The following changes since commit 42f6c9179be4401974dd3a75ee72defd16b5092d: Merge tag 'pull-ppc-20211112' of https://github.com/legoater/qemu into staging (2021-11-12 12:28:25 +0100) are available in the Git repository at: git://repo.or.cz/qemu/kevin.git tags/for-upstream for you to fetch changes up to 7461272c5f6032436ef9032c091c0118539483e4: softmmu/qdev-monitor: fix use-after-free in qdev_set_id() (2021-11-15 15:49:46 +0100) ---------------------------------------------------------------- Block layer patches - Fixes to image streaming job and block layer reconfiguration to make iotest 030 pass again - docs: Deprecate incorrectly typed device_add arguments - file-posix: Fix alignment after reopen changing O_DIRECT ---------------------------------------------------------------- Hanna Reitz (10): stream: Traverse graph after modification block: Manipulate children list in .attach/.detach block: Unite remove_empty_child and child_free block: Drop detached child from ignore list block: Pass BdrvChild ** to replace_child_noperm block: Restructure remove_file_or_backing_child() transactions: Invoke clean() after everything else block: Let replace_child_tran keep indirect pointer block: Let replace_child_noperm free children iotests/030: Unthrottle parallel jobs in reverse Kevin Wolf (2): docs: Deprecate incorrectly typed device_add arguments file-posix: Fix alignment after reopen changing O_DIRECT Stefan Hajnoczi (1): softmmu/qdev-monitor: fix use-after-free in qdev_set_id() docs/about/deprecated.rst | 14 +++ include/qemu/transactions.h | 3 + block.c | 233 +++++++++++++++++++++++++++++++++----------- block/file-posix.c | 20 +++- block/stream.c | 7 +- softmmu/qdev-monitor.c | 2 +- util/transactions.c | 8 +- tests/qemu-iotests/030 | 11 ++- tests/qemu-iotests/142 | 22 +++++ tests/qemu-iotests/142.out | 15 +++ 10 files changed, 269 insertions(+), 66 deletions(-) From: Hanna Reitz <hreitz@redhat.com> bdrv_cor_filter_drop() modifies the block graph. That means that other parties can also modify the block graph before it returns. Therefore, we cannot assume that the result of a graph traversal we did before remains valid afterwards. We should thus fetch `base` and `unfiltered_base` afterwards instead of before. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211111120829.81329-2-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/stream.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/block/stream.c b/block/stream.c index XXXXXXX..XXXXXXX 100644 --- a/block/stream.c +++ b/block/stream.c @@ -XXX,XX +XXX,XX @@ static int stream_prepare(Job *job) { StreamBlockJob *s = container_of(job, StreamBlockJob, common.job); BlockDriverState *unfiltered_bs = bdrv_skip_filters(s->target_bs); - BlockDriverState *base = bdrv_filter_or_cow_bs(s->above_base); - BlockDriverState *unfiltered_base = bdrv_skip_filters(base); + BlockDriverState *base; + BlockDriverState *unfiltered_base; Error *local_err = NULL; int ret = 0; @@ -XXX,XX +XXX,XX @@ static int stream_prepare(Job *job) bdrv_cor_filter_drop(s->cor_filter_bs); s->cor_filter_bs = NULL; + base = bdrv_filter_or_cow_bs(s->above_base); + unfiltered_base = bdrv_skip_filters(base); + if (bdrv_cow_child(unfiltered_bs)) { const char *base_id = NULL, *base_fmt = NULL; if (unfiltered_base) { -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> The children list is specific to BDS parents. We should not modify it in the general children modification code, but let BDS parents deal with it in their .attach() and .detach() methods. This also has the advantage that a BdrvChild is removed from the children list before its .bs pointer can become NULL. BDS parents generally assume that their children's .bs pointer is never NULL, so this is actually a bug fix. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211111120829.81329-3-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_child_cb_attach(BdrvChild *child) { BlockDriverState *bs = child->opaque; + QLIST_INSERT_HEAD(&bs->children, child, next); + if (child->role & BDRV_CHILD_COW) { bdrv_backing_attach(child); } @@ -XXX,XX +XXX,XX @@ static void bdrv_child_cb_detach(BdrvChild *child) } bdrv_unapply_subtree_drain(child, bs); + + QLIST_REMOVE(child, next); } static int bdrv_child_cb_update_filename(BdrvChild *c, BlockDriverState *base, @@ -XXX,XX +XXX,XX @@ static void bdrv_child_free(void *opaque) static void bdrv_remove_empty_child(BdrvChild *child) { assert(!child->bs); - QLIST_SAFE_REMOVE(child, next); + assert(!child->next.le_prev); /* not in children list */ bdrv_child_free(child); } @@ -XXX,XX +XXX,XX @@ static int bdrv_attach_child_noperm(BlockDriverState *parent_bs, return ret; } - QLIST_INSERT_HEAD(&parent_bs->children, *child, next); - /* - * child is removed in bdrv_attach_child_common_abort(), so don't care to - * abort this change separately. - */ - return 0; } @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_filter_or_cow_child_abort(void *opaque) BdrvRemoveFilterOrCowChild *s = opaque; BlockDriverState *parent_bs = s->child->opaque; - QLIST_INSERT_HEAD(&parent_bs->children, s->child, next); if (s->is_backing) { parent_bs->backing = s->child; } else { @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, }; tran_add(tran, &bdrv_remove_filter_or_cow_child_drv, s); - QLIST_SAFE_REMOVE(child, next); if (s->is_backing) { bs->backing = NULL; } else { -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> Now that bdrv_remove_empty_child() no longer removes the child from the parent's children list but only checks that it is not in such a list, it is only a wrapper around bdrv_child_free() that checks that the child is empty and unused. That should apply to all children that we free, so put those checks into bdrv_child_free() and drop bdrv_remove_empty_child(). Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211111120829.81329-4-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild *child, } } -static void bdrv_child_free(void *opaque) -{ - BdrvChild *c = opaque; - - g_free(c->name); - g_free(c); -} - -static void bdrv_remove_empty_child(BdrvChild *child) +/** + * Free the given @child. + * + * The child must be empty (i.e. `child->bs == NULL`) and it must be + * unused (i.e. not in a children list). + */ +static void bdrv_child_free(BdrvChild *child) { assert(!child->bs); assert(!child->next.le_prev); /* not in children list */ - bdrv_child_free(child); + + g_free(child->name); + g_free(child); } typedef struct BdrvAttachChildCommonState { @@ -XXX,XX +XXX,XX @@ static void bdrv_attach_child_common_abort(void *opaque) } bdrv_unref(bs); - bdrv_remove_empty_child(child); + bdrv_child_free(child); *s->child = NULL; } @@ -XXX,XX +XXX,XX @@ static int bdrv_attach_child_common(BlockDriverState *child_bs, if (ret < 0) { error_propagate(errp, local_err); - bdrv_remove_empty_child(new_child); + bdrv_child_free(new_child); return ret; } } @@ -XXX,XX +XXX,XX @@ static void bdrv_detach_child(BdrvChild *child) BlockDriverState *old_bs = child->bs; bdrv_replace_child_noperm(child, NULL); - bdrv_remove_empty_child(child); + bdrv_child_free(child); if (old_bs) { /* -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> bdrv_attach_child_common_abort() restores the parent's AioContext. To do so, the child (which was supposed to be attached, but is now detached again by this abort handler) is added to the ignore list for the AioContext changing functions. However, since we modify a BDS's children list in the BdrvChildClass's .attach and .detach handlers, the child is already effectively detached from the parent by this point. We do not need to put it into the ignore list. Use this opportunity to clean up the empty line structure: Keep setting the ignore list, invoking the AioContext function, and freeing the ignore list in blocks separated by empty lines. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Message-Id: <20211111120829.81329-5-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_attach_child_common_abort(void *opaque) } if (bdrv_child_get_parent_aio_context(child) != s->old_parent_ctx) { - GSList *ignore = g_slist_prepend(NULL, child); + GSList *ignore; + /* No need to ignore `child`, because it has been detached already */ + ignore = NULL; child->klass->can_set_aio_ctx(child, s->old_parent_ctx, &ignore, &error_abort); g_slist_free(ignore); - ignore = g_slist_prepend(NULL, child); - child->klass->set_aio_ctx(child, s->old_parent_ctx, &ignore); + ignore = NULL; + child->klass->set_aio_ctx(child, s->old_parent_ctx, &ignore); g_slist_free(ignore); } -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> bdrv_replace_child_noperm() modifies BdrvChild.bs, and can potentially set it to NULL. That is dangerous, because BDS parents generally assume that their children's .bs pointer is never NULL. We therefore want to let bdrv_replace_child_noperm() set the corresponding BdrvChild pointer to NULL, too. This patch lays the foundation for it by passing a BdrvChild ** pointer to bdrv_replace_child_noperm() so that it can later use it to NULL the BdrvChild pointer immediately after setting BdrvChild.bs to NULL. (We will still need to undertake some intermediate steps, though.) Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-6-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static BlockDriverState *bdrv_open_inherit(const char *filename, static bool bdrv_recurse_has_child(BlockDriverState *bs, BlockDriverState *child); -static void bdrv_replace_child_noperm(BdrvChild *child, +static void bdrv_replace_child_noperm(BdrvChild **child, BlockDriverState *new_bs); static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, BdrvChild *child, @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_abort(void *opaque) BlockDriverState *new_bs = s->child->bs; /* old_bs reference is transparently moved from @s to @s->child */ - bdrv_replace_child_noperm(s->child, s->old_bs); + bdrv_replace_child_noperm(&s->child, s->old_bs); bdrv_unref(new_bs); } @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_tran(BdrvChild *child, BlockDriverState *new_bs, if (new_bs) { bdrv_ref(new_bs); } - bdrv_replace_child_noperm(child, new_bs); + bdrv_replace_child_noperm(&child, new_bs); /* old_bs reference is transparently moved from @child to @s */ } @@ -XXX,XX +XXX,XX @@ uint64_t bdrv_qapi_perm_to_blk_perm(BlockPermission qapi_perm) return permissions[qapi_perm]; } -static void bdrv_replace_child_noperm(BdrvChild *child, +static void bdrv_replace_child_noperm(BdrvChild **childp, BlockDriverState *new_bs) { + BdrvChild *child = *childp; BlockDriverState *old_bs = child->bs; int new_bs_quiesce_counter; int drain_saldo; @@ -XXX,XX +XXX,XX @@ static void bdrv_attach_child_common_abort(void *opaque) BdrvChild *child = *s->child; BlockDriverState *bs = child->bs; - bdrv_replace_child_noperm(child, NULL); + bdrv_replace_child_noperm(s->child, NULL); if (bdrv_get_aio_context(bs) != s->old_child_ctx) { bdrv_try_set_aio_context(bs, s->old_child_ctx, &error_abort); @@ -XXX,XX +XXX,XX @@ static int bdrv_attach_child_common(BlockDriverState *child_bs, } bdrv_ref(child_bs); - bdrv_replace_child_noperm(new_child, child_bs); + bdrv_replace_child_noperm(&new_child, child_bs); *child = new_child; @@ -XXX,XX +XXX,XX @@ static int bdrv_attach_child_noperm(BlockDriverState *parent_bs, return 0; } -static void bdrv_detach_child(BdrvChild *child) +static void bdrv_detach_child(BdrvChild **childp) { - BlockDriverState *old_bs = child->bs; + BlockDriverState *old_bs = (*childp)->bs; - bdrv_replace_child_noperm(child, NULL); - bdrv_child_free(child); + bdrv_replace_child_noperm(childp, NULL); + bdrv_child_free(*childp); if (old_bs) { /* @@ -XXX,XX +XXX,XX @@ void bdrv_root_unref_child(BdrvChild *child) BlockDriverState *child_bs; child_bs = child->bs; - bdrv_detach_child(child); + bdrv_detach_child(&child); bdrv_unref(child_bs); } -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> As of a future patch, bdrv_replace_child_tran() will take a BdrvChild ** pointer. Prepare for that by getting such a pointer and using it where applicable, and (dereferenced) as a parameter for bdrv_replace_child_tran(). Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-7-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, BdrvChild *child, Transaction *tran) { + BdrvChild **childp; BdrvRemoveFilterOrCowChild *s; - assert(child == bs->backing || child == bs->file); - if (!child) { return; } + if (child == bs->backing) { + childp = &bs->backing; + } else if (child == bs->file) { + childp = &bs->file; + } else { + g_assert_not_reached(); + } + if (child->bs) { - bdrv_replace_child_tran(child, NULL, tran); + bdrv_replace_child_tran(*childp, NULL, tran); } s = g_new(BdrvRemoveFilterOrCowChild, 1); *s = (BdrvRemoveFilterOrCowChild) { .child = child, - .is_backing = (child == bs->backing), + .is_backing = (childp == &bs->backing), }; tran_add(tran, &bdrv_remove_filter_or_cow_child_drv, s); - if (s->is_backing) { - bs->backing = NULL; - } else { - bs->file = NULL; - } + *childp = NULL; } /* -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> Invoke the transaction drivers' .clean() methods only after all .commit() or .abort() handlers are done. This makes it easier to have nested transactions where the top-level transactions pass objects to lower transactions that the latter can still use throughout their commit/abort phases, while the top-level transaction keeps a reference that is released in its .clean() method. (Before this commit, that is also possible, but the top-level transaction would need to take care to invoke tran_add() before the lower-level transaction does. This commit makes the ordering irrelevant, which is just a bit nicer.) Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-8-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/qemu/transactions.h | 3 +++ util/transactions.c | 8 ++++++-- 2 files changed, 9 insertions(+), 2 deletions(-) diff --git a/include/qemu/transactions.h b/include/qemu/transactions.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/transactions.h +++ b/include/qemu/transactions.h @@ -XXX,XX +XXX,XX @@ * tran_create(), call your "prepare" functions on it, and finally call * tran_abort() or tran_commit() to finalize the transaction by corresponding * finalization actions in reverse order. + * + * The clean() functions registered by the drivers in a transaction are called + * last, after all abort() or commit() functions have been called. */ #ifndef QEMU_TRANSACTIONS_H diff --git a/util/transactions.c b/util/transactions.c index XXXXXXX..XXXXXXX 100644 --- a/util/transactions.c +++ b/util/transactions.c @@ -XXX,XX +XXX,XX @@ void tran_abort(Transaction *tran) { TransactionAction *act, *next; - QSLIST_FOREACH_SAFE(act, &tran->actions, entry, next) { + QSLIST_FOREACH(act, &tran->actions, entry) { if (act->drv->abort) { act->drv->abort(act->opaque); } + } + QSLIST_FOREACH_SAFE(act, &tran->actions, entry, next) { if (act->drv->clean) { act->drv->clean(act->opaque); } @@ -XXX,XX +XXX,XX @@ void tran_commit(Transaction *tran) { TransactionAction *act, *next; - QSLIST_FOREACH_SAFE(act, &tran->actions, entry, next) { + QSLIST_FOREACH(act, &tran->actions, entry) { if (act->drv->commit) { act->drv->commit(act->opaque); } + } + QSLIST_FOREACH_SAFE(act, &tran->actions, entry, next) { if (act->drv->clean) { act->drv->clean(act->opaque); } -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> As of a future commit, bdrv_replace_child_noperm() will clear the indirect BdrvChild pointer passed to it if the new child BDS is NULL. bdrv_replace_child_tran() will want to let it do that, but revert this change in its abort handler. For that, we need to have it receive a BdrvChild ** pointer, too, and keep it stored in the BdrvReplaceChildState object that we attach to the transaction. Note that we do not need to store it in the BdrvReplaceChildState when new_bs is not NULL, because then there is nothing to revert. This is important so that bdrv_replace_node_noperm() can pass a pointer to a loop-local variable to bdrv_replace_child_tran() without worrying that this pointer will outlive one loop iteration. (Of course, for that to work, bdrv_replace_node_noperm() and in turn bdrv_replace_node() and its relatives may not be called with a NULL @to node. Luckily, they already are not, but now we should assert this.) bdrv_remove_file_or_backing_child() on the other hand needs to ensure that the indirect pointer it passes will stay valid for the duration of the transaction. Ensure this by keeping a strong reference to the BDS whose &bs->backing or &bs->file it passes to bdrv_replace_child_tran(), and giving up that reference only in the transaction .clean() handler. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-9-hreitz@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 73 insertions(+), 10 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static int bdrv_drv_set_perm(BlockDriverState *bs, uint64_t perm, typedef struct BdrvReplaceChildState { BdrvChild *child; + BdrvChild **childp; BlockDriverState *old_bs; } BdrvReplaceChildState; @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_abort(void *opaque) BdrvReplaceChildState *s = opaque; BlockDriverState *new_bs = s->child->bs; - /* old_bs reference is transparently moved from @s to @s->child */ + /* + * old_bs reference is transparently moved from @s to s->child. + * + * Pass &s->child here instead of s->childp, because: + * (1) s->old_bs must be non-NULL, so bdrv_replace_child_noperm() will not + * modify the BdrvChild * pointer we indirectly pass to it, i.e. it + * will not modify s->child. From that perspective, it does not matter + * whether we pass s->childp or &s->child. + * (TODO: Right now, bdrv_replace_child_noperm() never modifies that + * pointer anyway (though it will in the future), so at this point it + * absolutely does not matter whether we pass s->childp or &s->child.) + * (2) If new_bs is not NULL, s->childp will be NULL. We then cannot use + * it here. + * (3) If new_bs is NULL, *s->childp will have been NULLed by + * bdrv_replace_child_tran()'s bdrv_replace_child_noperm() call, and we + * must not pass a NULL *s->childp here. + * (TODO: In its current state, bdrv_replace_child_noperm() will not + * have NULLed *s->childp, so this does not apply yet. It will in the + * future.) + * + * So whether new_bs was NULL or not, we cannot pass s->childp here; and in + * any case, there is no reason to pass it anyway. + */ bdrv_replace_child_noperm(&s->child, s->old_bs); bdrv_unref(new_bs); } @@ -XXX,XX +XXX,XX @@ static TransactionActionDrv bdrv_replace_child_drv = { * Note: real unref of old_bs is done only on commit. * * The function doesn't update permissions, caller is responsible for this. + * + * Note that if new_bs == NULL, @childp is stored in a state object attached + * to @tran, so that the old child can be reinstated in the abort handler. + * Therefore, if @new_bs can be NULL, @childp must stay valid until the + * transaction is committed or aborted. + * + * (TODO: The reinstating does not happen yet, but it will once + * bdrv_replace_child_noperm() NULLs *childp when new_bs is NULL.) */ -static void bdrv_replace_child_tran(BdrvChild *child, BlockDriverState *new_bs, +static void bdrv_replace_child_tran(BdrvChild **childp, + BlockDriverState *new_bs, Transaction *tran) { BdrvReplaceChildState *s = g_new(BdrvReplaceChildState, 1); *s = (BdrvReplaceChildState) { - .child = child, - .old_bs = child->bs, + .child = *childp, + .childp = new_bs == NULL ? childp : NULL, + .old_bs = (*childp)->bs, }; tran_add(tran, &bdrv_replace_child_drv, s); if (new_bs) { bdrv_ref(new_bs); } - bdrv_replace_child_noperm(&child, new_bs); - /* old_bs reference is transparently moved from @child to @s */ + bdrv_replace_child_noperm(childp, new_bs); + /* old_bs reference is transparently moved from *childp to @s */ } /* @@ -XXX,XX +XXX,XX @@ static bool should_update_child(BdrvChild *c, BlockDriverState *to) typedef struct BdrvRemoveFilterOrCowChild { BdrvChild *child; + BlockDriverState *bs; bool is_backing; } BdrvRemoveFilterOrCowChild; @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_filter_or_cow_child_commit(void *opaque) bdrv_child_free(s->child); } +static void bdrv_remove_filter_or_cow_child_clean(void *opaque) +{ + BdrvRemoveFilterOrCowChild *s = opaque; + + /* Drop the bs reference after the transaction is done */ + bdrv_unref(s->bs); + g_free(s); +} + static TransactionActionDrv bdrv_remove_filter_or_cow_child_drv = { .abort = bdrv_remove_filter_or_cow_child_abort, .commit = bdrv_remove_filter_or_cow_child_commit, - .clean = g_free, + .clean = bdrv_remove_filter_or_cow_child_clean, }; /* @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, return; } + /* + * Keep a reference to @bs so @childp will stay valid throughout the + * transaction (required by bdrv_replace_child_tran()) + */ + bdrv_ref(bs); if (child == bs->backing) { childp = &bs->backing; } else if (child == bs->file) { @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, } if (child->bs) { - bdrv_replace_child_tran(*childp, NULL, tran); + bdrv_replace_child_tran(childp, NULL, tran); } s = g_new(BdrvRemoveFilterOrCowChild, 1); *s = (BdrvRemoveFilterOrCowChild) { .child = child, + .bs = bs, .is_backing = (childp == &bs->backing), }; tran_add(tran, &bdrv_remove_filter_or_cow_child_drv, s); @@ -XXX,XX +XXX,XX @@ static int bdrv_replace_node_noperm(BlockDriverState *from, { BdrvChild *c, *next; + assert(to != NULL); + QLIST_FOREACH_SAFE(c, &from->parents, next_parent, next) { assert(c->bs == from); if (!should_update_child(c, to)) { @@ -XXX,XX +XXX,XX @@ static int bdrv_replace_node_noperm(BlockDriverState *from, c->name, from->node_name); return -EPERM; } - bdrv_replace_child_tran(c, to, tran); + + /* + * Passing a pointer to the local variable @c is fine here, because + * @to is not NULL, and so &c will not be attached to the transaction. + */ + bdrv_replace_child_tran(&c, to, tran); } return 0; @@ -XXX,XX +XXX,XX @@ static int bdrv_replace_node_noperm(BlockDriverState *from, * * With @detach_subchain=true @to must be in a backing chain of @from. In this * case backing link of the cow-parent of @to is removed. + * + * @to must not be NULL. */ static int bdrv_replace_node_common(BlockDriverState *from, BlockDriverState *to, @@ -XXX,XX +XXX,XX @@ static int bdrv_replace_node_common(BlockDriverState *from, BlockDriverState *to_cow_parent = NULL; int ret; + assert(to != NULL); + if (detach_subchain) { assert(bdrv_chain_contains(from, to)); assert(from != to); @@ -XXX,XX +XXX,XX @@ out: return ret; } +/** + * Replace node @from by @to (where neither may be NULL). + */ int bdrv_replace_node(BlockDriverState *from, BlockDriverState *to, Error **errp) { @@ -XXX,XX +XXX,XX @@ int bdrv_replace_child_bs(BdrvChild *child, BlockDriverState *new_bs, bdrv_drained_begin(old_bs); bdrv_drained_begin(new_bs); - bdrv_replace_child_tran(child, new_bs, tran); + bdrv_replace_child_tran(&child, new_bs, tran); found = g_hash_table_new(NULL, NULL); refresh_list = bdrv_topological_dfs(refresh_list, found, old_bs); -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> In most of the block layer, especially when traversing down from other BlockDriverStates, we assume that BdrvChild.bs can never be NULL. When it becomes NULL, it is expected that the corresponding BdrvChild pointer also becomes NULL and the BdrvChild object is freed. Therefore, once bdrv_replace_child_noperm() sets the BdrvChild.bs pointer to NULL, it should also immediately set the corresponding BdrvChild pointer (like bs->file or bs->backing) to NULL. In that context, it also makes sense for this function to free the child. Sometimes we cannot do so, though, because it is called in a transactional context where the caller might still want to reinstate the child in the abort branch (and free it only on commit), so this behavior has to remain optional. In bdrv_replace_child_tran()'s abort handler, we now rely on the fact that the BdrvChild passed to bdrv_replace_child_tran() must have had a non-NULL .bs pointer initially. Make a note of that and assert it. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-10-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block.c | 102 +++++++++++++++++++++++++++++++++++++++++++------------- 1 file changed, 79 insertions(+), 23 deletions(-) diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ static BlockDriverState *bdrv_open_inherit(const char *filename, static bool bdrv_recurse_has_child(BlockDriverState *bs, BlockDriverState *child); +static void bdrv_child_free(BdrvChild *child); static void bdrv_replace_child_noperm(BdrvChild **child, - BlockDriverState *new_bs); + BlockDriverState *new_bs, + bool free_empty_child); static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, BdrvChild *child, Transaction *tran); @@ -XXX,XX +XXX,XX @@ typedef struct BdrvReplaceChildState { BdrvChild *child; BdrvChild **childp; BlockDriverState *old_bs; + bool free_empty_child; } BdrvReplaceChildState; static void bdrv_replace_child_commit(void *opaque) { BdrvReplaceChildState *s = opaque; + if (s->free_empty_child && !s->child->bs) { + bdrv_child_free(s->child); + } bdrv_unref(s->old_bs); } @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_abort(void *opaque) * modify the BdrvChild * pointer we indirectly pass to it, i.e. it * will not modify s->child. From that perspective, it does not matter * whether we pass s->childp or &s->child. - * (TODO: Right now, bdrv_replace_child_noperm() never modifies that - * pointer anyway (though it will in the future), so at this point it - * absolutely does not matter whether we pass s->childp or &s->child.) * (2) If new_bs is not NULL, s->childp will be NULL. We then cannot use * it here. * (3) If new_bs is NULL, *s->childp will have been NULLed by * bdrv_replace_child_tran()'s bdrv_replace_child_noperm() call, and we * must not pass a NULL *s->childp here. - * (TODO: In its current state, bdrv_replace_child_noperm() will not - * have NULLed *s->childp, so this does not apply yet. It will in the - * future.) * * So whether new_bs was NULL or not, we cannot pass s->childp here; and in * any case, there is no reason to pass it anyway. */ - bdrv_replace_child_noperm(&s->child, s->old_bs); + bdrv_replace_child_noperm(&s->child, s->old_bs, true); + /* + * The child was pre-existing, so s->old_bs must be non-NULL, and + * s->child thus must not have been freed + */ + assert(s->child != NULL); + if (!new_bs) { + /* As described above, *s->childp was cleared, so restore it */ + assert(s->childp != NULL); + *s->childp = s->child; + } bdrv_unref(new_bs); } @@ -XXX,XX +XXX,XX @@ static TransactionActionDrv bdrv_replace_child_drv = { * * The function doesn't update permissions, caller is responsible for this. * + * (*childp)->bs must not be NULL. + * * Note that if new_bs == NULL, @childp is stored in a state object attached * to @tran, so that the old child can be reinstated in the abort handler. * Therefore, if @new_bs can be NULL, @childp must stay valid until the * transaction is committed or aborted. * - * (TODO: The reinstating does not happen yet, but it will once - * bdrv_replace_child_noperm() NULLs *childp when new_bs is NULL.) + * If @free_empty_child is true and @new_bs is NULL, the BdrvChild is + * freed (on commit). @free_empty_child should only be false if the + * caller will free the BDrvChild themselves (which may be important + * if this is in turn called in another transactional context). */ static void bdrv_replace_child_tran(BdrvChild **childp, BlockDriverState *new_bs, - Transaction *tran) + Transaction *tran, + bool free_empty_child) { BdrvReplaceChildState *s = g_new(BdrvReplaceChildState, 1); *s = (BdrvReplaceChildState) { .child = *childp, .childp = new_bs == NULL ? childp : NULL, .old_bs = (*childp)->bs, + .free_empty_child = free_empty_child, }; tran_add(tran, &bdrv_replace_child_drv, s); + /* The abort handler relies on this */ + assert(s->old_bs != NULL); + if (new_bs) { bdrv_ref(new_bs); } - bdrv_replace_child_noperm(childp, new_bs); + /* + * Pass free_empty_child=false, we will free the child (if + * necessary) in bdrv_replace_child_commit() (if our + * @free_empty_child parameter was true). + */ + bdrv_replace_child_noperm(childp, new_bs, false); /* old_bs reference is transparently moved from *childp to @s */ } @@ -XXX,XX +XXX,XX @@ uint64_t bdrv_qapi_perm_to_blk_perm(BlockPermission qapi_perm) return permissions[qapi_perm]; } +/** + * Replace (*childp)->bs by @new_bs. + * + * If @new_bs is NULL, *childp will be set to NULL, too: BDS parents + * generally cannot handle a BdrvChild with .bs == NULL, so clearing + * BdrvChild.bs should generally immediately be followed by the + * BdrvChild pointer being cleared as well. + * + * If @free_empty_child is true and @new_bs is NULL, the BdrvChild is + * freed. @free_empty_child should only be false if the caller will + * free the BdrvChild themselves (this may be important in a + * transactional context, where it may only be freed on commit). + */ static void bdrv_replace_child_noperm(BdrvChild **childp, - BlockDriverState *new_bs) + BlockDriverState *new_bs, + bool free_empty_child) { BdrvChild *child = *childp; BlockDriverState *old_bs = child->bs; @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild **childp, } child->bs = new_bs; + if (!new_bs) { + *childp = NULL; + } if (new_bs) { QLIST_INSERT_HEAD(&new_bs->parents, child, next_parent); @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_child_noperm(BdrvChild **childp, bdrv_parent_drained_end_single(child); drain_saldo++; } + + if (free_empty_child && !child->bs) { + bdrv_child_free(child); + } } /** @@ -XXX,XX +XXX,XX @@ static void bdrv_attach_child_common_abort(void *opaque) BdrvChild *child = *s->child; BlockDriverState *bs = child->bs; - bdrv_replace_child_noperm(s->child, NULL); + /* + * Pass free_empty_child=false, because we still need the child + * for the AioContext operations on the parent below; those + * BdrvChildClass methods all work on a BdrvChild object, so we + * need to keep it as an empty shell (after this function, it will + * not be attached to any parent, and it will not have a .bs). + */ + bdrv_replace_child_noperm(s->child, NULL, false); if (bdrv_get_aio_context(bs) != s->old_child_ctx) { bdrv_try_set_aio_context(bs, s->old_child_ctx, &error_abort); @@ -XXX,XX +XXX,XX @@ static void bdrv_attach_child_common_abort(void *opaque) bdrv_unref(bs); bdrv_child_free(child); - *s->child = NULL; } static TransactionActionDrv bdrv_attach_child_common_drv = { @@ -XXX,XX +XXX,XX @@ static int bdrv_attach_child_common(BlockDriverState *child_bs, } bdrv_ref(child_bs); - bdrv_replace_child_noperm(&new_child, child_bs); + bdrv_replace_child_noperm(&new_child, child_bs, true); + /* child_bs was non-NULL, so new_child must not have been freed */ + assert(new_child != NULL); *child = new_child; @@ -XXX,XX +XXX,XX @@ static void bdrv_detach_child(BdrvChild **childp) { BlockDriverState *old_bs = (*childp)->bs; - bdrv_replace_child_noperm(childp, NULL); - bdrv_child_free(*childp); + bdrv_replace_child_noperm(childp, NULL, true); if (old_bs) { /* @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, } if (child->bs) { - bdrv_replace_child_tran(childp, NULL, tran); + /* + * Pass free_empty_child=false, we will free the child in + * bdrv_remove_filter_or_cow_child_commit() + */ + bdrv_replace_child_tran(childp, NULL, tran, false); } s = g_new(BdrvRemoveFilterOrCowChild, 1); @@ -XXX,XX +XXX,XX @@ static void bdrv_remove_file_or_backing_child(BlockDriverState *bs, .is_backing = (childp == &bs->backing), }; tran_add(tran, &bdrv_remove_filter_or_cow_child_drv, s); - - *childp = NULL; } /* @@ -XXX,XX +XXX,XX @@ static int bdrv_replace_node_noperm(BlockDriverState *from, * Passing a pointer to the local variable @c is fine here, because * @to is not NULL, and so &c will not be attached to the transaction. */ - bdrv_replace_child_tran(&c, to, tran); + bdrv_replace_child_tran(&c, to, tran, true); } return 0; @@ -XXX,XX +XXX,XX @@ int bdrv_replace_child_bs(BdrvChild *child, BlockDriverState *new_bs, bdrv_drained_begin(old_bs); bdrv_drained_begin(new_bs); - bdrv_replace_child_tran(&child, new_bs, tran); + bdrv_replace_child_tran(&child, new_bs, tran, true); + /* @new_bs must have been non-NULL, so @child must not have been freed */ + assert(child != NULL); found = g_hash_table_new(NULL, NULL); refresh_list = bdrv_topological_dfs(refresh_list, found, old_bs); -- 2.31.1 From: Hanna Reitz <hreitz@redhat.com> See the comment for why this is necessary. Signed-off-by: Hanna Reitz <hreitz@redhat.com> Message-Id: <20211111120829.81329-11-hreitz@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- tests/qemu-iotests/030 | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/tests/qemu-iotests/030 b/tests/qemu-iotests/030 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/030 +++ b/tests/qemu-iotests/030 @@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase): speed=1024) self.assert_qmp(result, 'return', {}) - for job in pending_jobs: + # Do this in reverse: After unthrottling them, some jobs may finish + # before we have unthrottled all of them. This will drain their + # subgraph, and this will make jobs above them advance (despite those + # jobs on top being throttled). In the worst case, all jobs below the + # top one are finished before we can unthrottle it, and this makes it + # advance so far that it completes before we can unthrottle it - which + # results in an error. + # Starting from the top (i.e. in reverse) does not have this problem: + # When a job finishes, the ones below it are not advanced. + for job in reversed(pending_jobs): result = self.vm.qmp('block-job-set-speed', device=job, speed=0) self.assert_qmp(result, 'return', {}) -- 2.31.1 While introducing a non-QemuOpts code path for device creation for JSON -device, we noticed that QMP device_add doesn't check its input correctly (accepting arguments that should have been rejected), and that users may be relying on this behaviour (libvirt did until it was fixed recently). Let's use a deprecation period before we fix this bug in QEMU to avoid nasty surprises for users. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211111143530.18985-1-kwolf@redhat.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- docs/about/deprecated.rst | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/about/deprecated.rst +++ b/docs/about/deprecated.rst @@ -XXX,XX +XXX,XX @@ options are removed in favor of using explicit ``blockdev-create`` and ``blockdev-add`` calls. See :doc:`/interop/live-block-operations` for details. +Incorrectly typed ``device_add`` arguments (since 6.2) +'''''''''''''''''''''''''''''''''''''''''''''''''''''' + +Due to shortcomings in the internal implementation of ``device_add``, QEMU +incorrectly accepts certain invalid arguments: Any object or list arguments are +silently ignored. Other argument types are not checked, but an implicit +conversion happens, so that e.g. string values can be assigned to integer +device properties or vice versa. + +This is a bug in QEMU that will be fixed in the future so that previously +accepted incorrect commands will return an error. Users should make sure that +all arguments passed to ``device_add`` are consistent with the documented +property types. + System accelerators ------------------- -- 2.31.1 At the end of a reopen, we already call bdrv_refresh_limits(), which should update bs->request_alignment according to the new file descriptor. However, raw_probe_alignment() relies on s->needs_alignment and just uses 1 if it isn't set. We neglected to update this field, so starting with cache=writeback and then reopening with cache=none means that we get an incorrect bs->request_alignment == 1 and unaligned requests fail instead of being automatically aligned. Fix this by recalculating s->needs_alignment in raw_refresh_limits() before calling raw_probe_alignment(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20211104113109.56336-1-kwolf@redhat.com> Reviewed-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Stefano Garzarella <sgarzare@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/file-posix.c | 20 ++++++++++++++++---- tests/qemu-iotests/142 | 22 ++++++++++++++++++++++ tests/qemu-iotests/142.out | 15 +++++++++++++++ 3 files changed, 53 insertions(+), 4 deletions(-) diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ typedef struct BDRVRawState { int page_cache_inconsistent; /* errno from fdatasync failure */ bool has_fallocate; bool needs_alignment; + bool force_alignment; bool drop_cache; bool check_cache_dropped; struct { @@ -XXX,XX +XXX,XX @@ static bool dio_byte_aligned(int fd) return false; } +static bool raw_needs_alignment(BlockDriverState *bs) +{ + BDRVRawState *s = bs->opaque; + + if ((bs->open_flags & BDRV_O_NOCACHE) != 0 && !dio_byte_aligned(s->fd)) { + return true; + } + + return s->force_alignment; +} + /* Check if read is allowed with given memory buffer and length. * * This function is used to check O_DIRECT memory buffer and request alignment. @@ -XXX,XX +XXX,XX @@ static int raw_open_common(BlockDriverState *bs, QDict *options, s->has_discard = true; s->has_write_zeroes = true; - if ((bs->open_flags & BDRV_O_NOCACHE) != 0 && !dio_byte_aligned(s->fd)) { - s->needs_alignment = true; - } if (fstat(s->fd, &st) < 0) { ret = -errno; @@ -XXX,XX +XXX,XX @@ static int raw_open_common(BlockDriverState *bs, QDict *options, * so QEMU makes sure all IO operations on the device are aligned * to sector size, or else FreeBSD will reject them with EINVAL. */ - s->needs_alignment = true; + s->force_alignment = true; } #endif + s->needs_alignment = raw_needs_alignment(bs); #ifdef CONFIG_XFS if (platform_test_xfs_fd(s->fd)) { @@ -XXX,XX +XXX,XX @@ static void raw_refresh_limits(BlockDriverState *bs, Error **errp) BDRVRawState *s = bs->opaque; struct stat st; + s->needs_alignment = raw_needs_alignment(bs); raw_probe_alignment(bs, s->fd, errp); + bs->bl.min_mem_alignment = s->buf_align; bs->bl.opt_mem_alignment = MAX(s->buf_align, qemu_real_host_page_size); diff --git a/tests/qemu-iotests/142 b/tests/qemu-iotests/142 index XXXXXXX..XXXXXXX 100755 --- a/tests/qemu-iotests/142 +++ b/tests/qemu-iotests/142 @@ -XXX,XX +XXX,XX @@ info block backing-file" echo "$hmp_cmds" | run_qemu -drive "$files","$ids" | grep "Cache" +echo +echo "--- Alignment after changing O_DIRECT ---" +echo + +# Directly test the protocol level: Can unaligned requests succeed even if +# O_DIRECT was only enabled through a reopen and vice versa? + +$QEMU_IO --cache=writeback -f file $TEST_IMG <<EOF | _filter_qemu_io +read 42 42 +reopen -o cache.direct=on +read 42 42 +reopen -o cache.direct=off +read 42 42 +EOF +$QEMU_IO --cache=none -f file $TEST_IMG <<EOF | _filter_qemu_io +read 42 42 +reopen -o cache.direct=off +read 42 42 +reopen -o cache.direct=on +read 42 42 +EOF + # success, all done echo "*** done" rm -f $seq.full diff --git a/tests/qemu-iotests/142.out b/tests/qemu-iotests/142.out index XXXXXXX..XXXXXXX 100644 --- a/tests/qemu-iotests/142.out +++ b/tests/qemu-iotests/142.out @@ -XXX,XX +XXX,XX @@ cache.no-flush=on on backing-file Cache mode: writeback Cache mode: writeback, direct Cache mode: writeback, ignore flushes + +--- Alignment after changing O_DIRECT --- + +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 42/42 bytes at offset 42 +42 bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) *** done -- 2.31.1 From: Stefan Hajnoczi <stefanha@redhat.com> Reported by Coverity (CID 1465222). Fixes: 4a1d937796de0fecd8b22d7dbebf87f38e8282fd ("softmmu/qdev-monitor: add error handling in qdev_set_id") Cc: Damien Hedde <damien.hedde@greensocs.com> Cc: Kevin Wolf <kwolf@redhat.com> Cc: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20211102163342.31162-1-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Damien Hedde <damien.hedde@greensocs.com> Reviewed-by: Markus Armbruster <armbru@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- softmmu/qdev-monitor.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/softmmu/qdev-monitor.c b/softmmu/qdev-monitor.c index XXXXXXX..XXXXXXX 100644 --- a/softmmu/qdev-monitor.c +++ b/softmmu/qdev-monitor.c @@ -XXX,XX +XXX,XX @@ const char *qdev_set_id(DeviceState *dev, char *id, Error **errp) if (prop) { dev->id = id; } else { - g_free(id); error_setg(errp, "Duplicate device ID '%s'", id); + g_free(id); return NULL; } } else { -- 2.31.1 The following changes since commit 79b677d658d3d35e1e776826ac4abb28cdce69b8: Merge tag 'net-pull-request' of https://github.com/jasowang/qemu into staging (2023-02-21 11:28:31 +0000) are available in the Git repository at: https://repo.or.cz/qemu/kevin.git tags/for-upstream for you to fetch changes up to 0f385a2420d2c3f8ae7ed65fbe2712027664059e: block/rbd: Add support for layered encryption (2023-02-23 19:49:35 +0100) ---------------------------------------------------------------- Block layer patches - Lock the graph, part 2 (BlockDriver callbacks) - virtio-scsi: fix SCSIDevice hot unplug with IOThread - rbd: Add support for layered encryption ---------------------------------------------------------------- Emanuele Giuseppe Esposito (5): block/qed: add missing graph rdlock in qed_need_check_timer_entry block: Mark bdrv_co_flush() and callers GRAPH_RDLOCK block: Mark bdrv_co_pdiscard() and callers GRAPH_RDLOCK block: Mark bdrv_co_copy_range() GRAPH_RDLOCK block: Mark bdrv_co_is_inserted() and callers GRAPH_RDLOCK Kevin Wolf (18): block: Make bdrv_can_set_read_only() static mirror: Fix access of uninitialised fields during start block: Mark bdrv_co_truncate() and callers GRAPH_RDLOCK block: Mark bdrv_co_block_status() and callers GRAPH_RDLOCK block: Mark bdrv_co_ioctl() and callers GRAPH_RDLOCK block: Mark bdrv_co_pwrite_zeroes() and callers GRAPH_RDLOCK block: Mark read/write in block/io.c GRAPH_RDLOCK block: Mark public read/write functions GRAPH_RDLOCK block: Mark bdrv_co_pwrite_sync() and callers GRAPH_RDLOCK block: Mark bdrv_co_do_pwrite_zeroes() GRAPH_RDLOCK block: Mark preadv_snapshot/snapshot_block_status GRAPH_RDLOCK block: Mark bdrv_co_create() and callers GRAPH_RDLOCK block: Mark bdrv_co_io_(un)plug() and callers GRAPH_RDLOCK block: Mark bdrv_co_eject/lock_medium() and callers GRAPH_RDLOCK block: Mark bdrv_(un)register_buf() GRAPH_RDLOCK block: Mark bdrv_co_delete_file() and callers GRAPH_RDLOCK block: Mark bdrv_*_dirty_bitmap() and callers GRAPH_RDLOCK block: Mark bdrv_co_refresh_total_sectors() and callers GRAPH_RDLOCK Or Ozeri (3): block/rbd: Remove redundant stack variable passphrase_len block/rbd: Add luks-any encryption opening option block/rbd: Add support for layered encryption Stefan Hajnoczi (3): scsi: protect req->aiocb with AioContext lock dma-helpers: prevent dma_blk_cb() vs dma_aio_cancel() race virtio-scsi: reset SCSI devices from main loop thread qapi/block-core.json | 27 +++++- block/coroutines.h | 2 +- block/qcow2.h | 27 ++++-- block/qed.h | 45 +++++---- include/block/block-copy.h | 6 +- include/block/block-global-state.h | 14 +-- include/block/block-io.h | 110 ++++++++++++---------- include/block/block_int-common.h | 173 ++++++++++++++++++---------------- include/block/block_int-io.h | 53 ++++++----- include/block/dirty-bitmap.h | 12 +-- include/hw/virtio/virtio-scsi.h | 11 ++- include/sysemu/block-backend-io.h | 7 +- block.c | 12 ++- block/backup.c | 3 + block/blkdebug.c | 19 ++-- block/blklogwrites.c | 35 ++++--- block/blkreplay.c | 24 +++-- block/blkverify.c | 5 +- block/block-backend.c | 39 +++++--- block/block-copy.c | 32 ++++--- block/bochs.c | 2 +- block/commit.c | 5 +- block/copy-before-write.c | 33 ++++--- block/copy-on-read.c | 44 +++++---- block/create.c | 9 +- block/crypto.c | 16 ++-- block/dirty-bitmap.c | 2 + block/file-posix.c | 27 +++--- block/file-win32.c | 7 +- block/filter-compress.c | 36 ++++--- block/io.c | 108 +++++++++++++-------- block/iscsi.c | 28 +++--- block/mirror.c | 59 ++++++++---- block/parallels.c | 33 +++---- block/preallocate.c | 38 ++++---- block/qcow.c | 46 +++++---- block/qcow2-cluster.c | 17 ++-- block/qcow2.c | 136 +++++++++++++++------------ block/qed-check.c | 3 +- block/qed-table.c | 10 +- block/qed.c | 101 ++++++++++---------- block/quorum.c | 62 +++++++----- block/raw-format.c | 76 ++++++++------- block/rbd.c | 188 ++++++++++++++++++++++++++++++++++--- block/replication.c | 18 ++-- block/snapshot-access.c | 8 +- block/stream.c | 40 ++++---- block/throttle.c | 36 ++++--- block/vdi.c | 11 +-- block/vhdx.c | 18 ++-- block/vmdk.c | 132 ++++++++++++-------------- block/vpc.c | 11 +-- hw/scsi/scsi-disk.c | 23 +++-- hw/scsi/scsi-generic.c | 11 ++- hw/scsi/virtio-scsi.c | 169 ++++++++++++++++++++++++++------- qemu-img.c | 8 +- softmmu/dma-helpers.c | 12 ++- tests/unit/test-bdrv-drain.c | 20 ++-- tests/unit/test-block-iothread.c | 3 +- 59 files changed, 1355 insertions(+), 907 deletions(-) It is never called outside of block.c. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-2-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 2 -- block.c | 4 ++-- 2 files changed, 2 insertions(+), 4 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base, int coroutine_fn bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, int64_t bytes); -int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only, - bool ignore_allow_rdw, Error **errp); int bdrv_apply_auto_read_only(BlockDriverState *bs, const char *errmsg, Error **errp); bool bdrv_is_read_only(BlockDriverState *bs); diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ bool bdrv_is_read_only(BlockDriverState *bs) return !(bs->open_flags & BDRV_O_RDWR); } -int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only, - bool ignore_allow_rdw, Error **errp) +static int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only, + bool ignore_allow_rdw, Error **errp) { IO_CODE(); -- 2.39.2 bdrv_mirror_top_pwritev() accesses the job object when active mirroring is enabled. It disables this code during early initialisation while s->job isn't set yet. However, s->job is still set way too early when the job object isn't fully initialised. For example, &s->ops_in_flight isn't initialised yet and the in_flight bitmap doesn't exist yet. This causes crashes when a write request comes in too early. Move the assignment of s->job to when the mirror job is actually fully initialised to make sure that the mirror_top driver doesn't access it too early. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-3-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/mirror.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn mirror_run(Job *job, Error **errp) { MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job); BlockDriverState *bs = s->mirror_top_bs->backing->bs; + MirrorBDSOpaque *mirror_top_opaque = s->mirror_top_bs->opaque; BlockDriverState *target_bs = blk_bs(s->target); bool need_drain = true; BlockDeviceIoStatus iostatus; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn mirror_run(Job *job, Error **errp) } } + /* + * Only now the job is fully initialised and mirror_top_bs should start + * accessing it. + */ + mirror_top_opaque->job = s; + assert(!s->dbi); s->dbi = bdrv_dirty_iter_new(s->dirty_bitmap); for (;;) { @@ -XXX,XX +XXX,XX @@ static BlockJob *mirror_start_job( if (!s) { goto fail; } - bs_opaque->job = s; /* The block job now has a reference to this node */ bdrv_unref(mirror_top_bs); -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_truncate() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-4-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 6 +++--- include/block/block_int-common.h | 7 ++++--- block/block-backend.c | 1 + block/crypto.c | 2 +- block/io.c | 1 + block/parallels.c | 14 ++++++++------ block/preallocate.c | 2 +- block/qcow.c | 17 ++++++++++++----- block/qcow2.c | 14 ++++++++------ block/raw-format.c | 6 +++--- block/vmdk.c | 2 ++ 11 files changed, 44 insertions(+), 28 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, int64_t bytes, BdrvRequestFlags flags); -int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, - PreallocMode prealloc, BdrvRequestFlags flags, - Error **errp); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, + PreallocMode prealloc, BdrvRequestFlags flags, Error **errp); int64_t coroutine_fn bdrv_co_nb_sectors(BlockDriverState *bs); int64_t co_wrapper_mixed bdrv_nb_sectors(BlockDriverState *bs); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * If @exact is true and this function fails but would succeed * with @exact = false, it should return -ENOTSUP. */ - int coroutine_fn (*bdrv_co_truncate)(BlockDriverState *bs, int64_t offset, - bool exact, PreallocMode prealloc, - BdrvRequestFlags flags, Error **errp); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_truncate)( + BlockDriverState *bs, int64_t offset, bool exact, + PreallocMode prealloc, BdrvRequestFlags flags, Error **errp); + int64_t coroutine_fn (*bdrv_co_getlength)(BlockDriverState *bs); int64_t coroutine_fn (*bdrv_co_get_allocated_file_size)( BlockDriverState *bs); diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_truncate(BlockBackend *blk, int64_t offset, bool exact, Error **errp) { IO_OR_GS_CODE(); + GRAPH_RDLOCK_GUARD(); if (!blk_is_available(blk)) { error_setg(errp, "No medium inserted"); return -ENOMEDIUM; diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ block_crypto_co_create_generic(BlockDriverState *bs, int64_t size, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK block_crypto_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, PreallocMode prealloc, BdrvRequestFlags flags, Error **errp) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, int64_t old_size, new_bytes; int ret; IO_CODE(); + assert_bdrv_graph_readable(); /* if bs->drv == NULL, bs is closed, so there's nothing to do here */ if (!drv) { diff --git a/block/parallels.c b/block/parallels.c index XXXXXXX..XXXXXXX 100644 --- a/block/parallels.c +++ b/block/parallels.c @@ -XXX,XX +XXX,XX @@ static int64_t block_status(BDRVParallelsState *s, int64_t sector_num, return start_off; } -static coroutine_fn int64_t allocate_clusters(BlockDriverState *bs, - int64_t sector_num, - int nb_sectors, int *pnum) +static int64_t coroutine_fn GRAPH_RDLOCK +allocate_clusters(BlockDriverState *bs, int64_t sector_num, + int nb_sectors, int *pnum) { int ret = 0; BDRVParallelsState *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_writev(BlockDriverState *bs, QEMUIOVector hd_qiov; int ret = 0; + assume_graph_lock(); /* FIXME */ + qemu_iovec_init(&hd_qiov, qiov->niov); while (nb_sectors > 0) { @@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs, } -static int coroutine_fn parallels_co_check(BlockDriverState *bs, - BdrvCheckResult *res, - BdrvCheckMode fix) +static int coroutine_fn GRAPH_RDLOCK +parallels_co_check(BlockDriverState *bs, BdrvCheckResult *res, + BdrvCheckMode fix) { BDRVParallelsState *s = bs->opaque; int64_t size, prev_off, high_off; diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int preallocate_co_pwritev_part(BlockDriverState *bs, flags); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK preallocate_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, PreallocMode prealloc, BdrvRequestFlags flags, Error **errp) diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static int qcow_reopen_prepare(BDRVReopenState *state, * return 0 if not allocated, 1 if *result is assigned, and negative * errno on failure. */ -static int coroutine_fn get_cluster_offset(BlockDriverState *bs, - uint64_t offset, int allocate, - int compressed_size, - int n_start, int n_end, - uint64_t *result) +static int coroutine_fn GRAPH_RDLOCK +get_cluster_offset(BlockDriverState *bs, uint64_t offset, int allocate, + int compressed_size, int n_start, int n_end, + uint64_t *result) { BDRVQcowState *s = bs->opaque; int min_index, i, j, l1_index, l2_index, ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow_co_block_status(BlockDriverState *bs, int64_t n; uint64_t cluster_offset; + assume_graph_lock(); /* FIXME */ + qemu_co_mutex_lock(&s->lock); ret = get_cluster_offset(bs, offset, 0, 0, 0, 0, &cluster_offset); qemu_co_mutex_unlock(&s->lock); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_preadv(BlockDriverState *bs, int64_t offset, uint8_t *buf; void *orig_buf; + assume_graph_lock(); /* FIXME */ + if (qiov->niov > 1) { buf = orig_buf = qemu_try_blockalign(bs, qiov->size); if (buf == NULL) { @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_pwritev(BlockDriverState *bs, int64_t offset, uint8_t *buf; void *orig_buf; + assume_graph_lock(); /* FIXME */ + s->cluster_cache_offset = -1; /* disable compressed cache */ /* We must always copy the iov when encrypting, so we @@ -XXX,XX +XXX,XX @@ qcow_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, uint8_t *buf, *out_buf; uint64_t cluster_offset; + assume_graph_lock(); /* FIXME */ + buf = qemu_blockalign(bs, s->cluster_size); if (bytes != s->cluster_size) { if (bytes > s->cluster_size || diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int qcow2_set_up_encryption(BlockDriverState *bs, * * Returns: 0 on success, -errno on failure. */ -static int coroutine_fn preallocate_co(BlockDriverState *bs, uint64_t offset, - uint64_t new_length, PreallocMode mode, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +preallocate_co(BlockDriverState *bs, uint64_t offset, uint64_t new_length, + PreallocMode mode, Error **errp) { BDRVQcow2State *s = bs->opaque; uint64_t bytes; @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset, - bool exact, PreallocMode prealloc, - BdrvRequestFlags flags, Error **errp) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, + PreallocMode prealloc, BdrvRequestFlags flags, Error **errp) { BDRVQcow2State *s = bs->opaque; uint64_t old_length; @@ -XXX,XX +XXX,XX @@ qcow2_co_pwritev_compressed_part(BlockDriverState *bs, AioTaskPool *aio = NULL; int ret = 0; + assume_graph_lock(); /* FIXME */ + if (has_data_file(bs)) { return -ENOTSUP; } diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static void raw_refresh_limits(BlockDriverState *bs, Error **errp) } } -static int coroutine_fn raw_co_truncate(BlockDriverState *bs, int64_t offset, - bool exact, PreallocMode prealloc, - BdrvRequestFlags flags, Error **errp) +static int coroutine_fn GRAPH_RDLOCK +raw_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, + PreallocMode prealloc, BdrvRequestFlags flags, Error **errp) { BDRVRawState *s = bs->opaque; diff --git a/block/vmdk.c b/block/vmdk.c index XXXXXXX..XXXXXXX 100644 --- a/block/vmdk.c +++ b/block/vmdk.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vmdk_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov) { + assume_graph_lock(); /* FIXME */ + if (bytes == 0) { /* The caller will write bytes 0 to signal EOF. * When receive it, we align EOF to a sector boundary. */ -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_block_status() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-5-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/coroutines.h | 2 +- include/block/block-copy.h | 6 +++--- include/block/block-io.h | 22 +++++++++++----------- include/block/block_int-common.h | 3 ++- block/backup.c | 3 +++ block/block-backend.c | 2 ++ block/block-copy.c | 19 +++++++++++-------- block/io.c | 13 ++++++++----- block/mirror.c | 14 +++++++++----- block/qcow.c | 11 ++++------- block/quorum.c | 9 ++++----- block/stream.c | 32 ++++++++++++++++++-------------- qemu-img.c | 4 +++- tests/unit/test-block-iothread.c | 3 ++- 14 files changed, 81 insertions(+), 62 deletions(-) diff --git a/block/coroutines.h b/block/coroutines.h index XXXXXXX..XXXXXXX 100644 --- a/block/coroutines.h +++ b/block/coroutines.h @@ -XXX,XX +XXX,XX @@ bdrv_co_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix); int coroutine_fn GRAPH_RDLOCK bdrv_co_invalidate_cache(BlockDriverState *bs, Error **errp); -int coroutine_fn +int coroutine_fn GRAPH_RDLOCK bdrv_co_common_block_status_above(BlockDriverState *bs, BlockDriverState *base, bool include_base, diff --git a/include/block/block-copy.h b/include/block/block-copy.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-copy.h +++ b/include/block/block-copy.h @@ -XXX,XX +XXX,XX @@ void block_copy_set_progress_meter(BlockCopyState *s, ProgressMeter *pm); void block_copy_state_free(BlockCopyState *s); void block_copy_reset(BlockCopyState *s, int64_t offset, int64_t bytes); -int64_t coroutine_fn block_copy_reset_unallocated(BlockCopyState *s, - int64_t offset, - int64_t *count); + +int64_t coroutine_fn GRAPH_RDLOCK +block_copy_reset_unallocated(BlockCopyState *s, int64_t offset, int64_t *count); int coroutine_fn block_copy(BlockCopyState *s, int64_t offset, int64_t bytes, bool ignore_ratelimit, uint64_t timeout_ns, diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int bdrv_block_status(BlockDriverState *bs, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file); -int coroutine_fn bdrv_co_block_status_above(BlockDriverState *bs, - BlockDriverState *base, - int64_t offset, int64_t bytes, - int64_t *pnum, int64_t *map, - BlockDriverState **file); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_block_status_above(BlockDriverState *bs, BlockDriverState *base, + int64_t offset, int64_t bytes, int64_t *pnum, + int64_t *map, BlockDriverState **file); int bdrv_block_status_above(BlockDriverState *bs, BlockDriverState *base, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file); -int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t offset, - int64_t bytes, int64_t *pnum); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_is_allocated(BlockDriverState *bs, int64_t offset, int64_t bytes, + int64_t *pnum); int bdrv_is_allocated(BlockDriverState *bs, int64_t offset, int64_t bytes, int64_t *pnum); -int coroutine_fn bdrv_co_is_allocated_above(BlockDriverState *top, - BlockDriverState *base, - bool include_base, int64_t offset, - int64_t bytes, int64_t *pnum); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_is_allocated_above(BlockDriverState *top, BlockDriverState *base, + bool include_base, int64_t offset, int64_t bytes, + int64_t *pnum); int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base, bool include_base, int64_t offset, int64_t bytes, int64_t *pnum); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * *pnum value for the block-status cache on protocol nodes, prior * to clamping *pnum for return to its caller. */ - int coroutine_fn (*bdrv_co_block_status)(BlockDriverState *bs, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_block_status)( + BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file); diff --git a/block/backup.c b/block/backup.c index XXXXXXX..XXXXXXX 100644 --- a/block/backup.c +++ b/block/backup.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn backup_run(Job *job, Error **errp) return -ECANCELED; } + /* rdlock protects the subsequent call to bdrv_is_allocated() */ + bdrv_graph_co_rdlock(); ret = block_copy_reset_unallocated(s->bcs, offset, &count); + bdrv_graph_co_rdunlock(); if (ret < 0) { return ret; } diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_block_status_above(BlockBackend *blk, BlockDriverState **file) { IO_CODE(); + GRAPH_RDLOCK_GUARD(); return bdrv_co_block_status_above(blk_bs(blk), base, offset, bytes, pnum, map, file); } @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_is_allocated_above(BlockBackend *blk, int64_t bytes, int64_t *pnum) { IO_CODE(); + GRAPH_RDLOCK_GUARD(); return bdrv_co_is_allocated_above(blk_bs(blk), base, include_base, offset, bytes, pnum); } diff --git a/block/block-copy.c b/block/block-copy.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-copy.c +++ b/block/block-copy.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int block_copy_task_entry(AioTask *task) return ret; } -static coroutine_fn int block_copy_block_status(BlockCopyState *s, - int64_t offset, - int64_t bytes, int64_t *pnum) +static coroutine_fn GRAPH_RDLOCK +int block_copy_block_status(BlockCopyState *s, int64_t offset, int64_t bytes, + int64_t *pnum) { int64_t num; BlockDriverState *base; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int block_copy_block_status(BlockCopyState *s, * Check if the cluster starting at offset is allocated or not. * return via pnum the number of contiguous clusters sharing this allocation. */ -static int coroutine_fn block_copy_is_cluster_allocated(BlockCopyState *s, - int64_t offset, - int64_t *pnum) +static int coroutine_fn GRAPH_RDLOCK +block_copy_is_cluster_allocated(BlockCopyState *s, int64_t offset, + int64_t *pnum) { BlockDriverState *bs = s->source->bs; int64_t count, total_count = 0; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn block_copy_is_cluster_allocated(BlockCopyState *s, assert(QEMU_IS_ALIGNED(offset, s->cluster_size)); while (true) { + /* protected in backup_run() */ ret = bdrv_co_is_allocated(bs, offset, bytes, &count); if (ret < 0) { return ret; @@ -XXX,XX +XXX,XX @@ int64_t coroutine_fn block_copy_reset_unallocated(BlockCopyState *s, * Returns 1 if dirty clusters found and successfully copied, 0 if no dirty * clusters found and -errno on failure. */ -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK block_copy_dirty_clusters(BlockCopyCallState *call_state) { BlockCopyState *s = call_state->s; @@ -XXX,XX +XXX,XX @@ void block_copy_kick(BlockCopyCallState *call_state) * it means that some I/O operation failed in context of _this_ block_copy call, * not some parallel operation. */ -static int coroutine_fn block_copy_common(BlockCopyCallState *call_state) +static int coroutine_fn GRAPH_RDLOCK +block_copy_common(BlockCopyCallState *call_state) { int ret; BlockCopyState *s = call_state->s; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn block_copy_common(BlockCopyCallState *call_state) static void coroutine_fn block_copy_async_co_entry(void *opaque) { + GRAPH_RDLOCK_GUARD(); block_copy_common(opaque); } diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int bdrv_flush_all(void) * BDRV_BLOCK_OFFSET_VALID bit is set, 'map' and 'file' (if non-NULL) are * set to the host mapping and BDS corresponding to the guest offset. */ -static int coroutine_fn bdrv_co_block_status(BlockDriverState *bs, - bool want_zero, - int64_t offset, int64_t bytes, - int64_t *pnum, int64_t *map, - BlockDriverState **file) +static int coroutine_fn GRAPH_RDLOCK +bdrv_co_block_status(BlockDriverState *bs, bool want_zero, + int64_t offset, int64_t bytes, + int64_t *pnum, int64_t *map, BlockDriverState **file) { int64_t total_size; int64_t n; /* bytes */ @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_block_status(BlockDriverState *bs, bool has_filtered_child; assert(pnum); + assert_bdrv_graph_readable(); *pnum = 0; total_size = bdrv_getlength(bs); if (total_size < 0) { @@ -XXX,XX +XXX,XX @@ bdrv_co_common_block_status_above(BlockDriverState *bs, IO_CODE(); assert(!include_base || base); /* Can't include NULL base */ + assert_bdrv_graph_readable(); if (!depth) { depth = &dummy; @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, int64_t pnum = bytes; IO_CODE(); + assume_graph_lock(); /* FIXME */ + if (!bytes) { return 1; } diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s) MirrorMethod mirror_method = MIRROR_METHOD_COPY; assert(!(offset % s->granularity)); - ret = bdrv_block_status_above(source, NULL, offset, - nb_chunks * s->granularity, - &io_bytes, NULL, NULL); + WITH_GRAPH_RDLOCK_GUARD() { + ret = bdrv_block_status_above(source, NULL, offset, + nb_chunks * s->granularity, + &io_bytes, NULL, NULL); + } if (ret < 0) { io_bytes = MIN(nb_chunks * s->granularity, max_io_bytes); } else if (ret & BDRV_BLOCK_DATA) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn mirror_dirty_init(MirrorBlockJob *s) return 0; } - ret = bdrv_is_allocated_above(bs, s->base_overlay, true, offset, bytes, - &count); + WITH_GRAPH_RDLOCK_GUARD() { + ret = bdrv_is_allocated_above(bs, s->base_overlay, true, offset, + bytes, &count); + } if (ret < 0) { return ret; } diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ get_cluster_offset(BlockDriverState *bs, uint64_t offset, int allocate, return 1; } -static int coroutine_fn qcow_co_block_status(BlockDriverState *bs, - bool want_zero, - int64_t offset, int64_t bytes, - int64_t *pnum, int64_t *map, - BlockDriverState **file) +static int coroutine_fn GRAPH_RDLOCK +qcow_co_block_status(BlockDriverState *bs, bool want_zero, + int64_t offset, int64_t bytes, int64_t *pnum, + int64_t *map, BlockDriverState **file) { BDRVQcowState *s = bs->opaque; int index_in_cluster, ret; int64_t n; uint64_t cluster_offset; - assume_graph_lock(); /* FIXME */ - qemu_co_mutex_lock(&s->lock); ret = get_cluster_offset(bs, offset, 0, 0, 0, 0, &cluster_offset); qemu_co_mutex_unlock(&s->lock); diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ static void quorum_child_perm(BlockDriverState *bs, BdrvChild *c, * return BDRV_BLOCK_ZERO if *all* children agree that a certain * region contains zeroes, and BDRV_BLOCK_DATA otherwise. */ -static int coroutine_fn quorum_co_block_status(BlockDriverState *bs, - bool want_zero, - int64_t offset, int64_t count, - int64_t *pnum, int64_t *map, - BlockDriverState **file) +static int coroutine_fn GRAPH_RDLOCK +quorum_co_block_status(BlockDriverState *bs, bool want_zero, + int64_t offset, int64_t count, + int64_t *pnum, int64_t *map, BlockDriverState **file) { BDRVQuorumState *s = bs->opaque; int i, ret; diff --git a/block/stream.c b/block/stream.c index XXXXXXX..XXXXXXX 100644 --- a/block/stream.c +++ b/block/stream.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn stream_run(Job *job, Error **errp) copy = false; - ret = bdrv_is_allocated(unfiltered_bs, offset, STREAM_CHUNK, &n); - if (ret == 1) { - /* Allocated in the top, no need to copy. */ - } else if (ret >= 0) { - /* Copy if allocated in the intermediate images. Limit to the - * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE). */ - ret = bdrv_is_allocated_above(bdrv_cow_bs(unfiltered_bs), - s->base_overlay, true, - offset, n, &n); - /* Finish early if end of backing file has been reached */ - if (ret == 0 && n == 0) { - n = len - offset; + WITH_GRAPH_RDLOCK_GUARD() { + ret = bdrv_is_allocated(unfiltered_bs, offset, STREAM_CHUNK, &n); + if (ret == 1) { + /* Allocated in the top, no need to copy. */ + } else if (ret >= 0) { + /* + * Copy if allocated in the intermediate images. Limit to the + * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE). + */ + ret = bdrv_is_allocated_above(bdrv_cow_bs(unfiltered_bs), + s->base_overlay, true, + offset, n, &n); + /* Finish early if end of backing file has been reached */ + if (ret == 0 && n == 0) { + n = len - offset; + } + + copy = (ret > 0); } - - copy = (ret > 0); } trace_stream_one_iteration(s, offset, n, ret); if (copy) { diff --git a/qemu-img.c b/qemu-img.c index XXXXXXX..XXXXXXX 100644 --- a/qemu-img.c +++ b/qemu-img.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn convert_co_do_copy(void *opaque) qemu_co_mutex_unlock(&s->lock); break; } - n = convert_iteration_sectors(s, s->sector_num); + WITH_GRAPH_RDLOCK_GUARD() { + n = convert_iteration_sectors(s, s->sector_num); + } if (n < 0) { qemu_co_mutex_unlock(&s->lock); s->ret = n; diff --git a/tests/unit/test-block-iothread.c b/tests/unit/test-block-iothread.c index XXXXXXX..XXXXXXX 100644 --- a/tests/unit/test-block-iothread.c +++ b/tests/unit/test-block-iothread.c @@ -XXX,XX +XXX,XX @@ static void test_sync_op_blk_truncate(BlockBackend *blk) g_assert_cmpint(ret, ==, -EINVAL); } -static void test_sync_op_block_status(BdrvChild *c) +/* Disable TSA to make bdrv_test.bdrv_co_block_status writable */ +static void TSA_NO_TSA test_sync_op_block_status(BdrvChild *c) { int ret; int64_t n; -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_ioctl() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-6-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 3 ++- include/block/block_int-common.h | 9 +++++---- block/block-backend.c | 1 + block/io.c | 1 + block/raw-format.c | 4 ++-- 5 files changed, 11 insertions(+), 7 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ void bdrv_aio_cancel(BlockAIOCB *acb); void bdrv_aio_cancel_async(BlockAIOCB *acb); /* sg packet commands */ -int coroutine_fn bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf); /* Ensure contents are flushed to disk. */ int coroutine_fn bdrv_co_flush(BlockDriverState *bs); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { void coroutine_fn (*bdrv_co_lock_medium)(BlockDriverState *bs, bool locked); /* to control generic scsi devices */ - BlockAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs, - unsigned long int req, void *buf, + BlockAIOCB *coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_aio_ioctl)( + BlockDriverState *bs, unsigned long int req, void *buf, BlockCompletionFunc *cb, void *opaque); - int coroutine_fn (*bdrv_co_ioctl)(BlockDriverState *bs, - unsigned long int req, void *buf); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_ioctl)( + BlockDriverState *bs, unsigned long int req, void *buf); /* * Returns 0 for completed check, -errno for internal errors. diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ blk_co_do_ioctl(BlockBackend *blk, unsigned long int req, void *buf) IO_CODE(); blk_wait_while_drained(blk); + GRAPH_RDLOCK_GUARD(); if (!blk_is_available(blk)) { return -ENOMEDIUM; diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf) }; BlockAIOCB *acb; IO_CODE(); + assert_bdrv_graph_readable(); bdrv_inc_in_flight(bs); if (!drv || (!drv->bdrv_aio_ioctl && !drv->bdrv_co_ioctl)) { diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn raw_co_lock_medium(BlockDriverState *bs, bool locked) bdrv_co_lock_medium(bs->file->bs, locked); } -static int coroutine_fn raw_co_ioctl(BlockDriverState *bs, - unsigned long int req, void *buf) +static int coroutine_fn GRAPH_RDLOCK +raw_co_ioctl(BlockDriverState *bs, unsigned long int req, void *buf) { BDRVRawState *s = bs->opaque; if (s->offset || s->has_size) { -- 2.39.2 From: Emanuele Giuseppe Esposito <eesposit@redhat.com> This function is called in two different places: - timer callback, which does not take the graph rdlock. - bdrv_qed_drain_begin(), which is .bdrv_drain_begin() callback documented as function that does not take the lock. Since it calls recursive functions that traverse the graph, we need to protect them with the graph rdlock. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-7-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qed.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn qed_unplug_allocating_write_reqs(BDRVQEDState *s) qemu_co_mutex_unlock(&s->table_lock); } -static void coroutine_fn qed_need_check_timer(BDRVQEDState *s) +static void coroutine_fn GRAPH_RDLOCK qed_need_check_timer(BDRVQEDState *s) { int ret; trace_qed_need_check_timer_cb(s); + assert_bdrv_graph_readable(); if (!qed_plug_allocating_write_reqs(s)) { return; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn qed_need_check_timer(BDRVQEDState *s) static void coroutine_fn qed_need_check_timer_entry(void *opaque) { BDRVQEDState *s = opaque; + GRAPH_RDLOCK_GUARD(); qed_need_check_timer(opaque); bdrv_dec_in_flight(s->bs); -- 2.39.2 From: Emanuele Giuseppe Esposito <eesposit@redhat.com> This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_flush() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-8-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 5 +++- block/qed.h | 29 +++++++++++++-------- include/block/block-io.h | 2 +- include/block/block_int-common.h | 12 +++++---- block/blkdebug.c | 2 +- block/blklogwrites.c | 21 ++++++++++----- block/blkreplay.c | 2 +- block/blkverify.c | 2 +- block/block-backend.c | 3 ++- block/copy-before-write.c | 2 +- block/file-posix.c | 4 +-- block/io.c | 7 +++++ block/mirror.c | 2 +- block/preallocate.c | 2 +- block/qed-check.c | 3 ++- block/qed-table.c | 6 ++--- block/qed.c | 44 +++++++++++++++++++------------- block/quorum.c | 2 +- block/throttle.c | 2 +- block/vmdk.c | 6 +++-- 20 files changed, 98 insertions(+), 60 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn qcow2_detect_metadata_preallocation(BlockDriverState *bs); /* qcow2-cluster.c functions */ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size, bool exact_size); -int coroutine_fn qcow2_shrink_l1_table(BlockDriverState *bs, uint64_t max_size); + +int coroutine_fn GRAPH_RDLOCK +qcow2_shrink_l1_table(BlockDriverState *bs, uint64_t max_size); + int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index); int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num, uint8_t *buf, int nb_sectors, bool enc, Error **errp); diff --git a/block/qed.h b/block/qed.h index XXXXXXX..XXXXXXX 100644 --- a/block/qed.h +++ b/block/qed.h @@ -XXX,XX +XXX,XX @@ void qed_commit_l2_cache_entry(L2TableCache *l2_cache, CachedL2Table *l2_table); * Table I/O functions */ int coroutine_fn qed_read_l1_table_sync(BDRVQEDState *s); -int coroutine_fn qed_write_l1_table(BDRVQEDState *s, unsigned int index, - unsigned int n); -int coroutine_fn qed_write_l1_table_sync(BDRVQEDState *s, unsigned int index, - unsigned int n); + +int coroutine_fn GRAPH_RDLOCK +qed_write_l1_table(BDRVQEDState *s, unsigned int index, unsigned int n); + +int coroutine_fn GRAPH_RDLOCK +qed_write_l1_table_sync(BDRVQEDState *s, unsigned int index, unsigned int n); + int coroutine_fn qed_read_l2_table_sync(BDRVQEDState *s, QEDRequest *request, uint64_t offset); int coroutine_fn qed_read_l2_table(BDRVQEDState *s, QEDRequest *request, uint64_t offset); -int coroutine_fn qed_write_l2_table(BDRVQEDState *s, QEDRequest *request, - unsigned int index, unsigned int n, - bool flush); -int coroutine_fn qed_write_l2_table_sync(BDRVQEDState *s, QEDRequest *request, - unsigned int index, unsigned int n, - bool flush); + +int coroutine_fn GRAPH_RDLOCK +qed_write_l2_table(BDRVQEDState *s, QEDRequest *request, unsigned int index, + unsigned int n, bool flush); + +int coroutine_fn GRAPH_RDLOCK +qed_write_l2_table_sync(BDRVQEDState *s, QEDRequest *request, + unsigned int index, unsigned int n, bool flush); /** * Cluster functions @@ -XXX,XX +XXX,XX @@ int coroutine_fn qed_find_cluster(BDRVQEDState *s, QEDRequest *request, /** * Consistency check */ -int coroutine_fn qed_check(BDRVQEDState *s, BdrvCheckResult *result, bool fix); +int coroutine_fn GRAPH_RDLOCK +qed_check(BDRVQEDState *s, BdrvCheckResult *result, bool fix); + QEDTable *qed_alloc_table(BDRVQEDState *s); diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn GRAPH_RDLOCK bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf); /* Ensure contents are flushed to disk. */ -int coroutine_fn bdrv_co_flush(BlockDriverState *bs); +int coroutine_fn GRAPH_RDLOCK bdrv_co_flush(BlockDriverState *bs); int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int64_t bytes); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockAIOCB *(*bdrv_aio_pwritev)(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags, BlockCompletionFunc *cb, void *opaque); - BlockAIOCB *(*bdrv_aio_flush)(BlockDriverState *bs, - BlockCompletionFunc *cb, void *opaque); + BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_flush)( + BlockDriverState *bs, BlockCompletionFunc *cb, void *opaque); BlockAIOCB *(*bdrv_aio_pdiscard)(BlockDriverState *bs, int64_t offset, int bytes, BlockCompletionFunc *cb, void *opaque); @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * layers, if needed. This function is needed for deterministic * synchronization of the flush finishing callback. */ - int coroutine_fn (*bdrv_co_flush)(BlockDriverState *bs); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_flush)(BlockDriverState *bs); /* Delete a created file. */ int coroutine_fn (*bdrv_co_delete_file)(BlockDriverState *bs, @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * Flushes all data that was already written to the OS all the way down to * the disk (for example file-posix.c calls fsync()). */ - int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_flush_to_disk)( + BlockDriverState *bs); /* * Flushes all internal caches to the OS. The data may still sit in a * writeback cache of the host OS, but it will survive a crash of the qemu * process. */ - int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_flush_to_os)( + BlockDriverState *bs); /* * Truncate @bs to @offset bytes using the given @prealloc mode diff --git a/block/blkdebug.c b/block/blkdebug.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkdebug.c +++ b/block/blkdebug.c @@ -XXX,XX +XXX,XX @@ blkdebug_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, return bdrv_co_pwritev(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn blkdebug_co_flush(BlockDriverState *bs) +static int GRAPH_RDLOCK coroutine_fn blkdebug_co_flush(BlockDriverState *bs) { int err = rule_check(bs, 0, 0, BLKDEBUG_IO_TYPE_FLUSH); diff --git a/block/blklogwrites.c b/block/blklogwrites.c index XXXXXXX..XXXXXXX 100644 --- a/block/blklogwrites.c +++ b/block/blklogwrites.c @@ -XXX,XX +XXX,XX @@ typedef struct BlkLogWritesFileReq { uint64_t bytes; int file_flags; QEMUIOVector *qiov; - int (*func)(struct BlkLogWritesFileReq *r); + int GRAPH_RDLOCK_PTR (*func)(struct BlkLogWritesFileReq *r); int file_ret; } BlkLogWritesFileReq; @@ -XXX,XX +XXX,XX @@ typedef struct { int log_ret; } BlkLogWritesLogReq; -static void coroutine_fn blk_log_writes_co_do_log(BlkLogWritesLogReq *lr) +static void coroutine_fn GRAPH_RDLOCK +blk_log_writes_co_do_log(BlkLogWritesLogReq *lr) { BDRVBlkLogWritesState *s = lr->bs->opaque; uint64_t cur_log_offset = s->cur_log_sector << s->sectorbits; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn blk_log_writes_co_do_log(BlkLogWritesLogReq *lr) } } -static void coroutine_fn blk_log_writes_co_do_file(BlkLogWritesFileReq *fr) +static void coroutine_fn GRAPH_RDLOCK +blk_log_writes_co_do_file(BlkLogWritesFileReq *fr) { fr->file_ret = fr->func(fr); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_log(BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, int flags, - int (*file_func)(BlkLogWritesFileReq *r), + int /*GRAPH_RDLOCK*/ (*file_func)(BlkLogWritesFileReq *r), uint64_t entry_flags, bool is_zero_write) { QEMUIOVector log_qiov; @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_do_file_pwrite_zeroes(BlkLogWritesFileReq *fr) fr->file_flags); } -static int coroutine_fn blk_log_writes_co_do_file_flush(BlkLogWritesFileReq *fr) +static int coroutine_fn GRAPH_RDLOCK +blk_log_writes_co_do_file_flush(BlkLogWritesFileReq *fr) { return bdrv_co_flush(fr->bs->file->bs); } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blk_log_writes_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { + assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, qiov, flags, blk_log_writes_co_do_file_pwritev, 0, false); } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blk_log_writes_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvRequestFlags flags) { + assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, NULL, flags, blk_log_writes_co_do_file_pwrite_zeroes, 0, true); } -static int coroutine_fn blk_log_writes_co_flush_to_disk(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK +blk_log_writes_co_flush_to_disk(BlockDriverState *bs) { return blk_log_writes_co_log(bs, 0, 0, NULL, 0, blk_log_writes_co_do_file_flush, @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blk_log_writes_co_flush_to_disk(BlockDriverState *bs) static int coroutine_fn blk_log_writes_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { + assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, NULL, 0, blk_log_writes_co_do_file_pdiscard, LOG_DISCARD_FLAG, false); diff --git a/block/blkreplay.c b/block/blkreplay.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkreplay.c +++ b/block/blkreplay.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blkreplay_co_pdiscard(BlockDriverState *bs, return ret; } -static int coroutine_fn blkreplay_co_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK blkreplay_co_flush(BlockDriverState *bs) { uint64_t reqid = blkreplay_next_id(); int ret = bdrv_co_flush(bs->file->bs); diff --git a/block/blkverify.c b/block/blkverify.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkverify.c +++ b/block/blkverify.c @@ -XXX,XX +XXX,XX @@ blkverify_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, return blkverify_co_prwv(bs, &r, offset, bytes, qiov, qiov, flags, true); } -static int coroutine_fn blkverify_co_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK blkverify_co_flush(BlockDriverState *bs) { BDRVBlkverifyState *s = bs->opaque; diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_pdiscard(BlockBackend *blk, int64_t offset, /* To be called between exactly one pair of blk_inc/dec_in_flight() */ static int coroutine_fn blk_co_do_flush(BlockBackend *blk) { - blk_wait_while_drained(blk); IO_CODE(); + blk_wait_while_drained(blk); + GRAPH_RDLOCK_GUARD(); if (!blk_is_available(blk)) { return -ENOMEDIUM; diff --git a/block/copy-before-write.c b/block/copy-before-write.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-before-write.c +++ b/block/copy-before-write.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int cbw_co_pwritev(BlockDriverState *bs, return bdrv_co_pwritev(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn cbw_co_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK cbw_co_flush(BlockDriverState *bs) { if (!bs->file) { return 0; diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn check_cache_dropped(BlockDriverState *bs, Error **errp) } #endif /* __linux__ */ -static void coroutine_fn raw_co_invalidate_cache(BlockDriverState *bs, - Error **errp) +static void coroutine_fn GRAPH_RDLOCK +raw_co_invalidate_cache(BlockDriverState *bs, Error **errp) { BDRVRawState *s = bs->opaque; int ret; diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, int ret; IO_CODE(); + assume_graph_lock(); /* FIXME */ + ret = bdrv_co_pwrite(child, offset, bytes, buf, flags); if (ret < 0) { return ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_driver_pwritev(BlockDriverState *bs, QEMUIOVector local_qiov; int ret; + assume_graph_lock(); /* FIXME */ + bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int head = 0; int tail = 0; + assume_graph_lock(); /* FIXME */ + int64_t max_write_zeroes = MIN_NON_ZERO(bs->bl.max_pwrite_zeroes, INT64_MAX); int alignment = MAX(bs->bl.pwrite_zeroes_alignment, @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_flush(BlockDriverState *bs) int ret = 0; IO_CODE(); + assert_bdrv_graph_readable(); bdrv_inc_in_flight(bs); if (!bdrv_co_is_inserted(bs) || bdrv_is_read_only(bs) || diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_mirror_top_pwritev(BlockDriverState *bs, return ret; } -static int coroutine_fn bdrv_mirror_top_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK bdrv_mirror_top_flush(BlockDriverState *bs) { if (bs->backing == NULL) { /* we can be here after failed bdrv_append in mirror_start_job */ diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ preallocate_co_truncate(BlockDriverState *bs, int64_t offset, return 0; } -static int coroutine_fn preallocate_co_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK preallocate_co_flush(BlockDriverState *bs) { return bdrv_co_flush(bs->file->bs); } diff --git a/block/qed-check.c b/block/qed-check.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed-check.c +++ b/block/qed-check.c @@ -XXX,XX +XXX,XX @@ static unsigned int qed_check_l2_table(QEDCheck *check, QEDTable *table) /** * Descend tables and check each cluster is referenced once only */ -static int coroutine_fn qed_check_l1_table(QEDCheck *check, QEDTable *table) +static int coroutine_fn GRAPH_RDLOCK +qed_check_l1_table(QEDCheck *check, QEDTable *table) { BDRVQEDState *s = check->s; unsigned int i, num_invalid_l1 = 0; diff --git a/block/qed-table.c b/block/qed-table.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed-table.c +++ b/block/qed-table.c @@ -XXX,XX +XXX,XX @@ out: * * Called with table_lock held. */ -static int coroutine_fn qed_write_table(BDRVQEDState *s, uint64_t offset, - QEDTable *table, unsigned int index, - unsigned int n, bool flush) +static int coroutine_fn GRAPH_RDLOCK +qed_write_table(BDRVQEDState *s, uint64_t offset, QEDTable *table, + unsigned int index, unsigned int n, bool flush) { unsigned int sector_mask = BDRV_SECTOR_SIZE / sizeof(uint64_t) - 1; unsigned int start, end, i; diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ static void bdrv_qed_init_state(BlockDriverState *bs) } /* Called with table_lock held. */ -static int coroutine_fn bdrv_qed_do_open(BlockDriverState *bs, QDict *options, - int flags, Error **errp) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_do_open(BlockDriverState *bs, QDict *options, int flags, Error **errp) { BDRVQEDState *s = bs->opaque; QEDHeader le_header; @@ -XXX,XX +XXX,XX @@ typedef struct QEDOpenCo { int ret; } QEDOpenCo; -static void coroutine_fn bdrv_qed_open_entry(void *opaque) +static void coroutine_fn GRAPH_RDLOCK bdrv_qed_open_entry(void *opaque) { QEDOpenCo *qoc = opaque; BDRVQEDState *s = qoc->bs->opaque; @@ -XXX,XX +XXX,XX @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags, }; int ret; + assume_graph_lock(); /* FIXME */ + ret = bdrv_open_file_child(NULL, options, "file", bs, errp); if (ret < 0) { return ret; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn qed_aio_complete(QEDAIOCB *acb) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_l1_update(QEDAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK qed_aio_write_l1_update(QEDAIOCB *acb) { BDRVQEDState *s = acb_to_s(acb); CachedL2Table *l2_table = acb->request.l2_table; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qed_aio_write_l1_update(QEDAIOCB *acb) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_l2_update(QEDAIOCB *acb, uint64_t offset) +static int coroutine_fn GRAPH_RDLOCK +qed_aio_write_l2_update(QEDAIOCB *acb, uint64_t offset) { BDRVQEDState *s = acb_to_s(acb); bool need_alloc = acb->find_cluster_ret == QED_CLUSTER_L1; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qed_aio_write_main(QEDAIOCB *acb) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_cow(QEDAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK qed_aio_write_cow(QEDAIOCB *acb) { BDRVQEDState *s = acb_to_s(acb); uint64_t start, len, offset; @@ -XXX,XX +XXX,XX @@ static bool qed_should_set_need_check(BDRVQEDState *s) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_alloc(QEDAIOCB *acb, size_t len) +static int coroutine_fn GRAPH_RDLOCK +qed_aio_write_alloc(QEDAIOCB *acb, size_t len) { BDRVQEDState *s = acb_to_s(acb); int ret; @@ -XXX,XX +XXX,XX @@ out: * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_data(void *opaque, int ret, - uint64_t offset, size_t len) +static int coroutine_fn GRAPH_RDLOCK +qed_aio_write_data(void *opaque, int ret, uint64_t offset, size_t len) { QEDAIOCB *acb = opaque; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qed_aio_read_data(void *opaque, int ret, /** * Begin next I/O or complete the request */ -static int coroutine_fn qed_aio_next_io(QEDAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK qed_aio_next_io(QEDAIOCB *acb) { BDRVQEDState *s = acb_to_s(acb); uint64_t offset; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qed_aio_next_io(QEDAIOCB *acb) return ret; } -static int coroutine_fn qed_co_request(BlockDriverState *bs, int64_t sector_num, - QEMUIOVector *qiov, int nb_sectors, - int flags) +static int coroutine_fn GRAPH_RDLOCK +qed_co_request(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov, + int nb_sectors, int flags) { QEDAIOCB acb = { .bs = bs, @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_qed_co_readv(BlockDriverState *bs, int64_t sector_num, int nb_sectors, QEMUIOVector *qiov) { + assume_graph_lock(); /* FIXME */ return qed_co_request(bs, sector_num, qiov, nb_sectors, 0); } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_qed_co_writev(BlockDriverState *bs, int64_t sector_num, int nb_sectors, QEMUIOVector *qiov, int flags) { + assume_graph_lock(); /* FIXME */ return qed_co_request(bs, sector_num, qiov, nb_sectors, QED_AIOCB_WRITE); } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_qed_co_pwrite_zeroes(BlockDriverState *bs, { BDRVQEDState *s = bs->opaque; + assume_graph_lock(); /* FIXME */ + /* * Zero writes start without an I/O buffer. If a buffer becomes necessary * then it will be allocated during request processing. @@ -XXX,XX +XXX,XX @@ static int bdrv_qed_change_backing_file(BlockDriverState *bs, return ret; } -static void coroutine_fn bdrv_qed_co_invalidate_cache(BlockDriverState *bs, - Error **errp) +static void coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_invalidate_cache(BlockDriverState *bs, Error **errp) { BDRVQEDState *s = bs->opaque; int ret; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_qed_co_invalidate_cache(BlockDriverState *bs, } } -static int coroutine_fn bdrv_qed_co_check(BlockDriverState *bs, - BdrvCheckResult *result, - BdrvCheckMode fix) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_check(BlockDriverState *bs, BdrvCheckResult *result, + BdrvCheckMode fix) { BDRVQEDState *s = bs->opaque; int ret; diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ static int64_t coroutine_fn quorum_co_getlength(BlockDriverState *bs) return result; } -static coroutine_fn int quorum_co_flush(BlockDriverState *bs) +static coroutine_fn GRAPH_RDLOCK int quorum_co_flush(BlockDriverState *bs) { BDRVQuorumState *s = bs->opaque; QuorumVoteVersion *winner = NULL; diff --git a/block/throttle.c b/block/throttle.c index XXXXXXX..XXXXXXX 100644 --- a/block/throttle.c +++ b/block/throttle.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn throttle_co_pwritev_compressed(BlockDriverState *bs, BDRV_REQ_WRITE_COMPRESSED); } -static int coroutine_fn throttle_co_flush(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK throttle_co_flush(BlockDriverState *bs) { return bdrv_co_flush(bs->file->bs); } diff --git a/block/vmdk.c b/block/vmdk.c index XXXXXXX..XXXXXXX 100644 --- a/block/vmdk.c +++ b/block/vmdk.c @@ -XXX,XX +XXX,XX @@ exit: return ret; } -static int coroutine_fn vmdk_L2update(VmdkExtent *extent, VmdkMetaData *m_data, - uint32_t offset) +static int coroutine_fn GRAPH_RDLOCK +vmdk_L2update(VmdkExtent *extent, VmdkMetaData *m_data, uint32_t offset) { offset = cpu_to_le32(offset); /* update L2 table */ @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vmdk_pwritev(BlockDriverState *bs, uint64_t offset, uint64_t bytes_done = 0; VmdkMetaData m_data; + assume_graph_lock(); /* FIXME */ + if (DIV_ROUND_UP(offset, BDRV_SECTOR_SIZE) > bs->total_sectors) { error_report("Wrong offset: offset=0x%" PRIx64 " total_sectors=0x%" PRIx64, -- 2.39.2 From: Emanuele Giuseppe Esposito <eesposit@redhat.com> This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_pdiscard() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-9-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 5 +++-- include/block/block_int-common.h | 15 +++++++++------ include/block/block_int-io.h | 2 +- block/blkdebug.c | 4 ++-- block/blklogwrites.c | 5 ++--- block/blkreplay.c | 4 ++-- block/block-backend.c | 1 + block/copy-before-write.c | 8 ++++---- block/copy-on-read.c | 4 ++-- block/filter-compress.c | 4 ++-- block/io.c | 2 ++ block/mirror.c | 14 +++++++++----- block/preallocate.c | 4 ++-- block/raw-format.c | 4 ++-- block/snapshot-access.c | 4 ++-- block/throttle.c | 4 ++-- 16 files changed, 47 insertions(+), 37 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ bdrv_co_ioctl(BlockDriverState *bs, int req, void *buf); /* Ensure contents are flushed to disk. */ int coroutine_fn GRAPH_RDLOCK bdrv_co_flush(BlockDriverState *bs); -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, - int64_t bytes); +int coroutine_fn GRAPH_RDLOCK bdrv_co_pdiscard(BdrvChild *child, int64_t offset, + int64_t bytes); + bool bdrv_can_write_zeroes_with_unmap(BlockDriverState *bs); int bdrv_block_status(BlockDriverState *bs, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BdrvRequestFlags flags, BlockCompletionFunc *cb, void *opaque); BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_flush)( BlockDriverState *bs, BlockCompletionFunc *cb, void *opaque); - BlockAIOCB *(*bdrv_aio_pdiscard)(BlockDriverState *bs, - int64_t offset, int bytes, + + BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_pdiscard)( + BlockDriverState *bs, int64_t offset, int bytes, BlockCompletionFunc *cb, void *opaque); int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs, @@ -XXX,XX +XXX,XX @@ struct BlockDriver { */ int coroutine_fn (*bdrv_co_pwrite_zeroes)(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvRequestFlags flags); - int coroutine_fn (*bdrv_co_pdiscard)(BlockDriverState *bs, - int64_t offset, int64_t bytes); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pdiscard)( + BlockDriverState *bs, int64_t offset, int64_t bytes); /* * Map [offset, offset + nbytes) range onto a child of @bs to copy from, @@ -XXX,XX +XXX,XX @@ struct BlockDriver { int coroutine_fn (*bdrv_co_snapshot_block_status)(BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file); - int coroutine_fn (*bdrv_co_pdiscard_snapshot)(BlockDriverState *bs, - int64_t offset, int64_t bytes); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pdiscard_snapshot)( + BlockDriverState *bs, int64_t offset, int64_t bytes); /* * Invalidate any cached meta-data. diff --git a/include/block/block_int-io.h b/include/block/block_int-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-io.h +++ b/include/block/block_int-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_preadv_snapshot(BdrvChild *child, int coroutine_fn bdrv_co_snapshot_block_status(BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file); -int coroutine_fn bdrv_co_pdiscard_snapshot(BlockDriverState *bs, +int coroutine_fn GRAPH_RDLOCK bdrv_co_pdiscard_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes); diff --git a/block/blkdebug.c b/block/blkdebug.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkdebug.c +++ b/block/blkdebug.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blkdebug_co_pwrite_zeroes(BlockDriverState *bs, return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } -static int coroutine_fn blkdebug_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +blkdebug_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { uint32_t align = bs->bl.pdiscard_alignment; int err; diff --git a/block/blklogwrites.c b/block/blklogwrites.c index XXXXXXX..XXXXXXX 100644 --- a/block/blklogwrites.c +++ b/block/blklogwrites.c @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_do_file_flush(BlkLogWritesFileReq *fr) return bdrv_co_flush(fr->bs->file->bs); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_do_file_pdiscard(BlkLogWritesFileReq *fr) { return bdrv_co_pdiscard(fr->bs->file, fr->offset, fr->bytes); @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_flush_to_disk(BlockDriverState *bs) LOG_FLUSH_FLAG, false); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { - assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, NULL, 0, blk_log_writes_co_do_file_pdiscard, LOG_DISCARD_FLAG, false); diff --git a/block/blkreplay.c b/block/blkreplay.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkreplay.c +++ b/block/blkreplay.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blkreplay_co_pwrite_zeroes(BlockDriverState *bs, return ret; } -static int coroutine_fn blkreplay_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +blkreplay_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { uint64_t reqid = blkreplay_next_id(); int ret = bdrv_co_pdiscard(bs->file, offset, bytes); diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ blk_co_do_pdiscard(BlockBackend *blk, int64_t offset, int64_t bytes) IO_CODE(); blk_wait_while_drained(blk); + GRAPH_RDLOCK_GUARD(); ret = blk_check_byte_request(blk, offset, bytes); if (ret < 0) { diff --git a/block/copy-before-write.c b/block/copy-before-write.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-before-write.c +++ b/block/copy-before-write.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int cbw_do_copy_before_write(BlockDriverState *bs, return 0; } -static int coroutine_fn cbw_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +cbw_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { int ret = cbw_do_copy_before_write(bs, offset, bytes, 0); if (ret < 0) { @@ -XXX,XX +XXX,XX @@ cbw_co_snapshot_block_status(BlockDriverState *bs, return ret; } -static int coroutine_fn cbw_co_pdiscard_snapshot(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +cbw_co_pdiscard_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes) { BDRVCopyBeforeWriteState *s = bs->opaque; diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn cor_co_pwrite_zeroes(BlockDriverState *bs, } -static int coroutine_fn cor_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +cor_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { return bdrv_co_pdiscard(bs->file, offset, bytes); } diff --git a/block/filter-compress.c b/block/filter-compress.c index XXXXXXX..XXXXXXX 100644 --- a/block/filter-compress.c +++ b/block/filter-compress.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn compress_co_pwrite_zeroes(BlockDriverState *bs, } -static int coroutine_fn compress_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +compress_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { return bdrv_co_pdiscard(bs->file, offset, bytes); } diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int head, tail, align; BlockDriverState *bs = child->bs; IO_CODE(); + assert_bdrv_graph_readable(); if (!bs || !bs->drv || !bdrv_co_is_inserted(bs)) { return -ENOMEDIUM; @@ -XXX,XX +XXX,XX @@ bdrv_co_pdiscard_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes) BlockDriver *drv = bs->drv; int ret; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) { return -ENOMEDIUM; diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_mirror_top_preadv(BlockDriverState *bs, return bdrv_co_preadv(bs->backing, offset, bytes, qiov, flags); } -static int coroutine_fn bdrv_mirror_top_do_write(BlockDriverState *bs, - MirrorMethod method, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, - int flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_mirror_top_do_write(BlockDriverState *bs, MirrorMethod method, + uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, + int flags) { MirrorOp *op = NULL; MirrorBDSOpaque *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_mirror_top_pwritev(BlockDriverState *bs, int ret = 0; bool copy_to_target = false; + assume_graph_lock(); /* FIXME */ + if (s->job) { copy_to_target = s->job->ret >= 0 && !job_is_cancelled(&s->job->common.job) && @@ -XXX,XX +XXX,XX @@ static int coroutine_fn GRAPH_RDLOCK bdrv_mirror_top_flush(BlockDriverState *bs) static int coroutine_fn bdrv_mirror_top_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvRequestFlags flags) { + assume_graph_lock(); /* FIXME */ return bdrv_mirror_top_do_write(bs, MIRROR_METHOD_ZERO, offset, bytes, NULL, flags); } -static int coroutine_fn bdrv_mirror_top_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +bdrv_mirror_top_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { return bdrv_mirror_top_do_write(bs, MIRROR_METHOD_DISCARD, offset, bytes, NULL, 0); diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int preallocate_co_preadv_part( flags); } -static int coroutine_fn preallocate_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +preallocate_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { return bdrv_co_pdiscard(bs->file, offset, bytes); } diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_pwrite_zeroes(BlockDriverState *bs, return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } -static int coroutine_fn raw_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +raw_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { int ret; diff --git a/block/snapshot-access.c b/block/snapshot-access.c index XXXXXXX..XXXXXXX 100644 --- a/block/snapshot-access.c +++ b/block/snapshot-access.c @@ -XXX,XX +XXX,XX @@ snapshot_access_co_block_status(BlockDriverState *bs, bytes, pnum, map, file); } -static int coroutine_fn snapshot_access_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +snapshot_access_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { return bdrv_co_pdiscard_snapshot(bs->file->bs, offset, bytes); } diff --git a/block/throttle.c b/block/throttle.c index XXXXXXX..XXXXXXX 100644 --- a/block/throttle.c +++ b/block/throttle.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn throttle_co_pwrite_zeroes(BlockDriverState *bs, return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } -static int coroutine_fn throttle_co_pdiscard(BlockDriverState *bs, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +throttle_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) { ThrottleGroupMember *tgm = bs->opaque; throttle_group_co_io_limits_intercept(tgm, bytes, true); -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_pwrite_zeroes() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-10-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 6 ++++-- include/block/block-io.h | 9 +++++---- include/block/block_int-common.h | 5 +++-- block/blkdebug.c | 6 +++--- block/blklogwrites.c | 5 ++--- block/blkreplay.c | 5 +++-- block/block-copy.c | 13 +++++++------ block/copy-before-write.c | 5 +++-- block/copy-on-read.c | 6 +++--- block/filter-compress.c | 6 +++--- block/io.c | 12 +++++++++--- block/mirror.c | 6 +++--- block/preallocate.c | 11 +++++++---- block/qcow2.c | 30 ++++++++++++++++++------------ block/qed.c | 9 +++------ block/quorum.c | 15 ++++++++++----- block/raw-format.c | 6 +++--- block/throttle.c | 6 +++--- 18 files changed, 92 insertions(+), 69 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m); int qcow2_cluster_discard(BlockDriverState *bs, uint64_t offset, uint64_t bytes, enum qcow2_discard_type type, bool full_discard); -int coroutine_fn qcow2_subcluster_zeroize(BlockDriverState *bs, uint64_t offset, - uint64_t bytes, int flags); + +int coroutine_fn GRAPH_RDLOCK +qcow2_subcluster_zeroize(BlockDriverState *bs, uint64_t offset, uint64_t bytes, + int flags); int qcow2_expand_zero_clusters(BlockDriverState *bs, BlockDriverAmendStatusCB *status_cb, diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, * function is not suitable for zeroing the entire image in a single request * because it may allocate memory for the entire region. */ -int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, - int64_t bytes, BdrvRequestFlags flags); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, int64_t bytes, + BdrvRequestFlags flags); int coroutine_fn GRAPH_RDLOCK bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, @@ -XXX,XX +XXX,XX @@ int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base, bool include_base, int64_t offset, int64_t bytes, int64_t *pnum); -int coroutine_fn bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, - int64_t bytes); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, int64_t bytes); int bdrv_apply_auto_read_only(BlockDriverState *bs, const char *errmsg, Error **errp); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * function pointer may be NULL or return -ENOSUP and .bdrv_co_writev() * will be called instead. */ - int coroutine_fn (*bdrv_co_pwrite_zeroes)(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pwrite_zeroes)( + BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags); int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pdiscard)( BlockDriverState *bs, int64_t offset, int64_t bytes); diff --git a/block/blkdebug.c b/block/blkdebug.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkdebug.c +++ b/block/blkdebug.c @@ -XXX,XX +XXX,XX @@ static int GRAPH_RDLOCK coroutine_fn blkdebug_co_flush(BlockDriverState *bs) return bdrv_co_flush(bs->file->bs); } -static int coroutine_fn blkdebug_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +blkdebug_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { uint32_t align = MAX(bs->bl.request_alignment, bs->bl.pwrite_zeroes_alignment); diff --git a/block/blklogwrites.c b/block/blklogwrites.c index XXXXXXX..XXXXXXX 100644 --- a/block/blklogwrites.c +++ b/block/blklogwrites.c @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_do_file_pwritev(BlkLogWritesFileReq *fr) fr->qiov, fr->file_flags); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_do_file_pwrite_zeroes(BlkLogWritesFileReq *fr) { return bdrv_co_pwrite_zeroes(fr->bs->file, fr->offset, fr->bytes, @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, blk_log_writes_co_do_file_pwritev, 0, false); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvRequestFlags flags) { - assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, NULL, flags, blk_log_writes_co_do_file_pwrite_zeroes, 0, true); diff --git a/block/blkreplay.c b/block/blkreplay.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkreplay.c +++ b/block/blkreplay.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blkreplay_co_pwritev(BlockDriverState *bs, return ret; } -static int coroutine_fn blkreplay_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +blkreplay_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { uint64_t reqid = blkreplay_next_id(); int ret = bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); diff --git a/block/block-copy.c b/block/block-copy.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-copy.c +++ b/block/block-copy.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int block_copy_task_run(AioTaskPool *pool, * value of @method should be used for subsequent tasks. * Returns 0 on success. */ -static int coroutine_fn block_copy_do_copy(BlockCopyState *s, - int64_t offset, int64_t bytes, - BlockCopyMethod *method, - bool *error_is_read) +static int coroutine_fn GRAPH_RDLOCK +block_copy_do_copy(BlockCopyState *s, int64_t offset, int64_t bytes, + BlockCopyMethod *method, bool *error_is_read) { int ret; int64_t nbytes = MIN(offset + bytes, s->len) - offset; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int block_copy_task_entry(AioTask *task) BlockCopyMethod method = t->method; int ret; - ret = block_copy_do_copy(s, t->req.offset, t->req.bytes, &method, - &error_is_read); + WITH_GRAPH_RDLOCK_GUARD() { + ret = block_copy_do_copy(s, t->req.offset, t->req.bytes, &method, + &error_is_read); + } WITH_QEMU_LOCK_GUARD(&s->lock) { if (s->method == t->method) { diff --git a/block/copy-before-write.c b/block/copy-before-write.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-before-write.c +++ b/block/copy-before-write.c @@ -XXX,XX +XXX,XX @@ cbw_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) return bdrv_co_pdiscard(bs->file, offset, bytes); } -static int coroutine_fn cbw_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +cbw_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { int ret = cbw_do_copy_before_write(bs, offset, bytes, flags); if (ret < 0) { diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn cor_co_pwritev_part(BlockDriverState *bs, } -static int coroutine_fn cor_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +cor_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } diff --git a/block/filter-compress.c b/block/filter-compress.c index XXXXXXX..XXXXXXX 100644 --- a/block/filter-compress.c +++ b/block/filter-compress.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn compress_co_pwritev_part(BlockDriverState *bs, } -static int coroutine_fn compress_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +compress_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_copy_on_readv(BdrvChild *child, int64_t progress = 0; bool skip_write; + assume_graph_lock(); /* FIXME */ + bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, bs->bl.request_alignment); int max_transfer = MIN_NON_ZERO(bs->bl.max_transfer, MAX_BOUNCE_BUFFER); + assert_bdrv_graph_readable(); bdrv_check_request(offset, bytes, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_aligned_pwritev(BdrvChild *child, int64_t bytes_remaining = bytes; int max_transfer; + assume_graph_lock(); /* FIXME */ + bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset, { IO_CODE(); trace_bdrv_co_pwrite_zeroes(child->bs, offset, bytes, flags); + assert_bdrv_graph_readable(); if (!(child->bs->open_flags & BDRV_O_UNMAP)) { flags &= ~BDRV_REQ_MAY_UNMAP; @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_is_zero_fast(BlockDriverState *bs, int64_t offset, int64_t pnum = bytes; IO_CODE(); - assume_graph_lock(); /* FIXME */ - if (!bytes) { return 1; } @@ -XXX,XX +XXX,XX @@ void bdrv_unregister_buf(BlockDriverState *bs, void *host, size_t size) } } -static int coroutine_fn bdrv_co_copy_range_internal( +static int coroutine_fn GRAPH_RDLOCK bdrv_co_copy_range_internal( BdrvChild *src, int64_t src_offset, BdrvChild *dst, int64_t dst_offset, int64_t bytes, BdrvRequestFlags read_flags, BdrvRequestFlags write_flags, @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range_from(BdrvChild *src, int64_t src_offset, BdrvRequestFlags write_flags) { IO_CODE(); + assume_graph_lock(); /* FIXME */ trace_bdrv_co_copy_range_from(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); return bdrv_co_copy_range_internal(src, src_offset, dst, dst_offset, @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range_to(BdrvChild *src, int64_t src_offset, BdrvRequestFlags write_flags) { IO_CODE(); + assume_graph_lock(); /* FIXME */ trace_bdrv_co_copy_range_to(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); return bdrv_co_copy_range_internal(src, src_offset, dst, dst_offset, diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn GRAPH_RDLOCK bdrv_mirror_top_flush(BlockDriverState *bs) return bdrv_co_flush(bs->backing->bs); } -static int coroutine_fn bdrv_mirror_top_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_mirror_top_pwrite_zeroes(BlockDriverState *bs, int64_t offset, + int64_t bytes, BdrvRequestFlags flags) { - assume_graph_lock(); /* FIXME */ return bdrv_mirror_top_do_write(bs, MIRROR_METHOD_ZERO, offset, bytes, NULL, flags); } diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ static bool has_prealloc_perms(BlockDriverState *bs) * want_merge_zero is used to merge write-zero request with preallocation in * one bdrv_co_pwrite_zeroes() call. */ -static bool coroutine_fn handle_write(BlockDriverState *bs, int64_t offset, - int64_t bytes, bool want_merge_zero) +static bool coroutine_fn GRAPH_RDLOCK +handle_write(BlockDriverState *bs, int64_t offset, int64_t bytes, + bool want_merge_zero) { BDRVPreallocateState *s = bs->opaque; int64_t end = offset + bytes; @@ -XXX,XX +XXX,XX @@ static bool coroutine_fn handle_write(BlockDriverState *bs, int64_t offset, return want_merge_zero; } -static int coroutine_fn preallocate_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +preallocate_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, + int64_t bytes, BdrvRequestFlags flags) { bool want_merge_zero = !(flags & ~(BDRV_REQ_ZERO_WRITE | BDRV_REQ_NO_FALLBACK)); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int preallocate_co_pwritev_part(BlockDriverState *bs, size_t qiov_offset, BdrvRequestFlags flags) { + assume_graph_lock(); /* FIXME */ handle_write(bs, offset, bytes, false); return bdrv_co_pwritev_part(bs->file, offset, bytes, qiov, qiov_offset, diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static bool merge_cow(uint64_t offset, unsigned bytes, * Return 1 if the COW regions read as zeroes, 0 if not, < 0 on error. * Note that returning 0 does not guarantee non-zero data. */ -static int coroutine_fn is_zero_cow(BlockDriverState *bs, QCowL2Meta *m) +static int coroutine_fn GRAPH_RDLOCK +is_zero_cow(BlockDriverState *bs, QCowL2Meta *m) { /* * This check is designed for optimization shortcut so it must be @@ -XXX,XX +XXX,XX @@ static int coroutine_fn is_zero_cow(BlockDriverState *bs, QCowL2Meta *m) m->cow_end.nb_bytes); } -static int coroutine_fn handle_alloc_space(BlockDriverState *bs, - QCowL2Meta *l2meta) +static int coroutine_fn GRAPH_RDLOCK +handle_alloc_space(BlockDriverState *bs, QCowL2Meta *l2meta) { BDRVQcow2State *s = bs->opaque; QCowL2Meta *m; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn handle_alloc_space(BlockDriverState *bs, * l2meta - if not NULL, qcow2_co_pwritev_task() will consume it. Caller must * not use it somehow after qcow2_co_pwritev_task() call */ -static coroutine_fn int qcow2_co_pwritev_task(BlockDriverState *bs, - uint64_t host_offset, - uint64_t offset, uint64_t bytes, - QEMUIOVector *qiov, - uint64_t qiov_offset, - QCowL2Meta *l2meta) +static coroutine_fn GRAPH_RDLOCK +int qcow2_co_pwritev_task(BlockDriverState *bs, uint64_t host_offset, + uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, + uint64_t qiov_offset, QCowL2Meta *l2meta) { int ret; BDRVQcow2State *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ out_locked: return ret; } -static coroutine_fn int qcow2_co_pwritev_task_entry(AioTask *task) +/* + * This function can count as GRAPH_RDLOCK because qcow2_co_pwritev_part() holds + * the graph lock and keeps it until this coroutine has terminated. + */ +static coroutine_fn GRAPH_RDLOCK int qcow2_co_pwritev_task_entry(AioTask *task) { Qcow2AioTask *t = container_of(task, Qcow2AioTask, task); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_part( QCowL2Meta *l2meta = NULL; AioTaskPool *aio = NULL; + assume_graph_lock(); /* FIXME */ + trace_qcow2_writev_start_req(qemu_coroutine_self(), offset, bytes); while (bytes != 0 && aio_task_pool_status(aio) == 0) { @@ -XXX,XX +XXX,XX @@ static bool is_zero(BlockDriverState *bs, int64_t offset, int64_t bytes) return res >= 0 && (res & BDRV_BLOCK_ZERO) && bytes == 0; } -static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { int ret; BDRVQcow2State *s = bs->opaque; diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_qed_co_writev(BlockDriverState *bs, return qed_co_request(bs, sector_num, qiov, nb_sectors, QED_AIOCB_WRITE); } -static int coroutine_fn bdrv_qed_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { BDRVQEDState *s = bs->opaque; - assume_graph_lock(); /* FIXME */ - /* * Zero writes start without an I/O buffer. If a buffer becomes necessary * then it will be allocated during request processing. diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn quorum_co_preadv(BlockDriverState *bs, return ret; } -static void coroutine_fn write_quorum_entry(void *opaque) +/* + * This function can count as GRAPH_RDLOCK because quorum_co_pwritev() holds the + * graph lock and keeps it until this coroutine has terminated. + */ +static void coroutine_fn GRAPH_RDLOCK write_quorum_entry(void *opaque) { QuorumCo *co = opaque; QuorumAIOCB *acb = co->acb; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn quorum_co_pwritev(BlockDriverState *bs, int64_t offset, QuorumAIOCB *acb = quorum_aio_get(bs, qiov, offset, bytes, flags); int i, ret; + assume_graph_lock(); /* FIXME */ + for (i = 0; i < s->num_children; i++) { Coroutine *co; QuorumCo data = { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn quorum_co_pwritev(BlockDriverState *bs, int64_t offset, return ret; } -static int coroutine_fn quorum_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) - +static int coroutine_fn GRAPH_RDLOCK +quorum_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { return quorum_co_pwritev(bs, offset, bytes, NULL, flags | BDRV_REQ_ZERO_WRITE); diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_block_status(BlockDriverState *bs, return BDRV_BLOCK_RAW | BDRV_BLOCK_OFFSET_VALID; } -static int coroutine_fn raw_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { int ret; diff --git a/block/throttle.c b/block/throttle.c index XXXXXXX..XXXXXXX 100644 --- a/block/throttle.c +++ b/block/throttle.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn throttle_co_pwritev(BlockDriverState *bs, return bdrv_co_pwritev(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn throttle_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +throttle_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { ThrottleGroupMember *tgm = bs->opaque; throttle_group_co_io_limits_intercept(tgm, bytes, true); -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_driver_*() need to hold a reader lock for the graph. It doesn't add the annotation to public functions yet. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-11-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 5 ++- include/block/block_int-common.h | 40 ++++++++++--------- block/io.c | 66 +++++++++++++++----------------- block/parallels.c | 8 ++-- block/qcow.c | 20 ++++------ block/qcow2-cluster.c | 10 ++--- block/qcow2.c | 37 ++++++++++-------- block/qed.c | 14 +++---- block/quorum.c | 8 ++-- block/vmdk.c | 4 +- 10 files changed, 101 insertions(+), 111 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs, void qcow2_parse_compressed_l2_entry(BlockDriverState *bs, uint64_t l2_entry, uint64_t *coffset, int *csize); -int coroutine_fn qcow2_alloc_cluster_link_l2(BlockDriverState *bs, - QCowL2Meta *m); +int coroutine_fn GRAPH_RDLOCK +qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m); + void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m); int qcow2_cluster_discard(BlockDriverState *bs, uint64_t offset, uint64_t bytes, enum qcow2_discard_type type, diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { Error **errp); /* aio */ - BlockAIOCB *(*bdrv_aio_preadv)(BlockDriverState *bs, + BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_preadv)(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags, BlockCompletionFunc *cb, void *opaque); - BlockAIOCB *(*bdrv_aio_pwritev)(BlockDriverState *bs, + + BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_pwritev)(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags, BlockCompletionFunc *cb, void *opaque); + BlockAIOCB * GRAPH_RDLOCK_PTR (*bdrv_aio_flush)( BlockDriverState *bs, BlockCompletionFunc *cb, void *opaque); @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockDriverState *bs, int64_t offset, int bytes, BlockCompletionFunc *cb, void *opaque); - int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_readv)(BlockDriverState *bs, int64_t sector_num, int nb_sectors, QEMUIOVector *qiov); /** @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * * The buffer in @qiov may point directly to guest memory. */ - int coroutine_fn (*bdrv_co_preadv)(BlockDriverState *bs, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_preadv)(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags); - int coroutine_fn (*bdrv_co_preadv_part)(BlockDriverState *bs, - int64_t offset, int64_t bytes, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_preadv_part)( + BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, BdrvRequestFlags flags); - int coroutine_fn (*bdrv_co_writev)(BlockDriverState *bs, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_writev)(BlockDriverState *bs, int64_t sector_num, int nb_sectors, QEMUIOVector *qiov, int flags); /** @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * * The buffer in @qiov may point directly to guest memory. */ - int coroutine_fn (*bdrv_co_pwritev)(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags); - int coroutine_fn (*bdrv_co_pwritev_part)(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pwritev)( + BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pwritev_part)( + BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, + size_t qiov_offset, BdrvRequestFlags flags); /* * Efficiently zero a region of the disk image. Typically an image format @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockMeasureInfo *(*bdrv_measure)(QemuOpts *opts, BlockDriverState *in_bs, Error **errp); - int coroutine_fn (*bdrv_co_pwritev_compressed)(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov); - int coroutine_fn (*bdrv_co_pwritev_compressed_part)(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, - size_t qiov_offset); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pwritev_compressed)( + BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pwritev_compressed_part)( + BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset); int coroutine_fn (*bdrv_co_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi); @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockDriverState *bs, const char *name, Error **errp); }; -static inline bool block_driver_can_compress(BlockDriver *drv) +static inline bool TSA_NO_TSA block_driver_can_compress(BlockDriver *drv) { return drv->bdrv_co_pwritev_compressed || drv->bdrv_co_pwritev_compressed_part; diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void bdrv_refresh_limits(BlockDriverState *bs, Transaction *tran, Error **errp) bool have_limits; GLOBAL_STATE_CODE(); + assume_graph_lock(); /* FIXME */ if (tran) { BdrvRefreshLimitsState *s = g_new(BdrvRefreshLimitsState, 1); @@ -XXX,XX +XXX,XX @@ static void bdrv_co_io_em_complete(void *opaque, int ret) aio_co_wake(co->coroutine); } -static int coroutine_fn bdrv_driver_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, int flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_driver_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, int flags) { BlockDriver *drv = bs->drv; int64_t sector_num; @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn bdrv_driver_pwritev(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_driver_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { BlockDriver *drv = bs->drv; bool emulate_fua = false; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_driver_pwritev(BlockDriverState *bs, QEMUIOVector local_qiov; int ret; - assume_graph_lock(); /* FIXME */ - bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ emulate_flags: return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK bdrv_driver_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset) @@ -XXX,XX +XXX,XX @@ bdrv_driver_pwritev_compressed(BlockDriverState *bs, int64_t offset, return ret; } -static int coroutine_fn bdrv_co_do_copy_on_readv(BdrvChild *child, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, - size_t qiov_offset, int flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_co_do_copy_on_readv(BdrvChild *child, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, int flags) { BlockDriverState *bs = child->bs; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_copy_on_readv(BdrvChild *child, int64_t progress = 0; bool skip_write; - assume_graph_lock(); /* FIXME */ - bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ err: * handles copy on read, zeroing after EOF, and fragmentation of large * reads; any other features must be implemented by the caller. */ -static int coroutine_fn bdrv_aligned_preadv(BdrvChild *child, - BdrvTrackedRequest *req, int64_t offset, int64_t bytes, - int64_t align, QEMUIOVector *qiov, size_t qiov_offset, int flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_aligned_preadv(BdrvChild *child, BdrvTrackedRequest *req, + int64_t offset, int64_t bytes, int64_t align, + QEMUIOVector *qiov, size_t qiov_offset, int flags) { BlockDriverState *bs = child->bs; int64_t total_bytes, max_bytes; @@ -XXX,XX +XXX,XX @@ static bool bdrv_init_padding(BlockDriverState *bs, return true; } -static coroutine_fn int bdrv_padding_rmw_read(BdrvChild *child, - BdrvTrackedRequest *req, - BdrvRequestPadding *pad, - bool zero_middle) +static int coroutine_fn GRAPH_RDLOCK +bdrv_padding_rmw_read(BdrvChild *child, BdrvTrackedRequest *req, + BdrvRequestPadding *pad, bool zero_middle) { QEMUIOVector local_qiov; BlockDriverState *bs = child->bs; @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_preadv_part(BdrvChild *child, int ret; IO_CODE(); + assume_graph_lock(); /* FIXME */ + trace_bdrv_co_preadv_part(bs, offset, bytes, flags); if (!bdrv_co_is_inserted(bs)) { @@ -XXX,XX +XXX,XX @@ bdrv_co_write_req_finish(BdrvChild *child, int64_t offset, int64_t bytes, * Forwards an already correctly aligned write request to the BlockDriver, * after possibly fragmenting it. */ -static int coroutine_fn bdrv_aligned_pwritev(BdrvChild *child, - BdrvTrackedRequest *req, int64_t offset, int64_t bytes, - int64_t align, QEMUIOVector *qiov, size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_aligned_pwritev(BdrvChild *child, BdrvTrackedRequest *req, + int64_t offset, int64_t bytes, int64_t align, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { BlockDriverState *bs = child->bs; BlockDriver *drv = bs->drv; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_aligned_pwritev(BdrvChild *child, int64_t bytes_remaining = bytes; int max_transfer; - assume_graph_lock(); /* FIXME */ - bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); if (!drv) { @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_aligned_pwritev(BdrvChild *child, return ret; } -static int coroutine_fn bdrv_co_do_zero_pwritev(BdrvChild *child, - int64_t offset, - int64_t bytes, - BdrvRequestFlags flags, - BdrvTrackedRequest *req) +static int coroutine_fn GRAPH_RDLOCK +bdrv_co_do_zero_pwritev(BdrvChild *child, int64_t offset, int64_t bytes, + BdrvRequestFlags flags, BdrvTrackedRequest *req) { BlockDriverState *bs = child->bs; QEMUIOVector local_qiov; @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwritev_part(BdrvChild *child, bool padded = false; IO_CODE(); + assume_graph_lock(); /* FIXME */ + trace_bdrv_co_pwritev_part(child->bs, offset, bytes, flags); if (!bdrv_co_is_inserted(bs)) { diff --git a/block/parallels.c b/block/parallels.c index XXXXXXX..XXXXXXX 100644 --- a/block/parallels.c +++ b/block/parallels.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn parallels_co_block_status(BlockDriverState *bs, return BDRV_BLOCK_DATA | BDRV_BLOCK_OFFSET_VALID; } -static coroutine_fn int parallels_co_writev(BlockDriverState *bs, - int64_t sector_num, int nb_sectors, - QEMUIOVector *qiov, int flags) +static int coroutine_fn GRAPH_RDLOCK +parallels_co_writev(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov, int flags) { BDRVParallelsState *s = bs->opaque; uint64_t bytes_done = 0; QEMUIOVector hd_qiov; int ret = 0; - assume_graph_lock(); /* FIXME */ - qemu_iovec_init(&hd_qiov, qiov->niov); while (nb_sectors > 0) { diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ static void qcow_refresh_limits(BlockDriverState *bs, Error **errp) bs->bl.request_alignment = BDRV_SECTOR_SIZE; } -static coroutine_fn int qcow_co_preadv(BlockDriverState *bs, int64_t offset, - int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +qcow_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVQcowState *s = bs->opaque; int offset_in_cluster; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_preadv(BlockDriverState *bs, int64_t offset, uint8_t *buf; void *orig_buf; - assume_graph_lock(); /* FIXME */ - if (qiov->niov > 1) { buf = orig_buf = qemu_try_blockalign(bs, qiov->size); if (buf == NULL) { @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_preadv(BlockDriverState *bs, int64_t offset, return ret; } -static coroutine_fn int qcow_co_pwritev(BlockDriverState *bs, int64_t offset, - int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +qcow_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVQcowState *s = bs->opaque; int offset_in_cluster; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_pwritev(BlockDriverState *bs, int64_t offset, uint8_t *buf; void *orig_buf; - assume_graph_lock(); /* FIXME */ - s->cluster_cache_offset = -1; /* disable compressed cache */ /* We must always copy the iov when encrypting, so we @@ -XXX,XX +XXX,XX @@ static int qcow_make_empty(BlockDriverState *bs) /* XXX: put compressed sectors first, then all the cluster aligned tables to avoid losing bytes in alignment */ -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK qcow_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov) { @@ -XXX,XX +XXX,XX @@ qcow_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, uint8_t *buf, *out_buf; uint64_t cluster_offset; - assume_graph_lock(); /* FIXME */ - buf = qemu_blockalign(bs, s->cluster_size); if (bytes != s->cluster_size) { if (bytes > s->cluster_size || diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2-cluster.c +++ b/block/qcow2-cluster.c @@ -XXX,XX +XXX,XX @@ static int count_contiguous_subclusters(BlockDriverState *bs, int nb_clusters, return count; } -static int coroutine_fn do_perform_cow_read(BlockDriverState *bs, - uint64_t src_cluster_offset, - unsigned offset_in_cluster, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +do_perform_cow_read(BlockDriverState *bs, uint64_t src_cluster_offset, + unsigned offset_in_cluster, QEMUIOVector *qiov) { int ret; @@ -XXX,XX +XXX,XX @@ int coroutine_fn qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs, return 0; } -static int coroutine_fn perform_cow(BlockDriverState *bs, QCowL2Meta *m) +static int coroutine_fn GRAPH_RDLOCK +perform_cow(BlockDriverState *bs, QCowL2Meta *m) { BDRVQcow2State *s = bs->opaque; Qcow2COWRegion *start = &m->cow_start; diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs, return status; } -static coroutine_fn int qcow2_handle_l2meta(BlockDriverState *bs, - QCowL2Meta **pl2meta, - bool link_l2) +static int coroutine_fn GRAPH_RDLOCK +qcow2_handle_l2meta(BlockDriverState *bs, QCowL2Meta **pl2meta, bool link_l2) { int ret = 0; QCowL2Meta *l2meta = *pl2meta; @@ -XXX,XX +XXX,XX @@ static coroutine_fn GRAPH_RDLOCK int qcow2_co_pwritev_task_entry(AioTask *task) t->l2meta); } -static coroutine_fn int qcow2_co_pwritev_part( - BlockDriverState *bs, int64_t offset, int64_t bytes, - QEMUIOVector *qiov, size_t qiov_offset, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_pwritev_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { BDRVQcow2State *s = bs->opaque; int offset_in_cluster; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_part( QCowL2Meta *l2meta = NULL; AioTaskPool *aio = NULL; - assume_graph_lock(); /* FIXME */ - trace_qcow2_writev_start_req(qemu_coroutine_self(), offset, bytes); while (bytes != 0 && aio_task_pool_status(aio) == 0) { @@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_to(BlockDriverState *bs, uint64_t host_offset; QCowL2Meta *l2meta = NULL; + assume_graph_lock(); /* FIXME */ assert(!bs->encrypted); qemu_co_mutex_lock(&s->lock); @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK qcow2_co_pwritev_compressed_task(BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, size_t qiov_offset) @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static coroutine_fn int qcow2_co_pwritev_compressed_task_entry(AioTask *task) +/* + * This function can count as GRAPH_RDLOCK because + * qcow2_co_pwritev_compressed_part() holds the graph lock and keeps it until + * this coroutine has terminated. + */ +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_pwritev_compressed_task_entry(AioTask *task) { Qcow2AioTask *t = container_of(task, Qcow2AioTask, task); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_compressed_task_entry(AioTask *task) * XXX: put compressed sectors first, then all the cluster aligned * tables to avoid losing bytes in alignment */ -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK qcow2_co_pwritev_compressed_part(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset) @@ -XXX,XX +XXX,XX @@ qcow2_co_pwritev_compressed_part(BlockDriverState *bs, AioTaskPool *aio = NULL; int ret = 0; - assume_graph_lock(); /* FIXME */ - if (has_data_file(bs)) { return -ENOTSUP; } @@ -XXX,XX +XXX,XX @@ static int64_t qcow2_check_vmstate_request(BlockDriverState *bs, return pos; } -static coroutine_fn int qcow2_co_save_vmstate(BlockDriverState *bs, - QEMUIOVector *qiov, int64_t pos) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_save_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos) { int64_t offset = qcow2_check_vmstate_request(bs, qiov, pos); if (offset < 0) { @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_save_vmstate(BlockDriverState *bs, return bs->drv->bdrv_co_pwritev_part(bs, offset, qiov->size, qiov, 0, 0); } -static coroutine_fn int qcow2_co_load_vmstate(BlockDriverState *bs, - QEMUIOVector *qiov, int64_t pos) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_load_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos) { int64_t offset = qcow2_check_vmstate_request(bs, qiov, pos); if (offset < 0) { diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ qed_co_request(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov, return qed_aio_next_io(&acb); } -static int coroutine_fn bdrv_qed_co_readv(BlockDriverState *bs, - int64_t sector_num, int nb_sectors, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_readv(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov) { - assume_graph_lock(); /* FIXME */ return qed_co_request(bs, sector_num, qiov, nb_sectors, 0); } -static int coroutine_fn bdrv_qed_co_writev(BlockDriverState *bs, - int64_t sector_num, int nb_sectors, - QEMUIOVector *qiov, int flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_writev(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov, int flags) { - assume_graph_lock(); /* FIXME */ return qed_co_request(bs, sector_num, qiov, nb_sectors, QED_AIOCB_WRITE); } diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn GRAPH_RDLOCK write_quorum_entry(void *opaque) } } -static int coroutine_fn quorum_co_pwritev(BlockDriverState *bs, int64_t offset, - int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +quorum_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVQuorumState *s = bs->opaque; QuorumAIOCB *acb = quorum_aio_get(bs, qiov, offset, bytes, flags); int i, ret; - assume_graph_lock(); /* FIXME */ - for (i = 0; i < s->num_children; i++) { Coroutine *co; QuorumCo data = { diff --git a/block/vmdk.c b/block/vmdk.c index XXXXXXX..XXXXXXX 100644 --- a/block/vmdk.c +++ b/block/vmdk.c @@ -XXX,XX +XXX,XX @@ vmdk_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vmdk_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov) { - assume_graph_lock(); /* FIXME */ - if (bytes == 0) { /* The caller will write bytes 0 to signal EOF. * When receive it, we align EOF to a sector boundary. */ -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_pread*/pwrite*() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-12-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 9 ++--- block/qed.h | 18 +++++----- include/block/block_int-io.h | 14 ++++---- block/blkdebug.c | 4 +-- block/blklogwrites.c | 7 ++-- block/blkreplay.c | 10 +++--- block/block-backend.c | 2 ++ block/bochs.c | 2 +- block/commit.c | 5 +-- block/copy-before-write.c | 16 ++++----- block/copy-on-read.c | 26 ++++++--------- block/crypto.c | 4 +-- block/filter-compress.c | 19 +++++------ block/io.c | 7 ++-- block/mirror.c | 18 +++++----- block/parallels.c | 8 +++-- block/preallocate.c | 18 +++++----- block/qcow.c | 8 ++--- block/qcow2-cluster.c | 7 ++-- block/qcow2.c | 53 +++++++++++++++-------------- block/qed-table.c | 4 +-- block/qed.c | 31 +++++++++-------- block/quorum.c | 29 ++++++++++------ block/raw-format.c | 12 +++---- block/replication.c | 15 ++++----- block/throttle.c | 21 +++++------- block/vdi.c | 4 +-- block/vhdx.c | 11 +++--- block/vmdk.c | 65 +++++++++++++++--------------------- block/vpc.c | 4 +-- tests/unit/test-bdrv-drain.c | 20 ++++++----- 31 files changed, 233 insertions(+), 238 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ int qcow2_validate_table(BlockDriverState *bs, uint64_t offset, Error **errp); /* qcow2-refcount.c functions */ -int coroutine_fn qcow2_refcount_init(BlockDriverState *bs); +int coroutine_fn GRAPH_RDLOCK qcow2_refcount_init(BlockDriverState *bs); void qcow2_refcount_close(BlockDriverState *bs); int qcow2_get_refcount(BlockDriverState *bs, int64_t cluster_index, @@ -XXX,XX +XXX,XX @@ void qcow2_free_snapshots(BlockDriverState *bs); int qcow2_read_snapshots(BlockDriverState *bs, Error **errp); int qcow2_write_snapshots(BlockDriverState *bs); -int coroutine_fn qcow2_check_read_snapshot_table(BlockDriverState *bs, - BdrvCheckResult *result, - BdrvCheckMode fix); +int coroutine_fn GRAPH_RDLOCK +qcow2_check_read_snapshot_table(BlockDriverState *bs, BdrvCheckResult *result, + BdrvCheckMode fix); + int coroutine_fn qcow2_check_fix_snapshot_table(BlockDriverState *bs, BdrvCheckResult *result, BdrvCheckMode fix); diff --git a/block/qed.h b/block/qed.h index XXXXXXX..XXXXXXX 100644 --- a/block/qed.h +++ b/block/qed.h @@ -XXX,XX +XXX,XX @@ void qed_commit_l2_cache_entry(L2TableCache *l2_cache, CachedL2Table *l2_table); /** * Table I/O functions */ -int coroutine_fn qed_read_l1_table_sync(BDRVQEDState *s); +int coroutine_fn GRAPH_RDLOCK qed_read_l1_table_sync(BDRVQEDState *s); int coroutine_fn GRAPH_RDLOCK qed_write_l1_table(BDRVQEDState *s, unsigned int index, unsigned int n); @@ -XXX,XX +XXX,XX @@ qed_write_l1_table(BDRVQEDState *s, unsigned int index, unsigned int n); int coroutine_fn GRAPH_RDLOCK qed_write_l1_table_sync(BDRVQEDState *s, unsigned int index, unsigned int n); -int coroutine_fn qed_read_l2_table_sync(BDRVQEDState *s, QEDRequest *request, - uint64_t offset); -int coroutine_fn qed_read_l2_table(BDRVQEDState *s, QEDRequest *request, - uint64_t offset); +int coroutine_fn GRAPH_RDLOCK +qed_read_l2_table_sync(BDRVQEDState *s, QEDRequest *request, uint64_t offset); + +int coroutine_fn GRAPH_RDLOCK +qed_read_l2_table(BDRVQEDState *s, QEDRequest *request, uint64_t offset); int coroutine_fn GRAPH_RDLOCK qed_write_l2_table(BDRVQEDState *s, QEDRequest *request, unsigned int index, @@ -XXX,XX +XXX,XX @@ qed_write_l2_table_sync(BDRVQEDState *s, QEDRequest *request, /** * Cluster functions */ -int coroutine_fn qed_find_cluster(BDRVQEDState *s, QEDRequest *request, - uint64_t pos, size_t *len, - uint64_t *img_offset); +int coroutine_fn GRAPH_RDLOCK +qed_find_cluster(BDRVQEDState *s, QEDRequest *request, uint64_t pos, + size_t *len, uint64_t *img_offset); /** * Consistency check @@ -XXX,XX +XXX,XX @@ int coroutine_fn qed_find_cluster(BDRVQEDState *s, QEDRequest *request, int coroutine_fn GRAPH_RDLOCK qed_check(BDRVQEDState *s, BdrvCheckResult *result, bool fix); - QEDTable *qed_alloc_table(BDRVQEDState *s); /** diff --git a/include/block/block_int-io.h b/include/block/block_int-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-io.h +++ b/include/block/block_int-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn GRAPH_RDLOCK bdrv_co_pdiscard_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes); -int coroutine_fn bdrv_co_preadv(BdrvChild *child, +int coroutine_fn GRAPH_RDLOCK bdrv_co_preadv(BdrvChild *child, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags); -int coroutine_fn bdrv_co_preadv_part(BdrvChild *child, +int coroutine_fn GRAPH_RDLOCK bdrv_co_preadv_part(BdrvChild *child, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, BdrvRequestFlags flags); -int coroutine_fn bdrv_co_pwritev(BdrvChild *child, +int coroutine_fn GRAPH_RDLOCK bdrv_co_pwritev(BdrvChild *child, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags); -int coroutine_fn bdrv_co_pwritev_part(BdrvChild *child, +int coroutine_fn GRAPH_RDLOCK bdrv_co_pwritev_part(BdrvChild *child, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, BdrvRequestFlags flags); -static inline int coroutine_fn bdrv_co_pread(BdrvChild *child, +static inline int coroutine_fn GRAPH_RDLOCK bdrv_co_pread(BdrvChild *child, int64_t offset, int64_t bytes, void *buf, BdrvRequestFlags flags) { QEMUIOVector qiov = QEMU_IOVEC_INIT_BUF(qiov, buf, bytes); IO_CODE(); + assert_bdrv_graph_readable(); return bdrv_co_preadv(child, offset, bytes, &qiov, flags); } -static inline int coroutine_fn bdrv_co_pwrite(BdrvChild *child, +static inline int coroutine_fn GRAPH_RDLOCK bdrv_co_pwrite(BdrvChild *child, int64_t offset, int64_t bytes, const void *buf, BdrvRequestFlags flags) { QEMUIOVector qiov = QEMU_IOVEC_INIT_BUF(qiov, buf, bytes); IO_CODE(); + assert_bdrv_graph_readable(); return bdrv_co_pwritev(child, offset, bytes, &qiov, flags); } diff --git a/block/blkdebug.c b/block/blkdebug.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkdebug.c +++ b/block/blkdebug.c @@ -XXX,XX +XXX,XX @@ static int rule_check(BlockDriverState *bs, uint64_t offset, uint64_t bytes, return -error; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blkdebug_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ blkdebug_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blkdebug_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { diff --git a/block/blklogwrites.c b/block/blklogwrites.c index XXXXXXX..XXXXXXX 100644 --- a/block/blklogwrites.c +++ b/block/blklogwrites.c @@ -XXX,XX +XXX,XX @@ static void blk_log_writes_refresh_limits(BlockDriverState *bs, Error **errp) bs->bl.request_alignment = s->sectorsize; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_log(BlockDriverState *bs, uint64_t offset, uint64_t bytes, return fr.file_ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_do_file_pwritev(BlkLogWritesFileReq *fr) { return bdrv_co_pwritev(fr->bs->file, fr->offset, fr->bytes, @@ -XXX,XX +XXX,XX @@ blk_log_writes_co_do_file_pdiscard(BlkLogWritesFileReq *fr) return bdrv_co_pdiscard(fr->bs->file, fr->offset, fr->bytes); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK blk_log_writes_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { - assume_graph_lock(); /* FIXME */ return blk_log_writes_co_log(bs, offset, bytes, qiov, flags, blk_log_writes_co_do_file_pwritev, 0, false); } diff --git a/block/blkreplay.c b/block/blkreplay.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkreplay.c +++ b/block/blkreplay.c @@ -XXX,XX +XXX,XX @@ static void block_request_create(uint64_t reqid, BlockDriverState *bs, replay_block_event(req->bh, reqid); } -static int coroutine_fn blkreplay_co_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +blkreplay_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { uint64_t reqid = blkreplay_next_id(); int ret = bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blkreplay_co_preadv(BlockDriverState *bs, return ret; } -static int coroutine_fn blkreplay_co_pwritev(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +blkreplay_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { uint64_t reqid = blkreplay_next_id(); int ret = bdrv_co_pwritev(bs->file, offset, bytes, qiov, flags); diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ blk_co_do_preadv_part(BlockBackend *blk, int64_t offset, int64_t bytes, IO_CODE(); blk_wait_while_drained(blk); + GRAPH_RDLOCK_GUARD(); /* Call blk_bs() only after waiting, the graph may have changed */ bs = blk_bs(blk); @@ -XXX,XX +XXX,XX @@ blk_co_do_pwritev_part(BlockBackend *blk, int64_t offset, int64_t bytes, IO_CODE(); blk_wait_while_drained(blk); + GRAPH_RDLOCK_GUARD(); /* Call blk_bs() only after waiting, the graph may have changed */ bs = blk_bs(blk); diff --git a/block/bochs.c b/block/bochs.c index XXXXXXX..XXXXXXX 100644 --- a/block/bochs.c +++ b/block/bochs.c @@ -XXX,XX +XXX,XX @@ static int64_t seek_to_sector(BlockDriverState *bs, int64_t sector_num) return bitmap_offset + (512 * (s->bitmap_blocks + extent_offset)); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK bochs_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { diff --git a/block/commit.c b/block/commit.c index XXXXXXX..XXXXXXX 100644 --- a/block/commit.c +++ b/block/commit.c @@ -XXX,XX +XXX,XX @@ static const BlockJobDriver commit_job_driver = { }, }; -static int coroutine_fn bdrv_commit_top_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_commit_top_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { return bdrv_co_preadv(bs->backing, offset, bytes, qiov, flags); } diff --git a/block/copy-before-write.c b/block/copy-before-write.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-before-write.c +++ b/block/copy-before-write.c @@ -XXX,XX +XXX,XX @@ typedef struct BDRVCopyBeforeWriteState { int snapshot_error; } BDRVCopyBeforeWriteState; -static coroutine_fn int cbw_co_preadv( - BlockDriverState *bs, int64_t offset, int64_t bytes, - QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +cbw_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); } @@ -XXX,XX +XXX,XX @@ cbw_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } -static coroutine_fn int cbw_co_pwritev(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static coroutine_fn GRAPH_RDLOCK +int cbw_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { int ret = cbw_do_copy_before_write(bs, offset, bytes, flags); if (ret < 0) { @@ -XXX,XX +XXX,XX @@ cbw_co_preadv_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvChild *file; int ret; + assume_graph_lock(); /* FIXME */ + /* TODO: upgrade to async loop using AioTask */ while (bytes) { int64_t cur_bytes; diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ static int64_t coroutine_fn cor_co_getlength(BlockDriverState *bs) } -static int coroutine_fn cor_co_preadv_part(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +cor_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { int64_t n; int local_flags; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn cor_co_preadv_part(BlockDriverState *bs, } -static int coroutine_fn cor_co_pwritev_part(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +cor_co_pwritev_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { return bdrv_co_pwritev_part(bs->file, offset, bytes, qiov, qiov_offset, flags); @@ -XXX,XX +XXX,XX @@ cor_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) } -static int coroutine_fn cor_co_pwritev_compressed(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +cor_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov) { return bdrv_co_pwritev(bs->file, offset, bytes, qiov, BDRV_REQ_WRITE_COMPRESSED); diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ static int block_crypto_reopen_prepare(BDRVReopenState *state, */ #define BLOCK_CRYPTO_MAX_IO_SIZE (1024 * 1024) -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK block_crypto_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ block_crypto_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, } -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK block_crypto_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { diff --git a/block/filter-compress.c b/block/filter-compress.c index XXXXXXX..XXXXXXX 100644 --- a/block/filter-compress.c +++ b/block/filter-compress.c @@ -XXX,XX +XXX,XX @@ static int64_t coroutine_fn compress_co_getlength(BlockDriverState *bs) } -static int coroutine_fn compress_co_preadv_part(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +compress_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { return bdrv_co_preadv_part(bs->file, offset, bytes, qiov, qiov_offset, flags); } -static int coroutine_fn compress_co_pwritev_part(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +compress_co_pwritev_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { return bdrv_co_pwritev_part(bs->file, offset, bytes, qiov, qiov_offset, flags | BDRV_REQ_WRITE_COMPRESSED); diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ bdrv_driver_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, unsigned int nb_sectors; QEMUIOVector local_qiov; int ret; + assert_bdrv_graph_readable(); bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); assert(!(flags & ~bs->supported_read_flags)); @@ -XXX,XX +XXX,XX @@ bdrv_driver_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, unsigned int nb_sectors; QEMUIOVector local_qiov; int ret; + assert_bdrv_graph_readable(); bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); @@ -XXX,XX +XXX,XX @@ bdrv_driver_pwritev_compressed(BlockDriverState *bs, int64_t offset, BlockDriver *drv = bs->drv; QEMUIOVector local_qiov; int ret; + assert_bdrv_graph_readable(); bdrv_check_qiov_request(offset, bytes, qiov, qiov_offset, &error_abort); @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_preadv_part(BdrvChild *child, int ret; IO_CODE(); - assume_graph_lock(); /* FIXME */ - trace_bdrv_co_preadv_part(bs, offset, bytes, flags); if (!bdrv_co_is_inserted(bs)) { @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwritev_part(BdrvChild *child, bool padded = false; IO_CODE(); - assume_graph_lock(); /* FIXME */ - trace_bdrv_co_pwritev_part(child->bs, offset, bytes, flags); if (!bdrv_co_is_inserted(bs)) { diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn mirror_co_read(void *opaque) op->is_in_flight = true; trace_mirror_one_iteration(s, op->offset, op->bytes); - ret = bdrv_co_preadv(s->mirror_top_bs->backing, op->offset, op->bytes, - &op->qiov, 0); + WITH_GRAPH_RDLOCK_GUARD() { + ret = bdrv_co_preadv(s->mirror_top_bs->backing, op->offset, op->bytes, + &op->qiov, 0); + } mirror_read_complete(op, ret); } @@ -XXX,XX +XXX,XX @@ static void coroutine_fn active_write_settle(MirrorOp *op) g_free(op); } -static int coroutine_fn bdrv_mirror_top_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_mirror_top_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { return bdrv_co_preadv(bs->backing, offset, bytes, qiov, flags); } @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn bdrv_mirror_top_pwritev(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_mirror_top_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { MirrorBDSOpaque *s = bs->opaque; QEMUIOVector bounce_qiov; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_mirror_top_pwritev(BlockDriverState *bs, int ret = 0; bool copy_to_target = false; - assume_graph_lock(); /* FIXME */ - if (s->job) { copy_to_target = s->job->ret >= 0 && !job_is_cancelled(&s->job->common.job) && diff --git a/block/parallels.c b/block/parallels.c index XXXXXXX..XXXXXXX 100644 --- a/block/parallels.c +++ b/block/parallels.c @@ -XXX,XX +XXX,XX @@ allocate_clusters(BlockDriverState *bs, int64_t sector_num, } -static coroutine_fn int parallels_co_flush_to_os(BlockDriverState *bs) +static int coroutine_fn GRAPH_RDLOCK +parallels_co_flush_to_os(BlockDriverState *bs) { BDRVParallelsState *s = bs->opaque; unsigned long size = DIV_ROUND_UP(s->header_size, s->bat_dirty_block); @@ -XXX,XX +XXX,XX @@ parallels_co_writev(BlockDriverState *bs, int64_t sector_num, int nb_sectors, return ret; } -static coroutine_fn int parallels_co_readv(BlockDriverState *bs, - int64_t sector_num, int nb_sectors, QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +parallels_co_readv(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov) { BDRVParallelsState *s = bs->opaque; uint64_t bytes_done = 0; diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ static void preallocate_reopen_abort(BDRVReopenState *state) state->opaque = NULL; } -static coroutine_fn int preallocate_co_preadv_part( - BlockDriverState *bs, int64_t offset, int64_t bytes, - QEMUIOVector *qiov, size_t qiov_offset, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +preallocate_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { return bdrv_co_preadv_part(bs->file, offset, bytes, qiov, qiov_offset, flags); @@ -XXX,XX +XXX,XX @@ preallocate_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags); } -static coroutine_fn int preallocate_co_pwritev_part(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +preallocate_co_pwritev_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { - assume_graph_lock(); /* FIXME */ handle_write(bs, offset, bytes, false); return bdrv_co_pwritev_part(bs->file, offset, bytes, qiov, qiov_offset, diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ typedef struct BDRVQcowState { static QemuOptsList qcow_create_opts; -static int coroutine_fn decompress_cluster(BlockDriverState *bs, - uint64_t cluster_offset); +static int coroutine_fn GRAPH_RDLOCK +decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset); static int qcow_probe(const uint8_t *buf, int buf_size, const char *filename) { @@ -XXX,XX +XXX,XX @@ static int decompress_buffer(uint8_t *out_buf, int out_buf_size, return 0; } -static int coroutine_fn decompress_cluster(BlockDriverState *bs, - uint64_t cluster_offset) +static int coroutine_fn GRAPH_RDLOCK +decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset) { BDRVQcowState *s = bs->opaque; int ret, csize; diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2-cluster.c +++ b/block/qcow2-cluster.c @@ -XXX,XX +XXX,XX @@ do_perform_cow_read(BlockDriverState *bs, uint64_t src_cluster_offset, return 0; } -static int coroutine_fn do_perform_cow_write(BlockDriverState *bs, - uint64_t cluster_offset, - unsigned offset_in_cluster, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +do_perform_cow_write(BlockDriverState *bs, uint64_t cluster_offset, + unsigned offset_in_cluster, QEMUIOVector *qiov) { BDRVQcow2State *s = bs->opaque; int ret; diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static void qcow2_add_check_result(BdrvCheckResult *out, } } -static int coroutine_fn qcow2_co_check_locked(BlockDriverState *bs, - BdrvCheckResult *result, - BdrvCheckMode fix) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_check_locked(BlockDriverState *bs, BdrvCheckResult *result, + BdrvCheckMode fix) { BdrvCheckResult snapshot_res = {}; BdrvCheckResult refcount_res = {}; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_check_locked(BlockDriverState *bs, return ret; } -static int coroutine_fn qcow2_co_check(BlockDriverState *bs, - BdrvCheckResult *result, - BdrvCheckMode fix) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_check(BlockDriverState *bs, BdrvCheckResult *result, + BdrvCheckMode fix) { BDRVQcow2State *s = bs->opaque; int ret; @@ -XXX,XX +XXX,XX @@ static int validate_compression_type(BDRVQcow2State *s, Error **errp) } /* Called with s->lock held. */ -static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options, - int flags, bool open_data_file, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +qcow2_do_open(BlockDriverState *bs, QDict *options, int flags, + bool open_data_file, Error **errp) { ERRP_GUARD(); BDRVQcow2State *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn qcow2_open_entry(void *opaque) QCow2OpenCo *qoc = opaque; BDRVQcow2State *s = qoc->bs->opaque; + assume_graph_lock(); /* FIXME */ + qemu_co_mutex_lock(&s->lock); qoc->ret = qcow2_do_open(qoc->bs, qoc->options, qoc->flags, true, qoc->errp); @@ -XXX,XX +XXX,XX @@ out: return ret; } -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK qcow2_co_preadv_encrypted(BlockDriverState *bs, uint64_t host_offset, uint64_t offset, @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs, return 0; } -static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs, - QCow2SubclusterType subc_type, - uint64_t host_offset, - uint64_t offset, uint64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_preadv_task(BlockDriverState *bs, QCow2SubclusterType subc_type, + uint64_t host_offset, uint64_t offset, uint64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset) { BDRVQcow2State *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs, g_assert_not_reached(); } -static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task) +/* + * This function can count as GRAPH_RDLOCK because qcow2_co_preadv_part() holds + * the graph lock and keeps it until this coroutine has terminated. + */ +static int coroutine_fn GRAPH_RDLOCK qcow2_co_preadv_task_entry(AioTask *task) { Qcow2AioTask *t = container_of(task, Qcow2AioTask, task); @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task) t->qiov, t->qiov_offset); } -static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - size_t qiov_offset, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset, + BdrvRequestFlags flags) { BDRVQcow2State *s = bs->opaque; int ret = 0; @@ -XXX,XX +XXX,XX @@ static void qcow2_close(BlockDriverState *bs) qcow2_do_close(bs, true); } -static void coroutine_fn qcow2_co_invalidate_cache(BlockDriverState *bs, - Error **errp) +static void coroutine_fn GRAPH_RDLOCK +qcow2_co_invalidate_cache(BlockDriverState *bs, Error **errp) { ERRP_GUARD(); BDRVQcow2State *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ qcow2_co_pwritev_compressed_part(BlockDriverState *bs, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK qcow2_co_preadv_compressed(BlockDriverState *bs, uint64_t l2_entry, uint64_t offset, diff --git a/block/qed-table.c b/block/qed-table.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed-table.c +++ b/block/qed-table.c @@ -XXX,XX +XXX,XX @@ #include "qemu/memalign.h" /* Called with table_lock held. */ -static int coroutine_fn qed_read_table(BDRVQEDState *s, uint64_t offset, - QEDTable *table) +static int coroutine_fn GRAPH_RDLOCK +qed_read_table(BDRVQEDState *s, uint64_t offset, QEDTable *table) { unsigned int bytes = s->header.cluster_size * s->header.table_size; diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ int qed_write_header_sync(BDRVQEDState *s) * * No new allocating reqs can start while this function runs. */ -static int coroutine_fn qed_write_header(BDRVQEDState *s) +static int coroutine_fn GRAPH_RDLOCK qed_write_header(BDRVQEDState *s) { /* We must write full sectors for O_DIRECT but cannot necessarily generate * the data following the header if an unrecognized compat feature is @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int coroutine_fn bdrv_qed_co_block_status(BlockDriverState *bs, - bool want_zero, - int64_t pos, int64_t bytes, - int64_t *pnum, int64_t *map, - BlockDriverState **file) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_block_status(BlockDriverState *bs, bool want_zero, int64_t pos, + int64_t bytes, int64_t *pnum, int64_t *map, + BlockDriverState **file) { BDRVQEDState *s = bs->opaque; size_t len = MIN(bytes, SIZE_MAX); @@ -XXX,XX +XXX,XX @@ static BDRVQEDState *acb_to_s(QEDAIOCB *acb) * This function reads qiov->size bytes starting at pos from the backing file. * If there is no backing file then zeroes are read. */ -static int coroutine_fn qed_read_backing_file(BDRVQEDState *s, uint64_t pos, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +qed_read_backing_file(BDRVQEDState *s, uint64_t pos, QEMUIOVector *qiov) { if (s->bs->backing) { BLKDBG_EVENT(s->bs->file, BLKDBG_READ_BACKING_AIO); @@ -XXX,XX +XXX,XX @@ static int coroutine_fn qed_read_backing_file(BDRVQEDState *s, uint64_t pos, * @len: Number of bytes * @offset: Byte offset in image file */ -static int coroutine_fn qed_copy_from_backing_file(BDRVQEDState *s, - uint64_t pos, uint64_t len, - uint64_t offset) +static int coroutine_fn GRAPH_RDLOCK +qed_copy_from_backing_file(BDRVQEDState *s, uint64_t pos, uint64_t len, + uint64_t offset) { QEMUIOVector qiov; int ret; @@ -XXX,XX +XXX,XX @@ qed_aio_write_l2_update(QEDAIOCB *acb, uint64_t offset) * * Called with table_lock *not* held. */ -static int coroutine_fn qed_aio_write_main(QEDAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK qed_aio_write_main(QEDAIOCB *acb) { BDRVQEDState *s = acb_to_s(acb); uint64_t offset = acb->cur_cluster + @@ -XXX,XX +XXX,XX @@ qed_aio_write_alloc(QEDAIOCB *acb, size_t len) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_write_inplace(QEDAIOCB *acb, uint64_t offset, - size_t len) +static int coroutine_fn GRAPH_RDLOCK +qed_aio_write_inplace(QEDAIOCB *acb, uint64_t offset, size_t len) { BDRVQEDState *s = acb_to_s(acb); int r; @@ -XXX,XX +XXX,XX @@ qed_aio_write_data(void *opaque, int ret, uint64_t offset, size_t len) * * Called with table_lock held. */ -static int coroutine_fn qed_aio_read_data(void *opaque, int ret, - uint64_t offset, size_t len) +static int coroutine_fn GRAPH_RDLOCK +qed_aio_read_data(void *opaque, int ret, uint64_t offset, size_t len) { QEDAIOCB *acb = opaque; BDRVQEDState *s = acb_to_s(acb); diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ static void quorum_report_bad_versions(BDRVQuorumState *s, } } -static void coroutine_fn quorum_rewrite_entry(void *opaque) +/* + * This function can count as GRAPH_RDLOCK because read_quorum_children() holds + * the graph lock and keeps it until this coroutine has terminated. + */ +static void coroutine_fn GRAPH_RDLOCK quorum_rewrite_entry(void *opaque) { QuorumCo *co = opaque; QuorumAIOCB *acb = co->acb; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn quorum_rewrite_entry(void *opaque) } } -static bool quorum_rewrite_bad_versions(QuorumAIOCB *acb, - QuorumVoteValue *value) +static bool coroutine_fn GRAPH_RDLOCK +quorum_rewrite_bad_versions(QuorumAIOCB *acb, QuorumVoteValue *value) { QuorumVoteVersion *version; QuorumVoteItem *item; @@ -XXX,XX +XXX,XX @@ static int quorum_vote_error(QuorumAIOCB *acb) return ret; } -static void quorum_vote(QuorumAIOCB *acb) +static void coroutine_fn GRAPH_RDLOCK quorum_vote(QuorumAIOCB *acb) { bool quorum = true; int i, j, ret; @@ -XXX,XX +XXX,XX @@ free_exit: quorum_free_vote_list(&acb->votes); } -static void coroutine_fn read_quorum_children_entry(void *opaque) +/* + * This function can count as GRAPH_RDLOCK because read_quorum_children() holds + * the graph lock and keeps it until this coroutine has terminated. + */ +static void coroutine_fn GRAPH_RDLOCK read_quorum_children_entry(void *opaque) { QuorumCo *co = opaque; QuorumAIOCB *acb = co->acb; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn read_quorum_children_entry(void *opaque) } } -static int coroutine_fn read_quorum_children(QuorumAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK read_quorum_children(QuorumAIOCB *acb) { BDRVQuorumState *s = acb->bs->opaque; int i; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn read_quorum_children(QuorumAIOCB *acb) return acb->vote_ret; } -static int coroutine_fn read_fifo_child(QuorumAIOCB *acb) +static int coroutine_fn GRAPH_RDLOCK read_fifo_child(QuorumAIOCB *acb) { BDRVQuorumState *s = acb->bs->opaque; int n, ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn read_fifo_child(QuorumAIOCB *acb) return ret; } -static int coroutine_fn quorum_co_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +quorum_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVQuorumState *s = bs->opaque; QuorumAIOCB *acb = quorum_aio_get(bs, qiov, offset, bytes, flags); diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static inline int raw_adjust_offset(BlockDriverState *bs, int64_t *offset, return 0; } -static int coroutine_fn raw_co_preadv(BlockDriverState *bs, int64_t offset, - int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { int ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_preadv(BlockDriverState *bs, int64_t offset, return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn raw_co_pwritev(BlockDriverState *bs, int64_t offset, - int64_t bytes, QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { void *buf = NULL; BlockDriver *drv; diff --git a/block/replication.c b/block/replication.c index XXXXXXX..XXXXXXX 100644 --- a/block/replication.c +++ b/block/replication.c @@ -XXX,XX +XXX,XX @@ static int replication_return_value(BDRVReplicationState *s, int ret) return ret; } -static coroutine_fn int replication_co_readv(BlockDriverState *bs, - int64_t sector_num, - int remaining_sectors, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +replication_co_readv(BlockDriverState *bs, int64_t sector_num, + int remaining_sectors, QEMUIOVector *qiov) { BDRVReplicationState *s = bs->opaque; int ret; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int replication_co_readv(BlockDriverState *bs, return replication_return_value(s, ret); } -static coroutine_fn int replication_co_writev(BlockDriverState *bs, - int64_t sector_num, - int remaining_sectors, - QEMUIOVector *qiov, - int flags) +static int coroutine_fn GRAPH_RDLOCK +replication_co_writev(BlockDriverState *bs, int64_t sector_num, + int remaining_sectors, QEMUIOVector *qiov, int flags) { BDRVReplicationState *s = bs->opaque; QEMUIOVector hd_qiov; diff --git a/block/throttle.c b/block/throttle.c index XXXXXXX..XXXXXXX 100644 --- a/block/throttle.c +++ b/block/throttle.c @@ -XXX,XX +XXX,XX @@ static int64_t coroutine_fn throttle_co_getlength(BlockDriverState *bs) return bdrv_co_getlength(bs->file->bs); } -static int coroutine_fn throttle_co_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +throttle_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { ThrottleGroupMember *tgm = bs->opaque; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn throttle_co_preadv(BlockDriverState *bs, return bdrv_co_preadv(bs->file, offset, bytes, qiov, flags); } -static int coroutine_fn throttle_co_pwritev(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +throttle_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { ThrottleGroupMember *tgm = bs->opaque; throttle_group_co_io_limits_intercept(tgm, bytes, true); @@ -XXX,XX +XXX,XX @@ throttle_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) return bdrv_co_pdiscard(bs->file, offset, bytes); } -static int coroutine_fn throttle_co_pwritev_compressed(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +throttle_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, + int64_t bytes, QEMUIOVector *qiov) { return throttle_co_pwritev(bs, offset, bytes, qiov, BDRV_REQ_WRITE_COMPRESSED); diff --git a/block/vdi.c b/block/vdi.c index XXXXXXX..XXXXXXX 100644 --- a/block/vdi.c +++ b/block/vdi.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vdi_co_block_status(BlockDriverState *bs, (s->header.image_type == VDI_TYPE_STATIC ? BDRV_BLOCK_RECURSE : 0); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vdi_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ vdi_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vdi_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { diff --git a/block/vhdx.c b/block/vhdx.c index XXXXXXX..XXXXXXX 100644 --- a/block/vhdx.c +++ b/block/vhdx.c @@ -XXX,XX +XXX,XX @@ vhdx_co_get_info(BlockDriverState *bs, BlockDriverInfo *bdi) } -static coroutine_fn int vhdx_co_readv(BlockDriverState *bs, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov) +static int coroutine_fn GRAPH_RDLOCK +vhdx_co_readv(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov) { BDRVVHDXState *s = bs->opaque; int ret = 0; @@ -XXX,XX +XXX,XX @@ int vhdx_user_visible_write(BlockDriverState *bs, BDRVVHDXState *s) return ret; } -static coroutine_fn int vhdx_co_writev(BlockDriverState *bs, int64_t sector_num, - int nb_sectors, QEMUIOVector *qiov, - int flags) +static int coroutine_fn GRAPH_RDLOCK +vhdx_co_writev(BlockDriverState *bs, int64_t sector_num, int nb_sectors, + QEMUIOVector *qiov, int flags) { int ret = -ENOTSUP; BDRVVHDXState *s = bs->opaque; diff --git a/block/vmdk.c b/block/vmdk.c index XXXXXXX..XXXXXXX 100644 --- a/block/vmdk.c +++ b/block/vmdk.c @@ -XXX,XX +XXX,XX @@ static void vmdk_refresh_limits(BlockDriverState *bs, Error **errp) * [@skip_start_sector, @skip_end_sector) is not copied or written, and leave * it for call to write user data in the request. */ -static int coroutine_fn get_whole_cluster(BlockDriverState *bs, - VmdkExtent *extent, - uint64_t cluster_offset, - uint64_t offset, - uint64_t skip_start_bytes, - uint64_t skip_end_bytes, - bool zeroed) +static int coroutine_fn GRAPH_RDLOCK +get_whole_cluster(BlockDriverState *bs, VmdkExtent *extent, + uint64_t cluster_offset, uint64_t offset, + uint64_t skip_start_bytes, uint64_t skip_end_bytes, + bool zeroed) { int ret = VMDK_OK; int64_t cluster_bytes; @@ -XXX,XX +XXX,XX @@ vmdk_L2update(VmdkExtent *extent, VmdkMetaData *m_data, uint32_t offset) * VMDK_UNALLOC if cluster is not mapped and @allocate is false. * VMDK_ERROR if failed. */ -static int coroutine_fn get_cluster_offset(BlockDriverState *bs, - VmdkExtent *extent, - VmdkMetaData *m_data, - uint64_t offset, - bool allocate, - uint64_t *cluster_offset, - uint64_t skip_start_bytes, - uint64_t skip_end_bytes) +static int coroutine_fn GRAPH_RDLOCK +get_cluster_offset(BlockDriverState *bs, VmdkExtent *extent, + VmdkMetaData *m_data, uint64_t offset, bool allocate, + uint64_t *cluster_offset, uint64_t skip_start_bytes, + uint64_t skip_end_bytes) { unsigned int l1_index, l2_offset, l2_index; int min_index, i, j; @@ -XXX,XX +XXX,XX @@ static inline uint64_t vmdk_find_offset_in_cluster(VmdkExtent *extent, return extent_relative_offset % cluster_size; } -static int coroutine_fn vmdk_co_block_status(BlockDriverState *bs, - bool want_zero, - int64_t offset, int64_t bytes, - int64_t *pnum, int64_t *map, - BlockDriverState **file) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_block_status(BlockDriverState *bs, bool want_zero, + int64_t offset, int64_t bytes, int64_t *pnum, + int64_t *map, BlockDriverState **file) { BDRVVmdkState *s = bs->opaque; int64_t index_in_cluster, n, ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vmdk_co_block_status(BlockDriverState *bs, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vmdk_write_extent(VmdkExtent *extent, int64_t cluster_offset, int64_t offset_in_cluster, QEMUIOVector *qiov, uint64_t qiov_offset, uint64_t n_bytes, @@ -XXX,XX +XXX,XX @@ vmdk_write_extent(VmdkExtent *extent, int64_t cluster_offset, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vmdk_read_extent(VmdkExtent *extent, int64_t cluster_offset, - int64_t offset_in_cluster, QEMUIOVector *qiov, - int bytes) + int64_t offset_in_cluster, QEMUIOVector *qiov, int bytes) { int ret; int cluster_bytes, buf_bytes; @@ -XXX,XX +XXX,XX @@ vmdk_read_extent(VmdkExtent *extent, int64_t cluster_offset, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vmdk_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ fail: * * Returns: error code with 0 for success. */ -static int coroutine_fn vmdk_pwritev(BlockDriverState *bs, uint64_t offset, - uint64_t bytes, QEMUIOVector *qiov, - bool zeroed, bool zero_dry_run) +static int coroutine_fn GRAPH_RDLOCK +vmdk_pwritev(BlockDriverState *bs, uint64_t offset, uint64_t bytes, + QEMUIOVector *qiov, bool zeroed, bool zero_dry_run) { BDRVVmdkState *s = bs->opaque; VmdkExtent *extent = NULL; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vmdk_pwritev(BlockDriverState *bs, uint64_t offset, uint64_t bytes_done = 0; VmdkMetaData m_data; - assume_graph_lock(); /* FIXME */ - if (DIV_ROUND_UP(offset, BDRV_SECTOR_SIZE) > bs->total_sectors) { error_report("Wrong offset: offset=0x%" PRIx64 " total_sectors=0x%" PRIx64, @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vmdk_pwritev(BlockDriverState *bs, uint64_t offset, return 0; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vmdk_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ vmdk_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, return vmdk_co_pwritev(bs, offset, bytes, qiov, 0); } -static int coroutine_fn vmdk_co_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { int ret; BDRVVmdkState *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static VmdkExtentInfo *vmdk_get_extent_info(VmdkExtent *extent) return info; } -static int coroutine_fn vmdk_co_check(BlockDriverState *bs, - BdrvCheckResult *result, - BdrvCheckMode fix) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_check(BlockDriverState *bs, BdrvCheckResult *result, BdrvCheckMode fix) { BDRVVmdkState *s = bs->opaque; VmdkExtent *extent = NULL; diff --git a/block/vpc.c b/block/vpc.c index XXXXXXX..XXXXXXX 100644 --- a/block/vpc.c +++ b/block/vpc.c @@ -XXX,XX +XXX,XX @@ vpc_co_get_info(BlockDriverState *bs, BlockDriverInfo *bdi) return 0; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vpc_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK vpc_co_pwritev(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, BdrvRequestFlags flags) { diff --git a/tests/unit/test-bdrv-drain.c b/tests/unit/test-bdrv-drain.c index XXXXXXX..XXXXXXX 100644 --- a/tests/unit/test-bdrv-drain.c +++ b/tests/unit/test-bdrv-drain.c @@ -XXX,XX +XXX,XX @@ static void bdrv_test_top_close(BlockDriverState *bs) } } -static int coroutine_fn bdrv_test_top_co_preadv(BlockDriverState *bs, - int64_t offset, int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_test_top_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVTestTopState *tts = bs->opaque; return bdrv_co_preadv(tts->wait_child, offset, bytes, qiov, flags); @@ -XXX,XX +XXX,XX @@ static void coroutine_fn test_co_delete_by_drain(void *opaque) void *buffer = g_malloc(65536); QEMUIOVector qiov = QEMU_IOVEC_INIT_BUF(qiov, buffer, 65536); + GRAPH_RDLOCK_GUARD(); + /* Pretend some internal write operation from parent to child. * Important: We have to read from the child, not from the parent! * Draining works by first propagating it all up the tree to the @@ -XXX,XX +XXX,XX @@ static void bdrv_replace_test_close(BlockDriverState *bs) * Otherwise: * Set .has_read to true and return success. */ -static int coroutine_fn bdrv_replace_test_co_preadv(BlockDriverState *bs, - int64_t offset, - int64_t bytes, - QEMUIOVector *qiov, - BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_replace_test_co_preadv(BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, BdrvRequestFlags flags) { BDRVReplaceTestState *s = bs->opaque; @@ -XXX,XX +XXX,XX @@ static void coroutine_fn bdrv_replace_test_read_entry(void *opaque) int ret; /* Queue a read request post-drain */ + bdrv_graph_co_rdlock(); ret = bdrv_replace_test_co_preadv(bs, 0, 1, &qiov, 0); + bdrv_graph_co_rdunlock(); + g_assert(ret >= 0); bdrv_dec_in_flight(bs); } -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_pwrite_sync() need to hold a reader lock for the graph. For some places, we know that they will hold the lock, but we don't have the GRAPH_RDLOCK annotations yet. In this case, add assume_graph_lock() with a FIXME comment. These places will be removed once everything is properly annotated. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-13-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/qcow2.h | 2 +- include/block/block-io.h | 7 ++++--- block/io.c | 3 +-- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -XXX,XX +XXX,XX @@ int qcow2_inc_refcounts_imrt(BlockDriverState *bs, BdrvCheckResult *res, int qcow2_change_refcount_order(BlockDriverState *bs, int refcount_order, BlockDriverAmendStatusCB *status_cb, void *cb_opaque, Error **errp); -int coroutine_fn qcow2_shrink_reftable(BlockDriverState *bs); +int coroutine_fn GRAPH_RDLOCK qcow2_shrink_reftable(BlockDriverState *bs); int64_t qcow2_get_last_cluster(BlockDriverState *bs, int64_t size); int coroutine_fn qcow2_detect_metadata_preallocation(BlockDriverState *bs); diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int co_wrapper_mixed_bdrv_rdlock bdrv_pwrite_sync(BdrvChild *child, int64_t offset, int64_t bytes, const void *buf, BdrvRequestFlags flags); -int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, - int64_t bytes, const void *buf, - BdrvRequestFlags flags); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, int64_t bytes, + const void *buf, BdrvRequestFlags flags); + /* * Efficiently zero a region of the disk image. Note that this is a regular * I/O request like read or write and should have a reasonable size. This diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_pwrite_sync(BdrvChild *child, int64_t offset, { int ret; IO_CODE(); - - assume_graph_lock(); /* FIXME */ + assert_bdrv_graph_readable(); ret = bdrv_co_pwrite(child, offset, bytes, buf, flags); if (ret < 0) { -- 2.39.2 All callers are already GRAPH_RDLOCK, so just add the annotation and remove assume_graph_lock(). Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-14-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/io.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, - int64_t offset, int64_t bytes, BdrvRequestFlags flags) +static int coroutine_fn GRAPH_RDLOCK +bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, + BdrvRequestFlags flags) { BlockDriver *drv = bs->drv; QEMUIOVector qiov; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int head = 0; int tail = 0; - assume_graph_lock(); /* FIXME */ - int64_t max_write_zeroes = MIN_NON_ZERO(bs->bl.max_pwrite_zeroes, INT64_MAX); int alignment = MAX(bs->bl.pwrite_zeroes_alignment, -- 2.39.2 From: Emanuele Giuseppe Esposito <eesposit@redhat.com> This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_copy_range() need to hold a reader lock for the graph. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-15-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 9 +++++---- include/block/block_int-common.h | 24 ++++++++---------------- include/block/block_int-io.h | 20 ++++++++++---------- block/block-backend.c | 2 ++ block/file-posix.c | 16 +++++++--------- block/io.c | 7 +++++-- block/iscsi.c | 28 ++++++++++++---------------- block/qcow2.c | 5 ++--- block/raw-format.c | 28 ++++++++++++---------------- qemu-img.c | 4 +++- 10 files changed, 66 insertions(+), 77 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ bool co_wrapper bdrv_can_store_new_dirty_bitmap(BlockDriverState *bs, * * Returns: 0 if succeeded; negative error code if failed. **/ -int coroutine_fn bdrv_co_copy_range(BdrvChild *src, int64_t src_offset, - BdrvChild *dst, int64_t dst_offset, - int64_t bytes, BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_copy_range(BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags); /* * "I/O or GS" API functions. These functions can run without diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * See the comment of bdrv_co_copy_range for the parameter and return value * semantics. */ - int coroutine_fn (*bdrv_co_copy_range_from)(BlockDriverState *bs, - BdrvChild *src, - int64_t offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_copy_range_from)( + BlockDriverState *bs, BdrvChild *src, int64_t offset, + BdrvChild *dst, int64_t dst_offset, int64_t bytes, + BdrvRequestFlags read_flags, BdrvRequestFlags write_flags); /* * Map [offset, offset + nbytes) range onto a child of bs to copy data to, @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * See the comment of bdrv_co_copy_range for the parameter and return value * semantics. */ - int coroutine_fn (*bdrv_co_copy_range_to)(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_copy_range_to)( + BlockDriverState *bs, BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, int64_t bytes, + BdrvRequestFlags read_flags, BdrvRequestFlags write_flags); /* * Building block for bdrv_block_status[_above] and diff --git a/include/block/block_int-io.h b/include/block/block_int-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-io.h +++ b/include/block/block_int-io.h @@ -XXX,XX +XXX,XX @@ void bdrv_dirty_bitmap_merge_internal(BdrvDirtyBitmap *dest, void bdrv_inc_in_flight(BlockDriverState *bs); void bdrv_dec_in_flight(BlockDriverState *bs); -int coroutine_fn bdrv_co_copy_range_from(BdrvChild *src, int64_t src_offset, - BdrvChild *dst, int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags); -int coroutine_fn bdrv_co_copy_range_to(BdrvChild *src, int64_t src_offset, - BdrvChild *dst, int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_copy_range_from(BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_copy_range_to(BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags); int coroutine_fn bdrv_co_refresh_total_sectors(BlockDriverState *bs, int64_t hint); diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_copy_range(BlockBackend *blk_in, int64_t off_in, if (r) { return r; } + + GRAPH_RDLOCK_GUARD(); return bdrv_co_copy_range(blk_in->root, off_in, blk_out->root, off_out, bytes, read_flags, write_flags); diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ static void raw_abort_perm_update(BlockDriverState *bs) raw_handle_perm_lock(bs, RAW_PL_ABORT, 0, 0, NULL); } -static int coroutine_fn raw_co_copy_range_from( +static int coroutine_fn GRAPH_RDLOCK raw_co_copy_range_from( BlockDriverState *bs, BdrvChild *src, int64_t src_offset, BdrvChild *dst, int64_t dst_offset, int64_t bytes, BdrvRequestFlags read_flags, BdrvRequestFlags write_flags) @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_copy_range_from( read_flags, write_flags); } -static int coroutine_fn raw_co_copy_range_to(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_copy_range_to(BlockDriverState *bs, + BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags) { RawPosixAIOData acb; BDRVRawState *s = bs->opaque; diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn GRAPH_RDLOCK bdrv_co_copy_range_internal( { BdrvTrackedRequest req; int ret; + assert_bdrv_graph_readable(); /* TODO We can support BDRV_REQ_NO_FALLBACK here */ assert(!(read_flags & BDRV_REQ_NO_FALLBACK)); @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range_from(BdrvChild *src, int64_t src_offset, BdrvRequestFlags write_flags) { IO_CODE(); - assume_graph_lock(); /* FIXME */ + assert_bdrv_graph_readable(); trace_bdrv_co_copy_range_from(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); return bdrv_co_copy_range_internal(src, src_offset, dst, dst_offset, @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range_to(BdrvChild *src, int64_t src_offset, BdrvRequestFlags write_flags) { IO_CODE(); - assume_graph_lock(); /* FIXME */ + assert_bdrv_graph_readable(); trace_bdrv_co_copy_range_to(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); return bdrv_co_copy_range_internal(src, src_offset, dst, dst_offset, @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_copy_range(BdrvChild *src, int64_t src_offset, BdrvRequestFlags write_flags) { IO_CODE(); + assert_bdrv_graph_readable(); + return bdrv_co_copy_range_from(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); diff --git a/block/iscsi.c b/block/iscsi.c index XXXXXXX..XXXXXXX 100644 --- a/block/iscsi.c +++ b/block/iscsi.c @@ -XXX,XX +XXX,XX @@ static void coroutine_fn iscsi_co_invalidate_cache(BlockDriverState *bs, iscsi_allocmap_invalidate(iscsilun); } -static int coroutine_fn iscsi_co_copy_range_from(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags) +static int coroutine_fn GRAPH_RDLOCK +iscsi_co_copy_range_from(BlockDriverState *bs, + BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags) { return bdrv_co_copy_range_to(src, src_offset, dst, dst_offset, bytes, read_flags, write_flags); @@ -XXX,XX +XXX,XX @@ static void iscsi_xcopy_data(struct iscsi_data *data, src_lba, dst_lba); } -static int coroutine_fn iscsi_co_copy_range_to(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags) +static int coroutine_fn GRAPH_RDLOCK +iscsi_co_copy_range_to(BlockDriverState *bs, + BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags) { IscsiLun *dst_lun = dst->bs->opaque; IscsiLun *src_lun; diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pdiscard(BlockDriverState *bs, return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK qcow2_co_copy_range_from(BlockDriverState *bs, BdrvChild *src, int64_t src_offset, BdrvChild *dst, int64_t dst_offset, @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK qcow2_co_copy_range_to(BlockDriverState *bs, BdrvChild *src, int64_t src_offset, BdrvChild *dst, int64_t dst_offset, @@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_to(BlockDriverState *bs, uint64_t host_offset; QCowL2Meta *l2meta = NULL; - assume_graph_lock(); /* FIXME */ assert(!bs->encrypted); qemu_co_mutex_lock(&s->lock); diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static int raw_probe_geometry(BlockDriverState *bs, HDGeometry *geo) return bdrv_probe_geometry(bs->file->bs, geo); } -static int coroutine_fn raw_co_copy_range_from(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_copy_range_from(BlockDriverState *bs, + BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags) { int ret; @@ -XXX,XX +XXX,XX @@ static int coroutine_fn raw_co_copy_range_from(BlockDriverState *bs, bytes, read_flags, write_flags); } -static int coroutine_fn raw_co_copy_range_to(BlockDriverState *bs, - BdrvChild *src, - int64_t src_offset, - BdrvChild *dst, - int64_t dst_offset, - int64_t bytes, - BdrvRequestFlags read_flags, - BdrvRequestFlags write_flags) +static int coroutine_fn GRAPH_RDLOCK +raw_co_copy_range_to(BlockDriverState *bs, + BdrvChild *src, int64_t src_offset, + BdrvChild *dst, int64_t dst_offset, + int64_t bytes, BdrvRequestFlags read_flags, + BdrvRequestFlags write_flags) { int ret; diff --git a/qemu-img.c b/qemu-img.c index XXXXXXX..XXXXXXX 100644 --- a/qemu-img.c +++ b/qemu-img.c @@ -XXX,XX +XXX,XX @@ retry: if (s->ret == -EINPROGRESS) { if (copy_range) { - ret = convert_co_copy_range(s, sector_num, n); + WITH_GRAPH_RDLOCK_GUARD() { + ret = convert_co_copy_range(s, sector_num, n); + } if (ret) { s->copy_range = false; goto retry; -- 2.39.2 Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-16-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block_int-common.h | 12 +++++++----- include/block/block_int-io.h | 8 ++++---- block/copy-before-write.c | 6 ++---- block/io.c | 2 ++ block/snapshot-access.c | 4 ++-- 5 files changed, 17 insertions(+), 15 deletions(-) diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * - receive the snapshot's actual length (which may differ from bs's * length) */ - int coroutine_fn (*bdrv_co_preadv_snapshot)(BlockDriverState *bs, - int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset); - int coroutine_fn (*bdrv_co_snapshot_block_status)(BlockDriverState *bs, - bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, - int64_t *map, BlockDriverState **file); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_preadv_snapshot)( + BlockDriverState *bs, int64_t offset, int64_t bytes, + QEMUIOVector *qiov, size_t qiov_offset); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_snapshot_block_status)( + BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, + int64_t *pnum, int64_t *map, BlockDriverState **file); int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_pdiscard_snapshot)( BlockDriverState *bs, int64_t offset, int64_t bytes); diff --git a/include/block/block_int-io.h b/include/block/block_int-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-io.h +++ b/include/block/block_int-io.h @@ -XXX,XX +XXX,XX @@ * the I/O API. */ -int coroutine_fn bdrv_co_preadv_snapshot(BdrvChild *child, +int coroutine_fn GRAPH_RDLOCK bdrv_co_preadv_snapshot(BdrvChild *child, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset); -int coroutine_fn bdrv_co_snapshot_block_status(BlockDriverState *bs, - bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, - int64_t *map, BlockDriverState **file); +int coroutine_fn GRAPH_RDLOCK bdrv_co_snapshot_block_status( + BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, + int64_t *pnum, int64_t *map, BlockDriverState **file); int coroutine_fn GRAPH_RDLOCK bdrv_co_pdiscard_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes); diff --git a/block/copy-before-write.c b/block/copy-before-write.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-before-write.c +++ b/block/copy-before-write.c @@ -XXX,XX +XXX,XX @@ cbw_snapshot_read_unlock(BlockDriverState *bs, BlockReq *req) g_free(req); } -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK cbw_co_preadv_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset) { @@ -XXX,XX +XXX,XX @@ cbw_co_preadv_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes, BdrvChild *file; int ret; - assume_graph_lock(); /* FIXME */ - /* TODO: upgrade to async loop using AioTask */ while (bytes) { int64_t cur_bytes; @@ -XXX,XX +XXX,XX @@ cbw_co_preadv_snapshot(BlockDriverState *bs, int64_t offset, int64_t bytes, return 0; } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK cbw_co_snapshot_block_status(BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ bdrv_co_preadv_snapshot(BdrvChild *child, int64_t offset, int64_t bytes, BlockDriver *drv = bs->drv; int ret; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) { return -ENOMEDIUM; @@ -XXX,XX +XXX,XX @@ bdrv_co_snapshot_block_status(BlockDriverState *bs, BlockDriver *drv = bs->drv; int ret; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) { return -ENOMEDIUM; diff --git a/block/snapshot-access.c b/block/snapshot-access.c index XXXXXXX..XXXXXXX 100644 --- a/block/snapshot-access.c +++ b/block/snapshot-access.c @@ -XXX,XX +XXX,XX @@ #include "qemu/cutils.h" #include "block/block_int.h" -static coroutine_fn int +static int coroutine_fn GRAPH_RDLOCK snapshot_access_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, @@ -XXX,XX +XXX,XX @@ snapshot_access_co_preadv_part(BlockDriverState *bs, return bdrv_co_preadv_snapshot(bs->file, offset, bytes, qiov, qiov_offset); } -static int coroutine_fn +static int coroutine_fn GRAPH_RDLOCK snapshot_access_co_block_status(BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes, int64_t *pnum, -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_create() need to hold a reader lock for the graph. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-17-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-global-state.h | 14 ++++--- include/block/block_int-common.h | 11 +++--- block.c | 1 + block/create.c | 9 ++++- block/crypto.c | 7 ++-- block/file-posix.c | 7 ++-- block/file-win32.c | 7 ++-- block/parallels.c | 7 ++-- block/qcow.c | 6 +-- block/qcow2.c | 7 ++-- block/qed.c | 7 ++-- block/raw-format.c | 7 ++-- block/vdi.c | 7 ++-- block/vhdx.c | 7 ++-- block/vmdk.c | 63 ++++++++++++++---------------- block/vpc.c | 7 ++-- 16 files changed, 84 insertions(+), 90 deletions(-) diff --git a/include/block/block-global-state.h b/include/block/block-global-state.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-global-state.h +++ b/include/block/block-global-state.h @@ -XXX,XX +XXX,XX @@ BlockDriver *bdrv_find_protocol(const char *filename, Error **errp); BlockDriver *bdrv_find_format(const char *format_name); -int coroutine_fn bdrv_co_create(BlockDriver *drv, const char *filename, - QemuOpts *opts, Error **errp); -int co_wrapper bdrv_create(BlockDriver *drv, const char *filename, - QemuOpts *opts, Error **errp); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_create(BlockDriver *drv, const char *filename, QemuOpts *opts, + Error **errp); -int coroutine_fn bdrv_co_create_file(const char *filename, QemuOpts *opts, - Error **errp); +int co_wrapper_bdrv_rdlock bdrv_create(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp); + +int coroutine_fn GRAPH_RDLOCK +bdrv_co_create_file(const char *filename, QemuOpts *opts, Error **errp); BlockDriverState *bdrv_new(void); int bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top, diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { Error **errp); void (*bdrv_close)(BlockDriverState *bs); - int coroutine_fn (*bdrv_co_create)(BlockdevCreateOptions *opts, - Error **errp); - int coroutine_fn (*bdrv_co_create_opts)(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_create)( + BlockdevCreateOptions *opts, Error **errp); + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_create_opts)( + BlockDriver *drv, const char *filename, QemuOpts *opts, Error **errp); int (*bdrv_amend_options)(BlockDriverState *bs, QemuOpts *opts, diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_create(BlockDriver *drv, const char *filename, int ret; GLOBAL_STATE_CODE(); ERRP_GUARD(); + assert_bdrv_graph_readable(); if (!drv->bdrv_co_create_opts) { error_setg(errp, "Driver '%s' does not support image creation", diff --git a/block/create.c b/block/create.c index XXXXXXX..XXXXXXX 100644 --- a/block/create.c +++ b/block/create.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blockdev_create_run(Job *job, Error **errp) int ret; GLOBAL_STATE_CODE(); + GRAPH_RDLOCK_GUARD(); job_progress_set_remaining(&s->common, 1); ret = s->drv->bdrv_co_create(s->opts, errp); @@ -XXX,XX +XXX,XX @@ static const JobDriver blockdev_create_job_driver = { .run = blockdev_create_run, }; +/* Checking whether the function is present doesn't require the graph lock */ +static inline bool TSA_NO_TSA has_bdrv_co_create(BlockDriver *drv) +{ + return drv->bdrv_co_create; +} + void qmp_blockdev_create(const char *job_id, BlockdevCreateOptions *options, Error **errp) { @@ -XXX,XX +XXX,XX @@ void qmp_blockdev_create(const char *job_id, BlockdevCreateOptions *options, } /* Error out if the driver doesn't support .bdrv_co_create */ - if (!drv->bdrv_co_create) { + if (!has_bdrv_co_create(drv)) { error_setg(errp, "Driver does not support blockdev-create"); return; } diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int coroutine_fn block_crypto_co_create_opts_luks(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +block_crypto_co_create_opts_luks(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { QCryptoBlockCreateOptions *create_opts = NULL; BlockDriverState *bs = NULL; diff --git a/block/file-posix.c b/block/file-posix.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-posix.c +++ b/block/file-posix.c @@ -XXX,XX +XXX,XX @@ out: return result; } -static int coroutine_fn raw_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +raw_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions options; int64_t total_size = 0; diff --git a/block/file-win32.c b/block/file-win32.c index XXXXXXX..XXXXXXX 100644 --- a/block/file-win32.c +++ b/block/file-win32.c @@ -XXX,XX +XXX,XX @@ static int raw_co_create(BlockdevCreateOptions *options, Error **errp) return 0; } -static int coroutine_fn raw_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +raw_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions options; int64_t total_size = 0; diff --git a/block/parallels.c b/block/parallels.c index XXXXXXX..XXXXXXX 100644 --- a/block/parallels.c +++ b/block/parallels.c @@ -XXX,XX +XXX,XX @@ exit: goto out; } -static int coroutine_fn parallels_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +parallels_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions *create_options = NULL; BlockDriverState *bs = NULL; diff --git a/block/qcow.c b/block/qcow.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow.c +++ b/block/qcow.c @@ -XXX,XX +XXX,XX @@ exit: return ret; } -static int coroutine_fn qcow_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, Error **errp) +static int coroutine_fn GRAPH_RDLOCK +qcow_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions *create_options = NULL; BlockDriverState *bs = NULL; diff --git a/block/qcow2.c b/block/qcow2.c index XXXXXXX..XXXXXXX 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn qcow2_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +qcow2_co_create_opts(BlockDriver *drv, const char *filename, QemuOpts *opts, + Error **errp) { BlockdevCreateOptions *create_options = NULL; QDict *qdict; diff --git a/block/qed.c b/block/qed.c index XXXXXXX..XXXXXXX 100644 --- a/block/qed.c +++ b/block/qed.c @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn bdrv_qed_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +bdrv_qed_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions *create_options = NULL; QDict *qdict; diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ static int raw_has_zero_init(BlockDriverState *bs) return bdrv_has_zero_init(bs->file->bs); } -static int coroutine_fn raw_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +raw_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { return bdrv_co_create_file(filename, opts, errp); } diff --git a/block/vdi.c b/block/vdi.c index XXXXXXX..XXXXXXX 100644 --- a/block/vdi.c +++ b/block/vdi.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn vdi_co_create(BlockdevCreateOptions *create_options, return vdi_co_do_create(create_options, DEFAULT_CLUSTER_SIZE, errp); } -static int coroutine_fn vdi_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vdi_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { QDict *qdict = NULL; BlockdevCreateOptions *create_options = NULL; diff --git a/block/vhdx.c b/block/vhdx.c index XXXXXXX..XXXXXXX 100644 --- a/block/vhdx.c +++ b/block/vhdx.c @@ -XXX,XX +XXX,XX @@ delete_and_exit: return ret; } -static int coroutine_fn vhdx_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vhdx_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions *create_options = NULL; QDict *qdict; diff --git a/block/vmdk.c b/block/vmdk.c index XXXXXXX..XXXXXXX 100644 --- a/block/vmdk.c +++ b/block/vmdk.c @@ -XXX,XX +XXX,XX @@ exit: return ret; } -static int coroutine_fn vmdk_create_extent(const char *filename, - int64_t filesize, bool flat, - bool compress, bool zeroed_grain, - BlockBackend **pbb, - QemuOpts *opts, Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vmdk_create_extent(const char *filename, int64_t filesize, bool flat, + bool compress, bool zeroed_grain, BlockBackend **pbb, + QemuOpts *opts, Error **errp) { int ret; BlockBackend *blk = NULL; @@ -XXX,XX +XXX,XX @@ static int filename_decompose(const char *filename, char *path, char *prefix, * non-split format. * idx >= 1: get the n-th extent if in a split subformat */ -typedef BlockBackend * coroutine_fn (*vmdk_create_extent_fn)(int64_t size, - int idx, - bool flat, - bool split, - bool compress, - bool zeroed_grain, - void *opaque, - Error **errp); +typedef BlockBackend * coroutine_fn /* GRAPH_RDLOCK */ + (*vmdk_create_extent_fn)(int64_t size, int idx, bool flat, bool split, + bool compress, bool zeroed_grain, void *opaque, + Error **errp); static void vmdk_desc_add_extent(GString *desc, const char *extent_line_fmt, @@ -XXX,XX +XXX,XX @@ static void vmdk_desc_add_extent(GString *desc, g_free(basename); } -static int coroutine_fn vmdk_co_do_create(int64_t size, - BlockdevVmdkSubformat subformat, - BlockdevVmdkAdapterType adapter_type, - const char *backing_file, - const char *hw_version, - const char *toolsversion, - bool compat6, - bool zeroed_grain, - vmdk_create_extent_fn extent_fn, - void *opaque, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_do_create(int64_t size, + BlockdevVmdkSubformat subformat, + BlockdevVmdkAdapterType adapter_type, + const char *backing_file, + const char *hw_version, + const char *toolsversion, + bool compat6, + bool zeroed_grain, + vmdk_create_extent_fn extent_fn, + void *opaque, + Error **errp) { int extent_idx; BlockBackend *blk = NULL; @@ -XXX,XX +XXX,XX @@ typedef struct { QemuOpts *opts; } VMDKCreateOptsData; -static BlockBackend * coroutine_fn vmdk_co_create_opts_cb(int64_t size, int idx, - bool flat, bool split, bool compress, - bool zeroed_grain, void *opaque, - Error **errp) +static BlockBackend * coroutine_fn GRAPH_RDLOCK +vmdk_co_create_opts_cb(int64_t size, int idx, bool flat, bool split, + bool compress, bool zeroed_grain, void *opaque, + Error **errp) { BlockBackend *blk = NULL; BlockDriverState *bs = NULL; @@ -XXX,XX +XXX,XX @@ exit: return blk; } -static int coroutine_fn vmdk_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { Error *local_err = NULL; char *desc = NULL; @@ -XXX,XX +XXX,XX @@ static BlockBackend * coroutine_fn vmdk_co_create_cb(int64_t size, int idx, return blk; } -static int coroutine_fn vmdk_co_create(BlockdevCreateOptions *create_options, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vmdk_co_create(BlockdevCreateOptions *create_options, Error **errp) { BlockdevCreateOptionsVmdk *opts; diff --git a/block/vpc.c b/block/vpc.c index XXXXXXX..XXXXXXX 100644 --- a/block/vpc.c +++ b/block/vpc.c @@ -XXX,XX +XXX,XX @@ out: return ret; } -static int coroutine_fn vpc_co_create_opts(BlockDriver *drv, - const char *filename, - QemuOpts *opts, - Error **errp) +static int coroutine_fn GRAPH_RDLOCK +vpc_co_create_opts(BlockDriver *drv, const char *filename, + QemuOpts *opts, Error **errp) { BlockdevCreateOptions *create_options = NULL; QDict *qdict; -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_io_plug() and bdrv_co_io_unplug() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-18-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 4 ++-- include/block/block_int-common.h | 5 +++-- block/block-backend.c | 2 ++ block/io.c | 2 ++ 4 files changed, 9 insertions(+), 4 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_leave(BlockDriverState *bs, AioContext *old_ctx); AioContext *child_of_bds_get_parent_aio_context(BdrvChild *c); -void coroutine_fn bdrv_co_io_plug(BlockDriverState *bs); -void coroutine_fn bdrv_co_io_unplug(BlockDriverState *bs); +void coroutine_fn GRAPH_RDLOCK bdrv_co_io_plug(BlockDriverState *bs); +void coroutine_fn GRAPH_RDLOCK bdrv_co_io_unplug(BlockDriverState *bs); bool coroutine_fn bdrv_co_can_store_new_dirty_bitmap(BlockDriverState *bs, const char *name, diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlkdebugEvent event); /* io queue for linux-aio */ - void coroutine_fn (*bdrv_co_io_plug)(BlockDriverState *bs); - void coroutine_fn (*bdrv_co_io_unplug)(BlockDriverState *bs); + void coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_io_plug)(BlockDriverState *bs); + void coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_io_unplug)( + BlockDriverState *bs); /** * bdrv_drain_begin is called if implemented in the beginning of a diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ void coroutine_fn blk_co_io_plug(BlockBackend *blk) { BlockDriverState *bs = blk_bs(blk); IO_CODE(); + GRAPH_RDLOCK_GUARD(); if (bs) { bdrv_co_io_plug(bs); @@ -XXX,XX +XXX,XX @@ void coroutine_fn blk_co_io_unplug(BlockBackend *blk) { BlockDriverState *bs = blk_bs(blk); IO_CODE(); + GRAPH_RDLOCK_GUARD(); if (bs) { bdrv_co_io_unplug(bs); diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_io_plug(BlockDriverState *bs) { BdrvChild *child; IO_CODE(); + assert_bdrv_graph_readable(); QLIST_FOREACH(child, &bs->children, next) { bdrv_co_io_plug(child->bs); @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_io_unplug(BlockDriverState *bs) { BdrvChild *child; IO_CODE(); + assert_bdrv_graph_readable(); assert(bs->io_plugged); if (qatomic_fetch_dec(&bs->io_plugged) == 1) { -- 2.39.2 From: Emanuele Giuseppe Esposito <eesposit@redhat.com> This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_is_inserted() need to hold a reader lock for the graph. blk_is_inserted() is done as a co_wrapper_mixed_bdrv_rdlock (unlike most other blk_* functions) because it is called a lot from other blk_co_*() functions that already hold the lock. These calls go through blk_is_available(), which becomes a co_wrapper_mixed_bdrv_rdlock, too, for the same reason. Functions that run in a coroutine and can call bdrv_co_is_available() directly are changed to do so, which results in better TSA coverage. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-19-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 4 ++-- include/block/block_int-common.h | 3 ++- include/sysemu/block-backend-io.h | 7 ++++--- block.c | 1 + block/block-backend.c | 25 ++++++++++++++----------- 5 files changed, 23 insertions(+), 17 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ bool bdrv_is_writable(BlockDriverState *bs); bool bdrv_is_sg(BlockDriverState *bs); int bdrv_get_flags(BlockDriverState *bs); -bool coroutine_fn bdrv_co_is_inserted(BlockDriverState *bs); -bool co_wrapper bdrv_is_inserted(BlockDriverState *bs); +bool coroutine_fn GRAPH_RDLOCK bdrv_co_is_inserted(BlockDriverState *bs); +bool co_wrapper_bdrv_rdlock bdrv_is_inserted(BlockDriverState *bs); void coroutine_fn bdrv_co_lock_medium(BlockDriverState *bs, bool locked); void coroutine_fn bdrv_co_eject(BlockDriverState *bs, bool eject_flag); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos); /* removable device specific */ - bool coroutine_fn (*bdrv_co_is_inserted)(BlockDriverState *bs); + bool coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_is_inserted)( + BlockDriverState *bs); void coroutine_fn (*bdrv_co_eject)(BlockDriverState *bs, bool eject_flag); void coroutine_fn (*bdrv_co_lock_medium)(BlockDriverState *bs, bool locked); diff --git a/include/sysemu/block-backend-io.h b/include/sysemu/block-backend-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/sysemu/block-backend-io.h +++ b/include/sysemu/block-backend-io.h @@ -XXX,XX +XXX,XX @@ BlockAIOCB *blk_aio_ioctl(BlockBackend *blk, unsigned long int req, void *buf, void blk_inc_in_flight(BlockBackend *blk); void blk_dec_in_flight(BlockBackend *blk); -bool coroutine_fn blk_co_is_inserted(BlockBackend *blk); -bool co_wrapper_mixed blk_is_inserted(BlockBackend *blk); +bool coroutine_fn GRAPH_RDLOCK blk_co_is_inserted(BlockBackend *blk); +bool co_wrapper_mixed_bdrv_rdlock blk_is_inserted(BlockBackend *blk); -bool blk_is_available(BlockBackend *blk); +bool coroutine_fn GRAPH_RDLOCK blk_co_is_available(BlockBackend *blk); +bool co_wrapper_mixed_bdrv_rdlock blk_is_available(BlockBackend *blk); void coroutine_fn blk_co_lock_medium(BlockBackend *blk, bool locked); void co_wrapper blk_lock_medium(BlockBackend *blk, bool locked); diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ bool coroutine_fn bdrv_co_is_inserted(BlockDriverState *bs) BlockDriver *drv = bs->drv; BdrvChild *child; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) { return false; diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ void blk_set_disable_request_queuing(BlockBackend *blk, bool disable) blk->disable_request_queuing = disable; } -static coroutine_fn int blk_check_byte_request(BlockBackend *blk, - int64_t offset, int64_t bytes) +static int coroutine_fn GRAPH_RDLOCK +blk_check_byte_request(BlockBackend *blk, int64_t offset, int64_t bytes) { int64_t len; @@ -XXX,XX +XXX,XX @@ static coroutine_fn int blk_check_byte_request(BlockBackend *blk, return -EIO; } - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ BlockAIOCB *blk_aio_pwrite_zeroes(BlockBackend *blk, int64_t offset, int64_t coroutine_fn blk_co_getlength(BlockBackend *blk) { IO_CODE(); + GRAPH_RDLOCK_GUARD(); - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ void blk_get_geometry(BlockBackend *blk, uint64_t *nb_sectors_ptr) int64_t coroutine_fn blk_co_nb_sectors(BlockBackend *blk) { IO_CODE(); + GRAPH_RDLOCK_GUARD(); - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ blk_co_do_ioctl(BlockBackend *blk, unsigned long int req, void *buf) blk_wait_while_drained(blk); GRAPH_RDLOCK_GUARD(); - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ static int coroutine_fn blk_co_do_flush(BlockBackend *blk) blk_wait_while_drained(blk); GRAPH_RDLOCK_GUARD(); - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ bool coroutine_fn blk_co_is_inserted(BlockBackend *blk) { BlockDriverState *bs = blk_bs(blk); IO_CODE(); + assert_bdrv_graph_readable(); return bs && bdrv_co_is_inserted(bs); } -bool blk_is_available(BlockBackend *blk) +bool coroutine_fn blk_co_is_available(BlockBackend *blk) { IO_CODE(); - return blk_is_inserted(blk) && !blk_dev_is_tray_open(blk); + return blk_co_is_inserted(blk) && !blk_dev_is_tray_open(blk); } void coroutine_fn blk_co_lock_medium(BlockBackend *blk, bool locked) @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_truncate(BlockBackend *blk, int64_t offset, bool exact, { IO_OR_GS_CODE(); GRAPH_RDLOCK_GUARD(); - if (!blk_is_available(blk)) { + if (!blk_co_is_available(blk)) { error_setg(errp, "No medium inserted"); return -ENOMEDIUM; } @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_copy_range(BlockBackend *blk_in, int64_t off_in, { int r; IO_CODE(); + GRAPH_RDLOCK_GUARD(); r = blk_check_byte_request(blk_in, off_in, bytes); if (r) { @@ -XXX,XX +XXX,XX @@ int coroutine_fn blk_co_copy_range(BlockBackend *blk_in, int64_t off_in, return r; } - GRAPH_RDLOCK_GUARD(); return bdrv_co_copy_range(blk_in->root, off_in, blk_out->root, off_out, bytes, read_flags, write_flags); -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_eject() and bdrv_co_lock_medium() need to hold a reader lock for the graph. Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-20-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 7 +++++-- include/block/block_int-common.h | 6 ++++-- block.c | 2 ++ block/block-backend.c | 2 ++ block/copy-on-read.c | 6 ++++-- block/filter-compress.c | 4 ++-- block/raw-format.c | 6 ++++-- 7 files changed, 23 insertions(+), 10 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int bdrv_get_flags(BlockDriverState *bs); bool coroutine_fn GRAPH_RDLOCK bdrv_co_is_inserted(BlockDriverState *bs); bool co_wrapper_bdrv_rdlock bdrv_is_inserted(BlockDriverState *bs); -void coroutine_fn bdrv_co_lock_medium(BlockDriverState *bs, bool locked); -void coroutine_fn bdrv_co_eject(BlockDriverState *bs, bool eject_flag); +void coroutine_fn GRAPH_RDLOCK +bdrv_co_lock_medium(BlockDriverState *bs, bool locked); + +void coroutine_fn GRAPH_RDLOCK +bdrv_co_eject(BlockDriverState *bs, bool eject_flag); const char *bdrv_get_format_name(BlockDriverState *bs); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { /* removable device specific */ bool coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_is_inserted)( BlockDriverState *bs); - void coroutine_fn (*bdrv_co_eject)(BlockDriverState *bs, bool eject_flag); - void coroutine_fn (*bdrv_co_lock_medium)(BlockDriverState *bs, bool locked); + void coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_eject)( + BlockDriverState *bs, bool eject_flag); + void coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_lock_medium)( + BlockDriverState *bs, bool locked); /* to control generic scsi devices */ BlockAIOCB *coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_aio_ioctl)( diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_eject(BlockDriverState *bs, bool eject_flag) { BlockDriver *drv = bs->drv; IO_CODE(); + assert_bdrv_graph_readable(); if (drv && drv->bdrv_co_eject) { drv->bdrv_co_eject(bs, eject_flag); @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_lock_medium(BlockDriverState *bs, bool locked) { BlockDriver *drv = bs->drv; IO_CODE(); + assert_bdrv_graph_readable(); trace_bdrv_lock_medium(bs, locked); if (drv && drv->bdrv_co_lock_medium) { diff --git a/block/block-backend.c b/block/block-backend.c index XXXXXXX..XXXXXXX 100644 --- a/block/block-backend.c +++ b/block/block-backend.c @@ -XXX,XX +XXX,XX @@ void coroutine_fn blk_co_lock_medium(BlockBackend *blk, bool locked) { BlockDriverState *bs = blk_bs(blk); IO_CODE(); + GRAPH_RDLOCK_GUARD(); if (bs) { bdrv_co_lock_medium(bs, locked); @@ -XXX,XX +XXX,XX @@ void coroutine_fn blk_co_eject(BlockBackend *blk, bool eject_flag) BlockDriverState *bs = blk_bs(blk); char *id; IO_CODE(); + GRAPH_RDLOCK_GUARD(); if (bs) { bdrv_co_eject(bs, eject_flag); diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ cor_co_pwritev_compressed(BlockDriverState *bs, int64_t offset, int64_t bytes, } -static void coroutine_fn cor_co_eject(BlockDriverState *bs, bool eject_flag) +static void coroutine_fn GRAPH_RDLOCK +cor_co_eject(BlockDriverState *bs, bool eject_flag) { bdrv_co_eject(bs->file->bs, eject_flag); } -static void coroutine_fn cor_co_lock_medium(BlockDriverState *bs, bool locked) +static void coroutine_fn GRAPH_RDLOCK +cor_co_lock_medium(BlockDriverState *bs, bool locked) { bdrv_co_lock_medium(bs->file->bs, locked); } diff --git a/block/filter-compress.c b/block/filter-compress.c index XXXXXXX..XXXXXXX 100644 --- a/block/filter-compress.c +++ b/block/filter-compress.c @@ -XXX,XX +XXX,XX @@ static void compress_refresh_limits(BlockDriverState *bs, Error **errp) } -static void coroutine_fn +static void coroutine_fn GRAPH_RDLOCK compress_co_eject(BlockDriverState *bs, bool eject_flag) { bdrv_co_eject(bs->file->bs, eject_flag); } -static void coroutine_fn +static void coroutine_fn GRAPH_RDLOCK compress_co_lock_medium(BlockDriverState *bs, bool locked) { bdrv_co_lock_medium(bs->file->bs, locked); diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ raw_co_truncate(BlockDriverState *bs, int64_t offset, bool exact, return bdrv_co_truncate(bs->file, offset, exact, prealloc, flags, errp); } -static void coroutine_fn raw_co_eject(BlockDriverState *bs, bool eject_flag) +static void coroutine_fn GRAPH_RDLOCK +raw_co_eject(BlockDriverState *bs, bool eject_flag) { bdrv_co_eject(bs->file->bs, eject_flag); } -static void coroutine_fn raw_co_lock_medium(BlockDriverState *bs, bool locked) +static void coroutine_fn GRAPH_RDLOCK +raw_co_lock_medium(BlockDriverState *bs, bool locked) { bdrv_co_lock_medium(bs->file->bs, locked); } -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_register_buf() and bdrv_unregister_buf() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-21-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block_int-common.h | 7 ++++--- block/io.c | 14 ++++++++++---- 2 files changed, 14 insertions(+), 7 deletions(-) diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { * * Returns: true on success, false on failure */ - bool (*bdrv_register_buf)(BlockDriverState *bs, void *host, size_t size, - Error **errp); - void (*bdrv_unregister_buf)(BlockDriverState *bs, void *host, size_t size); + bool GRAPH_RDLOCK_PTR (*bdrv_register_buf)( + BlockDriverState *bs, void *host, size_t size, Error **errp); + void GRAPH_RDLOCK_PTR (*bdrv_unregister_buf)( + BlockDriverState *bs, void *host, size_t size); /* * This field is modified only under the BQL, and is part of diff --git a/block/io.c b/block/io.c index XXXXXXX..XXXXXXX 100644 --- a/block/io.c +++ b/block/io.c @@ -XXX,XX +XXX,XX @@ void coroutine_fn bdrv_co_io_unplug(BlockDriverState *bs) } /* Helper that undoes bdrv_register_buf() when it fails partway through */ -static void bdrv_register_buf_rollback(BlockDriverState *bs, - void *host, - size_t size, - BdrvChild *final_child) +static void GRAPH_RDLOCK +bdrv_register_buf_rollback(BlockDriverState *bs, void *host, size_t size, + BdrvChild *final_child) { BdrvChild *child; + GLOBAL_STATE_CODE(); + assert_bdrv_graph_readable(); + QLIST_FOREACH(child, &bs->children, next) { if (child == final_child) { break; @@ -XXX,XX +XXX,XX @@ bool bdrv_register_buf(BlockDriverState *bs, void *host, size_t size, BdrvChild *child; GLOBAL_STATE_CODE(); + GRAPH_RDLOCK_GUARD_MAINLOOP(); + if (bs->drv && bs->drv->bdrv_register_buf) { if (!bs->drv->bdrv_register_buf(bs, host, size, errp)) { return false; @@ -XXX,XX +XXX,XX @@ void bdrv_unregister_buf(BlockDriverState *bs, void *host, size_t size) BdrvChild *child; GLOBAL_STATE_CODE(); + GRAPH_RDLOCK_GUARD_MAINLOOP(); + if (bs->drv && bs->drv->bdrv_unregister_buf) { bs->drv->bdrv_unregister_buf(bs, host, size); } -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_delete_file() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-22-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 8 ++++++-- include/block/block_int-common.h | 4 ++-- block.c | 1 + 3 files changed, 9 insertions(+), 4 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int64_t co_wrapper bdrv_get_allocated_file_size(BlockDriverState *bs); BlockMeasureInfo *bdrv_measure(BlockDriver *drv, QemuOpts *opts, BlockDriverState *in_bs, Error **errp); void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr); -int coroutine_fn bdrv_co_delete_file(BlockDriverState *bs, Error **errp); -void coroutine_fn bdrv_co_delete_file_noerr(BlockDriverState *bs); + +int coroutine_fn GRAPH_RDLOCK +bdrv_co_delete_file(BlockDriverState *bs, Error **errp); + +void coroutine_fn GRAPH_RDLOCK +bdrv_co_delete_file_noerr(BlockDriverState *bs); /* async block I/O */ diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_flush)(BlockDriverState *bs); /* Delete a created file. */ - int coroutine_fn (*bdrv_co_delete_file)(BlockDriverState *bs, - Error **errp); + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_delete_file)( + BlockDriverState *bs, Error **errp); /* * Flushes all data that was already written to the OS all the way down to diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_delete_file(BlockDriverState *bs, Error **errp) IO_CODE(); assert(bs != NULL); + assert_bdrv_graph_readable(); if (!bs->drv) { error_setg(errp, "Block node '%s' is not opened", bs->filename); -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_*_dirty_bitmap() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-23-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 14 ++++++-------- include/block/block_int-common.h | 6 ++++-- include/block/dirty-bitmap.h | 12 ++++++------ block/dirty-bitmap.c | 2 ++ 4 files changed, 18 insertions(+), 16 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ AioContext *child_of_bds_get_parent_aio_context(BdrvChild *c); void coroutine_fn GRAPH_RDLOCK bdrv_co_io_plug(BlockDriverState *bs); void coroutine_fn GRAPH_RDLOCK bdrv_co_io_unplug(BlockDriverState *bs); -bool coroutine_fn bdrv_co_can_store_new_dirty_bitmap(BlockDriverState *bs, - const char *name, - uint32_t granularity, - Error **errp); -bool co_wrapper bdrv_can_store_new_dirty_bitmap(BlockDriverState *bs, - const char *name, - uint32_t granularity, - Error **errp); +bool coroutine_fn GRAPH_RDLOCK +bdrv_co_can_store_new_dirty_bitmap(BlockDriverState *bs, const char *name, + uint32_t granularity, Error **errp); +bool co_wrapper_bdrv_rdlock +bdrv_can_store_new_dirty_bitmap(BlockDriverState *bs, const char *name, + uint32_t granularity, Error **errp); /** * diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { void (*bdrv_drain_end)(BlockDriverState *bs); bool (*bdrv_supports_persistent_dirty_bitmap)(BlockDriverState *bs); - bool coroutine_fn (*bdrv_co_can_store_new_dirty_bitmap)( + + bool coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_can_store_new_dirty_bitmap)( BlockDriverState *bs, const char *name, uint32_t granularity, Error **errp); - int coroutine_fn (*bdrv_co_remove_persistent_dirty_bitmap)( + + int coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_remove_persistent_dirty_bitmap)( BlockDriverState *bs, const char *name, Error **errp); }; diff --git a/include/block/dirty-bitmap.h b/include/block/dirty-bitmap.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/dirty-bitmap.h +++ b/include/block/dirty-bitmap.h @@ -XXX,XX +XXX,XX @@ int bdrv_dirty_bitmap_check(const BdrvDirtyBitmap *bitmap, uint32_t flags, void bdrv_release_dirty_bitmap(BdrvDirtyBitmap *bitmap); void bdrv_release_named_dirty_bitmaps(BlockDriverState *bs); -int coroutine_fn bdrv_co_remove_persistent_dirty_bitmap(BlockDriverState *bs, - const char *name, - Error **errp); -int co_wrapper bdrv_remove_persistent_dirty_bitmap(BlockDriverState *bs, - const char *name, - Error **errp); +int coroutine_fn GRAPH_RDLOCK +bdrv_co_remove_persistent_dirty_bitmap(BlockDriverState *bs, const char *name, + Error **errp); +int co_wrapper_bdrv_rdlock +bdrv_remove_persistent_dirty_bitmap(BlockDriverState *bs, const char *name, + Error **errp); void bdrv_disable_dirty_bitmap(BdrvDirtyBitmap *bitmap); void bdrv_enable_dirty_bitmap(BdrvDirtyBitmap *bitmap); diff --git a/block/dirty-bitmap.c b/block/dirty-bitmap.c index XXXXXXX..XXXXXXX 100644 --- a/block/dirty-bitmap.c +++ b/block/dirty-bitmap.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_remove_persistent_dirty_bitmap(BlockDriverState *bs, const char *name, Error **errp) { + assert_bdrv_graph_readable(); if (bs->drv && bs->drv->bdrv_co_remove_persistent_dirty_bitmap) { return bs->drv->bdrv_co_remove_persistent_dirty_bitmap(bs, name, errp); } @@ -XXX,XX +XXX,XX @@ bdrv_co_can_store_new_dirty_bitmap(BlockDriverState *bs, const char *name, uint32_t granularity, Error **errp) { BlockDriver *drv = bs->drv; + assert_bdrv_graph_readable(); if (!drv) { error_setg_errno(errp, ENOMEDIUM, -- 2.39.2 This adds GRAPH_RDLOCK annotations to declare that callers of bdrv_co_refresh_total_sectors() need to hold a reader lock for the graph. Signed-off-by: Kevin Wolf <kwolf@redhat.com> Message-Id: <20230203152202.49054-24-kwolf@redhat.com> Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/block/block-io.h | 8 ++++---- include/block/block_int-common.h | 4 +++- include/block/block_int-io.h | 7 ++++--- block.c | 3 +++ block/blkdebug.c | 3 ++- block/blklogwrites.c | 3 ++- block/blkreplay.c | 3 ++- block/blkverify.c | 3 ++- block/copy-on-read.c | 2 +- block/crypto.c | 3 ++- block/filter-compress.c | 3 ++- block/mirror.c | 3 +++ block/preallocate.c | 3 ++- block/quorum.c | 3 ++- block/raw-format.c | 3 ++- block/replication.c | 3 ++- block/stream.c | 8 +++++--- block/throttle.c | 3 ++- 18 files changed, 45 insertions(+), 23 deletions(-) diff --git a/include/block/block-io.h b/include/block/block-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block-io.h +++ b/include/block/block-io.h @@ -XXX,XX +XXX,XX @@ int coroutine_fn GRAPH_RDLOCK bdrv_co_truncate(BdrvChild *child, int64_t offset, bool exact, PreallocMode prealloc, BdrvRequestFlags flags, Error **errp); -int64_t coroutine_fn bdrv_co_nb_sectors(BlockDriverState *bs); -int64_t co_wrapper_mixed bdrv_nb_sectors(BlockDriverState *bs); +int64_t coroutine_fn GRAPH_RDLOCK bdrv_co_nb_sectors(BlockDriverState *bs); +int64_t co_wrapper_mixed_bdrv_rdlock bdrv_nb_sectors(BlockDriverState *bs); -int64_t coroutine_fn bdrv_co_getlength(BlockDriverState *bs); -int64_t co_wrapper_mixed bdrv_getlength(BlockDriverState *bs); +int64_t coroutine_fn GRAPH_RDLOCK bdrv_co_getlength(BlockDriverState *bs); +int64_t co_wrapper_mixed_bdrv_rdlock bdrv_getlength(BlockDriverState *bs); int64_t coroutine_fn bdrv_co_get_allocated_file_size(BlockDriverState *bs); int64_t co_wrapper bdrv_get_allocated_file_size(BlockDriverState *bs); diff --git a/include/block/block_int-common.h b/include/block/block_int-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-common.h +++ b/include/block/block_int-common.h @@ -XXX,XX +XXX,XX @@ struct BlockDriver { BlockDriverState *bs, int64_t offset, bool exact, PreallocMode prealloc, BdrvRequestFlags flags, Error **errp); - int64_t coroutine_fn (*bdrv_co_getlength)(BlockDriverState *bs); + int64_t coroutine_fn GRAPH_RDLOCK_PTR (*bdrv_co_getlength)( + BlockDriverState *bs); + int64_t coroutine_fn (*bdrv_co_get_allocated_file_size)( BlockDriverState *bs); diff --git a/include/block/block_int-io.h b/include/block/block_int-io.h index XXXXXXX..XXXXXXX 100644 --- a/include/block/block_int-io.h +++ b/include/block/block_int-io.h @@ -XXX,XX +XXX,XX @@ bdrv_co_copy_range_to(BdrvChild *src, int64_t src_offset, int64_t bytes, BdrvRequestFlags read_flags, BdrvRequestFlags write_flags); -int coroutine_fn bdrv_co_refresh_total_sectors(BlockDriverState *bs, - int64_t hint); -int co_wrapper_mixed +int coroutine_fn GRAPH_RDLOCK +bdrv_co_refresh_total_sectors(BlockDriverState *bs, int64_t hint); + +int co_wrapper_mixed_bdrv_rdlock bdrv_refresh_total_sectors(BlockDriverState *bs, int64_t hint); BdrvChild *bdrv_cow_child(BlockDriverState *bs); diff --git a/block.c b/block.c index XXXXXXX..XXXXXXX 100644 --- a/block.c +++ b/block.c @@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_refresh_total_sectors(BlockDriverState *bs, { BlockDriver *drv = bs->drv; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) { return -ENOMEDIUM; @@ -XXX,XX +XXX,XX @@ int64_t coroutine_fn bdrv_co_nb_sectors(BlockDriverState *bs) { BlockDriver *drv = bs->drv; IO_CODE(); + assert_bdrv_graph_readable(); if (!drv) return -ENOMEDIUM; @@ -XXX,XX +XXX,XX @@ int64_t coroutine_fn bdrv_co_getlength(BlockDriverState *bs) { int64_t ret; IO_CODE(); + assert_bdrv_graph_readable(); ret = bdrv_co_nb_sectors(bs); if (ret < 0) { diff --git a/block/blkdebug.c b/block/blkdebug.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkdebug.c +++ b/block/blkdebug.c @@ -XXX,XX +XXX,XX @@ static bool blkdebug_debug_is_suspended(BlockDriverState *bs, const char *tag) return false; } -static int64_t coroutine_fn blkdebug_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +blkdebug_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/blklogwrites.c b/block/blklogwrites.c index XXXXXXX..XXXXXXX 100644 --- a/block/blklogwrites.c +++ b/block/blklogwrites.c @@ -XXX,XX +XXX,XX @@ static void blk_log_writes_close(BlockDriverState *bs) s->log_file = NULL; } -static int64_t coroutine_fn blk_log_writes_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +blk_log_writes_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/blkreplay.c b/block/blkreplay.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkreplay.c +++ b/block/blkreplay.c @@ -XXX,XX +XXX,XX @@ fail: return ret; } -static int64_t coroutine_fn blkreplay_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +blkreplay_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/blkverify.c b/block/blkverify.c index XXXXXXX..XXXXXXX 100644 --- a/block/blkverify.c +++ b/block/blkverify.c @@ -XXX,XX +XXX,XX @@ static void blkverify_close(BlockDriverState *bs) s->test_file = NULL; } -static int64_t coroutine_fn blkverify_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +blkverify_co_getlength(BlockDriverState *bs) { BDRVBlkverifyState *s = bs->opaque; diff --git a/block/copy-on-read.c b/block/copy-on-read.c index XXXXXXX..XXXXXXX 100644 --- a/block/copy-on-read.c +++ b/block/copy-on-read.c @@ -XXX,XX +XXX,XX @@ static void cor_child_perm(BlockDriverState *bs, BdrvChild *c, } -static int64_t coroutine_fn cor_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK cor_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/crypto.c b/block/crypto.c index XXXXXXX..XXXXXXX 100644 --- a/block/crypto.c +++ b/block/crypto.c @@ -XXX,XX +XXX,XX @@ static void block_crypto_refresh_limits(BlockDriverState *bs, Error **errp) } -static int64_t coroutine_fn block_crypto_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +block_crypto_co_getlength(BlockDriverState *bs) { BlockCrypto *crypto = bs->opaque; int64_t len = bdrv_co_getlength(bs->file->bs); diff --git a/block/filter-compress.c b/block/filter-compress.c index XXXXXXX..XXXXXXX 100644 --- a/block/filter-compress.c +++ b/block/filter-compress.c @@ -XXX,XX +XXX,XX @@ static int compress_open(BlockDriverState *bs, QDict *options, int flags, } -static int64_t coroutine_fn compress_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +compress_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/mirror.c b/block/mirror.c index XXXXXXX..XXXXXXX 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn mirror_run(Job *job, Error **errp) goto immediate_exit; } + bdrv_graph_co_rdlock(); s->bdev_length = bdrv_co_getlength(bs); + bdrv_graph_co_rdunlock(); + if (s->bdev_length < 0) { ret = s->bdev_length; goto immediate_exit; diff --git a/block/preallocate.c b/block/preallocate.c index XXXXXXX..XXXXXXX 100644 --- a/block/preallocate.c +++ b/block/preallocate.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn GRAPH_RDLOCK preallocate_co_flush(BlockDriverState *bs) return bdrv_co_flush(bs->file->bs); } -static int64_t coroutine_fn preallocate_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +preallocate_co_getlength(BlockDriverState *bs) { int64_t ret; BDRVPreallocateState *s = bs->opaque; diff --git a/block/quorum.c b/block/quorum.c index XXXXXXX..XXXXXXX 100644 --- a/block/quorum.c +++ b/block/quorum.c @@ -XXX,XX +XXX,XX @@ quorum_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes, flags | BDRV_REQ_ZERO_WRITE); } -static int64_t coroutine_fn quorum_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +quorum_co_getlength(BlockDriverState *bs) { BDRVQuorumState *s = bs->opaque; int64_t result; diff --git a/block/raw-format.c b/block/raw-format.c index XXXXXXX..XXXXXXX 100644 --- a/block/raw-format.c +++ b/block/raw-format.c @@ -XXX,XX +XXX,XX @@ raw_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes) return bdrv_co_pdiscard(bs->file, offset, bytes); } -static int64_t coroutine_fn raw_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +raw_co_getlength(BlockDriverState *bs) { int64_t len; BDRVRawState *s = bs->opaque; diff --git a/block/replication.c b/block/replication.c index XXXXXXX..XXXXXXX 100644 --- a/block/replication.c +++ b/block/replication.c @@ -XXX,XX +XXX,XX @@ static void replication_child_perm(BlockDriverState *bs, BdrvChild *c, return; } -static int64_t coroutine_fn replication_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +replication_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } diff --git a/block/stream.c b/block/stream.c index XXXXXXX..XXXXXXX 100644 --- a/block/stream.c +++ b/block/stream.c @@ -XXX,XX +XXX,XX @@ static int coroutine_fn stream_run(Job *job, Error **errp) return 0; } - len = bdrv_getlength(s->target_bs); - if (len < 0) { - return len; + WITH_GRAPH_RDLOCK_GUARD() { + len = bdrv_co_getlength(s->target_bs); + if (len < 0) { + return len; + } } job_progress_set_remaining(&s->common.job, len); diff --git a/block/throttle.c b/block/throttle.c index XXXXXXX..XXXXXXX 100644 --- a/block/throttle.c +++ b/block/throttle.c @@ -XXX,XX +XXX,XX @@ static void throttle_close(BlockDriverState *bs) } -static int64_t coroutine_fn throttle_co_getlength(BlockDriverState *bs) +static int64_t coroutine_fn GRAPH_RDLOCK +throttle_co_getlength(BlockDriverState *bs) { return bdrv_co_getlength(bs->file->bs); } -- 2.39.2 From: Stefan Hajnoczi <stefanha@redhat.com> If requests are being processed in the IOThread when a SCSIDevice is unplugged, scsi_device_purge_requests() -> scsi_req_cancel_async() races with I/O completion callbacks. Both threads load and store req->aiocb. This can lead to assert(r->req.aiocb == NULL) failures and undefined behavior. Protect r->req.aiocb with the AioContext lock to prevent the race. Reviewed-by: Eric Blake <eblake@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20230221212218.1378734-2-stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- hw/scsi/scsi-disk.c | 23 ++++++++++++++++------- hw/scsi/scsi-generic.c | 11 ++++++----- 2 files changed, 22 insertions(+), 12 deletions(-) diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c index XXXXXXX..XXXXXXX 100644 --- a/hw/scsi/scsi-disk.c +++ b/hw/scsi/scsi-disk.c @@ -XXX,XX +XXX,XX @@ static void scsi_aio_complete(void *opaque, int ret) SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + if (scsi_disk_req_check_error(r, ret, true)) { goto done; } @@ -XXX,XX +XXX,XX @@ static void scsi_dma_complete(void *opaque, int ret) SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); if (ret < 0) { block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct); } else { @@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void *opaque, int ret) SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); if (ret < 0) { block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct); } else { @@ -XXX,XX +XXX,XX @@ static void scsi_do_read_cb(void *opaque, int ret) SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert (r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); if (ret < 0) { block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct); } else { @@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret) SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert (r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); if (ret < 0) { block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct); } else { @@ -XXX,XX +XXX,XX @@ static void scsi_unmap_complete(void *opaque, int ret) SCSIDiskReq *r = data->r; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); if (scsi_disk_req_check_error(r, ret, true)) { scsi_req_unref(&r->req); g_free(data); @@ -XXX,XX +XXX,XX @@ static void scsi_write_same_complete(void *opaque, int ret) SCSIDiskReq *r = data->r; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); + aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); + if (scsi_disk_req_check_error(r, ret, true)) { goto done; } diff --git a/hw/scsi/scsi-generic.c b/hw/scsi/scsi-generic.c index XXXXXXX..XXXXXXX 100644 --- a/hw/scsi/scsi-generic.c +++ b/hw/scsi/scsi-generic.c @@ -XXX,XX +XXX,XX @@ static void scsi_command_complete(void *opaque, int ret) SCSIGenericReq *r = (SCSIGenericReq *)opaque; SCSIDevice *s = r->req.dev; + aio_context_acquire(blk_get_aio_context(s->conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->conf.blk)); scsi_command_complete_noio(r, ret); aio_context_release(blk_get_aio_context(s->conf.blk)); } @@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret) SCSIDevice *s = r->req.dev; int len; + aio_context_acquire(blk_get_aio_context(s->conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->conf.blk)); - if (ret || r->req.io_canceled) { scsi_command_complete_noio(r, ret); goto done; @@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret) trace_scsi_generic_write_complete(ret); + aio_context_acquire(blk_get_aio_context(s->conf.blk)); + assert(r->req.aiocb != NULL); r->req.aiocb = NULL; - aio_context_acquire(blk_get_aio_context(s->conf.blk)); - if (ret || r->req.io_canceled) { scsi_command_complete_noio(r, ret); goto done; -- 2.39.2 From: Stefan Hajnoczi <stefanha@redhat.com> dma_blk_cb() only takes the AioContext lock around ->io_func(). That means the rest of dma_blk_cb() is not protected. In particular, the DMAAIOCB field accesses happen outside the lock. There is a race when the main loop thread holds the AioContext lock and invokes scsi_device_purge_requests() -> bdrv_aio_cancel() -> dma_aio_cancel() while an IOThread executes dma_blk_cb(). The dbs->acb field determines how cancellation proceeds. If dma_aio_cancel() sees dbs->acb == NULL while dma_blk_cb() is still running, the request can be completed twice (-ECANCELED and the actual return value). The following assertion can occur with virtio-scsi when an IOThread is used: ../hw/scsi/scsi-disk.c:368: scsi_dma_complete: Assertion `r->req.aiocb != NULL' failed. Fix the race by holding the AioContext across dma_blk_cb(). Now dma_aio_cancel() under the AioContext lock will not see inconsistent/intermediate states. Cc: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20230221212218.1378734-3-stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- hw/scsi/scsi-disk.c | 4 +--- softmmu/dma-helpers.c | 12 +++++++----- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c index XXXXXXX..XXXXXXX 100644 --- a/hw/scsi/scsi-disk.c +++ b/hw/scsi/scsi-disk.c @@ -XXX,XX +XXX,XX @@ done: scsi_req_unref(&r->req); } +/* Called with AioContext lock held */ static void scsi_dma_complete(void *opaque, int ret) { SCSIDiskReq *r = (SCSIDiskReq *)opaque; SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev); - aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk)); - assert(r->req.aiocb != NULL); r->req.aiocb = NULL; @@ -XXX,XX +XXX,XX @@ static void scsi_dma_complete(void *opaque, int ret) block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct); } scsi_dma_complete_noio(r, ret); - aio_context_release(blk_get_aio_context(s->qdev.conf.blk)); } static void scsi_read_complete_noio(SCSIDiskReq *r, int ret) diff --git a/softmmu/dma-helpers.c b/softmmu/dma-helpers.c index XXXXXXX..XXXXXXX 100644 --- a/softmmu/dma-helpers.c +++ b/softmmu/dma-helpers.c @@ -XXX,XX +XXX,XX @@ static void dma_complete(DMAAIOCB *dbs, int ret) static void dma_blk_cb(void *opaque, int ret) { DMAAIOCB *dbs = (DMAAIOCB *)opaque; + AioContext *ctx = dbs->ctx; dma_addr_t cur_addr, cur_len; void *mem; trace_dma_blk_cb(dbs, ret); + aio_context_acquire(ctx); dbs->acb = NULL; dbs->offset += dbs->iov.size; if (dbs->sg_cur_index == dbs->sg->nsg || ret < 0) { dma_complete(dbs, ret); - return; + goto out; } dma_blk_unmap(dbs); @@ -XXX,XX +XXX,XX @@ static void dma_blk_cb(void *opaque, int ret) if (dbs->iov.size == 0) { trace_dma_map_wait(dbs); - dbs->bh = aio_bh_new(dbs->ctx, reschedule_dma, dbs); + dbs->bh = aio_bh_new(ctx, reschedule_dma, dbs); cpu_register_map_client(dbs->bh); - return; + goto out; } if (!QEMU_IS_ALIGNED(dbs->iov.size, dbs->align)) { @@ -XXX,XX +XXX,XX @@ static void dma_blk_cb(void *opaque, int ret) QEMU_ALIGN_DOWN(dbs->iov.size, dbs->align)); } - aio_context_acquire(dbs->ctx); dbs->acb = dbs->io_func(dbs->offset, &dbs->iov, dma_blk_cb, dbs, dbs->io_func_opaque); - aio_context_release(dbs->ctx); assert(dbs->acb); +out: + aio_context_release(ctx); } static void dma_aio_cancel(BlockAIOCB *acb) -- 2.39.2 From: Stefan Hajnoczi <stefanha@redhat.com> When an IOThread is configured, the ctrl virtqueue is processed in the IOThread. TMFs that reset SCSI devices are currently called directly from the IOThread and trigger an assertion failure in blk_drain() from the following call stack: virtio_scsi_handle_ctrl_req -> virtio_scsi_do_tmf -> device_code_reset -> scsi_disk_reset -> scsi_device_purge_requests -> blk_drain ../block/block-backend.c:1780: void blk_drain(BlockBackend *): Assertion `qemu_in_main_thread()' failed. The blk_drain() function is not designed to be called from an IOThread because it needs the Big QEMU Lock (BQL). This patch defers TMFs that reset SCSI devices to a Bottom Half (BH) that runs in the main loop thread under the BQL. This way it's safe to call blk_drain() and the assertion failure is avoided. Introduce s->tmf_bh_list for tracking TMF requests that have been deferred to the BH. When the BH runs it will grab the entire list and process all requests. Care must be taken to clear the list when the virtio-scsi device is reset or unrealized. Otherwise deferred TMF requests could execute later and lead to use-after-free or other undefined behavior. The s->resetting counter that's used by TMFs that reset SCSI devices is accessed from multiple threads. This patch makes that explicit by using atomic accessor functions. With this patch applied the counter is only modified by the main loop thread under the BQL but can be read by any thread. Reported-by: Qing Wang <qinwang@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Message-Id: <20230221212218.1378734-4-stefanha@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- include/hw/virtio/virtio-scsi.h | 11 ++- hw/scsi/virtio-scsi.c | 169 +++++++++++++++++++++++++------- 2 files changed, 143 insertions(+), 37 deletions(-) diff --git a/include/hw/virtio/virtio-scsi.h b/include/hw/virtio/virtio-scsi.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/virtio/virtio-scsi.h +++ b/include/hw/virtio/virtio-scsi.h @@ -XXX,XX +XXX,XX @@ struct VirtIOSCSICommon { VirtQueue **cmd_vqs; }; +struct VirtIOSCSIReq; + struct VirtIOSCSI { VirtIOSCSICommon parent_obj; SCSIBus bus; - int resetting; + int resetting; /* written from main loop thread, read from any thread */ bool events_dropped; + /* + * TMFs deferred to main loop BH. These fields are protected by + * virtio_scsi_acquire(). + */ + QEMUBH *tmf_bh; + QTAILQ_HEAD(, VirtIOSCSIReq) tmf_bh_list; + /* Fields for dataplane below */ AioContext *ctx; /* one iothread per virtio-scsi-pci for now */ diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c index XXXXXXX..XXXXXXX 100644 --- a/hw/scsi/virtio-scsi.c +++ b/hw/scsi/virtio-scsi.c @@ -XXX,XX +XXX,XX @@ typedef struct VirtIOSCSIReq { QEMUSGList qsgl; QEMUIOVector resp_iov; - union { - /* Used for two-stage request submission */ - QTAILQ_ENTRY(VirtIOSCSIReq) next; + /* Used for two-stage request submission and TMFs deferred to BH */ + QTAILQ_ENTRY(VirtIOSCSIReq) next; - /* Used for cancellation of request during TMFs */ - int remaining; - }; + /* Used for cancellation of request during TMFs */ + int remaining; SCSIRequest *sreq; size_t resp_size; @@ -XXX,XX +XXX,XX @@ static inline void virtio_scsi_ctx_check(VirtIOSCSI *s, SCSIDevice *d) } } +static void virtio_scsi_do_one_tmf_bh(VirtIOSCSIReq *req) +{ + VirtIOSCSI *s = req->dev; + SCSIDevice *d = virtio_scsi_device_get(s, req->req.tmf.lun); + BusChild *kid; + int target; + + switch (req->req.tmf.subtype) { + case VIRTIO_SCSI_T_TMF_LOGICAL_UNIT_RESET: + if (!d) { + req->resp.tmf.response = VIRTIO_SCSI_S_BAD_TARGET; + goto out; + } + if (d->lun != virtio_scsi_get_lun(req->req.tmf.lun)) { + req->resp.tmf.response = VIRTIO_SCSI_S_INCORRECT_LUN; + goto out; + } + qatomic_inc(&s->resetting); + device_cold_reset(&d->qdev); + qatomic_dec(&s->resetting); + break; + + case VIRTIO_SCSI_T_TMF_I_T_NEXUS_RESET: + target = req->req.tmf.lun[1]; + qatomic_inc(&s->resetting); + + rcu_read_lock(); + QTAILQ_FOREACH_RCU(kid, &s->bus.qbus.children, sibling) { + SCSIDevice *d1 = SCSI_DEVICE(kid->child); + if (d1->channel == 0 && d1->id == target) { + device_cold_reset(&d1->qdev); + } + } + rcu_read_unlock(); + + qatomic_dec(&s->resetting); + break; + + default: + g_assert_not_reached(); + break; + } + +out: + object_unref(OBJECT(d)); + + virtio_scsi_acquire(s); + virtio_scsi_complete_req(req); + virtio_scsi_release(s); +} + +/* Some TMFs must be processed from the main loop thread */ +static void virtio_scsi_do_tmf_bh(void *opaque) +{ + VirtIOSCSI *s = opaque; + QTAILQ_HEAD(, VirtIOSCSIReq) reqs = QTAILQ_HEAD_INITIALIZER(reqs); + VirtIOSCSIReq *req; + VirtIOSCSIReq *tmp; + + GLOBAL_STATE_CODE(); + + virtio_scsi_acquire(s); + + QTAILQ_FOREACH_SAFE(req, &s->tmf_bh_list, next, tmp) { + QTAILQ_REMOVE(&s->tmf_bh_list, req, next); + QTAILQ_INSERT_TAIL(&reqs, req, next); + } + + qemu_bh_delete(s->tmf_bh); + s->tmf_bh = NULL; + + virtio_scsi_release(s); + + QTAILQ_FOREACH_SAFE(req, &reqs, next, tmp) { + QTAILQ_REMOVE(&reqs, req, next); + virtio_scsi_do_one_tmf_bh(req); + } +} + +static void virtio_scsi_reset_tmf_bh(VirtIOSCSI *s) +{ + VirtIOSCSIReq *req; + VirtIOSCSIReq *tmp; + + GLOBAL_STATE_CODE(); + + virtio_scsi_acquire(s); + + if (s->tmf_bh) { + qemu_bh_delete(s->tmf_bh); + s->tmf_bh = NULL; + } + + QTAILQ_FOREACH_SAFE(req, &s->tmf_bh_list, next, tmp) { + QTAILQ_REMOVE(&s->tmf_bh_list, req, next); + + /* SAM-6 6.3.2 Hard reset */ + req->resp.tmf.response = VIRTIO_SCSI_S_TARGET_FAILURE; + virtio_scsi_complete_req(req); + } + + virtio_scsi_release(s); +} + +static void virtio_scsi_defer_tmf_to_bh(VirtIOSCSIReq *req) +{ + VirtIOSCSI *s = req->dev; + + QTAILQ_INSERT_TAIL(&s->tmf_bh_list, req, next); + + if (!s->tmf_bh) { + s->tmf_bh = qemu_bh_new(virtio_scsi_do_tmf_bh, s); + qemu_bh_schedule(s->tmf_bh); + } +} + /* Return 0 if the request is ready to be completed and return to guest; * -EINPROGRESS if the request is submitted and will be completed later, in the * case of async cancellation. */ @@ -XXX,XX +XXX,XX @@ static int virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req) { SCSIDevice *d = virtio_scsi_device_get(s, req->req.tmf.lun); SCSIRequest *r, *next; - BusChild *kid; - int target; int ret = 0; virtio_scsi_ctx_check(s, d); @@ -XXX,XX +XXX,XX @@ static int virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req) break; case VIRTIO_SCSI_T_TMF_LOGICAL_UNIT_RESET: - if (!d) { - goto fail; - } - if (d->lun != virtio_scsi_get_lun(req->req.tmf.lun)) { - goto incorrect_lun; - } - s->resetting++; - device_cold_reset(&d->qdev); - s->resetting--; + case VIRTIO_SCSI_T_TMF_I_T_NEXUS_RESET: + virtio_scsi_defer_tmf_to_bh(req); + ret = -EINPROGRESS; break; case VIRTIO_SCSI_T_TMF_ABORT_TASK_SET: @@ -XXX,XX +XXX,XX @@ static int virtio_scsi_do_tmf(VirtIOSCSI *s, VirtIOSCSIReq *req) } break; - case VIRTIO_SCSI_T_TMF_I_T_NEXUS_RESET: - target = req->req.tmf.lun[1]; - s->resetting++; - - rcu_read_lock(); - QTAILQ_FOREACH_RCU(kid, &s->bus.qbus.children, sibling) { - SCSIDevice *d1 = SCSI_DEVICE(kid->child); - if (d1->channel == 0 && d1->id == target) { - device_cold_reset(&d1->qdev); - } - } - rcu_read_unlock(); - - s->resetting--; - break; - case VIRTIO_SCSI_T_TMF_CLEAR_ACA: default: req->resp.tmf.response = VIRTIO_SCSI_S_FUNCTION_REJECTED; @@ -XXX,XX +XXX,XX @@ static void virtio_scsi_request_cancelled(SCSIRequest *r) if (!req) { return; } - if (req->dev->resetting) { + if (qatomic_read(&req->dev->resetting)) { req->resp.cmd.response = VIRTIO_SCSI_S_RESET; } else { req->resp.cmd.response = VIRTIO_SCSI_S_ABORTED; @@ -XXX,XX +XXX,XX @@ static void virtio_scsi_reset(VirtIODevice *vdev) VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(vdev); assert(!s->dataplane_started); - s->resetting++; + + virtio_scsi_reset_tmf_bh(s); + + qatomic_inc(&s->resetting); bus_cold_reset(BUS(&s->bus)); - s->resetting--; + qatomic_dec(&s->resetting); vs->sense_size = VIRTIO_SCSI_SENSE_DEFAULT_SIZE; vs->cdb_size = VIRTIO_SCSI_CDB_DEFAULT_SIZE; @@ -XXX,XX +XXX,XX @@ static void virtio_scsi_device_realize(DeviceState *dev, Error **errp) VirtIOSCSI *s = VIRTIO_SCSI(dev); Error *err = NULL; + QTAILQ_INIT(&s->tmf_bh_list); + virtio_scsi_common_realize(dev, virtio_scsi_handle_ctrl, virtio_scsi_handle_event, @@ -XXX,XX +XXX,XX @@ static void virtio_scsi_device_unrealize(DeviceState *dev) { VirtIOSCSI *s = VIRTIO_SCSI(dev); + virtio_scsi_reset_tmf_bh(s); + qbus_set_hotplug_handler(BUS(&s->bus), NULL); virtio_scsi_common_unrealize(dev); } -- 2.39.2 From: Or Ozeri <oro@il.ibm.com> Signed-off-by: Or Ozeri <oro@il.ibm.com> Message-Id: <20230129113120.722708-2-oro@oro.sl.cloud9.ibm.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- block/rbd.c | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-) diff --git a/block/rbd.c b/block/rbd.c index XXXXXXX..XXXXXXX 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_format(rbd_image_t image, { int r = 0; g_autofree char *passphrase = NULL; - size_t passphrase_len; rbd_encryption_format_t format; rbd_encryption_options_t opts; rbd_encryption_luks1_format_options_t luks_opts; @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_format(rbd_image_t image, opts_size = sizeof(luks_opts); r = qemu_rbd_convert_luks_create_options( qapi_RbdEncryptionCreateOptionsLUKS_base(&encrypt->u.luks), - &luks_opts.alg, &passphrase, &passphrase_len, errp); + &luks_opts.alg, &passphrase, &luks_opts.passphrase_size, + errp); if (r < 0) { return r; } luks_opts.passphrase = passphrase; - luks_opts.passphrase_size = passphrase_len; break; } case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2: { @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_format(rbd_image_t image, r = qemu_rbd_convert_luks_create_options( qapi_RbdEncryptionCreateOptionsLUKS2_base( &encrypt->u.luks2), - &luks2_opts.alg, &passphrase, &passphrase_len, errp); + &luks2_opts.alg, &passphrase, &luks2_opts.passphrase_size, + errp); if (r < 0) { return r; } luks2_opts.passphrase = passphrase; - luks2_opts.passphrase_size = passphrase_len; break; } default: { @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, { int r = 0; g_autofree char *passphrase = NULL; - size_t passphrase_len; rbd_encryption_luks1_format_options_t luks_opts; rbd_encryption_luks2_format_options_t luks2_opts; rbd_encryption_format_t format; @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, opts_size = sizeof(luks_opts); r = qemu_rbd_convert_luks_options( qapi_RbdEncryptionOptionsLUKS_base(&encrypt->u.luks), - &passphrase, &passphrase_len, errp); + &passphrase, &luks_opts.passphrase_size, errp); if (r < 0) { return r; } luks_opts.passphrase = passphrase; - luks_opts.passphrase_size = passphrase_len; break; } case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2: { @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, opts_size = sizeof(luks2_opts); r = qemu_rbd_convert_luks_options( qapi_RbdEncryptionOptionsLUKS2_base(&encrypt->u.luks2), - &passphrase, &passphrase_len, errp); + &passphrase, &luks2_opts.passphrase_size, errp); if (r < 0) { return r; } luks2_opts.passphrase = passphrase; - luks2_opts.passphrase_size = passphrase_len; break; } default: { -- 2.39.2 From: Or Ozeri <oro@il.ibm.com> Ceph RBD encryption API required specifying the encryption format for loading encryption. The supported formats were LUKS (v1) and LUKS2. Starting from Reef release, RBD also supports loading with "luks-any" format, which works for both versions of LUKS. This commit extends the qemu rbd driver API to enable qemu users to use this luks-any wildcard format. Signed-off-by: Or Ozeri <oro@il.ibm.com> Message-Id: <20230129113120.722708-3-oro@oro.sl.cloud9.ibm.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qapi/block-core.json | 16 ++++++++++++++-- block/rbd.c | 19 +++++++++++++++++++ 2 files changed, 33 insertions(+), 2 deletions(-) diff --git a/qapi/block-core.json b/qapi/block-core.json index XXXXXXX..XXXXXXX 100644 --- a/qapi/block-core.json +++ b/qapi/block-core.json @@ -XXX,XX +XXX,XX @@ ## # @RbdImageEncryptionFormat: # +# @luks-any: Used for opening either luks or luks2 (Since 8.0) +# # Since: 6.1 ## { 'enum': 'RbdImageEncryptionFormat', - 'data': [ 'luks', 'luks2' ] } + 'data': [ 'luks', 'luks2', 'luks-any' ] } ## # @RbdEncryptionOptionsLUKSBase: @@ -XXX,XX +XXX,XX @@ 'base': 'RbdEncryptionOptionsLUKSBase', 'data': { } } +## +# @RbdEncryptionOptionsLUKSAny: +# +# Since: 8.0 +## +{ 'struct': 'RbdEncryptionOptionsLUKSAny', + 'base': 'RbdEncryptionOptionsLUKSBase', + 'data': { } } + ## # @RbdEncryptionCreateOptionsLUKS: # @@ -XXX,XX +XXX,XX @@ 'base': { 'format': 'RbdImageEncryptionFormat' }, 'discriminator': 'format', 'data': { 'luks': 'RbdEncryptionOptionsLUKS', - 'luks2': 'RbdEncryptionOptionsLUKS2' } } + 'luks2': 'RbdEncryptionOptionsLUKS2', + 'luks-any': 'RbdEncryptionOptionsLUKSAny'} } ## # @RbdEncryptionCreateOptions: diff --git a/block/rbd.c b/block/rbd.c index XXXXXXX..XXXXXXX 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, g_autofree char *passphrase = NULL; rbd_encryption_luks1_format_options_t luks_opts; rbd_encryption_luks2_format_options_t luks2_opts; +#ifdef LIBRBD_SUPPORTS_ENCRYPTION_LOAD2 + rbd_encryption_luks_format_options_t luks_any_opts; +#endif rbd_encryption_format_t format; rbd_encryption_options_t opts; size_t opts_size; @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, luks2_opts.passphrase = passphrase; break; } +#ifdef LIBRBD_SUPPORTS_ENCRYPTION_LOAD2 + case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS_ANY: { + memset(&luks_any_opts, 0, sizeof(luks_any_opts)); + format = RBD_ENCRYPTION_FORMAT_LUKS; + opts = &luks_any_opts; + opts_size = sizeof(luks_any_opts); + r = qemu_rbd_convert_luks_options( + qapi_RbdEncryptionOptionsLUKSAny_base(&encrypt->u.luks_any), + &passphrase, &luks_any_opts.passphrase_size, errp); + if (r < 0) { + return r; + } + luks_any_opts.passphrase = passphrase; + break; + } +#endif default: { r = -ENOTSUP; error_setg_errno( -- 2.39.2 From: Or Ozeri <oro@il.ibm.com> Starting from ceph Reef, RBD has built-in support for layered encryption, where each ancestor image (in a cloned image setting) can be possibly encrypted using a unique passphrase. A new function, rbd_encryption_load2, was added to librbd API. This new function supports an array of passphrases (via "spec" structs). This commit extends the qemu rbd driver API to use this new librbd API, in order to support this new layered encryption feature. Signed-off-by: Or Ozeri <oro@il.ibm.com> Message-Id: <20230129113120.722708-4-oro@oro.sl.cloud9.ibm.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com> --- qapi/block-core.json | 11 +++- block/rbd.c | 153 ++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 162 insertions(+), 2 deletions(-) diff --git a/qapi/block-core.json b/qapi/block-core.json index XXXXXXX..XXXXXXX 100644 --- a/qapi/block-core.json +++ b/qapi/block-core.json @@ -XXX,XX +XXX,XX @@ ## # @RbdEncryptionOptions: # +# @format: Encryption format. +# +# @parent: Parent image encryption options (for cloned images). +# Can be left unspecified if this cloned image is encrypted +# using the same format and secret as its parent image (i.e. +# not explicitly formatted) or if its parent image is not +# encrypted. (Since 8.0) +# # Since: 6.1 ## { 'union': 'RbdEncryptionOptions', - 'base': { 'format': 'RbdImageEncryptionFormat' }, + 'base': { 'format': 'RbdImageEncryptionFormat', + '*parent': 'RbdEncryptionOptions' }, 'discriminator': 'format', 'data': { 'luks': 'RbdEncryptionOptionsLUKS', 'luks2': 'RbdEncryptionOptionsLUKS2', diff --git a/block/rbd.c b/block/rbd.c index XXXXXXX..XXXXXXX 100644 --- a/block/rbd.c +++ b/block/rbd.c @@ -XXX,XX +XXX,XX @@ static const char rbd_luks2_header_verification[ 'L', 'U', 'K', 'S', 0xBA, 0xBE, 0, 2 }; +static const char rbd_layered_luks_header_verification[ + RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN] = { + 'R', 'B', 'D', 'L', 0xBA, 0xBE, 0, 1 +}; + +static const char rbd_layered_luks2_header_verification[ + RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN] = { + 'R', 'B', 'D', 'L', 0xBA, 0xBE, 0, 2 +}; + typedef enum { RBD_AIO_READ, RBD_AIO_WRITE, @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_encryption_load(rbd_image_t image, return 0; } + +#ifdef LIBRBD_SUPPORTS_ENCRYPTION_LOAD2 +static int qemu_rbd_encryption_load2(rbd_image_t image, + RbdEncryptionOptions *encrypt, + Error **errp) +{ + int r = 0; + int encrypt_count = 1; + int i; + RbdEncryptionOptions *curr_encrypt; + rbd_encryption_spec_t *specs; + rbd_encryption_luks1_format_options_t *luks_opts; + rbd_encryption_luks2_format_options_t *luks2_opts; + rbd_encryption_luks_format_options_t *luks_any_opts; + + /* count encryption options */ + for (curr_encrypt = encrypt->parent; curr_encrypt; + curr_encrypt = curr_encrypt->parent) { + ++encrypt_count; + } + + specs = g_new0(rbd_encryption_spec_t, encrypt_count); + + curr_encrypt = encrypt; + for (i = 0; i < encrypt_count; ++i) { + switch (curr_encrypt->format) { + case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS: { + specs[i].format = RBD_ENCRYPTION_FORMAT_LUKS1; + + luks_opts = g_new0(rbd_encryption_luks1_format_options_t, 1); + specs[i].opts = luks_opts; + specs[i].opts_size = sizeof(*luks_opts); + + r = qemu_rbd_convert_luks_options( + qapi_RbdEncryptionOptionsLUKS_base( + &curr_encrypt->u.luks), + (char **)&luks_opts->passphrase, + &luks_opts->passphrase_size, + errp); + break; + } + case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2: { + specs[i].format = RBD_ENCRYPTION_FORMAT_LUKS2; + + luks2_opts = g_new0(rbd_encryption_luks2_format_options_t, 1); + specs[i].opts = luks2_opts; + specs[i].opts_size = sizeof(*luks2_opts); + + r = qemu_rbd_convert_luks_options( + qapi_RbdEncryptionOptionsLUKS2_base( + &curr_encrypt->u.luks2), + (char **)&luks2_opts->passphrase, + &luks2_opts->passphrase_size, + errp); + break; + } + case RBD_IMAGE_ENCRYPTION_FORMAT_LUKS_ANY: { + specs[i].format = RBD_ENCRYPTION_FORMAT_LUKS; + + luks_any_opts = g_new0(rbd_encryption_luks_format_options_t, 1); + specs[i].opts = luks_any_opts; + specs[i].opts_size = sizeof(*luks_any_opts); + + r = qemu_rbd_convert_luks_options( + qapi_RbdEncryptionOptionsLUKSAny_base( + &curr_encrypt->u.luks_any), + (char **)&luks_any_opts->passphrase, + &luks_any_opts->passphrase_size, + errp); + break; + } + default: { + r = -ENOTSUP; + error_setg_errno( + errp, -r, "unknown image encryption format: %u", + curr_encrypt->format); + } + } + + if (r < 0) { + goto exit; + } + + curr_encrypt = curr_encrypt->parent; + } + + r = rbd_encryption_load2(image, specs, encrypt_count); + if (r < 0) { + error_setg_errno(errp, -r, "layered encryption load fail"); + goto exit; + } + +exit: + for (i = 0; i < encrypt_count; ++i) { + if (!specs[i].opts) { + break; + } + + switch (specs[i].format) { + case RBD_ENCRYPTION_FORMAT_LUKS1: { + luks_opts = specs[i].opts; + g_free((void *)luks_opts->passphrase); + break; + } + case RBD_ENCRYPTION_FORMAT_LUKS2: { + luks2_opts = specs[i].opts; + g_free((void *)luks2_opts->passphrase); + break; + } + case RBD_ENCRYPTION_FORMAT_LUKS: { + luks_any_opts = specs[i].opts; + g_free((void *)luks_any_opts->passphrase); + break; + } + } + + g_free(specs[i].opts); + } + g_free(specs); + return r; +} +#endif #endif /* FIXME Deprecate and remove keypairs or make it available in QMP. */ @@ -XXX,XX +XXX,XX @@ static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags, if (opts->encrypt) { #ifdef LIBRBD_SUPPORTS_ENCRYPTION - r = qemu_rbd_encryption_load(s->image, opts->encrypt, errp); + if (opts->encrypt->parent) { +#ifdef LIBRBD_SUPPORTS_ENCRYPTION_LOAD2 + r = qemu_rbd_encryption_load2(s->image, opts->encrypt, errp); +#else + r = -ENOTSUP; + error_setg(errp, "RBD library does not support layered encryption"); +#endif + } else { + r = qemu_rbd_encryption_load(s->image, opts->encrypt, errp); + } if (r < 0) { goto failed_post_open; } @@ -XXX,XX +XXX,XX @@ static ImageInfoSpecific *qemu_rbd_get_specific_info(BlockDriverState *bs, spec_info->u.rbd.data->encryption_format = RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2; spec_info->u.rbd.data->has_encryption_format = true; + } else if (memcmp(buf, rbd_layered_luks_header_verification, + RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN) == 0) { + spec_info->u.rbd.data->encryption_format = + RBD_IMAGE_ENCRYPTION_FORMAT_LUKS; + spec_info->u.rbd.data->has_encryption_format = true; + } else if (memcmp(buf, rbd_layered_luks2_header_verification, + RBD_ENCRYPTION_LUKS_HEADER_VERIFICATION_LEN) == 0) { + spec_info->u.rbd.data->encryption_format = + RBD_IMAGE_ENCRYPTION_FORMAT_LUKS2; + spec_info->u.rbd.data->has_encryption_format = true; } else { spec_info->u.rbd.data->has_encryption_format = false; } -- 2.39.2