Re: [PATCH 2/2] ext4: correct the judgment of BUG in ext4_mb_normalize_request

[Jan Kara]

On Sat 21-05-22 21:42:17, Baokun Li wrote:
> When either of the "start + size <= ac->ac_o_ex.fe_logical" or
> "start > ac->ac_o_ex.fe_logical" conditions is met, it indicates
> that the fe_logical is not in the allocated range.
> In this case, it should be bug_ON.
> 
> Fixes: dfe076c106f6 ("ext4: get rid of code duplication")
> Signed-off-by: Baokun Li <libaokun1@xxxxxxxxxx>

I think this is actually wrong. The original condition checks whether
start + size does not overflow the used integer type. Your condition is
much stronger and I don't think it always has to be true. E.g. allocation
goal block (start variable) can be pushed to larger values by existing
preallocation or so.

								Honza 


> diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> index 32410b79b664..d0fb57970648 100644
> --- a/fs/ext4/mballoc.c
> +++ b/fs/ext4/mballoc.c
> @@ -4190,7 +4190,7 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac,
>  	}
>  	rcu_read_unlock();
>  
> -	if (start + size <= ac->ac_o_ex.fe_logical &&
> +	if (start + size <= ac->ac_o_ex.fe_logical ||
>  			start > ac->ac_o_ex.fe_logical) {
>  		ext4_msg(ac->ac_sb, KERN_ERR,
>  			 "start %lu, size %lu, fe_logical %lu",
> -- 
> 2.31.1
> 
-- 
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR