Re: [PATCH] btrfs: fix refcount_t usage when deleting btrfs_delayed_nodes
On Fri, Dec 15, 2017 at 11:58:27AM -0800, Chris Mason wrote: > refcounts have a generic implementation and an asm optimized one. The > generic version has extra debugging to make sure that once a refcount > goes to zero, refcount_inc won't increase it. > > The btrfs delayed inode code wasn't expecting this, and we're tripping > over the warnings when the generic refcounts are used. We ended up with > this race: > > Process A Process B > btrfs_get_delayed_node() > spin_lock(root->inode_lock) > radix_tree_lookup() > __btrfs_release_delayed_node() > refcount_dec_and_test(&delayed_node->refs) > our refcount is now zero > refcount_add(2) <--- > warning here, refcount > unchanged > > spin_lock(root->inode_lock) > radix_tree_delete() > > With the generic refcounts, we actually warn again when process B above > tries to release his refcount because refcount_add() turned into a > no-op. > > We saw this in production on older kernels without the asm optimized > refcounts. > > The fix used here is to use refcount_inc_not_zero() to detect when the > object is in the middle of being freed and return NULL. This is almost > always the right answer anyway, since we usually end up pitching the > delayed_node if it didn't have fresh data in it. > > This also changes __btrfs_release_delayed_node() to remove the extra > check for zero refcounts before radix tree deletion. > btrfs_get_delayed_node() was the only path that was allowing refcounts > to go from zero to one. > Reviewed-by: Liu Bo -liubo > Signed-off-by: Chris Mason > Fixes: 6de5f18e7b0da > cc: #4.12+ > --- > fs/btrfs/delayed-inode.c | 45 ++--- > 1 file changed, 34 insertions(+), 11 deletions(-) > > diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c > index 5d73f79..84c54af 100644 > --- a/fs/btrfs/delayed-inode.c > +++ b/fs/btrfs/delayed-inode.c > @@ -87,6 +87,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > > spin_lock(&root->inode_lock); > node = radix_tree_lookup(&root->delayed_nodes_tree, ino); > + > if (node) { > if (btrfs_inode->delayed_node) { > refcount_inc(&node->refs); /* can be accessed */ > @@ -94,9 +95,30 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > spin_unlock(&root->inode_lock); > return node; > } > - btrfs_inode->delayed_node = node; > - /* can be accessed and cached in the inode */ > - refcount_add(2, &node->refs); > + > + /* it's possible that we're racing into the middle of > + * removing this node from the radix tree. In this case, > + * the refcount was zero and it should never go back > + * to one. Just return NULL like it was never in the radix > + * at all; our release function is in the process of removing > + * it. > + * > + * Some implementations of refcount_inc refuse to > + * bump the refcount once it has hit zero. If we don't do > + * this dance here, refcount_inc() may decide to > + * just WARN_ONCE() instead of actually bumping the refcount. > + * > + * If this node is properly in the radix, we want to > + * bump the refcount twice, once for the inode > + * and once for this get operation. > + */ > + if (refcount_inc_not_zero(&node->refs)) { > + refcount_inc(&node->refs); > + btrfs_inode->delayed_node = node; > + } else { > + node = NULL; > + } > + > spin_unlock(&root->inode_lock); > return node; > } > @@ -254,17 +276,18 @@ static void __btrfs_release_delayed_node( > mutex_unlock(&delayed_node->mutex); > > if (refcount_dec_and_test(&delayed_node->refs)) { > - bool free = false; > struct btrfs_root *root = delayed_node->root; > + > spin_lock(&root->inode_lock); > - if (refcount_read(&delayed_node->refs) == 0) { > - radix_tree_delete(&root->delayed_nodes_tree, > - delayed_node->inode_id); > - free = true; > - } > + /* > + * once our refcount goes to zero, nobody is allowed to > + * bump it back up. We can delete it now > + */ > + ASSERT(refcount_read(&delayed_node->refs) == 0); > + radix_tree_
Re: [PATCH] btrfs: fix refcount_t usage when deleting btrfs_delayed_nodes
On Sat, Dec 16, 2017 at 08:42:51AM +0200, Nikolay Borisov wrote: > > > On 15.12.2017 21:58, Chris Mason wrote: > > refcounts have a generic implementation and an asm optimized one. The > > generic version has extra debugging to make sure that once a refcount > > goes to zero, refcount_inc won't increase it. > > > I guess you meant to say refcount_add refcount_inc may also just throw a warning without bumping the refcnt. Thanks, -liubo > > > > The btrfs delayed inode code wasn't expecting this, and we're tripping > > over the warnings when the generic refcounts are used. We ended up with > > this race: > > > > Process A Process B > > btrfs_get_delayed_node() > > spin_lock(root->inode_lock) > > radix_tree_lookup() > > __btrfs_release_delayed_node() > > refcount_dec_and_test(&delayed_node->refs) > > our refcount is now zero > > refcount_add(2) <--- > > warning here, refcount > > unchanged > > > > spin_lock(root->inode_lock) > > radix_tree_delete() > > > > With the generic refcounts, we actually warn again when process B above > > tries to release his refcount because refcount_add() turned into a > > no-op. > > > > We saw this in production on older kernels without the asm optimized > > refcounts. > > > > The fix used here is to use refcount_inc_not_zero() to detect when the > > object is in the middle of being freed and return NULL. This is almost > > always the right answer anyway, since we usually end up pitching the > > delayed_node if it didn't have fresh data in it. > > > > This also changes __btrfs_release_delayed_node() to remove the extra > > check for zero refcounts before radix tree deletion. > > btrfs_get_delayed_node() was the only path that was allowing refcounts > > to go from zero to one. > > > > Signed-off-by: Chris Mason > > Fixes: 6de5f18e7b0da > > cc: #4.12+ > > --- > > fs/btrfs/delayed-inode.c | 45 ++--- > > 1 file changed, 34 insertions(+), 11 deletions(-) > > > > diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c > > index 5d73f79..84c54af 100644 > > --- a/fs/btrfs/delayed-inode.c > > +++ b/fs/btrfs/delayed-inode.c > > @@ -87,6 +87,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > > > > spin_lock(&root->inode_lock); > > node = radix_tree_lookup(&root->delayed_nodes_tree, ino); > > + > > if (node) { > > if (btrfs_inode->delayed_node) { > > refcount_inc(&node->refs); /* can be accessed */ > > @@ -94,9 +95,30 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > > spin_unlock(&root->inode_lock); > > return node; > > } > > - btrfs_inode->delayed_node = node; > > - /* can be accessed and cached in the inode */ > > - refcount_add(2, &node->refs); > > + > > + /* it's possible that we're racing into the middle of > > +* removing this node from the radix tree. In this case, > > +* the refcount was zero and it should never go back > > +* to one. Just return NULL like it was never in the radix > > +* at all; our release function is in the process of removing > > +* it. > > +* > > +* Some implementations of refcount_inc refuse to > > +* bump the refcount once it has hit zero. If we don't do > > +* this dance here, refcount_inc() may decide to > > +* just WARN_ONCE() instead of actually bumping the refcount. > > +* > > +* If this node is properly in the radix, we want to > > +* bump the refcount twice, once for the inode > > +* and once for this get operation. > > +*/ > > + if (refcount_inc_not_zero(&node->refs)) { > > + refcount_inc(&node->refs); > > + btrfs_inode->delayed_node = node; > > + } else { > > + node = NULL; > > + } > > + > > spin_unlock(&root->inode_lock); > > return node; > > } > > @@ -254,17 +276,18 @@ static void __btrfs_release_delayed_node( > > mutex_unlock(&delayed_node->mutex); > > > > if (refcount_dec_and_test(&delayed_node->refs)) { > > - bool free = false; > > struct btrfs_root *root = delayed_node->root; > > + > > spin_lock(&root->inode_lock); > > - if (refcount_read(&delayed_node->refs) == 0) { > > - radix_tree_delete(&root->delayed_nodes_tree, > > - delayed_node->inode_id); > > -
Re: [PATCH] btrfs: fix refcount_t usage when deleting btrfs_delayed_nodes
On 15.12.2017 21:58, Chris Mason wrote: > refcounts have a generic implementation and an asm optimized one. The > generic version has extra debugging to make sure that once a refcount > goes to zero, refcount_inc won't increase it. I guess you meant to say refcount_add > > The btrfs delayed inode code wasn't expecting this, and we're tripping > over the warnings when the generic refcounts are used. We ended up with > this race: > > Process A Process B > btrfs_get_delayed_node() > spin_lock(root->inode_lock) > radix_tree_lookup() > __btrfs_release_delayed_node() > refcount_dec_and_test(&delayed_node->refs) > our refcount is now zero > refcount_add(2) <--- > warning here, refcount > unchanged > > spin_lock(root->inode_lock) > radix_tree_delete() > > With the generic refcounts, we actually warn again when process B above > tries to release his refcount because refcount_add() turned into a > no-op. > > We saw this in production on older kernels without the asm optimized > refcounts. > > The fix used here is to use refcount_inc_not_zero() to detect when the > object is in the middle of being freed and return NULL. This is almost > always the right answer anyway, since we usually end up pitching the > delayed_node if it didn't have fresh data in it. > > This also changes __btrfs_release_delayed_node() to remove the extra > check for zero refcounts before radix tree deletion. > btrfs_get_delayed_node() was the only path that was allowing refcounts > to go from zero to one. > > Signed-off-by: Chris Mason > Fixes: 6de5f18e7b0da > cc: #4.12+ > --- > fs/btrfs/delayed-inode.c | 45 ++--- > 1 file changed, 34 insertions(+), 11 deletions(-) > > diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c > index 5d73f79..84c54af 100644 > --- a/fs/btrfs/delayed-inode.c > +++ b/fs/btrfs/delayed-inode.c > @@ -87,6 +87,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > > spin_lock(&root->inode_lock); > node = radix_tree_lookup(&root->delayed_nodes_tree, ino); > + > if (node) { > if (btrfs_inode->delayed_node) { > refcount_inc(&node->refs); /* can be accessed */ > @@ -94,9 +95,30 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( > spin_unlock(&root->inode_lock); > return node; > } > - btrfs_inode->delayed_node = node; > - /* can be accessed and cached in the inode */ > - refcount_add(2, &node->refs); > + > + /* it's possible that we're racing into the middle of > + * removing this node from the radix tree. In this case, > + * the refcount was zero and it should never go back > + * to one. Just return NULL like it was never in the radix > + * at all; our release function is in the process of removing > + * it. > + * > + * Some implementations of refcount_inc refuse to > + * bump the refcount once it has hit zero. If we don't do > + * this dance here, refcount_inc() may decide to > + * just WARN_ONCE() instead of actually bumping the refcount. > + * > + * If this node is properly in the radix, we want to > + * bump the refcount twice, once for the inode > + * and once for this get operation. > + */ > + if (refcount_inc_not_zero(&node->refs)) { > + refcount_inc(&node->refs); > + btrfs_inode->delayed_node = node; > + } else { > + node = NULL; > + } > + > spin_unlock(&root->inode_lock); > return node; > } > @@ -254,17 +276,18 @@ static void __btrfs_release_delayed_node( > mutex_unlock(&delayed_node->mutex); > > if (refcount_dec_and_test(&delayed_node->refs)) { > - bool free = false; > struct btrfs_root *root = delayed_node->root; > + > spin_lock(&root->inode_lock); > - if (refcount_read(&delayed_node->refs) == 0) { > - radix_tree_delete(&root->delayed_nodes_tree, > - delayed_node->inode_id); > - free = true; > - } > + /* > + * once our refcount goes to zero, nobody is allowed to > + * bump it back up. We can delete it now > + */ > + ASSERT(refcount_read(&delayed_node->refs) == 0); > +
[PATCH] btrfs: fix refcount_t usage when deleting btrfs_delayed_nodes
refcounts have a generic implementation and an asm optimized one. The generic version has extra debugging to make sure that once a refcount goes to zero, refcount_inc won't increase it. The btrfs delayed inode code wasn't expecting this, and we're tripping over the warnings when the generic refcounts are used. We ended up with this race: Process A Process B btrfs_get_delayed_node() spin_lock(root->inode_lock) radix_tree_lookup() __btrfs_release_delayed_node() refcount_dec_and_test(&delayed_node->refs) our refcount is now zero refcount_add(2) <--- warning here, refcount unchanged spin_lock(root->inode_lock) radix_tree_delete() With the generic refcounts, we actually warn again when process B above tries to release his refcount because refcount_add() turned into a no-op. We saw this in production on older kernels without the asm optimized refcounts. The fix used here is to use refcount_inc_not_zero() to detect when the object is in the middle of being freed and return NULL. This is almost always the right answer anyway, since we usually end up pitching the delayed_node if it didn't have fresh data in it. This also changes __btrfs_release_delayed_node() to remove the extra check for zero refcounts before radix tree deletion. btrfs_get_delayed_node() was the only path that was allowing refcounts to go from zero to one. Signed-off-by: Chris Mason Fixes: 6de5f18e7b0da cc: #4.12+ --- fs/btrfs/delayed-inode.c | 45 ++--- 1 file changed, 34 insertions(+), 11 deletions(-) diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c index 5d73f79..84c54af 100644 --- a/fs/btrfs/delayed-inode.c +++ b/fs/btrfs/delayed-inode.c @@ -87,6 +87,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( spin_lock(&root->inode_lock); node = radix_tree_lookup(&root->delayed_nodes_tree, ino); + if (node) { if (btrfs_inode->delayed_node) { refcount_inc(&node->refs); /* can be accessed */ @@ -94,9 +95,30 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node( spin_unlock(&root->inode_lock); return node; } - btrfs_inode->delayed_node = node; - /* can be accessed and cached in the inode */ - refcount_add(2, &node->refs); + + /* it's possible that we're racing into the middle of +* removing this node from the radix tree. In this case, +* the refcount was zero and it should never go back +* to one. Just return NULL like it was never in the radix +* at all; our release function is in the process of removing +* it. +* +* Some implementations of refcount_inc refuse to +* bump the refcount once it has hit zero. If we don't do +* this dance here, refcount_inc() may decide to +* just WARN_ONCE() instead of actually bumping the refcount. +* +* If this node is properly in the radix, we want to +* bump the refcount twice, once for the inode +* and once for this get operation. +*/ + if (refcount_inc_not_zero(&node->refs)) { + refcount_inc(&node->refs); + btrfs_inode->delayed_node = node; + } else { + node = NULL; + } + spin_unlock(&root->inode_lock); return node; } @@ -254,17 +276,18 @@ static void __btrfs_release_delayed_node( mutex_unlock(&delayed_node->mutex); if (refcount_dec_and_test(&delayed_node->refs)) { - bool free = false; struct btrfs_root *root = delayed_node->root; + spin_lock(&root->inode_lock); - if (refcount_read(&delayed_node->refs) == 0) { - radix_tree_delete(&root->delayed_nodes_tree, - delayed_node->inode_id); - free = true; - } + /* +* once our refcount goes to zero, nobody is allowed to +* bump it back up. We can delete it now +*/ + ASSERT(refcount_read(&delayed_node->refs) == 0); + radix_tree_delete(&root->delayed_nodes_tree, + delayed_node->inode_id); spin_unlock(&root->inode_lock); - if (free) - kmem_cache