[PATCH] user32: Avoid holding display_dc_section when creating display DC.

[Paul Gofman]

get_display_dc() may be locking display_dc_section at the end of
user driver initialization in LoadCursorA() called from
register_builtin_classes(). If the driver initialization initiated
in CreateDCW() goes in parallel with the initialization started
elsewhere without holding display_dc_section, the process can deadlock.

Fixes random lockup on start in Hammerting.

Signed-off-by: Paul Gofman 
---
 dlls/user32/sysparams.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/dlls/user32/sysparams.c b/dlls/user32/sysparams.c
index 1381f387e03..a620116053a 100644
--- a/dlls/user32/sysparams.c
+++ b/dlls/user32/sysparams.c
@@ -548,7 +548,18 @@ static BOOL init_entry_string( struct sysparam_entry *entry, const WCHAR *str )
 HDC get_display_dc(void)
 {
     EnterCriticalSection( &display_dc_section );
-    if (!display_dc) display_dc = CreateDCW( L"DISPLAY", NULL, NULL, NULL );
+    if (!display_dc)
+    {
+        HDC dc;
+
+        LeaveCriticalSection( &display_dc_section );
+        dc = CreateDCW( L"DISPLAY", NULL, NULL, NULL );
+        EnterCriticalSection( &display_dc_section );
+        if (display_dc)
+            DeleteDC(dc);
+        else
+            display_dc = dc;
+    }
     return display_dc;
 }

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=81936

Your paranoid android.

=== debiant (32 bit report) ===

user32:
monitor: Timeout

=== debiant (32 bit Chinese:China report) ===

user32:
monitor: Timeout

=== debiant (32 bit WoW report) ===

user32:
monitor: Timeout

=== debiant (64 bit WoW report) ===

user32:
monitor: Timeout