Re: [PATCH v8] vfs: fix copy_file_range regression in cross-fs copies

[dai.ngo@oracle.com]

On 2/23/21 8:47 AM, Amir Goldstein wrote:
> On Tue, Feb 23, 2021 at 6:02 PM <dai.ngo@oracle.com> wrote:
>>
>> On 2/23/21 7:29 AM, dai.ngo@oracle.com wrote:
>>> On 2/23/21 2:32 AM, Luis Henriques wrote:
>>>> On Mon, Feb 22, 2021 at 08:25:27AM -0800, dai.ngo@oracle.com wrote:
>>>>> On 2/22/21 2:24 AM, Luis Henriques wrote:
>>>>>> A regression has been reported by Nicolas Boichat, found while
>>>>>> using the
>>>>>> copy_file_range syscall to copy a tracefs file.  Before commit
>>>>>> 5dae222a5ff0 ("vfs: allow copy_file_range to copy across devices") the
>>>>>> kernel would return -EXDEV to userspace when trying to copy a file
>>>>>> across
>>>>>> different filesystems.  After this commit, the syscall doesn't fail
>>>>>> anymore
>>>>>> and instead returns zero (zero bytes copied), as this file's
>>>>>> content is
>>>>>> generated on-the-fly and thus reports a size of zero.
>>>>>>
>>>>>> This patch restores some cross-filesystem copy restrictions that
>>>>>> existed
>>>>>> prior to commit 5dae222a5ff0 ("vfs: allow copy_file_range to copy
>>>>>> across
>>>>>> devices").  Filesystems are still allowed to fall-back to the VFS
>>>>>> generic_copy_file_range() implementation, but that has now to be done
>>>>>> explicitly.
>>>>>>
>>>>>> nfsd is also modified to fall-back into generic_copy_file_range()
>>>>>> in case
>>>>>> vfs_copy_file_range() fails with -EOPNOTSUPP or -EXDEV.
>>>>>>
>>>>>> Fixes: 5dae222a5ff0 ("vfs: allow copy_file_range to copy across
>>>>>> devices")
>>>>>> Link:
>>>>>> https://urldefense.com/v3/__https://lore.kernel.org/linux-fsdevel/20210212044405.4120619-1-drinkcat@chromium.org/__;!!GqivPVa7Brio!P1UWThiSkxbjfjFQWNYJmCxGEkiLFyvHjH6cS-G1ZTt1z-TeqwGQgQmi49dC6w$
>>>>>> Link:
>>>>>> https://urldefense.com/v3/__https://lore.kernel.org/linux-fsdevel/CANMq1KDZuxir2LM5jOTm0xx*BnvW=ZmpsG47CyHFJwnw7zSX6Q@mail.gmail.com/__;Kw!!GqivPVa7Brio!P1UWThiSkxbjfjFQWNYJmCxGEkiLFyvHjH6cS-G1ZTt1z-TeqwGQgQmgCmMHzA$
>>>>>> Link:
>>>>>> https://urldefense.com/v3/__https://lore.kernel.org/linux-fsdevel/20210126135012.1.If45b7cdc3ff707bc1efa17f5366057d60603c45f@changeid/__;!!GqivPVa7Brio!P1UWThiSkxbjfjFQWNYJmCxGEkiLFyvHjH6cS-G1ZTt1z-TeqwGQgQmzqItkrQ$
>>>>>> Reported-by: Nicolas Boichat <drinkcat@chromium.org>
>>>>>> Signed-off-by: Luis Henriques <lhenriques@suse.de>
>>>>>> ---
>>>>>> Changes since v7
>>>>>> - set 'ret' to '-EOPNOTSUPP' before the clone 'if' statement so
>>>>>> that the
>>>>>>      error returned is always related to the 'copy' operation
>>>>>> Changes since v6
>>>>>> - restored i_sb checks for the clone operation
>>>>>> Changes since v5
>>>>>> - check if ->copy_file_range is NULL before calling it
>>>>>> Changes since v4
>>>>>> - nfsd falls-back to generic_copy_file_range() only *if* it gets
>>>>>> -EOPNOTSUPP
>>>>>>      or -EXDEV.
>>>>>> Changes since v3
>>>>>> - dropped the COPY_FILE_SPLICE flag
>>>>>> - kept the f_op's checks early in generic_copy_file_checks,
>>>>>> implementing
>>>>>>      Amir's suggestions
>>>>>> - modified nfsd to use generic_copy_file_range()
>>>>>> Changes since v2
>>>>>> - do all the required checks earlier, in generic_copy_file_checks(),
>>>>>>      adding new checks for ->remap_file_range
>>>>>> - new COPY_FILE_SPLICE flag
>>>>>> - don't remove filesystem's fallback to generic_copy_file_range()
>>>>>> - updated commit changelog (and subject)
>>>>>> Changes since v1 (after Amir review)
>>>>>> - restored do_copy_file_range() helper
>>>>>> - return -EOPNOTSUPP if fs doesn't implement CFR
>>>>>> - updated commit description
>>>>>>
>>>>>>     fs/nfsd/vfs.c   |  8 +++++++-
>>>>>>     fs/read_write.c | 49
>>>>>> ++++++++++++++++++++++++-------------------------
>>>>>>     2 files changed, 31 insertions(+), 26 deletions(-)
>>>>>>
>>>>>> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
>>>>>> index 04937e51de56..23dab0fa9087 100644
>>>>>> --- a/fs/nfsd/vfs.c
>>>>>> +++ b/fs/nfsd/vfs.c
>>>>>> @@ -568,6 +568,7 @@ __be32 nfsd4_clone_file_range(struct nfsd_file
>>>>>> *nf_src, u64 src_pos,
>>>>>>     ssize_t nfsd_copy_file_range(struct file *src, u64 src_pos,
>>>>>> struct file *dst,
>>>>>>                      u64 dst_pos, u64 count)
>>>>>>     {
>>>>>> +    ssize_t ret;
>>>>>>         /*
>>>>>>          * Limit copy to 4MB to prevent indefinitely blocking an nfsd
>>>>>> @@ -578,7 +579,12 @@ ssize_t nfsd_copy_file_range(struct file *src,
>>>>>> u64 src_pos, struct file *dst,
>>>>>>          * limit like this and pipeline multiple COPY requests.
>>>>>>          */
>>>>>>         count = min_t(u64, count, 1 << 22);
>>>>>> -    return vfs_copy_file_range(src, src_pos, dst, dst_pos, count, 0);
>>>>>> +    ret = vfs_copy_file_range(src, src_pos, dst, dst_pos, count, 0);
>>>>>> +
>>>>>> +    if (ret == -EOPNOTSUPP || ret == -EXDEV)
>>>>>> +        ret = generic_copy_file_range(src, src_pos, dst, dst_pos,
>>>>>> +                          count, 0);
>>>>>> +    return ret;
>>>>>>     }
>>>>>>     __be32 nfsd4_vfs_fallocate(struct svc_rqst *rqstp, struct svc_fh
>>>>>> *fhp,
>>>>>> diff --git a/fs/read_write.c b/fs/read_write.c
>>>>>> index 75f764b43418..5a26297fd410 100644
>>>>>> --- a/fs/read_write.c
>>>>>> +++ b/fs/read_write.c
>>>>>> @@ -1388,28 +1388,6 @@ ssize_t generic_copy_file_range(struct file
>>>>>> *file_in, loff_t pos_in,
>>>>>>     }
>>>>>>     EXPORT_SYMBOL(generic_copy_file_range);
>>>>>> -static ssize_t do_copy_file_range(struct file *file_in, loff_t
>>>>>> pos_in,
>>>>>> -                  struct file *file_out, loff_t pos_out,
>>>>>> -                  size_t len, unsigned int flags)
>>>>>> -{
>>>>>> -    /*
>>>>>> -     * Although we now allow filesystems to handle cross sb copy,
>>>>>> passing
>>>>>> -     * a file of the wrong filesystem type to filesystem driver
>>>>>> can result
>>>>>> -     * in an attempt to dereference the wrong type of
>>>>>> ->private_data, so
>>>>>> -     * avoid doing that until we really have a good reason.  NFS
>>>>>> defines
>>>>>> -     * several different file_system_type structures, but they all
>>>>>> end up
>>>>>> -     * using the same ->copy_file_range() function pointer.
>>>>>> -     */
>>>>>> -    if (file_out->f_op->copy_file_range &&
>>>>>> -        file_out->f_op->copy_file_range ==
>>>>>> file_in->f_op->copy_file_range)
>>>>>> -        return file_out->f_op->copy_file_range(file_in, pos_in,
>>>>>> -                               file_out, pos_out,
>>>>>> -                               len, flags);
>>>>>> -
>>>>>> -    return generic_copy_file_range(file_in, pos_in, file_out,
>>>>>> pos_out, len,
>>>>>> -                       flags);
>>>>>> -}
>>>>>> -
>>>>>>     /*
>>>>>>      * Performs necessary checks before doing a file copy
>>>>>>      *
>>>>>> @@ -1427,6 +1405,25 @@ static int generic_copy_file_checks(struct
>>>>>> file *file_in, loff_t pos_in,
>>>>>>         loff_t size_in;
>>>>>>         int ret;
>>>>>> +    /*
>>>>>> +     * Although we now allow filesystems to handle cross sb copy,
>>>>>> passing
>>>>>> +     * a file of the wrong filesystem type to filesystem driver
>>>>>> can result
>>>>>> +     * in an attempt to dereference the wrong type of
>>>>>> ->private_data, so
>>>>>> +     * avoid doing that until we really have a good reason.  NFS
>>>>>> defines
>>>>>> +     * several different file_system_type structures, but they all
>>>>>> end up
>>>>>> +     * using the same ->copy_file_range() function pointer.
>>>>>> +     */
>>>>>> +    if (file_out->f_op->copy_file_range) {
>>>>>> +        if (file_in->f_op->copy_file_range !=
>>>>>> +            file_out->f_op->copy_file_range)
>>>>>> +            return -EXDEV;
>>>>>> +    } else if (file_in->f_op->remap_file_range) {
>>>>>> +        if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
>>>>>> +            return -EXDEV;
>>>>> I think this check is redundant, it's done in vfs_copy_file_range.
>>>>> If this check is removed then the else clause below should be removed
>>>>> also. Once this check and the else clause are removed then might as
>>>>> well move the the check of copy_file_range from here to
>>>>> vfs_copy_file_range.
>>>>>
>>>> I don't think it's really redundant, although I agree is messy due to
>>>> the
>>>> fact we try to clone first instead of copying them.
>>>>
>>>> So, in the clone path, this is the only place where we return -EXDEV if:
>>>>
>>>> 1) we don't have ->copy_file_range *and*
>>>> 2) we have ->remap_file_range but the i_sb are different.
>>>>
>>>> The check in vfs_copy_file_range() is only executed if:
>>>>
>>>> 1) we have *valid* ->copy_file_range ops and/or
>>>> 2) we have *valid* ->remap_file_range
>>>>
>>>> So... if we remove the check in generic_copy_file_checks() as you
>>>> suggest
>>>> and:
>>>> - we don't have ->copy_file_range,
>>>> - we have ->remap_file_range but
>>>> - the i_sb are different
>>>>
>>>> we'll return the -EOPNOTSUPP (the one set in line "ret =
>>>> -EOPNOTSUPP;" in
>>>> function vfs_copy_file_range() ) instead of -EXDEV.
>>> Yes, this is the different.The NFS code handles both -EOPNOTSUPP and
>>> -EXDEVV by doing generic_copy_file_range.  Do any other consumers of
>>> vfs_copy_file_range rely on -EXDEV and not -EOPNOTSUPP and which is
>>> the correct error code for this case? It seems to me that -EOPNOTSUPP
>>> is more appropriate than EXDEV when (sb1 != sb2).
> EXDEV is the right code for:
> filesystem supports the operation but not for sb1 != sb1.
>
>> So with the current patch, for a clone operation across 2 filesystems:
>>
>>     . if src and dst filesystem support both copy_file_range and
>>       map_file_range then the code returns -ENOTSUPPORT.
>>
> Why do you say that?
> Which code are you referring to exactly?

If the filesystems support both copy_file_range and map_file_range,
it passes the check in generic_file_check but it fails with the
check in vfs_copy_file_range and returns -ENOTSUPPORT (added by
the v8 patch)

-Dai

> Did you see this behavior in a test?
>
>>     . if the filesystems only support map_file_range then the
>>       code returns -EXDEV
>>
>> This seems confusing, shouldn't only 1 error code returned for this case?
>>
>  From my read of the code, user will get -EXDEV in both the cases you
> listed.
>
> Thanks,
> Amir.