[PATCH 0/2] MR9538: winegstreamer: Return S_FALSE from DMO when there is not enough data.

[PATCH 1/2] mf/tests: Test what's returned from ProcessOutput when input ran out.

[Yuxuan Shui]

From: Yuxuan Shui

---
 dlls/mf/tests/transform.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/dlls/mf/tests/transform.c b/dlls/mf/tests/transform.c
index e6ff4a0836d..1ee5a8bc4ac 100644
--- a/dlls/mf/tests/transform.c
+++ b/dlls/mf/tests/transform.c
@@ -8355,6 +8355,14 @@ static void test_wmv_decoder_media_object(void)
     diff = check_dmo_output_data_buffer(&output_data_buffer, &output_buffer_desc_nv12, L"nv12frame.bmp", 0, 300000);
     ok(diff == 0, "Got %lu%% diff.\n", diff);

+    output_media_buffer->length = 0;
+    output_data_buffer.pBuffer = &output_media_buffer->IMediaBuffer_iface;
+    output_data_buffer.dwStatus = 0xdeadbeef;
+    output_data_buffer.rtTimestamp = 0xdeadbeef;
+    output_data_buffer.rtTimelength = 0xdeadbeef;
+    hr = IMediaObject_ProcessOutput(media_object, 0, 1, &output_data_buffer, &status);
+    todo_wine ok(hr == S_FALSE, "ProcessOutput returned %#lx.\n", hr);
+
     hr = IMediaObject_AllocateStreamingResources(media_object);
     ok(hr == S_OK, "Got hr %#lx.\n", hr);
     hr = IMediaObject_FreeStreamingResources(media_object);

[PATCH 2/2] winegstreamer: Return S_FALSE from DMO when there is not enough data.

[Yuxuan Shui]

From: Yuxuan Shui

---
 dlls/mf/tests/transform.c      | 3 +--
 dlls/winegstreamer/wg_sample.c | 2 ++
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/dlls/mf/tests/transform.c b/dlls/mf/tests/transform.c
index 1ee5a8bc4ac..e5c65ace228 100644
--- a/dlls/mf/tests/transform.c
+++ b/dlls/mf/tests/transform.c
@@ -8328,7 +8328,6 @@ static void test_wmv_decoder_media_object(void)
     output_data_buffer.rtTimestamp = 0xdeadbeef;
     output_data_buffer.rtTimelength = 0xdeadbeef;
     hr = IMediaObject_ProcessOutput(media_object, 0, 1, &output_data_buffer, &status);
-    todo_wine
     ok(hr == S_FALSE, "ProcessOutput returned %#lx.\n", hr);
     ok(output_media_buffer->length == 0, "Unexpected length %#lx.\n", output_media_buffer->length);

@@ -8361,7 +8360,7 @@ static void test_wmv_decoder_media_object(void)
     output_data_buffer.rtTimestamp = 0xdeadbeef;
     output_data_buffer.rtTimelength = 0xdeadbeef;
     hr = IMediaObject_ProcessOutput(media_object, 0, 1, &output_data_buffer, &status);
-    todo_wine ok(hr == S_FALSE, "ProcessOutput returned %#lx.\n", hr);
+    ok(hr == S_FALSE, "ProcessOutput returned %#lx.\n", hr);

hr = IMediaObject_AllocateStreamingResources(media_object);
     ok(hr == S_OK, "Got hr %#lx.\n", hr);
diff --git a/dlls/winegstreamer/wg_sample.c b/dlls/winegstreamer/wg_sample.c
index 2fc2679337f..c28d5bd6a5b 100644
--- a/dlls/winegstreamer/wg_sample.c
+++ b/dlls/winegstreamer/wg_sample.c
@@ -501,6 +501,8 @@ HRESULT wg_transform_read_dmo(wg_transform_t transform, DMO_OUTPUT_DATA_BUFFER *
     {
         if (hr == MF_E_TRANSFORM_STREAM_CHANGE)
             TRACE_(mfplat)("Stream format changed.\n");
+        if (hr == MF_E_TRANSFORM_NEED_MORE_INPUT)
+            hr = S_FALSE;
         wg_sample_release(wg_sample);
         return hr;
     }

[Yuxuan Shui (＠yshui)]

Context: I was trying to make use of DMOs in `WMSyncReader`. I want to know if there is data available from a DMO decoder. Maybe there's a better way, but intuitively I call `ProcessOutput` and check if it returns `S_FALSE`.

Our DMOs return `MF_E_TRANSFORM_NEED_MORE_INPUT` which is not a documented return value for DMOs.

[Yuxuan Shui (＠yshui)]

On Thu Nov 20 08:40:37 2025 +0000, Yuxuan Shui wrote:
> Context: I was trying to make use of DMOs in `WMSyncReader`. I want to
> know if there is data available from a DMO decoder. Maybe there's a
> better way, but intuitively I call `ProcessOutput` and check if it
> returns `S_FALSE`.
> Our DMOs return `MF_E_TRANSFORM_NEED_MORE_INPUT` which is not a
> documented return value for DMOs.

Notes on the test case (on native):

- I also tried adding a test case calling `ProcessOutput` before any `ProcessInput` calls, and I got `E_POINTER`. Don't know what's going on there.
- If I call `ProcessInput` before setting an output type, subsequent `ProcessOutput` might crash. Either c0000005, or c0000374 (heap corruption). Must messed up some internal states.

[Rémi Bernon (＠rbernon)]

Rémi Bernon (@rbernon) commented about dlls/winegstreamer/wg_sample.c:
>  {
>      if (hr == MF_E_TRANSFORM_STREAM_CHANGE)
>          TRACE_(mfplat)("Stream format changed.\n");
> -
>    if (hr == MF_E_TRANSFORM_NEED_MORE_INPUT)
> -
>        hr = S_FALSE;

Several callers use SUCCEEDED(hr) after to check if a sample is returned, it will now succeed when no sample is available too and it needs to be changed.

[Yuxuan Shui (＠yshui)]

On Thu Nov 20 09:42:00 2025 +0000, Rémi Bernon wrote:
> Several callers use SUCCEEDED(hr) after to check if a sample is
> returned, it will now succeed when no sample is available too and it
> needs to be changed.

they need to be changed to `hr == S_OK` right?

[Rémi Bernon (＠rbernon)]

On Thu Nov 20 09:44:14 2025 +0000, Yuxuan Shui wrote:
> they need to be changed to `hr == S_OK` right?

That's one way, but as this function returns other MF specific return codes in other cases I think the translation from MF to DMO return code should perhaps be done in the callers.