Failing DCHECK exposing eval origin bug

Failing DCHECK exposing eval origin bug

BUG=

[caitp (gmail), 2016-06-07 23:11:11]

caitpotter88@gmail.com changed reviewers:
+ caitpotter88@gmail.com

[caitp (gmail), 2016-06-07 23:11:12]

A fun thing I notice (from printf-debugging), is that in
`LiveEdit::CheckAndDropActivations()`, sometimes "result_elements" and
"result->elements()" stop pointing to the same thing once it returns, so there's
nothing to iterate over in liveedit.js

I'm not sure that's the "only" problem, but that seems like a good starting
point for fixing this failure

[caitp (gmail), 2016-06-07 23:38:27]

On 2016/06/07 23:11:12, caitp wrote:
> A fun thing I notice (from printf-debugging), is that in
> `LiveEdit::CheckAndDropActivations()`, sometimes "result_elements" and
> "result->elements()" stop pointing to the same thing once it returns, so
there's
> nothing to iterate over in liveedit.js
> 
> I'm not sure that's the "only" problem, but that seems like a good starting
> point for fixing this failure

eh, nevermind, that's not it.

[caitp (gmail), 2016-06-08 14:01:24]

https://codereview.chromium.org/2035643003/diff/1/test/mjsunit/harmony/debug-...
File test/mjsunit/harmony/debug-async-liveedit.js (left):

https://codereview.chromium.org/2035643003/diff/1/test/mjsunit/harmony/debug-...
test/mjsunit/harmony/debug-async-liveedit.js:108: assertThrows(function() {
patch(fun_outside, "'Cat'", "'Cobra'") },
New theory on why this is broken:

By the time we get here, we have lots of unclosed generators on the heap, but
none actually on the stack (since V8 does not keep track of async stacks
itself)... So there isn't really a way to know if this should throw or not,
theres no recoded relationship between the async call, the inside call, or the
outside call.

I'm not sure how we'd make this work, tbh, at least not on the V8 side