Re: [f2fs-dev] [PATCH v2] f2fs: change the way of handling range.len in F2FS_IOC_SEC_TRIM_FILE

[Chao Yu]

On 2020/7/16 0:42, Eric Biggers wrote:
> On Wed, Jul 15, 2020 at 07:25:13PM +0900, Daeho Jeong wrote:
>> Chao,
>>
>> I can't find fscrypt_zeroout_range_inline_crypt() function. Do you
>> mean we need to implement this one for inline encryption?
>>
>> 2020ë 7ì 15ì (ì) ìí 4:17, Chao Yu <yuchao0@xxxxxxxxxx>ëì ìì:
>>>
>>> On 2020/7/15 14:54, Daeho Jeong wrote:
>>>> You mean we can support ZEROOUT option only for encrypted files of
>>>> non-multidevice f2fs,
>>>> and return -EOPNOTSUPP in the multidevice case, right now?
>>>
>>> Yes, something like:
>>>
>>> f2fs_sec_trim_file()
>>>
>>> if ((range.flags & F2FS_TRIM_FILE_ZEROOUT) &&
>>>         f2fs_encrypted_file() && f2fs_is_multi_device())
>>>         return -EOPNOTSUPP;
>>>
>>>
>>> f2fs_secure_erase()
>>>
>>> if (!ret && (flags & F2FS_TRIM_FILE_ZEROOUT)) {
>>>         if (f2fs_encrypted_file()) {
>>>                 if (fscrypt_inode_uses_fs_layer_crypto)
>>>                         ret = fscrypt_zeroout_range();
>>>                 else
>>>                         ret = fscrypt_zeroout_range_inline_crypt();
>>>         } else {
>>>                 ret = blkdev_issue_zeroout();
>>>         }
>>> }
> 
> fscrypt_zeroout_range_inline_crypt() is being added by
> "fscrypt: add inline encryption support", which is queued in the fscrypt tree
> (the master branch of https://git.kernel.org/pub/scm/fs/fscrypt/fscrypt.git).
> 
> But that's not actually relevant here because fscrypt_zeroout_range() calls
> fscrypt_zeroout_range_inline_crypt() when needed.

Oh, correct, thanks for pointing out.

Thanks,

> 
> Just use fscrypt_zeroout_range().
> 
> - Eric
> .
>