2647 - trying to run grok page loader (dom level 0 bug)

Status: VERIFIED FIXED

(Core Graveyard :: Viewer App, defect, P2)

Description

[glynn]

Jan 22 Mac, Jan 25 win32 and Linux RedHat 5.2, optimized Seamonkey build
marked PP since crash and need to be able to run grok page loader for basic
testing

1.  Launch Seamonkey / viewer and go to internal Netscape site http://grok/tests/
page_loader and try to select "about.html" to run the top 100 sites

•  Crash trying at select of link on Mac and PC, severe disk hammering and loss
of machine access on Linux requiring ctrl C to get out.

On win98 Invalid page fault in module RAPTORHTML.DLL.  On Mac, crawl focusing on
NS_NewNavHTMLDTD, On Linux it appears to be hammering disk VM just trying to go
to the page.

Debug window on
Linux has popped out "handle_size_allocate: top level resize" when I tried to
ctrl D out, it is still hammering my hard disk, minutes later.

Mac Crawl:

 Displaying resource information:
  >   Map $02F018B8, flags $0000, file $2342 = viewer.app
   +  Map $000031F0, flags $801A, file $0003 = ¥ROM resources that override
System¥
    S Map $000032E8, flags $200D, file $0002 = System
      Map $00003148, flags $0014, file $01D8 = System Resources
      [Skipped $0035 maps belonging to font files]
 Calling chain using A6/R1 links
  Back chain  ISA  Caller
  00000000    PPC  0A3855A0
  038DFC80    PPC  0A37C1C4
  038DFC20    PPC  0A37BABC
  038DFBE0    PPC  0A1C577C  nsMacMessageSink::DispatchMenuCommand(EventRecord&,
long)+004FC
  038DFB00    PPC  0A1C5CF0  nsMacMessageSink::DispatchMenuCommand(EventRecord&,
long)+00A70
  038DFAA0    PPC  0A30F8A0  Repeater::DoRepeaters(const EventRecord&)+00030
  038DFA60    PPC  0A30BFF4  TimerPeriodical::RepeatAction(const EventRecord&)+
00144
  038DF960    PPC  0A30B664  TimerImpl::Fire()+00024
  038DF920    PPC  0A29C4E4  NS_OpenURL(nsIURL*, nsIInputStream**,
nsIStreamListener*)+0016C
  038DF8E0    PPC  0A27D798  NET_RegisterProtocolImplementation+05978
  038DF890    PPC  0A2799C8  NET_RegisterProtocolImplementation+01BA8
  038DF500    PPC  0A2961BC  TimingElapsedTimeToString+17B24
  038DF090    PPC  0A295B3C  TimingElapsedTimeToString+174A4
  038DF040    PPC  0A270088
  038DEFE0    PPC  0A29EB14  NS_ShutdownINetService+024D0
  038DEF90    PPC  09F5D888
  038DEF40    PPC  09F7A0C0  NS_NewOtherHTMLDTD(nsIDTD**)+07B8C
  038DEEE0    PPC  09F79CF8  NS_NewOtherHTMLDTD(nsIDTD**)+077C4
  038DEE90    PPC  09F79DF0  NS_NewOtherHTMLDTD(nsIDTD**)+078BC
  038DEE50    PPC  09F6DB80  NS_NewNavHTMLDTD(nsIDTD**)+009D8
  038DEE00    PPC  09F6DE30  NS_NewNavHTMLDTD(nsIDTD**)+00C88
  038DEDB0    PPC  09F78D10  NS_NewOtherHTMLDTD(nsIDTD**)+067DC
  038DED70    PPC  09F6D2E4  NS_NewNavHTMLDTD(nsIDTD**)+0013C
  038DED20    PPC  09F6E664  NS_NewNavHTMLDTD(nsIDTD**)+014BC
  038DEC60    PPC  09F6E3A4  NS_NewNavHTMLDTD(nsIDTD**)+011FC
  038DEBA0    PPC  09F70694  NS_NewNavHTMLDTD(nsIDTD**)+034EC
  038DEB40    PPC  09F701FC  NS_NewNavHTMLDTD(nsIDTD**)+03054
  038DEA30    PPC  0A011AA8  nsGlobalVariables::Release()+07604
 Return addresses on the stack
  Stack Addr  Frame Addr   ISA   Caller
   038DED78                PPC   09F78D10 NS_NewOtherHTMLDTD(nsIDTD**)+067DC
   038DED48                PPC   09F7A3D8 NS_NewOtherHTMLDTD(nsIDTD**)+07EA4
   038DED28    038DED20    PPC   09F6D2E4 NS_NewNavHTMLDTD(nsIDTD**)+0013C
   038DEC98    038DEC90    PPC   0B78353C __dla__FPv+00014
   038DEC78    038DEC70    PPC   09F6E8F4 NS_NewNavHTMLDTD(nsIDTD**)+0174C
   038DEC68    038DEC60    PPC   09F6E664 NS_NewNavHTMLDTD(nsIDTD**)+014BC
   038DEC38    038DEC30    PPC   0B783498 operator delete(void*)+00014
   038DEC28    038DEC20    PPC   09F7114C NS_NewNavHTMLDTD(nsIDTD**)+03FA4
   038DEC18    038DEC10    PPC   09F76A40 NS_NewOtherHTMLDTD(nsIDTD**)+0450C
   038DEBF8    038DEBF0    PPC   09F6F290 NS_NewNavHTMLDTD(nsIDTD**)+020E8
   038DEBD8    038DEBD0    PPC   0A08DFE0 NS_NewImageDocument(nsIDocument**)+
2AE68
   038DEBB8    038DEBB0    PPC   09F6FAE0 NS_NewNavHTMLDTD(nsIDTD**)+02938
   038DEBA8    038DEBA0    PPC   09F6E3A4 NS_NewNavHTMLDTD(nsIDTD**)+011FC
   038DEB68    038DEB60    PPC   09F6EDA8 NS_NewNavHTMLDTD(nsIDTD**)+01C00
   038DEB48    038DEB40    PPC   09F70694 NS_NewNavHTMLDTD(nsIDTD**)+034EC
   038DEB28    038DEB20    PPC   0A0122F0 nsGlobalVariables::Release()+07E4C
   038DEB08    038DEB00    PPC   09F700D4 NS_NewNavHTMLDTD(nsIDTD**)+02F2C
   038DEAEA                68K   001BFFFE
   038DEAC8    038DEAC0    PPC   09F76E30 NS_NewOtherHTMLDTD(nsIDTD**)+048FC
   038DEAB8    038DEAB0    PPC   0A011A64 nsGlobalVariables::Release()+075C0
   038DEAA8    038DEAA0    PPC   0A01016C nsGlobalVariables::Release()+05CC8
   038DEA78    038DEA70    PPC   0A010550 nsGlobalVariables::Release()+060AC
   038DEA38    038DEA30    PPC   09F701FC NS_NewNavHTMLDTD(nsIDTD**)+03054
   038DE9E8    038DE9E0    PPC   0A011AA8 nsGlobalVariables::Release()+07604
   038DE9A8    038DE9A0    PPC   0A00F354 nsGlobalVariables::Release()+04EB0
   038DE998    038DE990    PPC   0B783470 operator new(unsigned long)+00010
   038DE988    038DE980    PPC   0A010134 nsGlobalVariables::Release()+05C90
 Displaying memory from 0
  00000000  FFC1 0000 03B9 5210  001D 7FB6 001D 7FB8  ÿÁ¥¥¥1R¥¥¥¶¥¥¸
  00000010  001D 7FBA 001D 7FBC  FFC0 3378 FFC0 337A  ¥¥º¥¥*ÿÀ3xÿÀ3z

Comment 1

[Eli Goldberg]

Suggest that this is a duplicate of 2766. (2766 has more detail, boiled down HTML
from this page to crashing element.)

Comment 2

[leger]

Inserting Milestone info.

Comment 3

[leger]

Setting all current Open Critical and Major to M3

Comment 4

[chris hofmann]

greg is this fixed with a recent rickg check-in.

Comment 5

[glynn]

Cannot verify....ftp list at grok won't display in Feb22 Seamonkey win32 Viewer
and apprunner won't even browse to a URL

Comment 6

[chris hofmann]

looks like we still don't like long directory listings in apprunner or viewer
on the win32 3/11 builds.
Takes a very long time to show a long list of files if there is no index.html

If you wait long enough apprunner will finally show a listing of files
at http://grok/tests/page_loader,  then you can load about.html.

from there you can select the cycling of top 100 sites.
the window gets pretty hosed with updates and redrawing from there.

still a ways to go before apprunner or viewer can run browser buster.

There are at least two bugs that have fallen into this report.
The performance problem with showing long directory listings
and frames/meta tag/other problem that keeps the app from running
browser buster.

Comment 7

[rickg]

The crash has been verified fix by ChrisD. I timed the parsing code, and eats
the page in ~800ms. The delay appears to be in layout. I'm lowering the priority
and passing to Kipp.

Comment 8

[glynn]

March 15 Seamonkey builds - I also checked and crash seems to be fixed but
attempt to run via about.html is mixed; on windows and Mac it displays pages in
top frame on page (wrong).  Linux is in other pain so hard to tell what is
happening there yet. Do you wish qa open new bug for improper function or use
this one still?

Comment 9

[kipp]

Since it doesn't crash anymore I've changed the priority back to P2 and updated
the summary.

Comment 10

[kipp]

The primary url will load (though slowly; see bug #985 for an already open bug
on the issue); loading "about.html" works, though there is an incremental reflow
bug (see bug #2676 for an approximately equivalent bug). The final problem is
that clicking on one of the "run the tests links" gets you to a page that says:

Timer::Init() called with bogus value "25000"!  Not enabling timer.
JavaScript error: document.forms[0] has no properties
URL: http://grok/tests/page_loader/count.html LineNo: 31

Which indicates that our level 0 dom support is not sufficient for these tests
to run yet.

Comment 11

[sujay]

Chris, re-assigning some of Greg's old bugs to you.

Comment 12

[vidur (gone)]

The problems associated with this page change every time I look at it!

My current theory is that there's something not too kosher going on with
document load and unload events.

Comment 13

[leger]

Putting on [Perf] radar

Comment 14

[vidur (gone)]

The page runs as well as it ever has.

The only problem I'm seeing now is the fact that the page refreshing is not in
sync with the page counter on the top right. This is especially curious since
the page refreshing is done via a META refresh tag in the top frameset. Since
some of the refresh support will change with Necko, I'm moving this problem off
to a post-Necko timeframe.

Comment 15

[vidur (gone)]

This page finally works as it should...well mostly. It does seem like the page
switch sometimes happens before all of the page is loaded. Since there are no
more DOM Level 0 bugs associated with this page, I'm going to close this bug.
I'd recommend opening another one regarding the premature page switching against
the Necko folk.

Comment 16

[cpratt]

Seems OK enough to me using the 1999092808 build and NT.