2021-02-01 20:53:13

by Dai Ngo

[permalink] [raw]
Subject: [PATCH v2] vfs: generic_copy_file_checks should return EINVAL when source offset is beyond EOF

Fix by returning -EINVAL instead of 0, per man page of copy_file_range,
when the requested range extends beyond the end of the source file.
Problem was discovered by subtest inter11 of nfstest_ssc.

Signed-off-by: Dai Ngo <[email protected]>
---
fs/read_write.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/read_write.c b/fs/read_write.c
index 75f764b43418..438c00910716 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -1445,7 +1445,7 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
/* Shorten the copy to EOF */
size_in = i_size_read(inode_in);
if (pos_in >= size_in)
- count = 0;
+ count = -EINVAL;
else
count = min(count, size_in - (uint64_t)pos_in);

--
2.9.5


2021-02-01 21:18:24

by Darrick J. Wong

[permalink] [raw]
Subject: Re: [PATCH v2] vfs: generic_copy_file_checks should return EINVAL when source offset is beyond EOF

On Mon, Feb 01, 2021 at 03:49:52PM -0500, Dai Ngo wrote:
> Fix by returning -EINVAL instead of 0, per man page of copy_file_range,

Huh? That's not what the manpage[1] says:

RETURN VALUE
Upon successful completion, copy_file_range() will return the
number of bytes copied between files. This could be less than the
length originally requested. If the file offset of fd_in is at or past
the end of file, no bytes are copied, and copy_file_range() returns
zero.

--D

[1] https://man7.org/linux/man-pages/man2/copy_file_range.2.html#RETURN_VALUE

> when the requested range extends beyond the end of the source file.
> Problem was discovered by subtest inter11 of nfstest_ssc.
>
> Signed-off-by: Dai Ngo <[email protected]>
> ---
> fs/read_write.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/read_write.c b/fs/read_write.c
> index 75f764b43418..438c00910716 100644
> --- a/fs/read_write.c
> +++ b/fs/read_write.c
> @@ -1445,7 +1445,7 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
> /* Shorten the copy to EOF */
> size_in = i_size_read(inode_in);
> if (pos_in >= size_in)
> - count = 0;
> + count = -EINVAL;
> else
> count = min(count, size_in - (uint64_t)pos_in);
>
> --
> 2.9.5
>

2021-02-01 21:53:30

by Dai Ngo

[permalink] [raw]
Subject: Re: [PATCH v2] vfs: generic_copy_file_checks should return EINVAL when source offset is beyond EOF


On 2/1/21 1:14 PM, Darrick J. Wong wrote:
> On Mon, Feb 01, 2021 at 03:49:52PM -0500, Dai Ngo wrote:
>> Fix by returning -EINVAL instead of 0, per man page of copy_file_range,
> Huh? That's not what the manpage[1] says:
>
> RETURN VALUE
> Upon successful completion, copy_file_range() will return the
> number of bytes copied between files. This could be less than the
> length originally requested. If the file offset of fd_in is at or past
> the end of file, no bytes are copied, and copy_file_range() returns
> zero.

In the ERROR section:

ERROR
EINVAL Requested range extends beyond the end of the source file; or the flags argument is not 0.

-Dai

>
> --D
>
> [1] https://urldefense.com/v3/__https://man7.org/linux/man-pages/man2/copy_file_range.2.html*RETURN_VALUE__;Iw!!GqivPVa7Brio!N8XGLyK9dUTWOYWmb3NCuq-9kL-tX8OGcq-sV7M53ub5KMgr7e93a2B8eUryKw$
>
>> when the requested range extends beyond the end of the source file.
>> Problem was discovered by subtest inter11 of nfstest_ssc.
>>
>> Signed-off-by: Dai Ngo <[email protected]>
>> ---
>> fs/read_write.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/fs/read_write.c b/fs/read_write.c
>> index 75f764b43418..438c00910716 100644
>> --- a/fs/read_write.c
>> +++ b/fs/read_write.c
>> @@ -1445,7 +1445,7 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
>> /* Shorten the copy to EOF */
>> size_in = i_size_read(inode_in);
>> if (pos_in >= size_in)
>> - count = 0;
>> + count = -EINVAL;
>> else
>> count = min(count, size_in - (uint64_t)pos_in);
>>
>> --
>> 2.9.5
>>

2021-02-11 19:58:15

by Dai Ngo

[permalink] [raw]
Subject: Re: [PATCH v2] vfs: generic_copy_file_checks should return EINVAL when source offset is beyond EOF

Hi Darrick,

My man page was out-of-date. The latest man page of coy_file_range is more clear:

This EINVAL was removed:

EINVAL Requested range extends beyond the end of the source file; or the flags argument is not 0.

and replaced with this:

*EINVAL *The/flags/ argument is not 0

Thanks,
-Dai

On 2/1/21 1:51 PM, [email protected] wrote:
>
> On 2/1/21 1:14 PM, Darrick J. Wong wrote:
>> On Mon, Feb 01, 2021 at 03:49:52PM -0500, Dai Ngo wrote:
>>> Fix by returning -EINVAL instead of 0, per man page of copy_file_range,
>> Huh?  That's not what the manpage[1] says:
>>
>> RETURN VALUE
>>         Upon successful completion, copy_file_range() will return the
>> number of bytes copied between files.  This could be less than the
>> length originally requested.  If the file offset of fd_in is at or past
>> the end of file, no bytes are copied, and copy_file_range() returns
>> zero.
>
> In the ERROR section:
>
> ERROR
>     EINVAL Requested range extends beyond the end of the source file;
> or the flags argument is not 0.
>
> -Dai
>
>>
>> --D
>>
>> [1]
>> https://urldefense.com/v3/__https://man7.org/linux/man-pages/man2/copy_file_range.2.html*RETURN_VALUE__;Iw!!GqivPVa7Brio!N8XGLyK9dUTWOYWmb3NCuq-9kL-tX8OGcq-sV7M53ub5KMgr7e93a2B8eUryKw$
>>
>>> when the requested range extends beyond the end of the source file.
>>> Problem was discovered by subtest inter11 of nfstest_ssc.
>>>
>>> Signed-off-by: Dai Ngo <[email protected]>
>>> ---
>>>   fs/read_write.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/fs/read_write.c b/fs/read_write.c
>>> index 75f764b43418..438c00910716 100644
>>> --- a/fs/read_write.c
>>> +++ b/fs/read_write.c
>>> @@ -1445,7 +1445,7 @@ static int generic_copy_file_checks(struct
>>> file *file_in, loff_t pos_in,
>>>       /* Shorten the copy to EOF */
>>>       size_in = i_size_read(inode_in);
>>>       if (pos_in >= size_in)
>>> -        count = 0;
>>> +        count = -EINVAL;
>>>       else
>>>           count = min(count, size_in - (uint64_t)pos_in);
>>>   --
>>> 2.9.5
>>>