Re: [PATCH 2/2] hugetlbfs: Improve exception handling in hugetlbfs_fill_super()

[Markus Elfring]

>> +++ b/fs/hugetlbfs/inode.c
>> @@ -1483,7 +1483,7 @@ hugetlbfs_fill_super(struct super_block *sb, struct fs_context *fc)
>>  						     ctx->max_hpages,
>>  						     ctx->min_hpages);
>>  		if (!sbinfo->spool)
>> -			goto out_free;
>> +			goto free_sbinfo;
>>  	}
>>  	sb->s_maxbytes = MAX_LFS_FILESIZE;
>>  	sb->s_blocksize = huge_page_size(ctx->hstate);
>> @@ -1499,10 +1499,12 @@ hugetlbfs_fill_super(struct super_block *sb, struct fs_context *fc)
>>  	sb->s_stack_depth = FILESYSTEM_MAX_STACK_DEPTH;
>>  	sb->s_root = d_make_root(hugetlbfs_get_root(sb, ctx));
>>  	if (!sb->s_root)
>> -		goto out_free;
>> +		goto free_spool;
>>  	return 0;
>> -out_free:
>> +
>> +free_spool:
>>  	kfree(sbinfo->spool);
>> +free_sbinfo:
>>  	kfree(sbinfo);
>>  	return -ENOMEM;
>>  }
>
> This is more complex.  NACK.

I am curious how coding style preferences will evolve further.

See also:
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

Regards,
Markus