Re: [PATCH v4 3/3] ext4: Add IOMAP_REPORT support for inline data

[Andreas Gruenbacher <agruenba@redhat.com>]

On Tue, Jul 25, 2017 at 2:16 PM, Jan Kara <jack@suse.cz> wrote:
> On Fri 07-07-17 23:28:01, Andreas Gruenbacher wrote:
>> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
>> index 82f3f7d..e2b0a8a 100644
>> --- a/fs/ext4/inode.c
>> +++ b/fs/ext4/inode.c
>> @@ -3361,8 +3361,13 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>>       bool delalloc = false;
>>       int ret;
>>
>> -     if (WARN_ON_ONCE(ext4_has_inline_data(inode)))
>> -             return -ERANGE;
>> +     if (ext4_has_inline_data(inode)) {
>> +             if (WARN_ON_ONCE(!(flags & IOMAP_REPORT)))
>> +                     return -ERANGE;
>> +             if (!ext4_inline_data_iomap(inode, iomap) &&
>> +                 offset < iomap->length)
>
> Hum, what's the thinking behind this "offset < iomap->length" check? If it
> fails, we'd just fall through to the normal case which I'm not sure is
> guaranteed to be safe? Shouldn't we return error instead?

Indeed. I'll send out an updated patch queue including this and
several other changes shortly.

Thanks,
Andreas