Closed Bug 48031 Opened 24 years ago Closed 24 years ago

Crash Browser with improper removeChild

Categories

(Core :: DOM: Core & HTML, defect, P2)

Product:
Core
Component:
DOM: Core & HTML
Type:
defect

Tracking

()

Status:
VERIFIED FIXED
Milestone:
mozilla0.9

People

(Reporter: bc, Assigned: jst)

References

(
URL
)

Details

(Keywords: crash, Whiteboard: [rtm-][HAVE FIX])

Attachments

(4 files)

Crash Mozilla when removing Document.documentElement
212 bytes, text/html
Details
Crash Mozilla when replacing Document.documentElement
259 bytes, text/html
Details
Demonstrate crashing using all combinations of removeChild, replaceChild
2.74 KB, text/html
Details
Proposed fix, added missing null-ptr checks.
2.18 KB, patch
Details | Diff | Splinter Review 
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; m17) Gecko/20000807
BuildID:    2000080712

If you remove the documentElement from a document or remove an Element that has
not been inserted into a document, you will crash Mozilla.

Reproducible: Always
Steps to Reproduce:
Crash 1.   document.removeChild(document.documentElement);

Crash 2.   elm=document.createElement('a');
           document.removeChild(elm);


Actual Results:  Mozilla crashes

Expected Results:  In Crash 1, perhaps throw NO_MODIFICATION_ALLOWED_ERR, but I
am not sure.

In Crash 2, it should have thrown NOT_FOUND_ERR.
Doesn't crash in M18 on Linux.  2000-08-08-08.
Just tried with 2000-08-08-08 on Win2k. It no longer crashes on either test
case.  Crash 1 removed and returned documentElement and Crash 2 throws a
NOT_FOUND_ERR.  Should I just mark this as WORKSFORME or INVALID?
Marking WORKSFORME based on the above comments.
Status: UNCONFIRMED → RESOLVED
Closed: 24 years ago
OS: Windows NT → All
Hardware: PC → All
Resolution: --- → WORKSFORME
Adding crash keyword for fix and/or verification escalation
Keywords: crash
Mass update of qa contact
QA Contact: gerardok → janc
I was able to crash 2000101320 by removing or replacing documentElement on a
created XML document.

<html>
<head>
<script language="javascript">

var doc = document.implementation.createDocument(null, 'TEST', null);

doc.removeChild(doc.documentElement);

</script>
</head>
<body>
</body>
</html>

nsVoidArray::Count() line 45 + 3 bytes
nsDocument::RemoveChild(nsDocument * const 0x02f0d6ec, nsIDOMNode * 0x00d34090,
nsIDOMNode * * 0x0012e300) line 2906 + 15 bytes
NodeRemoveChild(JSContext * 0x030154d8, JSObject * 0x030818d8, unsigned int 1,
long * 0x02f9f108, long * 0x0012e3b4) line 545 + 25 bytes


<html>
<head>
<script language="javascript">

var doc = document.implementation.createDocument(null, 'TEST', null);
var elm = doc.createElement('NEWTEST');

doc.replaceChild(elm, doc.documentElement);

</script>
</head>
<body>
</body>
</html>


nsVoidArray::Count() line 45 + 3 bytes
nsDocument::ReplaceChild(nsDocument * const 0x02e40f8c, nsIDOMNode * 0x02e39fe8,
nsIDOMNode * 0x02cbc768, nsIDOMNode * * 0x0012e300) line 2831 + 15 bytes
NodeReplaceChild(JSContext * 0x02e16290, JSObject * 0x02d574e8, unsigned int 2,
long * 0x02ecc0c8, long * 0x0012e3b4) line 497 + 34 bytes

Attaching test cases
Status: RESOLVED → UNCONFIRMED
Resolution: WORKSFORME → ---
Status: UNCONFIRMED → ASSIGNED
Ever confirmed: true
This crash is more general than replacing or removing documentElement.

Document can have Comment, ProcessingInstruction and DocumentType children in
addition to the one Element child (documentElement).

Mozilla 2000101320 crashes when trying to remove or replace any combination of
Comment, ProcessingInstruction or DocumentType.

Attaching testcase where you can choose the means of your crash.

Nominating for rtm, this is a crasher (admittedly not a frequent one, but still
a crasher) and it's trivial to fix the problem and there's no risk involved in
fixin this, it's just a matter of doing null-ptr checks before referencing a few
pointers.
Priority: P3 → P2
Target Milestone: --- → mozilla0.6
Oy.  I think this one was my fault.  Sorry.

Patch looks fine (simple null-pointer check).  r=dbaron

Adding rtm kw since I think jst meant to.
Keywords: rtm
Marking rtm need info.  This is a low risk fix for a crasher.  If we are going
to respin another rtm candidate build, we should take this.
Whiteboard: [rtm need info]
I've looked at the proposed fix.  r=nisheeth.  Now, all we need is a super
review from Vidur.
Vidur says sr=vidur, but form talking to PDT this won't become an rtm++ bug (too
low gain), so I'm marking it rtm- and setting milestone to mozilla0.9
Whiteboard: [rtm need info] → [rtm-]
Target Milestone: mozilla0.6 → mozilla0.9
Whiteboard: [rtm-] → [rtm-][HAVE FIX]
Patch checked into the trunk, --> FIXED.
Really marking FIXED
Status: ASSIGNED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
verified that attached test cases do not crash Mozilla 2000-11-28-04 on Win2k
verified linux build 2000-11-26-21
Component: DOM Level 1 → DOM HTML
QA contact Update
QA Contact: janc → desale
Updating QA contact to Shivakiran Tummala.
QA Contact: desale → stummala
verified
Status: RESOLVED → VERIFIED
Component: DOM: HTML → DOM: Core & HTML
QA Contact: stummala → general
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: