bench/qht-bench.c b/tests/bench/qht-bench.c
index 2e5b70c..8afe161 100644
--- a/tests/bench/qht-bench.c
+++ b/tests/bench/qht-bench.c
@@ -10,6 +10,7 @@
 #include "qemu/qht.h"
 #include "qemu/rcu.h"
 #include "qemu/xxhash.h"
+#include "qemu/memalign.h"
 
 struct thread_stats {
     size_t rd;
diff --git a/util/atomic64.c b/util/atomic64.c
index 22983a9..c20d071 100644
--- a/util/atomic64.c
+++ b/util/atomic64.c
@@ -8,6 +8,7 @@
 #include "qemu/atomic.h"
 #include "qemu/thread.h"
 #include "qemu/cacheinfo.h"
+#include "qemu/memalign.h"
 
 #ifdef CONFIG_ATOMIC64
 #error This file must only be compiled if !CONFIG_ATOMIC64
diff --git a/util/memalign.c b/util/memalign.c
index 440b8fd..c199ae7 100644
--- a/util/memalign.c
+++ b/util/memalign.c
@@ -26,6 +26,7 @@
 
 #include "qemu/osdep.h"
 #include "qemu/host-utils.h"
+#include "qemu/memalign.h"
 #include "trace.h"
 
 void *qemu_try_memalign(size_t alignment, size_t size)
diff --git a/util/qht.c b/util/qht.c
index 0796051..065fc50 100644
--- a/util/qht.c
+++ b/util/qht.c
@@ -69,6 +69,7 @@
 #include "qemu/qht.h"
 #include "qemu/atomic.h"
 #include "qemu/rcu.h"
+#include "qemu/memalign.h"
 
 //#define QHT_DEBUG