| Message ID | 20240322041639.23144-1-youngjin.gil@samsung.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 3127f1010c9b27d925e83081d413ea7fc361abb0 |
| Headers | show |
| Series | [f2fs-dev] f2fs: Prevent s_writer rw_sem count mismatch in f2fs_evict_inode | expand |
On 2024/3/22 12:16, Yeongjin Gil wrote: > If f2fs_evict_inode is called between freeze_super and thaw_super, the > s_writer rwsem count may become negative, resulting in hang. > > CPU1 CPU2 > > f2fs_resize_fs() f2fs_evict_inode() > f2fs_freeze > set SBI_IS_FREEZING > skip sb_start_intwrite > f2fs_unfreeze > clear SBI_IS_FREEZING > sb_end_intwrite > > To solve this problem, the call to sb_end_write is determined by whether > sb_start_intwrite is called, rather than the current freezing status. > > Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com> > Reviewed-by: Sunmin Jeong <s_min.jeong@samsung.com> > Signed-off-by: Yeongjin Gil <youngjin.gil@samsung.com> Reviewed-by: Chao Yu <chao@kernel.org> Thanks,
Hello: This patch was applied to jaegeuk/f2fs.git (dev) by Jaegeuk Kim <jaegeuk@kernel.org>: On Fri, 22 Mar 2024 13:16:39 +0900 you wrote: > If f2fs_evict_inode is called between freeze_super and thaw_super, the > s_writer rwsem count may become negative, resulting in hang. > > CPU1 CPU2 > > f2fs_resize_fs() f2fs_evict_inode() > f2fs_freeze > set SBI_IS_FREEZING > skip sb_start_intwrite > f2fs_unfreeze > clear SBI_IS_FREEZING > sb_end_intwrite > > [...] Here is the summary with links: - [f2fs-dev] f2fs: Prevent s_writer rw_sem count mismatch in f2fs_evict_inode https://git.kernel.org/jaegeuk/f2fs/c/3127f1010c9b You are awesome, thank you!
diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index c26effdce9aa..12b1fef31f43 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -804,6 +804,7 @@ void f2fs_evict_inode(struct inode *inode) struct f2fs_inode_info *fi = F2FS_I(inode); nid_t xnid = fi->i_xattr_nid; int err = 0; + bool freeze_protected = false; f2fs_abort_atomic_write(inode, true); @@ -843,8 +844,10 @@ void f2fs_evict_inode(struct inode *inode) f2fs_remove_ino_entry(sbi, inode->i_ino, UPDATE_INO); f2fs_remove_ino_entry(sbi, inode->i_ino, FLUSH_INO); - if (!is_sbi_flag_set(sbi, SBI_IS_FREEZING)) + if (!is_sbi_flag_set(sbi, SBI_IS_FREEZING)) { sb_start_intwrite(inode->i_sb); + freeze_protected = true; + } set_inode_flag(inode, FI_NO_ALLOC); i_size_write(inode, 0); retry: @@ -887,7 +890,7 @@ void f2fs_evict_inode(struct inode *inode) if (dquot_initialize_needed(inode)) set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR); } - if (!is_sbi_flag_set(sbi, SBI_IS_FREEZING)) + if (freeze_protected) sb_end_intwrite(inode->i_sb); no_delete: dquot_drop(inode);
If f2fs_evict_inode is called between freeze_super and thaw_super, the s_writer rwsem count may become negative, resulting in hang. CPU1 CPU2 f2fs_resize_fs() f2fs_evict_inode() f2fs_freeze set SBI_IS_FREEZING skip sb_start_intwrite f2fs_unfreeze clear SBI_IS_FREEZING sb_end_intwrite To solve this problem, the call to sb_end_write is determined by whether sb_start_intwrite is called, rather than the current freezing status. Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Sunmin Jeong <s_min.jeong@samsung.com> Signed-off-by: Yeongjin Gil <youngjin.gil@samsung.com> --- fs/f2fs/inode.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)