Merge lp:~ignacio-nin/sysbench/0.5-bug1262178 into lp:sysbench

[Ignacio Nin]

Fix for bug #1262178

To test:
$ sysbench --test=fileio --file-num=1 --file-total-size=1M prepare
$ sysbench --test=fileio --file-num=1 --file-total-size=1M --file-io-mode=mmap --file-test-mode=seqwr run

[Alexey Kopytov (akopytov)]

According to http://msdn.microsoft.com/en-us/library/windows/desktop/aa366563%28v=vs.85%29.aspx the return value of FlushViewOfFile differs from the UNIX paragidm, i.e. zero on error and a nonzero value on success. So the Windows codepath should be

return !FlushViewOfFile(...);

I will fix it in the merge.

review: Approve

[Ignacio Nin (ignacio-nin)]

> According to http://msdn.microsoft.com/en-
> us/library/windows/desktop/aa366563%28v=vs.85%29.aspx the return value of
> FlushViewOfFile differs from the UNIX paragidm, i.e. zero on error and a
> nonzero value on success. So the Windows codepath should be
>
> return !FlushViewOfFile(...);

Classic Microsoft ;)

>
> I will fix it in the merge.

Thank you for your time!!

=== modified file 'sysbench/tests/fileio/sb_fileio.c'
--- sysbench/tests/fileio/sb_fileio.c	2013-07-04 06:28:56 +0000
+++ sysbench/tests/fileio/sb_fileio.c	2013-12-18 12:17:48 +0000
@@ -1502,9 +1502,9 @@
   else if (file_io_mode == FILE_IO_MODE_MMAP)
   {
 #ifndef _WIN32
-    msync(mmaps[file_id], file_size, MS_SYNC | MS_INVALIDATE);
+    return msync(mmaps[file_id], file_size, MS_SYNC | MS_INVALIDATE);
 #else
-    FlushViewOfFile(mmaps[file_id], (size_t)file_size);
+    return FlushViewOfFile(mmaps[file_id], (size_t)file_size);
 #endif
   }
 #endif