500233 - make some DEBUG_CC information accessible to memory tools

Status: RESOLVED FIXED

(Core :: XPCOM, defect, P3)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I'd like to be able to use the cycle collection API for memory tools that run in normal Firefox builds (tools that are for advanced web developers to use).  This requires changing some of our DEBUG_CC ifdefs to be conditionally-executed code rather than conditionally-compiled code.

This patch changes two chunks of code away from DEBUG_CC ifdefs:
 * code that gives the object size, object name, and edge names
 * code that prevents us from tracing the entire graph when we know we don't need to
and makes them governed by flags on the nsCycleCollectionTraversalCallback instead.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: patch

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: patch (diff -w)

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 384907 [details] [diff] [review]
patch (diff -w)

I think there's something in this patch causing leaks, so never mind for now.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: patch

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: patch (diff -w)

Comment 6

[Peter Van der Beken [:peterv]]

Comment on attachment 385154 [details] [diff] [review]
patch (diff -w)

>+    return !(cb.Flags() &
>+             nsCycleCollectionTraversalCallback::WANT_ALL_TRACES) &&

Would it make sense to have an inline helper function so that this could just be cb.WantAllTraces()?
(Same thing for WANT_DEBUG_INFO)

>+    // If the callback cares about objsz or objname, it should
>+    // WANT_DEBUG_INFO in the flags.

should "return"?

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

http://hg.mozilla.org/mozilla-central/rev/e277a7a7ad4f