On Saturday, August 24, 2013, Branko Čibej wrote:
On 24.08.2013 06:29, Ivan Zhakov wrote:
> On Sat, Aug 24, 2013 at 3:48 AM, William A. Rowe Jr.
> <wrowe@rowe-clan.net> wrote:
>> On Fri, 23 Aug 2013 18:39:35 -0500
>> "William A. Rowe Jr." <wrowe@rowe-clan.net> wrote:
>>
>>> On Sat, 24 Aug 2013 00:33:38 +0400
>>> Ivan Zhakov <ivan@visualsvn.com> wrote:
>>>
>>>> Actually Windows supports atomic seek-to-end+write: file should be
>>>> opened with FILE_APPEND_DATA access right only [1] or Offset and
>>>> OffsetHigh should be 0xFFFFFFFF if overlapped I/O is used [2].
>>>>
>>>> I'm reopening this thread because in Subversion we found case where
>>>> we need true atomic append across processes/threads. So I'm willing
>>>> to create a patch implementing atomic append on Windows. Is right
>>>> idea for APR or not? Any concerns will be very helpful.
>>>>
>>>> [1]
>>>> http://msdn.microsoft.com/en-us/library/windows/desktop/aa363778%28v=vs.85%29.aspx
>>>> [2]
>>>> http://msdn.microsoft.com/en-us/library/windows/desktop/aa365747%28v=vs.85%29.aspx
>>> IIRC the difference is that you have writev on unix to atomically
>>> write multiple buffers.  On Windows we fake writev, so your proposed
>>> atomic writes are no longer atomic.
> Subversion doesn't use writev for file I/O, so implementing atomic
> writes is enough for our case.

Would that mean that writev on Windows uses a mutex while plain write
does not? How do you avoid a race between write and writev then?
Currently apr_file_writev is alias for calling apr_file_write multiple times without taking lock. Also please note that mutexes helps only if write performed using same file_t in one process, it does not race conditions when write operation from different threads or processes using different handles. 


--
Ivan Zhakov
CTO | VisualSVN | http://www.visualsvn.com