btrfs: qgroup fixes and a cleanup
cover btrfs: qgroup fixes and a cleanup
From: Filipe Manana <[email protected]>
A couple fixes and a cleanup, details in the change logs.
Filipe Manana (3):
btrfs: qgroup: set quota enabled bit if quota disable fails flushing reservations
btrfs: qgroup: fix qgroup create ioctl returning success after quotas disabled
btrfs: qgroup: use btrfs_qgroup_enabled() in ioctls
fs/btrfs/ioctl.c | 24 ++++-------------------- fs/btrfs/qgroup.c | 10 +++++----- fs/btrfs/transaction.c | 6 ++++-- 3 files changed, 13 insertions(+), 27 deletions(-)
Signature · 1 line
2.47.2
PATCH 1/3 btrfs: qgroup: set quota enabled bit if quota disable fails flushing reservations
From: Filipe Manana <[email protected]>
Before waiting for the rescan worker to finish and flushing reservations, we clear the BTRFS_FS_QUOTA_ENABLED flag from fs_info. If we fail flushing reservations we leave with the flag not set which is not correct since quotas are still enabled - we must set back the flag on error paths, such as when we fail to start a transaction, except for error paths that abort a transaction. The reservation flushing happens very early before we do any operation that actually disables quotas and before we start a transaction, so set back BTRFS_FS_QUOTA_ENABLED if it fails.
--- fs/btrfs/qgroup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
fs/btrfs/qgroup.c +5 -2
@@ -1334,11 +1334,14 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)/* * We have nothing held here and no trans handle, just return the error -* if there is one. +* if there is one and set back the quota enabled bit since we didn't +* actually disable quotas. */ ret =flush_reservations (fs_info); -if (ret) +if (ret) { +set_bit (BTRFS_FS_QUOTA_ENABLED , &fs_info->flags);return ret; + }/* * 1 For the root item
Signature · 1 line
2.47.2
Re: [PATCH 1/3] btrfs: qgroup: set quota enabled bit if quota disable fails flushing reservations
Quoted 11 lines · 在 2025/7/2 01:12, [email protected] 写道:
From: Filipe Manana <[email protected]>
Before waiting for the rescan worker to finish and flushing reservations, we clear the
BTRFS_FS_QUOTA_ENABLEDflag fromfs_info. If we fail flushing reservations we leave with the flag not set which is not correct since quotas are still enabled - we must set back the flag on error paths, such as when we fail to start a transaction, except for error paths that abort a transaction. The reservation flushing happens very early before we do any operation that actually disables quotas and before we start a transaction, so set backBTRFS_FS_QUOTA_ENABLEDif it fails.
Quoted 2 lines
Fixes af0e2aab3b70 ("btrfs: qgroup: flush reservations during quota disable")Signed-off-by Filipe Manana <[email protected]>
I initially thought it will be a little racy when setting/clearing the ENABLED flag without qgroup_ioctl_mutex, but since we're already holding subvol_sem() for both enable and disable, it should be fine.
Thanks,
Qu
Quoted 25 lines
--- fs/btrfs/qgroup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)fs/btrfs/qgroup.c +5
-2@@ -1334,11 +1334,14 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)/* * We have nothing held here and no trans handle, just return the error -* if there is one. +* if there is one and set back the quota enabled bit since we didn't +* actually disable quotas. */ ret =flush_reservations (fs_info); -if (ret) +if (ret) { +set_bit (BTRFS_FS_QUOTA_ENABLED , &fs_info->flags);return ret; + }/* * 1 For the root item
PATCH 2/3 btrfs: qgroup: fix qgroup create ioctl returning success after quotas disabled
In reply to [email protected]From: Filipe Manana <[email protected]>
When quotas are disabled qgroup ioctls are supposed to return -ENOTCONN, but the qgroup create ioctl stopped doing that when it races with a quota disable operation, returning 0 instead. This change of behaviour happened in commit 6ed05643ddb1 ("btrfs: create qgroup earlier in snapshot creation").
The issue happens as follows:
1) Task A enters btrfs_ioctl_qgroup_create(), qgroups are enabled and so qgroup_enabled() returns true since fs_info->quota_root is not NULL;
2) Task B enters btrfs_ioctl_quota_ctl() -> btrfs_quota_disable() and disables qgroups, so now fs_info->quota_root is NULL;
3) Task A enters btrfs_create_qgroup() and calls btrfs_qgroup_mode(), which returns BTRFS_QGROUP_MODE_DISABLED since quotas are disabled, and then btrfs_create_qgroup() returns 0 to the caller, which makes the ioctl return 0 instead of -ENOTCONN.
The check for fs_info->quota_root and returning -ENOTCONN if it's NULL is made only after the call btrfs_qgroup_mode().
Fix this by moving the check for disabled quotas with btrfs_qgroup_mode() into transaction.c:create_pending_snapshot(), so that we don't abort the transaction if btrfs_create_qgroup() returns -ENOTCONN and quotas are disabled.
--- fs/btrfs/qgroup.c | 3 --- fs/btrfs/transaction.c | 6 ++++-- 2 files changed, 4 insertions(+), 5 deletions(-)
fs/btrfs/qgroup.c +0 -3
@@ -1662,9 +1662,6 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)struct btrfs_qgroup *prealloc =NULL ;int ret =0 ; -if (btrfs_qgroup_mode (fs_info) ==BTRFS_QGROUP_MODE_DISABLED ) -return 0 ; -mutex_lock (&fs_info->qgroup_ioctl_lock);if (!fs_info->quota_root) { ret = -ENOTCONN ;
fs/btrfs/transaction.c +4 -2
@@ -1735,8 +1735,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, ret =btrfs_create_qgroup (trans, objectid);if (ret && ret != -EEXIST ) { -btrfs_abort_transaction (trans, ret); - goto fail; +if (ret != -ENOTCONN ||btrfs_qgroup_enabled (fs_info)) { +btrfs_abort_transaction (trans, ret); + goto fail; + } }/*
Signature · 1 line
2.47.2
Re: [PATCH 2/3] btrfs: qgroup: fix qgroup create ioctl returning success after quotas disabled
Quoted 24 lines · 在 2025/7/2 01:12, [email protected] 写道:
From: Filipe Manana <[email protected]>
When quotas are disabled qgroup ioctls are supposed to return
-ENOTCONN, but the qgroup create ioctl stopped doing that when it races with a quota disable operation, returning 0 instead. This change of behaviour happened in commit6ed05643ddb1("btrfs: create qgroup earlier in snapshot creation").The issue happens as follows:
1) Task A enters
btrfs_ioctl_qgroup_create(), qgroups are enabled and soqgroup_enabled() returns true since fs_info->quota_root is not NULL;2) Task B enters
btrfs_ioctl_quota_ctl()->btrfs_quota_disable()anddisables qgroups, so now fs_info->quota_root is NULL;3) Task A enters
btrfs_create_qgroup()and callsbtrfs_qgroup_mode(), which returnsBTRFS_QGROUP_MODE_DISABLEDsince quotas are disabled, and thenbtrfs_create_qgroup()returns 0 to the caller, which makes the ioctl return 0 instead of-ENOTCONN.The check for
fs_info->quota_rootand returning-ENOTCONNif it's NULL is made only after the callbtrfs_qgroup_mode().
Quoted 7 lines
Fix this by moving the check for disabled quotas with
btrfs_qgroup_mode()into transaction.c:create_pending_snapshot(), so that we don't abort the transaction ifbtrfs_create_qgroup()returns-ENOTCONNand quotas are disabled.Fixes 6ed05643ddb1 ("btrfs: create qgroup earlier in snapshot creation")Signed-off-by Filipe Manana <[email protected]>
Since create_pending_snapshot() is called during btrfs_commit_transaction(), we're safe from a btrfs_quota_enable() call as it requires a trans handle.
So there will not be a race between create_pending_snapshot() and btrfs_quota_enable().
Thanks,
Qu
Quoted 36 lines
--- fs/btrfs/qgroup.c | 3 --- fs/btrfs/transaction.c | 6 ++++-- 2 files changed, 4 insertions(+), 5 deletions(-)fs/btrfs/qgroup.c +0
-3@@ -1662,9 +1662,6 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)struct btrfs_qgroup *prealloc =NULL ;int ret =0 ; -if (btrfs_qgroup_mode (fs_info) ==BTRFS_QGROUP_MODE_DISABLED ) -return 0 ; -mutex_lock (&fs_info->qgroup_ioctl_lock);if (!fs_info->quota_root) { ret = -ENOTCONN ;fs/btrfs/transaction.c +4
-2@@ -1735,8 +1735,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, ret =btrfs_create_qgroup (trans, objectid);if (ret && ret != -EEXIST ) { -btrfs_abort_transaction (trans, ret); - goto fail; +if (ret != -ENOTCONN ||btrfs_qgroup_enabled (fs_info)) { +btrfs_abort_transaction (trans, ret); + goto fail; + } }/*
PATCH 3/3 btrfs: qgroup: use btrfs_qgroup_enabled() in ioctls
In reply to [email protected]From: Filipe Manana <[email protected]>
We have a publicly exported btrfs_qgroup_enabled() and an ioctl.c private qgroup_enabled() helper. Both of these test if qgroups are enabled, the first check if the flag BTRFS_FS_QUOTA_ENABLED is set in fs_info->flags while the second checks if fs_info->quota_root is not NULL while holding the mutex fs_info->qgroup_ioctl_lock.
We can get away with the private ioctl.c:qgroup_enabled(), as all entry points into the qgroup code check if fs_info->quota_root is NULL or not while holding the mutex fs_info->qgroup_ioctl_lock, and returning the error -ENOTCONN in case it's NULL.
--- fs/btrfs/ioctl.c | 24 ++++-------------------- 1 file changed, 4 insertions(+), 20 deletions(-)
fs/btrfs/ioctl.c +4 -20
@@ -3714,22 +3714,6 @@ static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)return ret; } -/* -* Quick check for ioctl handlers if quotas are enabled. Proper locking must be -* done before any operations. -*/ -static bool qgroup_enabled (struct btrfs_fs_info *fs_info) -{ -bool ret = true; - -mutex_lock (&fs_info->qgroup_ioctl_lock); -if (!fs_info->quota_root) - ret = false; -mutex_unlock (&fs_info->qgroup_ioctl_lock); - -return ret; -} -static long btrfs_ioctl_qgroup_assign (struct file *file,void __user *arg) {struct inode *inode =file_inode (file);
@@ -3744,7 +3728,7 @@ static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);
@@ -3814,7 +3798,7 @@ static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (root->fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);
@@ -3873,7 +3857,7 @@ static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (root->fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);
@@ -3921,7 +3905,7 @@ static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (fs_info)) +if (!btrfs_qgroup_enabled (fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);
Signature · 1 line
2.47.2
Re: [PATCH 3/3] btrfs: qgroup: use btrfs_qgroup_enabled() in ioctls
Quoted 8 lines · 在 2025/7/2 01:12, [email protected] 写道:
From: Filipe Manana <[email protected]>
We have a publicly exported
btrfs_qgroup_enabled()and anioctl.cprivateqgroup_enabled()helper. Both of these test if qgroups are enabled, the first check if the flagBTRFS_FS_QUOTA_ENABLEDis set infs_info->flagswhile the second checks iffs_info->quota_rootis not NULL while holding the mutexfs_info->qgroup_ioctl_lock.
Quoted 6 lines
We can get away with the private ioctl.c:qgroup_enabled(), as all entry points into the qgroup code check if
fs_info->quota_rootis NULL or not while holding the mutexfs_info->qgroup_ioctl_lock, and returning the error-ENOTCONNin case it's NULL.Signed-off-by Filipe Manana <[email protected]>
Thanks,
Qu
Quoted 67 lines
--- fs/btrfs/ioctl.c | 24 ++++-------------------- 1 file changed, 4 insertions(+), 20 deletions(-)fs/btrfs/ioctl.c +4
-20@@ -3714,22 +3714,6 @@ static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)return ret; } -/* -* Quick check for ioctl handlers if quotas are enabled. Proper locking must be -* done before any operations. -*/ -static bool qgroup_enabled (struct btrfs_fs_info *fs_info) -{ -bool ret = true; - -mutex_lock (&fs_info->qgroup_ioctl_lock); -if (!fs_info->quota_root) - ret = false; -mutex_unlock (&fs_info->qgroup_ioctl_lock); - -return ret; -} -static long btrfs_ioctl_qgroup_assign (struct file *file,void __user *arg) {struct inode *inode =file_inode (file);@@ -3744,7 +3728,7 @@ static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);@@ -3814,7 +3798,7 @@ static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (root->fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);@@ -3873,7 +3857,7 @@ static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (root->fs_info)) +if (!btrfs_qgroup_enabled (root->fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);@@ -3921,7 +3905,7 @@ static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)if (!capable (CAP_SYS_ADMIN ))return -EPERM ; -if (!qgroup_enabled (fs_info)) +if (!btrfs_qgroup_enabled (fs_info))return -ENOTCONN ; ret =mnt_want_write_file (file);
Quoted 8 lines · On Tue, Jul 01, 2025 at 04:42:19PM +0100, [email protected] wrote:
From: Filipe Manana <[email protected]>
A couple fixes and a cleanup, details in the change logs.
Filipe Manana (3):
btrfs: qgroup: set quota enabled bit if quota disable fails flushing reservations
btrfs: qgroup: fix qgroup create ioctl returning success after quotas disabled
btrfs: qgroup: usebtrfs_qgroup_enabled()in ioctls