Re: [PATCH v2 3/3] f2fs: support errors=remount-ro|continue|panic mountoption

From: Jaegeuk Kim
Date: Thu Sep 29 2022 - 14:31:53 EST

On 09/29, Chao Yu wrote:
> This patch supports errors=remount-ro|continue|panic mount option.
>
> Signed-off-by: Chao Yu <chao@xxxxxxxxxx>
> ---
> v2:
> - fix to load sb.s_stop_reason during fill_super().
> Documentation/filesystems/f2fs.rst | 4 +
> fs/f2fs/checkpoint.c | 7 +-
> fs/f2fs/f2fs.h | 18 ++++-
> fs/f2fs/super.c | 115 ++++++++++++++++++++++++++++-
> 4 files changed, 133 insertions(+), 11 deletions(-)
>
> diff --git a/Documentation/filesystems/f2fs.rst b/Documentation/filesystems/f2fs.rst
> index d0c09663dae8..6aa14e03f337 100644
> --- a/Documentation/filesystems/f2fs.rst
> +++ b/Documentation/filesystems/f2fs.rst
> @@ -341,6 +341,10 @@ memory=%s Control memory mode. This supports "normal" and "low" modes.
> Because of the nature of low memory devices, in this mode, f2fs
> will try to save memory sometimes by sacrificing performance.
> "normal" mode is the default mode and same as before.
> +errors=%s Specify f2fs behavior on critical errors. This supports modes:
> + "panic", "continue" and "remount-ro", respectively, triggers
> + panic immediately, continue without doing anything, or remount
> + the partition in read-only mode (default behavior).
> ======================== ============================================================
>
> Debugfs Entries
> diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> index 0c82dae082aa..109e96c15b84 100644
> --- a/fs/f2fs/checkpoint.c
> +++ b/fs/f2fs/checkpoint.c
> @@ -30,12 +30,9 @@ void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
> unsigned char reason)
> {
> f2fs_build_fault_attr(sbi, 0, 0);
> - set_ckpt_flags(sbi, CP_ERROR_FLAG);
> - if (!end_io) {
> + if (!end_io)
> f2fs_flush_merged_writes(sbi);
> -
> - f2fs_handle_stop(sbi, reason);
> - }
> + f2fs_handle_critical_error(sbi, reason, end_io);
> }
>
> /*
> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> index 2ed00111a399..254f7f153477 100644
> --- a/fs/f2fs/f2fs.h
> +++ b/fs/f2fs/f2fs.h
> @@ -160,6 +160,7 @@ struct f2fs_mount_info {
> int fs_mode; /* fs mode: LFS or ADAPTIVE */
> int bggc_mode; /* bggc mode: off, on or sync */
> int memory_mode; /* memory mode */
> + int errors; /* errors parameter */
> int discard_unit; /*
> * discard command's offset/size should
> * be aligned to this unit: block,
> @@ -1382,7 +1383,11 @@ enum {
> MEMORY_MODE_LOW, /* memory mode for low memry devices */
> };
>
> -
> +enum errors_option {
> + MOUNT_ERRORS_READONLY, /* remount fs ro on errors */
> + MOUNT_ERRORS_CONTINUE, /* continue on errors */
> + MOUNT_ERRORS_PANIC, /* panic on errors */
> +};
>
> static inline int f2fs_test_bit(unsigned int nr, char *addr);
> static inline void f2fs_set_bit(unsigned int nr, char *addr);
> @@ -1818,7 +1823,13 @@ struct f2fs_sb_info {
>
> struct workqueue_struct *post_read_wq; /* post read workqueue */
>
> - unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
> + /*
> + * If we are in irq context, let's update error information into
> + * on-disk superblock in the work.
> + */
> + struct work_struct s_error_work;
> + unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
> + unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */
> spinlock_t error_lock; /* protect errors array */
> bool error_dirty; /* errors of sb is dirty */
>
> @@ -3563,7 +3574,8 @@ int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
> int f2fs_quota_sync(struct super_block *sb, int type);
> loff_t max_file_blocks(struct inode *inode);
> void f2fs_quota_off_umount(struct super_block *sb);
> -void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason);
> +void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
> + bool irq_context);
> void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
> int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
> int f2fs_sync_fs(struct super_block *sb, int sync);
> diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
> index ee0a5a17ba39..60bd0344f573 100644
> --- a/fs/f2fs/super.c
> +++ b/fs/f2fs/super.c
> @@ -161,6 +161,7 @@ enum {
> Opt_nogc_merge,
> Opt_discard_unit,
> Opt_memory_mode,
> + Opt_errors,
> Opt_err,
> };
>
> @@ -238,6 +239,7 @@ static match_table_t f2fs_tokens = {
> {Opt_nogc_merge, "nogc_merge"},
> {Opt_discard_unit, "discard_unit=%s"},
> {Opt_memory_mode, "memory=%s"},
> + {Opt_errors, "errors=%s"},
> {Opt_err, NULL},
> };
>
> @@ -1253,6 +1255,25 @@ static int parse_options(struct super_block *sb, char *options, bool is_remount)
> }
> kfree(name);
> break;
> + case Opt_errors:
> + name = match_strdup(&args[0]);
> + if (!name)
> + return -ENOMEM;
> + if (!strcmp(name, "remount-ro")) {
> + F2FS_OPTION(sbi).errors =
> + MOUNT_ERRORS_READONLY;
> + } else if (!strcmp(name, "continue")) {
> + F2FS_OPTION(sbi).errors =
> + MOUNT_ERRORS_CONTINUE;
> + } else if (!strcmp(name, "panic")) {
> + F2FS_OPTION(sbi).errors =
> + MOUNT_ERRORS_PANIC;
> + } else {
> + kfree(name);
> + return -EINVAL;
> + }
> + kfree(name);
> + break;
> default:
> f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
> p);
> @@ -2031,6 +2052,13 @@ static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
> else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
> seq_printf(seq, ",memory=%s", "low");
>
> + if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
> + seq_printf(seq, ",errors=%s", "remount-ro");
> + else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
> + seq_printf(seq, ",errors=%s", "continue");
> + else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
> + seq_printf(seq, ",errors=%s", "panic");
> +
> return 0;
> }
>
> @@ -2053,6 +2081,7 @@ static void default_options(struct f2fs_sb_info *sbi)
> F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
> F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
> F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
> + F2FS_OPTION(sbi).errors = MOUNT_ERRORS_READONLY;
>
> sbi->sb->s_flags &= ~SB_INLINECRYPT;
>
> @@ -3846,7 +3875,16 @@ int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
> return err;
> }
>
> -void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason)
> +static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
> +
> +{
> + spin_lock(&sbi->error_lock);
> + if (sbi->stop_reason[reason] < ((1 << BITS_PER_BYTE) - 1))
> + sbi->stop_reason[reason]++;
> + spin_unlock(&sbi->error_lock);
> +}
> +
> +void f2fs_record_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
>
> {
> struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
> @@ -3854,8 +3892,9 @@ void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason)
>
> f2fs_down_write(&sbi->sb_lock);
>
> - if (raw_super->s_stop_reason[reason] < ((1 << BITS_PER_BYTE) - 1))
> - raw_super->s_stop_reason[reason]++;
> + spin_lock(&sbi->error_lock);
> + memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
> + spin_unlock(&sbi->error_lock);
>
> err = f2fs_commit_super(sbi, false);
> if (err)
> @@ -3909,6 +3948,74 @@ void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
> f2fs_up_write(&sbi->sb_lock);
> }
>
> +static bool system_going_down(void)
> +{
> + return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
> + || system_state == SYSTEM_RESTART;
> +}
> +
> +void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
> + bool irq_context)
> +{
> + struct super_block *sb = sbi->sb;
> + bool force_ro = reason == STOP_CP_REASON_SHUTDOWN;

Can we check f2fs stopped checkpoint intentionally and use it here?

> + bool continue_fs = !force_ro &&
> + F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
> +
> + if (!continue_fs && !sb_rdonly(sb))
> + set_ckpt_flags(sbi, CP_ERROR_FLAG);
> +
> + if (!bdev_read_only(sb->s_bdev)) {
> + save_stop_reason(sbi, reason);
> +
> + if (irq_context)
> + schedule_work(&sbi->s_error_work);
> + else
> + f2fs_record_stop_reason(sbi, reason);
> + }
> +
> + /*
> + * We force ERRORS_RO behavior when system is rebooting. Otherwise we
> + * could panic during 'reboot -f' as the underlying device got already
> + * disabled.
> + */
> + if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
> + !force_ro && !system_going_down())
> + panic("F2FS-fs (device %s): panic forced after error\n",
> + sb->s_id);
> +
> + /* continue filesystem operators if errors=continue */
> + if (continue_fs || sb_rdonly(sb))
> + return;
> +
> + f2fs_warn(sbi, "Remounting filesystem read-only");
> + /*
> + * Make sure updated value of ->s_mount_flags will be visible before
> + * ->s_flags update
> + */
> + smp_wmb();
> + sb->s_flags |= SB_RDONLY;
> +
> + f2fs_stop_gc_thread(sbi);
> + f2fs_stop_discard_thread(sbi);
> +}
> +
> +static void flush_error_work(struct work_struct *work)
> +{
> + struct f2fs_sb_info *sbi = container_of(work,
> + struct f2fs_sb_info, s_error_work);
> + int ret;
> +
> + if (sb_rdonly(sbi->sb))
> + return;
> +
> + f2fs_down_write(&sbi->sb_lock);
> + ret = f2fs_commit_super(sbi, false);
> + f2fs_up_write(&sbi->sb_lock);
> + if (ret)
> + f2fs_err(sbi, "flush_error_work() fails to commit superblock ret:%d", ret);
> +}
> +
> static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
> {
> struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
> @@ -4256,8 +4363,10 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
> goto free_devices;
> }
>
> + INIT_WORK(&sbi->s_error_work, flush_error_work);
> spin_lock_init(&sbi->error_lock);
> memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
> + memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
>
> sbi->total_valid_node_count =
> le32_to_cpu(sbi->ckpt->valid_node_count);
> --
> 2.36.1