The following changes since commit 609ef9f451759151d0bfe7c3843410ab94d68f18: Merge remote-tracking branch 'remotes/berrange/tags/qio-next-pull-request' into staging (2018-06-28 17:53:31 +0100) are available in the git repository at: git://repo.or.cz/qemu/kevin.git tags/for-upstream for you to fetch changes up to 583c99d39368526dfb57a715b04a6ceea27dbe1e: block: Remove unused sector-based vectored I/O (2018-06-29 14:20:56 +0200) ---------------------------------------------------------------- Block layer patches: - Make truncate operations asynchronous (so that preallocation in blockdev-create doesn't block the main loop any more) - usb-storage: Add rerror/werror properties - nvme: Add num_queues property - qemu-img convert: Copy offloading fixes (including data corruption fix) - qcow2: Fix cluster leak on temporary write error - Use byte-based functions instead of bdrv_co_readv/writev() - Various small fixes and cleanups ---------------------------------------------------------------- Eric Blake (8): parallels: Switch to byte-based calls qcow: Switch get_cluster_offset to be byte-based qcow: Switch qcow_co_readv to byte-based calls qcow: Switch qcow_co_writev to byte-based calls qcow: Switch to a byte-based driver replication: Switch to byte-based calls vhdx: Switch to byte-based calls block: Remove unused sector-based vectored I/O Fam Zheng (5): qcow2: Remove dead check on !ret block: Move request tracking to children in copy offloading qcow2: Fix src_offset in copy offloading iscsi: Don't blindly use designator length in response for memcpy file-posix: Fix EINTR handling Kevin Wolf (12): qapi/job: The next release will be 3.0 usb-storage: Add rerror/werror properties qcow2: Fix qcow2_truncate() error return value block: Convert .bdrv_truncate callback to coroutine_fn qcow2: Remove coroutine trampoline for preallocate_co() block: Move bdrv_truncate() implementation to io.c block: Use tracked request for truncate file-posix: Make .bdrv_co_truncate asynchronous qemu-iotests: Update 026.out.nocache reference output qcow2: Free allocated clusters on write error qemu-iotests: Test qcow2 not leaking clusters on write error file-posix: Implement co versions of discard/flush Markus Armbruster (3): block-qdict: Pacify Coverity after commit f1b34a248e9 block/crypto: Pacify Coverity after commit f853465aacb block/crypto: Simplify block_crypto_{open,create}_opts_init() Weiping Zhang (1): hw/block/nvme: add optional parameter num_queues for nvme device qapi/job.json | 18 +- block/crypto.h | 8 +- block/qcow2.h | 1 + include/block/block.h | 8 +- include/block/block_int.h | 7 +- include/block/raw-aio.h | 4 +- include/hw/scsi/scsi.h | 2 + block.c | 64 +------ block/copy-on-read.c | 8 +- block/crypto.c | 112 +++-------- block/file-posix.c | 367 +++++++++++++++++++------------------ block/file-win32.c | 6 +- block/gluster.c | 14 +- block/io.c | 219 +++++++++++++++------- block/iscsi.c | 10 +- block/nfs.c | 7 +- block/parallels.c | 16 +- block/qcow.c | 135 +++++++------- block/qcow2-cluster.c | 11 ++ block/qcow2.c | 140 ++++++-------- block/qed.c | 8 +- block/raw-format.c | 8 +- block/rbd.c | 8 +- block/replication.c | 14 +- block/sheepdog.c | 12 +- block/ssh.c | 6 +- block/vhdx.c | 12 +- hw/block/nvme.c | 5 +- hw/scsi/scsi-bus.c | 11 +- hw/usb/dev-storage.c | 2 + qobject/block-qdict.c | 16 +- tests/qemu-iotests/026 | 17 ++ tests/qemu-iotests/026.out | 8 + tests/qemu-iotests/026.out.nocache | 14 +- tests/qemu-iotests/063 | 9 + tests/qemu-iotests/063.out | 12 ++ 36 files changed, 685 insertions(+), 634 deletions(-) From: Markus Armbruster <armbru@redhat.com> Commit f1b34a248e9 replaced less-than-obvious test in qdict_flatten_qdict() by the obvious one. Sadly, it made something else non-obvious: the fact that @new_key passed to qdict_put_obj() can't be null, because that depends on the function's precondition (target == qdict) == !prefix. Tweak the function some more to help Coverity and human readers alike. Fixes: CID 1393620 Signed-off-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qobject/block-qdict.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/qobject/block-qdict.c b/qobject/block-qdict.c index XXXXXXX..XXXXXXX 100644 --- a/qobject/block-qdict.c +++ b/qobject/block-qdict.c @@ -XXX,XX +XXX,XX @@ static void qdict_flatten_qdict(QDict *qdict, QDict *target, const char *prefix) const QDictEntry *entry, *next; QDict *dict_val; QList *list_val; - char *new_key; + char *key, *new_key; entry = qdict_first(qdict); @@ -XXX,XX +XXX,XX @@ static void qdict_flatten_qdict(QDict *qdict, QDict *target, const char *prefix) value = qdict_entry_value(entry); dict_val = qobject_to(QDict, value); list_val = qobject_to(QList, value); - new_key = NULL; if (prefix) { - new_key = g_strdup_printf("%s.%s", prefix, entry->key); + key = new_key = g_strdup_printf("%s.%s", prefix, entry->key); + } else { + key = entry->key; + new_key = NULL; } /* @@ -XXX,XX +XXX,XX @@ static void qdict_flatten_qdict(QDict *qdict, QDict *target, const char *prefix) * well advised not to modify them altogether.) */ if (dict_val && qdict_size(dict_val)) { - qdict_flatten_qdict(dict_val, target, - new_key ? new_key : entry->key); + qdict_flatten_qdict(dict_val, target, key); if (target == qdict) { qdict_del(qdict, entry->key); } } else if (list_val && !qlist_empty(list_val)) { - qdict_flatten_qlist(list_val, target, - new_key ? new_key : entry->key); + qdict_flatten_qlist(list_val, target, key); if (target == qdict) { qdict_del(qdict, entry->key); } } else if (target != qdict) { - qdict_put_obj(target, new_key, qobject_ref(value)); + qdict_put_obj(target, key, qobject_ref(value)); } g_free(new_key); -- 2.13.6 From: Markus Armbruster <armbru@redhat.com> Coverity can't see that qobject_input_visitor_new_flat_confused() returns non-null when it doesn't set @local_err. Check the return value instead, like all the other callers do. Fixes: CID 1393615 Fixes: CID 1393616 Signed-off-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/crypto.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ block_crypto_open_opts_init(QCryptoBlockFormat format, ret->format = format; v = qobject_input_visitor_new_flat_confused(opts, &local_err); - if (local_err) { + if (!v) { goto out; } @@ -XXX,XX +XXX,XX @@ block_crypto_create_opts_init(QCryptoBlockFormat format, ret->format = format; v = qobject_input_visitor_new_flat_confused(opts, &local_err); - if (local_err) { + if (!v) { goto out; } -- 2.13.6 Commit 51f63ec7d tried to change all references to 2.13 into 3.0, but it failed to achieve this because it was not properly rebased on top of the series introducing qapi/job.json. Change the references now. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- qapi/job.json | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/qapi/job.json b/qapi/job.json index XXXXXXX..XXXXXXX 100644 --- a/qapi/job.json +++ b/qapi/job.json @@ -XXX,XX +XXX,XX @@ # @id: The job identifier # @status: The new job status # -# Since: 2.13 +# Since: 3.0 ## { 'event': 'JOB_STATUS_CHANGE', 'data': { 'id': 'str', @@ -XXX,XX +XXX,XX @@ # # @id: The job identifier. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-pause', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # # @id : The job identifier. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-resume', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # # @id: The job identifier. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-cancel', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # # @id: The job identifier. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-complete', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # # @id: The job identifier. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-dismiss', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # @id: The identifier of any job in the transaction, or of a job that is not # part of any transaction. # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'job-finalize', 'data': { 'id': 'str' } } @@ -XXX,XX +XXX,XX @@ # the reason for the job failure. It should not be parsed # by applications. # -# Since: 2.13 +# Since: 3.0 ## { 'struct': 'JobInfo', 'data': { 'id': 'str', 'type': 'JobType', 'status': 'JobStatus', @@ -XXX,XX +XXX,XX @@ # # Returns: a list with a @JobInfo for each active job # -# Since: 2.13 +# Since: 3.0 ## { 'command': 'query-jobs', 'returns': ['JobInfo'] } -- 2.13.6 The error handling policy was traditionally set with -drive, but with -blockdev it is no longer possible to set frontend options. scsi-disk (and other block devices) have long supported qdev properties to configure the error handling policy, so let's add these options to usb-storage as well and just forward them to the internal scsi-disk instance. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> --- include/hw/scsi/scsi.h | 2 ++ hw/scsi/scsi-bus.c | 11 ++++++++++- hw/usb/dev-storage.c | 2 ++ 3 files changed, 14 insertions(+), 1 deletion(-) diff --git a/include/hw/scsi/scsi.h b/include/hw/scsi/scsi.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/scsi/scsi.h +++ b/include/hw/scsi/scsi.h @@ -XXX,XX +XXX,XX @@ static inline SCSIBus *scsi_bus_from_device(SCSIDevice *d) SCSIDevice *scsi_bus_legacy_add_drive(SCSIBus *bus, BlockBackend *blk, int unit, bool removable, int bootindex, bool share_rw, + BlockdevOnError rerror, + BlockdevOnError werror, const char *serial, Error **errp); void scsi_bus_legacy_handle_cmdline(SCSIBus *bus); void scsi_legacy_handle_cmdline(void); diff --git a/hw/scsi/scsi-bus.c b/hw/scsi/scsi-bus.c index XXXXXXX..XXXXXXX 100644 --- a/hw/scsi/scsi-bus.c +++ b/hw/scsi/scsi-bus.c @@ -XXX,XX +XXX,XX @@ static void scsi_qdev_unrealize(DeviceState *qdev, Error **errp) SCSIDevice *scsi_bus_legacy_add_drive(SCSIBus *bus, BlockBackend *blk, int unit, bool removable, int bootindex, bool share_rw, + BlockdevOnError rerror, + BlockdevOnError werror, const char *serial, Error **errp) { const char *driver; @@ -XXX,XX +XXX,XX @@ SCSIDevice *scsi_bus_legacy_add_drive(SCSIBus *bus, BlockBackend *blk, object_unparent(OBJECT(dev)); return NULL; } + + qdev_prop_set_enum(dev, "rerror", rerror); + qdev_prop_set_enum(dev, "werror", werror); + object_property_set_bool(OBJECT(dev), true, "realized", &err); if (err != NULL) { error_propagate(errp, err); @@ -XXX,XX +XXX,XX @@ void scsi_bus_legacy_handle_cmdline(SCSIBus *bus) } qemu_opts_loc_restore(dinfo->opts); scsi_bus_legacy_add_drive(bus, blk_by_legacy_dinfo(dinfo), - unit, false, -1, false, NULL, &error_fatal); + unit, false, -1, false, + BLOCKDEV_ON_ERROR_AUTO, + BLOCKDEV_ON_ERROR_AUTO, + NULL, &error_fatal); } loc_pop(&loc); } diff --git a/hw/usb/dev-storage.c b/hw/usb/dev-storage.c index XXXXXXX..XXXXXXX 100644 --- a/hw/usb/dev-storage.c +++ b/hw/usb/dev-storage.c @@ -XXX,XX +XXX,XX @@ static void usb_msd_storage_realize(USBDevice *dev, Error **errp) &usb_msd_scsi_info_storage, NULL); scsi_dev = scsi_bus_legacy_add_drive(&s->bus, blk, 0, !!s->removable, s->conf.bootindex, s->conf.share_rw, + s->conf.rerror, s->conf.werror, dev->serial, errp); blk_unref(blk); @@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_usb_msd = { static Property msd_properties[] = { DEFINE_BLOCK_PROPERTIES(MSDState, conf), + DEFINE_BLOCK_ERROR_PROPERTIES(MSDState, conf), DEFINE_PROP_BIT("removable", MSDState, removable, 0, false), DEFINE_PROP_END_OF_LIST(), }; -- 2.13.6 From: Weiping Zhang <zhangweiping@didichuxing.com> Add an optional paramter num_queues for device, and set it to 64 by default. Signed-off-by: Weiping Zhang <zhangweiping@didichuxing.com> Acked-by: Keith Busch <keith.busch@intel.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- hw/block/nvme.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/hw/block/nvme.c b/hw/block/nvme.c index XXXXXXX..XXXXXXX 100644 --- a/hw/block/nvme.c +++ b/hw/block/nvme.c @@ -XXX,XX +XXX,XX @@ * Usage: add options: * -drive file=<file>,if=none,id=<drive_id> * -device nvme,drive=<drive_id>,serial=<serial>,id=<id[optional]>, \ - * cmb_size_mb=<cmb_size_mb[optional]> + * cmb_size_mb=<cmb_size_mb[optional]>, \ + * num_queues=<N[optional]> * * Note cmb_size_mb denotes size of CMB in MB. CMB is assumed to be at * offset 0 in BAR2 and supports only WDS, RDS and SQS for now. @@ -XXX,XX +XXX,XX @@ static void nvme_realize(PCIDevice *pci_dev, Error **errp) pcie_endpoint_cap_init(&n->parent_obj, 0x80); n->num_namespaces = 1; - n->num_queues = 64; n->reg_size = pow2ceil(0x1004 + 2 * (n->num_queues + 1) * 4); n->ns_size = bs_size / (uint64_t)n->num_namespaces; @@ -XXX,XX +XXX,XX @@ static Property nvme_props[] = { DEFINE_BLOCK_PROPERTIES(NvmeCtrl, conf), DEFINE_PROP_STRING("serial", NvmeCtrl, serial), DEFINE_PROP_UINT32("cmb_size_mb", NvmeCtrl, cmb_size_mb, 0), + DEFINE_PROP_UINT32("num_queues", NvmeCtrl, num_queues, 64), DEFINE_PROP_END_OF_LIST(), }; -- 2.13.6 If qcow2_alloc_clusters_at() returns an error, we do need to negate it to get back the positive errno code for error_setg_errno(), but we still need to return the negative error code. Fixes: 772d1f973f87269f6a4a4ea4b880680f3779bbdf Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/qcow2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (clusters_allocated < 0) { error_setg_errno(errp, -clusters_allocated, "Failed to allocate data clusters"); - return -clusters_allocated; + return clusters_allocated; } assert(clusters_allocated == nb_new_data_clusters); -- 2.13.6 bdrv_truncate() is an operation that can block (even for a quite long time, depending on the PreallocMode) in I/O paths that shouldn't block. Convert it to a coroutine_fn so that we have the infrastructure for drivers to make their .bdrv_co_truncate implementation asynchronous. This change could potentially introduce new race conditions because bdrv_truncate() isn't necessarily executed atomically any more. Whether this is a problem needs to be evaluated for each block driver that supports truncate: * file-posix/win32, gluster, iscsi, nfs, rbd, ssh, sheepdog: The protocol drivers are trivially safe because they don't actually yield yet, so there is no change in behaviour. * copy-on-read, crypto, raw-format: Essentially just filter drivers that pass the request to a child node, no problem. * qcow2: The implementation modifies metadata, so it needs to hold s->lock to be safe with concurrent I/O requests. In order to avoid double locking, this requires pulling the locking out into preallocate_co() and using qcow2_write_caches() instead of bdrv_flush(). * qed: Does a single header update, this is fine without locking. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- include/block/block.h | 4 +++ include/block/block_int.h | 4 +-- block.c | 63 +++++++++++++++++++++++++++++++++++----- block/copy-on-read.c | 8 ++--- block/crypto.c | 9 +++--- block/file-posix.c | 12 ++++---- block/file-win32.c | 6 ++-- block/gluster.c | 14 +++++---- block/iscsi.c | 8 ++--- block/nfs.c | 7 +++-- block/qcow2.c | 74 ++++++++++++++++++++++++++++------------------- block/qed.c | 8 +++-- block/raw-format.c | 8 ++--- block/rbd.c | 8 +++-- block/sheepdog.c | 12 ++++---- block/ssh.c | 6 ++-- 16 files changed, 162 insertions(+), 89 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs, const char *backing_file); void bdrv_refresh_filename(BlockDriverState *bs); + +int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, + PreallocMode prealloc, Error **errp); int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, Error **errp); + int64_t bdrv_nb_sectors(BlockDriverState *bs); int64_t bdrv_getlength(BlockDriverState *bs); int64_t bdrv_get_allocated_file_size(BlockDriverState *bs); diff --git a/include/block/block_int.h b/include/block/block_int.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * bdrv_parse_filename. */ const char *protocol_name; - int (*bdrv_truncate)(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp); + int coroutine_fn (*bdrv_co_truncate)(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp); int64_t (*bdrv_getlength)(BlockDriverState *bs); bool has_variable_length; diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ exit: /** * Truncate file to 'offset' bytes (needed only for file protocols) */ -int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, - Error **errp) +int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, + PreallocMode prealloc, Error **errp) { BlockDriverState *bs = child->bs; BlockDriver *drv = bs->drv; @@ -XXX,XX +XXX,XX @@ int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, return -EINVAL; } - if (!drv->bdrv_truncate) { + bdrv_inc_in_flight(bs); + + if (!drv->bdrv_co_truncate) { if (bs->file && drv->is_filter) { - return bdrv_truncate(bs->file, offset, prealloc, errp); + ret = bdrv_co_truncate(bs->file, offset, prealloc, errp); + goto out; } error_setg(errp, "Image format driver does not support resize"); - return -ENOTSUP; + ret = -ENOTSUP; + goto out; } if (bs->read_only) { error_setg(errp, "Image is read-only"); - return -EACCES; + ret = -EACCES; + goto out; } assert(!(bs->open_flags & BDRV_O_INACTIVE)); - ret = drv->bdrv_truncate(bs, offset, prealloc, errp); + ret = drv->bdrv_co_truncate(bs, offset, prealloc, errp); if (ret < 0) { - return ret; + goto out; } ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS); if (ret < 0) { @@ -XXX,XX +XXX,XX @@ int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, bdrv_dirty_bitmap_truncate(bs, offset); bdrv_parent_cb_resize(bs); atomic_inc(&bs->write_gen); + +out: + bdrv_dec_in_flight(bs); return ret; } +typedef struct TruncateCo { + BdrvChild *child; + int64_t offset; + PreallocMode prealloc; + Error **errp; + int ret; +} TruncateCo; + +static void coroutine_fn bdrv_truncate_co_entry(void *opaque) +{ + TruncateCo *tco = opaque; + tco->ret = bdrv_co_truncate(tco->child, tco->offset, tco->prealloc, + tco->errp); +} + +int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, + Error **errp) +{ + Coroutine *co; + TruncateCo tco = { + .child = child, + .offset = offset, + .prealloc = prealloc, + .errp = errp, + .ret = NOT_DONE, + }; + + if (qemu_in_coroutine()) { + /* Fast-path if already in coroutine context */ + bdrv_truncate_co_entry(&tco); + } else { + co = qemu_coroutine_create(bdrv_truncate_co_entry, &tco); + qemu_coroutine_enter(co); + BDRV_POLL_WHILE(child->bs, tco.ret == NOT_DONE); + } + + return tco.ret; +} + /** * Length of a allocated file in bytes. Sparse files are counted by actual * allocated space. Return < 0 if error or unknown. diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ static int64_t cor_getlength(BlockDriverState *bs) } -static int cor_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn cor_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { - return bdrv_truncate(bs->file, offset, prealloc, errp); + return bdrv_co_truncate(bs->file, offset, prealloc, errp); } @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_copy_on_read = { .bdrv_child_perm = cor_child_perm, .bdrv_getlength = cor_getlength, - .bdrv_truncate = cor_truncate, + .bdrv_co_truncate = cor_co_truncate, .bdrv_co_preadv = cor_co_preadv, .bdrv_co_pwritev = cor_co_pwritev, diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ static int block_crypto_co_create_generic(BlockDriverState *bs, return ret; } -static int block_crypto_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn +block_crypto_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BlockCrypto *crypto = bs->opaque; uint64_t payload_offset = @@ -XXX,XX +XXX,XX @@ static int block_crypto_truncate(BlockDriverState *bs, int64_t offset, offset += payload_offset; - return bdrv_truncate(bs->file, offset, prealloc, errp); + return bdrv_co_truncate(bs->file, offset, prealloc, errp); } static void block_crypto_close(BlockDriverState *bs) @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_crypto_luks = { .bdrv_child_perm = bdrv_format_default_perms, .bdrv_co_create = block_crypto_co_create_luks, .bdrv_co_create_opts = block_crypto_co_create_opts_luks, - .bdrv_truncate = block_crypto_truncate, + .bdrv_co_truncate = block_crypto_co_truncate, .create_opts = &block_crypto_create_opts_luks, .bdrv_reopen_prepare = block_crypto_reopen_prepare, diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ out: return result; } -static int raw_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVRawState *s = bs->opaque; struct stat st; @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_file = { .bdrv_io_unplug = raw_aio_unplug, .bdrv_attach_aio_context = raw_aio_attach_aio_context, - .bdrv_truncate = raw_truncate, + .bdrv_co_truncate = raw_co_truncate, .bdrv_getlength = raw_getlength, .bdrv_get_info = raw_get_info, .bdrv_get_allocated_file_size @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_device = { .bdrv_io_plug = raw_aio_plug, .bdrv_io_unplug = raw_aio_unplug, - .bdrv_truncate = raw_truncate, + .bdrv_co_truncate = raw_co_truncate, .bdrv_getlength = raw_getlength, .bdrv_get_info = raw_get_info, .bdrv_get_allocated_file_size @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_io_plug = raw_aio_plug, .bdrv_io_unplug = raw_aio_unplug, - .bdrv_truncate = raw_truncate, + .bdrv_co_truncate = raw_co_truncate, .bdrv_getlength = raw_getlength, .has_variable_length = true, .bdrv_get_allocated_file_size @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_io_plug = raw_aio_plug, .bdrv_io_unplug = raw_aio_unplug, - .bdrv_truncate = raw_truncate, + .bdrv_co_truncate = raw_co_truncate, .bdrv_getlength = raw_getlength, .has_variable_length = true, .bdrv_get_allocated_file_size diff --git a/block/file-win32.c b/block/file-win32.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-win32.c +++ b/block/file-win32.c @@ -XXX,XX +XXX,XX @@ static void raw_close(BlockDriverState *bs) } } -static int raw_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVRawState *s = bs->opaque; LONG low, high; @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_file = { .bdrv_aio_pwritev = raw_aio_pwritev, .bdrv_aio_flush = raw_aio_flush, - .bdrv_truncate = raw_truncate, + .bdrv_co_truncate = raw_co_truncate, .bdrv_getlength = raw_getlength, .bdrv_get_allocated_file_size = raw_get_allocated_file_size, diff --git a/block/gluster.c b/block/gluster.c index XXXXXXX..XXXXXXX 100644 --- a/block/gluster.c +++ b/block/gluster.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qemu_gluster_co_rw(BlockDriverState *bs, return acb.ret; } -static int qemu_gluster_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static coroutine_fn int qemu_gluster_co_truncate(BlockDriverState *bs, + int64_t offset, + PreallocMode prealloc, + Error **errp) { BDRVGlusterState *s = bs->opaque; return qemu_gluster_do_truncate(s->fd, offset, prealloc, errp); @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_gluster = { .bdrv_co_create_opts = qemu_gluster_co_create_opts, .bdrv_getlength = qemu_gluster_getlength, .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size, - .bdrv_truncate = qemu_gluster_truncate, + .bdrv_co_truncate = qemu_gluster_co_truncate, .bdrv_co_readv = qemu_gluster_co_readv, .bdrv_co_writev = qemu_gluster_co_writev, .bdrv_co_flush_to_disk = qemu_gluster_co_flush_to_disk, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_gluster_tcp = { .bdrv_co_create_opts = qemu_gluster_co_create_opts, .bdrv_getlength = qemu_gluster_getlength, .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size, - .bdrv_truncate = qemu_gluster_truncate, + .bdrv_co_truncate = qemu_gluster_co_truncate, .bdrv_co_readv = qemu_gluster_co_readv, .bdrv_co_writev = qemu_gluster_co_writev, .bdrv_co_flush_to_disk = qemu_gluster_co_flush_to_disk, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_gluster_unix = { .bdrv_co_create_opts = qemu_gluster_co_create_opts, .bdrv_getlength = qemu_gluster_getlength, .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size, - .bdrv_truncate = qemu_gluster_truncate, + .bdrv_co_truncate = qemu_gluster_co_truncate, .bdrv_co_readv = qemu_gluster_co_readv, .bdrv_co_writev = qemu_gluster_co_writev, .bdrv_co_flush_to_disk = qemu_gluster_co_flush_to_disk, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_gluster_rdma = { .bdrv_co_create_opts = qemu_gluster_co_create_opts, .bdrv_getlength = qemu_gluster_getlength, .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size, - .bdrv_truncate = qemu_gluster_truncate, + .bdrv_co_truncate = qemu_gluster_co_truncate, .bdrv_co_readv = qemu_gluster_co_readv, .bdrv_co_writev = qemu_gluster_co_writev, .bdrv_co_flush_to_disk = qemu_gluster_co_flush_to_disk, diff --git a/block/iscsi.c b/block/iscsi.c index XXXXXXX..XXXXXXX 100644 --- a/block/iscsi.c +++ b/block/iscsi.c @@ -XXX,XX +XXX,XX @@ static void iscsi_reopen_commit(BDRVReopenState *reopen_state) } } -static int iscsi_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn iscsi_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { IscsiLun *iscsilun = bs->opaque; Error *local_err = NULL; @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_iscsi = { .bdrv_getlength = iscsi_getlength, .bdrv_get_info = iscsi_get_info, - .bdrv_truncate = iscsi_truncate, + .bdrv_co_truncate = iscsi_co_truncate, .bdrv_refresh_limits = iscsi_refresh_limits, .bdrv_co_block_status = iscsi_co_block_status, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_iser = { .bdrv_getlength = iscsi_getlength, .bdrv_get_info = iscsi_get_info, - .bdrv_truncate = iscsi_truncate, + .bdrv_co_truncate = iscsi_co_truncate, .bdrv_refresh_limits = iscsi_refresh_limits, .bdrv_co_block_status = iscsi_co_block_status, diff --git a/block/nfs.c b/block/nfs.c index XXXXXXX..XXXXXXX 100644 --- a/block/nfs.c +++ b/block/nfs.c @@ -XXX,XX +XXX,XX @@ static int64_t nfs_get_allocated_file_size(BlockDriverState *bs) return (task.ret < 0 ? task.ret : st.st_blocks * 512); } -static int nfs_file_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn +nfs_file_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { NFSClient *client = bs->opaque; int ret; @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_nfs = { .bdrv_has_zero_init = nfs_has_zero_init, .bdrv_get_allocated_file_size = nfs_get_allocated_file_size, - .bdrv_truncate = nfs_file_truncate, + .bdrv_co_truncate = nfs_file_co_truncate, .bdrv_file_open = nfs_file_open, .bdrv_close = nfs_file_close, diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn preallocate_co(void *opaque) BlockDriverState *bs = params->bs; uint64_t offset = params->offset; uint64_t new_length = params->new_length; - BDRVQcow2State *s = bs->opaque; uint64_t bytes; uint64_t host_offset = 0; unsigned int cur_bytes; int ret; QCowL2Meta *meta; - qemu_co_mutex_lock(&s->lock); - assert(offset <= new_length); bytes = new_length - offset; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn preallocate_co(void *opaque) ret = 0; done: - qemu_co_mutex_unlock(&s->lock); params->ret = ret; } @@ -XXX,XX +XXX,XX @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp) /* And if we're supposed to preallocate metadata, do that now */ if (qcow2_opts->preallocation != PREALLOC_MODE_OFF) { + BDRVQcow2State *s = blk_bs(blk)->opaque; + qemu_co_mutex_lock(&s->lock); ret = preallocate(blk_bs(blk), 0, qcow2_opts->size); + qemu_co_mutex_unlock(&s->lock); + if (ret < 0) { error_setg_errno(errp, -ret, "Could not preallocate metadata"); goto out; @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int qcow2_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVQcow2State *s = bs->opaque; uint64_t old_length; @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, return -EINVAL; } + qemu_co_mutex_lock(&s->lock); + /* cannot proceed if image has snapshots */ if (s->nb_snapshots) { error_setg(errp, "Can't resize an image which has snapshots"); - return -ENOTSUP; + ret = -ENOTSUP; + goto fail; } /* cannot proceed if image has bitmaps */ if (s->nb_bitmaps) { /* TODO: resize bitmaps in the image */ error_setg(errp, "Can't resize an image which has bitmaps"); - return -ENOTSUP; + ret = -ENOTSUP; + goto fail; } old_length = bs->total_sectors * 512; @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (prealloc != PREALLOC_MODE_OFF) { error_setg(errp, "Preallocation can't be used for shrinking an image"); - return -EINVAL; + ret = -EINVAL; + goto fail; } ret = qcow2_cluster_discard(bs, ROUND_UP(offset, s->cluster_size), @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, QCOW2_DISCARD_ALWAYS, true); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to discard cropped clusters"); - return ret; + goto fail; } ret = qcow2_shrink_l1_table(bs, new_l1_size); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to reduce the number of L2 tables"); - return ret; + goto fail; } ret = qcow2_shrink_reftable(bs); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to discard unused refblocks"); - return ret; + goto fail; } old_file_size = bdrv_getlength(bs->file->bs); if (old_file_size < 0) { error_setg_errno(errp, -old_file_size, "Failed to inquire current file length"); - return old_file_size; + ret = old_file_size; + goto fail; } last_cluster = qcow2_get_last_cluster(bs, old_file_size); if (last_cluster < 0) { error_setg_errno(errp, -last_cluster, "Failed to find the last cluster"); - return last_cluster; + ret = last_cluster; + goto fail; } if ((last_cluster + 1) * s->cluster_size < old_file_size) { Error *local_err = NULL; - bdrv_truncate(bs->file, (last_cluster + 1) * s->cluster_size, - PREALLOC_MODE_OFF, &local_err); + bdrv_co_truncate(bs->file, (last_cluster + 1) * s->cluster_size, + PREALLOC_MODE_OFF, &local_err); if (local_err) { warn_reportf_err(local_err, "Failed to truncate the tail of the image: "); @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, ret = qcow2_grow_l1_table(bs, new_l1_size, true); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to grow the L1 table"); - return ret; + goto fail; } } @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, ret = preallocate(bs, old_length, offset); if (ret < 0) { error_setg_errno(errp, -ret, "Preallocation failed"); - return ret; + goto fail; } break; @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (old_file_size < 0) { error_setg_errno(errp, -old_file_size, "Failed to inquire current file length"); - return old_file_size; + ret = old_file_size; + goto fail; } old_file_size = ROUND_UP(old_file_size, s->cluster_size); @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (allocation_start < 0) { error_setg_errno(errp, -allocation_start, "Failed to resize refcount structures"); - return allocation_start; + ret = allocation_start; + goto fail; } clusters_allocated = qcow2_alloc_clusters_at(bs, allocation_start, @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (clusters_allocated < 0) { error_setg_errno(errp, -clusters_allocated, "Failed to allocate data clusters"); - return clusters_allocated; + ret = clusters_allocated; + goto fail; } assert(clusters_allocated == nb_new_data_clusters); @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, /* Allocate the data area */ new_file_size = allocation_start + nb_new_data_clusters * s->cluster_size; - ret = bdrv_truncate(bs->file, new_file_size, prealloc, errp); + ret = bdrv_co_truncate(bs->file, new_file_size, prealloc, errp); if (ret < 0) { error_prepend(errp, "Failed to resize underlying file: "); qcow2_free_clusters(bs, allocation_start, nb_new_data_clusters * s->cluster_size, QCOW2_DISCARD_OTHER); - return ret; + goto fail; } /* Create the necessary L2 entries */ @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, qcow2_free_clusters(bs, host_offset, nb_new_data_clusters * s->cluster_size, QCOW2_DISCARD_OTHER); - return ret; + goto fail; } guest_offset += nb_clusters * s->cluster_size; @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, if (prealloc != PREALLOC_MODE_OFF) { /* Flush metadata before actually changing the image size */ - ret = bdrv_flush(bs); + ret = qcow2_write_caches(bs); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to flush the preallocated area to disk"); - return ret; + goto fail; } } @@ -XXX,XX +XXX,XX @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset, &offset, sizeof(uint64_t)); if (ret < 0) { error_setg_errno(errp, -ret, "Failed to update the image size"); - return ret; + goto fail; } s->l1_vm_state_index = new_l1_size; - return 0; + ret = 0; +fail: + qemu_co_mutex_unlock(&s->lock); + return ret; } /* XXX: put compressed sectors first, then all the cluster aligned @@ -XXX,XX +XXX,XX @@ qcow2_co_pwritev_compressed(BlockDriverState *bs, uint64_t offset, if (cluster_offset < 0) { return cluster_offset; } - return bdrv_truncate(bs->file, cluster_offset, PREALLOC_MODE_OFF, NULL); + return bdrv_co_truncate(bs->file, cluster_offset, PREALLOC_MODE_OFF, + NULL); } if (offset_into_cluster(s, offset)) { @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_qcow2 = { .bdrv_co_pdiscard = qcow2_co_pdiscard, .bdrv_co_copy_range_from = qcow2_co_copy_range_from, .bdrv_co_copy_range_to = qcow2_co_copy_range_to, - .bdrv_truncate = qcow2_truncate, + .bdrv_co_truncate = qcow2_co_truncate, .bdrv_co_pwritev_compressed = qcow2_co_pwritev_compressed, .bdrv_make_empty = qcow2_make_empty, diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_qed_co_pwrite_zeroes(BlockDriverState *bs, QED_AIOCB_WRITE | QED_AIOCB_ZERO); } -static int bdrv_qed_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn bdrv_qed_co_truncate(BlockDriverState *bs, + int64_t offset, + PreallocMode prealloc, + Error **errp) { BDRVQEDState *s = bs->opaque; uint64_t old_image_size; @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_qed = { .bdrv_co_readv = bdrv_qed_co_readv, .bdrv_co_writev = bdrv_qed_co_writev, .bdrv_co_pwrite_zeroes = bdrv_qed_co_pwrite_zeroes, - .bdrv_truncate = bdrv_qed_truncate, + .bdrv_co_truncate = bdrv_qed_co_truncate, .bdrv_getlength = bdrv_qed_getlength, .bdrv_get_info = bdrv_qed_get_info, .bdrv_refresh_limits = bdrv_qed_refresh_limits, diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static void raw_refresh_limits(BlockDriverState *bs, Error **errp) } } -static int raw_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVRawState *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static int raw_truncate(BlockDriverState *bs, int64_t offset, s->size = offset; offset += s->offset; - return bdrv_truncate(bs->file, offset, prealloc, errp); + return bdrv_co_truncate(bs->file, offset, prealloc, errp); } static void raw_eject(BlockDriverState *bs, bool eject_flag) @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_raw = { .bdrv_co_block_status = &raw_co_block_status, .bdrv_co_copy_range_from = &raw_co_copy_range_from, .bdrv_co_copy_range_to = &raw_co_copy_range_to, - .bdrv_truncate = &raw_truncate, + .bdrv_co_truncate = &raw_co_truncate, .bdrv_getlength = &raw_getlength, .has_variable_length = true, .bdrv_measure = &raw_measure, diff --git a/block/rbd.c b/block/rbd.c index XXXXXXX..XXXXXXX 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -XXX,XX +XXX,XX @@ static int64_t qemu_rbd_getlength(BlockDriverState *bs) return info.size; } -static int qemu_rbd_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn qemu_rbd_co_truncate(BlockDriverState *bs, + int64_t offset, + PreallocMode prealloc, + Error **errp) { BDRVRBDState *s = bs->opaque; int r; @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_rbd = { .bdrv_get_info = qemu_rbd_getinfo, .create_opts = &qemu_rbd_create_opts, .bdrv_getlength = qemu_rbd_getlength, - .bdrv_truncate = qemu_rbd_truncate, + .bdrv_co_truncate = qemu_rbd_co_truncate, .protocol_name = "rbd", .bdrv_aio_preadv = qemu_rbd_aio_preadv, diff --git a/block/sheepdog.c b/block/sheepdog.c index XXXXXXX..XXXXXXX 100644 --- a/block/sheepdog.c +++ b/block/sheepdog.c @@ -XXX,XX +XXX,XX @@ static int64_t sd_getlength(BlockDriverState *bs) return s->inode.vdi_size; } -static int sd_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn sd_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVSheepdogState *s = bs->opaque; int ret, fd; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int sd_co_writev(BlockDriverState *bs, int64_t sector_num, assert(!flags); if (offset > s->inode.vdi_size) { - ret = sd_truncate(bs, offset, PREALLOC_MODE_OFF, NULL); + ret = sd_co_truncate(bs, offset, PREALLOC_MODE_OFF, NULL); if (ret < 0) { return ret; } @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_sheepdog = { .bdrv_has_zero_init = bdrv_has_zero_init_1, .bdrv_getlength = sd_getlength, .bdrv_get_allocated_file_size = sd_get_allocated_file_size, - .bdrv_truncate = sd_truncate, + .bdrv_co_truncate = sd_co_truncate, .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_sheepdog_tcp = { .bdrv_has_zero_init = bdrv_has_zero_init_1, .bdrv_getlength = sd_getlength, .bdrv_get_allocated_file_size = sd_get_allocated_file_size, - .bdrv_truncate = sd_truncate, + .bdrv_co_truncate = sd_co_truncate, .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_sheepdog_unix = { .bdrv_has_zero_init = bdrv_has_zero_init_1, .bdrv_getlength = sd_getlength, .bdrv_get_allocated_file_size = sd_get_allocated_file_size, - .bdrv_truncate = sd_truncate, + .bdrv_co_truncate = sd_co_truncate, .bdrv_co_readv = sd_co_readv, .bdrv_co_writev = sd_co_writev, diff --git a/block/ssh.c b/block/ssh.c index XXXXXXX..XXXXXXX 100644 --- a/block/ssh.c +++ b/block/ssh.c @@ -XXX,XX +XXX,XX @@ static int64_t ssh_getlength(BlockDriverState *bs) return length; } -static int ssh_truncate(BlockDriverState *bs, int64_t offset, - PreallocMode prealloc, Error **errp) +static int coroutine_fn ssh_co_truncate(BlockDriverState *bs, int64_t offset, + PreallocMode prealloc, Error **errp) { BDRVSSHState *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_ssh = { .bdrv_co_readv = ssh_co_readv, .bdrv_co_writev = ssh_co_writev, .bdrv_getlength = ssh_getlength, - .bdrv_truncate = ssh_truncate, + .bdrv_co_truncate = ssh_co_truncate, .bdrv_co_flush_to_disk = ssh_co_flush, .create_opts = &ssh_create_opts, }; -- 2.13.6 All callers are coroutine_fns now, so we can just directly call preallocate_co(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/qcow2.c | 51 ++++++++------------------------------------------- 1 file changed, 8 insertions(+), 43 deletions(-) diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int qcow2_set_up_encryption(BlockDriverState *bs, return ret; } - -typedef struct PreallocCo { - BlockDriverState *bs; - uint64_t offset; - uint64_t new_length; - - int ret; -} PreallocCo; - /** * Preallocates metadata structures for data clusters between @offset (in the * guest disk) and @new_length (which is thus generally the new guest disk @@ -XXX,XX +XXX,XX @@ typedef struct PreallocCo { * * Returns: 0 on success, -errno on failure. */ -static void coroutine_fn preallocate_co(void *opaque) +static int coroutine_fn preallocate_co(BlockDriverState *bs, uint64_t offset, + uint64_t new_length) { - PreallocCo *params = opaque; - BlockDriverState *bs = params->bs; - uint64_t offset = params->offset; - uint64_t new_length = params->new_length; uint64_t bytes; uint64_t host_offset = 0; unsigned int cur_bytes; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn preallocate_co(void *opaque) ret = qcow2_alloc_cluster_offset(bs, offset, &cur_bytes, &host_offset, &meta); if (ret < 0) { - goto done; + return ret; } while (meta) { @@ -XXX,XX +XXX,XX @@ static void coroutine_fn preallocate_co(void *opaque) if (ret < 0) { qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters, QCOW2_DISCARD_NEVER); - goto done; + return ret; } /* There are no dependent requests, but we need to remove our @@ -XXX,XX +XXX,XX @@ static void coroutine_fn preallocate_co(void *opaque) ret = bdrv_pwrite(bs->file, (host_offset + cur_bytes) - 1, &data, 1); if (ret < 0) { - goto done; + return ret; } } - ret = 0; - -done: - params->ret = ret; -} - -static int preallocate(BlockDriverState *bs, - uint64_t offset, uint64_t new_length) -{ - PreallocCo params = { - .bs = bs, - .offset = offset, - .new_length = new_length, - .ret = -EINPROGRESS, - }; - - if (qemu_in_coroutine()) { - preallocate_co(&params); - } else { - Coroutine *co = qemu_coroutine_create(preallocate_co, &params); - bdrv_coroutine_enter(bs, co); - BDRV_POLL_WHILE(bs, params.ret == -EINPROGRESS); - } - return params.ret; + return 0; } /* qcow2_refcount_metadata_size: @@ -XXX,XX +XXX,XX @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp) if (qcow2_opts->preallocation != PREALLOC_MODE_OFF) { BDRVQcow2State *s = blk_bs(blk)->opaque; qemu_co_mutex_lock(&s->lock); - ret = preallocate(blk_bs(blk), 0, qcow2_opts->size); + ret = preallocate_co(blk_bs(blk), 0, qcow2_opts->size); qemu_co_mutex_unlock(&s->lock); if (ret < 0) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset, break; case PREALLOC_MODE_METADATA: - ret = preallocate(bs, old_length, offset); + ret = preallocate_co(bs, old_length, offset); if (ret < 0) { error_setg_errno(errp, -ret, "Preallocation failed"); goto fail; -- 2.13.6 This moves the bdrv_truncate() implementation from block.c to block/io.c so it can have access to the tracked requests infrastructure. This involves making refresh_total_sectors() public (in block_int.h). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- include/block/block_int.h | 2 + block.c | 111 +--------------------------------------------- block/io.c | 109 +++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 112 insertions(+), 110 deletions(-) diff --git a/include/block/block_int.h b/include/block/block_int.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range_to(BdrvChild *src, uint64_t src_offset, BdrvChild *dst, uint64_t dst_offset, uint64_t bytes, BdrvRequestFlags flags); +int refresh_total_sectors(BlockDriverState *bs, int64_t hint); + #endif /* BLOCK_INT_H */ diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static int find_image_format(BlockBackend *file, const char *filename, * Set the current 'total_sectors' value * Return 0 on success, -errno on error. */ -static int refresh_total_sectors(BlockDriverState *bs, int64_t hint) +int refresh_total_sectors(BlockDriverState *bs, int64_t hint) { BlockDriver *drv = bs->drv; @@ -XXX,XX +XXX,XX @@ static void bdrv_parent_cb_change_media(BlockDriverState *bs, bool load) } } -static void bdrv_parent_cb_resize(BlockDriverState *bs) -{ - BdrvChild *c; - QLIST_FOREACH(c, &bs->parents, next_parent) { - if (c->role->resize) { - c->role->resize(c); - } - } -} - /* * Sets the backing file link of a BDS. A new reference is created; callers * which don't need their own reference any more must call bdrv_unref(). @@ -XXX,XX +XXX,XX @@ exit: } /** - * Truncate file to 'offset' bytes (needed only for file protocols) - */ -int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, - PreallocMode prealloc, Error **errp) -{ - BlockDriverState *bs = child->bs; - BlockDriver *drv = bs->drv; - int ret; - - assert(child->perm & BLK_PERM_RESIZE); - - /* if bs->drv == NULL, bs is closed, so there's nothing to do here */ - if (!drv) { - error_setg(errp, "No medium inserted"); - return -ENOMEDIUM; - } - if (offset < 0) { - error_setg(errp, "Image size cannot be negative"); - return -EINVAL; - } - - bdrv_inc_in_flight(bs); - - if (!drv->bdrv_co_truncate) { - if (bs->file && drv->is_filter) { - ret = bdrv_co_truncate(bs->file, offset, prealloc, errp); - goto out; - } - error_setg(errp, "Image format driver does not support resize"); - ret = -ENOTSUP; - goto out; - } - if (bs->read_only) { - error_setg(errp, "Image is read-only"); - ret = -EACCES; - goto out; - } - - assert(!(bs->open_flags & BDRV_O_INACTIVE)); - - ret = drv->bdrv_co_truncate(bs, offset, prealloc, errp); - if (ret < 0) { - goto out; - } - ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS); - if (ret < 0) { - error_setg_errno(errp, -ret, "Could not refresh total sector count"); - } else { - offset = bs->total_sectors * BDRV_SECTOR_SIZE; - } - bdrv_dirty_bitmap_truncate(bs, offset); - bdrv_parent_cb_resize(bs); - atomic_inc(&bs->write_gen); - -out: - bdrv_dec_in_flight(bs); - return ret; -} - -typedef struct TruncateCo { - BdrvChild *child; - int64_t offset; - PreallocMode prealloc; - Error **errp; - int ret; -} TruncateCo; - -static void coroutine_fn bdrv_truncate_co_entry(void *opaque) -{ - TruncateCo *tco = opaque; - tco->ret = bdrv_co_truncate(tco->child, tco->offset, tco->prealloc, - tco->errp); -} - -int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, - Error **errp) -{ - Coroutine *co; - TruncateCo tco = { - .child = child, - .offset = offset, - .prealloc = prealloc, - .errp = errp, - .ret = NOT_DONE, - }; - - if (qemu_in_coroutine()) { - /* Fast-path if already in coroutine context */ - bdrv_truncate_co_entry(&tco); - } else { - co = qemu_coroutine_create(bdrv_truncate_co_entry, &tco); - qemu_coroutine_enter(co); - BDRV_POLL_WHILE(child->bs, tco.ret == NOT_DONE); - } - - return tco.ret; -} - -/** * Length of a allocated file in bytes. Sparse files are counted by actual * allocated space. Return < 0 if error or unknown. */ diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range(BdrvChild *src, uint64_t src_offset, bdrv_dec_in_flight(dst_bs); return ret; } + +static void bdrv_parent_cb_resize(BlockDriverState *bs) +{ + BdrvChild *c; + QLIST_FOREACH(c, &bs->parents, next_parent) { + if (c->role->resize) { + c->role->resize(c); + } + } +} + +/** + * Truncate file to 'offset' bytes (needed only for file protocols) + */ +int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, + PreallocMode prealloc, Error **errp) +{ + BlockDriverState *bs = child->bs; + BlockDriver *drv = bs->drv; + int ret; + + assert(child->perm & BLK_PERM_RESIZE); + + /* if bs->drv == NULL, bs is closed, so there's nothing to do here */ + if (!drv) { + error_setg(errp, "No medium inserted"); + return -ENOMEDIUM; + } + if (offset < 0) { + error_setg(errp, "Image size cannot be negative"); + return -EINVAL; + } + + bdrv_inc_in_flight(bs); + + if (!drv->bdrv_co_truncate) { + if (bs->file && drv->is_filter) { + ret = bdrv_co_truncate(bs->file, offset, prealloc, errp); + goto out; + } + error_setg(errp, "Image format driver does not support resize"); + ret = -ENOTSUP; + goto out; + } + if (bs->read_only) { + error_setg(errp, "Image is read-only"); + ret = -EACCES; + goto out; + } + + assert(!(bs->open_flags & BDRV_O_INACTIVE)); + + ret = drv->bdrv_co_truncate(bs, offset, prealloc, errp); + if (ret < 0) { + goto out; + } + ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS); + if (ret < 0) { + error_setg_errno(errp, -ret, "Could not refresh total sector count"); + } else { + offset = bs->total_sectors * BDRV_SECTOR_SIZE; + } + bdrv_dirty_bitmap_truncate(bs, offset); + bdrv_parent_cb_resize(bs); + atomic_inc(&bs->write_gen); + +out: + bdrv_dec_in_flight(bs); + return ret; +} + +typedef struct TruncateCo { + BdrvChild *child; + int64_t offset; + PreallocMode prealloc; + Error **errp; + int ret; +} TruncateCo; + +static void coroutine_fn bdrv_truncate_co_entry(void *opaque) +{ + TruncateCo *tco = opaque; + tco->ret = bdrv_co_truncate(tco->child, tco->offset, tco->prealloc, + tco->errp); +} + +int bdrv_truncate(BdrvChild *child, int64_t offset, PreallocMode prealloc, + Error **errp) +{ + Coroutine *co; + TruncateCo tco = { + .child = child, + .offset = offset, + .prealloc = prealloc, + .errp = errp, + .ret = NOT_DONE, + }; + + if (qemu_in_coroutine()) { + /* Fast-path if already in coroutine context */ + bdrv_truncate_co_entry(&tco); + } else { + co = qemu_coroutine_create(bdrv_truncate_co_entry, &tco); + qemu_coroutine_enter(co); + BDRV_POLL_WHILE(child->bs, tco.ret == NOT_DONE); + } + + return tco.ret; +} -- 2.13.6 When growing an image, block drivers (especially protocol drivers) may initialise the newly added area. I/O requests to the same area need to wait for this initialisation to be completed so that data writes don't get overwritten and reads don't read uninitialised data. To avoid overhead in the fast I/O path by adding new locking in the protocol drivers and to restrict the impact to requests that actually touch the new area, reuse the existing tracked request infrastructure in block/io.c and mark all discard requests as serialising. With this change, it is safe for protocol drivers to make .bdrv_co_truncate actually asynchronous. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- include/block/block_int.h | 1 + block/io.c | 25 +++++++++++++++++++++++++ 2 files changed, 26 insertions(+) diff --git a/include/block/block_int.h b/include/block/block_int.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -XXX,XX +XXX,XX @@ enum BdrvTrackedRequestType { BDRV_TRACKED_READ, BDRV_TRACKED_WRITE, BDRV_TRACKED_DISCARD, + BDRV_TRACKED_TRUNCATE, }; typedef struct BdrvTrackedRequest { diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, { BlockDriverState *bs = child->bs; BlockDriver *drv = bs->drv; + BdrvTrackedRequest req; + int64_t old_size, new_bytes; int ret; assert(child->perm & BLK_PERM_RESIZE); @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, return -EINVAL; } + old_size = bdrv_getlength(bs); + if (old_size < 0) { + error_setg_errno(errp, -old_size, "Failed to get old image size"); + return old_size; + } + + if (offset > old_size) { + new_bytes = offset - old_size; + } else { + new_bytes = 0; + } + bdrv_inc_in_flight(bs); + tracked_request_begin(&req, bs, offset, new_bytes, BDRV_TRACKED_TRUNCATE); + + /* If we are growing the image and potentially using preallocation for the + * new area, we need to make sure that no write requests are made to it + * concurrently or they might be overwritten by preallocation. */ + if (new_bytes) { + mark_request_serialising(&req, 1); + wait_serialising_requests(&req); + } if (!drv->bdrv_co_truncate) { if (bs->file && drv->is_filter) { @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, atomic_inc(&bs->write_gen); out: + tracked_request_end(&req); bdrv_dec_in_flight(bs); + return ret; } -- 2.13.6 This moves the code to resize an image file to the thread pool to avoid blocking. Creating large images with preallocation with blockdev-create is now actually a background job instead of blocking the monitor (and most other things) until the preallocation has completed. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- include/block/raw-aio.h | 4 +- block/file-posix.c | 266 +++++++++++++++++++++++++++--------------------- 2 files changed, 154 insertions(+), 116 deletions(-) diff --git a/include/block/raw-aio.h b/include/block/raw-aio.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/raw-aio.h +++ b/include/block/raw-aio.h @@ -XXX,XX +XXX,XX @@ #define QEMU_AIO_DISCARD 0x0010 #define QEMU_AIO_WRITE_ZEROES 0x0020 #define QEMU_AIO_COPY_RANGE 0x0040 +#define QEMU_AIO_TRUNCATE 0x0080 #define QEMU_AIO_TYPE_MASK \ (QEMU_AIO_READ | \ QEMU_AIO_WRITE | \ @@ -XXX,XX +XXX,XX @@ QEMU_AIO_FLUSH | \ QEMU_AIO_DISCARD | \ QEMU_AIO_WRITE_ZEROES | \ - QEMU_AIO_COPY_RANGE) + QEMU_AIO_COPY_RANGE | \ + QEMU_AIO_TRUNCATE) /* AIO flags */ #define QEMU_AIO_MISALIGNED 0x1000 diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ typedef struct RawPosixAIOData { #define aio_ioctl_cmd aio_nbytes /* for QEMU_AIO_IOCTL */ off_t aio_offset; int aio_type; - int aio_fd2; - off_t aio_offset2; + union { + struct { + int aio_fd2; + off_t aio_offset2; + }; + struct { + PreallocMode prealloc; + Error **errp; + }; + }; } RawPosixAIOData; #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) @@ -XXX,XX +XXX,XX @@ static ssize_t handle_aiocb_discard(RawPosixAIOData *aiocb) return ret; } +static int handle_aiocb_truncate(RawPosixAIOData *aiocb) +{ + int result = 0; + int64_t current_length = 0; + char *buf = NULL; + struct stat st; + int fd = aiocb->aio_fildes; + int64_t offset = aiocb->aio_offset; + Error **errp = aiocb->errp; + + if (fstat(fd, &st) < 0) { + result = -errno; + error_setg_errno(errp, -result, "Could not stat file"); + return result; + } + + current_length = st.st_size; + if (current_length > offset && aiocb->prealloc != PREALLOC_MODE_OFF) { + error_setg(errp, "Cannot use preallocation for shrinking files"); + return -ENOTSUP; + } + + switch (aiocb->prealloc) { +#ifdef CONFIG_POSIX_FALLOCATE + case PREALLOC_MODE_FALLOC: + /* + * Truncating before posix_fallocate() makes it about twice slower on + * file systems that do not support fallocate(), trying to check if a + * block is allocated before allocating it, so don't do that here. + */ + if (offset != current_length) { + result = -posix_fallocate(fd, current_length, + offset - current_length); + if (result != 0) { + /* posix_fallocate() doesn't set errno. */ + error_setg_errno(errp, -result, + "Could not preallocate new data"); + } + } else { + result = 0; + } + goto out; +#endif + case PREALLOC_MODE_FULL: + { + int64_t num = 0, left = offset - current_length; + off_t seek_result; + + /* + * Knowing the final size from the beginning could allow the file + * system driver to do less allocations and possibly avoid + * fragmentation of the file. + */ + if (ftruncate(fd, offset) != 0) { + result = -errno; + error_setg_errno(errp, -result, "Could not resize file"); + goto out; + } + + buf = g_malloc0(65536); + + seek_result = lseek(fd, current_length, SEEK_SET); + if (seek_result < 0) { + result = -errno; + error_setg_errno(errp, -result, + "Failed to seek to the old end of file"); + goto out; + } + + while (left > 0) { + num = MIN(left, 65536); + result = write(fd, buf, num); + if (result < 0) { + result = -errno; + error_setg_errno(errp, -result, + "Could not write zeros for preallocation"); + goto out; + } + left -= result; + } + if (result >= 0) { + result = fsync(fd); + if (result < 0) { + result = -errno; + error_setg_errno(errp, -result, + "Could not flush file to disk"); + goto out; + } + } + goto out; + } + case PREALLOC_MODE_OFF: + if (ftruncate(fd, offset) != 0) { + result = -errno; + error_setg_errno(errp, -result, "Could not resize file"); + } + return result; + default: + result = -ENOTSUP; + error_setg(errp, "Unsupported preallocation mode: %s", + PreallocMode_str(aiocb->prealloc)); + return result; + } + +out: + if (result < 0) { + if (ftruncate(fd, current_length) < 0) { + error_report("Failed to restore old file length: %s", + strerror(errno)); + } + } + + g_free(buf); + return result; +} + static int aio_worker(void *arg) { RawPosixAIOData *aiocb = arg; @@ -XXX,XX +XXX,XX @@ static int aio_worker(void *arg) case QEMU_AIO_COPY_RANGE: ret = handle_aiocb_copy_range(aiocb); break; + case QEMU_AIO_TRUNCATE: + ret = handle_aiocb_truncate(aiocb); + break; default: fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type); ret = -EINVAL; @@ -XXX,XX +XXX,XX @@ static void raw_close(BlockDriverState *bs) * * Returns: 0 on success, -errno on failure. */ -static int raw_regular_truncate(int fd, int64_t offset, PreallocMode prealloc, - Error **errp) +static int coroutine_fn +raw_regular_truncate(BlockDriverState *bs, int fd, int64_t offset, + PreallocMode prealloc, Error **errp) { - int result = 0; - int64_t current_length = 0; - char *buf = NULL; - struct stat st; - - if (fstat(fd, &st) < 0) { - result = -errno; - error_setg_errno(errp, -result, "Could not stat file"); - return result; - } - - current_length = st.st_size; - if (current_length > offset && prealloc != PREALLOC_MODE_OFF) { - error_setg(errp, "Cannot use preallocation for shrinking files"); - return -ENOTSUP; - } - - switch (prealloc) { -#ifdef CONFIG_POSIX_FALLOCATE - case PREALLOC_MODE_FALLOC: - /* - * Truncating before posix_fallocate() makes it about twice slower on - * file systems that do not support fallocate(), trying to check if a - * block is allocated before allocating it, so don't do that here. - */ - if (offset != current_length) { - result = -posix_fallocate(fd, current_length, offset - current_length); - if (result != 0) { - /* posix_fallocate() doesn't set errno. */ - error_setg_errno(errp, -result, - "Could not preallocate new data"); - } - } else { - result = 0; - } - goto out; -#endif - case PREALLOC_MODE_FULL: - { - int64_t num = 0, left = offset - current_length; - off_t seek_result; - - /* - * Knowing the final size from the beginning could allow the file - * system driver to do less allocations and possibly avoid - * fragmentation of the file. - */ - if (ftruncate(fd, offset) != 0) { - result = -errno; - error_setg_errno(errp, -result, "Could not resize file"); - goto out; - } - - buf = g_malloc0(65536); - - seek_result = lseek(fd, current_length, SEEK_SET); - if (seek_result < 0) { - result = -errno; - error_setg_errno(errp, -result, - "Failed to seek to the old end of file"); - goto out; - } - - while (left > 0) { - num = MIN(left, 65536); - result = write(fd, buf, num); - if (result < 0) { - result = -errno; - error_setg_errno(errp, -result, - "Could not write zeros for preallocation"); - goto out; - } - left -= result; - } - if (result >= 0) { - result = fsync(fd); - if (result < 0) { - result = -errno; - error_setg_errno(errp, -result, - "Could not flush file to disk"); - goto out; - } - } - goto out; - } - case PREALLOC_MODE_OFF: - if (ftruncate(fd, offset) != 0) { - result = -errno; - error_setg_errno(errp, -result, "Could not resize file"); - } - return result; - default: - result = -ENOTSUP; - error_setg(errp, "Unsupported preallocation mode: %s", - PreallocMode_str(prealloc)); - return result; - } + RawPosixAIOData *acb = g_new(RawPosixAIOData, 1); + ThreadPool *pool; -out: - if (result < 0) { - if (ftruncate(fd, current_length) < 0) { - error_report("Failed to restore old file length: %s", - strerror(errno)); - } - } + *acb = (RawPosixAIOData) { + .bs = bs, + .aio_fildes = fd, + .aio_type = QEMU_AIO_TRUNCATE, + .aio_offset = offset, + .prealloc = prealloc, + .errp = errp, + }; - g_free(buf); - return result; + /* @bs can be NULL, bdrv_get_aio_context() returns the main context then */ + pool = aio_get_thread_pool(bdrv_get_aio_context(bs)); + return thread_pool_submit_co(pool, aio_worker, acb); } static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, } if (S_ISREG(st.st_mode)) { - return raw_regular_truncate(s->fd, offset, prealloc, errp); + return raw_regular_truncate(bs, s->fd, offset, prealloc, errp); } if (prealloc != PREALLOC_MODE_OFF) { @@ -XXX,XX +XXX,XX @@ static int64_t raw_get_allocated_file_size(BlockDriverState *bs) return (int64_t)st.st_blocks * 512; } -static int raw_co_create(BlockdevCreateOptions *options, Error **errp) +static int coroutine_fn +raw_co_create(BlockdevCreateOptions *options, Error **errp) { BlockdevCreateOptionsFile *file_opts; int fd; @@ -XXX,XX +XXX,XX @@ static int raw_co_create(BlockdevCreateOptions *options, Error **errp) } /* Clear the file by truncating it to 0 */ - result = raw_regular_truncate(fd, 0, PREALLOC_MODE_OFF, errp); + result = raw_regular_truncate(NULL, fd, 0, PREALLOC_MODE_OFF, errp); if (result < 0) { goto out_close; } @@ -XXX,XX +XXX,XX @@ static int raw_co_create(BlockdevCreateOptions *options, Error **errp) /* Resize and potentially preallocate the file to the desired * final size */ - result = raw_regular_truncate(fd, file_opts->size, file_opts->preallocation, - errp); + result = raw_regular_truncate(NULL, fd, file_opts->size, + file_opts->preallocation, errp); if (result < 0) { goto out_close; } -- 2.13.6 From: Fam Zheng <famz@redhat.com> In the beginning of the function, we initialize the local variable to 0, and in the body of the function, we check the assigned values and exit the loop immediately. So here it can never be non-zero. Reported-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Fam Zheng <famz@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_handle_l2meta(BlockDriverState *bs, while (l2meta != NULL) { QCowL2Meta *next; - if (!ret && link_l2) { + if (link_l2) { ret = qcow2_alloc_cluster_link_l2(bs, l2meta); if (ret) { goto out; -- 2.13.6 From: Fam Zheng <famz@redhat.com> in_flight and tracked requests need to be tracked in every layer during recursion. For now the only user is qemu-img convert where overlapping requests and IOThreads don't exist, therefore this change doesn't make much difference form user point of view, but it is incorrect as part of the API. Fix it. Reported-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Fam Zheng <famz@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/io.c | 59 ++++++++++++++++++++++++++++------------------------------- 1 file changed, 28 insertions(+), 31 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_copy_range_internal(BdrvChild *src, BdrvRequestFlags flags, bool recurse_src) { + BdrvTrackedRequest src_req, dst_req; + BlockDriverState *src_bs = src->bs; + BlockDriverState *dst_bs = dst->bs; int ret; if (!src || !dst || !src->bs || !dst->bs) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_copy_range_internal(BdrvChild *src, || src->bs->encrypted || dst->bs->encrypted) { return -ENOTSUP; } + bdrv_inc_in_flight(src_bs); + bdrv_inc_in_flight(dst_bs); + tracked_request_begin(&src_req, src_bs, src_offset, + bytes, BDRV_TRACKED_READ); + tracked_request_begin(&dst_req, dst_bs, dst_offset, + bytes, BDRV_TRACKED_WRITE); + + wait_serialising_requests(&src_req); + wait_serialising_requests(&dst_req); if (recurse_src) { - return src->bs->drv->bdrv_co_copy_range_from(src->bs, - src, src_offset, - dst, dst_offset, - bytes, flags); + ret = src->bs->drv->bdrv_co_copy_range_from(src->bs, + src, src_offset, + dst, dst_offset, + bytes, flags); } else { - return dst->bs->drv->bdrv_co_copy_range_to(dst->bs, - src, src_offset, - dst, dst_offset, - bytes, flags); + ret = dst->bs->drv->bdrv_co_copy_range_to(dst->bs, + src, src_offset, + dst, dst_offset, + bytes, flags); } + tracked_request_end(&src_req); + tracked_request_end(&dst_req); + bdrv_dec_in_flight(src_bs); + bdrv_dec_in_flight(dst_bs); + return ret; } /* Copy range from @src to @dst. @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range(BdrvChild *src, uint64_t src_offset, BdrvChild *dst, uint64_t dst_offset, uint64_t bytes, BdrvRequestFlags flags) { - BdrvTrackedRequest src_req, dst_req; - BlockDriverState *src_bs = src->bs; - BlockDriverState *dst_bs = dst->bs; - int ret; - - bdrv_inc_in_flight(src_bs); - bdrv_inc_in_flight(dst_bs); - tracked_request_begin(&src_req, src_bs, src_offset, - bytes, BDRV_TRACKED_READ); - tracked_request_begin(&dst_req, dst_bs, dst_offset, - bytes, BDRV_TRACKED_WRITE); - - wait_serialising_requests(&src_req); - wait_serialising_requests(&dst_req); - ret = bdrv_co_copy_range_from(src, src_offset, - dst, dst_offset, - bytes, flags); - - tracked_request_end(&src_req); - tracked_request_end(&dst_req); - bdrv_dec_in_flight(src_bs); - bdrv_dec_in_flight(dst_bs); - return ret; + return bdrv_co_copy_range_from(src, src_offset, + dst, dst_offset, + bytes, flags); } static void bdrv_parent_cb_resize(BlockDriverState *bs) -- 2.13.6 From: Markus Armbruster <armbru@redhat.com> block_crypto_open_opts_init() and block_crypto_create_opts_init() contain a virtual visit of QCryptoBlockOptions and QCryptoBlockCreateOptions less member "format", respectively. Change their callers to put member "format" in the QDict, so they can use the generated visitors for these types instead. Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/crypto.h | 8 ++--- block/crypto.c | 99 +++++++++------------------------------------------------- block/qcow.c | 5 ++- block/qcow2.c | 10 +++--- 4 files changed, 22 insertions(+), 100 deletions(-) diff --git a/block/crypto.h b/block/crypto.h index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.h +++ b/block/crypto.h @@ -XXX,XX +XXX,XX @@ } QCryptoBlockCreateOptions * -block_crypto_create_opts_init(QCryptoBlockFormat format, - QDict *opts, - Error **errp); +block_crypto_create_opts_init(QDict *opts, Error **errp); QCryptoBlockOpenOptions * -block_crypto_open_opts_init(QCryptoBlockFormat format, - QDict *opts, - Error **errp); +block_crypto_open_opts_init(QDict *opts, Error **errp); #endif /* BLOCK_CRYPTO_H__ */ diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ static QemuOptsList block_crypto_create_opts_luks = { QCryptoBlockOpenOptions * -block_crypto_open_opts_init(QCryptoBlockFormat format, - QDict *opts, - Error **errp) +block_crypto_open_opts_init(QDict *opts, Error **errp) { Visitor *v; - QCryptoBlockOpenOptions *ret = NULL; - Error *local_err = NULL; - - ret = g_new0(QCryptoBlockOpenOptions, 1); - ret->format = format; + QCryptoBlockOpenOptions *ret; - v = qobject_input_visitor_new_flat_confused(opts, &local_err); + v = qobject_input_visitor_new_flat_confused(opts, errp); if (!v) { - goto out; - } - - visit_start_struct(v, NULL, NULL, 0, &local_err); - if (local_err) { - goto out; - } - - switch (format) { - case Q_CRYPTO_BLOCK_FORMAT_LUKS: - visit_type_QCryptoBlockOptionsLUKS_members( - v, &ret->u.luks, &local_err); - break; - - case Q_CRYPTO_BLOCK_FORMAT_QCOW: - visit_type_QCryptoBlockOptionsQCow_members( - v, &ret->u.qcow, &local_err); - break; - - default: - error_setg(&local_err, "Unsupported block format %d", format); - break; - } - if (!local_err) { - visit_check_struct(v, &local_err); + return NULL; } - visit_end_struct(v, NULL); + visit_type_QCryptoBlockOpenOptions(v, NULL, &ret, errp); - out: - if (local_err) { - error_propagate(errp, local_err); - qapi_free_QCryptoBlockOpenOptions(ret); - ret = NULL; - } visit_free(v); return ret; } QCryptoBlockCreateOptions * -block_crypto_create_opts_init(QCryptoBlockFormat format, - QDict *opts, - Error **errp) +block_crypto_create_opts_init(QDict *opts, Error **errp) { Visitor *v; - QCryptoBlockCreateOptions *ret = NULL; - Error *local_err = NULL; - - ret = g_new0(QCryptoBlockCreateOptions, 1); - ret->format = format; + QCryptoBlockCreateOptions *ret; - v = qobject_input_visitor_new_flat_confused(opts, &local_err); + v = qobject_input_visitor_new_flat_confused(opts, errp); if (!v) { - goto out; - } - - visit_start_struct(v, NULL, NULL, 0, &local_err); - if (local_err) { - goto out; - } - - switch (format) { - case Q_CRYPTO_BLOCK_FORMAT_LUKS: - visit_type_QCryptoBlockCreateOptionsLUKS_members( - v, &ret->u.luks, &local_err); - break; - - case Q_CRYPTO_BLOCK_FORMAT_QCOW: - visit_type_QCryptoBlockOptionsQCow_members( - v, &ret->u.qcow, &local_err); - break; - - default: - error_setg(&local_err, "Unsupported block format %d", format); - break; - } - if (!local_err) { - visit_check_struct(v, &local_err); + return NULL; } - visit_end_struct(v, NULL); + visit_type_QCryptoBlockCreateOptions(v, NULL, &ret, errp); - out: - if (local_err) { - error_propagate(errp, local_err); - qapi_free_QCryptoBlockCreateOptions(ret); - ret = NULL; - } visit_free(v); return ret; } @@ -XXX,XX +XXX,XX @@ static int block_crypto_open_generic(QCryptoBlockFormat format, } cryptoopts = qemu_opts_to_qdict(opts, NULL); + qdict_put_str(cryptoopts, "format", QCryptoBlockFormat_str(format)); - open_opts = block_crypto_open_opts_init(format, cryptoopts, errp); + open_opts = block_crypto_open_opts_init(cryptoopts, errp); if (!open_opts) { goto cleanup; } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn block_crypto_co_create_opts_luks(const char *filename, &block_crypto_create_opts_luks, true); - create_opts = block_crypto_create_opts_init(Q_CRYPTO_BLOCK_FORMAT_LUKS, - cryptoopts, errp); + qdict_put_str(cryptoopts, "format", "luks"); + create_opts = block_crypto_create_opts_init(cryptoopts, errp); if (!create_opts) { ret = -EINVAL; goto fail; diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags, ret = -EINVAL; goto fail; } - qdict_del(encryptopts, "format"); - crypto_opts = block_crypto_open_opts_init( - Q_CRYPTO_BLOCK_FORMAT_QCOW, encryptopts, errp); + qdict_put_str(encryptopts, "format", "qcow"); + crypto_opts = block_crypto_open_opts_init(encryptopts, errp); if (!crypto_opts) { ret = -EINVAL; goto fail; diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int qcow2_update_options_prepare(BlockDriverState *bs, ret = -EINVAL; goto fail; } - qdict_del(encryptopts, "format"); - r->crypto_opts = block_crypto_open_opts_init( - Q_CRYPTO_BLOCK_FORMAT_QCOW, encryptopts, errp); + qdict_put_str(encryptopts, "format", "qcow"); + r->crypto_opts = block_crypto_open_opts_init(encryptopts, errp); break; case QCOW_CRYPT_LUKS: @@ -XXX,XX +XXX,XX @@ static int qcow2_update_options_prepare(BlockDriverState *bs, ret = -EINVAL; goto fail; } - qdict_del(encryptopts, "format"); - r->crypto_opts = block_crypto_open_opts_init( - Q_CRYPTO_BLOCK_FORMAT_LUKS, encryptopts, errp); + qdict_put_str(encryptopts, "format", "luks"); + r->crypto_opts = block_crypto_open_opts_init(encryptopts, errp); break; default: -- 2.13.6 Commit abf754fe406 updated 026.out, but forgot to also update 026.out.nocache. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> --- tests/qemu-iotests/026.out.nocache | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tests/qemu-iotests/026.out.nocache b/tests/qemu-iotests/026.out.nocache index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/026.out.nocache +++ b/tests/qemu-iotests/026.out.nocache @@ -XXX,XX +XXX,XX @@ Failed to flush the L2 table cache: No space left on device Failed to flush the refcount block cache: No space left on device write failed: No space left on device -11 leaked clusters were found on the image. +10 leaked clusters were found on the image. This means waste of disk space, but no harm to data. Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1073741824 @@ -XXX,XX +XXX,XX @@ Failed to flush the L2 table cache: No space left on device Failed to flush the refcount block cache: No space left on device write failed: No space left on device -11 leaked clusters were found on the image. +10 leaked clusters were found on the image. This means waste of disk space, but no harm to data. Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1073741824 @@ -XXX,XX +XXX,XX @@ Failed to flush the L2 table cache: No space left on device Failed to flush the refcount block cache: No space left on device write failed: No space left on device -11 leaked clusters were found on the image. +10 leaked clusters were found on the image. This means waste of disk space, but no harm to data. Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1073741824 -- 2.13.6 If we managed to allocate the clusters, but then failed to write the data, there's a good chance that we'll still be able to free the clusters again in order to avoid cluster leaks (the refcounts are cached, so even if we can't write them out right now, we may be able to do so when the VM is resumed after a werror=stop/enospc pause). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Tested-by: Eric Blake <eblake@redhat.com> --- block/qcow2.h | 1 + block/qcow2-cluster.c | 11 +++++++++++ block/qcow2.c | 2 ++ 3 files changed, 14 insertions(+) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs, int compressed_size); int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m); +void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m); int qcow2_cluster_discard(BlockDriverState *bs, uint64_t offset, uint64_t bytes, enum qcow2_discard_type type, bool full_discard); diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2-cluster.c +++ b/block/qcow2-cluster.c @@ -XXX,XX +XXX,XX @@ err: return ret; } +/** + * Frees the allocated clusters because the request failed and they won't + * actually be linked. + */ +void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m) +{ + BDRVQcow2State *s = bs->opaque; + qcow2_free_clusters(bs, m->alloc_offset, m->nb_clusters << s->cluster_bits, + QCOW2_DISCARD_NEVER); +} + /* * Returns the number of contiguous clusters that can be used for an allocating * write, but require COW to be performed (this includes yet unallocated space, diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_handle_l2meta(BlockDriverState *bs, if (ret) { goto out; } + } else { + qcow2_alloc_cluster_abort(bs, l2meta); } /* Take the request off the list of running requests */ -- 2.13.6 This adds a test for a temporary write failure, which simulates the situation after werror=stop/enospc has stopped the VM. We shouldn't leave leaked clusters behind in such cases. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Max Reitz <mreitz@redhat.com> --- tests/qemu-iotests/026 | 17 +++++++++++++++++ tests/qemu-iotests/026.out | 8 ++++++++ tests/qemu-iotests/026.out.nocache | 8 ++++++++ 3 files changed, 33 insertions(+) diff --git a/tests/qemu-iotests/026 b/tests/qemu-iotests/026 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/026 +++ b/tests/qemu-iotests/026 @@ -XXX,XX +XXX,XX @@ done done done +echo +echo === Avoid cluster leaks after temporary failure === +echo + +cat > "$TEST_DIR/blkdebug.conf" <<EOF +[inject-error] +event = "write_aio" +errno = "5" +once = "on" +EOF + +# After the failed first write, do a second write so that the updated refcount +# block is actually written back +_make_test_img 64M +$QEMU_IO -c "write 0 1M" -c "write 0 1M" "$BLKDBG_TEST_IMG" | _filter_qemu_io +_check_test_img + # success, all done echo "*** done" rm -f $seq.full diff --git a/tests/qemu-iotests/026.out b/tests/qemu-iotests/026.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/026.out +++ b/tests/qemu-iotests/026.out @@ -XXX,XX +XXX,XX @@ write failed: No space left on device 96 leaked clusters were found on the image. This means waste of disk space, but no harm to data. + +=== Avoid cluster leaks after temporary failure === + +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864 +write failed: Input/output error +wrote 1048576/1048576 bytes at offset 0 +1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +No errors were found on the image. *** done diff --git a/tests/qemu-iotests/026.out.nocache b/tests/qemu-iotests/026.out.nocache index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/026.out.nocache +++ b/tests/qemu-iotests/026.out.nocache @@ -XXX,XX +XXX,XX @@ write failed: No space left on device 96 leaked clusters were found on the image. This means waste of disk space, but no harm to data. + +=== Avoid cluster leaks after temporary failure === + +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864 +write failed: Input/output error +wrote 1048576/1048576 bytes at offset 0 +1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +No errors were found on the image. *** done -- 2.13.6 This simplifies file-posix by implementing the coroutine variants of the discard and flush BlockDriver callbacks. These were the last remaining users of paio_submit(), which can be removed now. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- block/file-posix.c | 72 ++++++++++++++++++------------------------------------ 1 file changed, 24 insertions(+), 48 deletions(-) diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ static inline int paio_submit_co(BlockDriverState *bs, int fd, return paio_submit_co_full(bs, fd, offset, -1, 0, qiov, bytes, type); } -static BlockAIOCB *paio_submit(BlockDriverState *bs, int fd, - int64_t offset, QEMUIOVector *qiov, int bytes, - BlockCompletionFunc *cb, void *opaque, int type) -{ - RawPosixAIOData *acb = g_new(RawPosixAIOData, 1); - ThreadPool *pool; - - acb->bs = bs; - acb->aio_type = type; - acb->aio_fildes = fd; - - acb->aio_nbytes = bytes; - acb->aio_offset = offset; - - if (qiov) { - acb->aio_iov = qiov->iov; - acb->aio_niov = qiov->niov; - assert(qiov->size == acb->aio_nbytes); - } - - trace_paio_submit(acb, opaque, offset, bytes, type); - pool = aio_get_thread_pool(bdrv_get_aio_context(bs)); - return thread_pool_submit_aio(pool, aio_worker, acb, cb, opaque); -} - static int coroutine_fn raw_co_prw(BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, int type) { @@ -XXX,XX +XXX,XX @@ static void raw_aio_unplug(BlockDriverState *bs) #endif } -static BlockAIOCB *raw_aio_flush(BlockDriverState *bs, - BlockCompletionFunc *cb, void *opaque) +static int raw_co_flush_to_disk(BlockDriverState *bs) { BDRVRawState *s = bs->opaque; + int ret; - if (fd_open(bs) < 0) - return NULL; + ret = fd_open(bs); + if (ret < 0) { + return ret; + } - return paio_submit(bs, s->fd, 0, NULL, 0, cb, opaque, QEMU_AIO_FLUSH); + return paio_submit_co(bs, s->fd, 0, NULL, 0, QEMU_AIO_FLUSH); } static void raw_aio_attach_aio_context(BlockDriverState *bs, @@ -XXX,XX +XXX,XX @@ static void coroutine_fn raw_co_invalidate_cache(BlockDriverState *bs, #endif /* !__linux__ */ } -static coroutine_fn BlockAIOCB *raw_aio_pdiscard(BlockDriverState *bs, - int64_t offset, int bytes, - BlockCompletionFunc *cb, void *opaque) +static coroutine_fn int +raw_co_pdiscard(BlockDriverState *bs, int64_t offset, int bytes) { BDRVRawState *s = bs->opaque; - return paio_submit(bs, s->fd, offset, NULL, bytes, - cb, opaque, QEMU_AIO_DISCARD); + return paio_submit_co(bs, s->fd, offset, NULL, bytes, QEMU_AIO_DISCARD); } static int coroutine_fn raw_co_pwrite_zeroes( @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_file = { .bdrv_co_preadv = raw_co_preadv, .bdrv_co_pwritev = raw_co_pwritev, - .bdrv_aio_flush = raw_aio_flush, - .bdrv_aio_pdiscard = raw_aio_pdiscard, + .bdrv_co_flush_to_disk = raw_co_flush_to_disk, + .bdrv_co_pdiscard = raw_co_pdiscard, .bdrv_co_copy_range_from = raw_co_copy_range_from, .bdrv_co_copy_range_to = raw_co_copy_range_to, .bdrv_refresh_limits = raw_refresh_limits, @@ -XXX,XX +XXX,XX @@ static int fd_open(BlockDriverState *bs) return -EIO; } -static coroutine_fn BlockAIOCB *hdev_aio_pdiscard(BlockDriverState *bs, - int64_t offset, int bytes, - BlockCompletionFunc *cb, void *opaque) +static coroutine_fn int +hdev_co_pdiscard(BlockDriverState *bs, int64_t offset, int bytes) { BDRVRawState *s = bs->opaque; + int ret; - if (fd_open(bs) < 0) { - return NULL; + ret = fd_open(bs); + if (ret < 0) { + return ret; } - return paio_submit(bs, s->fd, offset, NULL, bytes, - cb, opaque, QEMU_AIO_DISCARD|QEMU_AIO_BLKDEV); + return paio_submit_co(bs, s->fd, offset, NULL, bytes, + QEMU_AIO_DISCARD | QEMU_AIO_BLKDEV); } static coroutine_fn int hdev_co_pwrite_zeroes(BlockDriverState *bs, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_device = { .bdrv_co_preadv = raw_co_preadv, .bdrv_co_pwritev = raw_co_pwritev, - .bdrv_aio_flush = raw_aio_flush, - .bdrv_aio_pdiscard = hdev_aio_pdiscard, + .bdrv_co_flush_to_disk = raw_co_flush_to_disk, + .bdrv_co_pdiscard = hdev_co_pdiscard, .bdrv_co_copy_range_from = raw_co_copy_range_from, .bdrv_co_copy_range_to = raw_co_copy_range_to, .bdrv_refresh_limits = raw_refresh_limits, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_co_preadv = raw_co_preadv, .bdrv_co_pwritev = raw_co_pwritev, - .bdrv_aio_flush = raw_aio_flush, + .bdrv_co_flush_to_disk = raw_co_flush_to_disk, .bdrv_refresh_limits = raw_refresh_limits, .bdrv_io_plug = raw_aio_plug, .bdrv_io_unplug = raw_aio_unplug, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_co_preadv = raw_co_preadv, .bdrv_co_pwritev = raw_co_pwritev, - .bdrv_aio_flush = raw_aio_flush, + .bdrv_co_flush_to_disk = raw_co_flush_to_disk, .bdrv_refresh_limits = raw_refresh_limits, .bdrv_io_plug = raw_aio_plug, .bdrv_io_unplug = raw_aio_unplug, -- 2.13.6 From: Fam Zheng <famz@redhat.com> Not updating src_offset will result in wrong data being written to dst image. Reported-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Fam Zheng <famz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.c | 1 + tests/qemu-iotests/063 | 9 +++++++++ tests/qemu-iotests/063.out | 12 ++++++++++++ 3 files changed, 22 insertions(+) diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_to(BlockDriverState *bs, } bytes -= cur_bytes; + src_offset += cur_bytes; dst_offset += cur_bytes; } ret = 0; diff --git a/tests/qemu-iotests/063 b/tests/qemu-iotests/063 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/063 +++ b/tests/qemu-iotests/063 @@ -XXX,XX +XXX,XX @@ if $QEMU_IMG convert -f $IMGFMT -O $IMGFMT -n "$TEST_IMG.orig" "$TEST_IMG" >/dev exit 1 fi +echo "== Regression testing for copy offloading bug ==" + +_make_test_img 1M +TEST_IMG="$TEST_IMG.target" _make_test_img 1M +$QEMU_IO -c 'write -P 1 0 512k' -c 'write -P 2 512k 512k' "$TEST_IMG" | _filter_qemu_io +$QEMU_IO -c 'write -P 4 512k 512k' -c 'write -P 3 0 512k' "$TEST_IMG.target" | _filter_qemu_io +$QEMU_IMG convert -n -O $IMGFMT "$TEST_IMG" "$TEST_IMG.target" +$QEMU_IMG compare "$TEST_IMG" "$TEST_IMG.target" + echo "*** done" rm -f $seq.full status=0 diff --git a/tests/qemu-iotests/063.out b/tests/qemu-iotests/063.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/063.out +++ b/tests/qemu-iotests/063.out @@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=4194304 No errors were found on the image. == Testing conversion to a smaller file fails == Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2097152 +== Regression testing for copy offloading bug == +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576 +Formatting 'TEST_DIR/t.IMGFMT.target', fmt=IMGFMT size=1048576 +wrote 524288/524288 bytes at offset 0 +512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 524288/524288 bytes at offset 524288 +512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 524288/524288 bytes at offset 524288 +512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 524288/524288 bytes at offset 0 +512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +Images are identical. *** done -- 2.13.6 From: Fam Zheng <famz@redhat.com> Per SCSI definition the designator_length we receive from INQUIRY is 8, 12 or at most 16, but we should be careful because the remote iscsi target may misbehave, otherwise we could have a buffer overflow. Reported-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Fam Zheng <famz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/iscsi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block/iscsi.c b/block/iscsi.c index XXXXXXX..XXXXXXX 100644 --- a/block/iscsi.c +++ b/block/iscsi.c @@ -XXX,XX +XXX,XX @@ static void iscsi_populate_target_desc(unsigned char *desc, IscsiLun *lun) desc[5] = (dd->designator_type & 0xF) | ((dd->association & 3) << 4); desc[7] = dd->designator_length; - memcpy(desc + 8, dd->designator, dd->designator_length); + memcpy(desc + 8, dd->designator, MIN(dd->designator_length, 20)); desc[28] = 0; desc[29] = (lun->block_size >> 16) & 0xFF; -- 2.13.6 From: Fam Zheng <famz@redhat.com> EINTR should be checked against errno, not ret. While fixing the bug, collect the branches with a switch block. Also, change the return value from -ENOSTUP to -ENOSPC when the actual issue is request range passes EOF, which should be distinguishable from the case of error == ENOSYS by the caller, so that it could still retry with other byte ranges, whereas it shouldn't retry anymore upon ENOSYS. Signed-off-by: Fam Zheng <famz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ static ssize_t handle_aiocb_copy_range(RawPosixAIOData *aiocb) ssize_t ret = copy_file_range(aiocb->aio_fildes, &in_off, aiocb->aio_fd2, &out_off, bytes, 0); - if (ret == -EINTR) { - continue; + if (ret == 0) { + /* No progress (e.g. when beyond EOF), let the caller fall back to + * buffer I/O. */ + return -ENOSPC; } if (ret < 0) { - if (errno == ENOSYS) { + switch (errno) { + case ENOSYS: return -ENOTSUP; - } else { + case EINTR: + continue; + default: return -errno; } } - if (!ret) { - /* No progress (e.g. when beyond EOF), fall back to buffer I/O. */ - return -ENOTSUP; - } bytes -= ret; } return 0; -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the last few sector-based calls into the block layer from the parallels driver. Ideally, the parallels driver should switch to doing everything byte-based, but that's a more invasive change that requires a bit more auditing. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Denis V. Lunev <den@openvz.org> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/parallels.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/block/parallels.c b/block/parallels.c index XXXXXXX..XXXXXXX 100644 --- a/block/parallels.c +++ b/block/parallels.c @@ -XXX,XX +XXX,XX @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num, }; qemu_iovec_init_external(&qiov, &iov, 1); - ret = bdrv_co_readv(bs->backing, idx * s->tracks, nb_cow_sectors, - &qiov); + ret = bdrv_co_preadv(bs->backing, idx * s->tracks * BDRV_SECTOR_SIZE, + nb_cow_bytes, &qiov, 0); if (ret < 0) { qemu_vfree(iov.iov_base); return ret; } - ret = bdrv_co_writev(bs->file, s->data_end, nb_cow_sectors, &qiov); + ret = bdrv_co_pwritev(bs->file, s->data_end * BDRV_SECTOR_SIZE, + nb_cow_bytes, &qiov, 0); qemu_vfree(iov.iov_base); if (ret < 0) { return ret; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_writev(BlockDriverState *bs, qemu_iovec_reset(&hd_qiov); qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes); - ret = bdrv_co_writev(bs->file, position, n, &hd_qiov); + ret = bdrv_co_pwritev(bs->file, position * BDRV_SECTOR_SIZE, nbytes, + &hd_qiov, 0); if (ret < 0) { break; } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs, if (position < 0) { if (bs->backing) { - ret = bdrv_co_readv(bs->backing, sector_num, n, &hd_qiov); + ret = bdrv_co_preadv(bs->backing, sector_num * BDRV_SECTOR_SIZE, + nbytes, &hd_qiov, 0); if (ret < 0) { break; } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs, qemu_iovec_memset(&hd_qiov, 0, 0, nbytes); } } else { - ret = bdrv_co_readv(bs->file, position, n, &hd_qiov); + ret = bdrv_co_preadv(bs->file, position * BDRV_SECTOR_SIZE, nbytes, + &hd_qiov, 0); if (ret < 0) { break; } -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the internal helper function get_cluster_offset(), by changing n_start and n_end to be byte offsets rather than sector indices within the cluster being allocated. However, assert that these values are still sector-aligned (at least qcrypto_block_encrypt() still wants that). For now we get that alignment for free because we still use sector-based driver callbacks. A later patch will then switch the qcow driver as a whole over to byte-based operation; but will still leave things at sector alignments as it is not worth auditing the qcow image format to worry about sub-sector requests. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow.c | 29 +++++++++++++++-------------- 1 file changed, 15 insertions(+), 14 deletions(-) diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static int qcow_reopen_prepare(BDRVReopenState *state, * * 0 to not allocate. * - * 1 to allocate a normal cluster (for sector indexes 'n_start' to - * 'n_end') + * 1 to allocate a normal cluster (for sector-aligned byte offsets 'n_start' + * to 'n_end' within the cluster) * * 2 to allocate a compressed cluster of size * 'compressed_size'. 'compressed_size' must be > 0 and < @@ -XXX,XX +XXX,XX @@ static int get_cluster_offset(BlockDriverState *bs, if (!allocate) return 0; BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC); + assert(QEMU_IS_ALIGNED(n_start | n_end, BDRV_SECTOR_SIZE)); /* allocate a new cluster */ if ((cluster_offset & QCOW_OFLAG_COMPRESSED) && - (n_end - n_start) < s->cluster_sectors) { + (n_end - n_start) < s->cluster_size) { /* if the cluster is already compressed, we must decompress it in the case it is not completely overwritten */ @@ -XXX,XX +XXX,XX @@ static int get_cluster_offset(BlockDriverState *bs, /* if encrypted, we must initialize the cluster content which won't be written */ if (bs->encrypted && - (n_end - n_start) < s->cluster_sectors) { - uint64_t start_sect; + (n_end - n_start) < s->cluster_size) { + uint64_t start_offset; assert(s->crypto); - start_sect = (offset & ~(s->cluster_size - 1)) >> 9; - for(i = 0; i < s->cluster_sectors; i++) { + start_offset = offset & ~(s->cluster_size - 1); + for (i = 0; i < s->cluster_size; i += BDRV_SECTOR_SIZE) { if (i < n_start || i >= n_end) { - memset(s->cluster_data, 0x00, 512); + memset(s->cluster_data, 0x00, BDRV_SECTOR_SIZE); if (qcrypto_block_encrypt(s->crypto, - (start_sect + i) * - BDRV_SECTOR_SIZE, + start_offset + i, s->cluster_data, BDRV_SECTOR_SIZE, NULL) < 0) { @@ -XXX,XX +XXX,XX @@ static int get_cluster_offset(BlockDriverState *bs, } BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO); ret = bdrv_pwrite(bs->file, - cluster_offset + i * 512, - s->cluster_data, 512); + cluster_offset + i, + s->cluster_data, + BDRV_SECTOR_SIZE); if (ret < 0) { return ret; } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, n = nb_sectors; } ret = get_cluster_offset(bs, sector_num << 9, 1, 0, - index_in_cluster, - index_in_cluster + n, &cluster_offset); + index_in_cluster << 9, + (index_in_cluster + n) << 9, &cluster_offset); if (ret < 0) { break; } -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the internals of the qcow driver read function, by iterating over offset/bytes instead of sector_num/nb_sectors, and with a rename of index_in_cluster and repurposing of n to track bytes instead of sectors. A later patch will then switch the qcow driver as a whole over to byte-based operation. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow.c | 42 ++++++++++++++++++++---------------------- 1 file changed, 20 insertions(+), 22 deletions(-) diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, int nb_sectors, QEMUIOVector *qiov) { BDRVQcowState *s = bs->opaque; - int index_in_cluster; + int offset_in_cluster; int ret = 0, n; uint64_t cluster_offset; struct iovec hd_iov; QEMUIOVector hd_qiov; uint8_t *buf; void *orig_buf; + int64_t offset = sector_num * BDRV_SECTOR_SIZE; + int64_t bytes = nb_sectors * BDRV_SECTOR_SIZE; if (qiov->niov > 1) { buf = orig_buf = qemu_try_blockalign(bs, qiov->size); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, qemu_co_mutex_lock(&s->lock); - while (nb_sectors != 0) { + while (bytes != 0) { /* prepare next request */ - ret = get_cluster_offset(bs, sector_num << 9, - 0, 0, 0, 0, &cluster_offset); + ret = get_cluster_offset(bs, offset, 0, 0, 0, 0, &cluster_offset); if (ret < 0) { break; } - index_in_cluster = sector_num & (s->cluster_sectors - 1); - n = s->cluster_sectors - index_in_cluster; - if (n > nb_sectors) { - n = nb_sectors; + offset_in_cluster = offset & (s->cluster_size - 1); + n = s->cluster_size - offset_in_cluster; + if (n > bytes) { + n = bytes; } if (!cluster_offset) { if (bs->backing) { /* read from the base image */ hd_iov.iov_base = (void *)buf; - hd_iov.iov_len = n * 512; + hd_iov.iov_len = n; qemu_iovec_init_external(&hd_qiov, &hd_iov, 1); qemu_co_mutex_unlock(&s->lock); /* qcow2 emits this on bs->file instead of bs->backing */ BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO); - ret = bdrv_co_readv(bs->backing, sector_num, n, &hd_qiov); + ret = bdrv_co_preadv(bs->backing, offset, n, &hd_qiov, 0); qemu_co_mutex_lock(&s->lock); if (ret < 0) { break; } } else { /* Note: in this case, no need to wait */ - memset(buf, 0, 512 * n); + memset(buf, 0, n); } } else if (cluster_offset & QCOW_OFLAG_COMPRESSED) { /* add AIO support for compressed blocks ? */ @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, ret = -EIO; break; } - memcpy(buf, - s->cluster_cache + index_in_cluster * 512, 512 * n); + memcpy(buf, s->cluster_cache + offset_in_cluster, n); } else { if ((cluster_offset & 511) != 0) { ret = -EIO; break; } hd_iov.iov_base = (void *)buf; - hd_iov.iov_len = n * 512; + hd_iov.iov_len = n; qemu_iovec_init_external(&hd_qiov, &hd_iov, 1); qemu_co_mutex_unlock(&s->lock); BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO); - ret = bdrv_co_readv(bs->file, - (cluster_offset >> 9) + index_in_cluster, - n, &hd_qiov); + ret = bdrv_co_preadv(bs->file, cluster_offset + offset_in_cluster, + n, &hd_qiov, 0); qemu_co_mutex_lock(&s->lock); if (ret < 0) { break; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, if (bs->encrypted) { assert(s->crypto); if (qcrypto_block_decrypt(s->crypto, - sector_num * BDRV_SECTOR_SIZE, buf, - n * BDRV_SECTOR_SIZE, NULL) < 0) { + offset, buf, n, NULL) < 0) { ret = -EIO; break; } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, } ret = 0; - nb_sectors -= n; - sector_num += n; - buf += n * 512; + bytes -= n; + offset += n; + buf += n; } qemu_co_mutex_unlock(&s->lock); -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the internals of the qcow driver write function, by iterating over offset/bytes instead of sector_num/nb_sectors, and with a rename of index_in_cluster and repurposing of n to track bytes instead of sectors. A later patch will then switch the qcow driver as a whole over to byte-based operation. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow.c | 36 +++++++++++++++++------------------- 1 file changed, 17 insertions(+), 19 deletions(-) diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, int flags) { BDRVQcowState *s = bs->opaque; - int index_in_cluster; + int offset_in_cluster; uint64_t cluster_offset; int ret = 0, n; struct iovec hd_iov; QEMUIOVector hd_qiov; uint8_t *buf; void *orig_buf; + int64_t offset = sector_num * BDRV_SECTOR_SIZE; + int64_t bytes = nb_sectors * BDRV_SECTOR_SIZE; assert(!flags); s->cluster_cache_offset = -1; /* disable compressed cache */ @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, qemu_co_mutex_lock(&s->lock); - while (nb_sectors != 0) { - - index_in_cluster = sector_num & (s->cluster_sectors - 1); - n = s->cluster_sectors - index_in_cluster; - if (n > nb_sectors) { - n = nb_sectors; + while (bytes != 0) { + offset_in_cluster = offset & (s->cluster_size - 1); + n = s->cluster_size - offset_in_cluster; + if (n > bytes) { + n = bytes; } - ret = get_cluster_offset(bs, sector_num << 9, 1, 0, - index_in_cluster << 9, - (index_in_cluster + n) << 9, &cluster_offset); + ret = get_cluster_offset(bs, offset, 1, 0, offset_in_cluster, + offset_in_cluster + n, &cluster_offset); if (ret < 0) { break; } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, } if (bs->encrypted) { assert(s->crypto); - if (qcrypto_block_encrypt(s->crypto, sector_num * BDRV_SECTOR_SIZE, - buf, n * BDRV_SECTOR_SIZE, NULL) < 0) { + if (qcrypto_block_encrypt(s->crypto, offset, buf, n, NULL) < 0) { ret = -EIO; break; } } hd_iov.iov_base = (void *)buf; - hd_iov.iov_len = n * 512; + hd_iov.iov_len = n; qemu_iovec_init_external(&hd_qiov, &hd_iov, 1); qemu_co_mutex_unlock(&s->lock); BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO); - ret = bdrv_co_writev(bs->file, - (cluster_offset >> 9) + index_in_cluster, - n, &hd_qiov); + ret = bdrv_co_pwritev(bs->file, cluster_offset + offset_in_cluster, + n, &hd_qiov, 0); qemu_co_mutex_lock(&s->lock); if (ret < 0) { break; } ret = 0; - nb_sectors -= n; - sector_num += n; - buf += n * 512; + bytes -= n; + offset += n; + buf += n; } qemu_co_mutex_unlock(&s->lock); -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. The qcow driver is now ready to fully utilize the byte-based callback interface, as long as we override the default alignment to still be 512 (needed at least for asserts present because of encryption, but easier to do everywhere than to audit which sub-sector requests are handled correctly, especially since we no longer recommend qcow for new disk images). Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow.c | 35 ++++++++++++++++++++--------------- 1 file changed, 20 insertions(+), 15 deletions(-) diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ typedef struct QCowHeader { typedef struct BDRVQcowState { int cluster_bits; int cluster_size; - int cluster_sectors; int l2_bits; int l2_size; unsigned int l1_size; @@ -XXX,XX +XXX,XX @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags, } s->cluster_bits = header.cluster_bits; s->cluster_size = 1 << s->cluster_bits; - s->cluster_sectors = 1 << (s->cluster_bits - 9); s->l2_bits = header.l2_bits; s->l2_size = 1 << s->l2_bits; bs->total_sectors = header.size / 512; @@ -XXX,XX +XXX,XX @@ static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset) return 0; } -static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov) +static void qcow_refresh_limits(BlockDriverState *bs, Error **errp) +{ + /* At least encrypted images require 512-byte alignment. Apply the + * limit universally, rather than just on encrypted images, as + * it's easier to let the block layer handle rounding than to + * audit this code further. */ + bs->bl.request_alignment = BDRV_SECTOR_SIZE; +} + +static coroutine_fn int qcow_co_preadv(BlockDriverState *bs, uint64_t offset, + uint64_t bytes, QEMUIOVector *qiov, + int flags) { BDRVQcowState *s = bs->opaque; int offset_in_cluster; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, QEMUIOVector hd_qiov; uint8_t *buf; void *orig_buf; - int64_t offset = sector_num * BDRV_SECTOR_SIZE; - int64_t bytes = nb_sectors * BDRV_SECTOR_SIZE; + assert(!flags); if (qiov->niov > 1) { buf = orig_buf = qemu_try_blockalign(bs, qiov->size); if (buf == NULL) { @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num, return ret; } -static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov, - int flags) +static coroutine_fn int qcow_co_pwritev(BlockDriverState *bs, uint64_t offset, + uint64_t bytes, QEMUIOVector *qiov, + int flags) { BDRVQcowState *s = bs->opaque; int offset_in_cluster; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector hd_qiov; uint8_t *buf; void *orig_buf; - int64_t offset = sector_num * BDRV_SECTOR_SIZE; - int64_t bytes = nb_sectors * BDRV_SECTOR_SIZE; assert(!flags); s->cluster_cache_offset = -1; /* disable compressed cache */ @@ -XXX,XX +XXX,XX @@ qcow_co_pwritev_compressed(BlockDriverState *bs, uint64_t offset, if (ret != Z_STREAM_END || out_len >= s->cluster_size) { /* could not compress: write normal cluster */ - ret = qcow_co_writev(bs, offset >> BDRV_SECTOR_BITS, - bytes >> BDRV_SECTOR_BITS, qiov, 0); + ret = qcow_co_pwritev(bs, offset, bytes, qiov, 0); if (ret < 0) { goto fail; } @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_qcow = { .bdrv_co_create_opts = qcow_co_create_opts, .bdrv_has_zero_init = bdrv_has_zero_init_1, .supports_backing = true, + .bdrv_refresh_limits = qcow_refresh_limits, - .bdrv_co_readv = qcow_co_readv, - .bdrv_co_writev = qcow_co_writev, + .bdrv_co_preadv = qcow_co_preadv, + .bdrv_co_pwritev = qcow_co_pwritev, .bdrv_co_block_status = qcow_co_block_status, .bdrv_make_empty = qcow_make_empty, -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the last few sector-based calls into the block layer from the replication driver. Ideally, the replication driver should switch to doing everything byte-based, but that's a more invasive change that requires a bit more auditing. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/replication.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/block/replication.c b/block/replication.c index XXXXXXX..XXXXXXX 100644 --- a/block/replication.c +++ b/block/replication.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int replication_co_readv(BlockDriverState *bs, backup_cow_request_begin(&req, child->bs->job, sector_num * BDRV_SECTOR_SIZE, remaining_bytes); - ret = bdrv_co_readv(bs->file, sector_num, remaining_sectors, - qiov); + ret = bdrv_co_preadv(bs->file, sector_num * BDRV_SECTOR_SIZE, + remaining_bytes, qiov, 0); backup_cow_request_end(&req); goto out; } - ret = bdrv_co_readv(bs->file, sector_num, remaining_sectors, qiov); + ret = bdrv_co_preadv(bs->file, sector_num * BDRV_SECTOR_SIZE, + remaining_sectors * BDRV_SECTOR_SIZE, qiov, 0); out: return replication_return_value(s, ret); } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int replication_co_writev(BlockDriverState *bs, } if (ret == 0) { - ret = bdrv_co_writev(top, sector_num, - remaining_sectors, qiov); + ret = bdrv_co_pwritev(top, sector_num * BDRV_SECTOR_SIZE, + remaining_sectors * BDRV_SECTOR_SIZE, qiov, 0); return replication_return_value(s, ret); } @@ -XXX,XX +XXX,XX @@ static coroutine_fn int replication_co_writev(BlockDriverState *bs, qemu_iovec_concat(&hd_qiov, qiov, bytes_done, count); target = ret ? top : base; - ret = bdrv_co_writev(target, sector_num, n, &hd_qiov); + ret = bdrv_co_pwritev(target, sector_num * BDRV_SECTOR_SIZE, + n * BDRV_SECTOR_SIZE, &hd_qiov, 0); if (ret < 0) { goto out1; } -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Make the change for the last few sector-based calls into the block layer from the vhdx driver. Ideally, the vhdx driver should switch to doing everything byte-based, but that's a more invasive change that requires a bit more auditing. Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/vhdx.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/block/vhdx.c b/block/vhdx.c index XXXXXXX..XXXXXXX 100644 --- a/block/vhdx.c +++ b/block/vhdx.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int vhdx_co_readv(BlockDriverState *bs, int64_t sector_num, break; case PAYLOAD_BLOCK_FULLY_PRESENT: qemu_co_mutex_unlock(&s->lock); - ret = bdrv_co_readv(bs->file, - sinfo.file_offset >> BDRV_SECTOR_BITS, - sinfo.sectors_avail, &hd_qiov); + ret = bdrv_co_preadv(bs->file, sinfo.file_offset, + sinfo.sectors_avail * BDRV_SECTOR_SIZE, + &hd_qiov, 0); qemu_co_mutex_lock(&s->lock); if (ret < 0) { goto exit; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int vhdx_co_writev(BlockDriverState *bs, int64_t sector_num, } /* block exists, so we can just overwrite it */ qemu_co_mutex_unlock(&s->lock); - ret = bdrv_co_writev(bs->file, - sinfo.file_offset >> BDRV_SECTOR_BITS, - sectors_to_write, &hd_qiov); + ret = bdrv_co_pwritev(bs->file, sinfo.file_offset, + sectors_to_write * BDRV_SECTOR_SIZE, + &hd_qiov, 0); qemu_co_mutex_lock(&s->lock); if (ret < 0) { goto error_bat_restore; -- 2.13.6 From: Eric Blake <eblake@redhat.com> We are gradually moving away from sector-based interfaces, towards byte-based. Now that all callers of vectored I/O have been converted to use our preferred byte-based bdrv_co_p{read,write}v(), we can delete the unused bdrv_co_{read,write}v(). Furthermore, this gets rid of the signature difference between the public bdrv_co_writev() and the callback .bdrv_co_writev (the latter still exists, because some drivers still need more work before they are fully byte-based). Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Jeff Cody <jcody@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 4 ---- block/io.c | 36 ------------------------------------ 2 files changed, 40 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -XXX,XX +XXX,XX @@ int bdrv_pwrite(BdrvChild *child, int64_t offset, const void *buf, int bytes); int bdrv_pwritev(BdrvChild *child, int64_t offset, QEMUIOVector *qiov); int bdrv_pwrite_sync(BdrvChild *child, int64_t offset, const void *buf, int count); -int coroutine_fn bdrv_co_readv(BdrvChild *child, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov); -int coroutine_fn bdrv_co_writev(BdrvChild *child, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov); /* * Efficiently zero a region of the disk image. Note that this is a regular * I/O request like read or write and should have a reasonable size. This diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_preadv(BdrvChild *child, return ret; } -static int coroutine_fn bdrv_co_do_readv(BdrvChild *child, - int64_t sector_num, int nb_sectors, QEMUIOVector *qiov, - BdrvRequestFlags flags) -{ - if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) { - return -EINVAL; - } - - return bdrv_co_preadv(child, sector_num << BDRV_SECTOR_BITS, - nb_sectors << BDRV_SECTOR_BITS, qiov, flags); -} - -int coroutine_fn bdrv_co_readv(BdrvChild *child, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov) -{ - return bdrv_co_do_readv(child, sector_num, nb_sectors, qiov, 0); -} - static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int bytes, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn bdrv_co_do_writev(BdrvChild *child, - int64_t sector_num, int nb_sectors, QEMUIOVector *qiov, - BdrvRequestFlags flags) -{ - if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) { - return -EINVAL; - } - - return bdrv_co_pwritev(child, sector_num << BDRV_SECTOR_BITS, - nb_sectors << BDRV_SECTOR_BITS, qiov, flags); -} - -int coroutine_fn bdrv_co_writev(BdrvChild *child, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov) -{ - return bdrv_co_do_writev(child, sector_num, nb_sectors, qiov, 0); -} - int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, int bytes, BdrvRequestFlags flags) { -- 2.13.6 The following changes since commit 281f327487c9c9b1599f93c589a408bbf4a651b8: Merge remote-tracking branch 'remotes/vivier/tags/m68k-for-2.12-pull-request' into staging (2017-12-22 00:11:36 +0000) are available in the git repository at: git://repo.or.cz/qemu/kevin.git tags/for-upstream for you to fetch changes up to 1a63a907507fbbcfaee3f622907ec244b7eabda8: block: Keep nodes drained between reopen_queue/multiple (2017-12-22 15:05:32 +0100) ---------------------------------------------------------------- Block layer patches ---------------------------------------------------------------- Doug Gale (1): nvme: Add tracing Edgar Kaziakhmedov (1): qcow2: get rid of qcow2_backing_read1 routine Fam Zheng (2): block: Open backing image in force share mode for size probe block: Remove unused bdrv_requests_pending John Snow (1): iotests: fix 197 for vpc Kevin Wolf (27): block: Formats don't need CONSISTENT_READ with NO_IO block: Make bdrv_drain_invoke() recursive block: Call .drain_begin only once in bdrv_drain_all_begin() test-bdrv-drain: Test BlockDriver callbacks for drain block: bdrv_drain_recurse(): Remove unused begin parameter block: Don't wait for requests in bdrv_drain*_end() block: Unify order in drain functions block: Don't acquire AioContext in hmp_qemu_io() block: Document that x-blockdev-change breaks quorum children list block: Assert drain_all is only called from main AioContext block: Make bdrv_drain() driver callbacks non-recursive test-bdrv-drain: Test callback for bdrv_drain test-bdrv-drain: Test bs->quiesce_counter blockjob: Pause job on draining any job BDS test-bdrv-drain: Test drain vs. block jobs block: Don't block_job_pause_all() in bdrv_drain_all() block: Nested drain_end must still call callbacks test-bdrv-drain: Test nested drain sections block: Don't notify parents in drain call chain block: Add bdrv_subtree_drained_begin/end() test-bdrv-drain: Tests for bdrv_subtree_drain test-bdrv-drain: Test behaviour in coroutine context test-bdrv-drain: Recursive draining with multiple parents block: Allow graph changes in subtree drained section test-bdrv-drain: Test graph changes in drained section commit: Simplify reopen of base block: Keep nodes drained between reopen_queue/multiple Thomas Huth (3): block: Remove the obsolete -drive boot=on|off parameter block: Remove the deprecated -hdachs option block: Mention -drive cyls/heads/secs/trans/serial/addr in deprecation chapter qapi/block-core.json | 4 + block/qcow2.h | 3 - include/block/block.h | 15 +- include/block/block_int.h | 6 +- block.c | 75 ++++- block/commit.c | 8 +- block/io.c | 164 +++++++--- block/qcow2.c | 51 +-- block/replication.c | 6 + blockdev.c | 11 - blockjob.c | 22 +- hmp.c | 6 - hw/block/nvme.c | 349 +++++++++++++++++---- qemu-io-cmds.c | 3 + tests/test-bdrv-drain.c | 651 +++++++++++++++++++++++++++++++++++++++ vl.c | 86 +----- hw/block/trace-events | 93 ++++++ qemu-doc.texi | 29 +- qemu-options.hx | 19 +- tests/Makefile.include | 2 + tests/qemu-iotests/197 | 4 + tests/qemu-iotests/common.filter | 3 +- 22 files changed, 1294 insertions(+), 316 deletions(-) create mode 100644 tests/test-bdrv-drain.c Commit 1f4ad7d fixed 'qemu-img info' for raw images that are currently in use as a mirror target. It is not enough for image formats, though, as these still unconditionally request BLK_PERM_CONSISTENT_READ. As this permission is geared towards whether the guest-visible data is consistent, and has no impact on whether the metadata is sane, and 'qemu-img info' does not read guest-visible data (except for the raw format), it makes sense to not require BLK_PERM_CONSISTENT_READ if there is not going to be any guest I/O performed, regardless of image format. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ void bdrv_format_default_perms(BlockDriverState *bs, BdrvChild *c, assert(role == &child_backing || role == &child_file); if (!backing) { + int flags = bdrv_reopen_get_flags(reopen_queue, bs); + /* Apart from the modifications below, the same permissions are * forwarded and left alone as for filters */ bdrv_filter_default_perms(bs, c, role, reopen_queue, perm, shared, @@ -XXX,XX +XXX,XX @@ void bdrv_format_default_perms(BlockDriverState *bs, BdrvChild *c, /* bs->file always needs to be consistent because of the metadata. We * can never allow other users to resize or write to it. */ - perm |= BLK_PERM_CONSISTENT_READ; + if (!(flags & BDRV_O_NO_IO)) { + perm |= BLK_PERM_CONSISTENT_READ; + } shared &= ~(BLK_PERM_WRITE | BLK_PERM_RESIZE); } else { /* We want consistent read from backing files if the parent needs it. -- 2.13.6 From: John Snow <jsnow@redhat.com> VPC has some difficulty creating geometries of particular size. However, we can indeed force it to use a literal one, so let's do that for the sake of test 197, which is testing some specific offsets. Signed-off-by: John Snow <jsnow@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Lukáš Doktor <ldoktor@redhat.com> --- tests/qemu-iotests/197 | 4 ++++ tests/qemu-iotests/common.filter | 3 ++- 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/tests/qemu-iotests/197 b/tests/qemu-iotests/197 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/197 +++ b/tests/qemu-iotests/197 @@ -XXX,XX +XXX,XX @@ echo '=== Copy-on-read ===' echo # Prep the images +# VPC rounds image sizes to a specific geometry, force a specific size. +if [ "$IMGFMT" = "vpc" ]; then + IMGOPTS=$(_optstr_add "$IMGOPTS" "force_size") +fi _make_test_img 4G $QEMU_IO -c "write -P 55 3G 1k" "$TEST_IMG" | _filter_qemu_io IMGPROTO=file IMGFMT=qcow2 IMGOPTS= TEST_IMG_FILE="$TEST_WRAP" \ diff --git a/tests/qemu-iotests/common.filter b/tests/qemu-iotests/common.filter index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/common.filter +++ b/tests/qemu-iotests/common.filter @@ -XXX,XX +XXX,XX @@ _filter_img_create() -e "s# log_size=[0-9]\\+##g" \ -e "s# refcount_bits=[0-9]\\+##g" \ -e "s# key-secret=[a-zA-Z0-9]\\+##g" \ - -e "s# iter-time=[0-9]\\+##g" + -e "s# iter-time=[0-9]\\+##g" \ + -e "s# force_size=\\(on\\|off\\)##g" } _filter_img_info() -- 2.13.6 This change separates bdrv_drain_invoke(), which calls the BlockDriver drain callbacks, from bdrv_drain_recurse(). Instead, the function performs its own recursion now. One reason for this is that bdrv_drain_recurse() can be called multiple times by bdrv_drain_all_begin(), but the callbacks may only be called once. The separation is necessary to fix this bug. The other reason is that we intend to go to a model where we call all driver callbacks first, and only then start polling. This is not fully achieved yet with this patch, as bdrv_drain_invoke() contains a BDRV_POLL_WHILE() loop for the block driver callbacks, which can still call callbacks for any unrelated event. It's a step in this direction anyway. Cc: qemu-stable@nongnu.org Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/io.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_drain_invoke_entry(void *opaque) bdrv_wakeup(bs); } +/* Recursively call BlockDriver.bdrv_co_drain_begin/end callbacks */ static void bdrv_drain_invoke(BlockDriverState *bs, bool begin) { + BdrvChild *child, *tmp; BdrvCoDrainData data = { .bs = bs, .done = false, .begin = begin}; if (!bs->drv || (begin && !bs->drv->bdrv_co_drain_begin) || @@ -XXX,XX +XXX,XX @@ static void bdrv_drain_invoke(BlockDriverState *bs, bool begin) data.co = qemu_coroutine_create(bdrv_drain_invoke_entry, &data); bdrv_coroutine_enter(bs, data.co); BDRV_POLL_WHILE(bs, !data.done); + + QLIST_FOREACH_SAFE(child, &bs->children, next, tmp) { + bdrv_drain_invoke(child->bs, begin); + } } static bool bdrv_drain_recurse(BlockDriverState *bs, bool begin) @@ -XXX,XX +XXX,XX @@ static bool bdrv_drain_recurse(BlockDriverState *bs, bool begin) BdrvChild *child, *tmp; bool waited; - /* Ensure any pending metadata writes are submitted to bs->file. */ - bdrv_drain_invoke(bs, begin); - /* Wait for drained requests to finish */ waited = BDRV_POLL_WHILE(bs, atomic_read(&bs->in_flight) > 0); @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs) bdrv_parent_drained_begin(bs); } + bdrv_drain_invoke(bs, true); bdrv_drain_recurse(bs, true); } @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) } bdrv_parent_drained_end(bs); + bdrv_drain_invoke(bs, false); bdrv_drain_recurse(bs, false); aio_enable_external(bdrv_get_aio_context(bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) aio_context_acquire(aio_context); for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { if (aio_context == bdrv_get_aio_context(bs)) { + /* FIXME Calling this multiple times is wrong */ + bdrv_drain_invoke(bs, true); waited |= bdrv_drain_recurse(bs, true); } } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) aio_context_acquire(aio_context); aio_enable_external(aio_context); bdrv_parent_drained_end(bs); + bdrv_drain_invoke(bs, false); bdrv_drain_recurse(bs, false); aio_context_release(aio_context); } -- 2.13.6 bdrv_drain_all_begin() used to call the .bdrv_co_drain_begin() driver callback inside its polling loop. This means that how many times it got called for each node depended on long it had to poll the event loop. This is obviously not right and results in nodes that stay drained even after bdrv_drain_all_end(), which calls .bdrv_co_drain_begin() once per node. Fix bdrv_drain_all_begin() to call the callback only once, too. Cc: qemu-stable@nongnu.org Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/io.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) aio_context_acquire(aio_context); bdrv_parent_drained_begin(bs); aio_disable_external(aio_context); + bdrv_drain_invoke(bs, true); aio_context_release(aio_context); if (!g_slist_find(aio_ctxs, aio_context)) { @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) aio_context_acquire(aio_context); for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { if (aio_context == bdrv_get_aio_context(bs)) { - /* FIXME Calling this multiple times is wrong */ - bdrv_drain_invoke(bs, true); waited |= bdrv_drain_recurse(bs, true); } } -- 2.13.6 This adds a test case that the BlockDriver callbacks for drain are called in bdrv_drained_all_begin/end(), and that both of them are called exactly once. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- tests/test-bdrv-drain.c | 137 ++++++++++++++++++++++++++++++++++++++++++++++++ tests/Makefile.include | 2 + 2 files changed, 139 insertions(+) create mode 100644 tests/test-bdrv-drain.c diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ +/* + * Block node draining tests + * + * Copyright (c) 2017 Kevin Wolf <kwolf@redhat.com> + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "qemu/osdep.h" +#include "block/block.h" +#include "sysemu/block-backend.h" +#include "qapi/error.h" + +typedef struct BDRVTestState { + int drain_count; +} BDRVTestState; + +static void coroutine_fn bdrv_test_co_drain_begin(BlockDriverState *bs) +{ + BDRVTestState *s = bs->opaque; + s->drain_count++; +} + +static void coroutine_fn bdrv_test_co_drain_end(BlockDriverState *bs) +{ + BDRVTestState *s = bs->opaque; + s->drain_count--; +} + +static void bdrv_test_close(BlockDriverState *bs) +{ + BDRVTestState *s = bs->opaque; + g_assert_cmpint(s->drain_count, >, 0); +} + +static int coroutine_fn bdrv_test_co_preadv(BlockDriverState *bs, + uint64_t offset, uint64_t bytes, + QEMUIOVector *qiov, int flags) +{ + /* We want this request to stay until the polling loop in drain waits for + * it to complete. We need to sleep a while as bdrv_drain_invoke() comes + * first and polls its result, too, but it shouldn't accidentally complete + * this request yet. */ + qemu_co_sleep_ns(QEMU_CLOCK_REALTIME, 100000); + + return 0; +} + +static BlockDriver bdrv_test = { + .format_name = "test", + .instance_size = sizeof(BDRVTestState), + + .bdrv_close = bdrv_test_close, + .bdrv_co_preadv = bdrv_test_co_preadv, + + .bdrv_co_drain_begin = bdrv_test_co_drain_begin, + .bdrv_co_drain_end = bdrv_test_co_drain_end, +}; + +static void aio_ret_cb(void *opaque, int ret) +{ + int *aio_ret = opaque; + *aio_ret = ret; +} + +static void test_drv_cb_drain_all(void) +{ + BlockBackend *blk; + BlockDriverState *bs; + BDRVTestState *s; + BlockAIOCB *acb; + int aio_ret; + + QEMUIOVector qiov; + struct iovec iov = { + .iov_base = NULL, + .iov_len = 0, + }; + qemu_iovec_init_external(&qiov, &iov, 1); + + blk = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs = bdrv_new_open_driver(&bdrv_test, "test-node", BDRV_O_RDWR, + &error_abort); + s = bs->opaque; + blk_insert_bs(blk, bs, &error_abort); + + /* Simple bdrv_drain_all_begin/end pair, check that CBs are called */ + g_assert_cmpint(s->drain_count, ==, 0); + bdrv_drain_all_begin(); + g_assert_cmpint(s->drain_count, ==, 1); + bdrv_drain_all_end(); + g_assert_cmpint(s->drain_count, ==, 0); + + /* Now do the same while a request is pending */ + aio_ret = -EINPROGRESS; + acb = blk_aio_preadv(blk, 0, &qiov, 0, aio_ret_cb, &aio_ret); + g_assert(acb != NULL); + g_assert_cmpint(aio_ret, ==, -EINPROGRESS); + + g_assert_cmpint(s->drain_count, ==, 0); + bdrv_drain_all_begin(); + g_assert_cmpint(aio_ret, ==, 0); + g_assert_cmpint(s->drain_count, ==, 1); + bdrv_drain_all_end(); + g_assert_cmpint(s->drain_count, ==, 0); + + bdrv_unref(bs); + 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("/bdrv-drain/driver-cb/drain_all", test_drv_cb_drain_all); + + return g_test_run(); +} diff --git a/tests/Makefile.include b/tests/Makefile.include index XXXXXXX..XXXXXXX 100644 --- a/tests/Makefile.include +++ b/tests/Makefile.include @@ -XXX,XX +XXX,XX @@ gcov-files-test-thread-pool-y = thread-pool.c gcov-files-test-hbitmap-y = util/hbitmap.c check-unit-y += tests/test-hbitmap$(EXESUF) gcov-files-test-hbitmap-y = blockjob.c +check-unit-y += tests/test-bdrv-drain$(EXESUF) check-unit-y += tests/test-blockjob$(EXESUF) check-unit-y += tests/test-blockjob-txn$(EXESUF) check-unit-y += tests/test-x86-cpuid$(EXESUF) @@ -XXX,XX +XXX,XX @@ tests/test-coroutine$(EXESUF): tests/test-coroutine.o $(test-block-obj-y) tests/test-aio$(EXESUF): tests/test-aio.o $(test-block-obj-y) tests/test-aio-multithread$(EXESUF): tests/test-aio-multithread.o $(test-block-obj-y) tests/test-throttle$(EXESUF): tests/test-throttle.o $(test-block-obj-y) +tests/test-bdrv-drain$(EXESUF): tests/test-bdrv-drain.o $(test-block-obj-y) $(test-util-obj-y) tests/test-blockjob$(EXESUF): tests/test-blockjob.o $(test-block-obj-y) $(test-util-obj-y) tests/test-blockjob-txn$(EXESUF): tests/test-blockjob-txn.o $(test-block-obj-y) $(test-util-obj-y) tests/test-thread-pool$(EXESUF): tests/test-thread-pool.o $(test-block-obj-y) -- 2.13.6 Now that the bdrv_drain_invoke() calls are pulled up to the callers of bdrv_drain_recurse(), the 'begin' parameter isn't needed any more. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/io.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static void bdrv_drain_invoke(BlockDriverState *bs, bool begin) } } -static bool bdrv_drain_recurse(BlockDriverState *bs, bool begin) +static bool bdrv_drain_recurse(BlockDriverState *bs) { BdrvChild *child, *tmp; bool waited; @@ -XXX,XX +XXX,XX @@ static bool bdrv_drain_recurse(BlockDriverState *bs, bool begin) */ bdrv_ref(bs); } - waited |= bdrv_drain_recurse(bs, begin); + waited |= bdrv_drain_recurse(bs); if (in_main_loop) { bdrv_unref(bs); } @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs) } bdrv_drain_invoke(bs, true); - bdrv_drain_recurse(bs, true); + bdrv_drain_recurse(bs); } void bdrv_drained_end(BlockDriverState *bs) @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) bdrv_parent_drained_end(bs); bdrv_drain_invoke(bs, false); - bdrv_drain_recurse(bs, false); + bdrv_drain_recurse(bs); aio_enable_external(bdrv_get_aio_context(bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) aio_context_acquire(aio_context); for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { if (aio_context == bdrv_get_aio_context(bs)) { - waited |= bdrv_drain_recurse(bs, true); + waited |= bdrv_drain_recurse(bs); } } aio_context_release(aio_context); @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) aio_enable_external(aio_context); bdrv_parent_drained_end(bs); bdrv_drain_invoke(bs, false); - bdrv_drain_recurse(bs, false); + bdrv_drain_recurse(bs); aio_context_release(aio_context); } -- 2.13.6 The device is drained, so there is no point in waiting for requests at the end of the drained section. Remove the bdrv_drain_recurse() calls there. The bdrv_drain_recurse() calls were introduced in commit 481cad48e5e in order to call the .bdrv_co_drain_end() driver callback. This is now done by a separate bdrv_drain_invoke() call. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/io.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) bdrv_parent_drained_end(bs); bdrv_drain_invoke(bs, false); - bdrv_drain_recurse(bs); aio_enable_external(bdrv_get_aio_context(bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) aio_enable_external(aio_context); bdrv_parent_drained_end(bs); bdrv_drain_invoke(bs, false); - bdrv_drain_recurse(bs); aio_context_release(aio_context); } -- 2.13.6 Drain requests are propagated to child nodes, parent nodes and directly to the AioContext. The order in which this happened was different between all combinations of drain/drain_all and begin/end. The correct order is to keep children only drained when their parents are also drained. This means that at the start of a drained section, the AioContext needs to be drained first, the parents second and only then the children. The correct order for the end of a drained section is the opposite. This patch changes the three other functions to follow the example of bdrv_drained_begin(), which is the only one that got it right. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- block/io.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs) return; } + /* Stop things in parent-to-child order */ if (atomic_fetch_inc(&bs->quiesce_counter) == 0) { aio_disable_external(bdrv_get_aio_context(bs)); bdrv_parent_drained_begin(bs); @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) return; } - bdrv_parent_drained_end(bs); + /* Re-enable things in child-to-parent order */ bdrv_drain_invoke(bs, false); + bdrv_parent_drained_end(bs); aio_enable_external(bdrv_get_aio_context(bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { AioContext *aio_context = bdrv_get_aio_context(bs); + /* Stop things in parent-to-child order */ aio_context_acquire(aio_context); - bdrv_parent_drained_begin(bs); aio_disable_external(aio_context); + bdrv_parent_drained_begin(bs); bdrv_drain_invoke(bs, true); aio_context_release(aio_context); @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { AioContext *aio_context = bdrv_get_aio_context(bs); + /* Re-enable things in child-to-parent order */ aio_context_acquire(aio_context); - aio_enable_external(aio_context); - bdrv_parent_drained_end(bs); bdrv_drain_invoke(bs, false); + bdrv_parent_drained_end(bs); + aio_enable_external(aio_context); aio_context_release(aio_context); } -- 2.13.6 Commit 15afd94a047 added code to acquire and release the AioContext in qemuio_command(). This means that the lock is taken twice now in the call path from hmp_qemu_io(). This causes BDRV_POLL_WHILE() to hang for any requests issued to nodes in a non-mainloop AioContext. Dropping the first locking from hmp_qemu_io() fixes the problem. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> --- hmp.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/hmp.c b/hmp.c index XXXXXXX..XXXXXXX 100644 --- a/hmp.c +++ b/hmp.c @@ -XXX,XX +XXX,XX @@ void hmp_qemu_io(Monitor *mon, const QDict *qdict) { BlockBackend *blk; BlockBackend *local_blk = NULL; - AioContext *aio_context; const char* device = qdict_get_str(qdict, "device"); const char* command = qdict_get_str(qdict, "command"); Error *err = NULL; @@ -XXX,XX +XXX,XX @@ void hmp_qemu_io(Monitor *mon, const QDict *qdict) } } - aio_context = blk_get_aio_context(blk); - aio_context_acquire(aio_context); - /* * Notably absent: Proper permission management. This is sad, but it seems * almost impossible to achieve without changing the semantics and thereby @@ -XXX,XX +XXX,XX @@ void hmp_qemu_io(Monitor *mon, const QDict *qdict) */ qemuio_command(blk, command); - aio_context_release(aio_context); - fail: blk_unref(local_blk); hmp_handle_error(mon, &err); -- 2.13.6 From: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com> Since bdrv_co_preadv does all neccessary checks including reading after the end of the backing file, avoid duplication of verification before bdrv_co_preadv call. Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 3 --- block/qcow2.c | 51 ++++++++------------------------------------------- 2 files changed, 8 insertions(+), 46 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ uint32_t offset_to_reftable_index(BDRVQcow2State *s, uint64_t offset) } /* qcow2.c functions */ -int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov, - int64_t sector_num, int nb_sectors); - int64_t qcow2_refcount_metadata_size(int64_t clusters, size_t cluster_size, int refcount_order, bool generous_increase, uint64_t *refblock_count); diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int64_t coroutine_fn qcow2_co_get_block_status(BlockDriverState *bs, return status; } -/* handle reading after the end of the backing file */ -int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov, - int64_t offset, int bytes) -{ - uint64_t bs_size = bs->total_sectors * BDRV_SECTOR_SIZE; - int n1; - - if ((offset + bytes) <= bs_size) { - return bytes; - } - - if (offset >= bs_size) { - n1 = 0; - } else { - n1 = bs_size - offset; - } - - qemu_iovec_memset(qiov, n1, 0, bytes - n1); - - return n1; -} - static coroutine_fn int qcow2_co_preadv(BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, int flags) { BDRVQcow2State *s = bs->opaque; - int offset_in_cluster, n1; + int offset_in_cluster; int ret; unsigned int cur_bytes; /* number of bytes in current iteration */ uint64_t cluster_offset = 0; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv(BlockDriverState *bs, uint64_t offset, case QCOW2_CLUSTER_UNALLOCATED: if (bs->backing) { - /* read from the base image */ - n1 = qcow2_backing_read1(bs->backing->bs, &hd_qiov, - offset, cur_bytes); - if (n1 > 0) { - QEMUIOVector local_qiov; - - qemu_iovec_init(&local_qiov, hd_qiov.niov); - qemu_iovec_concat(&local_qiov, &hd_qiov, 0, n1); - - BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO); - qemu_co_mutex_unlock(&s->lock); - ret = bdrv_co_preadv(bs->backing, offset, n1, - &local_qiov, 0); - qemu_co_mutex_lock(&s->lock); - - qemu_iovec_destroy(&local_qiov); - - if (ret < 0) { - goto fail; - } + BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO); + qemu_co_mutex_unlock(&s->lock); + ret = bdrv_co_preadv(bs->backing, offset, cur_bytes, + &hd_qiov, 0); + qemu_co_mutex_lock(&s->lock); + if (ret < 0) { + goto fail; } } else { /* Note: in this case, no need to wait */ -- 2.13.6 Removing a quorum child node with x-blockdev-change results in a quorum driver state that cannot be recreated with create options because it would require a list with gaps. This causes trouble in at least .bdrv_refresh_filename(). Document this problem so that we won't accidentally mark the command stable without having addressed it. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> --- qapi/block-core.json | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/qapi/block-core.json b/qapi/block-core.json index XXXXXXX..XXXXXXX 100644 --- a/qapi/block-core.json +++ b/qapi/block-core.json @@ -XXX,XX +XXX,XX @@ # does not support all kinds of operations, all kinds of children, nor # all block drivers. # +# FIXME Removing children from a quorum node means introducing gaps in the +# child indices. This cannot be represented in the 'children' list of +# BlockdevOptionsQuorum, as returned by .bdrv_refresh_filename(). +# # Warning: The data in a new quorum child MUST be consistent with that of # the rest of the array. # -- 2.13.6 From: Doug Gale <doug16k@gmail.com> Add trace output for commands, errors, and undefined behavior. Add guest error log output for undefined behavior. Report invalid undefined accesses to MMIO. Annotate unlikely error checks with unlikely. Signed-off-by: Doug Gale <doug16k@gmail.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- hw/block/nvme.c | 349 ++++++++++++++++++++++++++++++++++++++++++-------- hw/block/trace-events | 93 ++++++++++++++ 2 files changed, 390 insertions(+), 52 deletions(-) diff --git a/hw/block/nvme.c b/hw/block/nvme.c index XXXXXXX..XXXXXXX 100644 --- a/hw/block/nvme.c +++ b/hw/block/nvme.c @@ -XXX,XX +XXX,XX @@ #include "qapi/visitor.h" #include "sysemu/block-backend.h" +#include "qemu/log.h" +#include "trace.h" #include "nvme.h" +#define NVME_GUEST_ERR(trace, fmt, ...) \ + do { \ + (trace_##trace)(__VA_ARGS__); \ + qemu_log_mask(LOG_GUEST_ERROR, #trace \ + " in %s: " fmt "\n", __func__, ## __VA_ARGS__); \ + } while (0) + static void nvme_process_sq(void *opaque); static void nvme_addr_read(NvmeCtrl *n, hwaddr addr, void *buf, int size) @@ -XXX,XX +XXX,XX @@ static void nvme_isr_notify(NvmeCtrl *n, NvmeCQueue *cq) { if (cq->irq_enabled) { if (msix_enabled(&(n->parent_obj))) { + trace_nvme_irq_msix(cq->vector); msix_notify(&(n->parent_obj), cq->vector); } else { + trace_nvme_irq_pin(); pci_irq_pulse(&n->parent_obj); } + } else { + trace_nvme_irq_masked(); } } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_map_prp(QEMUSGList *qsg, QEMUIOVector *iov, uint64_t prp1, trans_len = MIN(len, trans_len); int num_prps = (len >> n->page_bits) + 1; - if (!prp1) { + if (unlikely(!prp1)) { + trace_nvme_err_invalid_prp(); return NVME_INVALID_FIELD | NVME_DNR; } else if (n->cmbsz && prp1 >= n->ctrl_mem.addr && prp1 < n->ctrl_mem.addr + int128_get64(n->ctrl_mem.size)) { @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_map_prp(QEMUSGList *qsg, QEMUIOVector *iov, uint64_t prp1, } len -= trans_len; if (len) { - if (!prp2) { + if (unlikely(!prp2)) { + trace_nvme_err_invalid_prp2_missing(); goto unmap; } if (len > n->page_size) { @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_map_prp(QEMUSGList *qsg, QEMUIOVector *iov, uint64_t prp1, uint64_t prp_ent = le64_to_cpu(prp_list[i]); if (i == n->max_prp_ents - 1 && len > n->page_size) { - if (!prp_ent || prp_ent & (n->page_size - 1)) { + if (unlikely(!prp_ent || prp_ent & (n->page_size - 1))) { + trace_nvme_err_invalid_prplist_ent(prp_ent); goto unmap; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_map_prp(QEMUSGList *qsg, QEMUIOVector *iov, uint64_t prp1, prp_ent = le64_to_cpu(prp_list[i]); } - if (!prp_ent || prp_ent & (n->page_size - 1)) { + if (unlikely(!prp_ent || prp_ent & (n->page_size - 1))) { + trace_nvme_err_invalid_prplist_ent(prp_ent); goto unmap; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_map_prp(QEMUSGList *qsg, QEMUIOVector *iov, uint64_t prp1, i++; } } else { - if (prp2 & (n->page_size - 1)) { + if (unlikely(prp2 & (n->page_size - 1))) { + trace_nvme_err_invalid_prp2_align(prp2); goto unmap; } if (qsg->nsg) { @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_dma_read_prp(NvmeCtrl *n, uint8_t *ptr, uint32_t len, QEMUIOVector iov; uint16_t status = NVME_SUCCESS; + trace_nvme_dma_read(prp1, prp2); + if (nvme_map_prp(&qsg, &iov, prp1, prp2, len, n)) { return NVME_INVALID_FIELD | NVME_DNR; } if (qsg.nsg > 0) { - if (dma_buf_read(ptr, len, &qsg)) { + if (unlikely(dma_buf_read(ptr, len, &qsg))) { + trace_nvme_err_invalid_dma(); status = NVME_INVALID_FIELD | NVME_DNR; } qemu_sglist_destroy(&qsg); } else { - if (qemu_iovec_to_buf(&iov, 0, ptr, len) != len) { + if (unlikely(qemu_iovec_to_buf(&iov, 0, ptr, len) != len)) { + trace_nvme_err_invalid_dma(); status = NVME_INVALID_FIELD | NVME_DNR; } qemu_iovec_destroy(&iov); @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_write_zeros(NvmeCtrl *n, NvmeNamespace *ns, NvmeCmd *cmd, uint64_t aio_slba = slba << (data_shift - BDRV_SECTOR_BITS); uint32_t aio_nlb = nlb << (data_shift - BDRV_SECTOR_BITS); - if (slba + nlb > ns->id_ns.nsze) { + if (unlikely(slba + nlb > ns->id_ns.nsze)) { + trace_nvme_err_invalid_lba_range(slba, nlb, ns->id_ns.nsze); return NVME_LBA_RANGE | NVME_DNR; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_rw(NvmeCtrl *n, NvmeNamespace *ns, NvmeCmd *cmd, 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) { + trace_nvme_rw(is_write ? "write" : "read", nlb, data_size, slba); + + if (unlikely((slba + nlb) > ns->id_ns.nsze)) { block_acct_invalid(blk_get_stats(n->conf.blk), acct); + trace_nvme_err_invalid_lba_range(slba, nlb, ns->id_ns.nsze); return NVME_LBA_RANGE | NVME_DNR; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_io_cmd(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req) NvmeNamespace *ns; uint32_t nsid = le32_to_cpu(cmd->nsid); - if (nsid == 0 || nsid > n->num_namespaces) { + if (unlikely(nsid == 0 || nsid > n->num_namespaces)) { + trace_nvme_err_invalid_ns(nsid, n->num_namespaces); return NVME_INVALID_NSID | NVME_DNR; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_io_cmd(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req) case NVME_CMD_READ: return nvme_rw(n, ns, cmd, req); default: + trace_nvme_err_invalid_opc(cmd->opcode); return NVME_INVALID_OPCODE | NVME_DNR; } } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_del_sq(NvmeCtrl *n, NvmeCmd *cmd) NvmeCQueue *cq; uint16_t qid = le16_to_cpu(c->qid); - if (!qid || nvme_check_sqid(n, qid)) { + if (unlikely(!qid || nvme_check_sqid(n, qid))) { + trace_nvme_err_invalid_del_sq(qid); return NVME_INVALID_QID | NVME_DNR; } + trace_nvme_del_sq(qid); + sq = n->sq[qid]; while (!QTAILQ_EMPTY(&sq->out_req_list)) { req = QTAILQ_FIRST(&sq->out_req_list); @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_create_sq(NvmeCtrl *n, NvmeCmd *cmd) uint16_t qflags = le16_to_cpu(c->sq_flags); uint64_t prp1 = le64_to_cpu(c->prp1); - if (!cqid || nvme_check_cqid(n, cqid)) { + trace_nvme_create_sq(prp1, sqid, cqid, qsize, qflags); + + if (unlikely(!cqid || nvme_check_cqid(n, cqid))) { + trace_nvme_err_invalid_create_sq_cqid(cqid); return NVME_INVALID_CQID | NVME_DNR; } - if (!sqid || !nvme_check_sqid(n, sqid)) { + if (unlikely(!sqid || !nvme_check_sqid(n, sqid))) { + trace_nvme_err_invalid_create_sq_sqid(sqid); return NVME_INVALID_QID | NVME_DNR; } - if (!qsize || qsize > NVME_CAP_MQES(n->bar.cap)) { + if (unlikely(!qsize || qsize > NVME_CAP_MQES(n->bar.cap))) { + trace_nvme_err_invalid_create_sq_size(qsize); return NVME_MAX_QSIZE_EXCEEDED | NVME_DNR; } - if (!prp1 || prp1 & (n->page_size - 1)) { + if (unlikely(!prp1 || prp1 & (n->page_size - 1))) { + trace_nvme_err_invalid_create_sq_addr(prp1); return NVME_INVALID_FIELD | NVME_DNR; } - if (!(NVME_SQ_FLAGS_PC(qflags))) { + if (unlikely(!(NVME_SQ_FLAGS_PC(qflags)))) { + trace_nvme_err_invalid_create_sq_qflags(NVME_SQ_FLAGS_PC(qflags)); return NVME_INVALID_FIELD | NVME_DNR; } sq = g_malloc0(sizeof(*sq)); @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_del_cq(NvmeCtrl *n, NvmeCmd *cmd) NvmeCQueue *cq; uint16_t qid = le16_to_cpu(c->qid); - if (!qid || nvme_check_cqid(n, qid)) { + if (unlikely(!qid || nvme_check_cqid(n, qid))) { + trace_nvme_err_invalid_del_cq_cqid(qid); return NVME_INVALID_CQID | NVME_DNR; } cq = n->cq[qid]; - if (!QTAILQ_EMPTY(&cq->sq_list)) { + if (unlikely(!QTAILQ_EMPTY(&cq->sq_list))) { + trace_nvme_err_invalid_del_cq_notempty(qid); return NVME_INVALID_QUEUE_DEL; } + trace_nvme_del_cq(qid); nvme_free_cq(cq, n); return NVME_SUCCESS; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_create_cq(NvmeCtrl *n, NvmeCmd *cmd) uint16_t qflags = le16_to_cpu(c->cq_flags); uint64_t prp1 = le64_to_cpu(c->prp1); - if (!cqid || !nvme_check_cqid(n, cqid)) { + trace_nvme_create_cq(prp1, cqid, vector, qsize, qflags, + NVME_CQ_FLAGS_IEN(qflags) != 0); + + if (unlikely(!cqid || !nvme_check_cqid(n, cqid))) { + trace_nvme_err_invalid_create_cq_cqid(cqid); return NVME_INVALID_CQID | NVME_DNR; } - if (!qsize || qsize > NVME_CAP_MQES(n->bar.cap)) { + if (unlikely(!qsize || qsize > NVME_CAP_MQES(n->bar.cap))) { + trace_nvme_err_invalid_create_cq_size(qsize); return NVME_MAX_QSIZE_EXCEEDED | NVME_DNR; } - if (!prp1) { + if (unlikely(!prp1)) { + trace_nvme_err_invalid_create_cq_addr(prp1); return NVME_INVALID_FIELD | NVME_DNR; } - if (vector > n->num_queues) { + if (unlikely(vector > n->num_queues)) { + trace_nvme_err_invalid_create_cq_vector(vector); return NVME_INVALID_IRQ_VECTOR | NVME_DNR; } - if (!(NVME_CQ_FLAGS_PC(qflags))) { + if (unlikely(!(NVME_CQ_FLAGS_PC(qflags)))) { + trace_nvme_err_invalid_create_cq_qflags(NVME_CQ_FLAGS_PC(qflags)); return NVME_INVALID_FIELD | NVME_DNR; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_identify_ctrl(NvmeCtrl *n, NvmeIdentify *c) uint64_t prp1 = le64_to_cpu(c->prp1); uint64_t prp2 = le64_to_cpu(c->prp2); + trace_nvme_identify_ctrl(); + return nvme_dma_read_prp(n, (uint8_t *)&n->id_ctrl, sizeof(n->id_ctrl), prp1, prp2); } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_identify_ns(NvmeCtrl *n, NvmeIdentify *c) uint64_t prp1 = le64_to_cpu(c->prp1); uint64_t prp2 = le64_to_cpu(c->prp2); - if (nsid == 0 || nsid > n->num_namespaces) { + trace_nvme_identify_ns(nsid); + + if (unlikely(nsid == 0 || nsid > n->num_namespaces)) { + trace_nvme_err_invalid_ns(nsid, n->num_namespaces); return NVME_INVALID_NSID | NVME_DNR; } ns = &n->namespaces[nsid - 1]; + return nvme_dma_read_prp(n, (uint8_t *)&ns->id_ns, sizeof(ns->id_ns), prp1, prp2); } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_identify_nslist(NvmeCtrl *n, NvmeIdentify *c) uint16_t ret; int i, j = 0; + trace_nvme_identify_nslist(min_nsid); + list = g_malloc0(data_len); for (i = 0; i < n->num_namespaces; i++) { if (i < min_nsid) { @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_identify(NvmeCtrl *n, NvmeCmd *cmd) case 0x02: return nvme_identify_nslist(n, c); default: + trace_nvme_err_invalid_identify_cns(le32_to_cpu(c->cns)); return NVME_INVALID_FIELD | NVME_DNR; } } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_get_feature(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req) switch (dw10) { case NVME_VOLATILE_WRITE_CACHE: result = blk_enable_write_cache(n->conf.blk); + trace_nvme_getfeat_vwcache(result ? "enabled" : "disabled"); break; case NVME_NUMBER_OF_QUEUES: result = cpu_to_le32((n->num_queues - 2) | ((n->num_queues - 2) << 16)); + trace_nvme_getfeat_numq(result); break; default: + trace_nvme_err_invalid_getfeat(dw10); return NVME_INVALID_FIELD | NVME_DNR; } @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_set_feature(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req) blk_set_enable_write_cache(n->conf.blk, dw11 & 1); break; case NVME_NUMBER_OF_QUEUES: + trace_nvme_setfeat_numq((dw11 & 0xFFFF) + 1, + ((dw11 >> 16) & 0xFFFF) + 1, + n->num_queues - 1, n->num_queues - 1); req->cqe.result = cpu_to_le32((n->num_queues - 2) | ((n->num_queues - 2) << 16)); break; default: + trace_nvme_err_invalid_setfeat(dw10); return NVME_INVALID_FIELD | NVME_DNR; } return NVME_SUCCESS; @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_admin_cmd(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req) case NVME_ADM_CMD_GET_FEATURES: return nvme_get_feature(n, cmd, req); default: + trace_nvme_err_invalid_admin_opc(cmd->opcode); return NVME_INVALID_OPCODE | NVME_DNR; } } @@ -XXX,XX +XXX,XX @@ static int nvme_start_ctrl(NvmeCtrl *n) uint32_t page_bits = NVME_CC_MPS(n->bar.cc) + 12; uint32_t page_size = 1 << page_bits; - if (n->cq[0] || n->sq[0] || !n->bar.asq || !n->bar.acq || - n->bar.asq & (page_size - 1) || n->bar.acq & (page_size - 1) || - NVME_CC_MPS(n->bar.cc) < NVME_CAP_MPSMIN(n->bar.cap) || - NVME_CC_MPS(n->bar.cc) > NVME_CAP_MPSMAX(n->bar.cap) || - NVME_CC_IOCQES(n->bar.cc) < NVME_CTRL_CQES_MIN(n->id_ctrl.cqes) || - NVME_CC_IOCQES(n->bar.cc) > NVME_CTRL_CQES_MAX(n->id_ctrl.cqes) || - NVME_CC_IOSQES(n->bar.cc) < NVME_CTRL_SQES_MIN(n->id_ctrl.sqes) || - NVME_CC_IOSQES(n->bar.cc) > NVME_CTRL_SQES_MAX(n->id_ctrl.sqes) || - !NVME_AQA_ASQS(n->bar.aqa) || !NVME_AQA_ACQS(n->bar.aqa)) { + if (unlikely(n->cq[0])) { + trace_nvme_err_startfail_cq(); + return -1; + } + if (unlikely(n->sq[0])) { + trace_nvme_err_startfail_sq(); + return -1; + } + if (unlikely(!n->bar.asq)) { + trace_nvme_err_startfail_nbarasq(); + return -1; + } + if (unlikely(!n->bar.acq)) { + trace_nvme_err_startfail_nbaracq(); + return -1; + } + if (unlikely(n->bar.asq & (page_size - 1))) { + trace_nvme_err_startfail_asq_misaligned(n->bar.asq); + return -1; + } + if (unlikely(n->bar.acq & (page_size - 1))) { + trace_nvme_err_startfail_acq_misaligned(n->bar.acq); + return -1; + } + if (unlikely(NVME_CC_MPS(n->bar.cc) < + NVME_CAP_MPSMIN(n->bar.cap))) { + trace_nvme_err_startfail_page_too_small( + NVME_CC_MPS(n->bar.cc), + NVME_CAP_MPSMIN(n->bar.cap)); + return -1; + } + if (unlikely(NVME_CC_MPS(n->bar.cc) > + NVME_CAP_MPSMAX(n->bar.cap))) { + trace_nvme_err_startfail_page_too_large( + NVME_CC_MPS(n->bar.cc), + NVME_CAP_MPSMAX(n->bar.cap)); + return -1; + } + if (unlikely(NVME_CC_IOCQES(n->bar.cc) < + NVME_CTRL_CQES_MIN(n->id_ctrl.cqes))) { + trace_nvme_err_startfail_cqent_too_small( + NVME_CC_IOCQES(n->bar.cc), + NVME_CTRL_CQES_MIN(n->bar.cap)); + return -1; + } + if (unlikely(NVME_CC_IOCQES(n->bar.cc) > + NVME_CTRL_CQES_MAX(n->id_ctrl.cqes))) { + trace_nvme_err_startfail_cqent_too_large( + NVME_CC_IOCQES(n->bar.cc), + NVME_CTRL_CQES_MAX(n->bar.cap)); + return -1; + } + if (unlikely(NVME_CC_IOSQES(n->bar.cc) < + NVME_CTRL_SQES_MIN(n->id_ctrl.sqes))) { + trace_nvme_err_startfail_sqent_too_small( + NVME_CC_IOSQES(n->bar.cc), + NVME_CTRL_SQES_MIN(n->bar.cap)); + return -1; + } + if (unlikely(NVME_CC_IOSQES(n->bar.cc) > + NVME_CTRL_SQES_MAX(n->id_ctrl.sqes))) { + trace_nvme_err_startfail_sqent_too_large( + NVME_CC_IOSQES(n->bar.cc), + NVME_CTRL_SQES_MAX(n->bar.cap)); + return -1; + } + if (unlikely(!NVME_AQA_ASQS(n->bar.aqa))) { + trace_nvme_err_startfail_asqent_sz_zero(); + return -1; + } + if (unlikely(!NVME_AQA_ACQS(n->bar.aqa))) { + trace_nvme_err_startfail_acqent_sz_zero(); return -1; } @@ -XXX,XX +XXX,XX @@ static int nvme_start_ctrl(NvmeCtrl *n) static void nvme_write_bar(NvmeCtrl *n, hwaddr offset, uint64_t data, unsigned size) { + if (unlikely(offset & (sizeof(uint32_t) - 1))) { + NVME_GUEST_ERR(nvme_ub_mmiowr_misaligned32, + "MMIO write not 32-bit aligned," + " offset=0x%"PRIx64"", offset); + /* should be ignored, fall through for now */ + } + + if (unlikely(size < sizeof(uint32_t))) { + NVME_GUEST_ERR(nvme_ub_mmiowr_toosmall, + "MMIO write smaller than 32-bits," + " offset=0x%"PRIx64", size=%u", + offset, size); + /* should be ignored, fall through for now */ + } + switch (offset) { - case 0xc: + case 0xc: /* INTMS */ + if (unlikely(msix_enabled(&(n->parent_obj)))) { + NVME_GUEST_ERR(nvme_ub_mmiowr_intmask_with_msix, + "undefined access to interrupt mask set" + " when MSI-X is enabled"); + /* should be ignored, fall through for now */ + } n->bar.intms |= data & 0xffffffff; n->bar.intmc = n->bar.intms; + trace_nvme_mmio_intm_set(data & 0xffffffff, + n->bar.intmc); break; - case 0x10: + case 0x10: /* INTMC */ + if (unlikely(msix_enabled(&(n->parent_obj)))) { + NVME_GUEST_ERR(nvme_ub_mmiowr_intmask_with_msix, + "undefined access to interrupt mask clr" + " when MSI-X is enabled"); + /* should be ignored, fall through for now */ + } n->bar.intms &= ~(data & 0xffffffff); n->bar.intmc = n->bar.intms; + trace_nvme_mmio_intm_clr(data & 0xffffffff, + n->bar.intmc); break; - case 0x14: + case 0x14: /* CC */ + trace_nvme_mmio_cfg(data & 0xffffffff); /* Windows first sends data, then sends enable bit */ if (!NVME_CC_EN(data) && !NVME_CC_EN(n->bar.cc) && !NVME_CC_SHN(data) && !NVME_CC_SHN(n->bar.cc)) @@ -XXX,XX +XXX,XX @@ static void nvme_write_bar(NvmeCtrl *n, hwaddr offset, uint64_t data, if (NVME_CC_EN(data) && !NVME_CC_EN(n->bar.cc)) { n->bar.cc = data; - if (nvme_start_ctrl(n)) { + if (unlikely(nvme_start_ctrl(n))) { + trace_nvme_err_startfail(); n->bar.csts = NVME_CSTS_FAILED; } else { + trace_nvme_mmio_start_success(); n->bar.csts = NVME_CSTS_READY; } } else if (!NVME_CC_EN(data) && NVME_CC_EN(n->bar.cc)) { + trace_nvme_mmio_stopped(); nvme_clear_ctrl(n); n->bar.csts &= ~NVME_CSTS_READY; } if (NVME_CC_SHN(data) && !(NVME_CC_SHN(n->bar.cc))) { - nvme_clear_ctrl(n); - n->bar.cc = data; - n->bar.csts |= NVME_CSTS_SHST_COMPLETE; + trace_nvme_mmio_shutdown_set(); + nvme_clear_ctrl(n); + n->bar.cc = data; + n->bar.csts |= NVME_CSTS_SHST_COMPLETE; } else if (!NVME_CC_SHN(data) && NVME_CC_SHN(n->bar.cc)) { - n->bar.csts &= ~NVME_CSTS_SHST_COMPLETE; - n->bar.cc = data; + trace_nvme_mmio_shutdown_cleared(); + n->bar.csts &= ~NVME_CSTS_SHST_COMPLETE; + n->bar.cc = data; + } + break; + case 0x1C: /* CSTS */ + if (data & (1 << 4)) { + NVME_GUEST_ERR(nvme_ub_mmiowr_ssreset_w1c_unsupported, + "attempted to W1C CSTS.NSSRO" + " but CAP.NSSRS is zero (not supported)"); + } else if (data != 0) { + NVME_GUEST_ERR(nvme_ub_mmiowr_ro_csts, + "attempted to set a read only bit" + " of controller status"); + } + break; + case 0x20: /* NSSR */ + if (data == 0x4E564D65) { + trace_nvme_ub_mmiowr_ssreset_unsupported(); + } else { + /* The spec says that writes of other values have no effect */ + return; } break; - case 0x24: + case 0x24: /* AQA */ n->bar.aqa = data & 0xffffffff; + trace_nvme_mmio_aqattr(data & 0xffffffff); break; - case 0x28: + case 0x28: /* ASQ */ n->bar.asq = data; + trace_nvme_mmio_asqaddr(data); break; - case 0x2c: + case 0x2c: /* ASQ hi */ n->bar.asq |= data << 32; + trace_nvme_mmio_asqaddr_hi(data, n->bar.asq); break; - case 0x30: + case 0x30: /* ACQ */ + trace_nvme_mmio_acqaddr(data); n->bar.acq = data; break; - case 0x34: + case 0x34: /* ACQ hi */ n->bar.acq |= data << 32; + trace_nvme_mmio_acqaddr_hi(data, n->bar.acq); break; + case 0x38: /* CMBLOC */ + NVME_GUEST_ERR(nvme_ub_mmiowr_cmbloc_reserved, + "invalid write to reserved CMBLOC" + " when CMBSZ is zero, ignored"); + return; + case 0x3C: /* CMBSZ */ + NVME_GUEST_ERR(nvme_ub_mmiowr_cmbsz_readonly, + "invalid write to read only CMBSZ, ignored"); + return; default: + NVME_GUEST_ERR(nvme_ub_mmiowr_invalid, + "invalid MMIO write," + " offset=0x%"PRIx64", data=%"PRIx64"", + offset, data); break; } } @@ -XXX,XX +XXX,XX @@ static uint64_t nvme_mmio_read(void *opaque, hwaddr addr, unsigned size) uint8_t *ptr = (uint8_t *)&n->bar; uint64_t val = 0; + if (unlikely(addr & (sizeof(uint32_t) - 1))) { + NVME_GUEST_ERR(nvme_ub_mmiord_misaligned32, + "MMIO read not 32-bit aligned," + " offset=0x%"PRIx64"", addr); + /* should RAZ, fall through for now */ + } else if (unlikely(size < sizeof(uint32_t))) { + NVME_GUEST_ERR(nvme_ub_mmiord_toosmall, + "MMIO read smaller than 32-bits," + " offset=0x%"PRIx64"", addr); + /* should RAZ, fall through for now */ + } + if (addr < sizeof(n->bar)) { memcpy(&val, ptr + addr, size); + } else { + NVME_GUEST_ERR(nvme_ub_mmiord_invalid_ofs, + "MMIO read beyond last register," + " offset=0x%"PRIx64", returning 0", addr); } + return val; } @@ -XXX,XX +XXX,XX @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int val) { uint32_t qid; - if (addr & ((1 << 2) - 1)) { + if (unlikely(addr & ((1 << 2) - 1))) { + NVME_GUEST_ERR(nvme_ub_db_wr_misaligned, + "doorbell write not 32-bit aligned," + " offset=0x%"PRIx64", ignoring", addr); return; } if (((addr - 0x1000) >> 2) & 1) { + /* Completion queue doorbell write */ + uint16_t new_head = val & 0xffff; int start_sqs; NvmeCQueue *cq; qid = (addr - (0x1000 + (1 << 2))) >> 3; - if (nvme_check_cqid(n, qid)) { + if (unlikely(nvme_check_cqid(n, qid))) { + NVME_GUEST_ERR(nvme_ub_db_wr_invalid_cq, + "completion queue doorbell write" + " for nonexistent queue," + " sqid=%"PRIu32", ignoring", qid); return; } cq = n->cq[qid]; - if (new_head >= cq->size) { + if (unlikely(new_head >= cq->size)) { + NVME_GUEST_ERR(nvme_ub_db_wr_invalid_cqhead, + "completion queue doorbell write value" + " beyond queue size, sqid=%"PRIu32"," + " new_head=%"PRIu16", ignoring", + qid, new_head); return; } @@ -XXX,XX +XXX,XX @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int val) nvme_isr_notify(n, cq); } } else { + /* Submission queue doorbell write */ + uint16_t new_tail = val & 0xffff; NvmeSQueue *sq; qid = (addr - 0x1000) >> 3; - if (nvme_check_sqid(n, qid)) { + if (unlikely(nvme_check_sqid(n, qid))) { + NVME_GUEST_ERR(nvme_ub_db_wr_invalid_sq, + "submission queue doorbell write" + " for nonexistent queue," + " sqid=%"PRIu32", ignoring", qid); return; } sq = n->sq[qid]; - if (new_tail >= sq->size) { + if (unlikely(new_tail >= sq->size)) { + NVME_GUEST_ERR(nvme_ub_db_wr_invalid_sqtail, + "submission queue doorbell write value" + " beyond queue size, sqid=%"PRIu32"," + " new_tail=%"PRIu16", ignoring", + qid, new_tail); return; } diff --git a/hw/block/trace-events b/hw/block/trace-events index XXXXXXX..XXXXXXX 100644 --- a/hw/block/trace-events +++ b/hw/block/trace-events @@ -XXX,XX +XXX,XX @@ virtio_blk_submit_multireq(void *vdev, void *mrb, int start, int num_reqs, uint6 hd_geometry_lchs_guess(void *blk, int cyls, int heads, int secs) "blk %p LCHS %d %d %d" hd_geometry_guess(void *blk, uint32_t cyls, uint32_t heads, uint32_t secs, int trans) "blk %p CHS %u %u %u trans %d" +# hw/block/nvme.c +# nvme traces for successful events +nvme_irq_msix(uint32_t vector) "raising MSI-X IRQ vector %u" +nvme_irq_pin(void) "pulsing IRQ pin" +nvme_irq_masked(void) "IRQ is masked" +nvme_dma_read(uint64_t prp1, uint64_t prp2) "DMA read, prp1=0x%"PRIx64" prp2=0x%"PRIx64"" +nvme_rw(char const *verb, uint32_t blk_count, uint64_t byte_count, uint64_t lba) "%s %"PRIu32" blocks (%"PRIu64" bytes) from LBA %"PRIu64"" +nvme_create_sq(uint64_t addr, uint16_t sqid, uint16_t cqid, uint16_t qsize, uint16_t qflags) "create submission queue, addr=0x%"PRIx64", sqid=%"PRIu16", cqid=%"PRIu16", qsize=%"PRIu16", qflags=%"PRIu16"" +nvme_create_cq(uint64_t addr, uint16_t cqid, uint16_t vector, uint16_t size, uint16_t qflags, int ien) "create completion queue, addr=0x%"PRIx64", cqid=%"PRIu16", vector=%"PRIu16", qsize=%"PRIu16", qflags=%"PRIu16", ien=%d" +nvme_del_sq(uint16_t qid) "deleting submission queue sqid=%"PRIu16"" +nvme_del_cq(uint16_t cqid) "deleted completion queue, sqid=%"PRIu16"" +nvme_identify_ctrl(void) "identify controller" +nvme_identify_ns(uint16_t ns) "identify namespace, nsid=%"PRIu16"" +nvme_identify_nslist(uint16_t ns) "identify namespace list, nsid=%"PRIu16"" +nvme_getfeat_vwcache(char const* result) "get feature volatile write cache, result=%s" +nvme_getfeat_numq(int result) "get feature number of queues, result=%d" +nvme_setfeat_numq(int reqcq, int reqsq, int gotcq, int gotsq) "requested cq_count=%d sq_count=%d, responding with cq_count=%d sq_count=%d" +nvme_mmio_intm_set(uint64_t data, uint64_t new_mask) "wrote MMIO, interrupt mask set, data=0x%"PRIx64", new_mask=0x%"PRIx64"" +nvme_mmio_intm_clr(uint64_t data, uint64_t new_mask) "wrote MMIO, interrupt mask clr, data=0x%"PRIx64", new_mask=0x%"PRIx64"" +nvme_mmio_cfg(uint64_t data) "wrote MMIO, config controller config=0x%"PRIx64"" +nvme_mmio_aqattr(uint64_t data) "wrote MMIO, admin queue attributes=0x%"PRIx64"" +nvme_mmio_asqaddr(uint64_t data) "wrote MMIO, admin submission queue address=0x%"PRIx64"" +nvme_mmio_acqaddr(uint64_t data) "wrote MMIO, admin completion queue address=0x%"PRIx64"" +nvme_mmio_asqaddr_hi(uint64_t data, uint64_t new_addr) "wrote MMIO, admin submission queue high half=0x%"PRIx64", new_address=0x%"PRIx64"" +nvme_mmio_acqaddr_hi(uint64_t data, uint64_t new_addr) "wrote MMIO, admin completion queue high half=0x%"PRIx64", new_address=0x%"PRIx64"" +nvme_mmio_start_success(void) "setting controller enable bit succeeded" +nvme_mmio_stopped(void) "cleared controller enable bit" +nvme_mmio_shutdown_set(void) "shutdown bit set" +nvme_mmio_shutdown_cleared(void) "shutdown bit cleared" + +# nvme traces for error conditions +nvme_err_invalid_dma(void) "PRP/SGL is too small for transfer size" +nvme_err_invalid_prplist_ent(uint64_t prplist) "PRP list entry is null or not page aligned: 0x%"PRIx64"" +nvme_err_invalid_prp2_align(uint64_t prp2) "PRP2 is not page aligned: 0x%"PRIx64"" +nvme_err_invalid_prp2_missing(void) "PRP2 is null and more data to be transferred" +nvme_err_invalid_field(void) "invalid field" +nvme_err_invalid_prp(void) "invalid PRP" +nvme_err_invalid_sgl(void) "invalid SGL" +nvme_err_invalid_ns(uint32_t ns, uint32_t limit) "invalid namespace %u not within 1-%u" +nvme_err_invalid_opc(uint8_t opc) "invalid opcode 0x%"PRIx8"" +nvme_err_invalid_admin_opc(uint8_t opc) "invalid admin opcode 0x%"PRIx8"" +nvme_err_invalid_lba_range(uint64_t start, uint64_t len, uint64_t limit) "Invalid LBA start=%"PRIu64" len=%"PRIu64" limit=%"PRIu64"" +nvme_err_invalid_del_sq(uint16_t qid) "invalid submission queue deletion, sid=%"PRIu16"" +nvme_err_invalid_create_sq_cqid(uint16_t cqid) "failed creating submission queue, invalid cqid=%"PRIu16"" +nvme_err_invalid_create_sq_sqid(uint16_t sqid) "failed creating submission queue, invalid sqid=%"PRIu16"" +nvme_err_invalid_create_sq_size(uint16_t qsize) "failed creating submission queue, invalid qsize=%"PRIu16"" +nvme_err_invalid_create_sq_addr(uint64_t addr) "failed creating submission queue, addr=0x%"PRIx64"" +nvme_err_invalid_create_sq_qflags(uint16_t qflags) "failed creating submission queue, qflags=%"PRIu16"" +nvme_err_invalid_del_cq_cqid(uint16_t cqid) "failed deleting completion queue, cqid=%"PRIu16"" +nvme_err_invalid_del_cq_notempty(uint16_t cqid) "failed deleting completion queue, it is not empty, cqid=%"PRIu16"" +nvme_err_invalid_create_cq_cqid(uint16_t cqid) "failed creating completion queue, cqid=%"PRIu16"" +nvme_err_invalid_create_cq_size(uint16_t size) "failed creating completion queue, size=%"PRIu16"" +nvme_err_invalid_create_cq_addr(uint64_t addr) "failed creating completion queue, addr=0x%"PRIx64"" +nvme_err_invalid_create_cq_vector(uint16_t vector) "failed creating completion queue, vector=%"PRIu16"" +nvme_err_invalid_create_cq_qflags(uint16_t qflags) "failed creating completion queue, qflags=%"PRIu16"" +nvme_err_invalid_identify_cns(uint16_t cns) "identify, invalid cns=0x%"PRIx16"" +nvme_err_invalid_getfeat(int dw10) "invalid get features, dw10=0x%"PRIx32"" +nvme_err_invalid_setfeat(uint32_t dw10) "invalid set features, dw10=0x%"PRIx32"" +nvme_err_startfail_cq(void) "nvme_start_ctrl failed because there are non-admin completion queues" +nvme_err_startfail_sq(void) "nvme_start_ctrl failed because there are non-admin submission queues" +nvme_err_startfail_nbarasq(void) "nvme_start_ctrl failed because the admin submission queue address is null" +nvme_err_startfail_nbaracq(void) "nvme_start_ctrl failed because the admin completion queue address is null" +nvme_err_startfail_asq_misaligned(uint64_t addr) "nvme_start_ctrl failed because the admin submission queue address is misaligned: 0x%"PRIx64"" +nvme_err_startfail_acq_misaligned(uint64_t addr) "nvme_start_ctrl failed because the admin completion queue address is misaligned: 0x%"PRIx64"" +nvme_err_startfail_page_too_small(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the page size is too small: log2size=%u, min=%u" +nvme_err_startfail_page_too_large(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the page size is too large: log2size=%u, max=%u" +nvme_err_startfail_cqent_too_small(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the completion queue entry size is too small: log2size=%u, min=%u" +nvme_err_startfail_cqent_too_large(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the completion queue entry size is too large: log2size=%u, max=%u" +nvme_err_startfail_sqent_too_small(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the submission queue entry size is too small: log2size=%u, min=%u" +nvme_err_startfail_sqent_too_large(uint8_t log2ps, uint8_t maxlog2ps) "nvme_start_ctrl failed because the submission queue entry size is too large: log2size=%u, max=%u" +nvme_err_startfail_asqent_sz_zero(void) "nvme_start_ctrl failed because the admin submission queue size is zero" +nvme_err_startfail_acqent_sz_zero(void) "nvme_start_ctrl failed because the admin completion queue size is zero" +nvme_err_startfail(void) "setting controller enable bit failed" + +# Traces for undefined behavior +nvme_ub_mmiowr_misaligned32(uint64_t offset) "MMIO write not 32-bit aligned, offset=0x%"PRIx64"" +nvme_ub_mmiowr_toosmall(uint64_t offset, unsigned size) "MMIO write smaller than 32 bits, offset=0x%"PRIx64", size=%u" +nvme_ub_mmiowr_intmask_with_msix(void) "undefined access to interrupt mask set when MSI-X is enabled" +nvme_ub_mmiowr_ro_csts(void) "attempted to set a read only bit of controller status" +nvme_ub_mmiowr_ssreset_w1c_unsupported(void) "attempted to W1C CSTS.NSSRO but CAP.NSSRS is zero (not supported)" +nvme_ub_mmiowr_ssreset_unsupported(void) "attempted NVM subsystem reset but CAP.NSSRS is zero (not supported)" +nvme_ub_mmiowr_cmbloc_reserved(void) "invalid write to reserved CMBLOC when CMBSZ is zero, ignored" +nvme_ub_mmiowr_cmbsz_readonly(void) "invalid write to read only CMBSZ, ignored" +nvme_ub_mmiowr_invalid(uint64_t offset, uint64_t data) "invalid MMIO write, offset=0x%"PRIx64", data=0x%"PRIx64"" +nvme_ub_mmiord_misaligned32(uint64_t offset) "MMIO read not 32-bit aligned, offset=0x%"PRIx64"" +nvme_ub_mmiord_toosmall(uint64_t offset) "MMIO read smaller than 32-bits, offset=0x%"PRIx64"" +nvme_ub_mmiord_invalid_ofs(uint64_t offset) "MMIO read beyond last register, offset=0x%"PRIx64", returning 0" +nvme_ub_db_wr_misaligned(uint64_t offset) "doorbell write not 32-bit aligned, offset=0x%"PRIx64", ignoring" +nvme_ub_db_wr_invalid_cq(uint32_t qid) "completion queue doorbell write for nonexistent queue, cqid=%"PRIu32", ignoring" +nvme_ub_db_wr_invalid_cqhead(uint32_t qid, uint16_t new_head) "completion queue doorbell write value beyond queue size, cqid=%"PRIu32", new_head=%"PRIu16", ignoring" +nvme_ub_db_wr_invalid_sq(uint32_t qid) "submission queue doorbell write for nonexistent queue, sqid=%"PRIu32", ignoring" +nvme_ub_db_wr_invalid_sqtail(uint32_t qid, uint16_t new_tail) "submission queue doorbell write value beyond queue size, sqid=%"PRIu32", new_head=%"PRIu16", ignoring" + # hw/block/xen_disk.c xen_disk_alloc(char *name) "%s" xen_disk_init(char *name) "%s" -- 2.13.6 From: Fam Zheng <famz@redhat.com> Management tools create overlays of running guests with qemu-img: $ qemu-img create -b /image/in/use.qcow2 -f qcow2 /overlay/image.qcow2 but this doesn't work anymore due to image locking: qemu-img: /overlay/image.qcow2: Failed to get shared "write" lock Is another process using the image? Could not open backing image to determine size. Use the force share option to allow this use case again. Cc: qemu-stable@nongnu.org Signed-off-by: Fam Zheng <famz@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ void bdrv_img_create(const char *filename, const char *fmt, back_flags = flags; back_flags &= ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING); + backing_options = qdict_new(); if (backing_fmt) { - backing_options = qdict_new(); qdict_put_str(backing_options, "driver", backing_fmt); } + qdict_put_bool(backing_options, BDRV_OPT_FORCE_SHARE, true); bs = bdrv_open(full_backing, NULL, backing_options, back_flags, &local_err); -- 2.13.6 From: Thomas Huth <thuth@redhat.com> It's not working anymore since QEMU v1.3.0 - time to remove it now. Signed-off-by: Thomas Huth <thuth@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- blockdev.c | 11 ----------- qemu-doc.texi | 6 ------ 2 files changed, 17 deletions(-) diff --git a/blockdev.c b/blockdev.c index XXXXXXX..XXXXXXX 100644 --- a/blockdev.c +++ b/blockdev.c @@ -XXX,XX +XXX,XX @@ QemuOptsList qemu_legacy_drive_opts = { .type = QEMU_OPT_STRING, .help = "chs translation (auto, lba, none)", },{ - .name = "boot", - .type = QEMU_OPT_BOOL, - .help = "(deprecated, ignored)", - },{ .name = "addr", .type = QEMU_OPT_STRING, .help = "pci address (virtio only)", @@ -XXX,XX +XXX,XX @@ DriveInfo *drive_new(QemuOpts *all_opts, BlockInterfaceType block_default_type) goto fail; } - /* Deprecated option boot=[on|off] */ - if (qemu_opt_get(legacy_opts, "boot") != NULL) { - fprintf(stderr, "qemu-kvm: boot=on|off is deprecated and will be " - "ignored. Future versions will reject this parameter. Please " - "update your scripts.\n"); - } - /* Other deprecated options */ if (!qtest_enabled()) { for (i = 0; i < ARRAY_SIZE(deprecated); i++) { diff --git a/qemu-doc.texi b/qemu-doc.texi index XXXXXXX..XXXXXXX 100644 --- a/qemu-doc.texi +++ b/qemu-doc.texi @@ -XXX,XX +XXX,XX @@ deprecated. @section System emulator command line arguments -@subsection -drive boot=on|off (since 1.3.0) - -The ``boot=on|off'' option to the ``-drive'' argument is -ignored. Applications should use the ``bootindex=N'' parameter -to set an absolute ordering between devices instead. - @subsection -tdf (since 1.3.0) The ``-tdf'' argument is ignored. The behaviour implemented -- 2.13.6 From: Thomas Huth <thuth@redhat.com> It's been marked as deprecated since QEMU v2.10.0, and so far nobody complained that we should keep it, so let's remove this legacy option now to simplify the code quite a bit. Signed-off-by: Thomas Huth <thuth@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- vl.c | 86 ++------------------------------------------------------- qemu-doc.texi | 8 ------ qemu-options.hx | 19 ++----------- 3 files changed, 4 insertions(+), 109 deletions(-) diff --git a/vl.c b/vl.c index XXXXXXX..XXXXXXX 100644 --- a/vl.c +++ b/vl.c @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp) const char *boot_order = NULL; const char *boot_once = NULL; DisplayState *ds; - int cyls, heads, secs, translation; QemuOpts *opts, *machine_opts; - QemuOpts *hda_opts = NULL, *icount_opts = NULL, *accel_opts = NULL; + QemuOpts *icount_opts = NULL, *accel_opts = NULL; QemuOptsList *olist; int optind; const char *optarg; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp) cpu_model = NULL; snapshot = 0; - cyls = heads = secs = 0; - translation = BIOS_ATA_TRANSLATION_AUTO; nb_nics = 0; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp) if (optind >= argc) break; if (argv[optind][0] != '-') { - hda_opts = drive_add(IF_DEFAULT, 0, argv[optind++], HD_OPTS); + drive_add(IF_DEFAULT, 0, argv[optind++], HD_OPTS); } else { const QEMUOption *popt; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp) cpu_model = optarg; break; case QEMU_OPTION_hda: - { - char buf[256]; - if (cyls == 0) - snprintf(buf, sizeof(buf), "%s", HD_OPTS); - else - snprintf(buf, sizeof(buf), - "%s,cyls=%d,heads=%d,secs=%d%s", - HD_OPTS , cyls, heads, secs, - translation == BIOS_ATA_TRANSLATION_LBA ? - ",trans=lba" : - translation == BIOS_ATA_TRANSLATION_NONE ? - ",trans=none" : ""); - drive_add(IF_DEFAULT, 0, optarg, buf); - break; - } case QEMU_OPTION_hdb: case QEMU_OPTION_hdc: case QEMU_OPTION_hdd: @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp) case QEMU_OPTION_snapshot: snapshot = 1; break; - case QEMU_OPTION_hdachs: - { - const char *p; - p = optarg; - cyls = strtol(p, (char **)&p, 0); - if (cyls < 1 || cyls > 16383) - goto chs_fail; - if (*p != ',') - goto chs_fail; - p++; - heads = strtol(p, (char **)&p, 0); - if (heads < 1 || heads > 16) - goto chs_fail; - if (*p != ',') - goto chs_fail; - p++; - secs = strtol(p, (char **)&p, 0); - if (secs < 1 || secs > 63) - goto chs_fail; - if (*p == ',') { - p++; - if (!strcmp(p, "large")) { - translation = BIOS_ATA_TRANSLATION_LARGE; - } else if (!strcmp(p, "rechs")) { - translation = BIOS_ATA_TRANSLATION_RECHS; - } else if (!strcmp(p, "none")) { - translation = BIOS_ATA_TRANSLATION_NONE; - } else if (!strcmp(p, "lba")) { - translation = BIOS_ATA_TRANSLATION_LBA; - } else if (!strcmp(p, "auto")) { - translation = BIOS_ATA_TRANSLATION_AUTO; - } else { - goto chs_fail; - } - } else if (*p != '\0') { - chs_fail: - error_report("invalid physical CHS format"); - exit(1); - } - if (hda_opts != NULL) { - qemu_opt_set_number(hda_opts, "cyls", cyls, - &error_abort); - qemu_opt_set_number(hda_opts, "heads", heads, - &error_abort); - qemu_opt_set_number(hda_opts, "secs", secs, - &error_abort); - if (translation == BIOS_ATA_TRANSLATION_LARGE) { - qemu_opt_set(hda_opts, "trans", "large", - &error_abort); - } else if (translation == BIOS_ATA_TRANSLATION_RECHS) { - qemu_opt_set(hda_opts, "trans", "rechs", - &error_abort); - } else if (translation == BIOS_ATA_TRANSLATION_LBA) { - qemu_opt_set(hda_opts, "trans", "lba", - &error_abort); - } else if (translation == BIOS_ATA_TRANSLATION_NONE) { - qemu_opt_set(hda_opts, "trans", "none", - &error_abort); - } - } - } - error_report("'-hdachs' is deprecated, please use '-device" - " ide-hd,cyls=c,heads=h,secs=s,...' instead"); - break; case QEMU_OPTION_numa: opts = qemu_opts_parse_noisily(qemu_find_opts("numa"), optarg, true); diff --git a/qemu-doc.texi b/qemu-doc.texi index XXXXXXX..XXXXXXX 100644 --- a/qemu-doc.texi +++ b/qemu-doc.texi @@ -XXX,XX +XXX,XX @@ The ``--net dump'' argument is now replaced with the ``-object filter-dump'' argument which works in combination with the modern ``-netdev`` backends instead. -@subsection -hdachs (since 2.10.0) - -The ``-hdachs'' argument is now a synonym for setting -the ``cyls'', ``heads'', ``secs'', and ``trans'' properties -on the ``ide-hd'' device using the ``-device'' argument. -The new syntax allows different settings to be provided -per disk. - @subsection -usbdevice (since 2.10.0) The ``-usbdevice DEV'' argument is now a synonym for setting diff --git a/qemu-options.hx b/qemu-options.hx index XXXXXXX..XXXXXXX 100644 --- a/qemu-options.hx +++ b/qemu-options.hx @@ -XXX,XX +XXX,XX @@ of available connectors of a given interface type. @item media=@var{media} This option defines the type of the media: disk or cdrom. @item cyls=@var{c},heads=@var{h},secs=@var{s}[,trans=@var{t}] -These options have the same definition as they have in @option{-hdachs}. -These parameters are deprecated, use the corresponding parameters +Force disk physical geometry and the optional BIOS translation (trans=none or +lba). These parameters are deprecated, use the corresponding parameters of @code{-device} instead. @item snapshot=@var{snapshot} @var{snapshot} is "on" or "off" and controls snapshot mode for the given drive @@ -XXX,XX +XXX,XX @@ the raw disk image you use is not written back. You can however force the write back by pressing @key{C-a s} (@pxref{disk_images}). ETEXI -DEF("hdachs", HAS_ARG, QEMU_OPTION_hdachs, \ - "-hdachs c,h,s[,t]\n" \ - " force hard disk 0 physical geometry and the optional BIOS\n" \ - " translation (t=none or lba) (usually QEMU can guess them)\n", - QEMU_ARCH_ALL) -STEXI -@item -hdachs @var{c},@var{h},@var{s},[,@var{t}] -@findex -hdachs -Force hard disk 0 physical geometry (1 <= @var{c} <= 16383, 1 <= -@var{h} <= 16, 1 <= @var{s} <= 63) and optionally force the BIOS -translation mode (@var{t}=none, lba or auto). Usually QEMU can guess -all those parameters. This option is deprecated, please use -@code{-device ide-hd,cyls=c,heads=h,secs=s,...} instead. -ETEXI - DEF("fsdev", HAS_ARG, QEMU_OPTION_fsdev, "-fsdev fsdriver,id=id[,path=path,][security_model={mapped-xattr|mapped-file|passthrough|none}]\n" " [,writeout=immediate][,readonly][,socket=socket|sock_fd=sock_fd][,fmode=fmode][,dmode=dmode]\n" -- 2.13.6 From: Thomas Huth <thuth@redhat.com> Looks like we forgot to announce the deprecation of these options in the corresponding chapter of the qemu-doc text, so let's do that now. Signed-off-by: Thomas Huth <thuth@redhat.com> Reviewed-by: John Snow <jsnow@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qemu-doc.texi | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/qemu-doc.texi b/qemu-doc.texi index XXXXXXX..XXXXXXX 100644 --- a/qemu-doc.texi +++ b/qemu-doc.texi @@ -XXX,XX +XXX,XX @@ longer be directly supported in QEMU. The ``-drive if=scsi'' argument is replaced by the the ``-device BUS-TYPE'' argument combined with ``-drive if=none''. +@subsection -drive cyls=...,heads=...,secs=...,trans=... (since 2.10.0) + +The drive geometry arguments are replaced by the the geometry arguments +that can be specified with the ``-device'' parameter. + +@subsection -drive serial=... (since 2.10.0) + +The drive serial argument is replaced by the the serial argument +that can be specified with the ``-device'' parameter. + +@subsection -drive addr=... (since 2.10.0) + +The drive addr argument is replaced by the the addr argument +that can be specified with the ``-device'' parameter. + @subsection -net dump (since 2.10.0) The ``--net dump'' argument is now replaced with the -- 2.13.6 From: Fam Zheng <famz@redhat.com> Signed-off-by: Fam Zheng <famz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block_int.h | 1 - block/io.c | 18 ------------------ 2 files changed, 19 deletions(-) diff --git a/include/block/block_int.h b/include/block/block_int.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -XXX,XX +XXX,XX @@ bool blk_dev_is_tray_open(BlockBackend *blk); bool blk_dev_is_medium_locked(BlockBackend *blk); void bdrv_set_dirty(BlockDriverState *bs, int64_t offset, int64_t bytes); -bool bdrv_requests_pending(BlockDriverState *bs); void bdrv_clear_dirty_bitmap(BdrvDirtyBitmap *bitmap, HBitmap **out); void bdrv_undo_clear_dirty_bitmap(BdrvDirtyBitmap *bitmap, HBitmap *in); diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_disable_copy_on_read(BlockDriverState *bs) assert(old >= 1); } -/* Check if any requests are in-flight (including throttled requests) */ -bool bdrv_requests_pending(BlockDriverState *bs) -{ - BdrvChild *child; - - if (atomic_read(&bs->in_flight)) { - return true; - } - - QLIST_FOREACH(child, &bs->children, next) { - if (bdrv_requests_pending(child->bs)) { - return true; - } - } - - return false; -} - typedef struct { Coroutine *co; BlockDriverState *bs; -- 2.13.6 Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Fam Zheng <famz@redhat.com> --- block/io.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) BdrvNextIterator it; GSList *aio_ctxs = NULL, *ctx; + /* BDRV_POLL_WHILE() for a node can only be called from its own I/O thread + * or the main loop AioContext. We potentially use BDRV_POLL_WHILE() on + * nodes in several different AioContexts, so make sure we're in the main + * context. */ + assert(qemu_get_current_aio_context() == qemu_get_aio_context()); + block_job_pause_all(); for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { -- 2.13.6 bdrv_drained_begin() doesn't increase bs->quiesce_counter recursively and also doesn't notify other parent nodes of children, which both means that the child nodes are not actually drained, and bdrv_drained_begin() is providing useful functionality only on a single node. To keep things consistent, we also shouldn't call the block driver callbacks recursively. A proper recursive drain version that provides an actually working drained section for child nodes will be introduced later. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Fam Zheng <famz@redhat.com> --- block/io.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_drain_invoke_entry(void *opaque) } /* Recursively call BlockDriver.bdrv_co_drain_begin/end callbacks */ -static void bdrv_drain_invoke(BlockDriverState *bs, bool begin) +static void bdrv_drain_invoke(BlockDriverState *bs, bool begin, bool recursive) { BdrvChild *child, *tmp; BdrvCoDrainData data = { .bs = bs, .done = false, .begin = begin}; @@ -XXX,XX +XXX,XX @@ static void bdrv_drain_invoke(BlockDriverState *bs, bool begin) bdrv_coroutine_enter(bs, data.co); BDRV_POLL_WHILE(bs, !data.done); - QLIST_FOREACH_SAFE(child, &bs->children, next, tmp) { - bdrv_drain_invoke(child->bs, begin); + if (recursive) { + QLIST_FOREACH_SAFE(child, &bs->children, next, tmp) { + bdrv_drain_invoke(child->bs, begin, true); + } } } @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs) bdrv_parent_drained_begin(bs); } - bdrv_drain_invoke(bs, true); + bdrv_drain_invoke(bs, true, false); bdrv_drain_recurse(bs); } @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) } /* Re-enable things in child-to-parent order */ - bdrv_drain_invoke(bs, false); + bdrv_drain_invoke(bs, false, false); bdrv_parent_drained_end(bs); aio_enable_external(bdrv_get_aio_context(bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) aio_context_acquire(aio_context); aio_disable_external(aio_context); bdrv_parent_drained_begin(bs); - bdrv_drain_invoke(bs, true); + bdrv_drain_invoke(bs, true, true); aio_context_release(aio_context); if (!g_slist_find(aio_ctxs, aio_context)) { @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) /* Re-enable things in child-to-parent order */ aio_context_acquire(aio_context); - bdrv_drain_invoke(bs, false); + bdrv_drain_invoke(bs, false, true); bdrv_parent_drained_end(bs); aio_enable_external(aio_context); aio_context_release(aio_context); -- 2.13.6 The existing test is for bdrv_drain_all_begin/end() only. Generalise the test case so that it can be run for the other variants as well. At the moment this is only bdrv_drain_begin/end(), but in a while, we'll add another one. Also, add a backing file to the test node to test whether the operations work recursively. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 69 ++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 62 insertions(+), 7 deletions(-) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_test = { .bdrv_co_drain_begin = bdrv_test_co_drain_begin, .bdrv_co_drain_end = bdrv_test_co_drain_end, + + .bdrv_child_perm = bdrv_format_default_perms, }; static void aio_ret_cb(void *opaque, int ret) @@ -XXX,XX +XXX,XX @@ static void aio_ret_cb(void *opaque, int ret) *aio_ret = ret; } -static void test_drv_cb_drain_all(void) +enum drain_type { + BDRV_DRAIN_ALL, + BDRV_DRAIN, +}; + +static void do_drain_begin(enum drain_type drain_type, BlockDriverState *bs) +{ + switch (drain_type) { + case BDRV_DRAIN_ALL: bdrv_drain_all_begin(); break; + case BDRV_DRAIN: bdrv_drained_begin(bs); break; + default: g_assert_not_reached(); + } +} + +static void do_drain_end(enum drain_type drain_type, BlockDriverState *bs) +{ + switch (drain_type) { + case BDRV_DRAIN_ALL: bdrv_drain_all_end(); break; + case BDRV_DRAIN: bdrv_drained_end(bs); break; + default: g_assert_not_reached(); + } +} + +static void test_drv_cb_common(enum drain_type drain_type, bool recursive) { BlockBackend *blk; - BlockDriverState *bs; - BDRVTestState *s; + BlockDriverState *bs, *backing; + BDRVTestState *s, *backing_s; BlockAIOCB *acb; int aio_ret; @@ -XXX,XX +XXX,XX @@ static void test_drv_cb_drain_all(void) s = bs->opaque; blk_insert_bs(blk, bs, &error_abort); + backing = bdrv_new_open_driver(&bdrv_test, "backing", 0, &error_abort); + backing_s = backing->opaque; + bdrv_set_backing_hd(bs, backing, &error_abort); + /* Simple bdrv_drain_all_begin/end pair, check that CBs are called */ g_assert_cmpint(s->drain_count, ==, 0); - bdrv_drain_all_begin(); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + do_drain_begin(drain_type, bs); + g_assert_cmpint(s->drain_count, ==, 1); - bdrv_drain_all_end(); + g_assert_cmpint(backing_s->drain_count, ==, !!recursive); + + do_drain_end(drain_type, bs); + g_assert_cmpint(s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); /* Now do the same while a request is pending */ aio_ret = -EINPROGRESS; @@ -XXX,XX +XXX,XX @@ static void test_drv_cb_drain_all(void) g_assert_cmpint(aio_ret, ==, -EINPROGRESS); g_assert_cmpint(s->drain_count, ==, 0); - bdrv_drain_all_begin(); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + do_drain_begin(drain_type, bs); + g_assert_cmpint(aio_ret, ==, 0); g_assert_cmpint(s->drain_count, ==, 1); - bdrv_drain_all_end(); + g_assert_cmpint(backing_s->drain_count, ==, !!recursive); + + do_drain_end(drain_type, bs); + g_assert_cmpint(s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + bdrv_unref(backing); bdrv_unref(bs); blk_unref(blk); } +static void test_drv_cb_drain_all(void) +{ + test_drv_cb_common(BDRV_DRAIN_ALL, true); +} + +static void test_drv_cb_drain(void) +{ + test_drv_cb_common(BDRV_DRAIN, false); +} + int main(int argc, char **argv) { bdrv_init(); @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_init(&argc, &argv, NULL); g_test_add_func("/bdrv-drain/driver-cb/drain_all", test_drv_cb_drain_all); + g_test_add_func("/bdrv-drain/driver-cb/drain", test_drv_cb_drain); return g_test_run(); } -- 2.13.6 This is currently only working correctly for bdrv_drain(), not for bdrv_drain_all(). Leave a comment for the drain_all case, we'll address it later. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void test_drv_cb_drain(void) test_drv_cb_common(BDRV_DRAIN, false); } +static void test_quiesce_common(enum drain_type drain_type, bool recursive) +{ + BlockBackend *blk; + BlockDriverState *bs, *backing; + + blk = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs = bdrv_new_open_driver(&bdrv_test, "test-node", BDRV_O_RDWR, + &error_abort); + blk_insert_bs(blk, bs, &error_abort); + + backing = bdrv_new_open_driver(&bdrv_test, "backing", 0, &error_abort); + bdrv_set_backing_hd(bs, backing, &error_abort); + + g_assert_cmpint(bs->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + + do_drain_begin(drain_type, bs); + + g_assert_cmpint(bs->quiesce_counter, ==, 1); + g_assert_cmpint(backing->quiesce_counter, ==, !!recursive); + + do_drain_end(drain_type, bs); + + g_assert_cmpint(bs->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + + bdrv_unref(backing); + bdrv_unref(bs); + blk_unref(blk); +} + +static void test_quiesce_drain_all(void) +{ + // XXX drain_all doesn't quiesce + //test_quiesce_common(BDRV_DRAIN_ALL, true); +} + +static void test_quiesce_drain(void) +{ + test_quiesce_common(BDRV_DRAIN, false); +} + int main(int argc, char **argv) { bdrv_init(); @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/driver-cb/drain_all", test_drv_cb_drain_all); g_test_add_func("/bdrv-drain/driver-cb/drain", test_drv_cb_drain); + g_test_add_func("/bdrv-drain/quiesce/drain_all", test_quiesce_drain_all); + g_test_add_func("/bdrv-drain/quiesce/drain", test_quiesce_drain); + return g_test_run(); } -- 2.13.6 Block jobs already paused themselves when their main BlockBackend entered a drained section. This is not good enough: We also want to pause a block job and may not submit new requests if, for example, the mirror target node should be drained. This implements .drained_begin/end callbacks in child_job in order to consider all block nodes related to the job, and removes the BlockBackend callbacks which are unnecessary now because the root of the job main BlockBackend is always referenced with a child_job, too. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- blockjob.c | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) diff --git a/blockjob.c b/blockjob.c index XXXXXXX..XXXXXXX 100644 --- a/blockjob.c +++ b/blockjob.c @@ -XXX,XX +XXX,XX @@ static char *child_job_get_parent_desc(BdrvChild *c) job->id); } -static const BdrvChildRole child_job = { - .get_parent_desc = child_job_get_parent_desc, - .stay_at_node = true, -}; - -static void block_job_drained_begin(void *opaque) +static void child_job_drained_begin(BdrvChild *c) { - BlockJob *job = opaque; + BlockJob *job = c->opaque; block_job_pause(job); } -static void block_job_drained_end(void *opaque) +static void child_job_drained_end(BdrvChild *c) { - BlockJob *job = opaque; + BlockJob *job = c->opaque; block_job_resume(job); } -static const BlockDevOps block_job_dev_ops = { - .drained_begin = block_job_drained_begin, - .drained_end = block_job_drained_end, +static const BdrvChildRole child_job = { + .get_parent_desc = child_job_get_parent_desc, + .drained_begin = child_job_drained_begin, + .drained_end = child_job_drained_end, + .stay_at_node = true, }; void block_job_remove_all_bdrv(BlockJob *job) @@ -XXX,XX +XXX,XX @@ void *block_job_create(const char *job_id, const BlockJobDriver *driver, block_job_add_bdrv(job, "main node", bs, 0, BLK_PERM_ALL, &error_abort); bs->job = job; - blk_set_dev_ops(blk, &block_job_dev_ops, job); bdrv_op_unblock(bs, BLOCK_OP_TYPE_DATAPLANE, job->blocker); QLIST_INSERT_HEAD(&block_jobs, job, job_list); -- 2.13.6 Block jobs must be paused if any of the involved nodes are drained. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 121 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 121 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "block/block.h" +#include "block/blockjob_int.h" #include "sysemu/block-backend.h" #include "qapi/error.h" @@ -XXX,XX +XXX,XX @@ static void test_quiesce_drain(void) test_quiesce_common(BDRV_DRAIN, false); } + +typedef struct TestBlockJob { + BlockJob common; + bool should_complete; +} TestBlockJob; + +static void test_job_completed(BlockJob *job, void *opaque) +{ + block_job_completed(job, 0); +} + +static void coroutine_fn test_job_start(void *opaque) +{ + TestBlockJob *s = opaque; + + while (!s->should_complete) { + block_job_sleep_ns(&s->common, 100000); + } + + block_job_defer_to_main_loop(&s->common, test_job_completed, NULL); +} + +static void test_job_complete(BlockJob *job, Error **errp) +{ + TestBlockJob *s = container_of(job, TestBlockJob, common); + s->should_complete = true; +} + +BlockJobDriver test_job_driver = { + .instance_size = sizeof(TestBlockJob), + .start = test_job_start, + .complete = test_job_complete, +}; + +static void test_blockjob_common(enum drain_type drain_type) +{ + BlockBackend *blk_src, *blk_target; + BlockDriverState *src, *target; + BlockJob *job; + int ret; + + src = bdrv_new_open_driver(&bdrv_test, "source", BDRV_O_RDWR, + &error_abort); + blk_src = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + blk_insert_bs(blk_src, src, &error_abort); + + target = bdrv_new_open_driver(&bdrv_test, "target", BDRV_O_RDWR, + &error_abort); + blk_target = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + blk_insert_bs(blk_target, target, &error_abort); + + job = block_job_create("job0", &test_job_driver, src, 0, BLK_PERM_ALL, 0, + 0, NULL, NULL, &error_abort); + block_job_add_bdrv(job, "target", target, 0, BLK_PERM_ALL, &error_abort); + block_job_start(job); + + g_assert_cmpint(job->pause_count, ==, 0); + g_assert_false(job->paused); + g_assert_false(job->busy); /* We're in block_job_sleep_ns() */ + + do_drain_begin(drain_type, src); + + if (drain_type == BDRV_DRAIN_ALL) { + /* bdrv_drain_all() drains both src and target, and involves an + * additional block_job_pause_all() */ + g_assert_cmpint(job->pause_count, ==, 3); + } else { + g_assert_cmpint(job->pause_count, ==, 1); + } + /* XXX We don't wait until the job is actually paused. Is this okay? */ + /* g_assert_true(job->paused); */ + g_assert_false(job->busy); /* The job is paused */ + + do_drain_end(drain_type, src); + + g_assert_cmpint(job->pause_count, ==, 0); + g_assert_false(job->paused); + g_assert_false(job->busy); /* We're in block_job_sleep_ns() */ + + do_drain_begin(drain_type, target); + + if (drain_type == BDRV_DRAIN_ALL) { + /* bdrv_drain_all() drains both src and target, and involves an + * additional block_job_pause_all() */ + g_assert_cmpint(job->pause_count, ==, 3); + } else { + g_assert_cmpint(job->pause_count, ==, 1); + } + /* XXX We don't wait until the job is actually paused. Is this okay? */ + /* g_assert_true(job->paused); */ + g_assert_false(job->busy); /* The job is paused */ + + do_drain_end(drain_type, target); + + g_assert_cmpint(job->pause_count, ==, 0); + g_assert_false(job->paused); + g_assert_false(job->busy); /* We're in block_job_sleep_ns() */ + + ret = block_job_complete_sync(job, &error_abort); + g_assert_cmpint(ret, ==, 0); + + blk_unref(blk_src); + blk_unref(blk_target); + bdrv_unref(src); + bdrv_unref(target); +} + +static void test_blockjob_drain_all(void) +{ + test_blockjob_common(BDRV_DRAIN_ALL); +} + +static void test_blockjob_drain(void) +{ + test_blockjob_common(BDRV_DRAIN); +} + int main(int argc, char **argv) { bdrv_init(); @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/quiesce/drain_all", test_quiesce_drain_all); g_test_add_func("/bdrv-drain/quiesce/drain", test_quiesce_drain); + g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); + g_test_add_func("/bdrv-drain/blockjob/drain", test_blockjob_drain); + return g_test_run(); } -- 2.13.6 Block jobs are already paused using the BdrvChildRole drain callbacks, so we don't need an additional block_job_pause_all() call. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/io.c | 4 ---- tests/test-bdrv-drain.c | 10 ++++------ 2 files changed, 4 insertions(+), 10 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) * context. */ assert(qemu_get_current_aio_context() == qemu_get_aio_context()); - block_job_pause_all(); - for (bs = bdrv_first(&it); bs; bs = bdrv_next(&it)) { AioContext *aio_context = bdrv_get_aio_context(bs); @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) aio_enable_external(aio_context); aio_context_release(aio_context); } - - block_job_resume_all(); } void bdrv_drain_all(void) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void test_blockjob_common(enum drain_type drain_type) do_drain_begin(drain_type, src); if (drain_type == BDRV_DRAIN_ALL) { - /* bdrv_drain_all() drains both src and target, and involves an - * additional block_job_pause_all() */ - g_assert_cmpint(job->pause_count, ==, 3); + /* bdrv_drain_all() drains both src and target */ + g_assert_cmpint(job->pause_count, ==, 2); } else { g_assert_cmpint(job->pause_count, ==, 1); } @@ -XXX,XX +XXX,XX @@ static void test_blockjob_common(enum drain_type drain_type) do_drain_begin(drain_type, target); if (drain_type == BDRV_DRAIN_ALL) { - /* bdrv_drain_all() drains both src and target, and involves an - * additional block_job_pause_all() */ - g_assert_cmpint(job->pause_count, ==, 3); + /* bdrv_drain_all() drains both src and target */ + g_assert_cmpint(job->pause_count, ==, 2); } else { g_assert_cmpint(job->pause_count, ==, 1); } -- 2.13.6 bdrv_do_drained_begin() restricts the call of parent callbacks and aio_disable_external() to the outermost drain section, but the block driver callbacks are always called. bdrv_do_drained_end() must match this behaviour, otherwise nodes stay drained even if begin/end calls were balanced. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/io.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs) void bdrv_drained_end(BlockDriverState *bs) { + int old_quiesce_counter; + if (qemu_in_coroutine()) { bdrv_co_yield_to_drain(bs, false); return; } assert(bs->quiesce_counter > 0); - if (atomic_fetch_dec(&bs->quiesce_counter) > 1) { - return; - } + old_quiesce_counter = atomic_fetch_dec(&bs->quiesce_counter); /* Re-enable things in child-to-parent order */ bdrv_drain_invoke(bs, false, false); - bdrv_parent_drained_end(bs); - aio_enable_external(bdrv_get_aio_context(bs)); + if (old_quiesce_counter == 1) { + bdrv_parent_drained_end(bs); + aio_enable_external(bdrv_get_aio_context(bs)); + } } /* -- 2.13.6 Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 57 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 57 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void aio_ret_cb(void *opaque, int ret) enum drain_type { BDRV_DRAIN_ALL, BDRV_DRAIN, + DRAIN_TYPE_MAX, }; static void do_drain_begin(enum drain_type drain_type, BlockDriverState *bs) @@ -XXX,XX +XXX,XX @@ static void test_quiesce_drain(void) test_quiesce_common(BDRV_DRAIN, false); } +static void test_nested(void) +{ + BlockBackend *blk; + BlockDriverState *bs, *backing; + BDRVTestState *s, *backing_s; + enum drain_type outer, inner; + + blk = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs = bdrv_new_open_driver(&bdrv_test, "test-node", BDRV_O_RDWR, + &error_abort); + s = bs->opaque; + blk_insert_bs(blk, bs, &error_abort); + + backing = bdrv_new_open_driver(&bdrv_test, "backing", 0, &error_abort); + backing_s = backing->opaque; + bdrv_set_backing_hd(bs, backing, &error_abort); + + for (outer = 0; outer < DRAIN_TYPE_MAX; outer++) { + for (inner = 0; inner < DRAIN_TYPE_MAX; inner++) { + /* XXX bdrv_drain_all() doesn't increase the quiesce_counter */ + int bs_quiesce = (outer != BDRV_DRAIN_ALL) + + (inner != BDRV_DRAIN_ALL); + int backing_quiesce = 0; + int backing_cb_cnt = (outer != BDRV_DRAIN) + + (inner != BDRV_DRAIN); + + g_assert_cmpint(bs->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + do_drain_begin(outer, bs); + do_drain_begin(inner, bs); + + g_assert_cmpint(bs->quiesce_counter, ==, bs_quiesce); + g_assert_cmpint(backing->quiesce_counter, ==, backing_quiesce); + g_assert_cmpint(s->drain_count, ==, 2); + g_assert_cmpint(backing_s->drain_count, ==, backing_cb_cnt); + + do_drain_end(inner, bs); + do_drain_end(outer, bs); + + g_assert_cmpint(bs->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + } + } + + bdrv_unref(backing); + bdrv_unref(bs); + blk_unref(blk); +} + typedef struct TestBlockJob { BlockJob common; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/quiesce/drain_all", test_quiesce_drain_all); g_test_add_func("/bdrv-drain/quiesce/drain", test_quiesce_drain); + g_test_add_func("/bdrv-drain/nested", test_nested); + g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); g_test_add_func("/bdrv-drain/blockjob/drain", test_blockjob_drain); -- 2.13.6 This is in preparation for subtree drains, i.e. drained sections that affect not only a single node, but recursively all child nodes, too. Calling the parent callbacks for drain is pointless when we just came from that parent node recursively and leads to multiple increases of bs->quiesce_counter in a single drain call. Don't do it. In order for this to work correctly, the parent callback must be called for every bdrv_drain_begin/end() call, not only for the outermost one: If we have a node N with two parents A and B, recursive draining of A should cause the quiesce_counter of B to increase because its child N is drained independently of B. If now B is recursively drained, too, A must increase its quiesce_counter because N is drained independently of A only now, even if N is going from quiesce_counter 1 to 2. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 4 ++-- block.c | 13 +++++++++---- block/io.c | 47 ++++++++++++++++++++++++++++++++++------------- 3 files changed, 45 insertions(+), 19 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -XXX,XX +XXX,XX @@ void bdrv_io_unplug(BlockDriverState *bs); * Begin a quiesced section of all users of @bs. This is part of * bdrv_drained_begin. */ -void bdrv_parent_drained_begin(BlockDriverState *bs); +void bdrv_parent_drained_begin(BlockDriverState *bs, BdrvChild *ignore); /** * bdrv_parent_drained_end: @@ -XXX,XX +XXX,XX @@ void bdrv_parent_drained_begin(BlockDriverState *bs); * End a quiesced section of all users of @bs. This is part of * bdrv_drained_end. */ -void bdrv_parent_drained_end(BlockDriverState *bs); +void bdrv_parent_drained_end(BlockDriverState *bs, BdrvChild *ignore); /** * bdrv_drained_begin: diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild *child, BlockDriverState *new_bs) { BlockDriverState *old_bs = child->bs; + int i; if (old_bs && new_bs) { assert(bdrv_get_aio_context(old_bs) == bdrv_get_aio_context(new_bs)); } if (old_bs) { if (old_bs->quiesce_counter && child->role->drained_end) { - child->role->drained_end(child); + for (i = 0; i < old_bs->quiesce_counter; i++) { + child->role->drained_end(child); + } } if (child->role->detach) { child->role->detach(child); @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild *child, if (new_bs) { QLIST_INSERT_HEAD(&new_bs->parents, child, next_parent); if (new_bs->quiesce_counter && child->role->drained_begin) { - child->role->drained_begin(child); + for (i = 0; i < new_bs->quiesce_counter; i++) { + child->role->drained_begin(child); + } } if (child->role->attach) { @@ -XXX,XX +XXX,XX @@ void bdrv_set_aio_context(BlockDriverState *bs, AioContext *new_context) AioContext *ctx = bdrv_get_aio_context(bs); aio_disable_external(ctx); - bdrv_parent_drained_begin(bs); + bdrv_parent_drained_begin(bs, NULL); bdrv_drain(bs); /* ensure there are no in-flight requests */ while (aio_poll(ctx, false)) { @@ -XXX,XX +XXX,XX @@ void bdrv_set_aio_context(BlockDriverState *bs, AioContext *new_context) */ aio_context_acquire(new_context); bdrv_attach_aio_context(bs, new_context); - bdrv_parent_drained_end(bs); + bdrv_parent_drained_end(bs, NULL); aio_enable_external(ctx); aio_context_release(new_context); } diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int bytes, BdrvRequestFlags flags); -void bdrv_parent_drained_begin(BlockDriverState *bs) +void bdrv_parent_drained_begin(BlockDriverState *bs, BdrvChild *ignore) { BdrvChild *c, *next; QLIST_FOREACH_SAFE(c, &bs->parents, next_parent, next) { + if (c == ignore) { + continue; + } if (c->role->drained_begin) { c->role->drained_begin(c); } } } -void bdrv_parent_drained_end(BlockDriverState *bs) +void bdrv_parent_drained_end(BlockDriverState *bs, BdrvChild *ignore) { BdrvChild *c, *next; QLIST_FOREACH_SAFE(c, &bs->parents, next_parent, next) { + if (c == ignore) { + continue; + } if (c->role->drained_end) { c->role->drained_end(c); } @@ -XXX,XX +XXX,XX @@ typedef struct { BlockDriverState *bs; bool done; bool begin; + BdrvChild *parent; } BdrvCoDrainData; static void coroutine_fn bdrv_drain_invoke_entry(void *opaque) @@ -XXX,XX +XXX,XX @@ static bool bdrv_drain_recurse(BlockDriverState *bs) return waited; } +static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent); +static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent); + static void bdrv_co_drain_bh_cb(void *opaque) { BdrvCoDrainData *data = opaque; @@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque) bdrv_dec_in_flight(bs); if (data->begin) { - bdrv_drained_begin(bs); + bdrv_do_drained_begin(bs, data->parent); } else { - bdrv_drained_end(bs); + bdrv_do_drained_end(bs, data->parent); } data->done = true; @@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque) } static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, - bool begin) + bool begin, BdrvChild *parent) { BdrvCoDrainData data; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, .bs = bs, .done = false, .begin = begin, + .parent = parent, }; bdrv_inc_in_flight(bs); aio_bh_schedule_oneshot(bdrv_get_aio_context(bs), @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, assert(data.done); } -void bdrv_drained_begin(BlockDriverState *bs) +static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent) { if (qemu_in_coroutine()) { - bdrv_co_yield_to_drain(bs, true); + bdrv_co_yield_to_drain(bs, true, parent); return; } /* Stop things in parent-to-child order */ if (atomic_fetch_inc(&bs->quiesce_counter) == 0) { aio_disable_external(bdrv_get_aio_context(bs)); - bdrv_parent_drained_begin(bs); } + bdrv_parent_drained_begin(bs, parent); bdrv_drain_invoke(bs, true, false); bdrv_drain_recurse(bs); } -void bdrv_drained_end(BlockDriverState *bs) +void bdrv_drained_begin(BlockDriverState *bs) +{ + bdrv_do_drained_begin(bs, NULL); +} + +static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent) { int old_quiesce_counter; if (qemu_in_coroutine()) { - bdrv_co_yield_to_drain(bs, false); + bdrv_co_yield_to_drain(bs, false, parent); return; } assert(bs->quiesce_counter > 0); @@ -XXX,XX +XXX,XX @@ void bdrv_drained_end(BlockDriverState *bs) /* Re-enable things in child-to-parent order */ bdrv_drain_invoke(bs, false, false); + bdrv_parent_drained_end(bs, parent); if (old_quiesce_counter == 1) { - bdrv_parent_drained_end(bs); aio_enable_external(bdrv_get_aio_context(bs)); } } +void bdrv_drained_end(BlockDriverState *bs) +{ + bdrv_do_drained_end(bs, NULL); +} + /* * Wait for pending requests to complete on a single BlockDriverState subtree, * and suspend block driver's internal I/O until next request arrives. @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_begin(void) /* Stop things in parent-to-child order */ aio_context_acquire(aio_context); aio_disable_external(aio_context); - bdrv_parent_drained_begin(bs); + bdrv_parent_drained_begin(bs, NULL); bdrv_drain_invoke(bs, true, true); aio_context_release(aio_context); @@ -XXX,XX +XXX,XX @@ void bdrv_drain_all_end(void) /* Re-enable things in child-to-parent order */ aio_context_acquire(aio_context); bdrv_drain_invoke(bs, false, true); - bdrv_parent_drained_end(bs); + bdrv_parent_drained_end(bs, NULL); aio_enable_external(aio_context); aio_context_release(aio_context); } -- 2.13.6 bdrv_drained_begin() waits for the completion of requests in the whole subtree, but it only actually keeps its immediate bs parameter quiesced until bdrv_drained_end(). Add a version that keeps the whole subtree drained. As of this commit, graph changes cannot be allowed during a subtree drained section, but this will be fixed soon. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 13 +++++++++++++ block/io.c | 54 ++++++++++++++++++++++++++++++++++++++++----------- 2 files changed, 56 insertions(+), 11 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -XXX,XX +XXX,XX @@ void bdrv_parent_drained_end(BlockDriverState *bs, BdrvChild *ignore); void bdrv_drained_begin(BlockDriverState *bs); /** + * Like bdrv_drained_begin, but recursively begins a quiesced section for + * exclusive access to all child nodes as well. + * + * Graph changes are not allowed during a subtree drain section. + */ +void bdrv_subtree_drained_begin(BlockDriverState *bs); + +/** * bdrv_drained_end: * * End a quiescent section started by bdrv_drained_begin(). */ void bdrv_drained_end(BlockDriverState *bs); +/** + * End a quiescent section started by bdrv_subtree_drained_begin(). + */ +void bdrv_subtree_drained_end(BlockDriverState *bs); + void bdrv_add_child(BlockDriverState *parent, BlockDriverState *child, Error **errp); void bdrv_del_child(BlockDriverState *parent, BdrvChild *child, Error **errp); diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ typedef struct { BlockDriverState *bs; bool done; bool begin; + bool recursive; BdrvChild *parent; } BdrvCoDrainData; @@ -XXX,XX +XXX,XX @@ static bool bdrv_drain_recurse(BlockDriverState *bs) return waited; } -static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent); -static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent); +static void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, + BdrvChild *parent); +static void bdrv_do_drained_end(BlockDriverState *bs, bool recursive, + BdrvChild *parent); static void bdrv_co_drain_bh_cb(void *opaque) { @@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque) bdrv_dec_in_flight(bs); if (data->begin) { - bdrv_do_drained_begin(bs, data->parent); + bdrv_do_drained_begin(bs, data->recursive, data->parent); } else { - bdrv_do_drained_end(bs, data->parent); + bdrv_do_drained_end(bs, data->recursive, data->parent); } data->done = true; @@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque) } static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, - bool begin, BdrvChild *parent) + bool begin, bool recursive, + BdrvChild *parent) { BdrvCoDrainData data; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, .bs = bs, .done = false, .begin = begin, + .recursive = recursive, .parent = parent, }; bdrv_inc_in_flight(bs); @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, assert(data.done); } -static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent) +static void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, + BdrvChild *parent) { + BdrvChild *child, *next; + if (qemu_in_coroutine()) { - bdrv_co_yield_to_drain(bs, true, parent); + bdrv_co_yield_to_drain(bs, true, recursive, parent); return; } @@ -XXX,XX +XXX,XX @@ static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent) bdrv_parent_drained_begin(bs, parent); bdrv_drain_invoke(bs, true, false); bdrv_drain_recurse(bs); + + if (recursive) { + QLIST_FOREACH_SAFE(child, &bs->children, next, next) { + bdrv_do_drained_begin(child->bs, true, child); + } + } } void bdrv_drained_begin(BlockDriverState *bs) { - bdrv_do_drained_begin(bs, NULL); + bdrv_do_drained_begin(bs, false, NULL); +} + +void bdrv_subtree_drained_begin(BlockDriverState *bs) +{ + bdrv_do_drained_begin(bs, true, NULL); } -static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent) +static void bdrv_do_drained_end(BlockDriverState *bs, bool recursive, + BdrvChild *parent) { + BdrvChild *child, *next; int old_quiesce_counter; if (qemu_in_coroutine()) { - bdrv_co_yield_to_drain(bs, false, parent); + bdrv_co_yield_to_drain(bs, false, recursive, parent); return; } assert(bs->quiesce_counter > 0); @@ -XXX,XX +XXX,XX @@ static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent) if (old_quiesce_counter == 1) { aio_enable_external(bdrv_get_aio_context(bs)); } + + if (recursive) { + QLIST_FOREACH_SAFE(child, &bs->children, next, next) { + bdrv_do_drained_end(child->bs, true, child); + } + } } void bdrv_drained_end(BlockDriverState *bs) { - bdrv_do_drained_end(bs, NULL); + bdrv_do_drained_end(bs, false, NULL); +} + +void bdrv_subtree_drained_end(BlockDriverState *bs) +{ + bdrv_do_drained_end(bs, true, NULL); } /* -- 2.13.6 Add a subtree drain version to the existing test cases. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 27 ++++++++++++++++++++++++++- 1 file changed, 26 insertions(+), 1 deletion(-) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void aio_ret_cb(void *opaque, int ret) enum drain_type { BDRV_DRAIN_ALL, BDRV_DRAIN, + BDRV_SUBTREE_DRAIN, DRAIN_TYPE_MAX, }; @@ -XXX,XX +XXX,XX @@ static void do_drain_begin(enum drain_type drain_type, BlockDriverState *bs) switch (drain_type) { case BDRV_DRAIN_ALL: bdrv_drain_all_begin(); break; case BDRV_DRAIN: bdrv_drained_begin(bs); break; + case BDRV_SUBTREE_DRAIN: bdrv_subtree_drained_begin(bs); break; default: g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static void do_drain_end(enum drain_type drain_type, BlockDriverState *bs) switch (drain_type) { case BDRV_DRAIN_ALL: bdrv_drain_all_end(); break; case BDRV_DRAIN: bdrv_drained_end(bs); break; + case BDRV_SUBTREE_DRAIN: bdrv_subtree_drained_end(bs); break; default: g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static void test_drv_cb_drain(void) test_drv_cb_common(BDRV_DRAIN, false); } +static void test_drv_cb_drain_subtree(void) +{ + test_drv_cb_common(BDRV_SUBTREE_DRAIN, true); +} + static void test_quiesce_common(enum drain_type drain_type, bool recursive) { BlockBackend *blk; @@ -XXX,XX +XXX,XX @@ static void test_quiesce_drain(void) test_quiesce_common(BDRV_DRAIN, false); } +static void test_quiesce_drain_subtree(void) +{ + test_quiesce_common(BDRV_SUBTREE_DRAIN, true); +} + static void test_nested(void) { BlockBackend *blk; @@ -XXX,XX +XXX,XX @@ static void test_nested(void) /* XXX bdrv_drain_all() doesn't increase the quiesce_counter */ int bs_quiesce = (outer != BDRV_DRAIN_ALL) + (inner != BDRV_DRAIN_ALL); - int backing_quiesce = 0; + int backing_quiesce = (outer == BDRV_SUBTREE_DRAIN) + + (inner == BDRV_SUBTREE_DRAIN); int backing_cb_cnt = (outer != BDRV_DRAIN) + (inner != BDRV_DRAIN); @@ -XXX,XX +XXX,XX @@ static void test_blockjob_drain(void) test_blockjob_common(BDRV_DRAIN); } +static void test_blockjob_drain_subtree(void) +{ + test_blockjob_common(BDRV_SUBTREE_DRAIN); +} + int main(int argc, char **argv) { bdrv_init(); @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/driver-cb/drain_all", test_drv_cb_drain_all); g_test_add_func("/bdrv-drain/driver-cb/drain", test_drv_cb_drain); + g_test_add_func("/bdrv-drain/driver-cb/drain_subtree", + test_drv_cb_drain_subtree); g_test_add_func("/bdrv-drain/quiesce/drain_all", test_quiesce_drain_all); g_test_add_func("/bdrv-drain/quiesce/drain", test_quiesce_drain); + g_test_add_func("/bdrv-drain/quiesce/drain_subtree", + test_quiesce_drain_subtree); g_test_add_func("/bdrv-drain/nested", test_nested); g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); g_test_add_func("/bdrv-drain/blockjob/drain", test_blockjob_drain); + g_test_add_func("/bdrv-drain/blockjob/drain_subtree", + test_blockjob_drain_subtree); return g_test_run(); } -- 2.13.6 If bdrv_do_drained_begin/end() are called in coroutine context, they first use a BH to get out of the coroutine context. Call some existing tests again from a coroutine to cover this code path. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void aio_ret_cb(void *opaque, int ret) *aio_ret = ret; } +typedef struct CallInCoroutineData { + void (*entry)(void); + bool done; +} CallInCoroutineData; + +static coroutine_fn void call_in_coroutine_entry(void *opaque) +{ + CallInCoroutineData *data = opaque; + + data->entry(); + data->done = true; +} + +static void call_in_coroutine(void (*entry)(void)) +{ + Coroutine *co; + CallInCoroutineData data = { + .entry = entry, + .done = false, + }; + + co = qemu_coroutine_create(call_in_coroutine_entry, &data); + qemu_coroutine_enter(co); + while (!data.done) { + aio_poll(qemu_get_aio_context(), true); + } +} + enum drain_type { BDRV_DRAIN_ALL, BDRV_DRAIN, @@ -XXX,XX +XXX,XX @@ static void test_drv_cb_drain_subtree(void) test_drv_cb_common(BDRV_SUBTREE_DRAIN, true); } +static void test_drv_cb_co_drain(void) +{ + call_in_coroutine(test_drv_cb_drain); +} + +static void test_drv_cb_co_drain_subtree(void) +{ + call_in_coroutine(test_drv_cb_drain_subtree); +} + static void test_quiesce_common(enum drain_type drain_type, bool recursive) { BlockBackend *blk; @@ -XXX,XX +XXX,XX @@ static void test_quiesce_drain_subtree(void) test_quiesce_common(BDRV_SUBTREE_DRAIN, true); } +static void test_quiesce_co_drain(void) +{ + call_in_coroutine(test_quiesce_drain); +} + +static void test_quiesce_co_drain_subtree(void) +{ + call_in_coroutine(test_quiesce_drain_subtree); +} + static void test_nested(void) { BlockBackend *blk; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/driver-cb/drain_subtree", test_drv_cb_drain_subtree); + // XXX bdrv_drain_all() doesn't work in coroutine context + g_test_add_func("/bdrv-drain/driver-cb/co/drain", test_drv_cb_co_drain); + g_test_add_func("/bdrv-drain/driver-cb/co/drain_subtree", + test_drv_cb_co_drain_subtree); + + g_test_add_func("/bdrv-drain/quiesce/drain_all", test_quiesce_drain_all); g_test_add_func("/bdrv-drain/quiesce/drain", test_quiesce_drain); g_test_add_func("/bdrv-drain/quiesce/drain_subtree", test_quiesce_drain_subtree); + // XXX bdrv_drain_all() doesn't work in coroutine context + g_test_add_func("/bdrv-drain/quiesce/co/drain", test_quiesce_co_drain); + g_test_add_func("/bdrv-drain/quiesce/co/drain_subtree", + test_quiesce_co_drain_subtree); + g_test_add_func("/bdrv-drain/nested", test_nested); g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); -- 2.13.6 Test that drain sections are correctly propagated through the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 74 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 74 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void test_nested(void) blk_unref(blk); } +static void test_multiparent(void) +{ + BlockBackend *blk_a, *blk_b; + BlockDriverState *bs_a, *bs_b, *backing; + BDRVTestState *a_s, *b_s, *backing_s; + + blk_a = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs_a = bdrv_new_open_driver(&bdrv_test, "test-node-a", BDRV_O_RDWR, + &error_abort); + a_s = bs_a->opaque; + blk_insert_bs(blk_a, bs_a, &error_abort); + + blk_b = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs_b = bdrv_new_open_driver(&bdrv_test, "test-node-b", BDRV_O_RDWR, + &error_abort); + b_s = bs_b->opaque; + blk_insert_bs(blk_b, bs_b, &error_abort); + + backing = bdrv_new_open_driver(&bdrv_test, "backing", 0, &error_abort); + backing_s = backing->opaque; + bdrv_set_backing_hd(bs_a, backing, &error_abort); + bdrv_set_backing_hd(bs_b, backing, &error_abort); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 0); + g_assert_cmpint(bs_b->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(a_s->drain_count, ==, 0); + g_assert_cmpint(b_s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_a); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 1); + g_assert_cmpint(bs_b->quiesce_counter, ==, 1); + g_assert_cmpint(backing->quiesce_counter, ==, 1); + g_assert_cmpint(a_s->drain_count, ==, 1); + g_assert_cmpint(b_s->drain_count, ==, 1); + g_assert_cmpint(backing_s->drain_count, ==, 1); + + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_b); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 2); + g_assert_cmpint(bs_b->quiesce_counter, ==, 2); + g_assert_cmpint(backing->quiesce_counter, ==, 2); + g_assert_cmpint(a_s->drain_count, ==, 2); + g_assert_cmpint(b_s->drain_count, ==, 2); + g_assert_cmpint(backing_s->drain_count, ==, 2); + + do_drain_end(BDRV_SUBTREE_DRAIN, bs_b); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 1); + g_assert_cmpint(bs_b->quiesce_counter, ==, 1); + g_assert_cmpint(backing->quiesce_counter, ==, 1); + g_assert_cmpint(a_s->drain_count, ==, 1); + g_assert_cmpint(b_s->drain_count, ==, 1); + g_assert_cmpint(backing_s->drain_count, ==, 1); + + do_drain_end(BDRV_SUBTREE_DRAIN, bs_a); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 0); + g_assert_cmpint(bs_b->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(a_s->drain_count, ==, 0); + g_assert_cmpint(b_s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + bdrv_unref(backing); + bdrv_unref(bs_a); + bdrv_unref(bs_b); + blk_unref(blk_a); + blk_unref(blk_b); +} + typedef struct TestBlockJob { BlockJob common; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) test_quiesce_co_drain_subtree); g_test_add_func("/bdrv-drain/nested", test_nested); + g_test_add_func("/bdrv-drain/multiparent", test_multiparent); g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); g_test_add_func("/bdrv-drain/blockjob/drain", test_blockjob_drain); -- 2.13.6 We need to remember how many of the drain sections in which a node is were recursive (i.e. subtree drain rather than node drain), so that they can be correctly applied when children are added or removed during the drained section. With this change, it is safe to modify the graph even inside a bdrv_subtree_drained_begin/end() section. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 2 -- include/block/block_int.h | 5 +++++ block.c | 32 +++++++++++++++++++++++++++++--- block/io.c | 28 ++++++++++++++++++++++++---- 4 files changed, 58 insertions(+), 9 deletions(-) diff --git a/include/block/block.h b/include/block/block.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -XXX,XX +XXX,XX @@ void bdrv_drained_begin(BlockDriverState *bs); /** * Like bdrv_drained_begin, but recursively begins a quiesced section for * exclusive access to all child nodes as well. - * - * Graph changes are not allowed during a subtree drain section. */ void bdrv_subtree_drained_begin(BlockDriverState *bs); diff --git a/include/block/block_int.h b/include/block/block_int.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -XXX,XX +XXX,XX @@ struct BlockDriverState { /* Accessed with atomic ops. */ int quiesce_counter; + int recursive_quiesce_counter; + unsigned int write_gen; /* Current data generation */ /* Protected by reqs_lock. */ @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwritev(BdrvChild *child, int64_t offset, unsigned int bytes, QEMUIOVector *qiov, BdrvRequestFlags flags); +void bdrv_apply_subtree_drain(BdrvChild *child, BlockDriverState *new_parent); +void bdrv_unapply_subtree_drain(BdrvChild *child, BlockDriverState *old_parent); + int get_tmp_filename(char *filename, int size); BlockDriver *bdrv_probe_all(const uint8_t *buf, int buf_size, const char *filename); diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_child_cb_drained_end(BdrvChild *child) bdrv_drained_end(bs); } +static void bdrv_child_cb_attach(BdrvChild *child) +{ + BlockDriverState *bs = child->opaque; + bdrv_apply_subtree_drain(child, bs); +} + +static void bdrv_child_cb_detach(BdrvChild *child) +{ + BlockDriverState *bs = child->opaque; + bdrv_unapply_subtree_drain(child, bs); +} + static int bdrv_child_cb_inactivate(BdrvChild *child) { BlockDriverState *bs = child->opaque; @@ -XXX,XX +XXX,XX @@ const BdrvChildRole child_file = { .inherit_options = bdrv_inherited_options, .drained_begin = bdrv_child_cb_drained_begin, .drained_end = bdrv_child_cb_drained_end, + .attach = bdrv_child_cb_attach, + .detach = bdrv_child_cb_detach, .inactivate = bdrv_child_cb_inactivate, }; @@ -XXX,XX +XXX,XX @@ const BdrvChildRole child_format = { .inherit_options = bdrv_inherited_fmt_options, .drained_begin = bdrv_child_cb_drained_begin, .drained_end = bdrv_child_cb_drained_end, + .attach = bdrv_child_cb_attach, + .detach = bdrv_child_cb_detach, .inactivate = bdrv_child_cb_inactivate, }; @@ -XXX,XX +XXX,XX @@ static void bdrv_backing_attach(BdrvChild *c) parent->backing_blocker); bdrv_op_unblock(backing_hd, BLOCK_OP_TYPE_BACKUP_TARGET, parent->backing_blocker); + + bdrv_child_cb_attach(c); } static void bdrv_backing_detach(BdrvChild *c) @@ -XXX,XX +XXX,XX @@ static void bdrv_backing_detach(BdrvChild *c) bdrv_op_unblock_all(c->bs, parent->backing_blocker); error_free(parent->backing_blocker); parent->backing_blocker = NULL; + + bdrv_child_cb_detach(c); } /* @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild *child, assert(bdrv_get_aio_context(old_bs) == bdrv_get_aio_context(new_bs)); } 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->role->detach) { + child->role->detach(child); + } if (old_bs->quiesce_counter && child->role->drained_end) { for (i = 0; i < old_bs->quiesce_counter; i++) { child->role->drained_end(child); } } - if (child->role->detach) { - child->role->detach(child); - } QLIST_REMOVE(child, next_parent); } @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild *child, } } + /* 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->role->attach) { child->role->attach(child); } diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs, assert(data.done); } -static void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, - BdrvChild *parent) +void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, + BdrvChild *parent) { BdrvChild *child, *next; @@ -XXX,XX +XXX,XX @@ static void bdrv_do_drained_begin(BlockDriverState *bs, bool recursive, bdrv_drain_recurse(bs); if (recursive) { + bs->recursive_quiesce_counter++; QLIST_FOREACH_SAFE(child, &bs->children, next, next) { bdrv_do_drained_begin(child->bs, true, child); } @@ -XXX,XX +XXX,XX @@ void bdrv_subtree_drained_begin(BlockDriverState *bs) bdrv_do_drained_begin(bs, true, NULL); } -static void bdrv_do_drained_end(BlockDriverState *bs, bool recursive, - BdrvChild *parent) +void bdrv_do_drained_end(BlockDriverState *bs, bool recursive, + BdrvChild *parent) { BdrvChild *child, *next; int old_quiesce_counter; @@ -XXX,XX +XXX,XX @@ static void bdrv_do_drained_end(BlockDriverState *bs, bool recursive, } if (recursive) { + bs->recursive_quiesce_counter--; QLIST_FOREACH_SAFE(child, &bs->children, next, next) { bdrv_do_drained_end(child->bs, true, child); } @@ -XXX,XX +XXX,XX @@ void bdrv_subtree_drained_end(BlockDriverState *bs) bdrv_do_drained_end(bs, true, NULL); } +void bdrv_apply_subtree_drain(BdrvChild *child, BlockDriverState *new_parent) +{ + int i; + + for (i = 0; i < new_parent->recursive_quiesce_counter; i++) { + bdrv_do_drained_begin(child->bs, true, child); + } +} + +void bdrv_unapply_subtree_drain(BdrvChild *child, BlockDriverState *old_parent) +{ + int i; + + for (i = 0; i < old_parent->recursive_quiesce_counter; i++) { + bdrv_do_drained_end(child->bs, true, child); + } +} + /* * Wait for pending requests to complete on a single BlockDriverState subtree, * and suspend block driver's internal I/O until next request arrives. -- 2.13.6 Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/test-bdrv-drain.c | 80 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 80 insertions(+) diff --git a/tests/test-bdrv-drain.c b/tests/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/test-bdrv-drain.c +++ b/tests/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void test_multiparent(void) blk_unref(blk_b); } +static void test_graph_change(void) +{ + BlockBackend *blk_a, *blk_b; + BlockDriverState *bs_a, *bs_b, *backing; + BDRVTestState *a_s, *b_s, *backing_s; + + blk_a = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs_a = bdrv_new_open_driver(&bdrv_test, "test-node-a", BDRV_O_RDWR, + &error_abort); + a_s = bs_a->opaque; + blk_insert_bs(blk_a, bs_a, &error_abort); + + blk_b = blk_new(BLK_PERM_ALL, BLK_PERM_ALL); + bs_b = bdrv_new_open_driver(&bdrv_test, "test-node-b", BDRV_O_RDWR, + &error_abort); + b_s = bs_b->opaque; + blk_insert_bs(blk_b, bs_b, &error_abort); + + backing = bdrv_new_open_driver(&bdrv_test, "backing", 0, &error_abort); + backing_s = backing->opaque; + bdrv_set_backing_hd(bs_a, backing, &error_abort); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 0); + g_assert_cmpint(bs_b->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(a_s->drain_count, ==, 0); + g_assert_cmpint(b_s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_a); + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_a); + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_a); + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_b); + do_drain_begin(BDRV_SUBTREE_DRAIN, bs_b); + + bdrv_set_backing_hd(bs_b, backing, &error_abort); + g_assert_cmpint(bs_a->quiesce_counter, ==, 5); + g_assert_cmpint(bs_b->quiesce_counter, ==, 5); + g_assert_cmpint(backing->quiesce_counter, ==, 5); + g_assert_cmpint(a_s->drain_count, ==, 5); + g_assert_cmpint(b_s->drain_count, ==, 5); + g_assert_cmpint(backing_s->drain_count, ==, 5); + + bdrv_set_backing_hd(bs_b, NULL, &error_abort); + g_assert_cmpint(bs_a->quiesce_counter, ==, 3); + g_assert_cmpint(bs_b->quiesce_counter, ==, 2); + g_assert_cmpint(backing->quiesce_counter, ==, 3); + g_assert_cmpint(a_s->drain_count, ==, 3); + g_assert_cmpint(b_s->drain_count, ==, 2); + g_assert_cmpint(backing_s->drain_count, ==, 3); + + bdrv_set_backing_hd(bs_b, backing, &error_abort); + g_assert_cmpint(bs_a->quiesce_counter, ==, 5); + g_assert_cmpint(bs_b->quiesce_counter, ==, 5); + g_assert_cmpint(backing->quiesce_counter, ==, 5); + g_assert_cmpint(a_s->drain_count, ==, 5); + g_assert_cmpint(b_s->drain_count, ==, 5); + g_assert_cmpint(backing_s->drain_count, ==, 5); + + do_drain_end(BDRV_SUBTREE_DRAIN, bs_b); + do_drain_end(BDRV_SUBTREE_DRAIN, bs_b); + do_drain_end(BDRV_SUBTREE_DRAIN, bs_a); + do_drain_end(BDRV_SUBTREE_DRAIN, bs_a); + do_drain_end(BDRV_SUBTREE_DRAIN, bs_a); + + g_assert_cmpint(bs_a->quiesce_counter, ==, 0); + g_assert_cmpint(bs_b->quiesce_counter, ==, 0); + g_assert_cmpint(backing->quiesce_counter, ==, 0); + g_assert_cmpint(a_s->drain_count, ==, 0); + g_assert_cmpint(b_s->drain_count, ==, 0); + g_assert_cmpint(backing_s->drain_count, ==, 0); + + bdrv_unref(backing); + bdrv_unref(bs_a); + bdrv_unref(bs_b); + blk_unref(blk_a); + blk_unref(blk_b); +} + typedef struct TestBlockJob { BlockJob common; @@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv) g_test_add_func("/bdrv-drain/nested", test_nested); g_test_add_func("/bdrv-drain/multiparent", test_multiparent); + g_test_add_func("/bdrv-drain/graph-change", test_graph_change); g_test_add_func("/bdrv-drain/blockjob/drain_all", test_blockjob_drain_all); g_test_add_func("/bdrv-drain/blockjob/drain", test_blockjob_drain); -- 2.13.6 Since commit bde70715, base is the only node that is reopened in commit_start(). This means that the code, which still involves an explicit BlockReopenQueue, can now be simplified by using bdrv_reopen(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Fam Zheng <famz@redhat.com> --- block/commit.c | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) diff --git a/block/commit.c b/block/commit.c index XXXXXXX..XXXXXXX 100644 --- a/block/commit.c +++ b/block/commit.c @@ -XXX,XX +XXX,XX @@ void commit_start(const char *job_id, BlockDriverState *bs, const char *filter_node_name, Error **errp) { CommitBlockJob *s; - BlockReopenQueue *reopen_queue = NULL; int orig_base_flags; BlockDriverState *iter; BlockDriverState *commit_top_bs = NULL; @@ -XXX,XX +XXX,XX @@ void commit_start(const char *job_id, BlockDriverState *bs, /* convert base to r/w, if necessary */ orig_base_flags = bdrv_get_flags(base); if (!(orig_base_flags & BDRV_O_RDWR)) { - reopen_queue = bdrv_reopen_queue(reopen_queue, base, NULL, - orig_base_flags | BDRV_O_RDWR); - } - - if (reopen_queue) { - bdrv_reopen_multiple(bdrv_get_aio_context(bs), reopen_queue, &local_err); + bdrv_reopen(base, orig_base_flags | BDRV_O_RDWR, &local_err); if (local_err != NULL) { error_propagate(errp, local_err); goto fail; -- 2.13.6 The bdrv_reopen*() implementation doesn't like it if the graph is changed between queuing nodes for reopen and actually reopening them (one of the reasons is that queuing can be recursive). So instead of draining the device only in bdrv_reopen_multiple(), require that callers already drained all affected nodes, and assert this in bdrv_reopen_queue(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Fam Zheng <famz@redhat.com> --- block.c | 23 ++++++++++++++++------- block/replication.c | 6 ++++++ qemu-io-cmds.c | 3 +++ 3 files changed, 25 insertions(+), 7 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ BlockDriverState *bdrv_open(const char *filename, const char *reference, * returns a pointer to bs_queue, which is either the newly allocated * bs_queue, or the existing bs_queue being used. * + * bs must be drained between bdrv_reopen_queue() and bdrv_reopen_multiple(). */ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, BlockDriverState *bs, @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, BdrvChild *child; QDict *old_options, *explicit_options; + /* Make sure that the caller remembered to use a drained section. This is + * important to avoid graph changes between the recursive queuing here and + * bdrv_reopen_multiple(). */ + assert(bs->quiesce_counter > 0); + if (bs_queue == NULL) { bs_queue = g_new0(BlockReopenQueue, 1); QSIMPLEQ_INIT(bs_queue); @@ -XXX,XX +XXX,XX @@ BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue, * If all devices prepare successfully, then the changes are committed * to all devices. * + * All affected nodes must be drained between bdrv_reopen_queue() and + * bdrv_reopen_multiple(). */ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **errp) { @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **er assert(bs_queue != NULL); - aio_context_release(ctx); - bdrv_drain_all_begin(); - aio_context_acquire(ctx); - QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) { + assert(bs_entry->state.bs->quiesce_counter > 0); if (bdrv_reopen_prepare(&bs_entry->state, bs_queue, &local_err)) { error_propagate(errp, local_err); goto cleanup; @@ -XXX,XX +XXX,XX @@ cleanup: } g_free(bs_queue); - bdrv_drain_all_end(); - return ret; } @@ -XXX,XX +XXX,XX @@ int bdrv_reopen(BlockDriverState *bs, int bdrv_flags, Error **errp) { int ret = -1; Error *local_err = NULL; - BlockReopenQueue *queue = bdrv_reopen_queue(NULL, bs, NULL, bdrv_flags); + BlockReopenQueue *queue; + bdrv_subtree_drained_begin(bs); + + queue = bdrv_reopen_queue(NULL, bs, NULL, bdrv_flags); ret = bdrv_reopen_multiple(bdrv_get_aio_context(bs), queue, &local_err); if (local_err != NULL) { error_propagate(errp, local_err); } + + bdrv_subtree_drained_end(bs); + return ret; } diff --git a/block/replication.c b/block/replication.c index XXXXXXX..XXXXXXX 100644 --- a/block/replication.c +++ b/block/replication.c @@ -XXX,XX +XXX,XX @@ static void reopen_backing_file(BlockDriverState *bs, bool writable, new_secondary_flags = s->orig_secondary_flags; } + bdrv_subtree_drained_begin(s->hidden_disk->bs); + bdrv_subtree_drained_begin(s->secondary_disk->bs); + if (orig_hidden_flags != new_hidden_flags) { reopen_queue = bdrv_reopen_queue(reopen_queue, s->hidden_disk->bs, NULL, new_hidden_flags); @@ -XXX,XX +XXX,XX @@ static void reopen_backing_file(BlockDriverState *bs, bool writable, reopen_queue, &local_err); error_propagate(errp, local_err); } + + bdrv_subtree_drained_end(s->hidden_disk->bs); + bdrv_subtree_drained_end(s->secondary_disk->bs); } static void backup_job_cleanup(BlockDriverState *bs) diff --git a/qemu-io-cmds.c b/qemu-io-cmds.c index XXXXXXX..XXXXXXX 100644 --- a/qemu-io-cmds.c +++ b/qemu-io-cmds.c @@ -XXX,XX +XXX,XX @@ static int reopen_f(BlockBackend *blk, int argc, char **argv) opts = qopts ? qemu_opts_to_qdict(qopts, NULL) : NULL; qemu_opts_reset(&reopen_opts); + bdrv_subtree_drained_begin(bs); brq = bdrv_reopen_queue(NULL, bs, opts, flags); bdrv_reopen_multiple(bdrv_get_aio_context(bs), brq, &local_err); + bdrv_subtree_drained_end(bs); + if (local_err) { error_report_err(local_err); } else { -- 2.13.6