[PATCH 0/1] MR730: user32: Implement DefRawInputProc.

[PATCH 1/1] user32: Implement DefRawInputProc.

[Andrey Gusev]

From: Andrey Gusev

Used in The Testament of Sherlock Holmes on mouse move.
---
 dlls/user32/input.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/dlls/user32/input.c b/dlls/user32/input.c
index a329150f745..3512a967552 100644
--- a/dlls/user32/input.c
+++ b/dlls/user32/input.c
@@ -649,7 +649,14 @@ UINT WINAPI GetRawInputDeviceInfoA( HANDLE device, UINT command, void *data, UIN
  */
 LRESULT WINAPI DefRawInputProc( RAWINPUT **data, INT data_count, UINT header_size )
 {
-    FIXME( "data %p, data_count %d, header_size %u stub!\n", data, data_count, header_size );
+    TRACE( "data %p, data_count %d, header_size %u.\n", data, data_count, header_size );
+
+    if (header_size != sizeof(RAWINPUTHEADER))
+    {
+        WARN( "Invalid structure size %u.\n", header_size );
+        RtlSetLastWin32Error( ERROR_INVALID_PARAMETER );
+        return ~0u;
+    }

return 0;
 }

[Huw Davies (＠huw)]

Some tests would be nice.  This function is documented to do very little, so while surprising, this may be close to the correct implementation.  However, `SetLastError()` should be used instead of `RtlSetLastWin32Error()` if indeed the tests show that the last error is set.

[Andrey Gusev (＠agusev)]

I don't write tests. The game gives expected 16 bytes for header_size. Regarding `SetLastError()` it was changed not long ago with [1dfedae](1dfedae9030b18d6648d48fe613a1ade2af382e5)

[Huw Davies (＠huw)]

> I don't write tests.

Then it's going to hard to convince the reviewer that this is correct.
> Regarding `SetLastError()` it was changed not long ago with [1dfedae](/wine/wine/-/commit/1dfedae9030b18d6648d48fe613a1ade2af382e5)

That's on the unixlib side of win32u.  This code is in user32, which is PE.

[Andrey Gusev (＠agusev)]

I'm not going to convince. I'm only proposed my solution. If someone really thinks it worth to check this 'very complicated' function with tests, I don't mind.