mewin/glslang
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Don't release global mutex before initialize/finalize is done

Browse Source 
Otherwise this can race with other threads for access to the fields it's
supposed to be initializing/finalizing.

For example, imagine another thread is calling ShInitialize() while the
first thread is calling ShFinalize() - the finalize function would
destroy the state at the same time as the initialize function is trying
to initialize it.

Holding on to the global lock for the entire function prevents all of
these failure modes.
This commit is contained in:
Niklas Haas 2021-11-08 14:58:50 +01:00
parent 3971424207
commit 4302d51868
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
glslang/MachineIndependent/ShaderLang.cpp
View File

@ -1343,7 +1343,6 @@ int ShInitialize()
glslang::GetGlobalLock();
++NumberOfClients;
glslang::ReleaseGlobalLock();
if (PerProcessGPA == nullptr)
PerProcessGPA = new TPoolAllocator();
@ -1353,6 +1352,7 @@ int ShInitialize()
glslang::HlslScanContext::fillInKeywordMap();
#endif
glslang::ReleaseGlobalLock();
return 1;
}
@ -1415,9 +1415,10 @@ int ShFinalize()
--NumberOfClients;
assert(NumberOfClients >= 0);
bool finalize = NumberOfClients == 0;
glslang::ReleaseGlobalLock();
if (! finalize)
if (! finalize) {
glslang::ReleaseGlobalLock();
return 1;
}
for (int version = 0; version < VersionCount; ++version) {
for (int spvVersion = 0; spvVersion < SpvVersionCount; ++spvVersion) {
@ -1455,6 +1456,7 @@ int ShFinalize()
glslang::HlslScanContext::deleteKeywordMap();
#endif
glslang::ReleaseGlobalLock();
return 1;
}