From 67c401047a472d579cffe629519090ea8ef5c570 Mon Sep 17 00:00:00 2001
From: Nat Goodspeed <nat@lindenlab.com>
Date: Tue, 6 Sep 2016 20:48:16 -0400
Subject: [PATCH] MAINT-5011: Ensure BlockTimer::mStartTime is unconditionally
set.
Previous logic could possibly leave mStartTime uninitialized, producing fatal
warnings with gcc 4.7.
---
indra/llcommon/llfasttimer.h | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/indra/llcommon/llfasttimer.h b/indra/llcommon/llfasttimer.h
index 23702530784..f56e5596f54 100644
--- a/indra/llcommon/llfasttimer.h
+++ b/indra/llcommon/llfasttimer.h
@@ -296,7 +296,16 @@ LL_FORCE_INLINE BlockTimer::BlockTimer(BlockTimerStatHandle& timer)
{
#if LL_FAST_TIMER_ON
BlockTimerStackRecord* cur_timer_data = LLThreadLocalSingletonPointer<BlockTimerStackRecord>::getInstance();
- if (!cur_timer_data) return;
+ if (!cur_timer_data)
+ {
+ // How likely is it that
+ // LLThreadLocalSingletonPointer<T>::getInstance() will return NULL?
+ // Even without researching, what we can say is that if we exit
+ // without setting mStartTime at all, gcc 4.7 produces (fatal)
+ // warnings about a possibly-uninitialized data member.
+ mStartTime = 0;
+ return;
+ }
TimeBlockAccumulator& accumulator = timer.getCurrentAccumulator();
accumulator.mActiveCount++;
// keep current parent as long as it is active when we are
--
GitLab