462704 - [jsd] incorrect lineExtent when a while loop ends a method

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect, P2)

Description

[John J. Barton]

When a function ends with a while loop, jsd gives a lineExtent value that makes the last line appear to be on the while keyword. This makes the line table incorrect and these Javascript functions cannot be breakpointed or single step.

For example, this method will have the wrong line extent unless the comment is removed.
    function inner() {
      console.log("prompt returned: ",x);
      while (x) {  // bug not occurring if s/while/if/;
        var msg = 'this code is reached'; // I want a breakpoint here
        alert(msg);
        break;
      }
      // If this line is not commented, the lineExtent is correct
      // console.log("prompt returned: ",x);
    }

Comment 1

[John J. Barton]

A complete test case:

http://fbug.googlecode.com/svn/chromebug/tests/[e-mail]/content/jsd/jsdIScript/bugzilla-462704.js

The firefox extension:
http://fbug.googlecode.com/svn/chromebug/tests/[e-mail]

starts jsd and runs the test case. Tools -> Test Now or run firefox from the command line with -chrome chrome://jsdtest/content/testWebPage.html.

Comment 2

[John J. Barton]

For reference,the original bug report:
http://code.google.com/p/fbug/issues/detail?id=1239

Comment 3

[timeless]

Attachment: shell0.js

so... a xpcshell testcase is slightly easier for people to deal w/

from xpcshell:
load("shell0.js")
load("shell1.js") /* this is the testcase Test462704 */
test();

Comment 4

[timeless]

Attachment: DEBUG jsshell testcase

Comment 5

[Brendan Eich [:brendan]]

Attachment: fix

Codegen changed for loops, which made this ancient bug bite harder. We sometimes need a line number note before JSOP_STOP. This patch emits one when js_EmitTree unwinds to top level and is about to return, which for top level of global and eval code (not function code) happens after each t.l. statement or declaration, but it can't hurt -- CG_CURRENT_LINE(cg) is updated to avoid redundant notes.

/be

Comment 6

[Jeff Walden [:Waldo]]

Comment on attachment 378763 [details] [diff] [review]
fix

Looks good to me, as far as I understand what's going on.

Comment 7

[Brendan Eich [:brendan]]

http://hg.mozilla.org/tracemonkey/rev/d8316d783668

/be

Comment 8

[Robert Sayre]

http://hg.mozilla.org/mozilla-central/rev/d8316d783668

Comment 9

[Robert Sayre]

http://hg.mozilla.org/releases/mozilla-1.9.1/rev/c34912ba75e4