619749 - JSD: can't reach the last line when single stepping

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jan Honza Odvarko [:Honza] (always need-info? me)]

(origins of this bug comes from bug 610793)

It's not possible to reach the last line of debugged script (within a function) by stepping. The debugger is resumed before reaching the last line.

In order to reproduce this bug, please follow instructions on this page:
http://getfirebug.com/tests/content/script/singleStepping/single-stepping1.html

Tested with trace-monkey build that includes patch from bug 610793, see more here: https://bugzilla.mozilla.org/show_bug.cgi?id=610793#c42

Honza

Comment 1

[Steve Fink [:sfink] [:s:]]

Argh. I could reproduce this when using my own hacked version of an older Firebug 1.7. Unfortunately, when I updated to the latest firebug 1.7, I now hit yet another compartment mismatch (not firebug's fault), which is blocking me from digging further into this. I'm probably just off by one somewhere.

I'll work on filing the latest compartment problem.

Thanks for splitting off this bug, Honza.

Comment 2

[Steve Fink [:sfink] [:s:]]

Attachment: Fix src note scanning

Yep, it's an off by one. Or specifically, I thought the delta field on src notes was the delta to the next note. It turns out it's the delta from the *previous* note to this one, which is a little odd to my eye (what does the first note's delta mean? It's not always zero!)

This patch should fix the observed issue, and should also fix some cases where lines would be skipped over when single-stepping. (I haven't observed any, but I haven't played with it much either.)

Comment 3

[Brendan Eich [:brendan]]

(In reply to comment #2)
> Created attachment 498242 [details] [diff] [review]
> Fix src note scanning
> 
> Yep, it's an off by one. Or specifically, I thought the delta field on src
> notes was the delta to the next note. It turns out it's the delta from the
> *previous* note to this one, which is a little odd to my eye (what does the
> first note's delta mean? It's not always zero!)

The first note does not necessarily annotate the first bytecode in the script's code vector, so the delta (which may be big enough -- if not, then one or more SRC_XDELTA notes will be needed) tells bytes from start of script.

/be

Comment 4

[Steve Fink [:sfink] [:s:]]

Attachment: Fix src note scanning

(In reply to comment #3)
> (In reply to comment #2)
> > Created attachment 498242 [details] [diff] [review] [details]
> > Fix src note scanning
> > 
> > Yep, it's an off by one. Or specifically, I thought the delta field on src
> > notes was the delta to the next note. It turns out it's the delta from the
> > *previous* note to this one, which is a little odd to my eye (what does the
> > first note's delta mean? It's not always zero!)
> 
> The first note does not necessarily annotate the first bytecode in the script's
> code vector, so the delta (which may be big enough -- if not, then one or more
> SRC_XDELTA notes will be needed) tells bytes from start of script.
> 
> /be

Oops! Then sorry dmandelin, but I need to update this again to set the initial PC offset correctly.

Comment 5

[Steve Fink [:sfink] [:s:]]

http://hg.mozilla.org/tracemonkey/rev/955e406e35b1

I botched the commit message; it refers to bug 619479 when it should be this bug 619749.

Comment 6

[Steve Fink [:sfink] [:s:]]

http://hg.mozilla.org/mozilla-central/rev/955e406e35b1