|
| 1 | +2011-03-14 Geoffrey Garen <ggaren@apple.com> |
| 2 | + |
| 3 | + Reviewed by Oliver Hunt. |
| 4 | + |
| 5 | + Made JSWrapperObject and subclasses moving-GC-safe |
| 6 | + https://bugs.webkit.org/show_bug.cgi?id=56346 |
| 7 | + |
| 8 | + SunSpider reports no change. |
| 9 | + |
| 10 | + * runtime/BooleanObject.cpp: |
| 11 | + (JSC::BooleanObject::BooleanObject): |
| 12 | + * runtime/DateInstance.cpp: |
| 13 | + (JSC::DateInstance::DateInstance): No more need for JSGlobalData, since |
| 14 | + we don't initialize the wrapped value in our constructor. |
| 15 | + |
| 16 | + * runtime/DateInstance.h: Don't set the OverridesMarkChildren flag because |
| 17 | + we do not in fact override markChildren. |
| 18 | + |
| 19 | + * runtime/DatePrototype.h: Declare an anonymous slot, since wrapper object |
| 20 | + no longer does so for us. Also added an ASSERT to catch a latent bug, |
| 21 | + where DatePrototype stomped on its base class's anonymous slot. Hard-coded |
| 22 | + anonymous slots are a plague on our code. This doesn't cause any problems |
| 23 | + in our existing code since the base class never reads the anonymous slot |
| 24 | + it declares, but it caused crashes when I tried to start using the slot |
| 25 | + in an initial version of this patch. |
| 26 | + |
| 27 | + * runtime/JSWrapperObject.h: |
| 28 | + (JSC::JSWrapperObject::JSWrapperObject): |
| 29 | + (JSC::JSWrapperObject::internalValue): |
| 30 | + (JSC::JSWrapperObject::setInternalValue): Resolved a problem where |
| 31 | + our internal value was stored in two places: an anonymous slot, and a |
| 32 | + data member which was not always visited during GC. Now, we only use the |
| 33 | + data member, and we always visit it. (Instead of relying on certain |
| 34 | + subclasses to set the OverridesMarkChildren bit, we set it ourselves.) |
| 35 | + |
| 36 | + * runtime/NumberObject.cpp: |
| 37 | + (JSC::NumberObject::NumberObject): No more need for JSGlobalData, since |
| 38 | + we don't initialize the wrapped value in our constructor. |
| 39 | + |
| 40 | + * runtime/NumberObject.h: Removed meaningless declaration. |
| 41 | + |
| 42 | + * runtime/StringObject.cpp: |
| 43 | + (JSC::StringObject::StringObject): No more need for JSGlobalData, since |
| 44 | + we don't initialize the wrapped value in our constructor. |
| 45 | + |
| 46 | + * runtime/StringObject.h: Don't set the OverridesMarkChildren flag because |
| 47 | + we do not in fact override markChildren. |
| 48 | + |
| 49 | + * runtime/StringPrototype.h: Declare an anonymous slot, since wrapper object |
| 50 | + no longer does so for us. Also added an ASSERT to catch a latent bug, |
| 51 | + where DatePrototype stomped on its base class's anonymous slot. Hard-coded |
| 52 | + anonymous slots are a plague on our code. |
| 53 | + |
1 | 54 | 2011-03-14 Michael Saboff <msaboff@apple.com> |
2 | 55 |
|
3 | 56 | Reviewed by Gavin Barraclough. |
|
0 commit comments