126988 - NULL pc in frame passed to debugger hook

Status: VERIFIED INVALID

(Core :: JavaScript Engine, defect)

Description

[Robert Ginda]

As we're about to call into a function or start executing a script, the debugger
hook is called with a frame containing a NULL pc.  If that pc were instead
initialized to the start of the script we were about to execute life would be
easier for the debugger.

Comment 1

[Robert Ginda]

Brendan, I vaguely recall you were going to test for a null pc as a fix for some
other debugger bug, I don't remember if that's the fix that went in, do you?

Comment 2

[Robert Ginda]

Attachment: patch

Comment 3

[Robert Ginda]

oops, meant to assigne this to myself.

Comment 4

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

Comment on attachment 70706 [details] [diff] [review]
patch

Seems reasonable. r=shaver.

Comment 5

[Brendan Eich [:brendan]]

Comment on attachment 70706 [details] [diff] [review]
patch

The debugger can figure out this state and cope.  We do not want to spend
cycles here when the interpreter will get around to setting pc (if script), and
at one point the GC needed to know that we were not yet interpreting.  I don't
think this is a good change for the core engine.  Make the debugger pay the ?:
or other random-logic cost.

/be

Comment 6

[Robert Ginda]

invalid

Comment 7

[Phil Schwartau]

Marking Verified -