[PATCH 0/2] MR8328: win32u: Better debugging for SendMessageTimeout timeouts.

[Stefan Dösinger (＠stefan)]

I spent a week debugging an application only to find out that SendMessageTimeout timed out. These two patches should make it easier to catch such cases in the future.

[PATCH 1/2] win32u: Write a WARN if retrieve_reply fails.

[Stefan Dösinger]

From: Stefan Dösinger

I spent a week debugging an application only to find out that
SendMessageTimeout timed out. I think this situation warrants a WARN rather
than a TRACE.
---
 dlls/win32u/message.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/dlls/win32u/message.c b/dlls/win32u/message.c
index 2a824dbdab3..9567e5b125a 100644
--- a/dlls/win32u/message.c
+++ b/dlls/win32u/message.c
@@ -3585,12 +3585,20 @@ static LRESULT retrieve_reply( const struct send_message_info *info,

free( reply_data );

-    TRACE( "hwnd %p msg %x (%s) wp %lx lp %lx got reply %lx (err=%d)\n",
-           info->hwnd, info->msg, debugstr_msg_name(info->msg, info->hwnd), (long)info->wparam,
-           info->lparam, *result, status );
-
     /* MSDN states that last error is 0 on timeout, but at least NT4 returns ERROR_TIMEOUT */
-    if (status) RtlSetLastWin32Error( RtlNtStatusToDosError(status) );
+    if (status)
+    {
+        WARN( "hwnd %p msg %x (%s) wp %lx lp %lx got reply %lx (err=%d)\n",
+              info->hwnd, info->msg, debugstr_msg_name(info->msg, info->hwnd), (long)info->wparam,
+              info->lparam, *result, status );
+        RtlSetLastWin32Error( RtlNtStatusToDosError(status) );
+    }
+    else
+    {
+        TRACE( "hwnd %p msg %x (%s) wp %lx lp %lx got reply %lx (err=%d)\n",
+               info->hwnd, info->msg, debugstr_msg_name(info->msg, info->hwnd), (long)info->wparam,
+               info->lparam, *result, status );
+    }
     return !status;
 }

[PATCH 2/2] win32u: Write a FIXME if we get a timeout with SMTO_NOTIMEOUTIFNOTHUNG.

[Stefan Dösinger]

From: Stefan Dösinger

Alternatively we could write a FIXME (once or always) if the flag is
passed to SMTO, but this feels unnecessary in cases where message
replies come on time.
---
 dlls/win32u/message.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/dlls/win32u/message.c b/dlls/win32u/message.c
index 9567e5b125a..a996227016e 100644
--- a/dlls/win32u/message.c
+++ b/dlls/win32u/message.c
@@ -4532,6 +4532,10 @@ LRESULT WINAPI NtUserMessageCall( HWND hwnd, UINT msg, WPARAM wparam, LPARAM lpa
             DWORD_PTR res = 0;
             params->result = send_client_message( hwnd, msg, wparam, lparam, params->flags,
                                                   params->timeout, &res, ansi );
+            if (!params->result && (params->flags & SMTO_NOTIMEOUTIFNOTHUNG))
+            {
+                FIXME( "Timeout and SMTO_NOTIMEOUTIFNOTHUNG is not implemented\n" );
+            }
             return res;
         }