From 35faf3109a78516f60ca13f957083d5e5535fde0 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Tue, 21 Jun 2022 16:14:51 +0200 Subject: fs: port to iattr ownership update helpers Earlier we introduced new helpers to abstract ownership update and remove code duplication. This converts all filesystems supporting idmapped mounts to make use of these new helpers. For now we always pass the initial idmapping which makes the idmapping functions these helpers call nops. This is done because we currently always pass the actual value to be written to i_{g,u}id via struct iattr. While this allowed us to treat the {g,u}id values in struct iattr as values that can be directly written to inode->i_{g,u}id it also increases the potential for confusion for filesystems. Now that we are have dedicated types to prevent this confusion we will ultimately only map the value from the idmapped mount into a filesystem value that can be written to inode->i_{g,u}id when the filesystem actually updates the inode. So pass down the initial idmapping until we finished that conversion at which point we pass down the mount's idmapping. No functional changes intended. Link: https://lore.kernel.org/r/20220621141454.2914719-6-brauner@kernel.org Cc: Seth Forshee Cc: Christoph Hellwig Cc: Aleksa Sarai Cc: Linus Torvalds Cc: Al Viro CC: linux-fsdevel@vger.kernel.org Reviewed-by: Seth Forshee Signed-off-by: Christian Brauner (Microsoft) --- fs/ext2/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index e6b932219803..6dc66ab97d20 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -1684,8 +1684,8 @@ int ext2_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (error) return error; } - if ((iattr->ia_valid & ATTR_UID && !uid_eq(iattr->ia_uid, inode->i_uid)) || - (iattr->ia_valid & ATTR_GID && !gid_eq(iattr->ia_gid, inode->i_gid))) { + if (i_uid_needs_update(&init_user_ns, iattr, inode) || + i_gid_needs_update(&init_user_ns, iattr, inode)) { error = dquot_transfer(inode, iattr); if (error) return error; -- cgit v1.2.3 From 71e7b535b8900d7ce7d5279fa472711db5251ae5 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Tue, 21 Jun 2022 16:14:52 +0200 Subject: quota: port quota helpers mount ids Port the is_quota_modification() and dqout_transfer() helper to type safe vfs{g,u}id_t. Since these helpers are only called by a few filesystems don't introduce a new helper but simply extend the existing helpers to pass down the mount's idmapping. Note, that this is a non-functional change, i.e. nothing will have happened here or at the end of this series to how quota are done! This a change necessary because we will at the end of this series make ownership changes easier to reason about by keeping the original value in struct iattr for both non-idmapped and idmapped mounts. For now we always pass the initial idmapping which makes the idmapping functions these helpers call nops. This is done because we currently always pass the actual value to be written to i_{g,u}id via struct iattr. While this allowed us to treat the {g,u}id values in struct iattr as values that can be directly written to inode->i_{g,u}id it also increases the potential for confusion for filesystems. Now that we are have dedicated types to prevent this confusion we will ultimately only map the value from the idmapped mount into a filesystem value that can be written to inode->i_{g,u}id when the filesystem actually updates the inode. So pass down the initial idmapping until we finished that conversion at which point we pass down the mount's idmapping. Since struct iattr uses an anonymous union with overlapping types as supported by the C standard, filesystems that haven't converted to ia_vfs{g,u}id won't see any difference and things will continue to work as before. In other words, no functional changes intended with this change. Link: https://lore.kernel.org/r/20220621141454.2914719-7-brauner@kernel.org Cc: Seth Forshee Cc: Christoph Hellwig Cc: Jan Kara Cc: Aleksa Sarai Cc: Linus Torvalds Cc: Al Viro CC: linux-fsdevel@vger.kernel.org Reviewed-by: Jan Kara Reviewed-by: Seth Forshee Signed-off-by: Christian Brauner (Microsoft) --- fs/ext2/inode.c | 4 ++-- fs/ext4/inode.c | 4 ++-- fs/f2fs/file.c | 4 ++-- fs/f2fs/recovery.c | 2 +- fs/jfs/file.c | 4 ++-- fs/ocfs2/file.c | 2 +- fs/quota/dquot.c | 3 ++- fs/reiserfs/inode.c | 4 ++-- fs/zonefs/super.c | 2 +- include/linux/quotaops.h | 9 ++++++--- 10 files changed, 21 insertions(+), 17 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index 6dc66ab97d20..593b79416e0e 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -1679,14 +1679,14 @@ int ext2_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (error) return error; - if (is_quota_modification(inode, iattr)) { + if (is_quota_modification(&init_user_ns, inode, iattr)) { error = dquot_initialize(inode); if (error) return error; } if (i_uid_needs_update(&init_user_ns, iattr, inode) || i_gid_needs_update(&init_user_ns, iattr, inode)) { - error = dquot_transfer(inode, iattr); + error = dquot_transfer(&init_user_ns, inode, iattr); if (error) return error; } diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 05d932f81c53..72f08c184768 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -5350,7 +5350,7 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (error) return error; - if (is_quota_modification(inode, attr)) { + if (is_quota_modification(&init_user_ns, inode, attr)) { error = dquot_initialize(inode); if (error) return error; @@ -5374,7 +5374,7 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, * counts xattr inode references. */ down_read(&EXT4_I(inode)->xattr_sem); - error = dquot_transfer(inode, attr); + error = dquot_transfer(&init_user_ns, inode, attr); up_read(&EXT4_I(inode)->xattr_sem); if (error) { diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index a35d6b12bd63..02b2d56d4edc 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -915,7 +915,7 @@ int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (err) return err; - if (is_quota_modification(inode, attr)) { + if (is_quota_modification(&init_user_ns, inode, attr)) { err = f2fs_dquot_initialize(inode); if (err) return err; @@ -923,7 +923,7 @@ int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (i_uid_needs_update(&init_user_ns, attr, inode) || i_gid_needs_update(&init_user_ns, attr, inode)) { f2fs_lock_op(F2FS_I_SB(inode)); - err = dquot_transfer(inode, attr); + err = dquot_transfer(&init_user_ns, inode, attr); if (err) { set_sbi_flag(F2FS_I_SB(inode), SBI_QUOTA_NEED_REPAIR); diff --git a/fs/f2fs/recovery.c b/fs/f2fs/recovery.c index 3cb7f8a43b4d..8e5a089f1ac8 100644 --- a/fs/f2fs/recovery.c +++ b/fs/f2fs/recovery.c @@ -266,7 +266,7 @@ static int recover_quota_data(struct inode *inode, struct page *page) if (!attr.ia_valid) return 0; - err = dquot_transfer(inode, &attr); + err = dquot_transfer(&init_user_ns, inode, &attr); if (err) set_sbi_flag(F2FS_I_SB(inode), SBI_QUOTA_NEED_REPAIR); return err; diff --git a/fs/jfs/file.c b/fs/jfs/file.c index 1d732fd223d4..c18569b9895d 100644 --- a/fs/jfs/file.c +++ b/fs/jfs/file.c @@ -95,14 +95,14 @@ int jfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (rc) return rc; - if (is_quota_modification(inode, iattr)) { + if (is_quota_modification(&init_user_ns, inode, iattr)) { rc = dquot_initialize(inode); if (rc) return rc; } if ((iattr->ia_valid & ATTR_UID && !uid_eq(iattr->ia_uid, inode->i_uid)) || (iattr->ia_valid & ATTR_GID && !gid_eq(iattr->ia_gid, inode->i_gid))) { - rc = dquot_transfer(inode, iattr); + rc = dquot_transfer(&init_user_ns, inode, iattr); if (rc) return rc; } diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 7497cd592258..0e09cd8911da 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -1146,7 +1146,7 @@ int ocfs2_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (status) return status; - if (is_quota_modification(inode, attr)) { + if (is_quota_modification(&init_user_ns, inode, attr)) { status = dquot_initialize(inode); if (status) return status; diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c index 6cec2bfbf51b..df9af1ce2851 100644 --- a/fs/quota/dquot.c +++ b/fs/quota/dquot.c @@ -2085,7 +2085,8 @@ EXPORT_SYMBOL(__dquot_transfer); /* Wrapper for transferring ownership of an inode for uid/gid only * Called from FSXXX_setattr() */ -int dquot_transfer(struct inode *inode, struct iattr *iattr) +int dquot_transfer(struct user_namespace *mnt_userns, struct inode *inode, + struct iattr *iattr) { struct dquot *transfer_to[MAXQUOTAS] = {}; struct dquot *dquot; diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c index 0cffe054b78e..1e89e76972a0 100644 --- a/fs/reiserfs/inode.c +++ b/fs/reiserfs/inode.c @@ -3284,7 +3284,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, /* must be turned off for recursive notify_change calls */ ia_valid = attr->ia_valid &= ~(ATTR_KILL_SUID|ATTR_KILL_SGID); - if (is_quota_modification(inode, attr)) { + if (is_quota_modification(&init_user_ns, inode, attr)) { error = dquot_initialize(inode); if (error) return error; @@ -3367,7 +3367,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, reiserfs_write_unlock(inode->i_sb); if (error) goto out; - error = dquot_transfer(inode, attr); + error = dquot_transfer(&init_user_ns, inode, attr); reiserfs_write_lock(inode->i_sb); if (error) { journal_end(&th); diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c index 053299758deb..dd422b1d7320 100644 --- a/fs/zonefs/super.c +++ b/fs/zonefs/super.c @@ -616,7 +616,7 @@ static int zonefs_inode_setattr(struct user_namespace *mnt_userns, !uid_eq(iattr->ia_uid, inode->i_uid)) || ((iattr->ia_valid & ATTR_GID) && !gid_eq(iattr->ia_gid, inode->i_gid))) { - ret = dquot_transfer(inode, iattr); + ret = dquot_transfer(&init_user_ns, inode, iattr); if (ret) return ret; } diff --git a/include/linux/quotaops.h b/include/linux/quotaops.h index 61ee34861ca2..0342ff6584fd 100644 --- a/include/linux/quotaops.h +++ b/include/linux/quotaops.h @@ -20,7 +20,8 @@ static inline struct quota_info *sb_dqopt(struct super_block *sb) } /* i_mutex must being held */ -static inline bool is_quota_modification(struct inode *inode, struct iattr *ia) +static inline bool is_quota_modification(struct user_namespace *mnt_userns, + struct inode *inode, struct iattr *ia) { return ((ia->ia_valid & ATTR_SIZE) || i_uid_needs_update(&init_user_ns, ia, inode) || @@ -115,7 +116,8 @@ int dquot_set_dqblk(struct super_block *sb, struct kqid id, struct qc_dqblk *di); int __dquot_transfer(struct inode *inode, struct dquot **transfer_to); -int dquot_transfer(struct inode *inode, struct iattr *iattr); +int dquot_transfer(struct user_namespace *mnt_userns, struct inode *inode, + struct iattr *iattr); static inline struct mem_dqinfo *sb_dqinfo(struct super_block *sb, int type) { @@ -234,7 +236,8 @@ static inline void dquot_free_inode(struct inode *inode) { } -static inline int dquot_transfer(struct inode *inode, struct iattr *iattr) +static inline int dquot_transfer(struct user_namespace *mnt_userns, + struct inode *inode, struct iattr *iattr) { return 0; } -- cgit v1.2.3 From b27c82e1296572cfa3997e58db3118a33915f85c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Tue, 21 Jun 2022 16:14:54 +0200 Subject: attr: port attribute changes to new types Now that we introduced new infrastructure to increase the type safety for filesystems supporting idmapped mounts port the first part of the vfs over to them. This ports the attribute changes codepaths to rely on the new better helpers using a dedicated type. Before this change we used to take a shortcut and place the actual values that would be written to inode->i_{g,u}id into struct iattr. This had the advantage that we moved idmappings mostly out of the picture early on but it made reasoning about changes more difficult than it should be. The filesystem was never explicitly told that it dealt with an idmapped mount. The transition to the value that needed to be stored in inode->i_{g,u}id appeared way too early and increased the probability of bugs in various codepaths. We know place the same value in struct iattr no matter if this is an idmapped mount or not. The vfs will only deal with type safe vfs{g,u}id_t. This makes it massively safer to perform permission checks as the type will tell us what checks we need to perform and what helpers we need to use. Fileystems raising FS_ALLOW_IDMAP can't simply write ia_vfs{g,u}id to inode->i_{g,u}id since they are different types. Instead they need to use the dedicated vfs{g,u}id_to_k{g,u}id() helpers that map the vfs{g,u}id into the filesystem. The other nice effect is that filesystems like overlayfs don't need to care about idmappings explicitly anymore and can simply set up struct iattr accordingly directly. Link: https://lore.kernel.org/lkml/CAHk-=win6+ahs1EwLkcq8apqLi_1wXFWbrPf340zYEhObpz4jA@mail.gmail.com [1] Link: https://lore.kernel.org/r/20220621141454.2914719-9-brauner@kernel.org Cc: Seth Forshee Cc: Christoph Hellwig Cc: Aleksa Sarai Cc: Linus Torvalds Cc: Al Viro CC: linux-fsdevel@vger.kernel.org Reviewed-by: Seth Forshee Signed-off-by: Christian Brauner (Microsoft) --- fs/attr.c | 68 ++++++++++++++++++--------------------- fs/ext2/inode.c | 8 ++--- fs/ext4/inode.c | 12 +++---- fs/f2fs/file.c | 16 ++++----- fs/f2fs/recovery.c | 8 ++--- fs/fat/file.c | 8 +++-- fs/jfs/file.c | 4 +-- fs/ocfs2/file.c | 2 +- fs/open.c | 60 +++++++++++++++++++++++++--------- fs/overlayfs/copy_up.c | 4 +-- fs/overlayfs/overlayfs.h | 12 +------ fs/quota/dquot.c | 14 +++++--- fs/reiserfs/inode.c | 4 +-- fs/xfs/xfs_iops.c | 10 +++--- fs/zonefs/super.c | 2 +- include/linux/fs.h | 4 +++ include/linux/quotaops.h | 4 +-- security/integrity/evm/evm_main.c | 4 +-- 18 files changed, 137 insertions(+), 107 deletions(-) (limited to 'fs/ext2') diff --git a/fs/attr.c b/fs/attr.c index 88e2ca30d42e..1ba7ddef537f 100644 --- a/fs/attr.c +++ b/fs/attr.c @@ -31,15 +31,15 @@ * performed on the raw inode simply passs init_user_ns. */ static bool chown_ok(struct user_namespace *mnt_userns, - const struct inode *inode, - kuid_t uid) + const struct inode *inode, vfsuid_t ia_vfsuid) { - kuid_t kuid = i_uid_into_mnt(mnt_userns, inode); - if (uid_eq(current_fsuid(), kuid) && uid_eq(uid, inode->i_uid)) + vfsuid_t vfsuid = i_uid_into_vfsuid(mnt_userns, inode); + if (vfsuid_eq_kuid(vfsuid, current_fsuid()) && + vfsuid_eq(ia_vfsuid, vfsuid)) return true; if (capable_wrt_inode_uidgid(mnt_userns, inode, CAP_CHOWN)) return true; - if (uid_eq(kuid, INVALID_UID) && + if (!vfsuid_valid(vfsuid) && ns_capable(inode->i_sb->s_user_ns, CAP_CHOWN)) return true; return false; @@ -58,21 +58,19 @@ static bool chown_ok(struct user_namespace *mnt_userns, * performed on the raw inode simply passs init_user_ns. */ static bool chgrp_ok(struct user_namespace *mnt_userns, - const struct inode *inode, kgid_t gid) + const struct inode *inode, vfsgid_t ia_vfsgid) { - kgid_t kgid = i_gid_into_mnt(mnt_userns, inode); - if (uid_eq(current_fsuid(), i_uid_into_mnt(mnt_userns, inode))) { - kgid_t mapped_gid; - - if (gid_eq(gid, inode->i_gid)) + vfsgid_t vfsgid = i_gid_into_vfsgid(mnt_userns, inode); + vfsuid_t vfsuid = i_uid_into_vfsuid(mnt_userns, inode); + if (vfsuid_eq_kuid(vfsuid, current_fsuid())) { + if (vfsgid_eq(ia_vfsgid, vfsgid)) return true; - mapped_gid = mapped_kgid_fs(mnt_userns, i_user_ns(inode), gid); - if (in_group_p(mapped_gid)) + if (vfsgid_in_group_p(ia_vfsgid)) return true; } if (capable_wrt_inode_uidgid(mnt_userns, inode, CAP_CHOWN)) return true; - if (gid_eq(kgid, INVALID_GID) && + if (!vfsgid_valid(vfsgid) && ns_capable(inode->i_sb->s_user_ns, CAP_CHOWN)) return true; return false; @@ -120,28 +118,29 @@ int setattr_prepare(struct user_namespace *mnt_userns, struct dentry *dentry, goto kill_priv; /* Make sure a caller can chown. */ - if ((ia_valid & ATTR_UID) && !chown_ok(mnt_userns, inode, attr->ia_uid)) + if ((ia_valid & ATTR_UID) && + !chown_ok(mnt_userns, inode, attr->ia_vfsuid)) return -EPERM; /* Make sure caller can chgrp. */ - if ((ia_valid & ATTR_GID) && !chgrp_ok(mnt_userns, inode, attr->ia_gid)) + if ((ia_valid & ATTR_GID) && + !chgrp_ok(mnt_userns, inode, attr->ia_vfsgid)) return -EPERM; /* Make sure a caller can chmod. */ if (ia_valid & ATTR_MODE) { - kgid_t mapped_gid; + vfsgid_t vfsgid; if (!inode_owner_or_capable(mnt_userns, inode)) return -EPERM; if (ia_valid & ATTR_GID) - mapped_gid = mapped_kgid_fs(mnt_userns, - i_user_ns(inode), attr->ia_gid); + vfsgid = attr->ia_vfsgid; else - mapped_gid = i_gid_into_mnt(mnt_userns, inode); + vfsgid = i_gid_into_vfsgid(mnt_userns, inode); /* Also check the setgid bit! */ - if (!in_group_p(mapped_gid) && + if (!vfsgid_in_group_p(vfsgid) && !capable_wrt_inode_uidgid(mnt_userns, inode, CAP_FSETID)) attr->ia_mode &= ~S_ISGID; } @@ -219,9 +218,7 @@ EXPORT_SYMBOL(inode_newsize_ok); * setattr_copy must be called with i_mutex held. * * setattr_copy updates the inode's metadata with that specified - * in attr on idmapped mounts. If file ownership is changed setattr_copy - * doesn't map ia_uid and ia_gid. It will asssume the caller has already - * provided the intended values. Necessary permission checks to determine + * in attr on idmapped mounts. Necessary permission checks to determine * whether or not the S_ISGID property needs to be removed are performed with * the correct idmapped mount permission helpers. * Noticeably missing is inode size update, which is more complex @@ -242,8 +239,8 @@ void setattr_copy(struct user_namespace *mnt_userns, struct inode *inode, { unsigned int ia_valid = attr->ia_valid; - i_uid_update(&init_user_ns, attr, inode); - i_gid_update(&init_user_ns, attr, inode); + i_uid_update(mnt_userns, attr, inode); + i_gid_update(mnt_userns, attr, inode); if (ia_valid & ATTR_ATIME) inode->i_atime = attr->ia_atime; if (ia_valid & ATTR_MTIME) @@ -252,8 +249,8 @@ void setattr_copy(struct user_namespace *mnt_userns, struct inode *inode, inode->i_ctime = attr->ia_ctime; if (ia_valid & ATTR_MODE) { umode_t mode = attr->ia_mode; - kgid_t kgid = i_gid_into_mnt(mnt_userns, inode); - if (!in_group_p(kgid) && + vfsgid_t vfsgid = i_gid_into_vfsgid(mnt_userns, inode); + if (!vfsgid_in_group_p(vfsgid) && !capable_wrt_inode_uidgid(mnt_userns, inode, CAP_FSETID)) mode &= ~S_ISGID; inode->i_mode = mode; @@ -304,9 +301,6 @@ EXPORT_SYMBOL(may_setattr); * retry. Because breaking a delegation may take a long time, the * caller should drop the i_mutex before doing so. * - * If file ownership is changed notify_change() doesn't map ia_uid and - * ia_gid. It will asssume the caller has already provided the intended values. - * * Alternatively, a caller may pass NULL for delegated_inode. This may * be appropriate for callers that expect the underlying filesystem not * to be NFS exported. Also, passing NULL is fine for callers holding @@ -395,23 +389,25 @@ int notify_change(struct user_namespace *mnt_userns, struct dentry *dentry, * namespace of the superblock. */ if (ia_valid & ATTR_UID && - !kuid_has_mapping(inode->i_sb->s_user_ns, attr->ia_uid)) + !vfsuid_has_fsmapping(mnt_userns, inode->i_sb->s_user_ns, + attr->ia_vfsuid)) return -EOVERFLOW; if (ia_valid & ATTR_GID && - !kgid_has_mapping(inode->i_sb->s_user_ns, attr->ia_gid)) + !vfsgid_has_fsmapping(mnt_userns, inode->i_sb->s_user_ns, + attr->ia_vfsgid)) return -EOVERFLOW; /* Don't allow modifications of files with invalid uids or * gids unless those uids & gids are being made valid. */ if (!(ia_valid & ATTR_UID) && - !uid_valid(i_uid_into_mnt(mnt_userns, inode))) + !vfsuid_valid(i_uid_into_vfsuid(mnt_userns, inode))) return -EOVERFLOW; if (!(ia_valid & ATTR_GID) && - !gid_valid(i_gid_into_mnt(mnt_userns, inode))) + !vfsgid_valid(i_gid_into_vfsgid(mnt_userns, inode))) return -EOVERFLOW; - error = security_inode_setattr(&init_user_ns, dentry, attr); + error = security_inode_setattr(mnt_userns, dentry, attr); if (error) return error; error = try_break_deleg(inode, delegated_inode); diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index 593b79416e0e..7a192e4e7fa9 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -1679,14 +1679,14 @@ int ext2_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (error) return error; - if (is_quota_modification(&init_user_ns, inode, iattr)) { + if (is_quota_modification(mnt_userns, inode, iattr)) { error = dquot_initialize(inode); if (error) return error; } - if (i_uid_needs_update(&init_user_ns, iattr, inode) || - i_gid_needs_update(&init_user_ns, iattr, inode)) { - error = dquot_transfer(&init_user_ns, inode, iattr); + if (i_uid_needs_update(mnt_userns, iattr, inode) || + i_gid_needs_update(mnt_userns, iattr, inode)) { + error = dquot_transfer(mnt_userns, inode, iattr); if (error) return error; } diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 72f08c184768..3dcc1dd1f179 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -5350,14 +5350,14 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (error) return error; - if (is_quota_modification(&init_user_ns, inode, attr)) { + if (is_quota_modification(mnt_userns, inode, attr)) { error = dquot_initialize(inode); if (error) return error; } - if (i_uid_needs_update(&init_user_ns, attr, inode) || - i_gid_needs_update(&init_user_ns, attr, inode)) { + if (i_uid_needs_update(mnt_userns, attr, inode) || + i_gid_needs_update(mnt_userns, attr, inode)) { handle_t *handle; /* (user+group)*(old+new) structure, inode write (sb, @@ -5374,7 +5374,7 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, * counts xattr inode references. */ down_read(&EXT4_I(inode)->xattr_sem); - error = dquot_transfer(&init_user_ns, inode, attr); + error = dquot_transfer(mnt_userns, inode, attr); up_read(&EXT4_I(inode)->xattr_sem); if (error) { @@ -5383,8 +5383,8 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, } /* Update corresponding info in inode so that everything is in * one transaction */ - i_uid_update(&init_user_ns, attr, inode); - i_gid_update(&init_user_ns, attr, inode); + i_uid_update(mnt_userns, attr, inode); + i_gid_update(mnt_userns, attr, inode); error = ext4_mark_inode_dirty(handle, inode); ext4_journal_stop(handle); if (unlikely(error)) { diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 02b2d56d4edc..d66e37d80a2d 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -861,8 +861,8 @@ static void __setattr_copy(struct user_namespace *mnt_userns, { unsigned int ia_valid = attr->ia_valid; - i_uid_update(&init_user_ns, attr, inode); - i_gid_update(&init_user_ns, attr, inode); + i_uid_update(mnt_userns, attr, inode); + i_gid_update(mnt_userns, attr, inode); if (ia_valid & ATTR_ATIME) inode->i_atime = attr->ia_atime; if (ia_valid & ATTR_MTIME) @@ -915,15 +915,15 @@ int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (err) return err; - if (is_quota_modification(&init_user_ns, inode, attr)) { + if (is_quota_modification(mnt_userns, inode, attr)) { err = f2fs_dquot_initialize(inode); if (err) return err; } - if (i_uid_needs_update(&init_user_ns, attr, inode) || - i_gid_needs_update(&init_user_ns, attr, inode)) { + if (i_uid_needs_update(mnt_userns, attr, inode) || + i_gid_needs_update(mnt_userns, attr, inode)) { f2fs_lock_op(F2FS_I_SB(inode)); - err = dquot_transfer(&init_user_ns, inode, attr); + err = dquot_transfer(mnt_userns, inode, attr); if (err) { set_sbi_flag(F2FS_I_SB(inode), SBI_QUOTA_NEED_REPAIR); @@ -934,8 +934,8 @@ int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, * update uid/gid under lock_op(), so that dquot and inode can * be updated atomically. */ - i_uid_update(&init_user_ns, attr, inode); - i_gid_update(&init_user_ns, attr, inode); + i_uid_update(mnt_userns, attr, inode); + i_gid_update(mnt_userns, attr, inode); f2fs_mark_inode_dirty_sync(inode, true); f2fs_unlock_op(F2FS_I_SB(inode)); } diff --git a/fs/f2fs/recovery.c b/fs/f2fs/recovery.c index 8e5a089f1ac8..dcd0a1e35095 100644 --- a/fs/f2fs/recovery.c +++ b/fs/f2fs/recovery.c @@ -255,12 +255,12 @@ static int recover_quota_data(struct inode *inode, struct page *page) memset(&attr, 0, sizeof(attr)); - attr.ia_uid = make_kuid(inode->i_sb->s_user_ns, i_uid); - attr.ia_gid = make_kgid(inode->i_sb->s_user_ns, i_gid); + attr.ia_vfsuid = VFSUIDT_INIT(make_kuid(inode->i_sb->s_user_ns, i_uid)); + attr.ia_vfsgid = VFSGIDT_INIT(make_kgid(inode->i_sb->s_user_ns, i_gid)); - if (!uid_eq(attr.ia_uid, inode->i_uid)) + if (!vfsuid_eq(attr.ia_vfsuid, i_uid_into_vfsuid(&init_user_ns, inode))) attr.ia_valid |= ATTR_UID; - if (!gid_eq(attr.ia_gid, inode->i_gid)) + if (!vfsgid_eq(attr.ia_vfsgid, i_gid_into_vfsgid(&init_user_ns, inode))) attr.ia_valid |= ATTR_GID; if (!attr.ia_valid) diff --git a/fs/fat/file.c b/fs/fat/file.c index 530f18173db2..3e4eb3467cb4 100644 --- a/fs/fat/file.c +++ b/fs/fat/file.c @@ -90,7 +90,7 @@ static int fat_ioctl_set_attributes(struct file *file, u32 __user *user_attr) * out the RO attribute for checking by the security * module, just because it maps to a file mode. */ - err = security_inode_setattr(&init_user_ns, + err = security_inode_setattr(file_mnt_user_ns(file), file->f_path.dentry, &ia); if (err) goto out_unlock_inode; @@ -517,9 +517,11 @@ int fat_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, } if (((attr->ia_valid & ATTR_UID) && - (!uid_eq(attr->ia_uid, sbi->options.fs_uid))) || + (!uid_eq(from_vfsuid(mnt_userns, i_user_ns(inode), attr->ia_vfsuid), + sbi->options.fs_uid))) || ((attr->ia_valid & ATTR_GID) && - (!gid_eq(attr->ia_gid, sbi->options.fs_gid))) || + (!gid_eq(from_vfsgid(mnt_userns, i_user_ns(inode), attr->ia_vfsgid), + sbi->options.fs_gid))) || ((attr->ia_valid & ATTR_MODE) && (attr->ia_mode & ~FAT_VALID_MODE))) error = -EPERM; diff --git a/fs/jfs/file.c b/fs/jfs/file.c index c18569b9895d..332dc9ac47a9 100644 --- a/fs/jfs/file.c +++ b/fs/jfs/file.c @@ -95,14 +95,14 @@ int jfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (rc) return rc; - if (is_quota_modification(&init_user_ns, inode, iattr)) { + if (is_quota_modification(mnt_userns, inode, iattr)) { rc = dquot_initialize(inode); if (rc) return rc; } if ((iattr->ia_valid & ATTR_UID && !uid_eq(iattr->ia_uid, inode->i_uid)) || (iattr->ia_valid & ATTR_GID && !gid_eq(iattr->ia_gid, inode->i_gid))) { - rc = dquot_transfer(&init_user_ns, inode, iattr); + rc = dquot_transfer(mnt_userns, inode, iattr); if (rc) return rc; } diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 0e09cd8911da..9c67edd215d5 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -1146,7 +1146,7 @@ int ocfs2_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, if (status) return status; - if (is_quota_modification(&init_user_ns, inode, attr)) { + if (is_quota_modification(mnt_userns, inode, attr)) { status = dquot_initialize(inode); if (status) return status; diff --git a/fs/open.c b/fs/open.c index 1d57fbde2feb..2790aac66e58 100644 --- a/fs/open.c +++ b/fs/open.c @@ -663,6 +663,42 @@ SYSCALL_DEFINE2(chmod, const char __user *, filename, umode_t, mode) return do_fchmodat(AT_FDCWD, filename, mode); } +/** + * setattr_vfsuid - check and set ia_fsuid attribute + * @kuid: new inode owner + * + * Check whether @kuid is valid and if so generate and set vfsuid_t in + * ia_vfsuid. + * + * Return: true if @kuid is valid, false if not. + */ +static inline bool setattr_vfsuid(struct iattr *attr, kuid_t kuid) +{ + if (!uid_valid(kuid)) + return false; + attr->ia_valid |= ATTR_UID; + attr->ia_vfsuid = VFSUIDT_INIT(kuid); + return true; +} + +/** + * setattr_vfsgid - check and set ia_fsgid attribute + * @kgid: new inode owner + * + * Check whether @kgid is valid and if so generate and set vfsgid_t in + * ia_vfsgid. + * + * Return: true if @kgid is valid, false if not. + */ +static inline bool setattr_vfsgid(struct iattr *attr, kgid_t kgid) +{ + if (!gid_valid(kgid)) + return false; + attr->ia_valid |= ATTR_GID; + attr->ia_vfsgid = VFSGIDT_INIT(kgid); + return true; +} + int chown_common(const struct path *path, uid_t user, gid_t group) { struct user_namespace *mnt_userns, *fs_userns; @@ -678,28 +714,22 @@ int chown_common(const struct path *path, uid_t user, gid_t group) mnt_userns = mnt_user_ns(path->mnt); fs_userns = i_user_ns(inode); - uid = mapped_kuid_user(mnt_userns, fs_userns, uid); - gid = mapped_kgid_user(mnt_userns, fs_userns, gid); retry_deleg: newattrs.ia_valid = ATTR_CTIME; - if (user != (uid_t) -1) { - if (!uid_valid(uid)) - return -EINVAL; - newattrs.ia_valid |= ATTR_UID; - newattrs.ia_uid = uid; - } - if (group != (gid_t) -1) { - if (!gid_valid(gid)) - return -EINVAL; - newattrs.ia_valid |= ATTR_GID; - newattrs.ia_gid = gid; - } + if ((user != (uid_t)-1) && !setattr_vfsuid(&newattrs, uid)) + return -EINVAL; + if ((group != (gid_t)-1) && !setattr_vfsgid(&newattrs, gid)) + return -EINVAL; if (!S_ISDIR(inode->i_mode)) newattrs.ia_valid |= ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV; inode_lock(inode); - error = security_path_chown(path, uid, gid); + /* Continue to send actual fs values, not the mount values. */ + error = security_path_chown( + path, + from_vfsuid(mnt_userns, fs_userns, newattrs.ia_vfsuid), + from_vfsgid(mnt_userns, fs_userns, newattrs.ia_vfsgid)); if (!error) error = notify_change(mnt_userns, path->dentry, &newattrs, &delegated_inode); diff --git a/fs/overlayfs/copy_up.c b/fs/overlayfs/copy_up.c index 714ec569d25b..245e2cb62708 100644 --- a/fs/overlayfs/copy_up.c +++ b/fs/overlayfs/copy_up.c @@ -331,8 +331,8 @@ int ovl_set_attr(struct ovl_fs *ofs, struct dentry *upperdentry, if (!err) { struct iattr attr = { .ia_valid = ATTR_UID | ATTR_GID, - .ia_uid = stat->uid, - .ia_gid = stat->gid, + .ia_vfsuid = VFSUIDT_INIT(stat->uid), + .ia_vfsgid = VFSGIDT_INIT(stat->gid), }; err = ovl_do_notify_change(ofs, upperdentry, &attr); } diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index 4f34b7e02eee..e22e20f4811a 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -139,17 +139,7 @@ static inline int ovl_do_notify_change(struct ovl_fs *ofs, struct dentry *upperdentry, struct iattr *attr) { - struct user_namespace *upper_mnt_userns = ovl_upper_mnt_userns(ofs); - struct user_namespace *fs_userns = i_user_ns(d_inode(upperdentry)); - - if (attr->ia_valid & ATTR_UID) - attr->ia_uid = mapped_kuid_user(upper_mnt_userns, - fs_userns, attr->ia_uid); - if (attr->ia_valid & ATTR_GID) - attr->ia_gid = mapped_kgid_user(upper_mnt_userns, - fs_userns, attr->ia_gid); - - return notify_change(upper_mnt_userns, upperdentry, attr, NULL); + return notify_change(ovl_upper_mnt_userns(ofs), upperdentry, attr, NULL); } static inline int ovl_do_rmdir(struct ovl_fs *ofs, diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c index df9af1ce2851..28966da7834e 100644 --- a/fs/quota/dquot.c +++ b/fs/quota/dquot.c @@ -2096,8 +2096,11 @@ int dquot_transfer(struct user_namespace *mnt_userns, struct inode *inode, if (!dquot_active(inode)) return 0; - if (i_uid_needs_update(&init_user_ns, iattr, inode)) { - dquot = dqget(sb, make_kqid_uid(iattr->ia_uid)); + if (i_uid_needs_update(mnt_userns, iattr, inode)) { + kuid_t kuid = from_vfsuid(mnt_userns, i_user_ns(inode), + iattr->ia_vfsuid); + + dquot = dqget(sb, make_kqid_uid(kuid)); if (IS_ERR(dquot)) { if (PTR_ERR(dquot) != -ESRCH) { ret = PTR_ERR(dquot); @@ -2107,8 +2110,11 @@ int dquot_transfer(struct user_namespace *mnt_userns, struct inode *inode, } transfer_to[USRQUOTA] = dquot; } - if (i_gid_needs_update(&init_user_ns, iattr, inode)) { - dquot = dqget(sb, make_kqid_gid(iattr->ia_gid)); + if (i_gid_needs_update(mnt_userns, iattr, inode)) { + kgid_t kgid = from_vfsgid(mnt_userns, i_user_ns(inode), + iattr->ia_vfsgid); + + dquot = dqget(sb, make_kqid_gid(kgid)); if (IS_ERR(dquot)) { if (PTR_ERR(dquot) != -ESRCH) { ret = PTR_ERR(dquot); diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c index 1e89e76972a0..1141053b96ed 100644 --- a/fs/reiserfs/inode.c +++ b/fs/reiserfs/inode.c @@ -3284,7 +3284,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, /* must be turned off for recursive notify_change calls */ ia_valid = attr->ia_valid &= ~(ATTR_KILL_SUID|ATTR_KILL_SGID); - if (is_quota_modification(&init_user_ns, inode, attr)) { + if (is_quota_modification(mnt_userns, inode, attr)) { error = dquot_initialize(inode); if (error) return error; @@ -3367,7 +3367,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, reiserfs_write_unlock(inode->i_sb); if (error) goto out; - error = dquot_transfer(&init_user_ns, inode, attr); + error = dquot_transfer(mnt_userns, inode, attr); reiserfs_write_lock(inode->i_sb); if (error) { journal_end(&th); diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c index 31ec29565fb4..a7402f6ea510 100644 --- a/fs/xfs/xfs_iops.c +++ b/fs/xfs/xfs_iops.c @@ -667,13 +667,15 @@ xfs_setattr_nonsize( uint qflags = 0; if ((mask & ATTR_UID) && XFS_IS_UQUOTA_ON(mp)) { - uid = iattr->ia_uid; + uid = from_vfsuid(mnt_userns, i_user_ns(inode), + iattr->ia_vfsuid); qflags |= XFS_QMOPT_UQUOTA; } else { uid = inode->i_uid; } if ((mask & ATTR_GID) && XFS_IS_GQUOTA_ON(mp)) { - gid = iattr->ia_gid; + gid = from_vfsgid(mnt_userns, i_user_ns(inode), + iattr->ia_vfsgid); qflags |= XFS_QMOPT_GQUOTA; } else { gid = inode->i_gid; @@ -705,12 +707,12 @@ xfs_setattr_nonsize( * also. */ if (XFS_IS_UQUOTA_ON(mp) && - i_uid_needs_update(&init_user_ns, iattr, inode)) { + i_uid_needs_update(mnt_userns, iattr, inode)) { ASSERT(udqp); old_udqp = xfs_qm_vop_chown(tp, ip, &ip->i_udquot, udqp); } if (XFS_IS_GQUOTA_ON(mp) && - i_gid_needs_update(&init_user_ns, iattr, inode)) { + i_gid_needs_update(mnt_userns, iattr, inode)) { ASSERT(xfs_has_pquotino(mp) || !XFS_IS_PQUOTA_ON(mp)); ASSERT(gdqp); old_gdqp = xfs_qm_vop_chown(tp, ip, &ip->i_gdquot, gdqp); diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c index dd422b1d7320..f5d8338967cb 100644 --- a/fs/zonefs/super.c +++ b/fs/zonefs/super.c @@ -616,7 +616,7 @@ static int zonefs_inode_setattr(struct user_namespace *mnt_userns, !uid_eq(iattr->ia_uid, inode->i_uid)) || ((iattr->ia_valid & ATTR_GID) && !gid_eq(iattr->ia_gid, inode->i_gid))) { - ret = dquot_transfer(&init_user_ns, inode, iattr); + ret = dquot_transfer(mnt_userns, inode, iattr); if (ret) return ret; } diff --git a/include/linux/fs.h b/include/linux/fs.h index 0f8cc7f2665a..d6e3347cbf69 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -228,6 +228,10 @@ struct iattr { * are a dedicated type requiring the filesystem to use the dedicated * helpers. Other filesystem can continue to use ia_{g,u}id until they * have been ported. + * + * They always contain the same value. In other words FS_ALLOW_IDMAP + * pass down the same value on idmapped mounts as they would on regular + * mounts. */ union { kuid_t ia_uid; diff --git a/include/linux/quotaops.h b/include/linux/quotaops.h index 0342ff6584fd..0d8625d71733 100644 --- a/include/linux/quotaops.h +++ b/include/linux/quotaops.h @@ -24,8 +24,8 @@ static inline bool is_quota_modification(struct user_namespace *mnt_userns, struct inode *inode, struct iattr *ia) { return ((ia->ia_valid & ATTR_SIZE) || - i_uid_needs_update(&init_user_ns, ia, inode) || - i_gid_needs_update(&init_user_ns, ia, inode)); + i_uid_needs_update(mnt_userns, ia, inode) || + i_gid_needs_update(mnt_userns, ia, inode)); } #if defined(CONFIG_QUOTA) diff --git a/security/integrity/evm/evm_main.c b/security/integrity/evm/evm_main.c index 7f4af5b58583..93e8bc047a73 100644 --- a/security/integrity/evm/evm_main.c +++ b/security/integrity/evm/evm_main.c @@ -761,8 +761,8 @@ static int evm_attr_change(struct user_namespace *mnt_userns, struct inode *inode = d_backing_inode(dentry); unsigned int ia_valid = attr->ia_valid; - if (!i_uid_needs_update(&init_user_ns, attr, inode) && - !i_gid_needs_update(&init_user_ns, attr, inode) && + if (!i_uid_needs_update(mnt_userns, attr, inode) && + !i_gid_needs_update(mnt_userns, attr, inode) && (!(ia_valid & ATTR_MODE) || attr->ia_mode == inode->i_mode)) return 0; -- cgit v1.2.3 From 750cd7d0e65f2f65c52932ea48b2702fee6ab7a7 Mon Sep 17 00:00:00 2001 From: "Matthew Wilcox (Oracle)" Date: Tue, 17 May 2022 18:12:25 -0400 Subject: ext2: Remove check for PageError If read_mapping_page() encounters an error, it returns an errno, not a page with PageError set, so this test is not needed. Signed-off-by: Matthew Wilcox (Oracle) --- fs/ext2/dir.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/dir.c b/fs/ext2/dir.c index 43de293cef56..8326b63f0b70 100644 --- a/fs/ext2/dir.c +++ b/fs/ext2/dir.c @@ -204,8 +204,7 @@ static struct page * ext2_get_page(struct inode *dir, unsigned long n, if (!IS_ERR(page)) { *page_addr = kmap_local_page(page); if (unlikely(!PageChecked(page))) { - if (PageError(page) || !ext2_check_page(page, quiet, - *page_addr)) + if (!ext2_check_page(page, quiet, *page_addr)) goto fail; } } -- cgit v1.2.3 From 9888725d8a640d67abe6139e84ff41bafe89b11e Mon Sep 17 00:00:00 2001 From: Jiangshan Yi Date: Thu, 14 Jul 2022 14:33:18 +0800 Subject: fs/ext2: replace ternary operator with min_t() Fix the following coccicheck warning: fs/ext2/super.c:1494: WARNING opportunity for min(). fs/ext2/super.c:1533: WARNING opportunity for min(). min_t() macro is defined in include/linux/minmax.h. It avoids multiple evaluations of the arguments when non-constant and performs strict type-checking. Link: https://lore.kernel.org/r/20220714063318.1777139-1-13667453960@163.com Signed-off-by: Jiangshan Yi Signed-off-by: Jan Kara --- fs/ext2/super.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/super.c b/fs/ext2/super.c index f6a19f6d9f6d..300f2f0cf566 100644 --- a/fs/ext2/super.c +++ b/fs/ext2/super.c @@ -1490,8 +1490,7 @@ static ssize_t ext2_quota_read(struct super_block *sb, int type, char *data, len = i_size-off; toread = len; while (toread > 0) { - tocopy = sb->s_blocksize - offset < toread ? - sb->s_blocksize - offset : toread; + tocopy = min_t(size_t, sb->s_blocksize - offset, toread); tmp_bh.b_state = 0; tmp_bh.b_size = sb->s_blocksize; @@ -1529,8 +1528,7 @@ static ssize_t ext2_quota_write(struct super_block *sb, int type, struct buffer_head *bh; while (towrite > 0) { - tocopy = sb->s_blocksize - offset < towrite ? - sb->s_blocksize - offset : towrite; + tocopy = min_t(size_t, sb->s_blocksize - offset, towrite); tmp_bh.b_state = 0; tmp_bh.b_size = sb->s_blocksize; -- cgit v1.2.3 From 8012b866085523758780850087102421dbcce522 Mon Sep 17 00:00:00 2001 From: Shiyang Ruan Date: Fri, 3 Jun 2022 13:37:25 +0800 Subject: dax: introduce holder for dax_device Patch series "v14 fsdax-rmap + v11 fsdax-reflink", v2. The patchset fsdax-rmap is aimed to support shared pages tracking for fsdax. It moves owner tracking from dax_assocaite_entry() to pmem device driver, by introducing an interface ->memory_failure() for struct pagemap. This interface is called by memory_failure() in mm, and implemented by pmem device. Then call holder operations to find the filesystem which the corrupted data located in, and call filesystem handler to track files or metadata associated with this page. Finally we are able to try to fix the corrupted data in filesystem and do other necessary processing, such as killing processes who are using the files affected. The call trace is like this: memory_failure() |* fsdax case |------------ |pgmap->ops->memory_failure() => pmem_pgmap_memory_failure() | dax_holder_notify_failure() => | dax_device->holder_ops->notify_failure() => | - xfs_dax_notify_failure() | |* xfs_dax_notify_failure() | |-------------------------- | | xfs_rmap_query_range() | | xfs_dax_failure_fn() | | * corrupted on metadata | | try to recover data, call xfs_force_shutdown() | | * corrupted on file data | | try to recover data, call mf_dax_kill_procs() |* normal case |------------- |mf_generic_kill_procs() The patchset fsdax-reflink attempts to add CoW support for fsdax, and takes XFS, which has both reflink and fsdax features, as an example. One of the key mechanisms needed to be implemented in fsdax is CoW. Copy the data from srcmap before we actually write data to the destination iomap. And we just copy range in which data won't be changed. Another mechanism is range comparison. In page cache case, readpage() is used to load data on disk to page cache in order to be able to compare data. In fsdax case, readpage() does not work. So, we need another compare data with direct access support. With the two mechanisms implemented in fsdax, we are able to make reflink and fsdax work together in XFS. This patch (of 14): To easily track filesystem from a pmem device, we introduce a holder for dax_device structure, and also its operation. This holder is used to remember who is using this dax_device: - When it is the backend of a filesystem, the holder will be the instance of this filesystem. - When this pmem device is one of the targets in a mapped device, the holder will be this mapped device. In this case, the mapped device has its own dax_device and it will follow the first rule. So that we can finally track to the filesystem we needed. The holder and holder_ops will be set when filesystem is being mounted, or an target device is being activated. Link: https://lkml.kernel.org/r/20220603053738.1218681-1-ruansy.fnst@fujitsu.com Link: https://lkml.kernel.org/r/20220603053738.1218681-2-ruansy.fnst@fujitsu.com Signed-off-by: Shiyang Ruan Reviewed-by: Christoph Hellwig Reviewed-by: Dan Williams Reviewed-by: Darrick J. Wong Cc: Dave Chinner Cc: Jane Chu Cc: Goldwyn Rodrigues Cc: Al Viro Cc: Matthew Wilcox Cc: Naoya Horiguchi Cc: Miaohe Lin Cc: Dan Williams Cc: Goldwyn Rodrigues Cc: Ritesh Harjani Signed-off-by: Andrew Morton --- drivers/dax/super.c | 67 ++++++++++++++++++++++++++++++++++++++++++++++++++++- drivers/md/dm.c | 2 +- fs/erofs/super.c | 10 ++++---- fs/ext2/super.c | 7 +++--- fs/ext4/super.c | 9 +++---- fs/xfs/xfs_buf.c | 5 ++-- include/linux/dax.h | 33 +++++++++++++++++++------- 7 files changed, 110 insertions(+), 23 deletions(-) (limited to 'fs/ext2') diff --git a/drivers/dax/super.c b/drivers/dax/super.c index 50a08b2ec247..9b5e2a5eb0ae 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -22,6 +22,8 @@ * @private: dax driver private data * @flags: state and boolean properties * @ops: operations for this device + * @holder_data: holder of a dax_device: could be filesystem or mapped device + * @holder_ops: operations for the inner holder */ struct dax_device { struct inode inode; @@ -29,6 +31,8 @@ struct dax_device { void *private; unsigned long flags; const struct dax_operations *ops; + void *holder_data; + const struct dax_holder_operations *holder_ops; }; static dev_t dax_devt; @@ -71,8 +75,11 @@ EXPORT_SYMBOL_GPL(dax_remove_host); * fs_dax_get_by_bdev() - temporary lookup mechanism for filesystem-dax * @bdev: block device to find a dax_device for * @start_off: returns the byte offset into the dax_device that @bdev starts + * @holder: filesystem or mapped device inside the dax_device + * @ops: operations for the inner holder */ -struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, u64 *start_off) +struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, u64 *start_off, + void *holder, const struct dax_holder_operations *ops) { struct dax_device *dax_dev; u64 part_size; @@ -92,11 +99,26 @@ struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, u64 *start_off) dax_dev = xa_load(&dax_hosts, (unsigned long)bdev->bd_disk); if (!dax_dev || !dax_alive(dax_dev) || !igrab(&dax_dev->inode)) dax_dev = NULL; + else if (holder) { + if (!cmpxchg(&dax_dev->holder_data, NULL, holder)) + dax_dev->holder_ops = ops; + else + dax_dev = NULL; + } dax_read_unlock(id); return dax_dev; } EXPORT_SYMBOL_GPL(fs_dax_get_by_bdev); + +void fs_put_dax(struct dax_device *dax_dev, void *holder) +{ + if (dax_dev && holder && + cmpxchg(&dax_dev->holder_data, holder, NULL) == holder) + dax_dev->holder_ops = NULL; + put_dax(dax_dev); +} +EXPORT_SYMBOL_GPL(fs_put_dax); #endif /* CONFIG_BLOCK && CONFIG_FS_DAX */ enum dax_device_flags { @@ -204,6 +226,29 @@ size_t dax_recovery_write(struct dax_device *dax_dev, pgoff_t pgoff, } EXPORT_SYMBOL_GPL(dax_recovery_write); +int dax_holder_notify_failure(struct dax_device *dax_dev, u64 off, + u64 len, int mf_flags) +{ + int rc, id; + + id = dax_read_lock(); + if (!dax_alive(dax_dev)) { + rc = -ENXIO; + goto out; + } + + if (!dax_dev->holder_ops) { + rc = -EOPNOTSUPP; + goto out; + } + + rc = dax_dev->holder_ops->notify_failure(dax_dev, off, len, mf_flags); +out: + dax_read_unlock(id); + return rc; +} +EXPORT_SYMBOL_GPL(dax_holder_notify_failure); + #ifdef CONFIG_ARCH_HAS_PMEM_API void arch_wb_cache_pmem(void *addr, size_t size); void dax_flush(struct dax_device *dax_dev, void *addr, size_t size) @@ -277,8 +322,15 @@ void kill_dax(struct dax_device *dax_dev) if (!dax_dev) return; + if (dax_dev->holder_data != NULL) + dax_holder_notify_failure(dax_dev, 0, U64_MAX, 0); + clear_bit(DAXDEV_ALIVE, &dax_dev->flags); synchronize_srcu(&dax_srcu); + + /* clear holder data */ + dax_dev->holder_ops = NULL; + dax_dev->holder_data = NULL; } EXPORT_SYMBOL_GPL(kill_dax); @@ -420,6 +472,19 @@ void put_dax(struct dax_device *dax_dev) } EXPORT_SYMBOL_GPL(put_dax); +/** + * dax_holder() - obtain the holder of a dax device + * @dax_dev: a dax_device instance + + * Return: the holder's data which represents the holder if registered, + * otherwize NULL. + */ +void *dax_holder(struct dax_device *dax_dev) +{ + return dax_dev->holder_data; +} +EXPORT_SYMBOL_GPL(dax_holder); + /** * inode_dax: convert a public inode into its dax_dev * @inode: An inode with i_cdev pointing to a dax_dev diff --git a/drivers/md/dm.c b/drivers/md/dm.c index 2b75f1ef7386..0177a4ce9a18 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c @@ -758,7 +758,7 @@ static int open_table_device(struct table_device *td, dev_t dev, } td->dm_dev.bdev = bdev; - td->dm_dev.dax_dev = fs_dax_get_by_bdev(bdev, &part_off); + td->dm_dev.dax_dev = fs_dax_get_by_bdev(bdev, &part_off, NULL, NULL); return 0; } diff --git a/fs/erofs/super.c b/fs/erofs/super.c index 95addc5c9d34..3173debeaa5a 100644 --- a/fs/erofs/super.c +++ b/fs/erofs/super.c @@ -255,7 +255,8 @@ static int erofs_init_device(struct erofs_buf *buf, struct super_block *sb, if (IS_ERR(bdev)) return PTR_ERR(bdev); dif->bdev = bdev; - dif->dax_dev = fs_dax_get_by_bdev(bdev, &dif->dax_part_off); + dif->dax_dev = fs_dax_get_by_bdev(bdev, &dif->dax_part_off, + NULL, NULL); } dif->blocks = le32_to_cpu(dis->blocks); @@ -720,7 +721,8 @@ static int erofs_fc_fill_super(struct super_block *sb, struct fs_context *fc) } sbi->dax_dev = fs_dax_get_by_bdev(sb->s_bdev, - &sbi->dax_part_off); + &sbi->dax_part_off, + NULL, NULL); } err = erofs_read_superblock(sb); @@ -812,7 +814,7 @@ static int erofs_release_device_info(int id, void *ptr, void *data) { struct erofs_device_info *dif = ptr; - fs_put_dax(dif->dax_dev); + fs_put_dax(dif->dax_dev, NULL); if (dif->bdev) blkdev_put(dif->bdev, FMODE_READ | FMODE_EXCL); erofs_fscache_unregister_cookie(&dif->fscache); @@ -886,7 +888,7 @@ static void erofs_kill_sb(struct super_block *sb) return; erofs_free_dev_context(sbi->devs); - fs_put_dax(sbi->dax_dev); + fs_put_dax(sbi->dax_dev, NULL); erofs_fscache_unregister_cookie(&sbi->s_fscache); erofs_fscache_unregister_fs(sb); kfree(sbi->opt.fsid); diff --git a/fs/ext2/super.c b/fs/ext2/super.c index f6a19f6d9f6d..4638946251b9 100644 --- a/fs/ext2/super.c +++ b/fs/ext2/super.c @@ -171,7 +171,7 @@ static void ext2_put_super (struct super_block * sb) brelse (sbi->s_sbh); sb->s_fs_info = NULL; kfree(sbi->s_blockgroup_lock); - fs_put_dax(sbi->s_daxdev); + fs_put_dax(sbi->s_daxdev, NULL); kfree(sbi); } @@ -835,7 +835,8 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent) } sb->s_fs_info = sbi; sbi->s_sb_block = sb_block; - sbi->s_daxdev = fs_dax_get_by_bdev(sb->s_bdev, &sbi->s_dax_part_off); + sbi->s_daxdev = fs_dax_get_by_bdev(sb->s_bdev, &sbi->s_dax_part_off, + NULL, NULL); spin_lock_init(&sbi->s_lock); ret = -EINVAL; @@ -1204,7 +1205,7 @@ failed_mount_group_desc: failed_mount: brelse(bh); failed_sbi: - fs_put_dax(sbi->s_daxdev); + fs_put_dax(sbi->s_daxdev, NULL); sb->s_fs_info = NULL; kfree(sbi->s_blockgroup_lock); kfree(sbi); diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 845f2f8aee5f..1f8bf507ba5a 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1307,7 +1307,7 @@ static void ext4_put_super(struct super_block *sb) if (sbi->s_chksum_driver) crypto_free_shash(sbi->s_chksum_driver); kfree(sbi->s_blockgroup_lock); - fs_put_dax(sbi->s_daxdev); + fs_put_dax(sbi->s_daxdev, NULL); fscrypt_free_dummy_policy(&sbi->s_dummy_enc_policy); #if IS_ENABLED(CONFIG_UNICODE) utf8_unload(sb->s_encoding); @@ -4272,7 +4272,7 @@ static void ext4_free_sbi(struct ext4_sb_info *sbi) return; kfree(sbi->s_blockgroup_lock); - fs_put_dax(sbi->s_daxdev); + fs_put_dax(sbi->s_daxdev, NULL); kfree(sbi); } @@ -4284,7 +4284,8 @@ static struct ext4_sb_info *ext4_alloc_sbi(struct super_block *sb) if (!sbi) return NULL; - sbi->s_daxdev = fs_dax_get_by_bdev(sb->s_bdev, &sbi->s_dax_part_off); + sbi->s_daxdev = fs_dax_get_by_bdev(sb->s_bdev, &sbi->s_dax_part_off, + NULL, NULL); sbi->s_blockgroup_lock = kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL); @@ -4296,7 +4297,7 @@ static struct ext4_sb_info *ext4_alloc_sbi(struct super_block *sb) sbi->s_sb = sb; return sbi; err_out: - fs_put_dax(sbi->s_daxdev); + fs_put_dax(sbi->s_daxdev, NULL); kfree(sbi); return NULL; } diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index 4aa9c9cf5b6e..1ec2a7b6d44e 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -1911,7 +1911,7 @@ xfs_free_buftarg( list_lru_destroy(&btp->bt_lru); blkdev_issue_flush(btp->bt_bdev); - fs_put_dax(btp->bt_daxdev); + fs_put_dax(btp->bt_daxdev, NULL); kmem_free(btp); } @@ -1964,7 +1964,8 @@ xfs_alloc_buftarg( btp->bt_mount = mp; btp->bt_dev = bdev->bd_dev; btp->bt_bdev = bdev; - btp->bt_daxdev = fs_dax_get_by_bdev(bdev, &btp->bt_dax_part_off); + btp->bt_daxdev = fs_dax_get_by_bdev(bdev, &btp->bt_dax_part_off, NULL, + NULL); /* * Buffer IO error rate limiting. Limit it to no more than 10 messages diff --git a/include/linux/dax.h b/include/linux/dax.h index e7b81634c52a..cf85fc36da5f 100644 --- a/include/linux/dax.h +++ b/include/linux/dax.h @@ -43,8 +43,21 @@ struct dax_operations { void *addr, size_t bytes, struct iov_iter *iter); }; +struct dax_holder_operations { + /* + * notify_failure - notify memory failure into inner holder device + * @dax_dev: the dax device which contains the holder + * @offset: offset on this dax device where memory failure occurs + * @len: length of this memory failure event + * @flags: action flags for memory failure handler + */ + int (*notify_failure)(struct dax_device *dax_dev, u64 offset, + u64 len, int mf_flags); +}; + #if IS_ENABLED(CONFIG_DAX) struct dax_device *alloc_dax(void *private, const struct dax_operations *ops); +void *dax_holder(struct dax_device *dax_dev); void put_dax(struct dax_device *dax_dev); void kill_dax(struct dax_device *dax_dev); void dax_write_cache(struct dax_device *dax_dev, bool wc); @@ -66,6 +79,10 @@ static inline bool daxdev_mapping_supported(struct vm_area_struct *vma, return dax_synchronous(dax_dev); } #else +static inline void *dax_holder(struct dax_device *dax_dev) +{ + return NULL; +} static inline struct dax_device *alloc_dax(void *private, const struct dax_operations *ops) { @@ -114,12 +131,9 @@ struct writeback_control; #if defined(CONFIG_BLOCK) && defined(CONFIG_FS_DAX) int dax_add_host(struct dax_device *dax_dev, struct gendisk *disk); void dax_remove_host(struct gendisk *disk); -struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, - u64 *start_off); -static inline void fs_put_dax(struct dax_device *dax_dev) -{ - put_dax(dax_dev); -} +struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, u64 *start_off, + void *holder, const struct dax_holder_operations *ops); +void fs_put_dax(struct dax_device *dax_dev, void *holder); #else static inline int dax_add_host(struct dax_device *dax_dev, struct gendisk *disk) { @@ -129,11 +143,12 @@ static inline void dax_remove_host(struct gendisk *disk) { } static inline struct dax_device *fs_dax_get_by_bdev(struct block_device *bdev, - u64 *start_off) + u64 *start_off, void *holder, + const struct dax_holder_operations *ops) { return NULL; } -static inline void fs_put_dax(struct dax_device *dax_dev) +static inline void fs_put_dax(struct dax_device *dax_dev, void *holder) { } #endif /* CONFIG_BLOCK && CONFIG_FS_DAX */ @@ -203,6 +218,8 @@ size_t dax_copy_to_iter(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, size_t bytes, struct iov_iter *i); int dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff, size_t nr_pages); +int dax_holder_notify_failure(struct dax_device *dax_dev, u64 off, u64 len, + int mf_flags); void dax_flush(struct dax_device *dax_dev, void *addr, size_t size); ssize_t dax_iomap_rw(struct kiocb *iocb, struct iov_iter *iter, -- cgit v1.2.3 From fa78f336937240d1bc598db817d638086060e7e9 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 26 Jul 2022 13:13:50 +0200 Subject: ext2: Add more validity checks for inode counts Add checks verifying number of inodes stored in the superblock matches the number computed from number of inodes per group. Also verify we have at least one block worth of inodes per group. This prevents crashes on corrupted filesystems. Reported-by: syzbot+d273f7d7f58afd93be48@syzkaller.appspotmail.com Signed-off-by: Jan Kara --- fs/ext2/super.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/super.c b/fs/ext2/super.c index 300f2f0cf566..6f475d2e3b18 100644 --- a/fs/ext2/super.c +++ b/fs/ext2/super.c @@ -1059,9 +1059,10 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent) sbi->s_frags_per_group); goto failed_mount; } - if (sbi->s_inodes_per_group > sb->s_blocksize * 8) { + if (sbi->s_inodes_per_group < sbi->s_inodes_per_block || + sbi->s_inodes_per_group > sb->s_blocksize * 8) { ext2_msg(sb, KERN_ERR, - "error: #inodes per group too big: %lu", + "error: invalid #inodes per group: %lu", sbi->s_inodes_per_group); goto failed_mount; } @@ -1071,6 +1072,13 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent) sbi->s_groups_count = ((le32_to_cpu(es->s_blocks_count) - le32_to_cpu(es->s_first_data_block) - 1) / EXT2_BLOCKS_PER_GROUP(sb)) + 1; + if ((u64)sbi->s_groups_count * sbi->s_inodes_per_group != + le32_to_cpu(es->s_inodes_count)) { + ext2_msg(sb, KERN_ERR, "error: invalid #inodes: %u vs computed %llu", + le32_to_cpu(es->s_inodes_count), + (u64)sbi->s_groups_count * sbi->s_inodes_per_group); + goto failed_mount; + } db_count = (sbi->s_groups_count + EXT2_DESC_PER_BLOCK(sb) - 1) / EXT2_DESC_PER_BLOCK(sb); sbi->s_group_desc = kmalloc_array(db_count, -- cgit v1.2.3 From 37ce0b319b287666e2133ca740b73089355fe498 Mon Sep 17 00:00:00 2001 From: "Matthew Wilcox (Oracle)" Date: Tue, 17 May 2022 18:06:23 -0400 Subject: ext2: Use a folio in ext2_get_page() Remove a call to read_mapping_page(). Signed-off-by: Matthew Wilcox (Oracle) --- fs/ext2/dir.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/dir.c b/fs/ext2/dir.c index 8326b63f0b70..8f597753ac12 100644 --- a/fs/ext2/dir.c +++ b/fs/ext2/dir.c @@ -200,18 +200,19 @@ static struct page * ext2_get_page(struct inode *dir, unsigned long n, int quiet, void **page_addr) { struct address_space *mapping = dir->i_mapping; - struct page *page = read_mapping_page(mapping, n, NULL); - if (!IS_ERR(page)) { - *page_addr = kmap_local_page(page); - if (unlikely(!PageChecked(page))) { - if (!ext2_check_page(page, quiet, *page_addr)) - goto fail; - } + struct folio *folio = read_mapping_folio(mapping, n, NULL); + + if (IS_ERR(folio)) + return &folio->page; + *page_addr = kmap_local_folio(folio, n & (folio_nr_pages(folio) - 1)); + if (unlikely(!folio_test_checked(folio))) { + if (!ext2_check_page(&folio->page, quiet, *page_addr)) + goto fail; } - return page; + return &folio->page; fail: - ext2_put_page(page, *page_addr); + ext2_put_page(&folio->page, *page_addr); return ERR_PTR(-EIO); } -- cgit v1.2.3 From 67235182a41c1bd6b32806a1556a1d299b84212b Mon Sep 17 00:00:00 2001 From: "Matthew Wilcox (Oracle)" Date: Mon, 6 Jun 2022 10:20:31 -0400 Subject: mm/migrate: Convert buffer_migrate_page() to buffer_migrate_folio() Use a folio throughout __buffer_migrate_folio(), add kernel-doc for buffer_migrate_folio() and buffer_migrate_folio_norefs(), move their declarations to buffer.h and switch all filesystems that have wired them up. Signed-off-by: Matthew Wilcox (Oracle) Reviewed-by: Christoph Hellwig --- block/fops.c | 2 +- fs/ext2/inode.c | 4 +-- fs/ext4/inode.c | 4 +-- fs/ntfs/aops.c | 6 ++-- fs/ocfs2/aops.c | 2 +- include/linux/buffer_head.h | 10 ++++++ include/linux/fs.h | 12 ------- mm/migrate.c | 76 +++++++++++++++++++++++++++------------------ 8 files changed, 65 insertions(+), 51 deletions(-) (limited to 'fs/ext2') diff --git a/block/fops.c b/block/fops.c index d6b3276a6c68..743fc46d0aad 100644 --- a/block/fops.c +++ b/block/fops.c @@ -417,7 +417,7 @@ const struct address_space_operations def_blk_aops = { .write_end = blkdev_write_end, .writepages = blkdev_writepages, .direct_IO = blkdev_direct_IO, - .migratepage = buffer_migrate_page_norefs, + .migrate_folio = buffer_migrate_folio_norefs, .is_dirty_writeback = buffer_check_dirty_writeback, }; diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index e6b932219803..58a9d061f17d 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -973,7 +973,7 @@ const struct address_space_operations ext2_aops = { .bmap = ext2_bmap, .direct_IO = ext2_direct_IO, .writepages = ext2_writepages, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; @@ -989,7 +989,7 @@ const struct address_space_operations ext2_nobh_aops = { .bmap = ext2_bmap, .direct_IO = ext2_direct_IO, .writepages = ext2_writepages, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .error_remove_page = generic_error_remove_page, }; diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 06cc68878176..87a8b4382bce 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3633,7 +3633,7 @@ static const struct address_space_operations ext4_aops = { .invalidate_folio = ext4_invalidate_folio, .release_folio = ext4_release_folio, .direct_IO = noop_direct_IO, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, .swap_activate = ext4_iomap_swap_activate, @@ -3668,7 +3668,7 @@ static const struct address_space_operations ext4_da_aops = { .invalidate_folio = ext4_invalidate_folio, .release_folio = ext4_release_folio, .direct_IO = noop_direct_IO, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, .swap_activate = ext4_iomap_swap_activate, diff --git a/fs/ntfs/aops.c b/fs/ntfs/aops.c index 9e3964ea2ea0..5f4fb6ca6f2e 100644 --- a/fs/ntfs/aops.c +++ b/fs/ntfs/aops.c @@ -1659,7 +1659,7 @@ const struct address_space_operations ntfs_normal_aops = { .dirty_folio = block_dirty_folio, #endif /* NTFS_RW */ .bmap = ntfs_bmap, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; @@ -1673,7 +1673,7 @@ const struct address_space_operations ntfs_compressed_aops = { .writepage = ntfs_writepage, .dirty_folio = block_dirty_folio, #endif /* NTFS_RW */ - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; @@ -1688,7 +1688,7 @@ const struct address_space_operations ntfs_mst_aops = { .writepage = ntfs_writepage, /* Write dirty page to disk. */ .dirty_folio = filemap_dirty_folio, #endif /* NTFS_RW */ - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index 767df51f8657..1d489003f99d 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -2462,7 +2462,7 @@ const struct address_space_operations ocfs2_aops = { .direct_IO = ocfs2_direct_IO, .invalidate_folio = block_invalidate_folio, .release_folio = ocfs2_release_folio, - .migratepage = buffer_migrate_page, + .migrate_folio = buffer_migrate_folio, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index c9d1463bb20f..b0366c89d6a4 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -267,6 +267,16 @@ int nobh_truncate_page(struct address_space *, loff_t, get_block_t *); int nobh_writepage(struct page *page, get_block_t *get_block, struct writeback_control *wbc); +#ifdef CONFIG_MIGRATION +extern int buffer_migrate_folio(struct address_space *, + struct folio *dst, struct folio *src, enum migrate_mode); +extern int buffer_migrate_folio_norefs(struct address_space *, + struct folio *dst, struct folio *src, enum migrate_mode); +#else +#define buffer_migrate_folio NULL +#define buffer_migrate_folio_norefs NULL +#endif + void buffer_init(void); /* diff --git a/include/linux/fs.h b/include/linux/fs.h index 47431cf8fbb3..9e6b17da4e11 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -3215,18 +3215,6 @@ extern int generic_check_addressable(unsigned, u64); extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry); -#ifdef CONFIG_MIGRATION -extern int buffer_migrate_page(struct address_space *, - struct page *, struct page *, - enum migrate_mode); -extern int buffer_migrate_page_norefs(struct address_space *, - struct page *, struct page *, - enum migrate_mode); -#else -#define buffer_migrate_page NULL -#define buffer_migrate_page_norefs NULL -#endif - int may_setattr(struct user_namespace *mnt_userns, struct inode *inode, unsigned int ia_valid); int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *); diff --git a/mm/migrate.c b/mm/migrate.c index 75b171425c45..ea5398d0f7f1 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -656,23 +656,23 @@ static bool buffer_migrate_lock_buffers(struct buffer_head *head, return true; } -static int __buffer_migrate_page(struct address_space *mapping, - struct page *newpage, struct page *page, enum migrate_mode mode, +static int __buffer_migrate_folio(struct address_space *mapping, + struct folio *dst, struct folio *src, enum migrate_mode mode, bool check_refs) { struct buffer_head *bh, *head; int rc; int expected_count; - if (!page_has_buffers(page)) - return migrate_page(mapping, newpage, page, mode); + head = folio_buffers(src); + if (!head) + return migrate_page(mapping, &dst->page, &src->page, mode); /* Check whether page does not have extra refs before we do more work */ - expected_count = expected_page_refs(mapping, page); - if (page_count(page) != expected_count) + expected_count = expected_page_refs(mapping, &src->page); + if (folio_ref_count(src) != expected_count) return -EAGAIN; - head = page_buffers(page); if (!buffer_migrate_lock_buffers(head, mode)) return -EAGAIN; @@ -703,23 +703,22 @@ recheck_buffers: } } - rc = migrate_page_move_mapping(mapping, newpage, page, 0); + rc = folio_migrate_mapping(mapping, dst, src, 0); if (rc != MIGRATEPAGE_SUCCESS) goto unlock_buffers; - attach_page_private(newpage, detach_page_private(page)); + folio_attach_private(dst, folio_detach_private(src)); bh = head; do { - set_bh_page(bh, newpage, bh_offset(bh)); + set_bh_page(bh, &dst->page, bh_offset(bh)); bh = bh->b_this_page; - } while (bh != head); if (mode != MIGRATE_SYNC_NO_COPY) - migrate_page_copy(newpage, page); + folio_migrate_copy(dst, src); else - migrate_page_states(newpage, page); + folio_migrate_flags(dst, src); rc = MIGRATEPAGE_SUCCESS; unlock_buffers: @@ -729,34 +728,51 @@ unlock_buffers: do { unlock_buffer(bh); bh = bh->b_this_page; - } while (bh != head); return rc; } -/* - * Migration function for pages with buffers. This function can only be used - * if the underlying filesystem guarantees that no other references to "page" - * exist. For example attached buffer heads are accessed only under page lock. +/** + * buffer_migrate_folio() - Migration function for folios with buffers. + * @mapping: The address space containing @src. + * @dst: The folio to migrate to. + * @src: The folio to migrate from. + * @mode: How to migrate the folio. + * + * This function can only be used if the underlying filesystem guarantees + * that no other references to @src exist. For example attached buffer + * heads are accessed only under the folio lock. If your filesystem cannot + * provide this guarantee, buffer_migrate_folio_norefs() may be more + * appropriate. + * + * Return: 0 on success or a negative errno on failure. */ -int buffer_migrate_page(struct address_space *mapping, - struct page *newpage, struct page *page, enum migrate_mode mode) +int buffer_migrate_folio(struct address_space *mapping, + struct folio *dst, struct folio *src, enum migrate_mode mode) { - return __buffer_migrate_page(mapping, newpage, page, mode, false); + return __buffer_migrate_folio(mapping, dst, src, mode, false); } -EXPORT_SYMBOL(buffer_migrate_page); +EXPORT_SYMBOL(buffer_migrate_folio); -/* - * Same as above except that this variant is more careful and checks that there - * are also no buffer head references. This function is the right one for - * mappings where buffer heads are directly looked up and referenced (such as - * block device mappings). +/** + * buffer_migrate_folio_norefs() - Migration function for folios with buffers. + * @mapping: The address space containing @src. + * @dst: The folio to migrate to. + * @src: The folio to migrate from. + * @mode: How to migrate the folio. + * + * Like buffer_migrate_folio() except that this variant is more careful + * and checks that there are also no buffer head references. This function + * is the right one for mappings where buffer heads are directly looked + * up and referenced (such as block device mappings). + * + * Return: 0 on success or a negative errno on failure. */ -int buffer_migrate_page_norefs(struct address_space *mapping, - struct page *newpage, struct page *page, enum migrate_mode mode) +int buffer_migrate_folio_norefs(struct address_space *mapping, + struct folio *dst, struct folio *src, enum migrate_mode mode) { - return __buffer_migrate_page(mapping, newpage, page, mode, true); + return __buffer_migrate_folio(mapping, dst, src, mode, true); } #endif -- cgit v1.2.3 From 0cc5b4ce7a3735ba0c64ed4c5a1f673165c4d5b2 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 13 Jun 2022 07:37:11 +0200 Subject: ext2: remove nobh support The nobh mode is an obscure feature to save lowlevel for large memory 32-bit configurations while trading for much slower performance and has been long obsolete. Remove it. Signed-off-by: Christoph Hellwig Reviewed-by: Jan Kara Signed-off-by: Matthew Wilcox (Oracle) --- Documentation/filesystems/ext2.rst | 2 -- fs/ext2/ext2.h | 1 - fs/ext2/inode.c | 51 +++----------------------------------- fs/ext2/namei.c | 10 ++------ fs/ext2/super.c | 6 ++--- 5 files changed, 7 insertions(+), 63 deletions(-) (limited to 'fs/ext2') diff --git a/Documentation/filesystems/ext2.rst b/Documentation/filesystems/ext2.rst index 154101cf0e4f..92aae683e16a 100644 --- a/Documentation/filesystems/ext2.rst +++ b/Documentation/filesystems/ext2.rst @@ -59,8 +59,6 @@ acl Enable POSIX Access Control Lists support (requires CONFIG_EXT2_FS_POSIX_ACL). noacl Don't support POSIX ACLs. -nobh Do not attach buffer_heads to file pagecache. - quota, usrquota Enable user disk quota support (requires CONFIG_QUOTA). diff --git a/fs/ext2/ext2.h b/fs/ext2/ext2.h index d4f306aa5ace..28de11a22e5f 100644 --- a/fs/ext2/ext2.h +++ b/fs/ext2/ext2.h @@ -795,7 +795,6 @@ extern const struct file_operations ext2_file_operations; /* inode.c */ extern void ext2_set_file_ops(struct inode *inode); extern const struct address_space_operations ext2_aops; -extern const struct address_space_operations ext2_nobh_aops; extern const struct iomap_ops ext2_iomap_ops; /* namei.c */ diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index 58a9d061f17d..c5229033baf0 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -908,25 +908,6 @@ static int ext2_write_end(struct file *file, struct address_space *mapping, return ret; } -static int -ext2_nobh_write_begin(struct file *file, struct address_space *mapping, - loff_t pos, unsigned len, struct page **pagep, void **fsdata) -{ - int ret; - - ret = nobh_write_begin(mapping, pos, len, pagep, fsdata, - ext2_get_block); - if (ret < 0) - ext2_write_failed(mapping, pos + len); - return ret; -} - -static int ext2_nobh_writepage(struct page *page, - struct writeback_control *wbc) -{ - return nobh_writepage(page, ext2_get_block, wbc); -} - static sector_t ext2_bmap(struct address_space *mapping, sector_t block) { return generic_block_bmap(mapping,block,ext2_get_block); @@ -978,21 +959,6 @@ const struct address_space_operations ext2_aops = { .error_remove_page = generic_error_remove_page, }; -const struct address_space_operations ext2_nobh_aops = { - .dirty_folio = block_dirty_folio, - .invalidate_folio = block_invalidate_folio, - .read_folio = ext2_read_folio, - .readahead = ext2_readahead, - .writepage = ext2_nobh_writepage, - .write_begin = ext2_nobh_write_begin, - .write_end = nobh_write_end, - .bmap = ext2_bmap, - .direct_IO = ext2_direct_IO, - .writepages = ext2_writepages, - .migrate_folio = buffer_migrate_folio, - .error_remove_page = generic_error_remove_page, -}; - static const struct address_space_operations ext2_dax_aops = { .writepages = ext2_dax_writepages, .direct_IO = noop_direct_IO, @@ -1298,13 +1264,10 @@ static int ext2_setsize(struct inode *inode, loff_t newsize) inode_dio_wait(inode); - if (IS_DAX(inode)) { + if (IS_DAX(inode)) error = dax_zero_range(inode, newsize, PAGE_ALIGN(newsize) - newsize, NULL, &ext2_iomap_ops); - } else if (test_opt(inode->i_sb, NOBH)) - error = nobh_truncate_page(inode->i_mapping, - newsize, ext2_get_block); else error = block_truncate_page(inode->i_mapping, newsize, ext2_get_block); @@ -1396,8 +1359,6 @@ void ext2_set_file_ops(struct inode *inode) inode->i_fop = &ext2_file_operations; if (IS_DAX(inode)) inode->i_mapping->a_ops = &ext2_dax_aops; - else if (test_opt(inode->i_sb, NOBH)) - inode->i_mapping->a_ops = &ext2_nobh_aops; else inode->i_mapping->a_ops = &ext2_aops; } @@ -1497,10 +1458,7 @@ struct inode *ext2_iget (struct super_block *sb, unsigned long ino) } else if (S_ISDIR(inode->i_mode)) { inode->i_op = &ext2_dir_inode_operations; inode->i_fop = &ext2_dir_operations; - if (test_opt(inode->i_sb, NOBH)) - inode->i_mapping->a_ops = &ext2_nobh_aops; - else - inode->i_mapping->a_ops = &ext2_aops; + inode->i_mapping->a_ops = &ext2_aops; } else if (S_ISLNK(inode->i_mode)) { if (ext2_inode_is_fast_symlink(inode)) { inode->i_link = (char *)ei->i_data; @@ -1510,10 +1468,7 @@ struct inode *ext2_iget (struct super_block *sb, unsigned long ino) } else { inode->i_op = &ext2_symlink_inode_operations; inode_nohighmem(inode); - if (test_opt(inode->i_sb, NOBH)) - inode->i_mapping->a_ops = &ext2_nobh_aops; - else - inode->i_mapping->a_ops = &ext2_aops; + inode->i_mapping->a_ops = &ext2_aops; } } else { inode->i_op = &ext2_special_inode_operations; diff --git a/fs/ext2/namei.c b/fs/ext2/namei.c index 5f6b7560eb3f..5fd9a22d2b70 100644 --- a/fs/ext2/namei.c +++ b/fs/ext2/namei.c @@ -178,10 +178,7 @@ static int ext2_symlink (struct user_namespace * mnt_userns, struct inode * dir, /* slow symlink */ inode->i_op = &ext2_symlink_inode_operations; inode_nohighmem(inode); - if (test_opt(inode->i_sb, NOBH)) - inode->i_mapping->a_ops = &ext2_nobh_aops; - else - inode->i_mapping->a_ops = &ext2_aops; + inode->i_mapping->a_ops = &ext2_aops; err = page_symlink(inode, symname, l); if (err) goto out_fail; @@ -247,10 +244,7 @@ static int ext2_mkdir(struct user_namespace * mnt_userns, inode->i_op = &ext2_dir_inode_operations; inode->i_fop = &ext2_dir_operations; - if (test_opt(inode->i_sb, NOBH)) - inode->i_mapping->a_ops = &ext2_nobh_aops; - else - inode->i_mapping->a_ops = &ext2_aops; + inode->i_mapping->a_ops = &ext2_aops; inode_inc_link_count(inode); diff --git a/fs/ext2/super.c b/fs/ext2/super.c index f6a19f6d9f6d..a1c1263c07ab 100644 --- a/fs/ext2/super.c +++ b/fs/ext2/super.c @@ -296,9 +296,6 @@ static int ext2_show_options(struct seq_file *seq, struct dentry *root) seq_puts(seq, ",noacl"); #endif - if (test_opt(sb, NOBH)) - seq_puts(seq, ",nobh"); - if (test_opt(sb, USRQUOTA)) seq_puts(seq, ",usrquota"); @@ -551,7 +548,8 @@ static int parse_options(char *options, struct super_block *sb, clear_opt (opts->s_mount_opt, OLDALLOC); break; case Opt_nobh: - set_opt (opts->s_mount_opt, NOBH); + ext2_msg(sb, KERN_INFO, + "nobh option not supported"); break; #ifdef CONFIG_EXT2_FS_XATTR case Opt_user_xattr: -- cgit v1.2.3 From 90ae40d243d46f2268a5334fc8faa2dd67f875f5 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jul 2022 12:54:25 +0200 Subject: ext2: factor our freeing of xattr block reference Free of xattr block reference is opencode in two places. Factor it out into a separate function and use it. Signed-off-by: Jan Kara Link: https://lore.kernel.org/r/20220712105436.32204-6-jack@suse.cz Signed-off-by: Theodore Ts'o --- fs/ext2/xattr.c | 90 ++++++++++++++++++++++++--------------------------------- 1 file changed, 38 insertions(+), 52 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c index 841fa6d9d744..9885294993ef 100644 --- a/fs/ext2/xattr.c +++ b/fs/ext2/xattr.c @@ -651,6 +651,42 @@ cleanup: return error; } +static void ext2_xattr_release_block(struct inode *inode, + struct buffer_head *bh) +{ + struct mb_cache *ea_block_cache = EA_BLOCK_CACHE(inode); + + lock_buffer(bh); + if (HDR(bh)->h_refcount == cpu_to_le32(1)) { + __u32 hash = le32_to_cpu(HDR(bh)->h_hash); + + /* + * This must happen under buffer lock for + * ext2_xattr_set2() to reliably detect freed block + */ + mb_cache_entry_delete(ea_block_cache, hash, + bh->b_blocknr); + /* Free the old block. */ + ea_bdebug(bh, "freeing"); + ext2_free_blocks(inode, bh->b_blocknr, 1); + /* We let our caller release bh, so we + * need to duplicate the buffer before. */ + get_bh(bh); + bforget(bh); + unlock_buffer(bh); + } else { + /* Decrement the refcount only. */ + le32_add_cpu(&HDR(bh)->h_refcount, -1); + dquot_free_block(inode, 1); + mark_buffer_dirty(bh); + unlock_buffer(bh); + ea_bdebug(bh, "refcount now=%d", + le32_to_cpu(HDR(bh)->h_refcount)); + if (IS_SYNC(inode)) + sync_dirty_buffer(bh); + } +} + /* * Second half of ext2_xattr_set(): Update the file system. */ @@ -747,34 +783,7 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh, * If there was an old block and we are no longer using it, * release the old block. */ - lock_buffer(old_bh); - if (HDR(old_bh)->h_refcount == cpu_to_le32(1)) { - __u32 hash = le32_to_cpu(HDR(old_bh)->h_hash); - - /* - * This must happen under buffer lock for - * ext2_xattr_set2() to reliably detect freed block - */ - mb_cache_entry_delete(ea_block_cache, hash, - old_bh->b_blocknr); - /* Free the old block. */ - ea_bdebug(old_bh, "freeing"); - ext2_free_blocks(inode, old_bh->b_blocknr, 1); - mark_inode_dirty(inode); - /* We let our caller release old_bh, so we - * need to duplicate the buffer before. */ - get_bh(old_bh); - bforget(old_bh); - } else { - /* Decrement the refcount only. */ - le32_add_cpu(&HDR(old_bh)->h_refcount, -1); - dquot_free_block_nodirty(inode, 1); - mark_inode_dirty(inode); - mark_buffer_dirty(old_bh); - ea_bdebug(old_bh, "refcount now=%d", - le32_to_cpu(HDR(old_bh)->h_refcount)); - } - unlock_buffer(old_bh); + ext2_xattr_release_block(inode, old_bh); } cleanup: @@ -828,30 +837,7 @@ ext2_xattr_delete_inode(struct inode *inode) EXT2_I(inode)->i_file_acl); goto cleanup; } - lock_buffer(bh); - if (HDR(bh)->h_refcount == cpu_to_le32(1)) { - __u32 hash = le32_to_cpu(HDR(bh)->h_hash); - - /* - * This must happen under buffer lock for ext2_xattr_set2() to - * reliably detect freed block - */ - mb_cache_entry_delete(EA_BLOCK_CACHE(inode), hash, - bh->b_blocknr); - ext2_free_blocks(inode, EXT2_I(inode)->i_file_acl, 1); - get_bh(bh); - bforget(bh); - unlock_buffer(bh); - } else { - le32_add_cpu(&HDR(bh)->h_refcount, -1); - ea_bdebug(bh, "refcount now=%d", - le32_to_cpu(HDR(bh)->h_refcount)); - unlock_buffer(bh); - mark_buffer_dirty(bh); - if (IS_SYNC(inode)) - sync_dirty_buffer(bh); - dquot_free_block_nodirty(inode, 1); - } + ext2_xattr_release_block(inode, bh); EXT2_I(inode)->i_file_acl = 0; cleanup: -- cgit v1.2.3 From b67798d55185ba4471f70b67b3f7c2866f216499 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jul 2022 12:54:26 +0200 Subject: ext2: unindent codeblock in ext2_xattr_set() Replace one else in ext2_xattr_set() with a goto. This makes following code changes simpler to follow. No functional changes. Signed-off-by: Jan Kara Reviewed-by: Ritesh Harjani (IBM) Link: https://lore.kernel.org/r/20220712105436.32204-7-jack@suse.cz Signed-off-by: Theodore Ts'o --- fs/ext2/xattr.c | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c index 9885294993ef..37ce495eb279 100644 --- a/fs/ext2/xattr.c +++ b/fs/ext2/xattr.c @@ -517,7 +517,8 @@ bad_block: /* Here we know that we can set the new attribute. */ if (header) { - /* assert(header == HDR(bh)); */ + int offset; + lock_buffer(bh); if (header->h_refcount == cpu_to_le32(1)) { __u32 hash = le32_to_cpu(header->h_hash); @@ -531,22 +532,20 @@ bad_block: bh->b_blocknr); /* keep the buffer locked while modifying it. */ - } else { - int offset; - - unlock_buffer(bh); - ea_bdebug(bh, "cloning"); - header = kmemdup(HDR(bh), bh->b_size, GFP_KERNEL); - error = -ENOMEM; - if (header == NULL) - goto cleanup; - header->h_refcount = cpu_to_le32(1); - - offset = (char *)here - bh->b_data; - here = ENTRY((char *)header + offset); - offset = (char *)last - bh->b_data; - last = ENTRY((char *)header + offset); + goto update_block; } + unlock_buffer(bh); + ea_bdebug(bh, "cloning"); + header = kmemdup(HDR(bh), bh->b_size, GFP_KERNEL); + error = -ENOMEM; + if (header == NULL) + goto cleanup; + header->h_refcount = cpu_to_le32(1); + + offset = (char *)here - bh->b_data; + here = ENTRY((char *)header + offset); + offset = (char *)last - bh->b_data; + last = ENTRY((char *)header + offset); } else { /* Allocate a buffer where we construct the new block. */ header = kzalloc(sb->s_blocksize, GFP_KERNEL); @@ -559,6 +558,7 @@ bad_block: last = here = ENTRY(header+1); } +update_block: /* Iff we are modifying the block in-place, bh is locked here. */ if (not_found) { -- cgit v1.2.3 From 1189d8ec5105f7628fb0e3a6988cbf0d0cbfedeb Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 12 Jul 2022 12:54:27 +0200 Subject: ext2: avoid deleting xattr block that is being reused Currently when we decide to reuse xattr block we detect the case when the last reference to xattr block is being dropped at the same time and cancel the reuse attempt. Convert ext2 to a new scheme when as soon as matching mbcache entry is found, we wait with dropping the last xattr block reference until mbcache entry reference is dropped (meaning either the xattr block reference is increased or we decided not to reuse the block). Signed-off-by: Jan Kara Link: https://lore.kernel.org/r/20220712105436.32204-8-jack@suse.cz Signed-off-by: Theodore Ts'o --- fs/ext2/xattr.c | 58 ++++++++++++++++++++++++++++----------------------------- 1 file changed, 29 insertions(+), 29 deletions(-) (limited to 'fs/ext2') diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c index 37ce495eb279..641abfa4b718 100644 --- a/fs/ext2/xattr.c +++ b/fs/ext2/xattr.c @@ -522,17 +522,18 @@ bad_block: lock_buffer(bh); if (header->h_refcount == cpu_to_le32(1)) { __u32 hash = le32_to_cpu(header->h_hash); + struct mb_cache_entry *oe; - ea_bdebug(bh, "modifying in-place"); + oe = mb_cache_entry_delete_or_get(EA_BLOCK_CACHE(inode), + hash, bh->b_blocknr); + if (!oe) { + ea_bdebug(bh, "modifying in-place"); + goto update_block; + } /* - * This must happen under buffer lock for - * ext2_xattr_set2() to reliably detect modified block + * Someone is trying to reuse the block, leave it alone */ - mb_cache_entry_delete(EA_BLOCK_CACHE(inode), hash, - bh->b_blocknr); - - /* keep the buffer locked while modifying it. */ - goto update_block; + mb_cache_entry_put(EA_BLOCK_CACHE(inode), oe); } unlock_buffer(bh); ea_bdebug(bh, "cloning"); @@ -656,16 +657,29 @@ static void ext2_xattr_release_block(struct inode *inode, { struct mb_cache *ea_block_cache = EA_BLOCK_CACHE(inode); +retry_ref: lock_buffer(bh); if (HDR(bh)->h_refcount == cpu_to_le32(1)) { __u32 hash = le32_to_cpu(HDR(bh)->h_hash); + struct mb_cache_entry *oe; /* - * This must happen under buffer lock for - * ext2_xattr_set2() to reliably detect freed block + * This must happen under buffer lock to properly + * serialize with ext2_xattr_set() reusing the block. */ - mb_cache_entry_delete(ea_block_cache, hash, - bh->b_blocknr); + oe = mb_cache_entry_delete_or_get(ea_block_cache, hash, + bh->b_blocknr); + if (oe) { + /* + * Someone is trying to reuse the block. Wait + * and retry. + */ + unlock_buffer(bh); + mb_cache_entry_wait_unused(oe); + mb_cache_entry_put(ea_block_cache, oe); + goto retry_ref; + } + /* Free the old block. */ ea_bdebug(bh, "freeing"); ext2_free_blocks(inode, bh->b_blocknr, 1); @@ -929,7 +943,7 @@ ext2_xattr_cache_find(struct inode *inode, struct ext2_xattr_header *header) if (!header->h_hash) return NULL; /* never share */ ea_idebug(inode, "looking for cached blocks [%x]", (int)hash); -again: + ce = mb_cache_entry_find_first(ea_block_cache, hash); while (ce) { struct buffer_head *bh; @@ -941,22 +955,8 @@ again: inode->i_ino, (unsigned long) ce->e_value); } else { lock_buffer(bh); - /* - * We have to be careful about races with freeing or - * rehashing of xattr block. Once we hold buffer lock - * xattr block's state is stable so we can check - * whether the block got freed / rehashed or not. - * Since we unhash mbcache entry under buffer lock when - * freeing / rehashing xattr block, checking whether - * entry is still hashed is reliable. - */ - if (hlist_bl_unhashed(&ce->e_hash_list)) { - mb_cache_entry_put(ea_block_cache, ce); - unlock_buffer(bh); - brelse(bh); - goto again; - } else if (le32_to_cpu(HDR(bh)->h_refcount) > - EXT2_XATTR_REFCOUNT_MAX) { + if (le32_to_cpu(HDR(bh)->h_refcount) > + EXT2_XATTR_REFCOUNT_MAX) { ea_idebug(inode, "block %ld refcount %d>%d", (unsigned long) ce->e_value, le32_to_cpu(HDR(bh)->h_refcount), -- cgit v1.2.3