775807 - --dump-bytecode can observe partially-compiled scripts which breaks JSScript::enclosingScope

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

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

Attachment: stack from Windows 7

(function() {
    const x = ((function() {
        return {
            e: function() {
                (function() {
                    for (e in x) {}
                })()
            }
        }
    }(function() {
        return {
            t: {
                c
            }
        }
    })))
})()
quit()

crashes js debug shell on m-c changeset 01929e390ba5 with -D at js::EncapsulatedPtr

Comment 1

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

autoBisect shows this is probably related to the following changeset:

The first bad revision is:
changeset:   99533:99aaaee4e6b9
user:        Luke Wagner
date:        Thu Jul 05 20:35:08 2012 -0700
summary:     Bug 753158 - emit ALIASEDVAR ops for upvars (r=bhackett)

Comment 2

[Luke Wagner [:luke]]

Ugh, JS_DumpCompartmentPCCounts is finding a script where we aborted compilation, thereby leaving an unfinished interpreted function (whose script pointer is NULL).

Comment 3

[Luke Wagner [:luke]]

Attachment: fix and test

Oh well, simple enough fix.

Comment 4

[Jim Blandy :jimb]

Comment on attachment 644393 [details] [diff] [review]
fix and test

Review of attachment 644393 [details] [diff] [review]:
-----------------------------------------------------------------

::: js/src/jsscript.cpp
@@ +1415,5 @@
> +JSScript::enclosingScriptsCompiledSuccessfully() const
> +{
> +    /*
> +     * When a nested script is succesfully compiled, it is eagerly given the
> +     * static JSFunction of its enclosing script. The enclosing script's

"The enclosing function's", right?

::: js/src/jsscript.h
@@ +643,5 @@
> +    }
> +
> +    /*
> +     * If a compile error occurs in an enclosing function after parsing a
> +     * nested function, the enclosing function's JSFunction, which is embedded

"embedded in" is a strange term to use here. "Appears on"?

Comment 5

[Luke Wagner [:luke]]

Agreed on both, thanks!

https://hg.mozilla.org/integration/mozilla-inbound/rev/1dbd25c0205e

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/1dbd25c0205e

Comment 7

[Christian Holler (:decoder)]

A testcase for this bug was automatically identified at js/src/jit-test/tests/basic/testBug775807.js.