The following changes since commit eda1df0345f5a1e337e30367124dcb0e802bdfde: Merge remote-tracking branch 'remotes/armbru/tags/pull-pflash-2019-03-11' into staging (2019-03-12 11:12:36 +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 c31dfeb02a1d155bdb961edeb61a137a589c174b: qemu-iotests: Test the x-blockdev-reopen QMP command (2019-03-12 17:58:37 +0100) ---------------------------------------------------------------- Block layer patches: - file-posix: Make auto-read-only dynamic - Add x-blockdev-reopen QMP command - Finalize block-latency-histogram QMP command - gluster: Build fixes for newer lib version ---------------------------------------------------------------- Alberto Garcia (13): block: Allow freezing BdrvChild links block: Freeze the backing chain for the duration of the commit job block: Freeze the backing chain for the duration of the mirror job block: Freeze the backing chain for the duration of the stream job block: Add 'keep_old_opts' parameter to bdrv_reopen_queue() block: Handle child references in bdrv_reopen_queue() block: Allow omitting the 'backing' option in certain cases block: Allow changing the backing file on reopen block: Add a 'mutable_opts' field to BlockDriver block: Add bdrv_reset_options_allowed() block: Remove the AioContext parameter from bdrv_reopen_multiple() block: Add an 'x-blockdev-reopen' QMP command qemu-iotests: Test the x-blockdev-reopen QMP command Keith Busch (1): nvme: fix write zeroes offset and count Kevin Wolf (10): tests/virtio-blk-test: Disable auto-read-only qemu-iotests: commit to backing file with auto-read-only block: Avoid useless local_err block: Make permission changes in reopen less wrong file-posix: Fix bdrv_open_flags() for snapshot=on file-posix: Factor out raw_reconfigure_getfd() file-posix: Store BDRVRawState.reopen_state during reopen file-posix: Lock new fd in raw_reopen_prepare() file-posix: Prepare permission code for fd switching file-posix: Make auto-read-only dynamic Niels de Vos (1): gluster: the glfs_io_cbk callback function pointer adds pre/post stat args Prasanna Kumar Kalever (1): gluster: Handle changed glfs_ftruncate signature Vladimir Sementsov-Ogievskiy (2): qapi: move to QOM path for x-block-latency-histogram-set qapi: drop x- from x-block-latency-histogram-set qapi/block-core.json | 66 ++- configure | 42 ++ include/block/block.h | 13 +- include/block/block_int.h | 14 + block.c | 440 +++++++++++++++++-- block/commit.c | 16 + block/file-posix.c | 254 ++++++++--- block/gluster.c | 10 +- block/mirror.c | 8 + block/qapi.c | 12 +- block/qcow2.c | 25 ++ block/raw-format.c | 3 + block/replication.c | 7 +- block/stream.c | 21 + blockdev.c | 61 ++- hw/block/nvme.c | 6 +- qemu-io-cmds.c | 4 +- tests/virtio-blk-test.c | 2 +- tests/qemu-iotests/051 | 7 + tests/qemu-iotests/051.out | 9 + tests/qemu-iotests/051.pc.out | 9 + tests/qemu-iotests/232 | 31 ++ tests/qemu-iotests/232.out | 32 +- tests/qemu-iotests/245 | 991 ++++++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/245.out | 5 + tests/qemu-iotests/group | 1 + 26 files changed, 1929 insertions(+), 160 deletions(-) create mode 100644 tests/qemu-iotests/245 create mode 100644 tests/qemu-iotests/245.out From: Prasanna Kumar Kalever <prasanna.kalever@redhat.com> New versions of Glusters libgfapi.so have an updated glfs_ftruncate() function that returns additional 'struct stat' structures to enable advanced caching of attributes. This is useful for file servers, not so much for QEMU. Nevertheless, the API has changed and needs to be adopted. Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com> Signed-off-by: Niels de Vos <ndevos@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- configure | 18 ++++++++++++++++++ block/gluster.c | 4 ++++ 2 files changed, 22 insertions(+) diff --git a/configure b/configure index XXXXXXX..XXXXXXX 100755 --- a/configure +++ b/configure @@ -XXX,XX +XXX,XX @@ glusterfs_xlator_opt="no" glusterfs_discard="no" glusterfs_fallocate="no" glusterfs_zerofill="no" +glusterfs_ftruncate_has_stat="no" gtk="" gtk_gl="no" tls_priority="NORMAL" @@ -XXX,XX +XXX,XX @@ if test "$glusterfs" != "no" ; then glusterfs_fallocate="yes" glusterfs_zerofill="yes" fi + cat > $TMPC << EOF +#include <glusterfs/api/glfs.h> + +int +main(void) +{ + /* new glfs_ftruncate() passes two additional args */ + return glfs_ftruncate(NULL, 0, NULL, NULL); +} +EOF + if compile_prog "$glusterfs_cflags" "$glusterfs_libs" ; then + glusterfs_ftruncate_has_stat="yes" + fi else if test "$glusterfs" = "yes" ; then feature_not_found "GlusterFS backend support" \ @@ -XXX,XX +XXX,XX @@ if test "$glusterfs_zerofill" = "yes" ; then echo "CONFIG_GLUSTERFS_ZEROFILL=y" >> $config_host_mak fi +if test "$glusterfs_ftruncate_has_stat" = "yes" ; then + echo "CONFIG_GLUSTERFS_FTRUNCATE_HAS_STAT=y" >> $config_host_mak +fi + if test "$libssh2" = "yes" ; then echo "CONFIG_LIBSSH2=m" >> $config_host_mak echo "LIBSSH2_CFLAGS=$libssh2_cflags" >> $config_host_mak diff --git a/block/gluster.c b/block/gluster.c index XXXXXXX..XXXXXXX 100644 --- a/block/gluster.c +++ b/block/gluster.c @@ -XXX,XX +XXX,XX @@ #include "qemu/option.h" #include "qemu/cutils.h" +#ifdef CONFIG_GLUSTERFS_FTRUNCATE_HAS_STAT +# define glfs_ftruncate(fd, offset) glfs_ftruncate(fd, offset, NULL, NULL) +#endif + #define GLUSTER_OPT_FILENAME "filename" #define GLUSTER_OPT_VOLUME "volume" #define GLUSTER_OPT_PATH "path" -- 2.20.1 From: Niels de Vos <ndevos@redhat.com> The glfs_*_async() functions do a callback once finished. This callback has changed its arguments, pre- and post-stat structures have been added. This makes it possible to improve caching, which is useful for Samba and NFS-Ganesha, but not so much for QEMU. Gluster 6 is the first release that includes these new arguments. With an additional detection in ./configure, the new arguments can conditionally get included in the glfs_io_cbk handler. Signed-off-by: Niels de Vos <ndevos@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- configure | 24 ++++++++++++++++++++++++ block/gluster.c | 6 +++++- 2 files changed, 29 insertions(+), 1 deletion(-) diff --git a/configure b/configure index XXXXXXX..XXXXXXX 100755 --- a/configure +++ b/configure @@ -XXX,XX +XXX,XX @@ glusterfs_discard="no" glusterfs_fallocate="no" glusterfs_zerofill="no" glusterfs_ftruncate_has_stat="no" +glusterfs_iocb_has_stat="no" gtk="" gtk_gl="no" tls_priority="NORMAL" @@ -XXX,XX +XXX,XX @@ EOF if compile_prog "$glusterfs_cflags" "$glusterfs_libs" ; then glusterfs_ftruncate_has_stat="yes" fi + cat > $TMPC << EOF +#include <glusterfs/api/glfs.h> + +/* new glfs_io_cbk() passes two additional glfs_stat structs */ +static void +glusterfs_iocb(glfs_fd_t *fd, ssize_t ret, struct glfs_stat *prestat, struct glfs_stat *poststat, void *data) +{} + +int +main(void) +{ + glfs_io_cbk iocb = &glusterfs_iocb; + iocb(NULL, 0 , NULL, NULL, NULL); + return 0; +} +EOF + if compile_prog "$glusterfs_cflags" "$glusterfs_libs" ; then + glusterfs_iocb_has_stat="yes" + fi else if test "$glusterfs" = "yes" ; then feature_not_found "GlusterFS backend support" \ @@ -XXX,XX +XXX,XX @@ if test "$glusterfs_ftruncate_has_stat" = "yes" ; then echo "CONFIG_GLUSTERFS_FTRUNCATE_HAS_STAT=y" >> $config_host_mak fi +if test "$glusterfs_iocb_has_stat" = "yes" ; then + echo "CONFIG_GLUSTERFS_IOCB_HAS_STAT=y" >> $config_host_mak +fi + if test "$libssh2" = "yes" ; then echo "CONFIG_LIBSSH2=m" >> $config_host_mak echo "LIBSSH2_CFLAGS=$libssh2_cflags" >> $config_host_mak 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 struct glfs *qemu_gluster_init(BlockdevOptionsGluster *gconf, /* * AIO callback routine called from GlusterFS thread. */ -static void gluster_finish_aiocb(struct glfs_fd *fd, ssize_t ret, void *arg) +static void gluster_finish_aiocb(struct glfs_fd *fd, ssize_t ret, +#ifdef CONFIG_GLUSTERFS_IOCB_HAS_STAT + struct glfs_stat *pre, struct glfs_stat *post, +#endif + void *arg) { GlusterAIOCB *acb = (GlusterAIOCB *)arg; -- 2.20.1 From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qapi/block-core.json | 4 ++-- blockdev.c | 12 ++++++------ 2 files changed, 8 insertions(+), 8 deletions(-) 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 @@ # If only @device parameter is specified, remove all present latency histograms # for the device. Otherwise, add/reset some of (or all) latency histograms. # -# @device: device name to set latency histogram for. +# @id: The name or QOM path of the guest device. # # @boundaries: list of interval boundary values (see description in # BlockLatencyHistogramInfo definition). If specified, all @@ -XXX,XX +XXX,XX @@ # <- { "return": {} } ## { 'command': 'x-block-latency-histogram-set', - 'data': {'device': 'str', + 'data': {'id': 'str', '*boundaries': ['uint64'], '*boundaries-read': ['uint64'], '*boundaries-write': ['uint64'], diff --git a/blockdev.c b/blockdev.c index XXXXXXX..XXXXXXX 100644 --- a/blockdev.c +++ b/blockdev.c @@ -XXX,XX +XXX,XX @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread, } void qmp_x_block_latency_histogram_set( - const char *device, + const char *id, bool has_boundaries, uint64List *boundaries, bool has_boundaries_read, uint64List *boundaries_read, bool has_boundaries_write, uint64List *boundaries_write, bool has_boundaries_flush, uint64List *boundaries_flush, Error **errp) { - BlockBackend *blk = blk_by_name(device); + BlockBackend *blk = qmp_get_blk(NULL, id, errp); BlockAcctStats *stats; int ret; if (!blk) { - error_setg(errp, "Device '%s' not found", device); return; } + stats = blk_get_stats(blk); if (!has_boundaries && !has_boundaries_read && !has_boundaries_write && @@ -XXX,XX +XXX,XX @@ void qmp_x_block_latency_histogram_set( stats, BLOCK_ACCT_READ, has_boundaries_read ? boundaries_read : boundaries); if (ret) { - error_setg(errp, "Device '%s' set read boundaries fail", device); + error_setg(errp, "Device '%s' set read boundaries fail", id); return; } } @@ -XXX,XX +XXX,XX @@ void qmp_x_block_latency_histogram_set( stats, BLOCK_ACCT_WRITE, has_boundaries_write ? boundaries_write : boundaries); if (ret) { - error_setg(errp, "Device '%s' set write boundaries fail", device); + error_setg(errp, "Device '%s' set write boundaries fail", id); return; } } @@ -XXX,XX +XXX,XX @@ void qmp_x_block_latency_histogram_set( stats, BLOCK_ACCT_FLUSH, has_boundaries_flush ? boundaries_flush : boundaries); if (ret) { - error_setg(errp, "Device '%s' set flush boundaries fail", device); + error_setg(errp, "Device '%s' set flush boundaries fail", id); return; } } -- 2.20.1 From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Drop x- and x_ prefixes for latency histograms and update version to 3.1 Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qapi/block-core.json | 20 ++++++++++---------- block/qapi.c | 12 ++++++------ blockdev.c | 2 +- 3 files changed, 17 insertions(+), 17 deletions(-) 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 @@ # +------------------ # 10 50 100 # -# Since: 2.12 +# Since: 4.0 ## { 'struct': 'BlockLatencyHistogramInfo', 'data': {'boundaries': ['uint64'], 'bins': ['uint64'] } } ## -# @x-block-latency-histogram-set: +# @block-latency-histogram-set: # # Manage read, write and flush latency histograms for the device. # @@ -XXX,XX +XXX,XX @@ # # Returns: error if device is not found or any boundary arrays are invalid. # -# Since: 2.12 +# Since: 4.0 # # Example: set new histograms for all io types with intervals # [0, 10), [10, 50), [50, 100), [100, +inf): @@ -XXX,XX +XXX,XX @@ # "arguments": { "device": "drive0" } } # <- { "return": {} } ## -{ 'command': 'x-block-latency-histogram-set', +{ 'command': 'block-latency-histogram-set', 'data': {'id': 'str', '*boundaries': ['uint64'], '*boundaries-read': ['uint64'], @@ -XXX,XX +XXX,XX @@ # @timed_stats: Statistics specific to the set of previously defined # intervals of time (Since 2.5) # -# @x_rd_latency_histogram: @BlockLatencyHistogramInfo. (Since 2.12) +# @rd_latency_histogram: @BlockLatencyHistogramInfo. (Since 4.0) # -# @x_wr_latency_histogram: @BlockLatencyHistogramInfo. (Since 2.12) +# @wr_latency_histogram: @BlockLatencyHistogramInfo. (Since 4.0) # -# @x_flush_latency_histogram: @BlockLatencyHistogramInfo. (Since 2.12) +# @flush_latency_histogram: @BlockLatencyHistogramInfo. (Since 4.0) # # Since: 0.14.0 ## @@ -XXX,XX +XXX,XX @@ 'invalid_wr_operations': 'int', 'invalid_flush_operations': 'int', 'account_invalid': 'bool', 'account_failed': 'bool', 'timed_stats': ['BlockDeviceTimedStats'], - '*x_rd_latency_histogram': 'BlockLatencyHistogramInfo', - '*x_wr_latency_histogram': 'BlockLatencyHistogramInfo', - '*x_flush_latency_histogram': 'BlockLatencyHistogramInfo' } } + '*rd_latency_histogram': 'BlockLatencyHistogramInfo', + '*wr_latency_histogram': 'BlockLatencyHistogramInfo', + '*flush_latency_histogram': 'BlockLatencyHistogramInfo' } } ## # @BlockStats: diff --git a/block/qapi.c b/block/qapi.c index XXXXXXX..XXXXXXX 100644 --- a/block/qapi.c +++ b/block/qapi.c @@ -XXX,XX +XXX,XX @@ static void bdrv_query_blk_stats(BlockDeviceStats *ds, BlockBackend *blk) } bdrv_latency_histogram_stats(&stats->latency_histogram[BLOCK_ACCT_READ], - &ds->has_x_rd_latency_histogram, - &ds->x_rd_latency_histogram); + &ds->has_rd_latency_histogram, + &ds->rd_latency_histogram); bdrv_latency_histogram_stats(&stats->latency_histogram[BLOCK_ACCT_WRITE], - &ds->has_x_wr_latency_histogram, - &ds->x_wr_latency_histogram); + &ds->has_wr_latency_histogram, + &ds->wr_latency_histogram); bdrv_latency_histogram_stats(&stats->latency_histogram[BLOCK_ACCT_FLUSH], - &ds->has_x_flush_latency_histogram, - &ds->x_flush_latency_histogram); + &ds->has_flush_latency_histogram, + &ds->flush_latency_histogram); } static BlockStats *bdrv_query_bds_stats(BlockDriverState *bs, diff --git a/blockdev.c b/blockdev.c index XXXXXXX..XXXXXXX 100644 --- a/blockdev.c +++ b/blockdev.c @@ -XXX,XX +XXX,XX @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread, aio_context_release(old_context); } -void qmp_x_block_latency_histogram_set( +void qmp_block_latency_histogram_set( const char *id, bool has_boundaries, uint64List *boundaries, bool has_boundaries_read, uint64List *boundaries_read, -- 2.20.1 tests/virtio-blk-test uses a temporary image file that it deletes while QEMU is still running, so it can't be reopened when writers are attached or detached. Disable auto-read-only to keep it always writable. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- tests/virtio-blk-test.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/virtio-blk-test.c b/tests/virtio-blk-test.c index XXXXXXX..XXXXXXX 100644 --- a/tests/virtio-blk-test.c +++ b/tests/virtio-blk-test.c @@ -XXX,XX +XXX,XX @@ static void *virtio_blk_test_setup(GString *cmd_line, void *arg) char *tmp_path = drive_create(); g_string_append_printf(cmd_line, - " -drive if=none,id=drive0,file=%s,format=raw " + " -drive if=none,id=drive0,file=%s,format=raw,auto-read-only=off " "-drive if=none,id=drive1,file=null-co://,format=raw ", tmp_path); -- 2.20.1 Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- tests/qemu-iotests/232 | 31 +++++++++++++++++++++++++++++++ tests/qemu-iotests/232.out | 20 ++++++++++++++++++++ 2 files changed, 51 insertions(+) diff --git a/tests/qemu-iotests/232 b/tests/qemu-iotests/232 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/232 +++ b/tests/qemu-iotests/232 @@ -XXX,XX +XXX,XX @@ status=1 # failure is the default! _cleanup() { _cleanup_test_img + rm -f $TEST_IMG.[01234] } trap "_cleanup; exit \$status" 0 1 2 3 15 @@ -XXX,XX +XXX,XX @@ run_qemu_info_block -blockdev driver=file,filename="$TEST_IMG",node-name=node0,a run_qemu_info_block -blockdev driver=file,filename="$TEST_IMG",node-name=node0,auto-read-only=on run_qemu_info_block -blockdev driver=file,filename="$TEST_IMG",node-name=node0 +echo +echo "=== Try commit to backing file with auto-read-only ===" +echo + +TEST_IMG="$TEST_IMG.0" _make_test_img $size +TEST_IMG="$TEST_IMG.1" _make_test_img $size +TEST_IMG="$TEST_IMG.2" _make_test_img $size +TEST_IMG="$TEST_IMG.3" _make_test_img $size +TEST_IMG="$TEST_IMG.4" _make_test_img $size + +(cat <<EOF +{"execute":"qmp_capabilities"} +{"execute":"block-commit", + "arguments":{"device":"format-4", "top-node": "format-2", "base-node":"format-0", "job-id":"job0"}} +EOF +sleep 1 +echo '{"execute":"quit"}' +) | $QEMU -qmp stdio -nographic -nodefaults \ + -blockdev file,node-name=file-0,filename=$TEST_IMG.0,auto-read-only=on \ + -blockdev qcow2,node-name=format-0,file=file-0,read-only=on \ + -blockdev file,node-name=file-1,filename=$TEST_IMG.1,auto-read-only=on \ + -blockdev qcow2,node-name=format-1,file=file-1,read-only=on,backing=format-0 \ + -blockdev file,node-name=file-2,filename=$TEST_IMG.2,auto-read-only=on \ + -blockdev qcow2,node-name=format-2,file=file-2,read-only=on,backing=format-1 \ + -blockdev file,node-name=file-3,filename=$TEST_IMG.3,auto-read-only=on \ + -blockdev qcow2,node-name=format-3,file=file-3,read-only=on,backing=format-2 \ + -blockdev file,node-name=file-4,filename=$TEST_IMG.4,auto-read-only=on \ + -blockdev qcow2,node-name=format-4,file=file-4,read-only=on,backing=format-3 | + _filter_qmp + # success, all done echo "*** done" rm -f $seq.full diff --git a/tests/qemu-iotests/232.out b/tests/qemu-iotests/232.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/232.out +++ b/tests/qemu-iotests/232.out @@ -XXX,XX +XXX,XX @@ QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0,read QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0,auto-read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied node0: TEST_DIR/t.IMGFMT (file, read-only) QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0: Could not open 'TEST_DIR/t.IMGFMT': Permission denied + +=== Try commit to backing file with auto-read-only === + +Formatting 'TEST_DIR/t.IMGFMT.0', fmt=IMGFMT size=134217728 +Formatting 'TEST_DIR/t.IMGFMT.1', fmt=IMGFMT size=134217728 +Formatting 'TEST_DIR/t.IMGFMT.2', fmt=IMGFMT size=134217728 +Formatting 'TEST_DIR/t.IMGFMT.3', fmt=IMGFMT size=134217728 +Formatting 'TEST_DIR/t.IMGFMT.4', fmt=IMGFMT size=134217728 +QMP_VERSION +{"return": {}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "job0"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "job0"}} +{"return": {}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "waiting", "id": "job0"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "pending", "id": "job0"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "BLOCK_JOB_COMPLETED", "data": {"device": "job0", "len": 134217728, "offset": 134217728, "speed": 0, "type": "commit"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "concluded", "id": "job0"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "null", "id": "job0"}} +{"return": {}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "SHUTDOWN", "data": {"guest": false, "reason": "host-qmp-quit"}} *** done -- 2.20.1 Signed-off-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Alberto Garcia <berto@igalia.com> Reviewed-by: Eric Blake <eblake@redhat.com> --- block.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **er { int ret = -1; BlockReopenQueueEntry *bs_entry, *next; - Error *local_err = NULL; assert(bs_queue != NULL); 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); + if (bdrv_reopen_prepare(&bs_entry->state, bs_queue, errp)) { goto cleanup; } bs_entry->prepared = true; -- 2.20.1 The way that reopen interacts with permission changes has one big problem: Both operations are recursive, and the permissions are changes for each node in the reopen queue. For a simple graph that consists just of parent and child, .bdrv_check_perm will be called twice for the child, once recursively when adjusting the permissions of parent, and once again when the child itself is reopened. Even worse, the first .bdrv_check_perm call happens before .bdrv_reopen_prepare was called for the child and the second one is called afterwards. Making sure that .bdrv_check_perm (and the other permission callbacks) are called only once is hard. We can cope with multiple calls right now, but as soon as file-posix gets a dynamic auto-read-only that may need to open a new file descriptor, we get the additional requirement that all of them are after the .bdrv_reopen_prepare call. So reorder things in bdrv_reopen_multiple() to first call .bdrv_reopen_prepare for all involved nodes and only then adjust permissions. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 35 ++++++++++++++++++++++++----------- 1 file changed, 24 insertions(+), 11 deletions(-) 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_set_perm(BdrvChild *c, uint64_t perm, uint64_t shared); typedef struct BlockReopenQueueEntry { bool prepared; + bool perms_checked; BDRVReopenState state; QSIMPLEQ_ENTRY(BlockReopenQueueEntry) entry; } BlockReopenQueueEntry; @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **er bs_entry->prepared = true; } + QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) { + BDRVReopenState *state = &bs_entry->state; + ret = bdrv_check_perm(state->bs, bs_queue, state->perm, + state->shared_perm, NULL, errp); + if (ret < 0) { + goto cleanup_perm; + } + bs_entry->perms_checked = true; + } + /* If we reach this point, we have success and just need to apply the * changes */ @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **er } ret = 0; +cleanup_perm: + QSIMPLEQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) { + BDRVReopenState *state = &bs_entry->state; + + if (!bs_entry->perms_checked) { + continue; + } + if (ret == 0) { + bdrv_set_perm(state->bs, state->perm, state->shared_perm); + } else { + bdrv_abort_perm_update(state->bs); + } + } cleanup: QSIMPLEQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) { if (ret) { @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue, } while ((entry = qdict_next(reopen_state->options, entry))); } - ret = bdrv_check_perm(reopen_state->bs, queue, reopen_state->perm, - reopen_state->shared_perm, NULL, errp); - if (ret < 0) { - goto error; - } - ret = 0; /* Restore the original reopen_state->options QDict */ @@ -XXX,XX +XXX,XX @@ void bdrv_reopen_commit(BDRVReopenState *reopen_state) bdrv_refresh_limits(bs, NULL); - bdrv_set_perm(reopen_state->bs, reopen_state->perm, - reopen_state->shared_perm); - new_can_write = !bdrv_is_read_only(bs) && !(bdrv_get_flags(bs) & BDRV_O_INACTIVE); if (!old_can_write && new_can_write && drv->bdrv_reopen_bitmaps_rw) { @@ -XXX,XX +XXX,XX @@ void bdrv_reopen_abort(BDRVReopenState *reopen_state) if (drv->bdrv_reopen_abort) { drv->bdrv_reopen_abort(reopen_state); } - - bdrv_abort_perm_update(reopen_state->bs); } -- 2.20.1 Using a different read-only setting for bs->open_flags than for the flags to the driver's open function is just inconsistent and a bad idea. After this patch, the temporary snapshot keeps being opened read-only if read-only=on,snapshot=on is passed. If we wanted to change this behaviour to make only the orginal image file read-only, but the temporary overlay read-write (as the comment in the removed code suggests), that change would have to be made in bdrv_temp_snapshot_options() (where the comment suggests otherwise). Addressing this inconsistency before introducing dynamic auto-read-only is important because otherwise we would immediately try to reopen the temporary overlay even though the file is already unlinked. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 7 ------- tests/qemu-iotests/051 | 7 +++++++ tests/qemu-iotests/051.out | 9 +++++++++ tests/qemu-iotests/051.pc.out | 9 +++++++++ 4 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 @@ static int bdrv_open_flags(BlockDriverState *bs, int flags) */ open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_PROTOCOL); - /* - * Snapshots should be writable. - */ - if (flags & BDRV_O_TEMPORARY) { - open_flags |= BDRV_O_RDWR; - } - return open_flags; } diff --git a/tests/qemu-iotests/051 b/tests/qemu-iotests/051 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/051 +++ b/tests/qemu-iotests/051 @@ -XXX,XX +XXX,XX @@ $QEMU_IO -c "read -P 0x33 0 4k" "$TEST_IMG" | _filter_qemu_io # Using snapshot=on with a non-existent TMPDIR TMPDIR=/nonexistent run_qemu -drive driver=null-co,snapshot=on +# Using snapshot=on together with read-only=on +echo "info block" | + run_qemu -drive file="$TEST_IMG",snapshot=on,read-only=on,if=none,id=$device_id | + _filter_qemu_io | + sed -e 's#"/[^"]*/vl\.[A-Za-z]\{6\}"#SNAPSHOT_PATH#g' + + # success, all done echo "*** done" rm -f $seq.full diff --git a/tests/qemu-iotests/051.out b/tests/qemu-iotests/051.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/051.out +++ b/tests/qemu-iotests/051.out @@ -XXX,XX +XXX,XX @@ read 4096/4096 bytes at offset 0 Testing: -drive driver=null-co,snapshot=on QEMU_PROG: -drive driver=null-co,snapshot=on: Could not get temporary filename: No such file or directory +Testing: -drive file=TEST_DIR/t.qcow2,snapshot=on,read-only=on,if=none,id=drive0 +QEMU X.Y.Z monitor - type 'help' for more information +(qemu) info block +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}}, "driver": "qcow2", "file": {"driver": "file", "filename": SNAPSHOT_PATH}} (qcow2, read-only) + Removable device: not locked, tray closed + Cache mode: writeback, ignore flushes + Backing file: TEST_DIR/t.qcow2 (chain depth: 1) +(qemu) quit + *** done diff --git a/tests/qemu-iotests/051.pc.out b/tests/qemu-iotests/051.pc.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/051.pc.out +++ b/tests/qemu-iotests/051.pc.out @@ -XXX,XX +XXX,XX @@ read 4096/4096 bytes at offset 0 Testing: -drive driver=null-co,snapshot=on QEMU_PROG: -drive driver=null-co,snapshot=on: Could not get temporary filename: No such file or directory +Testing: -drive file=TEST_DIR/t.qcow2,snapshot=on,read-only=on,if=none,id=drive0 +QEMU X.Y.Z monitor - type 'help' for more information +(qemu) info block +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}}, "driver": "qcow2", "file": {"driver": "file", "filename": SNAPSHOT_PATH}} (qcow2, read-only) + Removable device: not locked, tray closed + Cache mode: writeback, ignore flushes + Backing file: TEST_DIR/t.qcow2 (chain depth: 1) +(qemu) quit + *** done -- 2.20.1 Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 107 ++++++++++++++++++++++++++------------------- 1 file changed, 62 insertions(+), 45 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 int raw_handle_perm_lock(BlockDriverState *bs, return ret; } +static int raw_reconfigure_getfd(BlockDriverState *bs, int flags, + int *open_flags, Error **errp) +{ + BDRVRawState *s = bs->opaque; + int fd = -1; + int ret; + int fcntl_flags = O_APPEND | O_NONBLOCK; +#ifdef O_NOATIME + fcntl_flags |= O_NOATIME; +#endif + + *open_flags = 0; + if (s->type == FTYPE_CD) { + *open_flags |= O_NONBLOCK; + } + + raw_parse_flags(flags, open_flags); + +#ifdef O_ASYNC + /* Not all operating systems have O_ASYNC, and those that don't + * will not let us track the state into rs->open_flags (typically + * you achieve the same effect with an ioctl, for example I_SETSIG + * on Solaris). But we do not use O_ASYNC, so that's fine. + */ + assert((s->open_flags & O_ASYNC) == 0); +#endif + + if ((*open_flags & ~fcntl_flags) == (s->open_flags & ~fcntl_flags)) { + /* dup the original fd */ + fd = qemu_dup(s->fd); + if (fd >= 0) { + ret = fcntl_setfl(fd, *open_flags); + if (ret) { + qemu_close(fd); + fd = -1; + } + } + } + + /* If we cannot use fcntl, or fcntl failed, fall back to qemu_open() */ + if (fd == -1) { + const char *normalized_filename = bs->filename; + ret = raw_normalize_devicepath(&normalized_filename, errp); + if (ret >= 0) { + assert(!(*open_flags & O_CREAT)); + fd = qemu_open(normalized_filename, *open_flags); + if (fd == -1) { + error_setg_errno(errp, errno, "Could not reopen file"); + return -1; + } + } + } + + return fd; +} + static int raw_reopen_prepare(BDRVReopenState *state, BlockReopenQueue *queue, Error **errp) { @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, state->opaque = g_new0(BDRVRawReopenState, 1); rs = state->opaque; - rs->fd = -1; /* Handle options changes */ opts = qemu_opts_create(&raw_runtime_opts, NULL, 0, &error_abort); @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, * bdrv_reopen_prepare() will detect changes and complain. */ qemu_opts_to_qdict(opts, state->options); - if (s->type == FTYPE_CD) { - rs->open_flags |= O_NONBLOCK; - } - - raw_parse_flags(state->flags, &rs->open_flags); - - int fcntl_flags = O_APPEND | O_NONBLOCK; -#ifdef O_NOATIME - fcntl_flags |= O_NOATIME; -#endif - -#ifdef O_ASYNC - /* Not all operating systems have O_ASYNC, and those that don't - * will not let us track the state into rs->open_flags (typically - * you achieve the same effect with an ioctl, for example I_SETSIG - * on Solaris). But we do not use O_ASYNC, so that's fine. - */ - assert((s->open_flags & O_ASYNC) == 0); -#endif - - if ((rs->open_flags & ~fcntl_flags) == (s->open_flags & ~fcntl_flags)) { - /* dup the original fd */ - rs->fd = qemu_dup(s->fd); - if (rs->fd >= 0) { - ret = fcntl_setfl(rs->fd, rs->open_flags); - if (ret) { - qemu_close(rs->fd); - rs->fd = -1; - } - } - } - - /* If we cannot use fcntl, or fcntl failed, fall back to qemu_open() */ - if (rs->fd == -1) { - const char *normalized_filename = state->bs->filename; - ret = raw_normalize_devicepath(&normalized_filename, errp); - if (ret >= 0) { - assert(!(rs->open_flags & O_CREAT)); - rs->fd = qemu_open(normalized_filename, rs->open_flags); - if (rs->fd == -1) { - error_setg_errno(errp, errno, "Could not reopen file"); - ret = -1; - } - } + rs->fd = raw_reconfigure_getfd(state->bs, state->flags, &rs->open_flags, + &local_err); + if (local_err) { + error_propagate(errp, local_err); + ret = -1; + goto out; } /* Fail already reopen_prepare() if we can't get a working O_DIRECT -- 2.20.1 We'll want to access the file descriptor in the reopen_state while processing permission changes in the context of the repoen. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 10 ++++++++++ 1 file changed, 10 insertions(+) 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 BDRVRawState { uint64_t locked_perm; uint64_t locked_shared_perm; + BDRVReopenState *reopen_state; + #ifdef CONFIG_XFS bool is_xfs:1; #endif @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, } } + s->reopen_state = state; out: qemu_opts_del(opts); return ret; @@ -XXX,XX +XXX,XX @@ static void raw_reopen_commit(BDRVReopenState *state) g_free(state->opaque); state->opaque = NULL; + + assert(s->reopen_state == state); + s->reopen_state = NULL; } static void raw_reopen_abort(BDRVReopenState *state) { BDRVRawReopenState *rs = state->opaque; + BDRVRawState *s = state->bs->opaque; /* nothing to do if NULL, we didn't get far enough */ if (rs == NULL) { @@ -XXX,XX +XXX,XX @@ static void raw_reopen_abort(BDRVReopenState *state) } g_free(state->opaque); state->opaque = NULL; + + assert(s->reopen_state == state); + s->reopen_state = NULL; } static int hdev_get_max_transfer_length(BlockDriverState *bs, int fd) -- 2.20.1 There is no reason why we can take locks on the new file descriptor only in raw_reopen_commit() where error handling isn't possible any more. Instead, we can already do this in raw_reopen_prepare(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 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 int raw_reopen_prepare(BDRVReopenState *state, BDRVRawState *s; BDRVRawReopenState *rs; QemuOpts *opts; - int ret = 0; + int ret; Error *local_err = NULL; assert(state != NULL); @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, if (rs->fd != -1) { raw_probe_alignment(state->bs, rs->fd, &local_err); if (local_err) { - qemu_close(rs->fd); - rs->fd = -1; error_propagate(errp, local_err); ret = -EINVAL; + goto out_fd; + } + + /* Copy locks to the new fd */ + ret = raw_apply_lock_bytes(NULL, rs->fd, s->locked_perm, + s->locked_shared_perm, false, errp); + if (ret < 0) { + ret = -EINVAL; + goto out_fd; } } s->reopen_state = state; + ret = 0; +out_fd: + if (ret < 0) { + qemu_close(rs->fd); + rs->fd = -1; + } out: qemu_opts_del(opts); return ret; @@ -XXX,XX +XXX,XX @@ static void raw_reopen_commit(BDRVReopenState *state) { BDRVRawReopenState *rs = state->opaque; BDRVRawState *s = state->bs->opaque; - Error *local_err = NULL; s->check_cache_dropped = rs->check_cache_dropped; s->open_flags = rs->open_flags; - /* Copy locks to the new fd before closing the old one. */ - raw_apply_lock_bytes(NULL, rs->fd, s->locked_perm, - s->locked_shared_perm, false, &local_err); - if (local_err) { - /* shouldn't fail in a sane host, but report it just in case. */ - error_report_err(local_err); - } qemu_close(s->fd); s->fd = rs->fd; -- 2.20.1 In order to be able to dynamically reopen the file read-only or read-write, depending on the users that are attached, we need to be able to switch to a different file descriptor during the permission change. This interacts with reopen, which also creates a new file descriptor and performs permission changes internally. In this case, the permission change code must reuse the reopen file descriptor instead of creating a third one. In turn, reopen can drop its code to copy file locks to the new file descriptor because that is now done when applying the new permissions. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 96 ++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 85 insertions(+), 11 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 @@ typedef struct BDRVRawState { uint64_t locked_perm; uint64_t locked_shared_perm; + int perm_change_fd; BDRVReopenState *reopen_state; #ifdef CONFIG_XFS @@ -XXX,XX +XXX,XX @@ static int raw_handle_perm_lock(BlockDriverState *bs, } static int raw_reconfigure_getfd(BlockDriverState *bs, int flags, - int *open_flags, Error **errp) + int *open_flags, bool force_dup, + Error **errp) { BDRVRawState *s = bs->opaque; int fd = -1; @@ -XXX,XX +XXX,XX @@ static int raw_reconfigure_getfd(BlockDriverState *bs, int flags, assert((s->open_flags & O_ASYNC) == 0); #endif + if (!force_dup && *open_flags == s->open_flags) { + /* We're lucky, the existing fd is fine */ + return s->fd; + } + if ((*open_flags & ~fcntl_flags) == (s->open_flags & ~fcntl_flags)) { /* dup the original fd */ fd = qemu_dup(s->fd); @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, qemu_opts_to_qdict(opts, state->options); rs->fd = raw_reconfigure_getfd(state->bs, state->flags, &rs->open_flags, - &local_err); + true, &local_err); if (local_err) { error_propagate(errp, local_err); ret = -1; @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, ret = -EINVAL; goto out_fd; } - - /* Copy locks to the new fd */ - ret = raw_apply_lock_bytes(NULL, rs->fd, s->locked_perm, - s->locked_shared_perm, false, errp); - if (ret < 0) { - ret = -EINVAL; - goto out_fd; - } } s->reopen_state = state; @@ -XXX,XX +XXX,XX @@ static QemuOptsList raw_create_opts = { static int raw_check_perm(BlockDriverState *bs, uint64_t perm, uint64_t shared, Error **errp) { - return raw_handle_perm_lock(bs, RAW_PL_PREPARE, perm, shared, errp); + BDRVRawState *s = bs->opaque; + BDRVRawReopenState *rs = NULL; + int open_flags; + int ret; + + if (s->perm_change_fd) { + /* + * In the context of reopen, this function may be called several times + * (directly and recursively while change permissions of the parent). + * This is even true for children that don't inherit from the original + * reopen node, so s->reopen_state is not set. + * + * Ignore all but the first call. + */ + return 0; + } + + if (s->reopen_state) { + /* We already have a new file descriptor to set permissions for */ + assert(s->reopen_state->perm == perm); + assert(s->reopen_state->shared_perm == shared); + rs = s->reopen_state->opaque; + s->perm_change_fd = rs->fd; + } else { + /* We may need a new fd if auto-read-only switches the mode */ + ret = raw_reconfigure_getfd(bs, bs->open_flags, &open_flags, + false, errp); + if (ret < 0) { + return ret; + } else if (ret != s->fd) { + s->perm_change_fd = ret; + } + } + + /* Prepare permissions on old fd to avoid conflicts between old and new, + * but keep everything locked that new will need. */ + ret = raw_handle_perm_lock(bs, RAW_PL_PREPARE, perm, shared, errp); + if (ret < 0) { + goto fail; + } + + /* Copy locks to the new fd */ + if (s->perm_change_fd) { + ret = raw_apply_lock_bytes(NULL, s->perm_change_fd, perm, ~shared, + false, errp); + if (ret < 0) { + raw_handle_perm_lock(bs, RAW_PL_ABORT, 0, 0, NULL); + goto fail; + } + } + return 0; + +fail: + if (s->perm_change_fd && !s->reopen_state) { + qemu_close(s->perm_change_fd); + } + s->perm_change_fd = 0; + return ret; } static void raw_set_perm(BlockDriverState *bs, uint64_t perm, uint64_t shared) { BDRVRawState *s = bs->opaque; + + /* For reopen, we have already switched to the new fd (.bdrv_set_perm is + * called after .bdrv_reopen_commit) */ + if (s->perm_change_fd && s->fd != s->perm_change_fd) { + qemu_close(s->fd); + s->fd = s->perm_change_fd; + } + s->perm_change_fd = 0; + raw_handle_perm_lock(bs, RAW_PL_COMMIT, perm, shared, NULL); s->perm = perm; s->shared_perm = shared; @@ -XXX,XX +XXX,XX @@ static void raw_set_perm(BlockDriverState *bs, uint64_t perm, uint64_t shared) static void raw_abort_perm_update(BlockDriverState *bs) { + BDRVRawState *s = bs->opaque; + + /* For reopen, .bdrv_reopen_abort is called afterwards and will close + * the file descriptor. */ + if (s->perm_change_fd && !s->reopen_state) { + qemu_close(s->perm_change_fd); + } + s->perm_change_fd = 0; + raw_handle_perm_lock(bs, RAW_PL_ABORT, 0, 0, NULL); } -- 2.20.1 Until now, with auto-read-only=on we tried to open the file read-write first and if that failed, read-only was tried. This is actually not good enough for libvirt, which gives QEMU SELinux permissions for read-write only as soon as it actually intends to write to the image. So we need to be able to switch between read-only and read-write at runtime. This patch makes auto-read-only dynamic, i.e. the file is opened read-only as long as no user of the node has requested write permissions, but it is automatically reopened read-write as soon as the first writer is attached. Conversely, if the last writer goes away, the file is reopened read-only again. bs->read_only is no longer set for auto-read-only=on files even if the file descriptor is opened read-only because it will be transparently upgraded as soon as a writer is attached. This changes the output of qemu-iotests 232. Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 36 +++++++++++++++++------------------- tests/qemu-iotests/232.out | 12 ++++++------ 2 files changed, 23 insertions(+), 25 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 void raw_probe_alignment(BlockDriverState *bs, int fd, Error **errp) } } -static void raw_parse_flags(int bdrv_flags, int *open_flags) +static void raw_parse_flags(int bdrv_flags, int *open_flags, bool has_writers) { + bool read_write = false; assert(open_flags != NULL); *open_flags |= O_BINARY; *open_flags &= ~O_ACCMODE; - if (bdrv_flags & BDRV_O_RDWR) { + + if (bdrv_flags & BDRV_O_AUTO_RDONLY) { + read_write = has_writers; + } else if (bdrv_flags & BDRV_O_RDWR) { + read_write = true; + } + + if (read_write) { *open_flags |= O_RDWR; } else { *open_flags |= O_RDONLY; @@ -XXX,XX +XXX,XX @@ static int raw_open_common(BlockDriverState *bs, QDict *options, false); s->open_flags = open_flags; - raw_parse_flags(bdrv_flags, &s->open_flags); + raw_parse_flags(bdrv_flags, &s->open_flags, false); s->fd = -1; fd = qemu_open(filename, s->open_flags, 0644); ret = fd < 0 ? -errno : 0; - if (ret == -EACCES || ret == -EROFS) { - /* Try to degrade to read-only, but if it doesn't work, still use the - * normal error message. */ - if (bdrv_apply_auto_read_only(bs, NULL, NULL) == 0) { - bdrv_flags &= ~BDRV_O_RDWR; - raw_parse_flags(bdrv_flags, &s->open_flags); - assert(!(s->open_flags & O_CREAT)); - fd = qemu_open(filename, s->open_flags); - ret = fd < 0 ? -errno : 0; - } - } - if (ret < 0) { error_setg_errno(errp, -ret, "Could not open '%s'", filename); if (ret == -EROFS) { @@ -XXX,XX +XXX,XX @@ static int raw_handle_perm_lock(BlockDriverState *bs, } static int raw_reconfigure_getfd(BlockDriverState *bs, int flags, - int *open_flags, bool force_dup, + int *open_flags, uint64_t perm, bool force_dup, Error **errp) { BDRVRawState *s = bs->opaque; int fd = -1; int ret; + bool has_writers = perm & + (BLK_PERM_WRITE | BLK_PERM_WRITE_UNCHANGED | BLK_PERM_RESIZE); int fcntl_flags = O_APPEND | O_NONBLOCK; #ifdef O_NOATIME fcntl_flags |= O_NOATIME; @@ -XXX,XX +XXX,XX @@ static int raw_reconfigure_getfd(BlockDriverState *bs, int flags, *open_flags |= O_NONBLOCK; } - raw_parse_flags(flags, open_flags); + raw_parse_flags(flags, open_flags, has_writers); #ifdef O_ASYNC /* Not all operating systems have O_ASYNC, and those that don't @@ -XXX,XX +XXX,XX @@ static int raw_reopen_prepare(BDRVReopenState *state, qemu_opts_to_qdict(opts, state->options); rs->fd = raw_reconfigure_getfd(state->bs, state->flags, &rs->open_flags, - true, &local_err); + state->perm, true, &local_err); if (local_err) { error_propagate(errp, local_err); ret = -1; @@ -XXX,XX +XXX,XX @@ static int raw_check_perm(BlockDriverState *bs, uint64_t perm, uint64_t shared, s->perm_change_fd = rs->fd; } else { /* We may need a new fd if auto-read-only switches the mode */ - ret = raw_reconfigure_getfd(bs, bs->open_flags, &open_flags, + ret = raw_reconfigure_getfd(bs, bs->open_flags, &open_flags, perm, false, errp); if (ret < 0) { return ret; diff --git a/tests/qemu-iotests/232.out b/tests/qemu-iotests/232.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/232.out +++ b/tests/qemu-iotests/232.out @@ -XXX,XX +XXX,XX @@ NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) QEMU_PROG: -drive driver=file,file=TEST_DIR/t.IMGFMT,if=none,read-only=off,auto-read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied -NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) -NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) +NODE_NAME: TEST_DIR/t.IMGFMT (file) +NODE_NAME: TEST_DIR/t.IMGFMT (file) QEMU_PROG: -drive driver=file,file=TEST_DIR/t.IMGFMT,if=none,auto-read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied -NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) -NODE_NAME: TEST_DIR/t.IMGFMT (file, read-only) +NODE_NAME: TEST_DIR/t.IMGFMT (file) +NODE_NAME: TEST_DIR/t.IMGFMT (file) === -blockdev with read-write image: read-only/auto-read-only combinations === @@ -XXX,XX +XXX,XX @@ node0: TEST_DIR/t.IMGFMT (file, read-only) node0: TEST_DIR/t.IMGFMT (file, read-only) QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0,read-only=off,auto-read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied -node0: TEST_DIR/t.IMGFMT (file, read-only) +node0: TEST_DIR/t.IMGFMT (file) QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0,read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0,auto-read-only=off: Could not open 'TEST_DIR/t.IMGFMT': Permission denied -node0: TEST_DIR/t.IMGFMT (file, read-only) +node0: TEST_DIR/t.IMGFMT (file) QEMU_PROG: -blockdev driver=file,filename=TEST_DIR/t.IMGFMT,node-name=node0: Could not open 'TEST_DIR/t.IMGFMT': Permission denied === Try commit to backing file with auto-read-only === -- 2.20.1 From: Keith Busch <keith.busch@intel.com> The implementation used blocks units rather than the expected bytes. Fixes: c03e7ef12a9 ("nvme: Implement Write Zeroes") Reported-by: Ming Lei <ming.lei@redhat.com> Signed-off-by: Keith Busch <keith.busch@intel.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- hw/block/nvme.c | 6 +++--- 1 file changed, 3 insertions(+), 3 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 @@ static uint16_t nvme_write_zeros(NvmeCtrl *n, NvmeNamespace *ns, NvmeCmd *cmd, const uint8_t data_shift = ns->id_ns.lbaf[lba_index].ds; uint64_t slba = le64_to_cpu(rw->slba); uint32_t nlb = le16_to_cpu(rw->nlb) + 1; - uint64_t aio_slba = slba << (data_shift - BDRV_SECTOR_BITS); - uint32_t aio_nlb = nlb << (data_shift - BDRV_SECTOR_BITS); + uint64_t offset = slba << data_shift; + uint32_t count = nlb << data_shift; if (unlikely(slba + nlb > ns->id_ns.nsze)) { trace_nvme_err_invalid_lba_range(slba, nlb, ns->id_ns.nsze); @@ -XXX,XX +XXX,XX @@ static uint16_t nvme_write_zeros(NvmeCtrl *n, NvmeNamespace *ns, NvmeCmd *cmd, req->has_sg = false; block_acct_start(blk_get_stats(n->conf.blk), &req->acct, 0, BLOCK_ACCT_WRITE); - req->aiocb = blk_aio_pwrite_zeroes(n->conf.blk, aio_slba, aio_nlb, + req->aiocb = blk_aio_pwrite_zeroes(n->conf.blk, offset, count, BDRV_REQ_MAY_UNMAP, nvme_rw_cb, req); return NVME_NO_COMPLETE; } -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Our permission system is useful to define what operations are allowed on a certain block node and includes things like BLK_PERM_WRITE or BLK_PERM_RESIZE among others. One of the permissions is BLK_PERM_GRAPH_MOD which allows "changing the node that this BdrvChild points to". The exact meaning of this has never been very clear, but it can be understood as "change any of the links connected to the node". This can be used to prevent changing a backing link, but it's too coarse. This patch adds a new 'frozen' attribute to BdrvChild, which forbids detaching the link from the node it points to, and new API to freeze and unfreeze a backing chain. After this change a few functions can fail, so they need additional checks. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 5 +++ include/block/block_int.h | 6 ++++ block.c | 76 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 87 insertions(+) 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_drop_intermediate(BlockDriverState *top, BlockDriverState *base, BlockDriverState *bdrv_find_overlay(BlockDriverState *active, BlockDriverState *bs); BlockDriverState *bdrv_find_base(BlockDriverState *bs); +bool bdrv_is_backing_chain_frozen(BlockDriverState *bs, BlockDriverState *base, + Error **errp); +int bdrv_freeze_backing_chain(BlockDriverState *bs, BlockDriverState *base, + Error **errp); +void bdrv_unfreeze_backing_chain(BlockDriverState *bs, BlockDriverState *base); typedef struct BdrvCheckResult { 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 BdrvChild { uint64_t backup_perm; uint64_t backup_shared_perm; + /* + * This link is frozen: the child can neither be replaced nor + * detached from the parent. + */ + bool frozen; + QLIST_ENTRY(BdrvChild) next; QLIST_ENTRY(BdrvChild) next_parent; }; 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 *old_bs = child->bs; int i; + assert(!child->frozen); + if (old_bs && new_bs) { assert(bdrv_get_aio_context(old_bs) == bdrv_get_aio_context(new_bs)); } @@ -XXX,XX +XXX,XX @@ void bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd, bool update_inherits_from = bdrv_chain_contains(bs, backing_hd) && bdrv_inherits_from_recursive(backing_hd, bs); + if (bdrv_is_backing_chain_frozen(bs, backing_bs(bs), errp)) { + return; + } + if (backing_hd) { bdrv_ref(backing_hd); } @@ -XXX,XX +XXX,XX @@ void bdrv_replace_node(BlockDriverState *from, BlockDriverState *to, if (!should_update_child(c, to)) { continue; } + if (c->frozen) { + error_setg(errp, "Cannot change '%s' link to '%s'", + c->name, from->node_name); + goto out; + } list = g_slist_prepend(list, c); perm |= c->perm; shared &= c->shared_perm; @@ -XXX,XX +XXX,XX @@ BlockDriverState *bdrv_find_base(BlockDriverState *bs) return bdrv_find_overlay(bs, NULL); } +/* + * Return true if at least one of the backing links between @bs and + * @base is frozen. @errp is set if that's the case. + */ +bool bdrv_is_backing_chain_frozen(BlockDriverState *bs, BlockDriverState *base, + Error **errp) +{ + BlockDriverState *i; + + for (i = bs; i != base && i->backing; i = backing_bs(i)) { + if (i->backing->frozen) { + error_setg(errp, "Cannot change '%s' link from '%s' to '%s'", + i->backing->name, i->node_name, + backing_bs(i)->node_name); + return true; + } + } + + return false; +} + +/* + * Freeze all backing links between @bs and @base. + * If any of the links is already frozen the operation is aborted and + * none of the links are modified. + * Returns 0 on success. On failure returns < 0 and sets @errp. + */ +int bdrv_freeze_backing_chain(BlockDriverState *bs, BlockDriverState *base, + Error **errp) +{ + BlockDriverState *i; + + if (bdrv_is_backing_chain_frozen(bs, base, errp)) { + return -EPERM; + } + + for (i = bs; i != base && i->backing; i = backing_bs(i)) { + i->backing->frozen = true; + } + + return 0; +} + +/* + * Unfreeze all backing links between @bs and @base. The caller must + * ensure that all links are frozen before using this function. + */ +void bdrv_unfreeze_backing_chain(BlockDriverState *bs, BlockDriverState *base) +{ + BlockDriverState *i; + + for (i = bs; i != base && i->backing; i = backing_bs(i)) { + assert(i->backing->frozen); + i->backing->frozen = false; + } +} + /* * Drops images above 'base' up to and including 'top', and sets the image * above 'top' to have base as its backing file. @@ -XXX,XX +XXX,XX @@ int bdrv_drop_intermediate(BlockDriverState *top, BlockDriverState *base, goto exit; } + /* This function changes all links that point to top and makes + * them point to base. Check that none of them is frozen. */ + QLIST_FOREACH(c, &top->parents, next_parent) { + if (c->frozen) { + goto exit; + } + } + /* If 'base' recursively inherits from 'top' then we should set * base->inherits_from to top->inherits_from after 'top' and all * other intermediate nodes have been dropped. -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/commit.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) 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 @@ typedef struct CommitBlockJob { BlockDriverState *base_bs; BlockdevOnError on_error; bool base_read_only; + bool chain_frozen; char *backing_file_str; } CommitBlockJob; @@ -XXX,XX +XXX,XX @@ static int commit_prepare(Job *job) { CommitBlockJob *s = container_of(job, CommitBlockJob, common.job); + bdrv_unfreeze_backing_chain(s->commit_top_bs, s->base_bs); + s->chain_frozen = false; + /* Remove base node parent that still uses BLK_PERM_WRITE/RESIZE before * the normal backing chain can be restored. */ blk_unref(s->base); @@ -XXX,XX +XXX,XX @@ static void commit_abort(Job *job) CommitBlockJob *s = container_of(job, CommitBlockJob, common.job); BlockDriverState *top_bs = blk_bs(s->top); + if (s->chain_frozen) { + bdrv_unfreeze_backing_chain(s->commit_top_bs, s->base_bs); + } + /* Make sure commit_top_bs and top stay around until bdrv_replace_node() */ bdrv_ref(top_bs); bdrv_ref(s->commit_top_bs); @@ -XXX,XX +XXX,XX @@ void commit_start(const char *job_id, BlockDriverState *bs, } } + if (bdrv_freeze_backing_chain(commit_top_bs, base, errp) < 0) { + goto fail; + } + s->chain_frozen = true; + ret = block_job_add_bdrv(&s->common, "base", base, 0, BLK_PERM_ALL, errp); if (ret < 0) { goto fail; @@ -XXX,XX +XXX,XX @@ void commit_start(const char *job_id, BlockDriverState *bs, return; fail: + if (s->chain_frozen) { + bdrv_unfreeze_backing_chain(commit_top_bs, base); + } if (s->base) { blk_unref(s->base); } -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/mirror.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int mirror_exit_common(Job *job) } s->prepared = true; + if (bdrv_chain_contains(src, target_bs)) { + bdrv_unfreeze_backing_chain(mirror_top_bs, target_bs); + } + bdrv_release_dirty_bitmap(src, s->dirty_bitmap); /* Make sure that the source BDS doesn't go away during bdrv_replace_node, @@ -XXX,XX +XXX,XX @@ static void mirror_start_job(const char *job_id, BlockDriverState *bs, goto fail; } } + + if (bdrv_freeze_backing_chain(mirror_top_bs, target, errp) < 0) { + goto fail; + } } QTAILQ_INIT(&s->ops_in_flight); -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/stream.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/block/stream.c b/block/stream.c index XXXXXXX..XXXXXXX 100644 --- a/block/stream.c +++ b/block/stream.c @@ -XXX,XX +XXX,XX @@ typedef struct StreamBlockJob { BlockdevOnError on_error; char *backing_file_str; bool bs_read_only; + bool chain_frozen; } StreamBlockJob; static int coroutine_fn stream_populate(BlockBackend *blk, @@ -XXX,XX +XXX,XX @@ static int coroutine_fn stream_populate(BlockBackend *blk, return blk_co_preadv(blk, offset, qiov.size, &qiov, BDRV_REQ_COPY_ON_READ); } +static void stream_abort(Job *job) +{ + StreamBlockJob *s = container_of(job, StreamBlockJob, common.job); + + if (s->chain_frozen) { + BlockJob *bjob = &s->common; + bdrv_unfreeze_backing_chain(blk_bs(bjob->blk), s->base); + } +} + static int stream_prepare(Job *job) { StreamBlockJob *s = container_of(job, StreamBlockJob, common.job); @@ -XXX,XX +XXX,XX @@ static int stream_prepare(Job *job) Error *local_err = NULL; int ret = 0; + bdrv_unfreeze_backing_chain(bs, base); + s->chain_frozen = false; + if (bs->backing) { const char *base_id = NULL, *base_fmt = NULL; if (base) { @@ -XXX,XX +XXX,XX @@ static const BlockJobDriver stream_job_driver = { .free = block_job_free, .run = stream_run, .prepare = stream_prepare, + .abort = stream_abort, .clean = stream_clean, .user_resume = block_job_user_resume, .drain = block_job_drain, @@ -XXX,XX +XXX,XX @@ void stream_start(const char *job_id, BlockDriverState *bs, &error_abort); } + if (bdrv_freeze_backing_chain(bs, base, errp) < 0) { + job_early_fail(&s->common.job); + goto fail; + } + s->base = base; s->backing_file_str = g_strdup(backing_file_str); s->bs_read_only = bs_read_only; + s->chain_frozen = true; s->on_error = on_error; trace_stream_start(bs, base, s); -- 2.20.1 From: Alberto Garcia <berto@igalia.com> The bdrv_reopen_queue() function is used to create a queue with the BDSs that are going to be reopened and their new options. Once the queue is ready bdrv_reopen_multiple() is called to perform the operation. The original options from each one of the BDSs are kept, with the new options passed to bdrv_reopen_queue() applied on top of them. For "x-blockdev-reopen" we want a function that behaves much like "blockdev-add". We want to ignore the previous set of options so that only the ones actually specified by the user are applied, with the rest having their default values. One of the things that we need is a way to tell bdrv_reopen_queue() whether we want to keep the old set of options or not, and that's what this patch does. All current callers are setting this new parameter to true and x-blockdev-reopen will set it to false. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 3 ++- block.c | 34 +++++++++++++++++++--------------- block/replication.c | 4 ++-- qemu-io-cmds.c | 2 +- 4 files changed, 24 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 @@ BlockDriverState *bdrv_open(const char *filename, const char *reference, BlockDriverState *bdrv_new_open_driver(BlockDriver *drv, const char *node_name, int flags, Error **errp); BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue, - BlockDriverState *bs, QDict *options); + BlockDriverState *bs, QDict *options, + bool keep_old_opts); int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **errp); int bdrv_reopen_set_read_only(BlockDriverState *bs, bool read_only, Error **errp); diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, QDict *options, const BdrvChildRole *role, QDict *parent_options, - int parent_flags) + int parent_flags, + bool keep_old_opts) { assert(bs != NULL); @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, */ /* Old explicitly set values (don't overwrite by inherited value) */ - if (bs_entry) { - old_options = qdict_clone_shallow(bs_entry->state.explicit_options); - } else { - old_options = qdict_clone_shallow(bs->explicit_options); + if (bs_entry || keep_old_opts) { + old_options = qdict_clone_shallow(bs_entry ? + bs_entry->state.explicit_options : + bs->explicit_options); + bdrv_join_options(bs, options, old_options); + qobject_unref(old_options); } - bdrv_join_options(bs, options, old_options); - qobject_unref(old_options); explicit_options = qdict_clone_shallow(options); @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, flags = bdrv_get_flags(bs); } - /* Old values are used for options that aren't set yet */ - old_options = qdict_clone_shallow(bs->options); - bdrv_join_options(bs, options, old_options); - qobject_unref(old_options); + if (keep_old_opts) { + /* Old values are used for options that aren't set yet */ + old_options = qdict_clone_shallow(bs->options); + bdrv_join_options(bs, options, old_options); + qobject_unref(old_options); + } /* We have the final set of options so let's update the flags */ options_copy = qdict_clone_shallow(options); @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, g_free(child_key_dot); bdrv_reopen_queue_child(bs_queue, child->bs, new_child_options, - child->role, options, flags); + child->role, options, flags, keep_old_opts); } return bs_queue; @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue, BlockDriverState *bs, - QDict *options) + QDict *options, bool keep_old_opts) { - return bdrv_reopen_queue_child(bs_queue, bs, options, NULL, NULL, 0); + return bdrv_reopen_queue_child(bs_queue, bs, options, NULL, NULL, 0, + keep_old_opts); } /* @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_set_read_only(BlockDriverState *bs, bool read_only, qdict_put_bool(opts, BDRV_OPT_READ_ONLY, read_only); bdrv_subtree_drained_begin(bs); - queue = bdrv_reopen_queue(NULL, bs, opts); + queue = bdrv_reopen_queue(NULL, bs, opts, true); ret = bdrv_reopen_multiple(bdrv_get_aio_context(bs), queue, errp); bdrv_subtree_drained_end(bs); 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, QDict *opts = qdict_new(); qdict_put_bool(opts, BDRV_OPT_READ_ONLY, !writable); reopen_queue = bdrv_reopen_queue(reopen_queue, s->hidden_disk->bs, - opts); + opts, true); } if (s->orig_secondary_read_only) { QDict *opts = qdict_new(); qdict_put_bool(opts, BDRV_OPT_READ_ONLY, !writable); reopen_queue = bdrv_reopen_queue(reopen_queue, s->secondary_disk->bs, - opts); + opts, true); } if (reopen_queue) { 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) } bdrv_subtree_drained_begin(bs); - brq = bdrv_reopen_queue(NULL, bs, opts); + brq = bdrv_reopen_queue(NULL, bs, opts, true); bdrv_reopen_multiple(bdrv_get_aio_context(bs), brq, &local_err); bdrv_subtree_drained_end(bs); -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Children in QMP are specified with BlockdevRef / BlockdevRefOrNull, which can contain a set of child options, a child reference, or NULL. In optional attributes like "backing" it can also be missing. Only the first case (set of child options) is being handled properly by bdrv_reopen_queue(). This patch deals with all the others. Here's how these cases should be handled when bdrv_reopen_queue() is deciding what to do with each child of a BlockDriverState: 1) Set of child options: if the child was implicitly created (i.e inherits_from points to the parent) then the options are removed from the parent's options QDict and are passed to the child with a recursive bdrv_reopen_queue() call. This case was already working fine. 2) Child reference: there's two possibilites here. 2a) Reference to the current child: if the child was implicitly created then it is put in the reopen queue, keeping its current set of options (since this was a child reference there was no way to specify a different set of options). If the child is not implicit then it keeps its current set of options but it is not reopened (and therefore does not inherit any new option from the parent). 2b) Reference to a different BDS: the current child is not put in the reopen queue at all. Passing a reference to a different BDS can be used to replace a child, although at the moment no driver implements this, so it results in an error. In any case, the current child is not going to be reopened (and might in fact disappear if it's replaced) 3) NULL: This is similar to (2b). Although no driver allows this yet it can be used to detach the current child so it should not be put in the reopen queue. 4) Missing option: at the moment "backing" is the only case where this can happen. With "blockdev-add", leaving "backing" out means that the default backing file is opened. We don't want to open a new image during reopen, so we require that "backing" is always present. We'll relax this requirement a bit in the next patch. If keep_old_opts is true and "backing" is missing then this behaves like 2a (the current child is reopened). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 1 + block.c | 52 +++++++++++++++++++++++++++++++++++++------ 2 files changed, 46 insertions(+), 7 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 @@ typedef struct BDRVReopenState { BlockDriverState *bs; int flags; BlockdevDetectZeroesOptions detect_zeroes; + bool backing_missing; uint64_t perm, shared_perm; QDict *options; QDict *explicit_options; diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, bs_entry->state.perm = UINT64_MAX; bs_entry->state.shared_perm = 0; + /* + * If keep_old_opts is false then it means that unspecified + * options must be reset to their original value. We don't allow + * resetting 'backing' but we need to know if the option is + * missing in order to decide if we have to return an error. + */ + if (!keep_old_opts) { + bs_entry->state.backing_missing = + !qdict_haskey(options, "backing") && + !qdict_haskey(options, "backing.driver"); + } + QLIST_FOREACH(child, &bs->children, next) { - QDict *new_child_options; - char *child_key_dot; + QDict *new_child_options = NULL; + bool child_keep_old = keep_old_opts; /* reopen can only change the options of block devices that were * implicitly created and inherited options. For other (referenced) @@ -XXX,XX +XXX,XX @@ static BlockReopenQueue *bdrv_reopen_queue_child(BlockReopenQueue *bs_queue, continue; } - child_key_dot = g_strdup_printf("%s.", child->name); - qdict_extract_subqdict(explicit_options, NULL, child_key_dot); - qdict_extract_subqdict(options, &new_child_options, child_key_dot); - g_free(child_key_dot); + /* Check if the options contain a child reference */ + if (qdict_haskey(options, child->name)) { + const char *childref = qdict_get_try_str(options, child->name); + /* + * The current child must not be reopened if the child + * reference is null or points to a different node. + */ + if (g_strcmp0(childref, child->bs->node_name)) { + continue; + } + /* + * If the child reference points to the current child then + * reopen it with its existing set of options (note that + * it can still inherit new options from the parent). + */ + child_keep_old = true; + } else { + /* Extract child options ("child-name.*") */ + char *child_key_dot = g_strdup_printf("%s.", child->name); + qdict_extract_subqdict(explicit_options, NULL, child_key_dot); + qdict_extract_subqdict(options, &new_child_options, child_key_dot); + g_free(child_key_dot); + } bdrv_reopen_queue_child(bs_queue, child->bs, new_child_options, - child->role, options, flags, keep_old_opts); + child->role, options, flags, child_keep_old); } return bs_queue; @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue, drv_prepared = true; + if (drv->supports_backing && reopen_state->backing_missing) { + error_setg(errp, "backing is missing for '%s'", + reopen_state->bs->node_name); + ret = -EINVAL; + goto error; + } + /* Options that are not handled are only okay if they are unchanged * compared to the old state. It is expected that some options are only * used for the initial open, but not reopen (e.g. filename) */ -- 2.20.1 From: Alberto Garcia <berto@igalia.com> Of all options of type BlockdevRef used to specify children in BlockdevOptions, 'backing' is the only one that is optional. For "x-blockdev-reopen" we want that if an option is omitted then it must be reset to its default value. The default value of 'backing' means that QEMU opens the backing file specified in the image metadata, but this is not something that we want to support for the reopen operation. Because of this the 'backing' option has to be specified during reopen, pointing to the existing backing file if we want to keep it, or pointing to a different one (or NULL) if we want to replace it (to be implemented in a subsequent patch). In order to simplify things a bit and not to require that the user passes the 'backing' option to every single block device even when it's clearly not necessary, this patch allows omitting this option if the block device being reopened doesn't have a backing file attached _and_ no default backing file is specified in the image metadata. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 8 +++++++- 1 file changed, 7 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 @@ int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue, drv_prepared = true; - if (drv->supports_backing && reopen_state->backing_missing) { + /* + * We must provide the 'backing' option if the BDS has a backing + * file or if the image file has a backing file name as part of + * its metadata. Otherwise the 'backing' option can be omitted. + */ + if (drv->supports_backing && reopen_state->backing_missing && + (backing_bs(reopen_state->bs) || reopen_state->bs->backing_file[0])) { error_setg(errp, "backing is missing for '%s'", reopen_state->bs->node_name); ret = -EINVAL; -- 2.20.1 From: Alberto Garcia <berto@igalia.com> This patch allows the user to change the backing file of an image that is being reopened. Here's what it does: - In bdrv_reopen_prepare(): check that the value of 'backing' points to an existing node or is null. If it points to an existing node it also needs to make sure that replacing the backing file will not create a cycle in the node graph (i.e. you cannot reach the parent from the new backing file). - In bdrv_reopen_commit(): perform the actual node replacement by calling bdrv_set_backing_hd(). There may be temporary implicit nodes between a BDS and its backing file (e.g. a commit filter node). In these cases bdrv_reopen_prepare() looks for the real (non-implicit) backing file and requires that the 'backing' option points to it. Replacing or detaching a backing file is forbidden if there are implicit nodes in the middle. Although x-blockdev-reopen is meant to be used like blockdev-add, there's an important thing that must be taken into account: the only way to set a new backing file is by using a reference to an existing node (previously added with e.g. blockdev-add). If 'backing' contains a dictionary with a new set of options ({"driver": "qcow2", "file": { ... }}) then it is interpreted that the _existing_ backing file must be reopened with those options. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 2 + block.c | 166 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 168 insertions(+) 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 @@ typedef struct BDRVReopenState { int flags; BlockdevDetectZeroesOptions detect_zeroes; bool backing_missing; + bool replace_backing_bs; /* new_backing_bs is ignored if this is false */ + BlockDriverState *new_backing_bs; /* If NULL then detach the current bs */ uint64_t perm, shared_perm; QDict *options; QDict *explicit_options; 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, NULL, errp); } +/* + * Returns true if @child can be reached recursively from @bs + */ +static bool bdrv_recurse_has_child(BlockDriverState *bs, + BlockDriverState *child) +{ + BdrvChild *c; + + if (bs == child) { + return true; + } + + QLIST_FOREACH(c, &bs->children, next) { + if (bdrv_recurse_has_child(c->bs, child)) { + return true; + } + } + + return false; +} + /* * Adds a BlockDriverState to a simple queue for an atomic, transactional * reopen of multiple devices. @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **er if (ret < 0) { goto cleanup_perm; } + /* Check if new_backing_bs would accept the new permissions */ + if (state->replace_backing_bs && state->new_backing_bs) { + uint64_t nperm, nshared; + bdrv_child_perm(state->bs, state->new_backing_bs, + NULL, &child_backing, bs_queue, + state->perm, state->shared_perm, + &nperm, &nshared); + ret = bdrv_check_update_perm(state->new_backing_bs, NULL, + nperm, nshared, NULL, errp); + if (ret < 0) { + goto cleanup_perm; + } + } bs_entry->perms_checked = true; } @@ -XXX,XX +XXX,XX @@ cleanup_perm: bdrv_set_perm(state->bs, state->perm, state->shared_perm); } else { bdrv_abort_perm_update(state->bs); + if (state->replace_backing_bs && state->new_backing_bs) { + bdrv_abort_perm_update(state->new_backing_bs); + } } } cleanup: @@ -XXX,XX +XXX,XX @@ cleanup: qobject_unref(bs_entry->state.explicit_options); qobject_unref(bs_entry->state.options); } + if (bs_entry->state.new_backing_bs) { + bdrv_unref(bs_entry->state.new_backing_bs); + } g_free(bs_entry); } g_free(bs_queue); @@ -XXX,XX +XXX,XX @@ static void bdrv_reopen_perm(BlockReopenQueue *q, BlockDriverState *bs, *shared = cumulative_shared_perms; } +/* + * Take a BDRVReopenState and check if the value of 'backing' in the + * reopen_state->options QDict is valid or not. + * + * If 'backing' is missing from the QDict then return 0. + * + * If 'backing' contains the node name of the backing file of + * reopen_state->bs then return 0. + * + * If 'backing' contains a different node name (or is null) then check + * whether the current backing file can be replaced with the new one. + * If that's the case then reopen_state->replace_backing_bs is set to + * true and reopen_state->new_backing_bs contains a pointer to the new + * backing BlockDriverState (or NULL). + * + * Return 0 on success, otherwise return < 0 and set @errp. + */ +static int bdrv_reopen_parse_backing(BDRVReopenState *reopen_state, + Error **errp) +{ + BlockDriverState *bs = reopen_state->bs; + BlockDriverState *overlay_bs, *new_backing_bs; + QObject *value; + const char *str; + + value = qdict_get(reopen_state->options, "backing"); + if (value == NULL) { + return 0; + } + + switch (qobject_type(value)) { + case QTYPE_QNULL: + new_backing_bs = NULL; + break; + case QTYPE_QSTRING: + str = qobject_get_try_str(value); + new_backing_bs = bdrv_lookup_bs(NULL, str, errp); + if (new_backing_bs == NULL) { + return -EINVAL; + } else if (bdrv_recurse_has_child(new_backing_bs, bs)) { + error_setg(errp, "Making '%s' a backing file of '%s' " + "would create a cycle", str, bs->node_name); + return -EINVAL; + } + break; + default: + /* 'backing' does not allow any other data type */ + g_assert_not_reached(); + } + + /* + * TODO: before removing the x- prefix from x-blockdev-reopen we + * should move the new backing file into the right AioContext + * instead of returning an error. + */ + if (new_backing_bs) { + if (bdrv_get_aio_context(new_backing_bs) != bdrv_get_aio_context(bs)) { + error_setg(errp, "Cannot use a new backing file " + "with a different AioContext"); + return -EINVAL; + } + } + + /* + * Find the "actual" backing file by skipping all links that point + * to an implicit node, if any (e.g. a commit filter node). + */ + overlay_bs = bs; + while (backing_bs(overlay_bs) && backing_bs(overlay_bs)->implicit) { + overlay_bs = backing_bs(overlay_bs); + } + + /* If we want to replace the backing file we need some extra checks */ + if (new_backing_bs != backing_bs(overlay_bs)) { + /* Check for implicit nodes between bs and its backing file */ + if (bs != overlay_bs) { + error_setg(errp, "Cannot change backing link if '%s' has " + "an implicit backing file", bs->node_name); + return -EPERM; + } + /* Check if the backing link that we want to replace is frozen */ + if (bdrv_is_backing_chain_frozen(overlay_bs, backing_bs(overlay_bs), + errp)) { + return -EPERM; + } + reopen_state->replace_backing_bs = true; + if (new_backing_bs) { + bdrv_ref(new_backing_bs); + reopen_state->new_backing_bs = new_backing_bs; + } + } + + return 0; +} + /* * Prepares a BlockDriverState for reopen. All changes are staged in the * 'opaque' field of the BDRVReopenState, which is used and allocated by @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue, goto error; } + /* + * Allow changing the 'backing' option. The new value can be + * either a reference to an existing node (using its node name) + * or NULL to simply detach the current backing file. + */ + ret = bdrv_reopen_parse_backing(reopen_state, errp); + if (ret < 0) { + goto error; + } + qdict_del(reopen_state->options, "backing"); + /* Options that are not handled are only okay if they are unchanged * compared to the old state. It is expected that some options are only * used for the initial open, but not reopen (e.g. filename) */ @@ -XXX,XX +XXX,XX @@ void bdrv_reopen_commit(BDRVReopenState *reopen_state) bs->read_only = !(reopen_state->flags & BDRV_O_RDWR); bs->detect_zeroes = reopen_state->detect_zeroes; + if (reopen_state->replace_backing_bs) { + qdict_del(bs->explicit_options, "backing"); + qdict_del(bs->options, "backing"); + } + /* Remove child references from bs->options and bs->explicit_options. * Child options were already removed in bdrv_reopen_queue_child() */ QLIST_FOREACH(child, &bs->children, next) { @@ -XXX,XX +XXX,XX @@ void bdrv_reopen_commit(BDRVReopenState *reopen_state) qdict_del(bs->options, child->name); } + /* + * Change the backing file if a new one was specified. We do this + * after updating bs->options, so bdrv_refresh_filename() (called + * from bdrv_set_backing_hd()) has the new values. + */ + if (reopen_state->replace_backing_bs) { + BlockDriverState *old_backing_bs = backing_bs(bs); + assert(!old_backing_bs || !old_backing_bs->implicit); + /* Abort the permission update on the backing bs we're detaching */ + if (old_backing_bs) { + bdrv_abort_perm_update(old_backing_bs); + } + bdrv_set_backing_hd(bs, reopen_state->new_backing_bs, &error_abort); + } + bdrv_refresh_limits(bs, NULL); new_can_write = -- 2.20.1 From: Alberto Garcia <berto@igalia.com> If we reopen a BlockDriverState and there is an option that is present in bs->options but missing from the new set of options then we have to return an error unless the driver is able to reset it to its default value. This patch adds a new 'mutable_opts' field to BlockDriver. This is a list of runtime options that can be modified during reopen. If an option in this list is unspecified on reopen then it must be reset (or return an error). Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block_int.h | 8 ++++++++ block/file-posix.c | 6 ++++++ block/qcow2.c | 25 +++++++++++++++++++++++++ block/raw-format.c | 3 +++ 4 files changed, 42 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 @@ struct BlockDriver { /* List of options for creating images, terminated by name == NULL */ QemuOptsList *create_opts; + /* + * If this driver supports reopening images this contains a + * NULL-terminated list of the runtime options that can be + * modified. If an option in this list is unspecified during + * reopen then it _must_ be reset to its default value or return + * an error. + */ + const char *const *mutable_opts; /* * Returns 0 for completed check, -errno for internal errors. 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 QemuOptsList raw_runtime_opts = { }, }; +static const char *const mutable_opts[] = { "x-check-cache-dropped", NULL }; + static int raw_open_common(BlockDriverState *bs, QDict *options, int bdrv_flags, int open_flags, bool device, Error **errp) @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_file = { .bdrv_set_perm = raw_set_perm, .bdrv_abort_perm_update = raw_abort_perm_update, .create_opts = &raw_create_opts, + .mutable_opts = mutable_opts, }; /***********************************************/ @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_device = { .bdrv_reopen_abort = raw_reopen_abort, .bdrv_co_create_opts = hdev_co_create_opts, .create_opts = &raw_create_opts, + .mutable_opts = mutable_opts, .bdrv_co_invalidate_cache = raw_co_invalidate_cache, .bdrv_co_pwrite_zeroes = hdev_co_pwrite_zeroes, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_reopen_abort = raw_reopen_abort, .bdrv_co_create_opts = hdev_co_create_opts, .create_opts = &raw_create_opts, + .mutable_opts = mutable_opts, .bdrv_co_invalidate_cache = raw_co_invalidate_cache, @@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_host_cdrom = { .bdrv_reopen_abort = raw_reopen_abort, .bdrv_co_create_opts = hdev_co_create_opts, .create_opts = &raw_create_opts, + .mutable_opts = mutable_opts, .bdrv_co_preadv = raw_co_preadv, .bdrv_co_pwritev = raw_co_pwritev, 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 @@ int qcow2_validate_table(BlockDriverState *bs, uint64_t offset, return 0; } +static const char *const mutable_opts[] = { + QCOW2_OPT_LAZY_REFCOUNTS, + QCOW2_OPT_DISCARD_REQUEST, + QCOW2_OPT_DISCARD_SNAPSHOT, + QCOW2_OPT_DISCARD_OTHER, + QCOW2_OPT_OVERLAP, + QCOW2_OPT_OVERLAP_TEMPLATE, + QCOW2_OPT_OVERLAP_MAIN_HEADER, + QCOW2_OPT_OVERLAP_ACTIVE_L1, + QCOW2_OPT_OVERLAP_ACTIVE_L2, + QCOW2_OPT_OVERLAP_REFCOUNT_TABLE, + QCOW2_OPT_OVERLAP_REFCOUNT_BLOCK, + QCOW2_OPT_OVERLAP_SNAPSHOT_TABLE, + QCOW2_OPT_OVERLAP_INACTIVE_L1, + QCOW2_OPT_OVERLAP_INACTIVE_L2, + QCOW2_OPT_OVERLAP_BITMAP_DIRECTORY, + QCOW2_OPT_CACHE_SIZE, + QCOW2_OPT_L2_CACHE_SIZE, + QCOW2_OPT_L2_CACHE_ENTRY_SIZE, + QCOW2_OPT_REFCOUNT_CACHE_SIZE, + QCOW2_OPT_CACHE_CLEAN_INTERVAL, + NULL +}; + static QemuOptsList qcow2_runtime_opts = { .name = "qcow2", .head = QTAILQ_HEAD_INITIALIZER(qcow2_runtime_opts.head), @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_qcow2 = { .create_opts = &qcow2_create_opts, .strong_runtime_opts = qcow2_strong_runtime_opts, + .mutable_opts = mutable_opts, .bdrv_co_check = qcow2_co_check, .bdrv_amend_options = qcow2_amend_options, 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 @@ typedef struct BDRVRawState { bool has_size; } BDRVRawState; +static const char *const mutable_opts[] = { "offset", "size", NULL }; + static QemuOptsList raw_runtime_opts = { .name = "raw", .head = QTAILQ_HEAD_INITIALIZER(raw_runtime_opts.head), @@ -XXX,XX +XXX,XX @@ BlockDriver bdrv_raw = { .create_opts = &raw_create_opts, .bdrv_has_zero_init = &raw_has_zero_init, .strong_runtime_opts = raw_strong_runtime_opts, + .mutable_opts = mutable_opts, }; static void bdrv_raw_init(void) -- 2.20.1 From: Alberto Garcia <berto@igalia.com> bdrv_reopen_prepare() receives a BDRVReopenState with (among other things) a new set of options to be applied to that BlockDriverState. If an option is missing then it means that we want to reset it to its default value rather than keep the previous one. This way the state of the block device after being reopened is comparable to that of a device added with "blockdev-add" using the same set of options. Not all options from all drivers can be changed this way, however. If the user attempts to reset an immutable option to its default value using this method then we must forbid it. This new function takes a BlockDriverState and a new set of options and checks if there's any option that was previously set but is missing from the new set of options. If the option is present in both sets we don't need to check that they have the same value. The loop at the end of bdrv_reopen_prepare() already takes care of that. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 58 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 58 insertions(+) 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, NULL, errp); } +/* Return true if the NULL-terminated @list contains @str */ +static bool is_str_in_list(const char *str, const char *const *list) +{ + if (str && list) { + int i; + for (i = 0; list[i] != NULL; i++) { + if (!strcmp(str, list[i])) { + return true; + } + } + } + return false; +} + +/* + * Check that every option set in @bs->options is also set in + * @new_opts. + * + * Options listed in the common_options list and in + * @bs->drv->mutable_opts are skipped. + * + * Return 0 on success, otherwise return -EINVAL and set @errp. + */ +static int bdrv_reset_options_allowed(BlockDriverState *bs, + const QDict *new_opts, Error **errp) +{ + const QDictEntry *e; + /* These options are common to all block drivers and are handled + * in bdrv_reopen_prepare() so they can be left out of @new_opts */ + const char *const common_options[] = { + "node-name", "discard", "cache.direct", "cache.no-flush", + "read-only", "auto-read-only", "detect-zeroes", NULL + }; + + for (e = qdict_first(bs->options); e; e = qdict_next(bs->options, e)) { + if (!qdict_haskey(new_opts, e->key) && + !is_str_in_list(e->key, common_options) && + !is_str_in_list(e->key, bs->drv->mutable_opts)) { + error_setg(errp, "Option '%s' cannot be reset " + "to its default value", e->key); + return -EINVAL; + } + } + + return 0; +} + /* * Returns true if @child can be reached recursively from @bs */ @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue, } if (drv->bdrv_reopen_prepare) { + /* + * If a driver-specific option is missing, it means that we + * should reset it to its default value. + * But not all options allow that, so we need to check it first. + */ + ret = bdrv_reset_options_allowed(reopen_state->bs, + reopen_state->options, errp); + if (ret) { + goto error; + } + ret = drv->bdrv_reopen_prepare(reopen_state, queue, &local_err); if (ret) { if (local_err != NULL) { -- 2.20.1 From: Alberto Garcia <berto@igalia.com> This parameter has been unused since 1a63a907507fbbcfaee3f622907ec244b Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block.h | 2 +- block.c | 4 ++-- block/replication.c | 3 +-- qemu-io-cmds.c | 2 +- 4 files changed, 5 insertions(+), 6 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 @@ BlockDriverState *bdrv_new_open_driver(BlockDriver *drv, const char *node_name, BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue, BlockDriverState *bs, QDict *options, bool keep_old_opts); -int bdrv_reopen_multiple(AioContext *ctx, BlockReopenQueue *bs_queue, Error **errp); +int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp); int bdrv_reopen_set_read_only(BlockDriverState *bs, bool read_only, Error **errp); int bdrv_reopen_prepare(BDRVReopenState *reopen_state, diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue, * 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) +int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp) { int ret = -1; BlockReopenQueueEntry *bs_entry, *next; @@ -XXX,XX +XXX,XX @@ int bdrv_reopen_set_read_only(BlockDriverState *bs, bool read_only, bdrv_subtree_drained_begin(bs); queue = bdrv_reopen_queue(NULL, bs, opts, true); - ret = bdrv_reopen_multiple(bdrv_get_aio_context(bs), queue, errp); + ret = bdrv_reopen_multiple(queue, errp); 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, } if (reopen_queue) { - bdrv_reopen_multiple(bdrv_get_aio_context(bs), - reopen_queue, &local_err); + bdrv_reopen_multiple(reopen_queue, &local_err); error_propagate(errp, local_err); } 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) bdrv_subtree_drained_begin(bs); brq = bdrv_reopen_queue(NULL, bs, opts, true); - bdrv_reopen_multiple(bdrv_get_aio_context(bs), brq, &local_err); + bdrv_reopen_multiple(brq, &local_err); bdrv_subtree_drained_end(bs); if (local_err) { -- 2.20.1 From: Alberto Garcia <berto@igalia.com> This command allows reopening an arbitrary BlockDriverState with a new set of options. Some options (e.g node-name) cannot be changed and some block drivers don't allow reopening, but otherwise this command is modelled after 'blockdev-add' and the state of the reopened BlockDriverState should generally be the same as if it had just been added by 'blockdev-add' with the same set of options. One notable exception is the 'backing' option: 'x-blockdev-reopen' requires that it is always present unless the BlockDriverState in question doesn't have a current or default backing file. This command allows reconfiguring the graph by using the appropriate options to change the children of a node. At the moment it's possible to change a backing file by setting the 'backing' option to the name of the new node, but it should also be possible to add a similar functionality to other block drivers (e.g. Quorum, blkverify). Although the API is unlikely to change, this command is marked experimental for the time being so there's room to see if the semantics need changes. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qapi/block-core.json | 42 +++++++++++++++++++++++++++++++++++++++ blockdev.c | 47 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 89 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 @@ ## { 'command': 'blockdev-add', 'data': 'BlockdevOptions', 'boxed': true } +## +# @x-blockdev-reopen: +# +# Reopens a block device using the given set of options. Any option +# not specified will be reset to its default value regardless of its +# previous status. If an option cannot be changed or a particular +# driver does not support reopening then the command will return an +# error. +# +# The top-level @node-name option (from BlockdevOptions) must be +# specified and is used to select the block device to be reopened. +# Other @node-name options must be either omitted or set to the +# current name of the appropriate node. This command won't change any +# node name and any attempt to do it will result in an error. +# +# In the case of options that refer to child nodes, the behavior of +# this command depends on the value: +# +# 1) A set of options (BlockdevOptions): the child is reopened with +# the specified set of options. +# +# 2) A reference to the current child: the child is reopened using +# its existing set of options. +# +# 3) A reference to a different node: the current child is replaced +# with the specified one. +# +# 4) NULL: the current child (if any) is detached. +# +# Options (1) and (2) are supported in all cases, but at the moment +# only @backing allows replacing or detaching an existing child. +# +# Unlike with blockdev-add, the @backing option must always be present +# unless the node being reopened does not have a backing file and its +# image does not have a default backing file name as part of its +# metadata. +# +# Since: 4.0 +## +{ 'command': 'x-blockdev-reopen', + 'data': 'BlockdevOptions', 'boxed': true } + ## # @blockdev-del: # diff --git a/blockdev.c b/blockdev.c index XXXXXXX..XXXXXXX 100644 --- a/blockdev.c +++ b/blockdev.c @@ -XXX,XX +XXX,XX @@ fail: visit_free(v); } +void qmp_x_blockdev_reopen(BlockdevOptions *options, Error **errp) +{ + BlockDriverState *bs; + AioContext *ctx; + QObject *obj; + Visitor *v = qobject_output_visitor_new(&obj); + Error *local_err = NULL; + BlockReopenQueue *queue; + QDict *qdict; + + /* Check for the selected node name */ + if (!options->has_node_name) { + error_setg(errp, "Node name not specified"); + goto fail; + } + + bs = bdrv_find_node(options->node_name); + if (!bs) { + error_setg(errp, "Cannot find node named '%s'", options->node_name); + goto fail; + } + + /* Put all options in a QDict and flatten it */ + visit_type_BlockdevOptions(v, NULL, &options, &local_err); + if (local_err) { + error_propagate(errp, local_err); + goto fail; + } + + visit_complete(v, &obj); + qdict = qobject_to(QDict, obj); + + qdict_flatten(qdict); + + /* Perform the reopen operation */ + ctx = bdrv_get_aio_context(bs); + aio_context_acquire(ctx); + bdrv_subtree_drained_begin(bs); + queue = bdrv_reopen_queue(NULL, bs, qdict, false); + bdrv_reopen_multiple(queue, errp); + bdrv_subtree_drained_end(bs); + aio_context_release(ctx); + +fail: + visit_free(v); +} + void qmp_blockdev_del(const char *node_name, Error **errp) { AioContext *aio_context; -- 2.20.1 From: Alberto Garcia <berto@igalia.com> This patch adds several tests for the x-blockdev-reopen QMP command. Signed-off-by: Alberto Garcia <berto@igalia.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/qemu-iotests/245 | 991 +++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/245.out | 5 + tests/qemu-iotests/group | 1 + 3 files changed, 997 insertions(+) create mode 100644 tests/qemu-iotests/245 create mode 100644 tests/qemu-iotests/245.out diff --git a/tests/qemu-iotests/245 b/tests/qemu-iotests/245 new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/tests/qemu-iotests/245 @@ -XXX,XX +XXX,XX @@ +#!/usr/bin/env python +# +# Test cases for the QMP 'x-blockdev-reopen' command +# +# Copyright (C) 2018-2019 Igalia, S.L. +# Author: Alberto Garcia <berto@igalia.com> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +import os +import re +import iotests +import copy +import json +from iotests import qemu_img, qemu_io + +hd_path = [ + os.path.join(iotests.test_dir, 'hd0.img'), + os.path.join(iotests.test_dir, 'hd1.img'), + os.path.join(iotests.test_dir, 'hd2.img') +] + +def hd_opts(idx): + return {'driver': iotests.imgfmt, + 'node-name': 'hd%d' % idx, + 'file': {'driver': 'file', + 'node-name': 'hd%d-file' % idx, + 'filename': hd_path[idx] } } + +class TestBlockdevReopen(iotests.QMPTestCase): + total_io_cmds = 0 + + def setUp(self): + qemu_img('create', '-f', iotests.imgfmt, hd_path[0], '3M') + qemu_img('create', '-f', iotests.imgfmt, '-b', hd_path[0], hd_path[1]) + qemu_img('create', '-f', iotests.imgfmt, hd_path[2], '3M') + qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xa0 0 1M', hd_path[0]) + qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xa1 1M 1M', hd_path[1]) + qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xa2 2M 1M', hd_path[2]) + self.vm = iotests.VM() + self.vm.launch() + + def tearDown(self): + self.vm.shutdown() + self.check_qemu_io_errors() + os.remove(hd_path[0]) + os.remove(hd_path[1]) + os.remove(hd_path[2]) + + # The output of qemu-io is not returned by vm.hmp_qemu_io() but + # it's stored in the log and can only be read when the VM has been + # shut down. This function runs qemu-io and keeps track of the + # number of times it's been called. + def run_qemu_io(self, img, cmd): + result = self.vm.hmp_qemu_io(img, cmd) + self.assert_qmp(result, 'return', '') + self.total_io_cmds += 1 + + # Once the VM is shut down we can parse the log and see if qemu-io + # ran without errors. + def check_qemu_io_errors(self): + self.assertFalse(self.vm.is_running()) + found = 0 + log = self.vm.get_log() + for line in log.split("\n"): + if line.startswith("Pattern verification failed"): + raise Exception("%s (command #%d)" % (line, found)) + if re.match("read .*/.* bytes at offset", line): + found += 1 + self.assertEqual(found, self.total_io_cmds, + "Expected output of %d qemu-io commands, found %d" % + (found, self.total_io_cmds)) + + # Run x-blockdev-reopen with 'opts' but applying 'newopts' + # on top of it. The original 'opts' dict is unmodified + def reopen(self, opts, newopts = {}, errmsg = None): + opts = copy.deepcopy(opts) + + # Apply the changes from 'newopts' on top of 'opts' + for key in newopts: + value = newopts[key] + # If key has the form "foo.bar" then we need to do + # opts["foo"]["bar"] = value, not opts["foo.bar"] = value + subdict = opts + while key.find('.') != -1: + [prefix, key] = key.split('.', 1) + subdict = opts[prefix] + subdict[key] = value + + result = self.vm.qmp('x-blockdev-reopen', conv_keys = False, **opts) + if errmsg: + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', errmsg) + else: + self.assert_qmp(result, 'return', {}) + + + # Run query-named-block-nodes and return the specified entry + def get_node(self, node_name): + result = self.vm.qmp('query-named-block-nodes') + for node in result['return']: + if node['node-name'] == node_name: + return node + return None + + # Run 'query-named-block-nodes' and compare its output with the + # value passed by the user in 'graph' + def check_node_graph(self, graph): + result = self.vm.qmp('query-named-block-nodes') + self.assertEqual(json.dumps(graph, sort_keys=True), + json.dumps(result, sort_keys=True)) + + # This test opens one single disk image (without backing files) + # and tries to reopen it with illegal / incorrect parameters. + def test_incorrect_parameters_single_file(self): + # Open 'hd0' only (no backing files) + opts = hd_opts(0) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + original_graph = self.vm.qmp('query-named-block-nodes') + + # We can reopen the image passing the same options + self.reopen(opts) + + # We can also reopen passing a child reference in 'file' + self.reopen(opts, {'file': 'hd0-file'}) + + # We cannot change any of these + self.reopen(opts, {'node-name': 'not-found'}, "Cannot find node named 'not-found'") + self.reopen(opts, {'node-name': ''}, "Cannot find node named ''") + self.reopen(opts, {'node-name': None}, "Invalid parameter type for 'node-name', expected: string") + self.reopen(opts, {'driver': 'raw'}, "Cannot change the option 'driver'") + self.reopen(opts, {'driver': ''}, "Invalid parameter ''") + self.reopen(opts, {'driver': None}, "Invalid parameter type for 'driver', expected: string") + self.reopen(opts, {'file': 'not-found'}, "Cannot change the option 'file'") + self.reopen(opts, {'file': ''}, "Cannot change the option 'file'") + self.reopen(opts, {'file': None}, "Invalid parameter type for 'file', expected: BlockdevRef") + self.reopen(opts, {'file.node-name': 'newname'}, "Cannot change the option 'node-name'") + self.reopen(opts, {'file.driver': 'host_device'}, "Cannot change the option 'driver'") + self.reopen(opts, {'file.filename': hd_path[1]}, "Cannot change the option 'filename'") + self.reopen(opts, {'file.aio': 'native'}, "Cannot change the option 'aio'") + self.reopen(opts, {'file.locking': 'off'}, "Cannot change the option 'locking'") + self.reopen(opts, {'file.filename': None}, "Invalid parameter type for 'file.filename', expected: string") + + # node-name is optional in BlockdevOptions, but x-blockdev-reopen needs it + del opts['node-name'] + self.reopen(opts, {}, "Node name not specified") + + # Check that nothing has changed + self.check_node_graph(original_graph) + + # Remove the node + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + # This test opens an image with a backing file and tries to reopen + # it with illegal / incorrect parameters. + def test_incorrect_parameters_backing_file(self): + # Open hd1 omitting the backing options (hd0 will be opened + # with the default options) + opts = hd_opts(1) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + original_graph = self.vm.qmp('query-named-block-nodes') + + # We can't reopen the image passing the same options, 'backing' is mandatory + self.reopen(opts, {}, "backing is missing for 'hd1'") + + # Everything works if we pass 'backing' using the existing node name + for node in original_graph['return']: + if node['drv'] == iotests.imgfmt and node['file'] == hd_path[0]: + backing_node_name = node['node-name'] + self.reopen(opts, {'backing': backing_node_name}) + + # We can't use a non-existing or empty (non-NULL) node as the backing image + self.reopen(opts, {'backing': 'not-found'}, "Cannot find device= nor node_name=not-found") + self.reopen(opts, {'backing': ''}, "Cannot find device= nor node_name=") + + # We can reopen the image just fine if we specify the backing options + opts['backing'] = {'driver': iotests.imgfmt, + 'file': {'driver': 'file', + 'filename': hd_path[0]}} + self.reopen(opts) + + # We cannot change any of these options + self.reopen(opts, {'backing.node-name': 'newname'}, "Cannot change the option 'node-name'") + self.reopen(opts, {'backing.driver': 'raw'}, "Cannot change the option 'driver'") + self.reopen(opts, {'backing.file.node-name': 'newname'}, "Cannot change the option 'node-name'") + self.reopen(opts, {'backing.file.driver': 'host_device'}, "Cannot change the option 'driver'") + + # Check that nothing has changed since the beginning + self.check_node_graph(original_graph) + + # Remove the node + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd1') + self.assert_qmp(result, 'return', {}) + + # Reopen an image several times changing some of its options + def test_reopen(self): + # Open the hd1 image passing all backing options + opts = hd_opts(1) + opts['backing'] = hd_opts(0) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + original_graph = self.vm.qmp('query-named-block-nodes') + + # We can reopen the image passing the same options + self.reopen(opts) + + # Reopen in read-only mode + self.assert_qmp(self.get_node('hd1'), 'ro', False) + + self.reopen(opts, {'read-only': True}) + self.assert_qmp(self.get_node('hd1'), 'ro', True) + self.reopen(opts) + self.assert_qmp(self.get_node('hd1'), 'ro', False) + + # Change the cache options + self.assert_qmp(self.get_node('hd1'), 'cache/writeback', True) + self.assert_qmp(self.get_node('hd1'), 'cache/direct', False) + self.assert_qmp(self.get_node('hd1'), 'cache/no-flush', False) + self.reopen(opts, {'cache': { 'direct': True, 'no-flush': True }}) + self.assert_qmp(self.get_node('hd1'), 'cache/writeback', True) + self.assert_qmp(self.get_node('hd1'), 'cache/direct', True) + self.assert_qmp(self.get_node('hd1'), 'cache/no-flush', True) + + # Reopen again with the original options + self.reopen(opts) + self.assert_qmp(self.get_node('hd1'), 'cache/writeback', True) + self.assert_qmp(self.get_node('hd1'), 'cache/direct', False) + self.assert_qmp(self.get_node('hd1'), 'cache/no-flush', False) + + # Change 'detect-zeroes' and 'discard' + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'off') + self.reopen(opts, {'detect-zeroes': 'on'}) + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'on') + self.reopen(opts, {'detect-zeroes': 'unmap'}, + "setting detect-zeroes to unmap is not allowed " + + "without setting discard operation to unmap") + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'on') + self.reopen(opts, {'detect-zeroes': 'unmap', 'discard': 'unmap'}) + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'unmap') + self.reopen(opts) + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'off') + + # Changing 'force-share' is currently not supported + self.reopen(opts, {'force-share': True}, "Cannot change the option 'force-share'") + + # Change some qcow2-specific options + # No way to test for success other than checking the return message + if iotests.imgfmt == 'qcow2': + self.reopen(opts, {'l2-cache-entry-size': 128 * 1024}, + "L2 cache entry size must be a power of two "+ + "between 512 and the cluster size (65536)") + self.reopen(opts, {'l2-cache-size': 1024 * 1024, + 'cache-size': 512 * 1024}, + "l2-cache-size may not exceed cache-size") + self.reopen(opts, {'l2-cache-size': 4 * 1024 * 1024, + 'refcount-cache-size': 4 * 1024 * 1024, + 'l2-cache-entry-size': 32 * 1024}) + self.reopen(opts, {'pass-discard-request': True}) + + # Check that nothing has changed since the beginning + # (from the parameters that we can check) + self.check_node_graph(original_graph) + + # Check that the node names (other than the top-level one) are optional + del opts['file']['node-name'] + del opts['backing']['node-name'] + del opts['backing']['file']['node-name'] + self.reopen(opts) + self.check_node_graph(original_graph) + + # Reopen setting backing = null, this removes the backing image from the chain + self.reopen(opts, {'backing': None}) + self.assert_qmp_absent(self.get_node('hd1'), 'image/backing-image') + + # Open the 'hd0' image + result = self.vm.qmp('blockdev-add', conv_keys = False, **hd_opts(0)) + self.assert_qmp(result, 'return', {}) + + # Reopen the hd1 image setting 'hd0' as its backing image + self.reopen(opts, {'backing': 'hd0'}) + self.assert_qmp(self.get_node('hd1'), 'image/backing-image/filename', hd_path[0]) + + # Check that nothing has changed since the beginning + self.reopen(hd_opts(0), {'read-only': True}) + self.check_node_graph(original_graph) + + # The backing file (hd0) is now a reference, we cannot change backing.* anymore + self.reopen(opts, {}, "Cannot change the option 'backing.driver'") + + # We can't remove 'hd0' while it's a backing image of 'hd1' + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', "Node 'hd0' is busy: node is used as backing hd of 'hd1'") + + # But we can remove both nodes if done in the proper order + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd1') + self.assert_qmp(result, 'return', {}) + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + # Reopen a raw image and see the effect of changing the 'offset' option + def test_reopen_raw(self): + opts = {'driver': 'raw', 'node-name': 'hd0', + 'file': { 'driver': 'file', + 'filename': hd_path[0], + 'node-name': 'hd0-file' } } + + # First we create a 2MB raw file, and fill each half with a + # different value + qemu_img('create', '-f', 'raw', hd_path[0], '2M') + qemu_io('-f', 'raw', '-c', 'write -P 0xa0 0 1M', hd_path[0]) + qemu_io('-f', 'raw', '-c', 'write -P 0xa1 1M 1M', hd_path[0]) + + # Open the raw file with QEMU + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Read 1MB from offset 0 + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + + # Reopen the image with a 1MB offset. + # Now the results are different + self.reopen(opts, {'offset': 1024*1024}) + self.run_qemu_io("hd0", "read -P 0xa1 0 1M") + + # Reopen again with the original options. + # We get the original results again + self.reopen(opts) + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + + # Remove the block device + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + # Omitting an option should reset it to the default value, but if + # an option cannot be changed it shouldn't be possible to reset it + # to its default value either + def test_reset_default_values(self): + opts = {'driver': 'qcow2', 'node-name': 'hd0', + 'file': { 'driver': 'file', + 'filename': hd_path[0], + 'x-check-cache-dropped': True, # This one can be changed + 'locking': 'off', # This one can NOT be changed + 'node-name': 'hd0-file' } } + + # Open the file with QEMU + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # file.x-check-cache-dropped can be changed... + self.reopen(opts, { 'file.x-check-cache-dropped': False }) + # ...and dropped completely (resetting to the default value) + del opts['file']['x-check-cache-dropped'] + self.reopen(opts) + + # file.locking cannot be changed nor reset to the default value + self.reopen(opts, { 'file.locking': 'on' }, "Cannot change the option 'locking'") + del opts['file']['locking'] + self.reopen(opts, {}, "Option 'locking' cannot be reset to its default value") + # But we can reopen it if we maintain its previous value + self.reopen(opts, { 'file.locking': 'off' }) + + # Remove the block device + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + # This test modifies the node graph a few times by changing the + # 'backing' option on reopen and verifies that the guest data that + # is read afterwards is consistent with the graph changes. + def test_io_with_graph_changes(self): + opts = [] + + # Open hd0, hd1 and hd2 without any backing image + for i in range(3): + opts.append(hd_opts(i)) + opts[i]['backing'] = None + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts[i]) + self.assert_qmp(result, 'return', {}) + + # hd0 + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + self.run_qemu_io("hd0", "read -P 0 1M 1M") + self.run_qemu_io("hd0", "read -P 0 2M 1M") + + # hd1 <- hd0 + self.reopen(opts[0], {'backing': 'hd1'}) + + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + self.run_qemu_io("hd0", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd0", "read -P 0 2M 1M") + + # hd1 <- hd0 , hd1 <- hd2 + self.reopen(opts[2], {'backing': 'hd1'}) + + self.run_qemu_io("hd2", "read -P 0 0 1M") + self.run_qemu_io("hd2", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd2", "read -P 0xa2 2M 1M") + + # hd1 <- hd2 <- hd0 + self.reopen(opts[0], {'backing': 'hd2'}) + + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + self.run_qemu_io("hd0", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd0", "read -P 0xa2 2M 1M") + + # hd2 <- hd0 + self.reopen(opts[2], {'backing': None}) + + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + self.run_qemu_io("hd0", "read -P 0 1M 1M") + self.run_qemu_io("hd0", "read -P 0xa2 2M 1M") + + # hd2 <- hd1 <- hd0 + self.reopen(opts[1], {'backing': 'hd2'}) + self.reopen(opts[0], {'backing': 'hd1'}) + + self.run_qemu_io("hd0", "read -P 0xa0 0 1M") + self.run_qemu_io("hd0", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd0", "read -P 0xa2 2M 1M") + + # Illegal operation: hd2 is a child of hd1 + self.reopen(opts[2], {'backing': 'hd1'}, + "Making 'hd1' a backing file of 'hd2' would create a cycle") + + # hd2 <- hd0, hd2 <- hd1 + self.reopen(opts[0], {'backing': 'hd2'}) + + self.run_qemu_io("hd1", "read -P 0 0 1M") + self.run_qemu_io("hd1", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd1", "read -P 0xa2 2M 1M") + + # More illegal operations + self.reopen(opts[2], {'backing': 'hd1'}, + "Making 'hd1' a backing file of 'hd2' would create a cycle") + self.reopen(opts[2], {'file': 'hd0-file'}, "Cannot change the option 'file'") + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd2') + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', "Node 'hd2' is busy: node is used as backing hd of 'hd0'") + + # hd1 doesn't have a backing file now + self.reopen(opts[1], {'backing': None}) + + self.run_qemu_io("hd1", "read -P 0 0 1M") + self.run_qemu_io("hd1", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd1", "read -P 0 2M 1M") + + # We can't remove the 'backing' option if the image has a + # default backing file + del opts[1]['backing'] + self.reopen(opts[1], {}, "backing is missing for 'hd1'") + + self.run_qemu_io("hd1", "read -P 0 0 1M") + self.run_qemu_io("hd1", "read -P 0xa1 1M 1M") + self.run_qemu_io("hd1", "read -P 0 2M 1M") + + # This test verifies that we can't change the children of a block + # device during a reopen operation in a way that would create + # cycles in the node graph + def test_graph_cycles(self): + opts = [] + + # Open all three images without backing file + for i in range(3): + opts.append(hd_opts(i)) + opts[i]['backing'] = None + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts[i]) + self.assert_qmp(result, 'return', {}) + + # hd1 <- hd0, hd1 <- hd2 + self.reopen(opts[0], {'backing': 'hd1'}) + self.reopen(opts[2], {'backing': 'hd1'}) + + # Illegal: hd2 is backed by hd1 + self.reopen(opts[1], {'backing': 'hd2'}, + "Making 'hd2' a backing file of 'hd1' would create a cycle") + + # hd1 <- hd0 <- hd2 + self.reopen(opts[2], {'backing': 'hd0'}) + + # Illegal: hd2 is backed by hd0, which is backed by hd1 + self.reopen(opts[1], {'backing': 'hd2'}, + "Making 'hd2' a backing file of 'hd1' would create a cycle") + + # Illegal: hd1 cannot point to itself + self.reopen(opts[1], {'backing': 'hd1'}, + "Making 'hd1' a backing file of 'hd1' would create a cycle") + + # Remove all backing files + self.reopen(opts[0]) + self.reopen(opts[2]) + + ########################################## + # Add a blkverify node using hd0 and hd1 # + ########################################## + bvopts = {'driver': 'blkverify', + 'node-name': 'bv', + 'test': 'hd0', + 'raw': 'hd1'} + result = self.vm.qmp('blockdev-add', conv_keys = False, **bvopts) + self.assert_qmp(result, 'return', {}) + + # blkverify doesn't currently allow reopening. TODO: implement this + self.reopen(bvopts, {}, "Block format 'blkverify' used by node 'bv'" + + " does not support reopening files") + + # Illegal: hd0 is a child of the blkverify node + self.reopen(opts[0], {'backing': 'bv'}, + "Making 'bv' a backing file of 'hd0' would create a cycle") + + # Delete the blkverify node + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'bv') + self.assert_qmp(result, 'return', {}) + + # Misc reopen tests with different block drivers + def test_misc_drivers(self): + #################### + ###### quorum ###### + #################### + for i in range(3): + opts = hd_opts(i) + # Open all three images without backing file + opts['backing'] = None + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + opts = {'driver': 'quorum', + 'node-name': 'quorum0', + 'children': ['hd0', 'hd1', 'hd2'], + 'vote-threshold': 2} + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Quorum doesn't currently allow reopening. TODO: implement this + self.reopen(opts, {}, "Block format 'quorum' used by node 'quorum0'" + + " does not support reopening files") + + # You can't make quorum0 a backing file of hd0: + # hd0 is already a child of quorum0. + self.reopen(hd_opts(0), {'backing': 'quorum0'}, + "Making 'quorum0' a backing file of 'hd0' would create a cycle") + + # Delete quorum0 + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'quorum0') + self.assert_qmp(result, 'return', {}) + + # Delete hd0, hd1 and hd2 + for i in range(3): + result = self.vm.qmp('blockdev-del', conv_keys = True, + node_name = 'hd%d' % i) + self.assert_qmp(result, 'return', {}) + + ###################### + ###### blkdebug ###### + ###################### + opts = {'driver': 'blkdebug', + 'node-name': 'bd', + 'config': '/dev/null', + 'image': hd_opts(0)} + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # blkdebug allows reopening if we keep the same options + self.reopen(opts) + + # but it currently does not allow changes + self.reopen(opts, {'image': 'hd1'}, "Cannot change the option 'image'") + self.reopen(opts, {'align': 33554432}, "Cannot change the option 'align'") + self.reopen(opts, {'config': '/non/existent'}, "Cannot change the option 'config'") + del opts['config'] + self.reopen(opts, {}, "Option 'config' cannot be reset to its default value") + + # Delete the blkdebug node + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'bd') + self.assert_qmp(result, 'return', {}) + + ################## + ###### null ###### + ################## + opts = {'driver': 'null-aio', 'node-name': 'root', 'size': 1024} + + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # 1 << 30 is the default value, but we cannot change it explicitly + self.reopen(opts, {'size': (1 << 30)}, "Cannot change the option 'size'") + + # We cannot change 'size' back to its default value either + del opts['size'] + self.reopen(opts, {}, "Option 'size' cannot be reset to its default value") + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'root') + self.assert_qmp(result, 'return', {}) + + ################## + ###### file ###### + ################## + opts = hd_opts(0) + opts['file']['locking'] = 'on' + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # 'locking' cannot be changed + del opts['file']['locking'] + self.reopen(opts, {'file.locking': 'on'}) + self.reopen(opts, {'file.locking': 'off'}, "Cannot change the option 'locking'") + self.reopen(opts, {}, "Option 'locking' cannot be reset to its default value") + + # Trying to reopen the 'file' node directly does not make a difference + opts = opts['file'] + self.reopen(opts, {'locking': 'on'}) + self.reopen(opts, {'locking': 'off'}, "Cannot change the option 'locking'") + self.reopen(opts, {}, "Option 'locking' cannot be reset to its default value") + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + ###################### + ###### throttle ###### + ###################### + opts = { 'qom-type': 'throttle-group', 'id': 'group0', + 'props': { 'limits': { 'iops-total': 1000 } } } + result = self.vm.qmp('object-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + opts = { 'qom-type': 'throttle-group', 'id': 'group1', + 'props': { 'limits': { 'iops-total': 2000 } } } + result = self.vm.qmp('object-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Add a throttle filter with group = group0 + opts = { 'driver': 'throttle', 'node-name': 'throttle0', + 'throttle-group': 'group0', 'file': hd_opts(0) } + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # We can reopen it if we keep the same options + self.reopen(opts) + + # We can also reopen if 'file' is a reference to the child + self.reopen(opts, {'file': 'hd0'}) + + # This is illegal + self.reopen(opts, {'throttle-group': 'notfound'}, "Throttle group 'notfound' does not exist") + + # But it's possible to change the group to group1 + self.reopen(opts, {'throttle-group': 'group1'}) + + # Now group1 is in use, it cannot be deleted + result = self.vm.qmp('object-del', id = 'group1') + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', "object 'group1' is in use, can not be deleted") + + # Default options, this switches the group back to group0 + self.reopen(opts) + + # So now we cannot delete group0 + result = self.vm.qmp('object-del', id = 'group0') + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', "object 'group0' is in use, can not be deleted") + + # But group1 is free this time, and it can be deleted + result = self.vm.qmp('object-del', id = 'group1') + self.assert_qmp(result, 'return', {}) + + # Let's delete the filter node + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'throttle0') + self.assert_qmp(result, 'return', {}) + + # And we can finally get rid of group0 + result = self.vm.qmp('object-del', id = 'group0') + self.assert_qmp(result, 'return', {}) + + # If an image has a backing file then the 'backing' option must be + # passed on reopen. We don't allow leaving the option out in this + # case because it's unclear what the correct semantics would be. + def test_missing_backing_options_1(self): + # hd2 + opts = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # hd0 + opts = hd_opts(0) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # hd0 has no backing file: we can omit the 'backing' option + self.reopen(opts) + + # hd2 <- hd0 + self.reopen(opts, {'backing': 'hd2'}) + + # hd0 has a backing file: we must set the 'backing' option + self.reopen(opts, {}, "backing is missing for 'hd0'") + + # hd2 can't be removed because it's the backing file of hd0 + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd2') + self.assert_qmp(result, 'error/class', 'GenericError') + self.assert_qmp(result, 'error/desc', "Node 'hd2' is busy: node is used as backing hd of 'hd0'") + + # Detach hd2 from hd0. + self.reopen(opts, {'backing': None}) + self.reopen(opts, {}, "backing is missing for 'hd0'") + + # Remove both hd0 and hd2 + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd2') + self.assert_qmp(result, 'return', {}) + + # If an image has default backing file (as part of its metadata) + # then the 'backing' option must be passed on reopen. We don't + # allow leaving the option out in this case because it's unclear + # what the correct semantics would be. + def test_missing_backing_options_2(self): + # hd0 <- hd1 + # (hd0 is hd1's default backing file) + opts = hd_opts(1) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # hd1 has a backing file: we can't omit the 'backing' option + self.reopen(opts, {}, "backing is missing for 'hd1'") + + # Let's detach the backing file + self.reopen(opts, {'backing': None}) + + # No backing file attached to hd1 now, but we still can't omit the 'backing' option + self.reopen(opts, {}, "backing is missing for 'hd1'") + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd1') + self.assert_qmp(result, 'return', {}) + + # Test that making 'backing' a reference to an existing child + # keeps its current options + def test_backing_reference(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + # Enable 'detect-zeroes' on all three nodes + opts['detect-zeroes'] = 'on' + opts['backing']['detect-zeroes'] = 'on' + opts['backing']['backing']['detect-zeroes'] = 'on' + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Reopen the chain passing the minimum amount of required options. + # By making 'backing' a reference to hd1 (instead of a sub-dict) + # we tell QEMU to keep its current set of options. + opts = {'driver': iotests.imgfmt, + 'node-name': 'hd0', + 'file': 'hd0-file', + 'backing': 'hd1' } + self.reopen(opts) + + # This has reset 'detect-zeroes' on hd0, but not on hd1 and hd2. + self.assert_qmp(self.get_node('hd0'), 'detect_zeroes', 'off') + self.assert_qmp(self.get_node('hd1'), 'detect_zeroes', 'on') + self.assert_qmp(self.get_node('hd2'), 'detect_zeroes', 'on') + + # Test what happens if the graph changes due to other operations + # such as block-stream + def test_block_stream_1(self): + # hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = None + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Stream hd1 into hd0 and wait until it's done + result = self.vm.qmp('block-stream', conv_keys = True, job_id = 'stream0', device = 'hd0') + self.assert_qmp(result, 'return', {}) + self.wait_until_completed(drive = 'stream0') + + # Now we have only hd0 + self.assertEqual(self.get_node('hd1'), None) + + # We have backing.* options but there's no backing file anymore + self.reopen(opts, {}, "Cannot change the option 'backing.driver'") + + # If we remove the 'backing' option then we can reopen hd0 just fine + del opts['backing'] + self.reopen(opts) + + # We can also reopen hd0 if we set 'backing' to null + self.reopen(opts, {'backing': None}) + + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + + # Another block_stream test + def test_block_stream_2(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # Stream hd1 into hd0 and wait until it's done + result = self.vm.qmp('block-stream', conv_keys = True, job_id = 'stream0', + device = 'hd0', base_node = 'hd2') + self.assert_qmp(result, 'return', {}) + self.wait_until_completed(drive = 'stream0') + + # The chain is hd2 <- hd0 now. hd1 is missing + self.assertEqual(self.get_node('hd1'), None) + + # The backing options in the dict were meant for hd1, but we cannot + # use them with hd2 because hd1 had a backing file while hd2 does not. + self.reopen(opts, {}, "Cannot change the option 'backing.driver'") + + # If we remove hd1's options from the dict then things work fine + opts['backing'] = opts['backing']['backing'] + self.reopen(opts) + + # We can also reopen hd0 if we use a reference to the backing file + self.reopen(opts, {'backing': 'hd2'}) + + # But we cannot leave the option out + del opts['backing'] + self.reopen(opts, {}, "backing is missing for 'hd0'") + + # Now we can delete hd0 (and hd2) + result = self.vm.qmp('blockdev-del', conv_keys = True, node_name = 'hd0') + self.assert_qmp(result, 'return', {}) + self.assertEqual(self.get_node('hd2'), None) + + # Reopen the chain during a block-stream job (from hd1 to hd0) + def test_block_stream_3(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # hd2 <- hd0 + result = self.vm.qmp('block-stream', conv_keys = True, job_id = 'stream0', + device = 'hd0', base_node = 'hd2', speed = 512 * 1024) + self.assert_qmp(result, 'return', {}) + + # We can't remove hd2 while the stream job is ongoing + opts['backing']['backing'] = None + self.reopen(opts, {}, "Cannot change 'backing' link from 'hd1' to 'hd2'") + + # We can't remove hd1 while the stream job is ongoing + opts['backing'] = None + self.reopen(opts, {}, "Cannot change 'backing' link from 'hd0' to 'hd1'") + + self.wait_until_completed(drive = 'stream0') + + # Reopen the chain during a block-stream job (from hd2 to hd1) + def test_block_stream_4(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + # hd1 <- hd0 + result = self.vm.qmp('block-stream', conv_keys = True, job_id = 'stream0', + device = 'hd1', speed = 512 * 1024) + self.assert_qmp(result, 'return', {}) + + # We can't reopen with the original options because that would + # make hd1 read-only and block-stream requires it to be read-write + self.reopen(opts, {}, "Can't set node 'hd1' to r/o with copy-on-read enabled") + + # We can't remove hd2 while the stream job is ongoing + opts['backing']['backing'] = None + self.reopen(opts, {'backing.read-only': False}, "Cannot change 'backing' link from 'hd1' to 'hd2'") + + # We can detach hd1 from hd0 because it doesn't affect the stream job + opts['backing'] = None + self.reopen(opts) + + self.wait_until_completed(drive = 'stream0') + + # Reopen the chain during a block-commit job (from hd0 to hd2) + def test_block_commit_1(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('block-commit', conv_keys = True, job_id = 'commit0', + device = 'hd0', speed = 1024 * 1024) + self.assert_qmp(result, 'return', {}) + + # We can't remove hd2 while the commit job is ongoing + opts['backing']['backing'] = None + self.reopen(opts, {}, "Cannot change 'backing' link from 'hd1' to 'hd2'") + + # We can't remove hd1 while the commit job is ongoing + opts['backing'] = None + self.reopen(opts, {}, "Cannot change 'backing' link from 'hd0' to 'hd1'") + + event = self.vm.event_wait(name='BLOCK_JOB_READY') + self.assert_qmp(event, 'data/device', 'commit0') + self.assert_qmp(event, 'data/type', 'commit') + self.assert_qmp_absent(event, 'data/error') + + result = self.vm.qmp('block-job-complete', device='commit0') + self.assert_qmp(result, 'return', {}) + + self.wait_until_completed(drive = 'commit0') + + # Reopen the chain during a block-commit job (from hd1 to hd2) + def test_block_commit_2(self): + # hd2 <- hd1 <- hd0 + opts = hd_opts(0) + opts['backing'] = hd_opts(1) + opts['backing']['backing'] = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('block-commit', conv_keys = True, job_id = 'commit0', + device = 'hd0', top_node = 'hd1', speed = 1024 * 1024) + self.assert_qmp(result, 'return', {}) + + # We can't remove hd2 while the commit job is ongoing + opts['backing']['backing'] = None + self.reopen(opts, {}, "Cannot change the option 'backing.driver'") + + # We can't remove hd1 while the commit job is ongoing + opts['backing'] = None + self.reopen(opts, {}, "Cannot change backing link if 'hd0' has an implicit backing file") + + # hd2 <- hd0 + self.wait_until_completed(drive = 'commit0') + + self.assert_qmp(self.get_node('hd0'), 'ro', False) + self.assertEqual(self.get_node('hd1'), None) + self.assert_qmp(self.get_node('hd2'), 'ro', True) + + # We don't allow setting a backing file that uses a different AioContext + def test_iothreads(self): + opts = hd_opts(0) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts) + self.assert_qmp(result, 'return', {}) + + opts2 = hd_opts(2) + result = self.vm.qmp('blockdev-add', conv_keys = False, **opts2) + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('object-add', qom_type='iothread', id='iothread0') + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('object-add', qom_type='iothread', id='iothread1') + self.assert_qmp(result, 'return', {}) + + result = self.vm.qmp('x-blockdev-set-iothread', node_name='hd0', iothread='iothread0') + self.assert_qmp(result, 'return', {}) + + self.reopen(opts, {'backing': 'hd2'}, "Cannot use a new backing file with a different AioContext") + + result = self.vm.qmp('x-blockdev-set-iothread', node_name='hd2', iothread='iothread1') + self.assert_qmp(result, 'return', {}) + + self.reopen(opts, {'backing': 'hd2'}, "Cannot use a new backing file with a different AioContext") + + result = self.vm.qmp('x-blockdev-set-iothread', node_name='hd2', iothread='iothread0') + self.assert_qmp(result, 'return', {}) + + self.reopen(opts, {'backing': 'hd2'}) + +if __name__ == '__main__': + iotests.main(supported_fmts=["qcow2"]) diff --git a/tests/qemu-iotests/245.out b/tests/qemu-iotests/245.out new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/tests/qemu-iotests/245.out @@ -XXX,XX +XXX,XX @@ +.................. +---------------------------------------------------------------------- +Ran 18 tests + +OK diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/group +++ b/tests/qemu-iotests/group @@ -XXX,XX +XXX,XX @@ 242 rw auto quick 243 rw auto quick 244 rw auto quick +245 rw auto -- 2.20.1 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