920294 - Fix DumpJSStack

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Tom S]

Attachment: dumpstack

I fixed paramters and kind of fixed locals. For some reason all the values are undefined, is that because of an optimization?

Comment 1

[Tom S]

So I don't really remember what Jan told me, but I think it boiled down to that supporting this for the JITs needs some work. Also I think he recommended using something different from debug scopes. Jan would you mind clarifying this again? :)

Comment 2

[Jan de Mooij [:jandem]]

Comment on attachment 809517 [details] [diff] [review]
dumpstack

Review of attachment 809517 [details] [diff] [review]:
-----------------------------------------------------------------

(In reply to Tom Schuster [:evilpie] from comment #1)
> So I don't really remember what Jan told me, but I think it boiled down to
> that supporting this for the JITs needs some work. Also I think he
> recommended using something different from debug scopes. Jan would you mind
> clarifying this again? :)

Yeah I wouldn't use GetDebugScopeForFrame here. I think you can just add a ScriptFrameIter::unaliasedLocal method like the unaliasedActual method you're already using.

Note that these methods work fine for Interpreter and Baseline frames but will assert for Ion frames I think, so please make sure you're handling that correctly (the simplest thing to do is to not print args/locals if iter.isIon()).

Comment 3

[Tom S]

Attachment: wip v2

Made some progress here, I think right now I am missing something like AliasedFormalIter but for formals.

Comment 4

[Tom S]

Oh looks like the locals are wrong even for simple functions, need to investigate this more.

Comment 5

[Tom S]

Maybe we can just punt on locals and fix the arguments at least?

Comment 6

[Tom S]

Attachment: v1

Comment 7

[Tom S]

http://hg.mozilla.org/integration/mozilla-inbound/rev/1268bd66a094

Comment 8

[Phil Ringnalda (:philor)]

https://hg.mozilla.org/mozilla-central/rev/1268bd66a094