block: Fix some filename generation issues
v8 cover block: Fix some filename generation issues
I'm sparing myself writing this cover letter again, and I'll just give you a link to one of the previous versions:
http://lists.nongnu.org/archive/html/qemu-block/2017-09/msg01030.html
The main difference is that I dropped patch 16 which added a QAPI @base-directory option for any node that could be used to override the base directory to be used for resolving relative filenames. (Because Berto and Kevin convinced me it's not that useful now -- and the patch is pretty stand-alone, so we can always add it later if needed.)
In turn, I had to add patch 4 because test 191 was broken by this series.
v8: Rebased, and:
- Patch 3: Actually just contextual difference
- Patch 4: Rebase conflict: 191 now filters its output a bit more, so there are differences in the reference output ("format-specific" is missing and %s/qcow2/IMGFMT/g)
- Patch 5:
- Drop superfluous "&& !bs->backing" in one else branch following a
"bs->backing" if [Berto]
- Same 191.out conflicts as in patch 4
- Patch 18: Indentation conflicts in block/sheepdog.c
- Patch 20:
- Dropped now-missing "base-directory" option from the global_options array [Berto]
- Same old 191.out conflicts
- Patch 21:
- Trivial rebase conflict because patch 5 has changed
- For blkdebug, both "image" and "x-image" may contain child-related information, so we need to consider both in its .bdrv_refresh_filename implementation [Berto]
- Patch 24: Added [Berto]
- Patch 25: In curl, don't "silently" encode the default values in its .bdrv_refresh_filename(), but use explicit macros (except for "readahead" and "timeout", which we can ignore, and for everything that is a pointer, because the default value is rather obviously NULL; so that actually just leaves us with "sslverify") [Berto]
git-backport-diff against v7:
Key:
[----] : patches are identical
[####] : number of functional differences between upstream/downstream patch [down] : patch is downstream-only
The flags [FC] indicate (F)unctional and (C)ontextual differences, respectively
001/26:[----] [--] 'block/mirror: Small absolute-paths simplification' 002/26:[----] [--] 'block: Use children list in bdrv_refresh_filename' 003/26:[0002] [FC] 'block: Add BDS.backing_overridden' 004/26:[0074] [FC] 'iotests: Drop explicit base blockdev in 191' 005/26:[0050] [FC] 'block: Respect backing bs in bdrv_refresh_filename' 006/26:[----] [--] 'block: Make path_combine() return the path' 007/26:[----] [--] 'block: bdrv_get_full_backing_filename_from_...'s ret. val.' 008/26:[----] [--] 'block: bdrv_get_full_backing_filename's ret. val.' 009/26:[----] [--] 'block: Add bdrv_make_absolute_filename()' 010/26:[----] [--] 'block: Fix bdrv_find_backing_image()' 011/26:[----] [--] 'block: Add bdrv_dirname()'
012/26:[----] [--] 'blkverify: Make bdrv_dirname() return NULL' 013/26:[----] [--] 'quorum: Make bdrv_dirname() return NULL' 014/26:[----] [-C] 'block/nbd: Make bdrv_dirname() return NULL' 015/26:[----] [--] 'block/nfs: Implement bdrv_dirname()' 016/26:[----] [--] 'block: Use bdrv_dirname() for relative filenames' 017/26:[----] [--] 'iotests: Add quorum case to test 110' 018/26:[0006] [FC] 'block: Add sgfnt_runtime_opts to BlockDriver' 019/26:[----] [--] 'block: Add BlockDriver.bdrv_gather_child_options' 020/26:[0050] [FC] 'block: Generically refresh runtime options' 021/26:[0005] [FC] 'block: Purify .bdrv_refresh_filename()' 022/26:[----] [--] 'block: Do not copy exact_filename from format file' 023/26:[----] [--] 'block: Fix FIXME from "Add BDS.backing_overridden"' 024/26:[down] 'block/curl: Harmonize option defaults' 025/26:[0007] [FC] 'block/curl: Implement bdrv_refresh_filename()' 026/26:[----] [--] 'block/null: Generate filename even with latency-ns'
Max Reitz (26):
block/mirror: Small absolute-paths simplification
block: Use children list in bdrv_refresh_filename
block: Add BDS.backing_overridden
iotests: Drop explicit base blockdev in 191
block: Respect backing bs in bdrv_refresh_filename
block: Make path_combine() return the path
block: bdrv_get_full_backing_filename_from_...'s ret. val.
block: bdrv_get_full_backing_filename's ret. val.
block: Add bdrv_make_absolute_filename()
block: Fix bdrv_find_backing_image()
block: Add bdrv_dirname()
blkverify: Make bdrv_dirname() return NULL
quorum: Make bdrv_dirname() return NULL block/nbd: Make bdrv_dirname() return NULL block/nfs: Implement bdrv_dirname()
block: Use bdrv_dirname() for relative filenames
iotests: Add quorum case to test 110
block: Add sgfnt_runtime_opts to BlockDriver
block: Add BlockDriver.bdrv_gather_child_options
block: Generically refresh runtime options
block: Purify .bdrv_refresh_filename()
block: Do not copy exact_filename from format file
block: Fix FIXME from "Add BDS.backing_overridden" block/curl: Harmonize option defaults block/curl: Implement bdrv_refresh_filename() block/null: Generate filename even with latency-ns
include/block/block.h | 15 +- include/block/block_int.h | 38 +++- block.c | 507 ++++++++++++++++++++++++++++-------------- block/blkdebug.c | 72 +++--- block/blkverify.c | 29 +-- block/commit.c | 3 +- block/crypto.c | 8 + block/curl.c | 55 ++++- block/gluster.c | 19 ++ block/iscsi.c | 18 ++ block/mirror.c | 19 +- block/nbd.c | 50 +++-- block/nfs.c | 53 ++--- block/null.c | 33 ++- block/qapi.c | 12 +- block/qcow.c | 7 + block/qcow2.c | 7 + block/quorum.c | 69 ++++-- block/raw-format.c | 10 +- block/rbd.c | 14 ++ block/replication.c | 8 + block/sheepdog.c | 12 + block/ssh.c | 12 + block/throttle.c | 7 + block/vmdk.c | 25 ++- block/vpc.c | 7 + block/vvfat.c | 12 + block/vxhs.c | 11 + blockdev.c | 16 ++ tests/qemu-iotests/051.out | 8 +- tests/qemu-iotests/051.pc.out | 8 +- tests/qemu-iotests/110 | 29 ++- tests/qemu-iotests/110.out | 9 +- tests/qemu-iotests/191 | 3 +- tests/qemu-iotests/191.out | 76 +++---- 35 files changed, 884 insertions(+), 397 deletions(-)
Signature · 1 line
2.14.3
PATCH v8 1/26 block/mirror: Small absolute-paths simplification
When invoking drive-mirror in absolute-paths mode, the target's backing BDS is assigned to it in mirror_exit(). The current logic only does so if the target does not have that backing BDS already; but it actually cannot have a backing BDS at all (the BDS is opened with O_NO_BACKING in qmp_drive_mirror()), so just assert that and assign the new backing BDS unconditionally.
--- block/mirror.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
block/mirror.c +6 -6
@@ -523,12 +523,12 @@ static void mirror_exit(BlockJob *job, void *opaque) &error_abort);if (s->backing_mode ==MIRROR_SOURCE_BACKING_CHAIN ) {BlockDriverState *backing = s->is_none_mode ? src : s->base; -if (backing_bs (target_bs) != backing) { -bdrv_set_backing_hd (target_bs, backing, &local_err); -if (local_err) { -error_report_err (local_err); - data->ret = -EPERM ; - } + +assert (!target_bs->backing); +bdrv_set_backing_hd (target_bs, backing, &local_err); +if (local_err) { +error_report_err (local_err); + data->ret = -EPERM ; } }
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 01/26] block/mirror: Small absolute-paths simplification
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 9 lines
When invoking drive-mirror in absolute-paths mode, the target's backing BDS is assigned to it in
mirror_exit(). The current logic only does so if the target does not have that backing BDS already; but it actually cannot have a backing BDS at all (the BDS is opened withO_NO_BACKINGinqmp_drive_mirror()), so just assert that and assign the new backing BDS unconditionally.Signed-off-by Max Reitz <[email protected]>Reviewed-by Alberto Garcia <[email protected]>
Introducing new assumptions that the graph stays unmodified between the start of a block job and its completion feels a bit adventurous when all of the blockdev work is moving towards making things more flexible.
If we really do want to enforce this, I guess you finally found a use case for BLK_PERM_GRAPH_MOD.
Kevin
Re: [Qemu-devel] [PATCH v8 01/26] block/mirror: Small absolute-paths simplification
Quoted 8 lines · Kevin Wolf
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 7 lines
When invoking drive-mirror in absolute-paths mode, the target's backing BDS is assigned to it in
mirror_exit(). The current logic only does so if the target does not have that backing BDS already; but it actually cannot have a backing BDS at all (the BDS is opened withO_NO_BACKINGinqmp_drive_mirror()), so just assert that and assign the new backing BDS unconditionally.
Quoted 6 lines
Quoted 2 lines
Signed-off-by Max Reitz <[email protected]>Reviewed-by Alberto Garcia <[email protected]>Introducing new assumptions that the graph stays unmodified between the start of a block job and its completion feels a bit adventurous when all of the blockdev work is moving towards making things more flexible.
But as the commit message hints at, MIRROR_SOURCE_BACKING_CHAIN is used only by drive-mirror. How flexible do you intend that to be?
Yes, technically you can specify a node name for the newly created image and then go wild while the block job is still running. But then something inside me feels like you should see your VM crashing for that... OTOH, a crash is always bad.
I can't make it crash in the current state because the mirror job creates a BB and blockdev-snapshot thus refuses to attach a backing file to the target. But I guess I'll drop this patch anyway (unless I find it was important for some other patch in this series, which I hope it wasn't...).
On a totally unrelated note, we need some protocol for hurting the user if they do something that deserves a bit of pain. Crashing is always bad and our fault, so it needs to be something else. Removing $HOME? Marking all attached qcow2 images corrupt even though they aren't so you have to run a qemu-img check? I like that one.
Max
PATCH v8 2/26 block: Use children list in bdrv_refresh_filename
In reply to Max Reitzbdrv_refresh_filename() should invoke itself recursively on all children, not just on file.
With that change, we can remove the manual invocations in blkverify, quorum, commit, and mirror.
--- block.c | 9 +++++---- block/blkverify.c | 3 --- block/commit.c | 1 - block/mirror.c | 1 - block/quorum.c | 1 - 5 files changed, 5 insertions(+), 10 deletions(-)
block.c +5 -4
@@ -5020,16 +5020,17 @@ static bool append_open_options(QDict *d, BlockDriverState *bs)void bdrv_refresh_filename (BlockDriverState *bs) {BlockDriver *drv = bs->drv; +BdrvChild *child;QDict *opts;if (!drv) {return ; } -/* This BDS's file name will most probably depend on its file's name, so -* refresh that first */ -if (bs->file) { -bdrv_refresh_filename (bs->file->bs); +/* This BDS's file name may depend on any of its children's file names, so +* refresh those first */ + QLIST_FOREACH(child, &bs->children, next) { +bdrv_refresh_filename (child->bs); }if (drv->bdrv_refresh_filename) {
block/blkverify.c +0 -3
@@ -281,9 +281,6 @@ static void blkverify_refresh_filename(BlockDriverState *bs, QDict *options) {BDRVBlkverifyState *s = bs->opaque; -/* bs->file->bs has already been refreshed */ -bdrv_refresh_filename (s->test_file->bs); -if (bs->file->bs->full_open_options && s->test_file->bs->full_open_options) {
block/commit.c +0 -1
@@ -241,7 +241,6 @@ static int coroutine_fn bdrv_commit_top_preadv(BlockDriverState *bs,static void bdrv_commit_top_refresh_filename (BlockDriverState *bs,QDict *opts) { -bdrv_refresh_filename (bs->backing->bs);pstrcpy (bs->exact_filename,sizeof (bs->exact_filename), bs->backing->bs->filename); }
block/mirror.c +0 -1
@@ -1061,7 +1061,6 @@ static void bdrv_mirror_top_refresh_filename(BlockDriverState *bs, QDict *opts)* bdrv_set_backing_hd */ return ; } -bdrv_refresh_filename (bs->backing->bs);pstrcpy (bs->exact_filename,sizeof (bs->exact_filename), bs->backing->bs->filename); }
block/quorum.c +0 -1
@@ -1074,7 +1074,6 @@ static void quorum_refresh_filename(BlockDriverState *bs, QDict *options)int i;for (i =0 ; i < s->num_children; i++) { -bdrv_refresh_filename (s->children[i]->bs);if (!s->children[i]->bs->full_open_options) {return ; }
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 02/26] block: Use children list in bdrv_refresh_filename
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 8 lines
bdrv_refresh_filename()should invoke itself recursively on all children, not just on file.With that change, we can remove the manual invocations in blkverify, quorum, commit, and mirror.
Signed-off-by Max Reitz <[email protected]>Reviewed-by Alberto Garcia <[email protected]>
PATCH v8 3/26 block: Add BDS.backing_overridden
In reply to Max ReitzIf the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in bdrv_refresh_filename().
Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
This commit adds a FIXME which will be remedied by a follow-up commit. Until then, the respective piece of code will not result in any behavior that is worse than what we currently have.
--- include/block/block_int.h | 1 + block.c | 13 +++++++++++++ block/mirror.c | 4 ++++ blockdev.c | 16 ++++++++++++++++ 4 files changed, 34 insertions(+)
include/block/block_int.h +1 -0
@@ -619,6 +619,7 @@ struct BlockDriverState {char backing_file[PATH_MAX ];/* if non zero, the image is a diff of this file image */ char backing_format[16 ];/* if non-zero and backing_file exists */ +bool backing_overridden;/* backing file has been specified by the user */ QDict *full_open_options;char exact_filename[PATH_MAX ];
block.c +13 -0
@@ -2268,6 +2268,11 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options, reference =qdict_get_try_str (parent_options, bdref_key);if (reference ||qdict_haskey (options,"file.filename" )) { backing_filename[0 ] ='\0' ; + +/* FIXME: Should also be set to true if @options contains other runtime +* options which control the data that is read from the backing +* BDS */ + bs->backing_overridden = true; }else if (bs->backing_file[0 ] =='\0' &&qdict_size (options) ==0 ) {QDECREF (options); goto free_exit;
@@ -2469,6 +2474,9 @@ static BlockDriverState *bdrv_append_temp_snapshot(BlockDriverState *bs, goto out; } + bs_snapshot->backing_overridden = true; +bdrv_refresh_filename (bs_snapshot); + out:QDECREF (snapshot_options);g_free (tmp_filename);
@@ -2599,6 +2607,7 @@ static BlockDriverState *bdrv_open_inherit(const char *filename, backing =qdict_get_try_str (options,"backing" );if (backing && *backing =='\0' ) { flags |=BDRV_O_NO_BACKING ; + bs->backing_overridden = true;qdict_del (options,"backing" ); }
@@ -5031,6 +5040,10 @@ void bdrv_refresh_filename(BlockDriverState *bs)* refresh those first */ QLIST_FOREACH(child, &bs->children, next) {bdrv_refresh_filename (child->bs); + +if (child->role == &child_backing && child->bs->backing_overridden) { + bs->backing_overridden = true; + } }if (drv->bdrv_refresh_filename) {
block/mirror.c +4 -0
@@ -530,6 +530,10 @@ static void mirror_exit(BlockJob *job, void *opaque)error_report_err (local_err); data->ret = -EPERM ; } + +/* The target image's file already has been created with the backing +* file we just set, so there is no need to set backing_overridden or +* call bdrv_refresh_filename(). */ }if (s->to_replace) {
blockdev.c +16 -0
@@ -1795,6 +1795,8 @@ static void external_snapshot_commit(BlkActionState *common) {ExternalSnapshotState *state =DO_UPCAST (ExternalSnapshotState, common, common); +TransactionAction *action = common->action; +bool image_was_existing = false;AioContext *aio_context; aio_context =bdrv_get_aio_context (state->old_bs);
@@ -1808,6 +1810,20 @@ static void external_snapshot_commit(BlkActionState *common) NULL); } +if (action->type ==TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_SYNC ) { +BlockdevSnapshotSync *s = action->u.blockdev_snapshot_sync.data; +if (s->has_mode && s->mode ==NEW_IMAGE_MODE_EXISTING ) { + image_was_existing = true; + } + }else { + image_was_existing = true; + } + +if (image_was_existing) { + state->new_bs->backing_overridden = true; +bdrv_refresh_filename (state->new_bs); + } +aio_context_release (aio_context); }
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like qmp_change_backing_file().
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of bs->options and bs->backing_file and see if they match.
Quoted 7 lines
This commit adds a FIXME which will be remedied by a follow-up commit. Until then, the respective piece of code will not result in any behavior that is worse than what we currently have.
Signed-off-by Max Reitz <[email protected]>Reviewed-by Eric Blake <[email protected]>Reviewed-by Alberto Garcia <[email protected]>
Kevin
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Quoted 17 lines · Kevin Wolf
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Quoted 3 lines
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like
qmp_change_backing_file().
With you so far.
Quoted 4 lines
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of
bs->optionsandbs->backing_fileand see if they match.
I don't see how that is simple. First, bs->options does not necessarily reflect the "current options", those would be bs->full_open_options. And for generating that, we need a way to determine whether the backing file has been overridden or not, so whether we need to put the backing options into it or whether we do not.
(I am right that bs->backing_file is what the image header says, right? So we need to compare it against something that reflects the runtime state.)
What I could see would be comparing bs->backing_file to bs->backing->bs->filename. But this sounds very hacky to me.
One thing the comes to mind is that it can break whenever bdrv_refresh_filename() is clever. So you specify 'json:{"driver":"null-co"}' in the image header, and bdrv_refresh_filename() optimizes that to "null-co://". Now the filenames differ even though it's still the original filename. So this wouldn't work very well either.
Max
Quoted 10 lines
Quoted 7 lines
This commit adds a FIXME which will be remedied by a follow-up commit. Until then, the respective piece of code will not result in any behavior that is worse than what we currently have.
Signed-off-by Max Reitz <[email protected]>Reviewed-by Eric Blake <[email protected]>Reviewed-by Alberto Garcia <[email protected]>Kevin
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Am 22.02.2018 um 15:55 hat Max Reitz geschrieben:
Quoted 29 lines · Kevin Wolf
Quoted 17 lines
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Quoted 3 lines
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like
qmp_change_backing_file().With you so far.
Quoted 4 lines
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of
bs->optionsandbs->backing_fileand see if they match.
Quoted 5 lines
I don't see how that is simple. First,
bs->optionsdoes not necessarily reflect the "current options", those would bebs->full_open_options. And for generating that, we need a way to determine whether the backing file has been overridden or not, so whether we need to put the backing options into it or whether we do not.
For the purpose of this comparison, we need a set of options that contains the backing file options unconditionally.
Quoted 2 lines
(I am right that
bs->backing_fileis what the image header says, right? So we need to compare it against something that reflects the runtime state.)
I think so, yes.
Quoted 9 lines
What I could see would be comparing
bs->backing_filetobs->backing->bs->filename. But this sounds very hacky to me.One thing the comes to mind is that it can break whenever
bdrv_refresh_filename()is clever. So you specify 'json:{"driver":"null-co"}' in the image header, andbdrv_refresh_filename()optimizes that to "null-co://". Now the filenames differ even though it's still the original filename. So this wouldn't work very well either.
On the other hand, the problem with your current approach is that it results in a JSON filename even if you override the backing file and specify the same file name as we already have in the image header.
In the future, libvirt is going to manually build the graph, so we will always have the backing file overridden according to the logic in this patch. I don't think we want to get JSON filenames for all libvirt managed VMs, so can we realistically do without any kind of comparison?
Kevin
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Quoted 54 lines · Kevin Wolf
Am 22.02.2018 um 15:55 hat Max Reitz geschrieben:
Quoted 29 lines · Kevin Wolf
Quoted 17 lines
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Quoted 3 lines
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like
qmp_change_backing_file().With you so far.
Quoted 4 lines
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of
bs->optionsandbs->backing_fileand see if they match.Quoted 5 lines
I don't see how that is simple. First,
bs->optionsdoes not necessarily reflect the "current options", those would bebs->full_open_options. And for generating that, we need a way to determine whether the backing file has been overridden or not, so whether we need to put the backing options into it or whether we do not.For the purpose of this comparison, we need a set of options that contains the backing file options unconditionally.
Quoted 2 lines
(I am right that
bs->backing_fileis what the image header says, right? So we need to compare it against something that reflects the runtime state.)I think so, yes.
Quoted 9 lines
What I could see would be comparing
bs->backing_filetobs->backing->bs->filename. But this sounds very hacky to me.One thing the comes to mind is that it can break whenever
bdrv_refresh_filename()is clever. So you specify 'json:{"driver":"null-co"}' in the image header, andbdrv_refresh_filename()optimizes that to "null-co://". Now the filenames differ even though it's still the original filename. So this wouldn't work very well either.
Quoted 3 lines
On the other hand, the problem with your current approach is that it results in a JSON filename even if you override the backing file and specify the same file name as we already have in the image header.
Yes.
Quoted 4 lines
In the future, libvirt is going to manually build the graph, so we will always have the backing file overridden according to the logic in this patch. I don't think we want to get JSON filenames for all libvirt managed VMs, so can we realistically do without any kind of comparison?
libvirt doesn't need to query the filename, though, does it?
In my mind, we wanted to phase out filenames and basically only present them as convenience/legacy information to users who use qemu directly.
I really don't see the point of burdening qemu with simplifying and niceifying filenames when you want to use node names for everything anyway.
Max
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Am 22.02.2018 um 16:17 hat Max Reitz geschrieben:
Quoted 48 lines · Kevin Wolf
Quoted 47 lines
Am 22.02.2018 um 15:55 hat Max Reitz geschrieben:
Quoted 29 lines · Kevin Wolf
Quoted 17 lines
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Quoted 3 lines
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like
qmp_change_backing_file().With you so far.
Quoted 4 lines
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of
bs->optionsandbs->backing_fileand see if they match.Quoted 5 lines
I don't see how that is simple. First,
bs->optionsdoes not necessarily reflect the "current options", those would bebs->full_open_options. And for generating that, we need a way to determine whether the backing file has been overridden or not, so whether we need to put the backing options into it or whether we do not.For the purpose of this comparison, we need a set of options that contains the backing file options unconditionally.
Quoted 2 lines
(I am right that
bs->backing_fileis what the image header says, right? So we need to compare it against something that reflects the runtime state.)I think so, yes.
Quoted 3 lines
What I could see would be comparing
bs->backing_filetobs->backing->bs->filename. But this sounds very hacky to me.
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
One thing the comes to mind is that it can break whenever
bdrv_refresh_filename()is clever. So you specify 'json:{"driver":"null-co"}' in the image header, andbdrv_refresh_filename()optimizes that to "null-co://". Now the filenames differ even though it's still the original filename. So this wouldn't work very well either.
So what's the full effect here?
You example says that if you use an overcomplicated way to specify an image (by using json: instead of an URL), you get back an overcomplicated filename for the parent image (which includes the backing file even though it's not really necessary). Sounds fair enough to me.
Can bad things happen with absolute vs. relative paths?
Quoted 4 lines
Quoted 3 lines
On the other hand, the problem with your current approach is that it results in a JSON filename even if you override the backing file and specify the same file name as we already have in the image header.
Quoted 8 lines
Yes.
Quoted 4 lines
In the future, libvirt is going to manually build the graph, so we will always have the backing file overridden according to the logic in this patch. I don't think we want to get JSON filenames for all libvirt managed VMs, so can we realistically do without any kind of comparison?
libvirt doesn't need to query the filename, though, does it?
I know that libvirt uses the output in qemu-img info. And I learnt about that because they were surprised that json: filenames you get there can't necessarily be fed to QMP (because they contain only strings).
Other than that, I hope they don't. I suppose the filename can end up in error messages in logfiles, though.
Quoted 6 lines
In my mind, we wanted to phase out filenames and basically only present them as convenience/legacy information to users who use qemu directly.
I really don't see the point of burdening qemu with simplifying and niceifying filenames when you want to use node names for everything anyway.
But if you essentially say "filenames are only for those who don't use advanced features", then why bother with overridden backing files?
There are two problems I have with this patch: The first is that it introduces additional state that needs to be managed correctly in all future patches that modify the graph, and the second (and worse one) is that it fails to manage this state correctly even now.
I mentioned snapshots and change-backing-file that can result in a wrong bs->backing_overridden, and those were only the obvious first places I had a look at. Even if you fix them, I wouldn't trust my own review to find all relevant places. And that's a really bad sign for a design.
This is the most important reason why I'm looking for some method to derive the flag from already existing state.
Kevin
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
Quoted 56 lines · Kevin Wolf
Am 22.02.2018 um 16:17 hat Max Reitz geschrieben:
Quoted 48 lines · Kevin Wolf
Quoted 47 lines
Am 22.02.2018 um 15:55 hat Max Reitz geschrieben:
Quoted 29 lines · Kevin Wolf
Quoted 17 lines
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
If the backing file is overridden, this most probably does change the guest-visible data of a BDS. Therefore, we will need to consider this in
bdrv_refresh_filename().Adding a new field to the BDS is not nice, but it is very simple and exactly keeps track of whether the backing file has been overridden.
...as long as we manage to actually keep it up to date all the time.
First of all, what I'm missing here (or in fact in the comment in the code) is a definition what "overridden" really means. "specified by the user" is kind of vague: You consider the backing file relationship for snapshot=on as user specified, even though the user wasn't explicit about this. On the other hand, creating a live snapshot results in a node that isn't user specified.
Quoted 3 lines
Isn't the real question to ask whether the default backing file (taken from the image header) would result in the same tree? The answer to this changes after more operations, like
qmp_change_backing_file().With you so far.
Quoted 4 lines
Considering that there are so many ways to change the answer, I think the simplest reliable option isn't a new BDS field that needs to updated everywhere, but looking at the current value of
bs->optionsandbs->backing_fileand see if they match.Quoted 5 lines
I don't see how that is simple. First,
bs->optionsdoes not necessarily reflect the "current options", those would bebs->full_open_options. And for generating that, we need a way to determine whether the backing file has been overridden or not, so whether we need to put the backing options into it or whether we do not.For the purpose of this comparison, we need a set of options that contains the backing file options unconditionally.
Quoted 2 lines
(I am right that
bs->backing_fileis what the image header says, right? So we need to compare it against something that reflects the runtime state.)I think so, yes.
Quoted 3 lines
What I could see would be comparing
bs->backing_filetobs->backing->bs->filename. But this sounds very hacky to me.Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
One thing the comes to mind is that it can break whenever
bdrv_refresh_filename()is clever. So you specify 'json:{"driver":"null-co"}' in the image header, andbdrv_refresh_filename()optimizes that to "null-co://". Now the filenames differ even though it's still the original filename. So this wouldn't work very well either.
Quoted 7 lines
So what's the full effect here?
You example says that if you use an overcomplicated way to specify an image (by using json: instead of an URL), you get back an overcomplicated filename for the parent image (which includes the backing file even though it's not really necessary). Sounds fair enough to me.
OK, but one issue is that you've used an overcomplicated way for the backing file; but you get an overcomplicated filename for the overlay.
Quoted 1 line
Can bad things happen with absolute vs. relative paths?
"Can"? Absolutely. "Do"? I don't know? :-)
Another thing are non-unique URLs. For instance, nbd allows you to specify "nbd:localhost:10809", but it will generate "nbd://localhost:10809". (Same for just "nbd://localhost".)
Or of course the good old "file:foo.qcow2".
Quoted 11 lines
Quoted 11 lines
Quoted 3 lines
On the other hand, the problem with your current approach is that it results in a JSON filename even if you override the backing file and specify the same file name as we already have in the image header.
Yes.
Quoted 4 lines
In the future, libvirt is going to manually build the graph, so we will always have the backing file overridden according to the logic in this patch. I don't think we want to get JSON filenames for all libvirt managed VMs, so can we realistically do without any kind of comparison?
Quoted 8 lines
Quoted 1 line
libvirt doesn't need to query the filename, though, does it?
I know that libvirt uses the output in qemu-img info. And I learnt about that because they were surprised that json: filenames you get there can't necessarily be fed to QMP (because they contain only strings).
Other than that, I hope they don't. I suppose the filename can end up in error messages in logfiles, though.
Fair point. Although it isn't impossible to decrypt json:{} filenames, it isn't very nice.
(And I suspect most error messages today actually contain the node name, which is even less useful than json:{} filenames.)
Quoted 9 lines
Quoted 6 lines
In my mind, we wanted to phase out filenames and basically only present them as convenience/legacy information to users who use qemu directly.
I really don't see the point of burdening qemu with simplifying and niceifying filenames when you want to use node names for everything anyway.
But if you essentially say "filenames are only for those who don't use advanced features", then why bother with overridden backing files?
Because giving overcomplicated information is better than giving plainly wrong information.
If we return a filename, it has to be correct. Sure, we could just not return filenames, but even then we would need to have a way of recognizing when to do that.
But given backwards compatibility and all, I can't imagine a way where we don't have to deal with the backing file issue in one way or another.
If you want to remove filenames from the internal state and for what queries return (apart from bs->options for protocol nodes) in 3.0, sign me up for it.
Quoted 4 lines
There are two problems I have with this patch: The first is that it introduces additional state that needs to be managed correctly in all future patches that modify the graph, and the second (and worse one) is that it fails to manage this state correctly even now.
Well, agreed. I know there were a couple revisions already where I fixed things.
But then again, I also seem to remember that I've had a discussion about what to do here with you before, but it's been so long that I can't remember the actual content.
Quoted 7 lines
I mentioned snapshots and change-backing-file that can result in a wrong
bs->backing_overridden, and those were only the obvious first places I had a look at. Even if you fix them, I wouldn't trust my own review to find all relevant places. And that's a really bad sign for a design.This is the most important reason why I'm looking for some method to derive the flag from already existing state.
The only things I can say to that are "I agree" and "I've tried before".
The thing is, I started working on this series in November 2015
(apparently), so I've completely forgotten what exactly it was that I tried and I only know that my result was "I'll have to add this field".
I concede that result may be wrong, and I myself sure hope so.
So thee two things I see now are:
(1) Compare bs->backing_file against bs->backing->bs->filename. Seems basically as error-prone as backing_overridden to me, probably not only because bdrv_refresh_filename() may change .filename.
(2) Keep backing_overridden; but we'd need a central point for changing bs->backing, i.e. two functions or one function with a boolean parameter. That parameter tells it whether we want to change the current backing file, independently of the file's metadata, or whether we want to change the backing file in accordance with the file's metadata. In the former case, we'd set backing_overridden, in the latter, we'd clear it. This function would then be the only one allowed to modifiy bs->backing. The issue here is that maybe it's still not as simple as it sounds, and it is still possible to have backing_overridden set even though the backing node is indeed what it would be without the @backing option specified -- but I personally don't see a simple way to support this.
(We could strcmp bs->backing_file and bs->backing->bs->filename and if they match, we clear backing_overridden even though it is supposed to be set. But that would mean inconsistent behavior (sometimes you're lucky, sometimes you aren't), and I don't like that very much. I actually prefer consistently undesirable behavior over inconsistently desirable behavior.)
Max
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
So, while tackling this once more, this is the current state, in short.
I wanted to add a new field BDS.backing_file_canonical which basically gets set to bs->backing->bs->filename whenever we know that bs->backing was opened through bs->backing_file. That is, bs->backing_file_canonical would be bs->backing_file, but after going through the whole hoop of bdrv_open() and bdrv_refresh_filename().
With such a field, we could compare bs->backing->bs->filename against bs->backing_file_canonical, and thus see whether the backing BDS matches the image header filename. I hope this would do the right thing whenever qemu tries to figure out backing filenames on its own; it will break, though, when the user just uses some filename and that doesn't happen to coincide with qemu's generated filenames.
(For instance, if you create a new overlay manually, give it some backing filename, and that backing filename isn't 1:1 what qemu would reconstruct; and then you open that file with backing=null, and do a blockdev-snapshot; then qemu will give you a json:{} filename for the overlay because due to backing=null, it didn't open the overlay's backing_file and thus doesn't know how it would look reconstructed. But I think that's something we could live with.)
((The only real way to fix this I can imagine is to clone the whole bdrv_parse_filename()/bdrv_open()/bdrv_refresh_filename() infrastructure, but just for filenames instead of real BDSs. Yeah, no, I won't do that.))
OK, so that was an idea, but now it turns out that bs->backing_file actually isn't that image header's idea of the backing file. It appears to mostly be a clone of bs->backing->bs->filename, but not really. Honestly, I have no idea what it is, but as a matter of fact, it is modified by bdrv_backing_attach(), so it will probably usually match bs->backing->bs->filename (unless there is no bs->backing, in which case it just retains its old value?). So it pretty much is completely useless for any comparison here.
So I suppose I'll rename my BDS.backing_file_canonical attempt to BDS.auto_backing_file, and this will be what the image header contains (unless we have opened a BDS from it, in which case it will be that BDS's filename, so it is canonicalized).
Well, or I could just go with backing_overridden, because honestly that didn't seem so bad.
Max
PS: Or, we could argue that nobody needs filenames anyway and that they're just for show and debugging nowadays, so nobody actually needs backing chain information in them. May sound a bit stupid, but then again nobody has ever complained that that's in fact the current state.
Re: [Qemu-devel] [PATCH v8 03/26] block: Add BDS.backing_overridden
On 2018-06-26 19:34, Max Reitz wrote:
[...]
Quoted 4 lines
So I suppose I'll rename my BDS.backing_file_canonical attempt to BDS.auto_backing_file, and this will be what the image header contains (unless we have opened a BDS from it, in which case it will be that BDS's filename, so it is canonicalized).
Update: That seems to go down the drain. We (stupidly) allow taking a backing filename from the image header and then enriching it with options from the command line (so you could take null-co:// from the image header and then the user could give backing.size. I won't comment on how little sense that makes (apart from [1]), but in any case, this means that "we have opened a BDS from it" is very hard to reliably recognize, because you need to find out whether the user has given any strong options for the backing file. That pretty much brings the complexity back to backing_overridden, I'd say.
Well, or: Whenever the user has given any backing options, the resulting BDS is categorized as overridden and the filename is not used as a canonicalized version of auto_backing_file.
I now pretty much fail to see how any of this is better than backing_overridden, but whatever.
Max
[1] In fact, it's completely broken, because the user-given backing options only override the backing filename once the user-given options contain "file.filename". So you actually cannot really override anything unless your new backing file has a protocol driver with a "filename" option. Well, you can, but that means creating an own BDS and then giving a reference to that...
PATCH v8 4/26 iotests: Drop explicit base blockdev in 191
In reply to Max ReitzOverriding the backing image should result in a json:{} pseudo-filename. Then, you can no longer use the commit block job with filename parameters. Therefore, do not explicitly add the base and override the middle image in iotest 191, since we do not need to anyway. This will allow us to continue to use the middle image's filename to identify it.
In the long run, we want block-commit to accept node names for base and top (just like block-stream does).
--- tests/qemu-iotests/191 | 3 +-- tests/qemu-iotests/191.out | 52 +++++++++++++++++++++++----------------------- 2 files changed, 27 insertions(+), 28 deletions(-)
tests/qemu-iotests/191 +1 -2
@@ -66,8 +66,7 @@ qemu_comm_method="qmp" qmp_pretty="y" _launch_qemu \ - -blockdev "driver=${IMGFMT},file.driver=file,file.filename=${TEST_IMG}.base,node-name=base" \ - -blockdev "driver=${IMGFMT},file.driver=file,file.filename=${TEST_IMG}.mid,node-name=mid,backing=base" \ + -blockdev "driver=${IMGFMT},file.driver=file,file.filename=${TEST_IMG}.mid,node-name=mid" \ -blockdev "driver=${IMGFMT},file.driver=file,file.filename=${TEST_IMG},node-name=top,backing=mid" \ -blockdev "driver=${IMGFMT},file.driver=file,file.filename=${TEST_IMG}.ovl2,node-name=top2,backing=mid" h=$QEMU_HANDLE
tests/qemu-iotests/191.out +26 -26
128 lines not on this page · raw message
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 04/26] iotests: Drop explicit base blockdev in 191
Quoted 10 lines · Max Reitz
Overriding the backing image should result in a json:{} pseudo-filename. Then, you can no longer use the commit block job with filename parameters. Therefore, do not explicitly add the base and override the middle image in iotest 191, since we do not need to anyway. This will allow us to continue to use the middle image's filename to identify it.
In the long run, we want block-commit to accept node names for base and top (just like block-stream does).
Signed-off-by Max Reitz <[email protected]>
Berto
Re: [Qemu-devel] [PATCH v8 04/26] iotests: Drop explicit base blockdev in 191
In reply to Max ReitzAm 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 10 lines
Overriding the backing image should result in a json:{} pseudo-filename. Then, you can no longer use the commit block job with filename parameters. Therefore, do not explicitly add the base and override the middle image in iotest 191, since we do not need to anyway. This will allow us to continue to use the middle image's filename to identify it.
In the long run, we want block-commit to accept node names for base and top (just like block-stream does).
Signed-off-by Max Reitz <[email protected]>
Is this actually needed once the FIXME in patch 3 is addressed and only significant fields are considered, or can we revert the patch at the end of the series?
Once we implement a node-named based blockdev-commit, we'll want to test both versions here, so it would be good to be able to set a node-name and still be able to use the filename for the older interface.
Kevin
Re: [Qemu-devel] [PATCH v8 04/26] iotests: Drop explicit base blockdev in 191
Quoted 7 lines · Kevin Wolf
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 6 lines
Overriding the backing image should result in a json:{} pseudo-filename. Then, you can no longer use the commit block job with filename parameters. Therefore, do not explicitly add the base and override the middle image in iotest 191, since we do not need to anyway. This will allow us to continue to use the middle image's filename to identify it.
Quoted 8 lines
Quoted 4 lines
In the long run, we want block-commit to accept node names for base and top (just like block-stream does).
Signed-off-by Max Reitz <[email protected]>Is this actually needed once the FIXME in patch 3 is addressed and only significant fields are considered, or can we revert the patch at the end of the series?
The empiric result is "no". :-)
In the series's current state, overriding the backing file always means that it is being treated as overridden -- even if you override it with itself and thus the resulting graph is basically the same.
I'm still open to suggestions whether we can do anything about this, but the only thing that comes to my mind is:
(1) Open bs->backing_file as a BDS.
(2) Run bdrv_refresh_filename() on it (done automatically).
(3) Compare the result with bs->backing->bs->filename.
But you can't just open bs->backing_file twice (that is, have to BDS that refer to the same file), except maybe with BDRV_O_NO_IO. Still, that would be a huge hack. (And isn't guaranteed to work, even with BDRV_O_NO_IO.)
So what we'd need is an off-line bdrv_refresh_filename() that constructs a filename from filename+options, without a BDS. But honestly, I don't want to write that just so that we can continue to do blockdev-commit with filenames. I'd rather break it so we get node names there.
Max
Quoted 6 lines
Once we implement a node-named based blockdev-commit, we'll want to test both versions here, so it would be good to be able to set a node-name and still be able to use the filename for the older interface.
Kevin
PATCH v8 5/26 block: Respect backing bs in bdrv_refresh_filename
In reply to Max ReitzBasically, bdrv_refresh_filename() should respect all children of a BlockDriverState. However, generally those children are driver-specific, so this function cannot handle the general case. On the other hand, there are only few drivers which use other children than @file and @backing (that being vmdk, quorum, and blkverify).
Most block drivers only use @file and/or @backing (if they use any children at all). Both can be implemented directly in bdrv_refresh_filename.
The user overriding the file's filename is already handled, however, the user overriding the backing file is not. If this is done, opening the BDS with the plain filename of its file will not be correct, so we may not set bs->exact_filename in that case.
iotests 051 and 191 contain test cases for overwriting the backing file, and so their output changes with this patch applied (which I consider a good thing). Note that in the case of 191, the implicitly opened (non-overridden) base file is included in the json:{} filename as well -- this will be remedied by a later patch.
--- block.c | 12 +++++++++++- tests/qemu-iotests/051.out | 8 ++++---- tests/qemu-iotests/051.pc.out | 8 ++++---- tests/qemu-iotests/191.out | 24 ++++++++++++------------ 4 files changed, 31 insertions(+), 21 deletions(-)
block.c +11 -1
@@ -5071,6 +5071,7 @@ void bdrv_refresh_filename(BlockDriverState *bs) opts =qdict_new (); has_open_options =append_open_options (opts, bs); + has_open_options |= bs->backing_overridden;/* If no specific options have been given for this BDS, the filename of * the underlying file should suffice for this one as well */
@@ -5082,11 +5083,20 @@ void bdrv_refresh_filename(BlockDriverState *bs)* file BDS. The full options QDict of that file BDS should somehow * contain a representation of the filename, therefore the following * suffices without querying the (exact_)filename of this BDS. */ -if (bs->file->bs->full_open_options) { +if (bs->file->bs->full_open_options && + (!bs->backing || bs->backing->bs->full_open_options)) + {qdict_put_str (opts,"driver" , drv->format_name);QINCREF (bs->file->bs->full_open_options);qdict_put (opts,"file" , bs->file->bs->full_open_options); +if (bs->backing) { +QINCREF (bs->backing->bs->full_open_options); +qdict_put (opts,"backing" , bs->backing->bs->full_open_options); + }else if (bs->backing_overridden) { +qdict_put (opts,"backing" ,qstring_new ()); + } + bs->full_open_options = opts; }else {QDECREF (opts);
tests/qemu-iotests/051.out +4 -4
@@ -59,7 +59,7 @@ QEMU X.Y.Z monitor - type 'help' for more information Testing: -drive file=TEST_DIR/t.qcow2,driver=qcow2,backing.file.filename=TEST_DIR/t.qcow2.orig,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.orig"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback Backing file: TEST_DIR/t.qcow2.orig (chain depth: 1)
@@ -149,7 +149,7 @@ QEMU_PROG: -drive driver=null-co,cache=invalid_value: invalid cache option Testing: -drive file=TEST_DIR/t.qcow2,cache=writeback,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
@@ -169,7 +169,7 @@ backing-file: TEST_DIR/t.qcow2.base (file, read-only) Testing: -drive file=TEST_DIR/t.qcow2,cache=writethrough,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writethrough Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
@@ -189,7 +189,7 @@ backing-file: TEST_DIR/t.qcow2.base (file, read-only) Testing: -drive file=TEST_DIR/t.qcow2,cache=unsafe,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback, ignore flushes Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
tests/qemu-iotests/051.pc.out +4 -4
@@ -59,7 +59,7 @@ QEMU X.Y.Z monitor - type 'help' for more information Testing: -drive file=TEST_DIR/t.qcow2,driver=qcow2,backing.file.filename=TEST_DIR/t.qcow2.orig,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.orig"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback Backing file: TEST_DIR/t.qcow2.orig (chain depth: 1)
@@ -241,7 +241,7 @@ QEMU_PROG: -drive driver=null-co,cache=invalid_value: invalid cache option Testing: -drive file=TEST_DIR/t.qcow2,cache=writeback,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
@@ -261,7 +261,7 @@ backing-file: TEST_DIR/t.qcow2.base (file, read-only) Testing: -drive file=TEST_DIR/t.qcow2,cache=writethrough,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writethrough Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
@@ -281,7 +281,7 @@ backing-file: TEST_DIR/t.qcow2.base (file, read-only) Testing: -drive file=TEST_DIR/t.qcow2,cache=unsafe,backing.file.filename=TEST_DIR/t.qcow2.base,backing.cache.no-flush=on,backing.node-name=backing,backing.file.node-name=backing-file,file.node-name=file,if=none,id=drive0 -nodefaults QEMU X.Y.Z monitor - type 'help' for more information (qemu) info block -drive0 (NODE_NAME): TEST_DIR/t.qcow2 (qcow2) +drive0 (NODE_NAME): json:{"backing": {"driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2.base"}}, "driver": "qcow2", "file": {"driver": "file", "filename": "TEST_DIR/t.qcow2"}} (qcow2) Removable device: not locked, tray closed Cache mode: writeback, ignore flushes Backing file: TEST_DIR/t.qcow2.base (chain depth: 1)
tests/qemu-iotests/191.out +12 -12
@@ -52,7 +52,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "TEST_DIR/t.IMGFMT.ovl2", + "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -77,7 +77,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "TEST_DIR/t.IMGFMT.ovl2", + "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encryption_key_missing": false }, {
@@ -123,7 +123,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "TEST_DIR/t.IMGFMT", + "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -148,7 +148,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "TEST_DIR/t.IMGFMT", + "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", "encryption_key_missing": false }, {
@@ -392,7 +392,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "TEST_DIR/t.IMGFMT.ovl2", + "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -417,7 +417,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "TEST_DIR/t.IMGFMT.ovl2", + "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encryption_key_missing": false }, {
@@ -464,7 +464,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "TEST_DIR/t.IMGFMT.ovl2", + "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -474,12 +474,12 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "TEST_DIR/t.IMGFMT.ovl3", + "filename": "json:{"backing": {"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE, - "full-backing-filename": "TEST_DIR/t.IMGFMT.ovl2", - "backing-filename": "TEST_DIR/t.IMGFMT.ovl2", + "full-backing-filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "backing-filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "dirty-flag": false }, "iops_wr": 0,
@@ -490,7 +490,7 @@ wrote 65536/65536 bytes at offset 1048576 "iops": 0, "bps_wr": 0, "write_threshold": 0, - "backing_file": "TEST_DIR/t.IMGFMT.ovl2", + "backing_file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encrypted": false, "bps": 0, "bps_rd": 0,
@@ -499,7 +499,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "TEST_DIR/t.IMGFMT.ovl3", + "file": "json:{"backing": {"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", "encryption_key_missing": false }, {
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 05/26] block: Respect backing bs in bdrv_refresh_filename
Quoted 15 lines · Max Reitz
Basically,
bdrv_refresh_filename()should respect all children of a BlockDriverState. However, generally those children are driver-specific, so this function cannot handle the general case. On the other hand, there are only few drivers which use other children than @file and @backing (that being vmdk, quorum, and blkverify).Most block drivers only use @file and/or @backing (if they use any children at all). Both can be implemented directly in
bdrv_refresh_filename.The user overriding the file's filename is already handled, however, the user overriding the backing file is not. If this is done, opening the BDS with the plain filename of its file will not be correct, so we may not set
bs->exact_filenamein that case.
Quoted 7 lines
iotests 051 and 191 contain test cases for overwriting the backing file, and so their output changes with this patch applied (which I consider a good thing). Note that in the case of 191, the implicitly opened (non-overridden) base file is included in the json:{} filename as well -- this will be remedied by a later patch.
Signed-off-by Max Reitz <[email protected]>
Berto
PATCH v8 6/26 block: Make path_combine() return the path
In reply to Max ReitzBesides being safe for arbitrary path lengths, after some follow-up patches all callers will want a freshly allocated buffer anyway.
In the meantime, path_combine_deprecated() is added which has the same interface as path_combine() had before this patch. All callers to that function will be converted in follow-up patches.
--- include/block/block.h | 4 +-- block.c | 85 ++++++++++++++++++++++++++++----------------------- block/vmdk.c | 3 +- 3 files changed, 49 insertions(+), 43 deletions(-)
include/block/block.h +1 -3
@@ -490,9 +490,7 @@ void bdrv_get_full_backing_filename_from_filename(const char *backed,int path_has_protocol (const char *path);int path_is_absolute (const char *path); -void path_combine (char *dest,int dest_size, -const char *base_path, -const char *filename); +char *path_combine (const char *base_path,const char *filename);int bdrv_readv_vmstate (BlockDriverState *bs,QEMUIOVector *qiov,int64_t pos);int bdrv_writev_vmstate (BlockDriverState *bs,QEMUIOVector *qiov,int64_t pos);
block.c +47 -38
123 lines not on this page · raw message
block/vmdk.c +1 -2
@@ -857,8 +857,7 @@ static int vmdk_parse_extents(const char *desc, BlockDriverState *bs,return -EINVAL ; } - extent_path =g_malloc0 (PATH_MAX ); -path_combine (extent_path,PATH_MAX , desc_file_path, fname); + extent_path =path_combine (desc_file_path, fname); ret =snprintf (extent_opt_prefix,32 ,"extents.%d" , s->num_extents);assert (ret <32 );
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 06/26] block: Make path_combine() return the path
Am 05.02.2018 um 16:18 hat Max Reitz geschrieben:
Quoted 9 lines
Besides being safe for arbitrary path lengths, after some follow-up patches all callers will want a freshly allocated buffer anyway.
In the meantime,
path_combine_deprecated()is added which has the same interface aspath_combine()had before this patch. All callers to that function will be converted in follow-up patches.Signed-off-by Max Reitz <[email protected]>Reviewed-by Alberto Garcia <[email protected]>
PATCH v8 7/26 block: bdrv_get_full_backing_filename_from_...'s ret. val.
In reply to Max ReitzMake bdrv_get_full_backing_filename_from_filename() return an allocated string instead of placing the result in a caller-provided buffer.
--- include/block/block.h | 7 +++---- block.c | 51 +++++++++++++++++++++++++++++++++++---------------- block/vmdk.c | 9 ++++----- 3 files changed, 42 insertions(+), 25 deletions(-)
include/block/block.h +3 -4
@@ -483,10 +483,9 @@ void bdrv_get_backing_filename(BlockDriverState *bs, char *filename, int filename_size);void bdrv_get_full_backing_filename (BlockDriverState *bs,char *dest,size_t sz,Error **errp); -void bdrv_get_full_backing_filename_from_filename (const char *backed, -const char *backing, -char *dest,size_t sz, -Error **errp); +char *bdrv_get_full_backing_filename_from_filename (const char *backed, +const char *backing, +Error **errp);int path_has_protocol (const char *path);int path_is_absolute (const char *path);
block.c +35 -16
81 lines not on this page · raw message
block/vmdk.c +4 -5
@@ -2017,16 +2017,15 @@ static int vmdk_create(const char *filename, QemuOpts *opts, Error **errp) }if (backing_file) {BlockBackend *blk; -char *full_backing =g_new0 (char,PATH_MAX ); -bdrv_get_full_backing_filename_from_filename (filename, backing_file, - full_backing,PATH_MAX , - &local_err); +char *full_backing = +bdrv_get_full_backing_filename_from_filename (filename, backing_file, + &local_err);if (local_err) { -g_free (full_backing);error_propagate (errp, local_err); ret = -ENOENT ; goto exit; } +assert (full_backing); blk =blk_new_open (full_backing,NULL ,NULL ,BDRV_O_NO_BACKING , errp);
Signature · 1 line
2.14.3
PATCH v8 8/26 block: bdrv_get_full_backing_filename's ret. val.
In reply to Max ReitzMake bdrv_get_full_backing_filename() return an allocated string instead of placing the result in a caller-provided buffer.
--- include/block/block.h | 3 +-- block.c | 47 +++++++++++++++++------------------------------ block/qapi.c | 12 ++---------- 3 files changed, 20 insertions(+), 42 deletions(-)
include/block/block.h +1 -2
@@ -481,8 +481,7 @@ void bdrv_round_to_clusters(BlockDriverState *bs,const char *bdrv_get_encrypted_filename (BlockDriverState *bs);void bdrv_get_backing_filename (BlockDriverState *bs,char *filename,int filename_size); -void bdrv_get_full_backing_filename (BlockDriverState *bs, -char *dest,size_t sz,Error **errp); +char *bdrv_get_full_backing_filename (BlockDriverState *bs,Error **errp);char *bdrv_get_full_backing_filename_from_filename (const char *backed,const char *backing,Error **errp);
block.c +17 -30
100 lines not on this page · raw message
block/qapi.c +2 -10
@@ -282,18 +282,10 @@ void bdrv_query_image_info(BlockDriverState *bs, backing_filename = bs->backing_file;if (backing_filename[0 ] !='\0' ) { -char *backing_filename2 =g_malloc0 (PATH_MAX ); +char *backing_filename2; info->backing_filename =g_strdup (backing_filename); info->has_backing_filename = true; -bdrv_get_full_backing_filename (bs, backing_filename2,PATH_MAX , &err); -if (err) { -/* Can't reconstruct the full backing filename, so we must omit -* this field and apply a Best Effort to this query. */ -g_free (backing_filename2); - backing_filename2 =NULL ; -error_free (err); - err =NULL ; - } + backing_filename2 =bdrv_get_full_backing_filename (bs,NULL );/* Always report the full_backing_filename if present, even if it's the * same as backing_filename. That they are same is useful info. */
Signature · 1 line
2.14.3
PATCH v8 9/26 block: Add bdrv_make_absolute_filename()
In reply to Max ReitzThis is a general function for making a filename that is relative to a certain BDS absolute.
It calls bdrv_get_full_backing_filename_from_filename() for now, but that will be changed in a follow-up patch.
--- block.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-)
block.c +16 -5
@@ -313,13 +313,24 @@ char *bdrv_get_full_backing_filename_from_filename(const char *backed, } } -char *bdrv_get_full_backing_filename (BlockDriverState *bs,Error **errp) +/* If @filename is empty or NULL, this function returns NULL without +* setting @errp. In all other cases, NULL will only be returned with +* @errp set. +*/ +static char *bdrv_make_absolute_filename (BlockDriverState *relative_to, +const char *filename,Error **errp) { -char *backed = bs->exact_filename[0 ] ? bs->exact_filename : bs->filename; +char *bs_filename = relative_to->exact_filename[0 ] + ? relative_to->exact_filename + : relative_to->filename; + +return bdrv_get_full_backing_filename_from_filename (bs_filename, + filename ?:"" , errp); +} -return bdrv_get_full_backing_filename_from_filename (backed, - bs->backing_file, - errp); +char *bdrv_get_full_backing_filename (BlockDriverState *bs,Error **errp) +{ +return bdrv_make_absolute_filename (bs, bs->backing_file, errp); }void bdrv_register (BlockDriver *bdrv)
Signature · 1 line
2.14.3
PATCH v8 10/26 block: Fix bdrv_find_backing_image()
In reply to Max Reitzbdrv_find_backing_image() should use bdrv_get_full_backing_filename() or bdrv_make_absolute_filename() instead of trying to do what those functions do by itself.
path_combine_deprecated() can now be dropped, so let's do that.
--- block.c | 30 ++++++++++-------------------- 1 file changed, 10 insertions(+), 20 deletions(-)
block.c +10 -20
@@ -196,15 +196,6 @@ char *path_combine(const char *base_path, const char *filename)return result; } -static void path_combine_deprecated (char *dest,int dest_size, -const char *base_path, -const char *filename) -{ -char *combined =path_combine (base_path, filename); -pstrcpy (dest, dest_size, combined); -g_free (combined); -} -/* * Helper function for bdrv_parse_filename() implementations to remove optional * protocol prefixes (especially "file:") from a filename and for putting the
@@ -4179,7 +4170,6 @@ BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs, filename_full =g_malloc (PATH_MAX ); backing_file_full =g_malloc (PATH_MAX ); - filename_tmp =g_malloc (PATH_MAX ); is_protocol =path_has_protocol (backing_file);
@@ -4208,22 +4198,23 @@ BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs, }else {/* If not an absolute filename path, make it relative to the current * image's filename path */ -path_combine_deprecated (filename_tmp,PATH_MAX , curr_bs->filename, - backing_file); - -/* We are going to compare absolute pathnames */ -if (!realpath (filename_tmp, filename_full)) { + filename_tmp =bdrv_make_absolute_filename (curr_bs, backing_file, +NULL ); +/* We are going to compare canonicalized absolute pathnames */ +if (!filename_tmp || !realpath (filename_tmp, filename_full)) { +g_free (filename_tmp);continue ; } +g_free (filename_tmp);/* We need to make sure the backing filename we are comparing against * is relative to the current image filename (or absolute) */ -path_combine_deprecated (filename_tmp,PATH_MAX , curr_bs->filename, - curr_bs->backing_file); - -if (!realpath (filename_tmp, backing_file_full)) { + filename_tmp =bdrv_get_full_backing_filename (curr_bs,NULL ); +if (!filename_tmp || !realpath (filename_tmp, backing_file_full)) { +g_free (filename_tmp);continue ; } +g_free (filename_tmp);if (strcmp (backing_file_full, filename_full) ==0 ) { retval = curr_bs->backing->bs;
@@ -4234,7 +4225,6 @@ BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,g_free (filename_full);g_free (backing_file_full); -g_free (filename_tmp);return retval; }
Signature · 1 line
2.14.3
PATCH v8 11/26 block: Add bdrv_dirname()
In reply to Max ReitzThis function may be implemented by block drivers to derive a directory name from a BDS. Concatenating this g_free()-able string with a relative filename must result in a valid (not necessarily existing) filename, so this is a function that should generally be not implemented by format drivers, because this is protocol-specific.
If a BDS's driver does not implement this function, bdrv_dirname() will fall through to the BDS's file if it exists. If it does not, the exact_filename field will be used to generate a directory name.
--- include/block/block.h | 1 + include/block/block_int.h | 7 +++++++ block.c | 26 ++++++++++++++++++++++++++ 3 files changed, 34 insertions(+)
include/block/block.h +1 -0
@@ -485,6 +485,7 @@ char *bdrv_get_full_backing_filename(BlockDriverState *bs, Error **errp);char *bdrv_get_full_backing_filename_from_filename (const char *backed,const char *backing,Error **errp); +char *bdrv_dirname (BlockDriverState *bs,Error **errp);int path_has_protocol (const char *path);int path_is_absolute (const char *path);
include/block/block_int.h +7 -0
@@ -135,6 +135,13 @@ struct BlockDriver {void (*bdrv_refresh_filename)(BlockDriverState *bs,QDict *options); +/* +* Returns an allocated string which is the directory name of this BDS: It +* will be used to make relative filenames absolute by prepending this +* function's return value to them. +*/ +char *(*bdrv_dirname)(BlockDriverState *bs,Error **errp); +/* aio */ BlockAIOCB *(*bdrv_aio_readv)(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
block.c +26 -0
@@ -5153,6 +5153,32 @@ void bdrv_refresh_filename(BlockDriverState *bs) } } +char *bdrv_dirname (BlockDriverState *bs,Error **errp) +{ +BlockDriver *drv = bs->drv; + +if (!drv) { +error_setg (errp,"Node '%s' is ejected" , bs->node_name); +return NULL ; + } + +if (drv->bdrv_dirname) { +return drv->bdrv_dirname (bs, errp); + } + +if (bs->file) { +return bdrv_dirname (bs->file->bs, errp); + } + +if (bs->exact_filename[0 ] !='\0' ) { +return path_combine (bs->exact_filename,"" ); + } + +error_setg (errp,"Cannot generate a base directory for %s nodes" , + drv->format_name); +return NULL ; +} +/* * Hot add/remove a BDS's child. So the user can take a child offline when * it is broken and take a new child online
Signature · 1 line
2.14.3
PATCH v8 12/26 blkverify: Make bdrv_dirname() return NULL
In reply to Max Reitzblkverify's BDSs have a file BDS, but we do not want this to be preferred over the raw node. There is no way to decide between the two (and not really a reason to, either), so just return NULL in blkverify's implementation of bdrv_dirname().
--- block/blkverify.c | 10 ++++++++++ 1 file changed, 10 insertions(+)
block/blkverify.c +10 -0
@@ -309,6 +309,15 @@ static void blkverify_refresh_filename(BlockDriverState *bs, QDict *options) } } +static char *blkverify_dirname (BlockDriverState *bs,Error **errp) +{ +/* In general, there are two BDSs with different dirnames below this one; +* so there is no unique dirname we could return (unless both are equal by +* chance). Therefore, to be consistent, just always return NULL. */ +error_setg (errp,"Cannot generate a base directory for blkverify nodes" ); +return NULL ; +} +static BlockDriver bdrv_blkverify = { .format_name ="blkverify" , .protocol_name ="blkverify" ,
@@ -320,6 +329,7 @@ static BlockDriver bdrv_blkverify = { .bdrv_child_perm = bdrv_filter_default_perms, .bdrv_getlength = blkverify_getlength, .bdrv_refresh_filename = blkverify_refresh_filename, + .bdrv_dirname = blkverify_dirname, .bdrv_co_preadv = blkverify_co_preadv, .bdrv_co_pwritev = blkverify_co_pwritev,
Signature · 1 line
2.14.3
PATCH v8 13/26 quorum: Make bdrv_dirname() return NULL
In reply to Max ReitzWhile the common implementation for bdrv_dirname() should return NULL for quorum BDSs already (because they do not have a file node and their exact_filename field should be empty), there is no reason not to make that explicit.
--- block/quorum.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
block/quorum.c +11 -0
@@ -1095,6 +1095,16 @@ static void quorum_refresh_filename(BlockDriverState *bs, QDict *options) bs->full_open_options = opts; } +static char *quorum_dirname (BlockDriverState *bs,Error **errp) +{ +/* In general, there are multiple BDSs with different dirnames below this +* one; so there is no unique dirname we could return (unless all are equal +* by chance, or there is only one). Therefore, to be consistent, just +* always return NULL. */ +error_setg (errp,"Cannot generate a base directory for quorum nodes" ); +return NULL ; +} +static BlockDriver bdrv_quorum = { .format_name ="quorum" , .protocol_name ="quorum" ,
@@ -1104,6 +1114,7 @@ static BlockDriver bdrv_quorum = { .bdrv_file_open = quorum_open, .bdrv_close = quorum_close, .bdrv_refresh_filename = quorum_refresh_filename, + .bdrv_dirname = quorum_dirname, .bdrv_co_flush_to_disk = quorum_co_flush,
Signature · 1 line
2.14.3
PATCH v8 14/26 block/nbd: Make bdrv_dirname() return NULL
In reply to Max ReitzThe generic bdrv_dirname() implementation would be able to generate some form of directory name for many NBD nodes, but it would be always wrong. Therefore, we have to explicitly make it an error (until NBD has some form of specification for export paths, if it ever will).
--- block/nbd.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)
block/nbd.c +13 -0
@@ -574,6 +574,16 @@ static int nbd_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)return 0 ; } +static char *nbd_dirname (BlockDriverState *bs,Error **errp) +{ +/* The generic bdrv_dirname() implementation is able to work out some +* directory name for NBD nodes, but that would be wrong. So far there is no +* specification for how "export paths" would work, so NBD does not have +* directory names. */ +error_setg (errp,"Cannot generate a base directory for NBD nodes" ); +return NULL ; +} +static BlockDriver bdrv_nbd = { .format_name ="nbd" , .protocol_name ="nbd" ,
@@ -592,6 +602,7 @@ static BlockDriver bdrv_nbd = { .bdrv_attach_aio_context = nbd_attach_aio_context, .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, + .bdrv_dirname = nbd_dirname, };static BlockDriver bdrv_nbd_tcp = {
@@ -612,6 +623,7 @@ static BlockDriver bdrv_nbd_tcp = { .bdrv_attach_aio_context = nbd_attach_aio_context, .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, + .bdrv_dirname = nbd_dirname, };static BlockDriver bdrv_nbd_unix = {
@@ -632,6 +644,7 @@ static BlockDriver bdrv_nbd_unix = { .bdrv_attach_aio_context = nbd_attach_aio_context, .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, + .bdrv_dirname = nbd_dirname, };static void bdrv_nbd_init (void )
Signature · 1 line
2.14.3
PATCH v8 15/26 block/nfs: Implement bdrv_dirname()
In reply to Max ReitzWhile the basic idea is obvious and could be handled by the default bdrv_dirname() implementation, we cannot generate a directory name if the gid or uid are set, so we have to explicitly return NULL in those cases.
--- block/nfs.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
block/nfs.c +14 -0
@@ -874,6 +874,19 @@ static void nfs_refresh_filename(BlockDriverState *bs, QDict *options) bs->full_open_options = opts; } +static char *nfs_dirname (BlockDriverState *bs,Error **errp) +{ +NFSClient *client = bs->opaque; + +if (client->uid || client->gid) { +error_setg (errp,"Cannot generate a base directory for NBD node '%s'" , + bs->filename); +return NULL ; + } + +return g_strdup_printf ("nfs://%s%s/" , client->server->host, client->path); +} +#ifdef LIBNFS_FEATURE_PAGECACHE static void nfs_invalidate_cache (BlockDriverState *bs,Error **errp)
@@ -907,6 +920,7 @@ static BlockDriver bdrv_nfs = { .bdrv_detach_aio_context = nfs_detach_aio_context, .bdrv_attach_aio_context = nfs_attach_aio_context, .bdrv_refresh_filename = nfs_refresh_filename, + .bdrv_dirname = nfs_dirname, #ifdef LIBNFS_FEATURE_PAGECACHE .bdrv_invalidate_cache = nfs_invalidate_cache,
Signature · 1 line
2.14.3
PATCH v8 16/26 block: Use bdrv_dirname() for relative filenames
In reply to Max Reitzbdrv_get_full_backing_filename_from_filename() breaks down when it comes to JSON filenames. Using bdrv_dirname() as the basis is better because since we have BDS, we can descend through the BDS tree to the protocol layer, which gives us a greater probability of finding a non-JSON name; also, bdrv_dirname() is more correct as it allows block drivers to override the generation of that directory name in a protocol-specific way.
We still need to keep bdrv_get_full_backing_filename_from_filename(), though, because it has valid callers which need it during image creation when no BDS is available yet.
This makes a test case in qemu-iotest 110, which was supposed to fail, work. That is actually good, but we need to change the reference output (and the comment in 110) accordingly.
--- block.c | 20 +++++++++++++++----- tests/qemu-iotests/110 | 3 ++- tests/qemu-iotests/110.out | 2 +- 3 files changed, 18 insertions(+), 7 deletions(-)
block.c +15 -5
@@ -311,12 +311,22 @@ char *bdrv_get_full_backing_filename_from_filename(const char *backed,static char *bdrv_make_absolute_filename (BlockDriverState *relative_to,const char *filename,Error **errp) { -char *bs_filename = relative_to->exact_filename[0 ] - ? relative_to->exact_filename - : relative_to->filename; +char *dir, *full_name; -return bdrv_get_full_backing_filename_from_filename (bs_filename, - filename ?:"" , errp); +if (!filename || filename[0 ] =='\0' ) { +return NULL ; + }else if (path_has_protocol (filename) ||path_is_absolute (filename)) { +return g_strdup (filename); + } + + dir =bdrv_dirname (relative_to, errp); +if (!dir) { +return NULL ; + } + + full_name =g_strconcat (dir, filename,NULL ); +g_free (dir); +return full_name; }char *bdrv_get_full_backing_filename (BlockDriverState *bs,Error **errp)
tests/qemu-iotests/110 +2 -1
@@ -61,7 +61,8 @@ echo '=== Non-reconstructable filename ===' echo # Across blkdebug without a config file, you cannot reconstruct filenames, so -# qemu is incapable of knowing the directory of the top image +# qemu is incapable of knowing the directory of the top image from the filename +# alone. However, using bdrv_dirname(), it should still work. TEST_IMG="json:{ 'driver': '$IMGFMT', 'file': {
tests/qemu-iotests/110.out +1 -1
@@ -14,7 +14,7 @@ backing file: t.IMGFMT.base (actual path: TEST_DIR/t.IMGFMT.base) image: json:{"driver": "IMGFMT", "file": {"set-state.0.event": "read_aio", "image": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}, "driver": "blkdebug", "set-state.0.new_state": 42}} file format: IMGFMT virtual size: 64M (67108864 bytes) -backing file: t.IMGFMT.base (cannot determine actual path) +backing file: t.IMGFMT.base (actual path: TEST_DIR/t.IMGFMT.base) === Backing name is always relative to the backed image ===
Signature · 1 line
2.14.3
PATCH v8 17/26 iotests: Add quorum case to test 110
In reply to Max ReitzTest 110 tests relative backing filenames for complex BDS trees. Now that the originally supposedly failing test passes, let us add a new failing test: Quorum can never work automatically (without detecting whether all child nodes have the same base directory, but that would be rather inconsistent behavior).
--- tests/qemu-iotests/110 | 26 ++++++++++++++++++++++++++ tests/qemu-iotests/110.out | 7 +++++++ 2 files changed, 33 insertions(+)
tests/qemu-iotests/110 +26 -0
@@ -30,6 +30,7 @@ status=1 # failure is the default! _cleanup() { _cleanup_test_img + rm -f "$TEST_IMG.copy" } trap "_cleanup; exit \$status" 0 1 2 3 15
@@ -87,6 +88,31 @@ echo # omit the image size; it should work anyway _make_test_img -b "$TEST_IMG_REL.base" +echo +echo '=== Nodes without a common directory ===' +echo + +cp "$TEST_IMG" "$TEST_IMG.copy" + +# Should inform us that the actual path of the backing file cannot be determined +TEST_IMG="json:{ + 'driver': '$IMGFMT', + 'file': { + 'driver': 'quorum', + 'vote-threshold': 1, + 'children': [ + { + 'driver': 'file', + 'filename': '$TEST_IMG' + }, + { + 'driver': 'file', + 'filename': '$TEST_IMG.copy' + } + ] + } +}" _img_info | _filter_img_info + # success, all done echo '*** done'
tests/qemu-iotests/110.out +7 -0
@@ -19,4 +19,11 @@ backing file: t.IMGFMT.base (actual path: TEST_DIR/t.IMGFMT.base) === Backing name is always relative to the backed image === Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864 backing_file=t.IMGFMT.base + +=== Nodes without a common directory === + +image: json:{"driver": "IMGFMT", "file": {"children": [{"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}, {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.copy"}], "driver": "quorum", "blkverify": false, "rewrite-corrupted": false, "vote-threshold": 1}} +file format: IMGFMT +virtual size: 64M (67108864 bytes) +backing file: t.IMGFMT.base (cannot determine actual path) *** done
Signature · 1 line
2.14.3
PATCH v8 18/26 block: Add sgfnt_runtime_opts to BlockDriver
In reply to Max ReitzThis new field can be set by block drivers to list the runtime options they accept that may influence the contents of the respective BDS. As of a follow-up patch, this list will be used by the common bdrv_refresh_filename() implementation to decide which options to put into BDS.full_open_options (and consequently whether a JSON filename has to be created), thus freeing the drivers of having to implement that logic themselves.
Additionally, this patch adds the field to all of the block drivers that need it and sets it accordingly.
--- include/block/block_int.h | 7 +++++++ block/blkdebug.c | 17 +++++++++++++++++ block/crypto.c | 8 ++++++++ block/curl.c | 21 +++++++++++++++++++++ block/gluster.c | 19 +++++++++++++++++++ block/iscsi.c | 18 ++++++++++++++++++ block/nbd.c | 14 ++++++++++++++ block/nfs.c | 11 +++++++++++ block/null.c | 9 +++++++++ block/qcow.c | 7 +++++++ block/qcow2.c | 7 +++++++ block/quorum.c | 11 +++++++++++ block/raw-format.c | 10 +++++++++- block/rbd.c | 14 ++++++++++++++ block/replication.c | 8 ++++++++ block/sheepdog.c | 12 ++++++++++++ block/ssh.c | 12 ++++++++++++ block/throttle.c | 7 +++++++ block/vpc.c | 7 +++++++ block/vvfat.c | 12 ++++++++++++ block/vxhs.c | 11 +++++++++++ 21 files changed, 241 insertions(+), 1 deletion(-)
include/block/block_int.h +7 -0
@@ -454,6 +454,13 @@ struct BlockDriver { Error **errp);QLIST_ENTRY (BlockDriver) list; + +/* Pointer to a NULL-terminated array of names of significant options that +* can be specified for bdrv_open(). A significant option is one that +* changes the data of a BDS. +* If this pointer is NULL, the array is considered empty. +* "filename" and "driver" are always considered significant. */ +const char *const *sgfnt_runtime_opts; }; typedef struct BlockLimits {
block/blkdebug.c +17 -0
@@ -886,6 +886,21 @@ static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state,return 0 ; } +static const char *const blkdebug_sgfnt_runtime_opts[] = { +"config" , +"inject-error." , +"set-state." , +"suspend." , +"align" , +"max-transfer" , +"opt-write-zero" , +"max-write-zero" , +"opt-discard" , +"max-discard" , + +NULL +}; +static BlockDriver bdrv_blkdebug = { .format_name ="blkdebug" , .protocol_name ="blkdebug" ,
@@ -915,6 +930,8 @@ static BlockDriver bdrv_blkdebug = { = blkdebug_debug_remove_breakpoint, .bdrv_debug_resume = blkdebug_debug_resume, .bdrv_debug_is_suspended = blkdebug_debug_is_suspended, + + .sgfnt_runtime_opts = blkdebug_sgfnt_runtime_opts, };static void bdrv_blkdebug_init (void )
block/crypto.c +8 -0
@@ -609,6 +609,12 @@ block_crypto_get_specific_info_luks(BlockDriverState *bs)return spec_info; } +static const char *const block_crypto_sgfnt_runtime_opts[] = { +BLOCK_CRYPTO_OPT_LUKS_KEY_SECRET , + +NULL +}; +BlockDriver bdrv_crypto_luks = { .format_name ="luks" , .instance_size =sizeof (BlockCrypto),
@@ -626,6 +632,8 @@ BlockDriver bdrv_crypto_luks = { .bdrv_getlength = block_crypto_getlength, .bdrv_get_info = block_crypto_get_info_luks, .bdrv_get_specific_info = block_crypto_get_specific_info_luks, + + .sgfnt_runtime_opts = block_crypto_sgfnt_runtime_opts, };static void block_crypto_init (void )
block/curl.c +21 -0
@@ -969,6 +969,19 @@ static int64_t curl_getlength(BlockDriverState *bs)return s->len; } +static const char *const curl_sgfnt_runtime_opts[] = { +CURL_BLOCK_OPT_URL , +CURL_BLOCK_OPT_SSLVERIFY , +CURL_BLOCK_OPT_COOKIE , +CURL_BLOCK_OPT_COOKIE_SECRET , +CURL_BLOCK_OPT_USERNAME , +CURL_BLOCK_OPT_PASSWORD_SECRET , +CURL_BLOCK_OPT_PROXY_USERNAME , +CURL_BLOCK_OPT_PROXY_PASSWORD_SECRET , + +NULL +}; +static BlockDriver bdrv_http = { .format_name ="http" , .protocol_name ="http" ,
@@ -983,6 +996,8 @@ static BlockDriver bdrv_http = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + + .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };static BlockDriver bdrv_https = {
@@ -999,6 +1014,8 @@ static BlockDriver bdrv_https = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + + .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };static BlockDriver bdrv_ftp = {
@@ -1015,6 +1032,8 @@ static BlockDriver bdrv_ftp = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + + .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };static BlockDriver bdrv_ftps = {
@@ -1031,6 +1050,8 @@ static BlockDriver bdrv_ftps = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + + .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };static void curl_block_init (void )
block/gluster.c +19 -0
@@ -1414,6 +1414,21 @@ static int64_t coroutine_fn qemu_gluster_co_get_block_status( } +static const char *const gluster_sgfnt_open_opts[] = { +GLUSTER_OPT_VOLUME , +GLUSTER_OPT_PATH , +GLUSTER_OPT_TYPE , +GLUSTER_OPT_SERVER_PATTERN , +GLUSTER_OPT_HOST , +GLUSTER_OPT_PORT , +GLUSTER_OPT_TO , +GLUSTER_OPT_IPV4 , +GLUSTER_OPT_IPV6 , +GLUSTER_OPT_SOCKET , + +NULL +}; +static BlockDriver bdrv_gluster = { .format_name ="gluster" , .protocol_name ="gluster" ,
@@ -1440,6 +1455,7 @@ static BlockDriver bdrv_gluster = {#endif .bdrv_co_get_block_status = qemu_gluster_co_get_block_status, .create_opts = &qemu_gluster_create_opts, + .sgfnt_runtime_opts = gluster_sgfnt_open_opts, };static BlockDriver bdrv_gluster_tcp = {
@@ -1468,6 +1484,7 @@ static BlockDriver bdrv_gluster_tcp = {#endif .bdrv_co_get_block_status = qemu_gluster_co_get_block_status, .create_opts = &qemu_gluster_create_opts, + .sgfnt_runtime_opts = gluster_sgfnt_open_opts, };static BlockDriver bdrv_gluster_unix = {
@@ -1496,6 +1513,7 @@ static BlockDriver bdrv_gluster_unix = {#endif .bdrv_co_get_block_status = qemu_gluster_co_get_block_status, .create_opts = &qemu_gluster_create_opts, + .sgfnt_runtime_opts = gluster_sgfnt_open_opts, };/* rdma is deprecated (actually never supported for volfile fetch).
@@ -1530,6 +1548,7 @@ static BlockDriver bdrv_gluster_rdma = {#endif .bdrv_co_get_block_status = qemu_gluster_co_get_block_status, .create_opts = &qemu_gluster_create_opts, + .sgfnt_runtime_opts = gluster_sgfnt_open_opts, };static void bdrv_gluster_init (void )
block/iscsi.c +18 -0
@@ -2183,6 +2183,20 @@ static QemuOptsList iscsi_create_opts = { } }; +static const char *const iscsi_sgfnt_runtime_opts[] = { +"transport" , +"portal" , +"target" , +"user" , +"password" , +"password-secret" , +"lun" , +"initiator-name" , +"header-digest" , + +NULL +}; +static BlockDriver bdrv_iscsi = { .format_name ="iscsi" , .protocol_name ="iscsi" ,
@@ -2215,6 +2229,8 @@ static BlockDriver bdrv_iscsi = { .bdrv_detach_aio_context = iscsi_detach_aio_context, .bdrv_attach_aio_context = iscsi_attach_aio_context, + + .sgfnt_runtime_opts = iscsi_sgfnt_runtime_opts, };#if LIBISCSI_API_VERSION >= (20160603 )
@@ -2250,6 +2266,8 @@ static BlockDriver bdrv_iser = { .bdrv_detach_aio_context = iscsi_detach_aio_context, .bdrv_attach_aio_context = iscsi_attach_aio_context, + + .sgfnt_runtime_opts = iscsi_sgfnt_runtime_opts, };#endif
block/nbd.c +14 -0
@@ -584,6 +584,17 @@ static char *nbd_dirname(BlockDriverState *bs, Error **errp)return NULL ; } +static const char *const nbd_sgfnt_runtime_opts[] = { +"path" , +"host" , +"port" , +"export" , +"tls-creds" , +"server." , + +NULL +}; +static BlockDriver bdrv_nbd = { .format_name ="nbd" , .protocol_name ="nbd" ,
@@ -603,6 +614,7 @@ static BlockDriver bdrv_nbd = { .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, .bdrv_dirname = nbd_dirname, + .sgfnt_runtime_opts = nbd_sgfnt_runtime_opts, };static BlockDriver bdrv_nbd_tcp = {
@@ -624,6 +636,7 @@ static BlockDriver bdrv_nbd_tcp = { .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, .bdrv_dirname = nbd_dirname, + .sgfnt_runtime_opts = nbd_sgfnt_runtime_opts, };static BlockDriver bdrv_nbd_unix = {
@@ -645,6 +658,7 @@ static BlockDriver bdrv_nbd_unix = { .bdrv_refresh_filename = nbd_refresh_filename, .bdrv_get_info = nbd_get_info, .bdrv_dirname = nbd_dirname, + .sgfnt_runtime_opts = nbd_sgfnt_runtime_opts, };static void bdrv_nbd_init (void )
block/nfs.c +11 -0
@@ -896,6 +896,15 @@ static void nfs_invalidate_cache(BlockDriverState *bs, }#endif +static const char *nfs_sgfnt_runtime_opts[] = { +"path" , +"user" , +"group" , +"server." , + +NULL +}; +static BlockDriver bdrv_nfs = { .format_name ="nfs" , .protocol_name ="nfs" ,
@@ -922,6 +931,8 @@ static BlockDriver bdrv_nfs = { .bdrv_refresh_filename = nfs_refresh_filename, .bdrv_dirname = nfs_dirname, + .sgfnt_runtime_opts = nfs_sgfnt_runtime_opts, + #ifdef LIBNFS_FEATURE_PAGECACHE .bdrv_invalidate_cache = nfs_invalidate_cache, #endif
block/null.c +9 -0
@@ -254,6 +254,13 @@ static void null_refresh_filename(BlockDriverState *bs, QDict *opts) bs->full_open_options = opts; } +static const char *const null_sgfnt_runtime_opts[] = { +BLOCK_OPT_SIZE , +NULL_OPT_ZEROES , + +NULL +}; +static BlockDriver bdrv_null_co = { .format_name ="null-co" , .protocol_name ="null-co" ,
@@ -272,6 +279,7 @@ static BlockDriver bdrv_null_co = { .bdrv_co_get_block_status = null_co_get_block_status, .bdrv_refresh_filename = null_refresh_filename, + .sgfnt_runtime_opts = null_sgfnt_runtime_opts, };static BlockDriver bdrv_null_aio = {
@@ -292,6 +300,7 @@ static BlockDriver bdrv_null_aio = { .bdrv_co_get_block_status = null_co_get_block_status, .bdrv_refresh_filename = null_refresh_filename, + .sgfnt_runtime_opts = null_sgfnt_runtime_opts, };static void bdrv_null_init (void )
block/qcow.c +7 -0
@@ -1113,6 +1113,12 @@ static QemuOptsList qcow_create_opts = { } }; +static const char *const qcow_sgfnt_runtime_opts[] = { +"encrypt." BLOCK_CRYPTO_OPT_QCOW_KEY_SECRET , + +NULL +}; +static BlockDriver bdrv_qcow = { .format_name ="qcow" , .instance_size =sizeof (BDRVQcowState),
@@ -1134,6 +1140,7 @@ static BlockDriver bdrv_qcow = { .bdrv_get_info = qcow_get_info, .create_opts = &qcow_create_opts, + .sgfnt_runtime_opts = qcow_sgfnt_runtime_opts, };static void bdrv_qcow_init (void )
block/qcow2.c +7 -0
@@ -4315,6 +4315,12 @@ static QemuOptsList qcow2_create_opts = { } }; +static const char *const qcow2_sgfnt_runtime_opts[] = { +"encrypt." BLOCK_CRYPTO_OPT_QCOW_KEY_SECRET , + +NULL +}; +BlockDriver bdrv_qcow2 = { .format_name ="qcow2" , .instance_size =sizeof (BDRVQcow2State),
@@ -4360,6 +4366,7 @@ BlockDriver bdrv_qcow2 = { .bdrv_inactivate = qcow2_inactivate, .create_opts = &qcow2_create_opts, + .sgfnt_runtime_opts = qcow2_sgfnt_runtime_opts, .bdrv_check = qcow2_check, .bdrv_amend_options = qcow2_amend_options,
block/quorum.c +11 -0
@@ -1105,6 +1105,15 @@ static char *quorum_dirname(BlockDriverState *bs, Error **errp)return NULL ; } +static const char *const quorum_sgfnt_runtime_opts[] = { +QUORUM_OPT_VOTE_THRESHOLD , +QUORUM_OPT_BLKVERIFY , +QUORUM_OPT_REWRITE , +QUORUM_OPT_READ_PATTERN , + +NULL +}; +static BlockDriver bdrv_quorum = { .format_name ="quorum" , .protocol_name ="quorum" ,
@@ -1130,6 +1139,8 @@ static BlockDriver bdrv_quorum = { .is_filter = true, .bdrv_recurse_is_first_non_filter = quorum_recurse_is_first_non_filter, + + .sgfnt_runtime_opts = quorum_sgfnt_runtime_opts, };static void bdrv_quorum_init (void )
block/raw-format.c +9 -1
@@ -481,6 +481,13 @@ static int raw_probe_geometry(BlockDriverState *bs, HDGeometry *geo)return bdrv_probe_geometry (bs->file->bs, geo); } +static const char *const raw_sgfnt_runtime_opts[] = { +"offset" , +"size" , + +NULL +}; +BlockDriver bdrv_raw = { .format_name ="raw" , .instance_size =sizeof (BDRVRawState),
@@ -509,7 +516,8 @@ BlockDriver bdrv_raw = { .bdrv_lock_medium = &raw_lock_medium, .bdrv_co_ioctl = &raw_co_ioctl, .create_opts = &raw_create_opts, - .bdrv_has_zero_init = &raw_has_zero_init + .bdrv_has_zero_init = &raw_has_zero_init, + .sgfnt_runtime_opts = raw_sgfnt_runtime_opts, };static void bdrv_raw_init (void )
block/rbd.c +14 -0
@@ -1122,6 +1122,18 @@ static QemuOptsList qemu_rbd_create_opts = { } }; +static const char *const qemu_rbd_sgfnt_runtime_opts[] = { +"pool" , +"image" , +"conf" , +"snapshot" , +"user" , +"server." , +"password-secret" , + +NULL +}; +static BlockDriver bdrv_rbd = { .format_name ="rbd" , .instance_size =sizeof (BDRVRBDState),
@@ -1157,6 +1169,8 @@ static BlockDriver bdrv_rbd = {#ifdef LIBRBD_SUPPORTS_INVALIDATE .bdrv_invalidate_cache = qemu_rbd_invalidate_cache, #endif + + .sgfnt_runtime_opts = qemu_rbd_sgfnt_runtime_opts, };static void bdrv_rbd_init (void )
block/replication.c +8 -0
@@ -701,6 +701,13 @@ static void replication_stop(ReplicationState *rs, bool failover, Error **errp)aio_context_release (aio_context); } +static const char *const replication_sgfnt_runtime_opts[] = { +REPLICATION_MODE , +REPLICATION_TOP_ID , + +NULL +}; +BlockDriver bdrv_replication = { .format_name ="replication" , .protocol_name ="replication" ,
@@ -718,6 +725,7 @@ BlockDriver bdrv_replication = { .bdrv_recurse_is_first_non_filter = replication_recurse_is_first_non_filter, .has_variable_length = true, + .sgfnt_runtime_opts = replication_sgfnt_runtime_opts, };static void bdrv_replication_init (void )
block/sheepdog.c +12 -0
@@ -3076,6 +3076,15 @@ static QemuOptsList sd_create_opts = { } }; +static const char *const sd_sgfnt_runtime_opts[] = { +"vdi" , +"snap-id" , +"tag" , +"server." , + +NULL +}; +static BlockDriver bdrv_sheepdog = { .format_name ="sheepdog" , .protocol_name ="sheepdog" ,
@@ -3110,6 +3119,7 @@ static BlockDriver bdrv_sheepdog = { .bdrv_attach_aio_context = sd_attach_aio_context, .create_opts = &sd_create_opts, + .sgfnt_runtime_opts = sd_sgfnt_runtime_opts, };static BlockDriver bdrv_sheepdog_tcp = {
@@ -3146,6 +3156,7 @@ static BlockDriver bdrv_sheepdog_tcp = { .bdrv_attach_aio_context = sd_attach_aio_context, .create_opts = &sd_create_opts, + .sgfnt_runtime_opts = sd_sgfnt_runtime_opts, };static BlockDriver bdrv_sheepdog_unix = {
@@ -3182,6 +3193,7 @@ static BlockDriver bdrv_sheepdog_unix = { .bdrv_attach_aio_context = sd_attach_aio_context, .create_opts = &sd_create_opts, + .sgfnt_runtime_opts = sd_sgfnt_runtime_opts, };static void bdrv_sheepdog_init (void )
block/ssh.c +12 -0
@@ -1196,6 +1196,17 @@ static int64_t ssh_getlength(BlockDriverState *bs)return length; } +static const char *const ssh_sgfnt_runtime_opts[] = { +"host" , +"port" , +"path" , +"user" , +"host_key_check" , +"server." , + +NULL +}; +static BlockDriver bdrv_ssh = { .format_name ="ssh" , .protocol_name ="ssh" ,
@@ -1210,6 +1221,7 @@ static BlockDriver bdrv_ssh = { .bdrv_getlength = ssh_getlength, .bdrv_co_flush_to_disk = ssh_co_flush, .create_opts = &ssh_create_opts, + .sgfnt_runtime_opts = ssh_sgfnt_runtime_opts, };static void bdrv_ssh_init (void )
block/throttle.c +7 -0
@@ -212,6 +212,12 @@ static void coroutine_fn throttle_co_drain_end(BlockDriverState *bs)atomic_dec (&tgm->io_limits_disabled); } +static const char *const throttle_sgfnt_runtime_opts[] = { +QEMU_OPT_THROTTLE_GROUP_NAME , + +NULL +}; +static BlockDriver bdrv_throttle = { .format_name ="throttle" , .protocol_name ="throttle" ,
@@ -245,6 +251,7 @@ static BlockDriver bdrv_throttle = { .bdrv_co_drain_end = throttle_co_drain_end, .is_filter = true, + .sgfnt_runtime_opts = throttle_sgfnt_runtime_opts, };static void bdrv_throttle_init (void )
block/vpc.c +7 -0
@@ -1084,6 +1084,12 @@ static QemuOptsList vpc_create_opts = { } }; +static const char *const vpc_sgfnt_runtime_opts[] = { +VPC_OPT_SIZE_CALC , + +NULL +}; +static BlockDriver bdrv_vpc = { .format_name ="vpc" , .instance_size =sizeof (BDRVVPCState),
@@ -1103,6 +1109,7 @@ static BlockDriver bdrv_vpc = { .create_opts = &vpc_create_opts, .bdrv_has_zero_init = vpc_has_zero_init, + .sgfnt_runtime_opts = vpc_sgfnt_runtime_opts, };static void bdrv_vpc_init (void )
block/vvfat.c +12 -0
@@ -3242,6 +3242,16 @@ static void vvfat_close(BlockDriverState *bs) } } +static const char *const vvfat_sgfnt_runtime_opts[] = { +"dir" , +"fat-type" , +"floppy" , +"label" , +"rw" , + +NULL +}; +static BlockDriver bdrv_vvfat = { .format_name ="vvfat" , .protocol_name ="fat" ,
@@ -3256,6 +3266,8 @@ static BlockDriver bdrv_vvfat = { .bdrv_co_preadv = vvfat_co_preadv, .bdrv_co_pwritev = vvfat_co_pwritev, .bdrv_co_get_block_status = vvfat_co_get_block_status, + + .sgfnt_runtime_opts = vvfat_sgfnt_runtime_opts, };static void bdrv_vvfat_init (void )
block/vxhs.c +11 -0
@@ -555,6 +555,16 @@ static int64_t vxhs_getlength(BlockDriverState *bs)return vdisk_size; } +static const char *const vxhs_sgfnt_runtime_opts[] = { +VXHS_OPT_VDISK_ID , +"tls-creds" , +VXHS_OPT_HOST , +VXHS_OPT_PORT , +VXHS_OPT_SERVER "." , + +NULL +}; +static BlockDriver bdrv_vxhs = { .format_name ="vxhs" , .protocol_name ="vxhs" ,
@@ -565,6 +575,7 @@ static BlockDriver bdrv_vxhs = { .bdrv_getlength = vxhs_getlength, .bdrv_aio_readv = vxhs_aio_readv, .bdrv_aio_writev = vxhs_aio_writev, + .sgfnt_runtime_opts = vxhs_sgfnt_runtime_opts, };static void bdrv_vxhs_init (void )
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
block/blkdebug.c static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state, · line 893
Quoted 11 lines · Max Reitz
block/blkdebug.c +5
-0@@ -886,6 +886,21 @@ static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state,return 0 ; } +static const char *const blkdebug_sgfnt_runtime_opts[] = { +"config" , +"inject-error." , +"set-state." , +"suspend." ,
Where did this "suspend." come from?
Berto
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
Quoted 7 lines · Alberto Garcia
Quoted 6 lines · Max Reitz
block/blkdebug.c +0
-0@@ -886,6 +886,21 @@ static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state,return 0 ; }
Quoted 7 lines
block/blkdebug.c static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state, · line 893
Quoted 5 lines
block/blkdebug.c +5
-0+static const char *const blkdebug_sgfnt_runtime_opts[] = { +"config" , +"inject-error." , +"set-state." , +"suspend." ,Where did this "suspend." come from?
Er, well, er.
After having looked at it a couple of times now, I can't see anything. I guess I'll remove it and hope for the best...?
Max
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
Quoted 16 lines · Max Reitz
Quoted 7 lines · Alberto Garcia
Quoted 6 lines · Max Reitz
block/blkdebug.c +0
-0@@ -886,6 +886,21 @@ static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state,return 0 ; }Quoted 7 lines
block/blkdebug.c static int blkdebug_reopen_prepare(BDRVReopenState *reopen_state, · line 893
Quoted 5 lines
block/blkdebug.c +5
-0+static const char *const blkdebug_sgfnt_runtime_opts[] = { +"config" , +"inject-error." , +"set-state." , +"suspend." ,Where did this "suspend." come from?
Quoted 4 lines
Er, well, er.
After having looked at it a couple of times now, I can't see anything. I guess I'll remove it and hope for the best...?
My interpretation is that you saw
enum {
ACTION_INJECT_ERROR,
ACTION_SET_STATE,
ACTION_SUSPEND,
};
and assumed that "suspend" was an option (the other two are). I haven't seen any string with the "suspend" text anywhere else in the code.
Berto
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
In reply to Max ReitzQuoted 14 lines · Max Reitz
This new field can be set by block drivers to list the runtime options they accept that may influence the contents of the respective BDS. As of a follow-up patch, this list will be used by the common
bdrv_refresh_filename()implementation to decide which options to put into BDS.full_open_options (and consequently whether a JSON filename has to be created), thus freeing the drivers of having to implement that logic themselves.Additionally, this patch adds the field to all of the block drivers that need it and sets it accordingly.
Signed-off-by Max Reitz <[email protected]>---
include/block/block_int.h struct BlockDriver { · line 463
Quoted 6 lines
include/block/block_int.h +6
-0+/* Pointer to a NULL-terminated array of names of significant options that +* can be specified for bdrv_open(). A significant option is one that +* changes the data of a BDS. +* If this pointer is NULL, the array is considered empty. +* "filename" and "driver" are always considered significant. */ +const char *const *sgfnt_runtime_opts;
Warning: Bikeshedding follows:
s/sgfnt/vital/
might read easier (same number of letters, but has some vowels rthr thn bng crptc bbrvtn).
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
Quoted 24 lines · Eric Blake
Quoted 14 lines · Max Reitz
This new field can be set by block drivers to list the runtime options they accept that may influence the contents of the respective BDS. As of a follow-up patch, this list will be used by the common
bdrv_refresh_filename()implementation to decide which options to put into BDS.full_open_options (and consequently whether a JSON filename has to be created), thus freeing the drivers of having to implement that logic themselves.Additionally, this patch adds the field to all of the block drivers that need it and sets it accordingly.
Signed-off-by Max Reitz <[email protected]>---
Quoted 8 lines
+ /* Pointer to a NULL-terminated array of names of significant options that
+ * can be specified forbdrv_open(). A significant option is one that
+ * changes the data of a BDS.
+ * If this pointer is NULL, the array is considered empty. + * "filename" and "driver" are always considered significant. */ + const char *const*sgfnt_runtime_opts;
Quoted 6 lines
Warning: Bikeshedding follows:
s/sgfnt/vital/
might read easier (same number of letters, but has some vowels rthr thn bng crptc bbrvtn).
So, did we reach any verdict on this? :-)
To sum up IRC (as far as I've understood), I don't quite like "vital" or "needed" because they sound like those options would be required; but they usually have defaults so they are not.
I preferred "strong", but you said that might mean something else (although I still don't know what exactly :-)).
So...?
Max
Re: [Qemu-devel] [PATCH v8 18/26] block: Add sgfnt_runtime_opts to BlockDriver
Quoted 34 lines · Max Reitz
Quoted 24 lines · Eric Blake
Quoted 14 lines · Max Reitz
This new field can be set by block drivers to list the runtime options they accept that may influence the contents of the respective BDS. As of a follow-up patch, this list will be used by the common
bdrv_refresh_filename()implementation to decide which options to put into BDS.full_open_options (and consequently whether a JSON filename has to be created), thus freeing the drivers of having to implement that logic themselves.Additionally, this patch adds the field to all of the block drivers that need it and sets it accordingly.
Signed-off-by Max Reitz <[email protected]>---
Quoted 8 lines
+ /* Pointer to a NULL-terminated array of names of significant options that
+ * can be specified forbdrv_open(). A significant option is one that
+ * changes the data of a BDS.
+ * If this pointer is NULL, the array is considered empty. + * "filename" and "driver" are always considered significant. */ + const char *const*sgfnt_runtime_opts;Quoted 6 lines
Warning: Bikeshedding follows:
s/sgfnt/vital/
might read easier (same number of letters, but has some vowels rthr thn bng crptc bbrvtn).
So, did we reach any verdict on this? :-)
Quoted 8 lines
To sum up IRC (as far as I've understood), I don't quite like "vital" or "needed" because they sound like those options would be required; but they usually have defaults so they are not.
I preferred "strong", but you said that might mean something else (although I still don't know what exactly :-)).
So...?
major? apt?
(the thesaurus mentions other words like meaningful or influential, but those don't buy you any length reduction or easier abbreviations)
I don't recall what (if anything) I said against strong, so I'm also fine if that's what you choose.
At the end of the day, it's all bikeshedding - pick a term, and I'll live with it ;)
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
PATCH v8 19/26 block: Add BlockDriver.bdrv_gather_child_options
In reply to Max ReitzSome follow-up patches will rework the way bs->full_open_options is refreshed in bdrv_refresh_filename(). The new implementation will remove the need for the block drivers' bdrv_refresh_filename() implementations to set bs->full_open_options; instead, it will be generic and use static information from each block driver.
However, by implementing bdrv_gather_child_options(), block drivers will still be able to override the way the full_open_options of their children are incorporated into their own.
We need to implement this function for VMDK because we have to prevent the generic implementation from gathering the options of all children:
It is not possible to specify options for the extents through the runtime options.
For quorum, the child names that would be used by the generic implementation and the ones that we actually (currently) want to use differ. See quorum_gather_child_options() for more information.
Note that both of these are cases which are not ideal: In case of VMDK it would probably be nice to be able to specify options for all extents. In case of quorum, the current runtime option structure is simply broken and needs to be fixed (but that is left for another patch).
--- include/block/block_int.h | 17 +++++++++++++++++ block/quorum.c | 38 ++++++++++++++++++++++++++++++++++++++ block/vmdk.c | 13 +++++++++++++ 3 files changed, 68 insertions(+)
include/block/block_int.h +17 -0
@@ -135,6 +135,23 @@ struct BlockDriver {void (*bdrv_refresh_filename)(BlockDriverState *bs,QDict *options); +/* +* Gathers the open options for all children into @target. +* A simple format driver (without backing file support) might +* implement this function like this: +* +* QINCREF(bs->file->bs->full_open_options); +* qdict_put(target, "file", bs->file->bs->full_open_options); +* +* If not specified, the generic implementation will simply put +* all children's options under their respective name. +* +* Note that ideally this function would not be needed. Every +* block driver which implements it is probably doing something +* shady regarding its runtime option structure. +*/ +void (*bdrv_gather_child_options)(BlockDriverState *bs,QDict *target); +/* * Returns an allocated string which is the directory name of this BDS: It * will be used to make relative filenames absolute by prepending this
block/quorum.c +38 -0
@@ -1095,6 +1095,43 @@ static void quorum_refresh_filename(BlockDriverState *bs, QDict *options) bs->full_open_options = opts; } +static void quorum_gather_child_options (BlockDriverState *bs,QDict *target) +{ +BDRVQuorumState *s = bs->opaque; +QList *children_list; +int i; + +/* The generic implementation for gathering child options in +* bdrv_refresh_filename() would use the names of the children +* as specified for bdrv_open_child() or bdrv_attach_child(), +* which is "children.%u" with %u being a value +* (s->next_child_index) that is incremented each time a new child +* is added (and never decremented). Since children can be +* deleted at runtime, there may be gaps in that enumeration. +* When creating a new quorum BDS and specifying the children for +* it through runtime options, the enumeration used there may not +* have any gaps, though. +* +* Therefore, we have to create a new gap-less enumeration here +* (which we can achieve by simply putting all of the children's +* full_open_options into a QList). +* +* XXX: Note that there are issues with the current child option +* structure quorum uses (such as the fact that children do +* not really have unique permanent names). Therefore, this +* is going to have to change in the future and ideally we +* want quorum to be covered by the generic implementation. +*/ + + children_list =qlist_new (); +qdict_put (target,"children" , children_list); + +for (i =0 ; i < s->num_children; i++) { +QINCREF (s->children[i]->bs->full_open_options); +qlist_append (children_list, s->children[i]->bs->full_open_options); + } +} +static char *quorum_dirname (BlockDriverState *bs,Error **errp) {/* In general, there are multiple BDSs with different dirnames below this
@@ -1123,6 +1160,7 @@ static BlockDriver bdrv_quorum = { .bdrv_file_open = quorum_open, .bdrv_close = quorum_close, .bdrv_refresh_filename = quorum_refresh_filename, + .bdrv_gather_child_options = quorum_gather_child_options, .bdrv_dirname = quorum_dirname, .bdrv_co_flush_to_disk = quorum_co_flush,
block/vmdk.c +13 -0
@@ -2343,6 +2343,18 @@ static int vmdk_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)return 0 ; } +static void vmdk_gather_child_options (BlockDriverState *bs,QDict *target) +{ +/* No children but file and backing can be explicitly specified (TODO) */ +QINCREF (bs->file->bs->full_open_options); +qdict_put (target,"file" , bs->file->bs->full_open_options); + +if (bs->backing && bs->backing_overridden) { +QINCREF (bs->backing->bs->full_open_options); +qdict_put (target,"backing" , bs->backing->bs->full_open_options); + } +} +static QemuOptsList vmdk_create_opts = { .name ="vmdk-create-opts" , .head =QTAILQ_HEAD_INITIALIZER (vmdk_create_opts.head),
@@ -2413,6 +2425,7 @@ static BlockDriver bdrv_vmdk = { .bdrv_get_specific_info = vmdk_get_specific_info, .bdrv_refresh_limits = vmdk_refresh_limits, .bdrv_get_info = vmdk_get_info, + .bdrv_gather_child_options = vmdk_gather_child_options, .supports_backing = true, .create_opts = &vmdk_create_opts,
Signature · 1 line
2.14.3
PATCH v8 20/26 block: Generically refresh runtime options
In reply to Max ReitzInstead of having every block driver which implements bdrv_refresh_filename() copy all of the significant runtime options over to bs->full_open_options, implement this process generically in bdrv_refresh_filename().
This patch only adds this new generic implementation, it does not remove the old functionality. This is done in a follow-up patch.
With this patch, some superfluous information (that should never have been there) may be removed from some JSON filenames, as can be seen in the change to iotest 110's reference output. In case of 191, backing nodes that have not been overridden are now removed from the filename.
--- block.c | 116 ++++++++++++++++++++++++++++++++++++++++++++- tests/qemu-iotests/110.out | 2 +- tests/qemu-iotests/191.out | 24 +++++----- 3 files changed, 128 insertions(+), 14 deletions(-)
block.c +115 -1
@@ -4999,6 +4999,92 @@ out:return to_replace_bs; } +/** +* Iterates through the list of runtime option keys that are said to be +* "significant" for a BDS. An option is called "significant" if it changes a +* BDS's data. For example, the null block driver's "size" and "read-zeroes" +* options are significant, but its "latency-ns" option is not. +* +* If a key returned by this function ends with a dot, all options starting with +* that prefix are significant. +*/ +static const char *const *significant_options (BlockDriverState *bs, +const char *const *curopt) +{ +static const char *const global_options[] = { +"driver" ,"filename" ,NULL + }; + +if (!curopt) { +return &global_options[0 ]; + } + + curopt++; +if (curopt == &global_options[ARRAY_SIZE (global_options) -1 ] && bs->drv) { + curopt = bs->drv->sgfnt_runtime_opts; + } + +return (curopt && *curopt) ? curopt :NULL ; +} + +/** +* Copies all significant runtime options from bs->options to the given QDict. +* The set of significant option keys is determined by invoking +* significant_options(). +* +* Returns true iff any significant option was present in bs->options (and thus +* copied to the target QDict) with the exception of "filename" and "driver". +* The caller is expected to use this value to decide whether the existence of +* significant options prevents the generation of a plain filename. +*/ +static bool append_significant_runtime_options (QDict *d,BlockDriverState *bs) +{ +bool found_any = false; +const char *const *option_name =NULL ; + +if (!bs->drv) { +return false; + } + +while ((option_name =significant_options (bs, option_name))) { +bool option_given = false; + +assert (strlen (*option_name) >0 ); +if ((*option_name)[strlen (*option_name) -1 ] !='.' ) { +QObject *entry =qdict_get (bs->options, *option_name); +if (!entry) { +continue ; + } + +qobject_incref (entry); +qdict_put_obj (d, *option_name, entry); + option_given = true; + }else { +const QDictEntry *entry; +for (entry =qdict_first (bs->options); entry; + entry =qdict_next (bs->options, entry)) + { +if (strstart (qdict_entry_key (entry), *option_name,NULL )) { +qobject_incref (qdict_entry_value (entry)); +qdict_put_obj (d,qdict_entry_key (entry), +qdict_entry_value (entry)); + option_given = true; + } + } + } + +/* While "driver" and "filename" need to be included in a JSON filename, +* their existence does not prohibit generation of a plain filename. */ +if (!found_any && option_given && +strcmp (*option_name,"driver" ) &&strcmp (*option_name,"filename" )) + { + found_any = true; + } + } + +return found_any; +} +static bool append_open_options (QDict *d,BlockDriverState *bs) {const QDictEntry *entry;
@@ -5153,9 +5239,37 @@ void bdrv_refresh_filename(BlockDriverState *bs) bs->full_open_options = opts; } +/* Gather the options QDict */ + opts =qdict_new (); +append_significant_runtime_options (opts, bs); + +if (drv->bdrv_gather_child_options) { +/* Some block drivers may not want to present all of their children's +* options, or name them differently from BdrvChild.name */ + drv->bdrv_gather_child_options (bs, opts); + }else { + QLIST_FOREACH(child, &bs->children, next) { +if (child->role == &child_backing && !bs->backing_overridden) { +/* We can skip the backing BDS if it has not been overridden */ +continue ; + } + +QINCREF (child->bs->full_open_options); +qdict_put (opts, child->name, child->bs->full_open_options); + } + +if (bs->backing_overridden && !bs->backing) { +/* Force no backing file */ +qdict_put (opts,"backing" ,qstring_new ()); + } + } + +QDECREF (bs->full_open_options); + bs->full_open_options = opts; +if (bs->exact_filename[0 ]) {pstrcpy (bs->filename,sizeof (bs->filename), bs->exact_filename); - }else if (bs->full_open_options) { + }else {QString *json =qobject_to_json (QOBJECT (bs->full_open_options));snprintf (bs->filename,sizeof (bs->filename),"json:%s" ,qstring_get_str (json));
tests/qemu-iotests/110.out +1 -1
@@ -22,7 +22,7 @@ Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864 backing_file=t.IMGFMT.b === Nodes without a common directory === -image: json:{"driver": "IMGFMT", "file": {"children": [{"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}, {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.copy"}], "driver": "quorum", "blkverify": false, "rewrite-corrupted": false, "vote-threshold": 1}} +image: json:{"driver": "IMGFMT", "file": {"children": [{"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}, {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.copy"}], "driver": "quorum", "vote-threshold": 1}} file format: IMGFMT virtual size: 64M (67108864 bytes) backing file: t.IMGFMT.base (cannot determine actual path)
tests/qemu-iotests/191.out +12 -12
@@ -52,7 +52,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -77,7 +77,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "file": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encryption_key_missing": false }, {
@@ -123,7 +123,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", + "filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -148,7 +148,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", + "file": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT"}}", "encryption_key_missing": false }, {
@@ -392,7 +392,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -417,7 +417,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "file": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encryption_key_missing": false }, {
@@ -464,7 +464,7 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE,
@@ -474,12 +474,12 @@ wrote 65536/65536 bytes at offset 1048576 }, "backing-filename-format": "IMGFMT", "virtual-size": 67108864, - "filename": "json:{"backing": {"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", + "filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", "cluster-size": 65536, "format": "IMGFMT", "actual-size": SIZE, - "full-backing-filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", - "backing-filename": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "full-backing-filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "backing-filename": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "dirty-flag": false }, "iops_wr": 0,
@@ -490,7 +490,7 @@ wrote 65536/65536 bytes at offset 1048576 "iops": 0, "bps_wr": 0, "write_threshold": 0, - "backing_file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", + "backing_file": "json:{"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}", "encrypted": false, "bps": 0, "bps_rd": 0,
@@ -499,7 +499,7 @@ wrote 65536/65536 bytes at offset 1048576 "direct": false, "writeback": true }, - "file": "json:{"backing": {"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.base"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", + "file": "json:{"backing": {"backing": {"driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.mid"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl2"}}, "driver": "IMGFMT", "file": {"driver": "file", "filename": "TEST_DIR/t.IMGFMT.ovl3"}}", "encryption_key_missing": false }, {
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 20/26] block: Generically refresh runtime options
Quoted 8 lines · Max Reitz
Instead of having every block driver which implements
bdrv_refresh_filename()copy all of the significant runtime options over tobs->full_open_options, implement this process generically inbdrv_refresh_filename().This patch only adds this new generic implementation, it does not remove the old functionality. This is done in a follow-up patch.
Quoted 6 lines
With this patch, some superfluous information (that should never have been there) may be removed from some JSON filenames, as can be seen in the change to iotest 110's reference output. In case of 191, backing nodes that have not been overridden are now removed from the filename.
Signed-off-by Max Reitz <[email protected]>
Berto
PATCH v8 21/26 block: Purify .bdrv_refresh_filename()
In reply to Max ReitzCurrently, BlockDriver.bdrv_refresh_filename() is supposed to both refresh the filename (BDS.exact_filename) and set BDS.full_open_options. Now that we have generic code in the central bdrv_refresh_filename() for creating BDS.full_open_options, we can drop the latter part from all BlockDriver.bdrv_refresh_filename() implementations.
This also means that we can drop all of the existing default code for this from the global bdrv_refresh_filename() itself.
Furthermore, we now have to call BlockDriver.bdrv_refresh_filename() after having set BDS.full_open_options, because the block driver's implementation should now be allowed to depend on BDS.full_open_options being set correctly.
Finally, with this patch we can drop the @options parameter from BlockDriver.bdrv_refresh_filename(); also, add a comment on this function's purpose in block/block_int.h while touching its interface.
--- include/block/block_int.h | 6 +- block.c | 145 +++++++--------------------------------------- block/blkdebug.c | 55 +++++++----------- block/blkverify.c | 16 +---- block/commit.c | 2 +- block/mirror.c | 2 +- block/nbd.c | 23 +------- block/nfs.c | 36 +----------- block/null.c | 23 +++++--- block/quorum.c | 30 ---------- 10 files changed, 66 insertions(+), 272 deletions(-)
include/block/block_int.h +5 -1
@@ -133,7 +133,11 @@ struct BlockDriver {int (*bdrv_create)(const char *filename,QemuOpts *opts,Error **errp);int (*bdrv_make_empty)(BlockDriverState *bs); -void (*bdrv_refresh_filename)(BlockDriverState *bs,QDict *options); +/* +* Refreshes the bs->exact_filename field. If that is impossible, +* bs->exact_filename has to be left empty. +*/ +void (*bdrv_refresh_filename)(BlockDriverState *bs);/* * Gathers the open options for all children into @target.
block.c +22 -123
171 lines not on this page · raw message
block/blkdebug.c +20 -35
72 lines not on this page · raw message
block/blkverify.c +1 -15
@@ -277,24 +277,10 @@ static bool blkverify_recurse_is_first_non_filter(BlockDriverState *bs,return bdrv_recurse_is_first_non_filter (s->test_file->bs, candidate); } -static void blkverify_refresh_filename (BlockDriverState *bs,QDict *options) +static void blkverify_refresh_filename (BlockDriverState *bs) {BDRVBlkverifyState *s = bs->opaque; -if (bs->file->bs->full_open_options - && s->test_file->bs->full_open_options) - { -QDict *opts =qdict_new (); -qdict_put_str (opts,"driver" ,"blkverify" ); - -QINCREF (bs->file->bs->full_open_options); -qdict_put (opts,"raw" , bs->file->bs->full_open_options); -QINCREF (s->test_file->bs->full_open_options); -qdict_put (opts,"test" , s->test_file->bs->full_open_options); - - bs->full_open_options = opts; - } -if (bs->file->bs->exact_filename[0 ] && s->test_file->bs->exact_filename[0 ]) {
block/commit.c +1 -1
@@ -239,7 +239,7 @@ static int coroutine_fn bdrv_commit_top_preadv(BlockDriverState *bs,return bdrv_co_preadv (bs->backing, offset, bytes, qiov, flags); } -static void bdrv_commit_top_refresh_filename (BlockDriverState *bs,QDict *opts) +static void bdrv_commit_top_refresh_filename (BlockDriverState *bs) {pstrcpy (bs->exact_filename,sizeof (bs->exact_filename), bs->backing->bs->filename);
block/mirror.c +1 -1
@@ -1058,7 +1058,7 @@ static int coroutine_fn bdrv_mirror_top_pdiscard(BlockDriverState *bs,return bdrv_co_pdiscard (bs->backing->bs, offset, bytes); } -static void bdrv_mirror_top_refresh_filename (BlockDriverState *bs,QDict *opts) +static void bdrv_mirror_top_refresh_filename (BlockDriverState *bs) {if (bs->backing ==NULL ) {/* we can be here after failed bdrv_attach_child in
block/nbd.c +1 -22
@@ -515,12 +515,9 @@ static void nbd_attach_aio_context(BlockDriverState *bs,nbd_client_attach_aio_context (bs, new_context); } -static void nbd_refresh_filename (BlockDriverState *bs,QDict *options) +static void nbd_refresh_filename (BlockDriverState *bs) {BDRVNBDState *s = bs->opaque; -QDict *opts =qdict_new (); -QObject *saddr_qdict; -Visitor *ov;const char *host =NULL , *port =NULL , *path =NULL ;if (s->saddr->type ==SOCKET_ADDRESS_TYPE_INET ) {
@@ -533,8 +530,6 @@ static void nbd_refresh_filename(BlockDriverState *bs, QDict *options) path = s->saddr->u.q_unix.path; }/* else can't represent as pseudo-filename */ -qdict_put_str (opts,"driver" ,"nbd" ); -if (path && s->export) {snprintf (bs->exact_filename,sizeof (bs->exact_filename),"nbd+unix:///%s?socket=%s" , s->export, path);
@@ -548,22 +543,6 @@ static void nbd_refresh_filename(BlockDriverState *bs, QDict *options)snprintf (bs->exact_filename,sizeof (bs->exact_filename),"nbd://%s:%s" , host, port); } - - ov =qobject_output_visitor_new (&saddr_qdict); -visit_type_SocketAddress (ov,NULL , &s->saddr, &error_abort); -visit_complete (ov, &saddr_qdict); -visit_free (ov); -qdict_put_obj (opts,"server" , saddr_qdict); - -if (s->export) { -qdict_put_str (opts,"export" , s->export); - } -if (s->tlscredsid) { -qdict_put_str (opts,"tls-creds" , s->tlscredsid); - } - -qdict_flatten (opts); - bs->full_open_options = opts; }static int nbd_get_info (BlockDriverState *bs,BlockDriverInfo *bdi)
block/nfs.c +1 -35
@@ -818,14 +818,9 @@ static int nfs_reopen_prepare(BDRVReopenState *state,return 0 ; } -static void nfs_refresh_filename (BlockDriverState *bs,QDict *options) +static void nfs_refresh_filename (BlockDriverState *bs) {NFSClient *client = bs->opaque; -QDict *opts =qdict_new (); -QObject *server_qdict; -Visitor *ov; - -qdict_put_str (opts,"driver" ,"nfs" );if (client->uid && !client->gid) { snprintf(bs->exact_filename, sizeof(bs->exact_filename),
@@ -843,35 +838,6 @@ static void nfs_refresh_filename(BlockDriverState *bs, QDict *options)snprintf (bs->exact_filename,sizeof (bs->exact_filename),"nfs://%s%s" , client->server->host, client->path); } - - ov =qobject_output_visitor_new (&server_qdict); -visit_type_NFSServer (ov,NULL , &client->server, &error_abort); -visit_complete (ov, &server_qdict); -qdict_put_obj (opts,"server" , server_qdict); -qdict_put_str (opts,"path" , client->path); - -if (client->uid) { -qdict_put_int (opts,"user" , client->uid); - } -if (client->gid) { -qdict_put_int (opts,"group" , client->gid); - } -if (client->tcp_syncnt) { -qdict_put_int (opts,"tcp-syn-cnt" , client->tcp_syncnt); - } -if (client->readahead) { -qdict_put_int (opts,"readahead-size" , client->readahead); - } -if (client->pagecache) { -qdict_put_int (opts,"page-cache-size" , client->pagecache); - } -if (client->debug) { -qdict_put_int (opts,"debug" , client->debug); - } - -visit_free (ov); -qdict_flatten (opts); - bs->full_open_options = opts; }static char *nfs_dirname (BlockDriverState *bs,Error **errp)
block/null.c +14 -9
@@ -240,18 +240,23 @@ static int64_t coroutine_fn null_co_get_block_status(BlockDriverState *bs, } } -static void null_refresh_filename (BlockDriverState *bs,QDict *opts) +static void null_refresh_filename (BlockDriverState *bs) { -QINCREF (opts); -qdict_del (opts,"filename" ); - -if (!qdict_size (opts)) { -snprintf (bs->exact_filename,sizeof (bs->exact_filename),"%s://" , - bs->drv->format_name); +const QDictEntry *e; + +for (e =qdict_first (bs->full_open_options); e; + e =qdict_next (bs->full_open_options, e)) + { +/* These options can be ignored */ +if (strcmp (qdict_entry_key (e),"filename" ) && +strcmp (qdict_entry_key (e),"driver" )) + { +return ; + } } -qdict_put_str (opts,"driver" , bs->drv->format_name); - bs->full_open_options = opts; +snprintf (bs->exact_filename,sizeof (bs->exact_filename),"%s://" , + bs->drv->format_name); }static const char *const null_sgfnt_runtime_opts[] = {
block/quorum.c +0 -30
@@ -1066,35 +1066,6 @@ static void quorum_del_child(BlockDriverState *bs, BdrvChild *child,bdrv_drained_end (bs); } -static void quorum_refresh_filename (BlockDriverState *bs,QDict *options) -{ -BDRVQuorumState *s = bs->opaque; -QDict *opts; -QList *children; -int i; - -for (i =0 ; i < s->num_children; i++) { -if (!s->children[i]->bs->full_open_options) { -return ; - } - } - - children =qlist_new (); -for (i =0 ; i < s->num_children; i++) { -QINCREF (s->children[i]->bs->full_open_options); -qlist_append (children, s->children[i]->bs->full_open_options); - } - - opts =qdict_new (); -qdict_put_str (opts,"driver" ,"quorum" ); -qdict_put_int (opts,QUORUM_OPT_VOTE_THRESHOLD , s->threshold); -qdict_put_bool (opts,QUORUM_OPT_BLKVERIFY , s->is_blkverify); -qdict_put_bool (opts,QUORUM_OPT_REWRITE , s->rewrite_corrupted); -qdict_put (opts,"children" , children); - - bs->full_open_options = opts; -} -static void quorum_gather_child_options (BlockDriverState *bs,QDict *target) {BDRVQuorumState *s = bs->opaque;
@@ -1159,7 +1130,6 @@ static BlockDriver bdrv_quorum = { .bdrv_file_open = quorum_open, .bdrv_close = quorum_close, - .bdrv_refresh_filename = quorum_refresh_filename, .bdrv_gather_child_options = quorum_gather_child_options, .bdrv_dirname = quorum_dirname,
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 21/26] block: Purify .bdrv_refresh_filename()
Quoted 14 lines · Max Reitz
Currently,
BlockDriver.bdrv_refresh_filename()is supposed to both refresh the filename (BDS.exact_filename) and set BDS.full_open_options. Now that we have generic code in the centralbdrv_refresh_filename()for creating BDS.full_open_options, we can drop the latter part from allBlockDriver.bdrv_refresh_filename()implementations.This also means that we can drop all of the existing default code for this from the global
bdrv_refresh_filename()itself.Furthermore, we now have to call
BlockDriver.bdrv_refresh_filename()after having set BDS.full_open_options, because the block driver's implementation should now be allowed to depend on BDS.full_open_options being set correctly.
Quoted 5 lines
Finally, with this patch we can drop the @options parameter from
BlockDriver.bdrv_refresh_filename(); also, add a comment on this function's purpose inblock/block_int.hwhile touching its interface.Signed-off-by Max Reitz <[email protected]>
Berto
PATCH v8 22/26 block: Do not copy exact_filename from format file
In reply to Max ReitzIf a format BDS's file BDS is in turn a format BDS, we cannot simply use the same filename, because when opening a BDS tree based on a filename alone, qemu will create only one format node on top of one protocol node (disregarding a potential backing file).
--- block.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-)
block.c +14 -3
@@ -5159,9 +5159,20 @@ void bdrv_refresh_filename(BlockDriverState *bs) bs->exact_filename[0 ] ='\0' ; -/* If no specific options have been given for this BDS, the filename of -* the underlying file should suffice for this one as well */ -if (bs->file->bs->exact_filename[0 ] && !generate_json_filename) { +/* We can use the underlying file's filename if: +* - it has a filename, +* - the file is a protocol BDS, and +* - opening that file (as this BDS's format) will automatically create +* the BDS tree we have right now, that is: +* - the user did not significantly change this BDS's behavior with +* some explicit options +* - no non-file child of this BDS has been overridden by the user +* Both of these conditions are represented by generate_json_filename. +*/ +if (bs->file->bs->exact_filename[0 ] && + bs->file->bs->drv->bdrv_file_open && + !generate_json_filename) + {strcpy (bs->exact_filename, bs->file->bs->exact_filename); } }
Signature · 1 line
2.14.3
PATCH v8 23/26 block: Fix FIXME from "Add BDS.backing_overridden"
In reply to Max ReitzSaid commit introduced a FIXME stating that bdrv_open_backing_file() should set bs->backing_overridden to true not only if the file.filename option was set, but if the "options" QDict contained any option that is significant for any node in the BDS tree emerging from the backing BDS. This behavior is implemented by this patch.
--- block.c | 81 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 75 insertions(+), 6 deletions(-)
block.c +75 -6
@@ -73,6 +73,9 @@ static BlockDriverState *bdrv_open_inherit(const char *filename,const BdrvChildRole *child_role, Error **errp); +static bool has_significant_runtime_options (BlockDriverState *bs, +const QDict *d); +/* If non-zero, use only whitelisted block drivers */ static int use_bdrv_whitelist;
@@ -2249,6 +2252,42 @@ out:bdrv_refresh_limits (bs,NULL ); } +/** +* Checks whether @options contains any significant option for any of the nodes +* in the BDS tree emerging from @bs. +*/ +static bool is_significant_option_tree (BlockDriverState *bs,QDict *options) +{ +BdrvChild *child; + +if (!qdict_size (options)) { +/* No need to recurse */ +return false; + } + +if (has_significant_runtime_options (bs, options)) { +return true; + } + + QLIST_FOREACH(child, &bs->children, next) { +QDict *child_options; +char *option_prefix; + + option_prefix =g_strdup_printf ("%s." , child->name); +qdict_extract_subqdict (options, &child_options, option_prefix); +g_free (option_prefix); + +if (is_significant_option_tree (child->bs, child_options)) { +QDECREF (child_options); +return true; + } + +QDECREF (child_options); + } + +return false; +} +/* * Opens the backing file for a BlockDriverState if not yet open *
@@ -2267,7 +2306,7 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options,const char *reference =NULL ;int ret =0 ;BlockDriverState *backing_hd; -QDict *options; +QDict *options, *cloned_options =NULL ;QDict *tmp_parent_options =NULL ;Error *local_err =NULL ;
@@ -2297,11 +2336,6 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options, reference =qdict_get_try_str (parent_options, bdref_key);if (reference ||qdict_haskey (options,"file.filename" )) {/* keep backing_filename NULL */ - -/* FIXME: Should also be set to true if @options contains other runtime -* options which control the data that is read from the backing -* BDS */ - bs->backing_overridden = true; }else if (bs->backing_file[0 ] =='\0' &&qdict_size (options) ==0 ) {QDECREF (options); goto free_exit;
@@ -2322,6 +2356,8 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options, goto free_exit; } + cloned_options =qdict_clone_shallow (options); +if (!reference && bs->backing_format[0 ] !='\0' && !qdict_haskey (options,"driver" )) {qdict_put_str (options,"driver" , bs->backing_format);
@@ -2337,6 +2373,10 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options, }bdrv_set_aio_context (backing_hd,bdrv_get_aio_context (bs)); +if (reference ||is_significant_option_tree (backing_hd, cloned_options)) { + bs->backing_overridden = true; + } +/* Hook up the backing file link; drop our reference, bs owns the * backing_hd reference now */ bdrv_set_backing_hd (bs, backing_hd, &local_err);
@@ -2352,6 +2392,7 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *parent_options, free_exit:g_free (backing_filename);QDECREF (tmp_parent_options); +QDECREF (cloned_options);return ret; }
@@ -5027,6 +5068,34 @@ static const char *const *significant_options(BlockDriverState *bs,return (curopt && *curopt) ? curopt :NULL ; } +/** +* Returns true if @d contains any options the block driver of @bs considers to +* be significant runtime options. +*/ +static bool has_significant_runtime_options (BlockDriverState *bs, +const QDict *d) +{ +const char *const *option_name =NULL ; + +while ((option_name =significant_options (bs, option_name))) { +assert (strlen (*option_name) >0 ); +if ((*option_name)[strlen (*option_name) -1 ] !='.' ) { +if (qdict_haskey (d, *option_name)) { +return true; + } + }else { +const QDictEntry *entry; +for (entry =qdict_first (d); entry; entry =qdict_next (d, entry)) { +if (strstart (qdict_entry_key (entry), *option_name,NULL )) { +return true; + } + } + } + } + +return false; +} +/** * Copies all significant runtime options from bs->options to the given QDict. * The set of significant option keys is determined by invoking
Signature · 1 line
2.14.3
PATCH v8 24/26 block/curl: Harmonize option defaults
In reply to Max ReitzBoth of the defaults we currently have in the curl driver are named based on a slightly different schema, let's unify that and call both CURL_BLOCK_OPT_${NAME}_DEFAULT.
While at it, we can add a macro for the third option for which a default exists, namely "sslverify".
--- block/curl.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)
block/curl.c +8 -5
@@ -72,8 +72,6 @@ static CURLMcode __curl_multi_socket_action(CURLM *multi_handle,#define CURL_NUM_STATES 8#define CURL_NUM_ACB 8 -#define READ_AHEAD_DEFAULT (256 * 1024) -#define CURL_TIMEOUT_DEFAULT 5#define CURL_TIMEOUT_MAX 10000#define CURL_BLOCK_OPT_URL "url"
@@ -87,6 +85,10 @@ static CURLMcode __curl_multi_socket_action(CURLM *multi_handle,#define CURL_BLOCK_OPT_PROXY_USERNAME "proxy-username"#define CURL_BLOCK_OPT_PROXY_PASSWORD_SECRET "proxy-password-secret" +#define CURL_BLOCK_OPT_READAHEAD_DEFAULT (256 * 1024) +#define CURL_BLOCK_OPT_SSLVERIFY_DEFAULT true +#define CURL_BLOCK_OPT_TIMEOUT_DEFAULT 5 +struct BDRVCURLState ;static bool libcurl_initialized;
@@ -714,7 +716,7 @@ static int curl_open(BlockDriverState *bs, QDict *options, int flags, } s->readahead_size =qemu_opt_get_size (opts,CURL_BLOCK_OPT_READAHEAD , -READ_AHEAD_DEFAULT ); +CURL_BLOCK_OPT_READAHEAD_DEFAULT );if ((s->readahead_size &0x1ff ) !=0 ) {error_setg (errp,"HTTP_READAHEAD_SIZE %zd is not a multiple of 512" , s->readahead_size);
@@ -722,13 +724,14 @@ static int curl_open(BlockDriverState *bs, QDict *options, int flags, } s->timeout =qemu_opt_get_number (opts,CURL_BLOCK_OPT_TIMEOUT , -CURL_TIMEOUT_DEFAULT ); +CURL_BLOCK_OPT_TIMEOUT_DEFAULT );if (s->timeout >CURL_TIMEOUT_MAX ) {error_setg (errp,"timeout parameter is too large or negative" ); goto out_noclean; } - s->sslverify =qemu_opt_get_bool (opts,CURL_BLOCK_OPT_SSLVERIFY , true); + s->sslverify =qemu_opt_get_bool (opts,CURL_BLOCK_OPT_SSLVERIFY , +CURL_BLOCK_OPT_SSLVERIFY_DEFAULT ); cookie =qemu_opt_get (opts,CURL_BLOCK_OPT_COOKIE ); cookie_secret =qemu_opt_get (opts,CURL_BLOCK_OPT_COOKIE_SECRET );
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 24/26] block/curl: Harmonize option defaults
Quoted 8 lines · Max Reitz
Both of the defaults we currently have in the curl driver are named based on a slightly different schema, let's unify that and call both CURL_BLOCK_OPT_${NAME}_DEFAULT.
While at it, we can add a macro for the third option for which a default exists, namely "sslverify".
Signed-off-by Max Reitz <[email protected]>
Berto
PATCH v8 25/26 block/curl: Implement bdrv_refresh_filename()
In reply to Max Reitz--- block/curl.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+)
block/curl.c +21 -0
@@ -972,6 +972,23 @@ static int64_t curl_getlength(BlockDriverState *bs)return s->len; } +static void curl_refresh_filename (BlockDriverState *bs) +{ +BDRVCURLState *s = bs->opaque; + +/* "readahead" and "timeout" do not change the guest-visible data, +* so ignore them */ +if (s->sslverify !=CURL_BLOCK_OPT_SSLVERIFY_DEFAULT || + s->cookie || s->username || s->password || s->proxyusername || + s->proxypassword) + { +return ; + } + +pstrcpy (bs->exact_filename,sizeof (bs->exact_filename), s->url); +} + +static const char *const curl_sgfnt_runtime_opts[] = {CURL_BLOCK_OPT_URL ,CURL_BLOCK_OPT_SSLVERIFY ,
@@ -1000,6 +1017,7 @@ static BlockDriver bdrv_http = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + .bdrv_refresh_filename = curl_refresh_filename, .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };
@@ -1018,6 +1036,7 @@ static BlockDriver bdrv_https = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + .bdrv_refresh_filename = curl_refresh_filename, .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };
@@ -1036,6 +1055,7 @@ static BlockDriver bdrv_ftp = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + .bdrv_refresh_filename = curl_refresh_filename, .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };
@@ -1054,6 +1074,7 @@ static BlockDriver bdrv_ftps = { .bdrv_detach_aio_context = curl_detach_aio_context, .bdrv_attach_aio_context = curl_attach_aio_context, + .bdrv_refresh_filename = curl_refresh_filename, .sgfnt_runtime_opts = curl_sgfnt_runtime_opts, };
Signature · 1 line
2.14.3
Re: [Qemu-devel] [PATCH v8 25/26] block/curl: Implement bdrv_refresh_filename()
Quoted 1 line · Max Reitz
Signed-off-by Max Reitz <[email protected]>
Berto
PATCH v8 26/26 block/null: Generate filename even with latency-ns
In reply to Max ReitzWhile we cannot represent the latency-ns option in a filename, it is not a significant option so not being able to should not stop us from generating a filename nonetheless.
--- block/null.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
block/null.c +2 -1
@@ -249,7 +249,8 @@ static void null_refresh_filename(BlockDriverState *bs) {/* These options can be ignored */ if (strcmp (qdict_entry_key (e),"filename" ) && -strcmp (qdict_entry_key (e),"driver" )) +strcmp (qdict_entry_key (e),"driver" ) && +strcmp (qdict_entry_key (e),NULL_OPT_LATENCY )) {return ; }
Signature · 1 line
2.14.3