601393 - JM: "Assertion failure: nuses <= stackDepth,"

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

Attachment: testcase

Testcase asserts js debug shell on TM changeset 0230a9e80c1f with -m at Assertion failure: nuses <= stackDepth,

Comment 1

[Sean Stangl [:sstangl]]

This is caused by that patch. The previous version, 60b9a07d3d29, does not assert.

Comment 2

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

autoBisect shows this is probably related to the following changeset:

The first bad revision is:
changeset:   54650:427282865362
user:        Bill McCloskey
date:        Wed Sep 29 13:21:36 2010 -0700
summary:     Bug 535912 - Eliminate blockChain from JSStackFrame (r=cdleary)

Comment 3

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

Attachment: fix

I think that this is the correct fix. The problem was that the return JS_TRUE statement caused us to skip the decrement of cg->emitLevel at the bottom of the function. That meant that OptimizeSpanDeps was not called when it should have been.

To avoid this in the future, I used some C++ magic to ensure that the decrement happens no matter what.

Comment 4

[Brendan Eich [:brendan]]

Comment on attachment 480780 [details] [diff] [review]
fix

>+class EmitLevelManager
>+{
>+private:
>+    JSCodeGenerator *cg;
>+    
>+public:
>+    EmitLevelManager(JSCodeGenerator *cg) : cg(cg) { cg->emitLevel++; }
>+
>+    ~EmitLevelManager() { cg->emitLevel--; }
>+};

Nice to have C++ -- we were C for so long, MUST_FLOW_THROUGH static analysis macrology and the analysis itself evolved to help cope with lack of RAII balancers. We should rip out the MUST_FLOW_... calls now, replacing where necessary with RAII.

>+    EmitLevelManager emgr(cg);

Custom dictates naming this elm, not emgr. Plus, Unix mail nerd homage.

>-    cg->emitLevel++;
>+    //cg->emitLevel++;

Remove, or move the elm decl here with a better-spaced version of this comment.

r=me with these mods.

/be

Comment 5

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

http://hg.mozilla.org/tracemonkey/rev/ca4d24b04c2c

Comment 6

[Robert Sayre]

http://hg.mozilla.org/mozilla-central/rev/ca4d24b04c2c

Comment 7

[Christian Holler (:decoder)]

Automatically extracted testcase for this bug was committed:

https://hg.mozilla.org/mozilla-central/rev/efaf8960a929