210115 - Browser is no response by clicking anywhere on the logitech homepage.

Status: RESOLVED WORKSFORME

(SeaMonkey :: General, defect)

Description

[u90614]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5a) Gecko/20030620
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5a) Gecko/20030620

Clicking anywhere on that web but the browser is no response. After that, CPU
load is 100%.

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Jim Brown]

I am seeing something similar with a trunk build this morning.
I can visit some web pages but other web pages cause mozilla to hang during the
load process. Happened on several different sites.

The hangs do not happen if I revert to a build from yesterday.
One of last nights changes mucked it up.
This affects more than Windows.

OSF1/DEC/unix

Comment 2

[Jim Brown]

Web sites which generate browser hangs:
http://www.macnn.com
http://www.theregister.co.uk

It looks like the document is fetched and then mozilla hangs during
layout/rendering (100% cpu) - GUI is totally locked up and non-responsive.

A checkin after Jun 19 19:15 is causing this regression.
I have a build pulled at Jun 19 19:15 which works fine (and an earlier build).

Builds pulled at Jun 20 10:07 and Fri Jun 20 16:42 both fail with hangs.

This should be keyword: hang dataloss

Comment 3

[Andrew Schultz]

> This should be keyword: hang dataloss

all hangs are also dataloss

this bug needs a stacktrace so it can have a component.

Comment 4

[Jim Brown]

Attachment: debugger traceback, 3 traces

I was able to reproduce the 100% cpu hang this time by 1) start mozilla with
debugger, 2) select menu QA/CVS checkins today, 3) create a new tab (ctrl-T).
As soon as I did ctrl-T mozilla was 100% cpu bound and would not come back.
I may have paused 3-5 seconds between each step.

Attached are two stack traces (by interrupting using debugger).
I got each of these traces three times across six different interrupts.
I could try a debug build to see if I get more useful information in
the traceback.

Comment 5

[Jim Brown]

Hanging builds were optimized.  A debug build has not hung yet.
Some odd looking messages in the debug output:

WARNING: NS_ENSURE_TRUE(focusedShell && doc) failed, file nsTypeAheadFind.cpp,
line 562

WARNING: NS_ENSURE_TRUE(NS_SUCCEEDED(EnsureScriptEnvironment())) failed, file
nsWebShell.cpp, line 306

WARNING: NS_ENSURE_TRUE(presContext) failed, file nsGenericHTMLElement.cpp, line
4395

Document http://bugzilla.mozilla.org/query.cgi loaded successfully
WARNING: Content has no document., file nsTextFrame.cpp, line 5372
WARNING: Reflow of frame failed in nsLineLayout, file nsLineLayout.cpp, line 1033
###!!! ASSERTION: reflow dirty lines failed: 'NS_SUCCEEDED(rv)', file
nsBlockFrame.cpp, line 850
Break: at file nsBlockFrame.cpp, line 850
nsBlockReflowContext: Block(body)(2)@1422b20a8 metrics=-559038737,-559038737!
nsBlockReflowContext: Block(body)(2)@1422b20a8 didn't set whad
-559038737,-559038737,-559038737,-559038737!
###!!! ASSERTION: reflow dirty lines failed: 'NS_SUCCEEDED(rv)', file
nsBlockFrame.cpp, line 850
Break: at file nsBlockFrame.cpp, line 850

###!!! ASSERTION: non-root frame's desired size changed during an incremental
reflow: 'first == root || (aDesiredSize.width == size.width &&
aDesiredSize.height == size.height)', file nsPresShell.cpp, line 915
Break: at file nsPresShell.cpp, line 915

Comment 6

[Jim Brown]

More testing only shows the hangs with the optimized build -- I cannot get the
debug build (of the same sources) to hang.

The hang happens when I left-click in either the (empty) main window or in the
(empty) URL window.
If I start mozilla and then immediately click in either of these windows then I
get the hang (one step).
I forgot that in step 3 I was left-clicking before I entered the text.

Comment 7

[Jim Brown]

Still cannot get a debug build to hang.
Non-debug build hangs every time I click in the URL or main window.
It also hangs if I 1) start mozilla and load QA/CVS checkins today, and 2) open
a new tab (^T) and load QA/Tree status (never finishes loading this page).
The only difference between a debug and non-debug build is this one config line
-- leave out for debug, include for non-debug build (.mozconfig).

ac_add_options --disable-debug

Comment 8

[Jim Brown]

The hangs started for me after 6/19/2003 16:15 [19:15 EDT] (last good trunk
update) and before 6/20/2003 7:07 [10:07 EDT] (first failing trunk update).
I still have the hang with a non-debug trunk build pulled on Jun 23.

Comment 9

[u90614]

Build ID: 2003062312 seems to be fixed.

Comment 10

[Andrew Schultz]

reopening.  FIXED is reserved for bugs with code patches

Comment 11

[Andrew Schultz]

Jim: it looks like what you're seeing is specific to OSF and probably not this bug

marking this WORKSFORME