summaryrefslogtreecommitdiff
path: root/vcl/inc/win
diff options
context:
space:
mode:
authorJan-Marek Glogowski <glogow@fbihome.de>2018-08-24 09:14:47 +0200
committerStephan Bergmann <sbergman@redhat.com>2018-08-24 15:14:53 +0200
commitbf0a63dc93a746a008fb1656457c77de8df693ba (patch)
treeb3adb6af94881b9195809d6624735c8dbecfb742 /vcl/inc/win
parent34dff8df7b9fc46fed6e6ffbaa8b6847ad6ed1b1 (diff)
tdf#118786 WIN just assert in Yield
This reminds me - again - that Jenkins doesn't run make check. It turns out InSendMessage() also returns true, if you process a nested SendMessage in the same thread. Therefore we have to remove the SalComWndProc assert and just keep the one in the Yield call. Why? Because there seem to be no way to get the information ReplyMessage has access to, so we could detect the caller / origin of the send message and implement proper nested call checks. The alternative would be to change all call sites of SendMessage to: if ( !pSalData->mpInstance->IsMainThread() ) SendMessage(...) else SalComWndProc(...) which is the same SendMessage already does. Change-Id: I991d68a64952dc5d47ba51edd8635c9e8c46614c Reviewed-on: https://gerrit.libreoffice.org/59538 Tested-by: Jenkins Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Diffstat (limited to 'vcl/inc/win')
-rw-r--r--vcl/inc/win/salinst.h2
1 files changed, 1 insertions, 1 deletions
diff --git a/vcl/inc/win/salinst.h b/vcl/inc/win/salinst.h
index cea4aff964bc..a43d52e2f8ce 100644
--- a/vcl/inc/win/salinst.h
+++ b/vcl/inc/win/salinst.h
@@ -39,7 +39,7 @@ public:
SalYieldMutex* mpSalYieldMutex;
osl::Condition maWaitingYieldCond;
- bool mbNoYieldLock;
+ unsigned m_nNoYieldLock;
public:
WinSalInstance();