No bug - Bump the Windows stack frame size. r=shu
authorRyan VanderMeulen <ryanvm@gmail.com>
Fri, 15 May 2015 18:17:13 -0400
changeset 265191 a28dbf1faa66289c1cab95153581b9abd8d77d7a
parent 265190 a6bf6230c4a57a61468d57fa4279de3710a7fcd1
child 265192 bd6fb3d49f402a476530545386c06671e08befda
push id2102
push usermartin.thomson@gmail.com
push dateMon, 18 May 2015 17:35:37 +0000
reviewersshu
milestone41.0a1
No bug - Bump the Windows stack frame size. r=shu
js/xpconnect/src/XPCJSRuntime.cpp
--- a/js/xpconnect/src/XPCJSRuntime.cpp
+++ b/js/xpconnect/src/XPCJSRuntime.cpp
@@ -3389,18 +3389,18 @@ XPCJSRuntime::XPCJSRuntime(nsXPConnect* 
     // On normal builds, the largest stack frame size we might encounter is
     // 9.0k (see above), so let's use a buffer of 9.0 * 3 * 10 = 270k.
     const size_t kStackQuota =  2 * kDefaultStackQuota;
     const size_t kTrustedScriptBuffer = 270 * 1024;
 #elif defined(XP_WIN)
     // 1MB is the default stack size on Windows, so use 900k.
     // Windows PGO stack frames have unfortunately gotten pretty large lately. :-(
     const size_t kStackQuota = 900 * 1024;
-    const size_t kTrustedScriptBuffer = (sizeof(size_t) == 8) ? 160 * 1024
-                                                              : 100 * 1024;
+    const size_t kTrustedScriptBuffer = (sizeof(size_t) == 8) ? 180 * 1024   //win64
+                                                              : 120 * 1024;  //win32
     // The following two configurations are linux-only. Given the numbers above,
     // we use 50k and 100k trusted buffers on 32-bit and 64-bit respectively.
 #elif defined(DEBUG)
     // Bug 803182: account for the 4x difference in the size of js::Interpret
     // between optimized and debug builds.
     // XXXbholley - Then why do we only account for 2x of difference?
     const size_t kStackQuota = 2 * kDefaultStackQuota;
     const size_t kTrustedScriptBuffer = sizeof(size_t) * 12800;