From 0163acf4f3fa65299e2574d9cee12c41c16b6ee7 Mon Sep 17 00:00:00 2001 From: Shaun Tancheff Date: Thu, 8 Jun 2023 13:32:17 +0700 Subject: [PATCH] LU-16883 ldiskfs: update for ext4-delayed-iput for SUSE 15 ext4-delayed-iput patch does not apply cleanly to SUSE 15 SP4 and SP5 series 5.14.21 kernel. Adjust the minor conflict in ext4_put_super() Test-Parameters: trivial Fixes: 616fa9b581 ("LU-15404 ldiskfs: use per-filesystem workqueues to avoid deadlocks") HPE-bug-id: LUS-11661 Signed-off-by: Shaun Tancheff Change-Id: Iee424bd6d455853d9f82e6e5b08e4ab44deb432c Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/51252 Tested-by: jenkins Tested-by: Maloo Reviewed-by: James Simmons Reviewed-by: Neil Brown Reviewed-by: Oleg Drokin --- .../patches/sles15sp4/ext4-delayed-iput.patch | 178 +++++++++++++++++++++ .../series/ldiskfs-5.14.21-sles15sp4.series | 2 +- .../series/ldiskfs-5.14.21-sles15sp5.series | 2 +- 3 files changed, 180 insertions(+), 2 deletions(-) create mode 100644 ldiskfs/kernel_patches/patches/sles15sp4/ext4-delayed-iput.patch diff --git a/ldiskfs/kernel_patches/patches/sles15sp4/ext4-delayed-iput.patch b/ldiskfs/kernel_patches/patches/sles15sp4/ext4-delayed-iput.patch new file mode 100644 index 0000000..193b561 --- /dev/null +++ b/ldiskfs/kernel_patches/patches/sles15sp4/ext4-delayed-iput.patch @@ -0,0 +1,178 @@ +Subject: [PATCH] ext4: delayed iput + +When changing a large xattr value to a different large xattr +value, the old xattr inode is freed. Truncate during the final iput causes +current transaction restart. Eventually, parent inode bh is marked dirty and +kernel panic happens when jbd2 figures out that this bh belongs to the +committed transaction. + +A possible fix is to call this final iput in a separate thread. +This way, setxattr transactions will never be split into two. +Since the setxattr code adds xattr inodes with nlink=0 into the +orphan list, old xattr inodes will be properly cleaned up in +any case. + +Signed-off-by: Andrew Perepechko +HPE-bug-id: LUS-10534 +--- + fs/ext4/ext4.h | 7 +++++-- + fs/ext4/page-io.c | 2 +- + fs/ext4/super.c | 15 ++++++++------- + fs/ext4/xattr.c | 39 +++++++++++++++++++++++++++++++++++++-- + 4 files changed, 51 insertions(+), 12 deletions(-) + +diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h +index 885c9f2..ee27198 100644 +--- a/fs/ext4/ext4.h ++++ b/fs/ext4/ext4.h +@@ -1624,8 +1624,11 @@ struct ext4_sb_info { + struct flex_groups * __rcu *s_flex_groups; + ext4_group_t s_flex_groups_allocated; + +- /* workqueue for reserved extent conversions (buffered io) */ +- struct workqueue_struct *rsv_conversion_wq; ++ /* ++ * workqueue for reserved extent conversions (buffered io) ++ * and large ea inodes reclaim ++ */ ++ struct workqueue_struct *s_misc_wq; + + /* timer for periodic error stats printing */ + struct timer_list s_err_report; +diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c +index 18977ff..0d76faf 100644 +--- a/fs/ext4/page-io.c ++++ b/fs/ext4/page-io.c +@@ -230,7 +230,7 @@ static void ext4_add_complete_io(ext4_io_end_t *io_end) + WARN_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN)); + WARN_ON(!io_end->handle && sbi->s_journal); + spin_lock_irqsave(&ei->i_completed_io_lock, flags); +- wq = sbi->rsv_conversion_wq; ++ wq = sbi->s_misc_wq; + if (list_empty(&ei->i_rsv_conversion_list)) + queue_work(wq, &ei->i_rsv_conversion_work); + list_add_tail(&io_end->list, &ei->i_rsv_conversion_list); +diff --git a/fs/ext4/super.c b/fs/ext4/super.c +index 9b46ecd..cc8a566 100644 +--- a/fs/ext4/super.c ++++ b/fs/ext4/super.c +@@ -1171,10 +1171,11 @@ static void ext4_put_super(struct super_block *sb) + int i, err; + + ext4_unregister_li_request(sb); ++ flush_workqueue(sbi->s_misc_wq); + ext4_quota_off_umount(sb); + + flush_work(&sbi->s_error_work); +- destroy_workqueue(sbi->rsv_conversion_wq); ++ destroy_workqueue(sbi->s_misc_wq); + + /* + * Unregister sysfs before destroying jbd2 journal. +@@ -4994,9 +4995,9 @@ no_journal: + * The maximum number of concurrent works can be high and + * concurrency isn't really necessary. Limit it to 1. + */ +- EXT4_SB(sb)->rsv_conversion_wq = +- alloc_workqueue("ext4-rsv-conversion", WQ_MEM_RECLAIM | WQ_UNBOUND, 1); +- if (!EXT4_SB(sb)->rsv_conversion_wq) { ++ EXT4_SB(sb)->s_misc_wq = ++ alloc_workqueue("ext4-misc", WQ_MEM_RECLAIM | WQ_UNBOUND, 1); ++ if (!EXT4_SB(sb)->s_misc_wq) { + printk(KERN_ERR "EXT4-fs: failed to create workqueue\n"); + ret = -ENOMEM; + goto failed_mount4; +@@ -5227,8 +5228,8 @@ failed_mount4a: + sb->s_root = NULL; + failed_mount4: + ext4_msg(sb, KERN_ERR, "mount failed"); +- if (EXT4_SB(sb)->rsv_conversion_wq) +- destroy_workqueue(EXT4_SB(sb)->rsv_conversion_wq); ++ if (EXT4_SB(sb)->s_misc_wq) ++ destroy_workqueue(EXT4_SB(sb)->s_misc_wq); + failed_mount_wq: + ext4_xattr_destroy_cache(sbi->s_ea_inode_cache); + sbi->s_ea_inode_cache = NULL; +@@ -5789,7 +5790,7 @@ static int ext4_sync_fs(struct super_block *sb, int wait) + return 0; + + trace_ext4_sync_fs(sb, wait); +- flush_workqueue(sbi->rsv_conversion_wq); ++ flush_workqueue(sbi->s_misc_wq); + /* + * Writeback quota in non-journalled quota case - journalled quota has + * no dirty dquots +diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c +index 0bb9a00..561159e 100644 +--- a/fs/ext4/xattr.c ++++ b/fs/ext4/xattr.c +@@ -1601,6 +1601,36 @@ static int ext4_xattr_inode_lookup_create(handle_t *handle, struct inode *inode, + return 0; + } + ++struct delayed_iput_work { ++ struct work_struct work; ++ struct inode *inode; ++}; ++ ++static void delayed_iput_fn(struct work_struct *work) ++{ ++ struct delayed_iput_work *diwork; ++ ++ diwork = container_of(work, struct delayed_iput_work, work); ++ iput(diwork->inode); ++ kfree(diwork); ++} ++ ++static void delayed_iput(struct inode *inode, struct delayed_iput_work *work) ++{ ++ if (!inode) { ++ kfree(work); ++ return; ++ } ++ ++ if (!work) { ++ iput(inode); ++ } else { ++ INIT_WORK(&work->work, delayed_iput_fn); ++ work->inode = inode; ++ queue_work(EXT4_SB(inode->i_sb)->s_misc_wq, &work->work); ++ } ++} ++ + /* + * Reserve min(block_size/8, 1024) bytes for xattr entries/names if ea_inode + * feature is enabled. +@@ -1618,6 +1648,7 @@ static int ext4_xattr_set_entry(struct ext4_xattr_info *i, + int in_inode = i->in_inode; + struct inode *old_ea_inode = NULL; + struct inode *new_ea_inode = NULL; ++ struct delayed_iput_work *diwork = NULL; + size_t old_size, new_size; + int ret; + +@@ -1694,7 +1725,11 @@ static int ext4_xattr_set_entry(struct ext4_xattr_info *i, + * Finish that work before doing any modifications to the xattr data. + */ + if (!s->not_found && here->e_value_inum) { +- ret = ext4_xattr_inode_iget(inode, ++ diwork = kmalloc(sizeof(*diwork), GFP_NOFS); ++ if (!diwork) ++ ret = -ENOMEM; ++ else ++ ret = ext4_xattr_inode_iget(inode, + le32_to_cpu(here->e_value_inum), + le32_to_cpu(here->e_hash), + &old_ea_inode); +@@ -1847,7 +1882,7 @@ update_hash: + + ret = 0; + out: +- iput(old_ea_inode); ++ delayed_iput(old_ea_inode, diwork); + iput(new_ea_inode); + return ret; + } +-- +2.25.1 + diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series index e07353d..bdde820 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp4.series @@ -26,7 +26,7 @@ linux-5.14/export-ext4fs-dirhash-helper.patch linux-5.8/ext4-no-max-dir-size-limit-for-iam-objects.patch linux-5.14/ext4-ialloc-uid-gid-and-pass-owner-down.patch linux-5.14/ext4-projid-xattrs.patch -base/ext4-delayed-iput.patch +sles15sp4/ext4-delayed-iput.patch rhel8/ext4-ext-merge.patch linux-5.14/ext4-xattr-disable-credits-check.patch linux-5.10/ext4-fiemap-kernel-data.patch diff --git a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp5.series b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp5.series index efb741e..26837f4 100644 --- a/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp5.series +++ b/ldiskfs/kernel_patches/series/ldiskfs-5.14.21-sles15sp5.series @@ -26,7 +26,7 @@ linux-5.14/export-ext4fs-dirhash-helper.patch linux-5.8/ext4-no-max-dir-size-limit-for-iam-objects.patch linux-5.14/ext4-ialloc-uid-gid-and-pass-owner-down.patch linux-5.14/ext4-projid-xattrs.patch -base/ext4-delayed-iput.patch +sles15sp4/ext4-delayed-iput.patch rhel8/ext4-ext-merge.patch linux-5.14/ext4-xattr-disable-credits-check.patch linux-5.10/ext4-fiemap-kernel-data.patch -- 1.8.3.1