#include "block/trace.h"
#include "block/block_int.h"
#include "block/blockjob.h"
+#include "block/dirty-bitmap.h"
#include "block/fuse.h"
#include "block/nbd.h"
#include "block/qdict.h"
Error *err;
} CreateCo;
-static void coroutine_fn bdrv_create_co_entry(void *opaque)
-{
- Error *local_err = NULL;
- int ret;
-
- CreateCo *cco = opaque;
- assert(cco->drv);
- GLOBAL_STATE_CODE();
-
- ret = cco->drv->bdrv_co_create_opts(cco->drv,
- cco->filename, cco->opts, &local_err);
- error_propagate(&cco->err, local_err);
- cco->ret = ret;
-}
-
-int bdrv_create(BlockDriver *drv, const char* filename,
- QemuOpts *opts, Error **errp)
+int coroutine_fn bdrv_co_create(BlockDriver *drv, const char *filename,
+ QemuOpts *opts, Error **errp)
{
int ret;
-
GLOBAL_STATE_CODE();
-
- Coroutine *co;
- CreateCo cco = {
- .drv = drv,
- .filename = g_strdup(filename),
- .opts = opts,
- .ret = NOT_DONE,
- .err = NULL,
- };
+ ERRP_GUARD();
if (!drv->bdrv_co_create_opts) {
- error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
- ret = -ENOTSUP;
- goto out;
+ error_setg(errp, "Driver '%s' does not support image creation",
+ drv->format_name);
+ return -ENOTSUP;
}
- if (qemu_in_coroutine()) {
- /* Fast-path if already in coroutine context */
- bdrv_create_co_entry(&cco);
- } else {
- co = qemu_coroutine_create(bdrv_create_co_entry, &cco);
- qemu_coroutine_enter(co);
- while (cco.ret == NOT_DONE) {
- aio_poll(qemu_get_aio_context(), true);
- }
+ ret = drv->bdrv_co_create_opts(drv, filename, opts, errp);
+ if (ret < 0 && !*errp) {
+ error_setg_errno(errp, -ret, "Could not create image");
}
- ret = cco.ret;
- if (ret < 0) {
- if (cco.err) {
- error_propagate(errp, cco.err);
- } else {
- error_setg_errno(errp, -ret, "Could not create image");
- }
- }
-
-out:
- g_free(cco.filename);
return ret;
}
return ret;
}
-int bdrv_create_file(const char *filename, QemuOpts *opts, Error **errp)
+int coroutine_fn bdrv_co_create_file(const char *filename, QemuOpts *opts,
+ Error **errp)
{
QemuOpts *protocol_opts;
BlockDriver *drv;
goto out;
}
- ret = bdrv_create(drv, filename, protocol_opts, errp);
+ ret = bdrv_co_create(drv, filename, protocol_opts, errp);
out:
qemu_opts_del(protocol_opts);
qobject_unref(qdict);
* Set the current 'total_sectors' value
* Return 0 on success, -errno on error.
*/
-int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
+int coroutine_fn bdrv_co_refresh_total_sectors(BlockDriverState *bs,
+ int64_t hint)
{
BlockDriver *drv = bs->drv;
IO_CODE();
return -ENOMEDIUM;
}
- /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
+ /* Do not attempt drv->bdrv_co_getlength() on scsi-generic devices */
if (bdrv_is_sg(bs))
return 0;
/* query actual device if possible, otherwise just trust the hint */
- if (drv->bdrv_getlength) {
- int64_t length = drv->bdrv_getlength(bs);
+ if (drv->bdrv_co_getlength) {
+ int64_t length = drv->bdrv_co_getlength(bs);
if (length < 0) {
return length;
}
static void bdrv_child_cb_drained_begin(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
- bdrv_do_drained_begin_quiesce(bs, NULL, false);
+ bdrv_do_drained_begin_quiesce(bs, NULL);
}
static bool bdrv_child_cb_drained_poll(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
- return bdrv_drain_poll(bs, false, NULL, false);
+ return bdrv_drain_poll(bs, NULL, false);
}
static void bdrv_child_cb_drained_end(BdrvChild *child)
*child_flags = flags;
}
-static void bdrv_child_cb_attach(BdrvChild *child)
+static void GRAPH_WRLOCK bdrv_child_cb_attach(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
- assert_bdrv_graph_writable(bs);
+ assert_bdrv_graph_writable();
QLIST_INSERT_HEAD(&bs->children, child, next);
if (bs->drv->is_filter || (child->role & BDRV_CHILD_FILTERED)) {
/*
assert(!bs->file);
bs->file = child;
}
-
- bdrv_apply_subtree_drain(child, bs);
}
-static void bdrv_child_cb_detach(BdrvChild *child)
+static void GRAPH_WRLOCK bdrv_child_cb_detach(BdrvChild *child)
{
BlockDriverState *bs = child->opaque;
bdrv_backing_detach(child);
}
- bdrv_unapply_subtree_drain(child, bs);
-
- assert_bdrv_graph_writable(bs);
+ assert_bdrv_graph_writable();
QLIST_REMOVE(child, next);
if (child == bs->backing) {
assert(child != bs->file);
g_free(gen_node_name);
}
+/*
+ * The caller must always hold @bs AioContext lock, because this function calls
+ * bdrv_refresh_total_sectors() which polls when called from non-coroutine
+ * context.
+ */
static int bdrv_open_driver(BlockDriverState *bs, BlockDriver *drv,
const char *node_name, QDict *options,
int open_flags, Error **errp)
bs->supported_read_flags |= BDRV_REQ_REGISTERED_BUF;
bs->supported_write_flags |= BDRV_REQ_REGISTERED_BUF;
- ret = refresh_total_sectors(bs, bs->total_sectors);
+ ret = bdrv_refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
error_setg_errno(errp, -ret, "Could not refresh total sector count");
return ret;
GLOBAL_STATE_CODE();
/* old_bs reference is transparently moved from @s to @s->child */
+ if (!s->child->bs) {
+ /*
+ * The parents were undrained when removing old_bs from the child. New
+ * requests can't have been made, though, because the child was empty.
+ *
+ * TODO Make bdrv_replace_child_noperm() transactionable to avoid
+ * undraining the parent in the first place. Once this is done, having
+ * new_bs drained when calling bdrv_replace_child_tran() is not a
+ * requirement any more.
+ */
+ bdrv_parent_drained_begin_single(s->child);
+ assert(!bdrv_parent_drained_poll_single(s->child));
+ }
+ assert(s->child->quiesced_parent);
bdrv_replace_child_noperm(s->child, s->old_bs);
bdrv_unref(new_bs);
}
*
* Note: real unref of old_bs is done only on commit.
*
+ * Both @child->bs and @new_bs (if non-NULL) must be drained. @new_bs must be
+ * kept drained until the transaction is completed.
+ *
* The function doesn't update permissions, caller is responsible for this.
*/
static void bdrv_replace_child_tran(BdrvChild *child, BlockDriverState *new_bs,
Transaction *tran)
{
BdrvReplaceChildState *s = g_new(BdrvReplaceChildState, 1);
+
+ assert(child->quiesced_parent);
+ assert(!new_bs || new_bs->quiesce_counter);
+
*s = (BdrvReplaceChildState) {
.child = child,
.old_bs = child->bs,
return permissions[qapi_perm];
}
+/*
+ * Replaces the node that a BdrvChild points to without updating permissions.
+ *
+ * If @new_bs is non-NULL, the parent of @child must already be drained through
+ * @child.
+ */
static void bdrv_replace_child_noperm(BdrvChild *child,
BlockDriverState *new_bs)
{
BlockDriverState *old_bs = child->bs;
int new_bs_quiesce_counter;
- int drain_saldo;
assert(!child->frozen);
+
+ /*
+ * If we want to change the BdrvChild to point to a drained node as its new
+ * child->bs, we need to make sure that its new parent is drained, too. In
+ * other words, either child->quiesce_parent must already be true or we must
+ * be able to set it and keep the parent's quiesce_counter consistent with
+ * that, but without polling or starting new requests (this function
+ * guarantees that it doesn't poll, and starting new requests would be
+ * against the invariants of drain sections).
+ *
+ * To keep things simple, we pick the first option (child->quiesce_parent
+ * must already be true). We also generalise the rule a bit to make it
+ * easier to verify in callers and more likely to be covered in test cases:
+ * The parent must be quiesced through this child even if new_bs isn't
+ * currently drained.
+ *
+ * The only exception is for callers that always pass new_bs == NULL. In
+ * this case, we obviously never need to consider the case of a drained
+ * new_bs, so we can keep the callers simpler by allowing them not to drain
+ * the parent.
+ */
+ assert(!new_bs || child->quiesced_parent);
assert(old_bs != new_bs);
GLOBAL_STATE_CODE();
assert(bdrv_get_aio_context(old_bs) == bdrv_get_aio_context(new_bs));
}
- new_bs_quiesce_counter = (new_bs ? new_bs->quiesce_counter : 0);
- drain_saldo = new_bs_quiesce_counter - child->parent_quiesce_counter;
-
- /*
- * If the new child node is drained but the old one was not, flush
- * all outstanding requests to the old child node.
- */
- while (drain_saldo > 0 && child->klass->drained_begin) {
- bdrv_parent_drained_begin_single(child, true);
- drain_saldo--;
- }
-
+ /* TODO Pull this up into the callers to avoid polling here */
+ bdrv_graph_wrlock();
if (old_bs) {
- /* Detach first so that the recursive drain sections coming from @child
- * are already gone and we only end the drain sections that came from
- * elsewhere. */
if (child->klass->detach) {
child->klass->detach(child);
}
- assert_bdrv_graph_writable(old_bs);
QLIST_REMOVE(child, next_parent);
}
child->bs = new_bs;
if (new_bs) {
- assert_bdrv_graph_writable(new_bs);
QLIST_INSERT_HEAD(&new_bs->parents, child, next_parent);
-
- /*
- * Detaching the old node may have led to the new node's
- * quiesce_counter having been decreased. Not a problem, we
- * just need to recognize this here and then invoke
- * drained_end appropriately more often.
- */
- assert(new_bs->quiesce_counter <= new_bs_quiesce_counter);
- drain_saldo += new_bs->quiesce_counter - new_bs_quiesce_counter;
-
- /* Attach only after starting new drained sections, so that recursive
- * drain sections coming from @child don't get an extra .drained_begin
- * callback. */
if (child->klass->attach) {
child->klass->attach(child);
}
}
+ bdrv_graph_wrunlock();
/*
- * If the old child node was drained but the new one is not, allow
- * requests to come in only after the new node has been attached.
+ * If the parent was drained through this BdrvChild previously, but new_bs
+ * is not drained, allow requests to come in only after the new node has
+ * been attached.
*/
- while (drain_saldo < 0 && child->klass->drained_end) {
+ new_bs_quiesce_counter = (new_bs ? new_bs->quiesce_counter : 0);
+ if (!new_bs_quiesce_counter && child->quiesced_parent) {
bdrv_parent_drained_end_single(child);
- drain_saldo++;
}
}
}
bdrv_ref(child_bs);
+ /*
+ * Let every new BdrvChild start with a drained parent. Inserting the child
+ * in the graph with bdrv_replace_child_noperm() will undrain it if
+ * @child_bs is not drained.
+ *
+ * The child was only just created and is not yet visible in global state
+ * until bdrv_replace_child_noperm() inserts it into the graph, so nobody
+ * could have sent requests and polling is not necessary.
+ *
+ * Note that this means that the parent isn't fully drained yet, we only
+ * stop new requests from coming in. This is fine, we don't care about the
+ * old requests here, they are not for this child. If another place enters a
+ * drain section for the same parent, but wants it to be fully quiesced, it
+ * will not run most of the the code in .drained_begin() again (which is not
+ * a problem, we already did this), but it will still poll until the parent
+ * is fully quiesced, so it will not be negatively affected either.
+ */
+ bdrv_parent_drained_begin_single(new_child);
bdrv_replace_child_noperm(new_child, child_bs);
BdrvAttachChildCommonState *s = g_new(BdrvAttachChildCommonState, 1);
return bdrv_set_file_or_backing_noperm(bs, backing_hd, true, tran, errp);
}
-int bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd,
- Error **errp)
+int bdrv_set_backing_hd_drained(BlockDriverState *bs,
+ BlockDriverState *backing_hd,
+ Error **errp)
{
int ret;
Transaction *tran = tran_new();
GLOBAL_STATE_CODE();
- bdrv_drained_begin(bs);
+ assert(bs->quiesce_counter > 0);
ret = bdrv_set_backing_noperm(bs, backing_hd, tran, errp);
if (ret < 0) {
ret = bdrv_refresh_perms(bs, tran, errp);
out:
tran_finalize(tran, ret);
+ return ret;
+}
+int bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd,
+ Error **errp)
+{
+ int ret;
+ GLOBAL_STATE_CODE();
+
+ bdrv_drained_begin(bs);
+ ret = bdrv_set_backing_hd_drained(bs, backing_hd, errp);
bdrv_drained_end(bs);
return ret;
* The reference parameter may be used to specify an existing block device which
* should be opened. If specified, neither options nor a filename may be given,
* nor can an existing BDS be reused (that is, *pbs has to be NULL).
+ *
+ * The caller must always hold @filename AioContext lock, because this
+ * function eventually calls bdrv_refresh_total_sectors() which polls
+ * when called from non-coroutine context.
*/
static BlockDriverState *bdrv_open_inherit(const char *filename,
const char *reference,
return NULL;
}
+/*
+ * The caller must always hold @filename AioContext lock, because this
+ * function eventually calls bdrv_refresh_total_sectors() which polls
+ * when called from non-coroutine context.
+ */
BlockDriverState *bdrv_open(const char *filename, const char *reference,
QDict *options, int flags, Error **errp)
{
}
if (child->bs) {
+ BlockDriverState *bs = child->bs;
+ bdrv_drained_begin(bs);
bdrv_replace_child_tran(child, NULL, tran);
+ bdrv_drained_end(bs);
}
tran_add(tran, &bdrv_remove_child_drv, child);
}
+static void undrain_on_clean_cb(void *opaque)
+{
+ bdrv_drained_end(opaque);
+}
+
+static TransactionActionDrv undrain_on_clean = {
+ .clean = undrain_on_clean_cb,
+};
+
static int bdrv_replace_node_noperm(BlockDriverState *from,
BlockDriverState *to,
bool auto_skip, Transaction *tran,
GLOBAL_STATE_CODE();
+ bdrv_drained_begin(from);
+ bdrv_drained_begin(to);
+ tran_add(tran, &undrain_on_clean, from);
+ tran_add(tran, &undrain_on_clean, to);
+
QLIST_FOREACH_SAFE(c, &from->parents, next_parent, next) {
assert(c->bs == from);
if (!should_update_child(c, to)) {
BdrvCheckResult *res, BdrvCheckMode fix)
{
IO_CODE();
+ assert_bdrv_graph_readable();
if (bs->drv == NULL) {
return -ENOMEDIUM;
}
GLOBAL_STATE_CODE();
bdrv_ref(top);
- bdrv_subtree_drained_begin(top);
+ bdrv_drained_begin(base);
if (!top->drv || !base->drv) {
goto exit;
ret = 0;
exit:
- bdrv_subtree_drained_end(top);
+ bdrv_drained_end(base);
bdrv_unref(top);
return ret;
}
/**
- * Implementation of BlockDriver.bdrv_get_allocated_file_size() that
+ * Implementation of BlockDriver.bdrv_co_get_allocated_file_size() that
* sums the size of all data-bearing children. (This excludes backing
* children.)
*/
if (child->role & (BDRV_CHILD_DATA | BDRV_CHILD_METADATA |
BDRV_CHILD_FILTERED))
{
- child_size = bdrv_get_allocated_file_size(child->bs);
+ child_size = bdrv_co_get_allocated_file_size(child->bs);
if (child_size < 0) {
return child_size;
}
* Length of a allocated file in bytes. Sparse files are counted by actual
* allocated space. Return < 0 if error or unknown.
*/
-int64_t bdrv_get_allocated_file_size(BlockDriverState *bs)
+int64_t coroutine_fn bdrv_co_get_allocated_file_size(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
IO_CODE();
if (!drv) {
return -ENOMEDIUM;
}
- if (drv->bdrv_get_allocated_file_size) {
- return drv->bdrv_get_allocated_file_size(bs);
+ if (drv->bdrv_co_get_allocated_file_size) {
+ return drv->bdrv_co_get_allocated_file_size(bs);
}
if (drv->bdrv_file_open) {
return -ENOTSUP;
} else if (drv->is_filter) {
/* Filter drivers default to the size of their filtered child */
- return bdrv_get_allocated_file_size(bdrv_filter_bs(bs));
+ return bdrv_co_get_allocated_file_size(bdrv_filter_bs(bs));
} else {
/* Other drivers default to summing their children's sizes */
return bdrv_sum_allocated_file_size(bs);
/**
* Return number of sectors on success, -errno on error.
*/
-int64_t bdrv_nb_sectors(BlockDriverState *bs)
+int64_t coroutine_fn bdrv_co_nb_sectors(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
IO_CODE();
return -ENOMEDIUM;
if (drv->has_variable_length) {
- int ret = refresh_total_sectors(bs, bs->total_sectors);
+ int ret = bdrv_co_refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
return ret;
}
* Return length in bytes on success, -errno on error.
* The length is always a multiple of BDRV_SECTOR_SIZE.
*/
-int64_t bdrv_getlength(BlockDriverState *bs)
+int64_t coroutine_fn bdrv_co_getlength(BlockDriverState *bs)
{
- int64_t ret = bdrv_nb_sectors(bs);
+ int64_t ret;
IO_CODE();
+ ret = bdrv_co_nb_sectors(bs);
if (ret < 0) {
return ret;
}
pstrcpy(filename, filename_size, bs->backing_file);
}
-int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
+int coroutine_fn bdrv_co_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
int ret;
BlockDriver *drv = bs->drv;
if (!drv) {
return -ENOMEDIUM;
}
- if (!drv->bdrv_get_info) {
+ if (!drv->bdrv_co_get_info) {
BlockDriverState *filtered = bdrv_filter_bs(bs);
if (filtered) {
- return bdrv_get_info(filtered, bdi);
+ return bdrv_co_get_info(filtered, bdi);
}
return -ENOTSUP;
}
memset(bdi, 0, sizeof(*bdi));
- ret = drv->bdrv_get_info(bs, bdi);
+ ret = drv->bdrv_co_get_info(bs, bdi);
if (ret < 0) {
return ret;
}
return drv->bdrv_get_specific_stats(bs);
}
-void bdrv_debug_event(BlockDriverState *bs, BlkdebugEvent event)
+void coroutine_fn bdrv_co_debug_event(BlockDriverState *bs, BlkdebugEvent event)
{
IO_CODE();
- if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
+ if (!bs || !bs->drv || !bs->drv->bdrv_co_debug_event) {
return;
}
- bs->drv->bdrv_debug_event(bs, event);
+ bs->drv->bdrv_co_debug_event(bs, event);
}
static BlockDriverState *bdrv_find_debug_node(BlockDriverState *bs)
bdrv_dirty_bitmap_skip_store(bm, false);
}
- ret = refresh_total_sectors(bs, bs->total_sectors);
+ ret = bdrv_refresh_total_sectors(bs, bs->total_sectors);
if (ret < 0) {
bs->open_flags |= BDRV_O_INACTIVE;
error_setg_errno(errp, -ret, "Could not refresh total sector count");
IO_CODE();
assert(!(bs->open_flags & BDRV_O_INACTIVE));
+ assert_bdrv_graph_readable();
if (bs->drv->bdrv_co_invalidate_cache) {
bs->drv->bdrv_co_invalidate_cache(bs, &local_err);
/**
* Return TRUE if the media is present
*/
-bool bdrv_is_inserted(BlockDriverState *bs)
+bool coroutine_fn bdrv_co_is_inserted(BlockDriverState *bs)
{
BlockDriver *drv = bs->drv;
BdrvChild *child;
if (!drv) {
return false;
}
- if (drv->bdrv_is_inserted) {
- return drv->bdrv_is_inserted(bs);
+ if (drv->bdrv_co_is_inserted) {
+ return drv->bdrv_co_is_inserted(bs);
}
QLIST_FOREACH(child, &bs->children, next) {
- if (!bdrv_is_inserted(child->bs)) {
+ if (!bdrv_co_is_inserted(child->bs)) {
return false;
}
}
/**
* If eject_flag is TRUE, eject the media. Otherwise, close the tray
*/
-void bdrv_eject(BlockDriverState *bs, bool eject_flag)
+void coroutine_fn bdrv_co_eject(BlockDriverState *bs, bool eject_flag)
{
BlockDriver *drv = bs->drv;
IO_CODE();
- if (drv && drv->bdrv_eject) {
- drv->bdrv_eject(bs, eject_flag);
+ if (drv && drv->bdrv_co_eject) {
+ drv->bdrv_co_eject(bs, eject_flag);
}
}
* Lock or unlock the media (if it is locked, the user won't be able
* to eject it manually).
*/
-void bdrv_lock_medium(BlockDriverState *bs, bool locked)
+void coroutine_fn bdrv_co_lock_medium(BlockDriverState *bs, bool locked)
{
BlockDriver *drv = bs->drv;
IO_CODE();
trace_bdrv_lock_medium(bs, locked);
- if (drv && drv->bdrv_lock_medium) {
- drv->bdrv_lock_medium(bs, locked);
+ if (drv && drv->bdrv_co_lock_medium) {
+ drv->bdrv_co_lock_medium(bs, locked);
}
}
return true;
}
+/*
+ * Must not be called while holding the lock of an AioContext other than the
+ * current one.
+ */
void bdrv_img_create(const char *filename, const char *fmt,
const char *base_filename, const char *base_fmt,
char *options, uint64_t img_size, int flags, bool quiet,
}
}
-void bdrv_coroutine_enter(BlockDriverState *bs, Coroutine *co)
-{
- IO_CODE();
- aio_co_enter(bdrv_get_aio_context(bs), co);
-}
-
static void bdrv_do_remove_aio_context_notifier(BdrvAioNotifier *ban)
{
GLOBAL_STATE_CODE();
if (bs->quiesce_counter) {
aio_enable_external(bs->aio_context);
}
- assert_bdrv_graph_writable(bs);
bs->aio_context = NULL;
}
aio_disable_external(new_context);
}
- assert_bdrv_graph_writable(bs);
bs->aio_context = new_context;
if (bs->drv && bs->drv->bdrv_attach_aio_context) {
BlockDriverState *bs = (BlockDriverState *) state->bs;
AioContext *new_context = state->new_ctx;
AioContext *old_context = bdrv_get_aio_context(bs);
- assert_bdrv_graph_writable(bs);
/*
* Take the old AioContex when detaching it from bs.