Bug 1484835 Comment 14 Edit History

Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.

Original comment by
on 
I think that's a great idea to improve symmetry and testing between x86_64 / arm64, so yes.

The JIT region *almost* lives forever; technically the code is realized when JS_ShutDown() is called:
  https://searchfox.org/mozilla-central/source/js/src/vm/Initialization.cpp#211
We already don't do this if there are any live JSRuntimes (as a hack to paper over leaking situations).  I'm not sure what would go wrong if we simply never released the machine code.  It's not like we support a sequence of init->shutdown->init->shutdown (libraryInitState asserts this, even) or we'll need the memory for something else after shutdown.  Jan, can you think any reason we need to release the executable code region?
Revision 1 by
on 
I think that's a great idea to improve symmetry and testing between x86_64 / arm64, so yes.

The JIT region *almost* lives forever; technically the code is realized when JS_ShutDown() is called:
  * https://searchfox.org/mozilla-central/source/js/src/vm/Initialization.cpp#211
We already don't do this if there are any live JSRuntimes (as a hack to paper over leaking situations).  I'm not sure what would go wrong if we simply never released the machine code.  It's not like we support a sequence of init->shutdown->init->shutdown (libraryInitState asserts this, even) or we'll need the memory for something else after shutdown.  Jan, can you think any reason we need to release the executable code region?
Revision 2 by
on 
I think that's a great idea to improve symmetry and testing between x86_64 / arm64, so yes.

The JIT region *almost* lives forever; technically the code is realized when JS_ShutDown() is called:
  * https://searchfox.org/mozilla-central/source/js/src/vm/Initialization.cpp#211

We already don't do this if there are any live JSRuntimes (as a hack to paper over leaking situations).  I'm not sure what would go wrong if we simply never released the machine code.  It's not like we support a sequence of init->shutdown->init->shutdown (libraryInitState asserts this, even) or we'll need the memory for something else after shutdown.  Jan, can you think any reason we need to release the executable code region?

Back to Bug 1484835 Comment 14