[PATCH 0/1] MR8939: winegstreamer: Free stream buffers before wg_parser_disconnect.

[Yuxuan Shui (＠yshui)]

Renaming destroy_stream to free_stream_buffers because that's what it does.

wg_parser_disconnect frees all streams on the wg_parser, but free_stream_buffers
later calls wg_parser_stream_release_buffer on those streams. It has saved
pointers to those in wm_stream->wg_stream, which became dangling pointers after
wg_parser_disconnect.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939

[PATCH 1/1] winegstreamer: Free stream buffers before wg_parser_disconnect.

[Yuxuan Shui]

From: Yuxuan Shui <yshui(a)codeweavers.com>

Renaming destroy_stream to free_stream_buffers because that's what it does.

wg_parser_disconnect frees all streams on the wg_parser, but free_stream_buffers
later calls wg_parser_stream_release_buffer on those streams. It has saved
pointers to those in wm_stream->wg_stream, which became dangling pointers after
wg_parser_disconnect.
---
 dlls/winegstreamer/wm_reader.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/dlls/winegstreamer/wm_reader.c b/dlls/winegstreamer/wm_reader.c
index 8507d35233a..467662b964c 100644
--- a/dlls/winegstreamer/wm_reader.c
+++ b/dlls/winegstreamer/wm_reader.c
@@ -1454,7 +1454,7 @@ static const IWMReaderTimecodeVtbl timecode_vtbl =
     timecode_GetTimecodeRangeBounds,
 };
 
-static void destroy_stream(struct wm_reader *reader)
+static void free_stream_buffers(struct wm_reader *reader)
 {
     unsigned int i;
 
@@ -1587,6 +1587,9 @@ static HRESULT reinit_stream(struct wm_reader *reader, bool read_compressed)
     WORD i;
 
     ReleaseSemaphore(reader->read_sem, 1, NULL);
+
+    free_stream_buffers(reader);
+
     wg_parser_disconnect(reader->wg_parser);
 
     EnterCriticalSection(&reader->shutdown_cs);
@@ -1598,7 +1601,6 @@ static HRESULT reinit_stream(struct wm_reader *reader, bool read_compressed)
     reader->read_thread = NULL;
     reader->read_sem = NULL;
 
-    destroy_stream(reader);
     wg_parser_destroy(reader->wg_parser);
     reader->wg_parser = 0;
 
@@ -1665,7 +1667,7 @@ out_destroy_parser:
         CloseHandle(reader->read_sem);
         reader->read_sem = NULL;
     }
-    destroy_stream(reader);
+    free_stream_buffers(reader);
     wg_parser_destroy(reader->wg_parser);
     reader->wg_parser = 0;
 
@@ -1945,6 +1947,9 @@ static HRESULT WINAPI reader_Close(IWMSyncReader2 *iface)
     }
 
     ReleaseSemaphore(reader->read_sem, 1, NULL);
+
+    free_stream_buffers(reader);
+
     wg_parser_disconnect(reader->wg_parser);
 
     EnterCriticalSection(&reader->shutdown_cs);
@@ -1956,7 +1961,6 @@ static HRESULT WINAPI reader_Close(IWMSyncReader2 *iface)
     reader->read_thread = NULL;
     reader->read_sem = NULL;
 
-    destroy_stream(reader);
     wg_parser_destroy(reader->wg_parser);
     reader->wg_parser = 0;
 
-- 
GitLab

https://gitlab.winehq.org/wine/wine/-/merge_requests/8939

[Yuxuan Shui (＠yshui)]

(not discovered by ASan this time, since it's from unix heap which is not covered yet. found by @alexhenrie)

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939#note_115435

[PATCH 0/1] MR8939: winegstreamer: Free stream buffers before wg_parser_disconnect. - approved

[Alex Henrie (＠alexhenrie)]

This merge request was approved by Alex Henrie.
-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939

[PATCH 0/1] MR8939: winegstreamer: Free stream buffers before wg_parser_disconnect. - approved

[Elizabeth Figura (＠zfigura)]

This merge request was approved by Elizabeth Figura.
-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939

[Rémi Bernon]

Rémi Bernon (@rbernon) commented about dlls/winegstreamer/wm_reader.c:
> CloseHandle(reader->read_sem);
>          reader->read_sem = NULL;
>      }
> -    destroy_stream(reader);
> +    free_stream_buffers(reader);

I doesn't seem to be needed here, is it? Or if it does, it should also be added to init_stream don't you think?

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939#note_115690

[Elizabeth Figura (＠zfigura)]

On Thu Sep 11 17:51:08 2025 +0000, Rémi Bernon wrote:
> I doesn't seem to be needed here, is it? Or if it does, it should also
> be added to init_stream don't you think?
> Yes, it should be removed here. I won't call that necessary for this patch series, although it'd certainly be appreciated.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/8939#note_115693