Re: [PATCH v5 1/2] ext4: Discard preallocations before releasing group lock

[Ritesh Harjani <riteshh@linux.ibm.com>]

On 9/16/20 5:08 PM, Ye Bin wrote:
> From: Jan Kara <jack@suse.cz>
> 
> ext4_mb_discard_group_preallocations() can be releasing group lock with
> preallocations accumulated on its local list. Thus although
> discard_pa_seq was incremented and concurrent allocating processes will
> be retrying allocations, it can happen that premature ENOSPC error is
> returned because blocks used for preallocations are not available for
> reuse yet. Make sure we always free locally accumulated preallocations
> before releasing group lock.
> 
> Fixes: 07b5b8e1ac40 ("ext4: mballoc: introduce pcpu seqcnt for freeing PA to improve ENOSPC handling")
> Signed-off-by: Jan Kara <jack@suse.cz>
> Signed-off-by: Ye Bin <yebin10@huawei.com>
> Reviewed-by: Ritesh Harjani <riteshh@linux.ibm.com>
> ---
>   fs/ext4/mballoc.c | 28 +++++++++++-----------------
>   1 file changed, 11 insertions(+), 17 deletions(-)
> 
> diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> index 132c118d12e1..f736819a381b 100644
> --- a/fs/ext4/mballoc.c
> +++ b/fs/ext4/mballoc.c
> @@ -4215,22 +4215,6 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
>   		list_add(&pa->u.pa_tmp_list, &list);
>   	}
>   
> -	/* if we still need more blocks and some PAs were used, try again */
> -	if (free < needed && busy) {
> -		busy = 0;
> -		ext4_unlock_group(sb, group);
> -		cond_resched();
> -		goto repeat;
> -	}
> -
> -	/* found anything to free? */
> -	if (list_empty(&list)) {
> -		BUG_ON(free != 0);
> -		mb_debug(sb, "Someone else may have freed PA for this group %u\n",
> -			 group);
> -		goto out;
> -	}
> -
>   	/* now free all selected PAs */
>   	list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) {
>   
> @@ -4248,7 +4232,17 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
>   		call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback);
>   	}
>   
> -out:
> +	/* if we still need more blocks and some PAs were used, try again */
> +	if (free < needed && busy) {
> +		ext4_unlock_group(sb, group);
> +		cond_resched();
> +		busy = 0;
> +		/* Make sure we increment discard_pa_seq again */
> +		needed -= free;
> +		free = 0;

Oops sorry about getting back to this.
But if we are making free 0 here so we may return a wrong free value
when we return from this function. We should fix that by also accounting
previous freed blocks at the time of final return from this function.


-ritesh

> +		goto repeat;
> +	}
> +
>   	ext4_unlock_group(sb, group);
>   	ext4_mb_unload_buddy(&e4b);
>   	put_bh(bitmap_bh);
>