tech-security archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

firefox behaviour, www.haaretz.co.il, js firefox36 dies, ghost process marked?



Hello, 

       Sometime on Fri. Dec. 28, and over in and around the weekend
        through Dec. 31, 2012 when I brought peano up and checked
        mail/news and ongoingly, I ran into a problem with firefox36
        from pkgsrc, (firefox non-3 series for some reason stopped
        working in Feb. 2012), this works, for NetBSD-6.0_(now STABLE),
        i386, on the Haaretz newspaper site (and some others,
        documented), http://www.haaretz.co.il/.  The spec needed is
        utf8, bidi with its rendering, Hebr. char set, and javascript is
        used.  Firefox36 hangs and dies with about a 80% load of the
        page, needs a wm kill (pkgsrc/fvwm but twm would work), and then
        on restart there is an error box, "There seems to be a firefox
        process already running.", with all the stuff about independent
        instances of any process/browser running under Un*x, X11 not
        being separate as per desktop windows on fvwm etc. when
        firefox36 is generally run, and its thus tricky build and
        problematic interaction with X11 (tried by hand throughout
        Dec. 2011-Jan 2012.).  Emacs-24.2.91 also died unexpectedly
        once, and I first tried src, xsrc rebuilds from
        cvs. 6.0.1-RELEASE came out, and that did not work and had
        problems, so I switched back to cvs -r netbsd-6
        (NetBSD-6.0_STABLE).  Took time.

        Upon the full rebuilds, then firefox36 would build, and work
        until I hit http://www.haaretz.co.il/, and then it died same
        problem.  no zombies  in ps -augx for any of these, ??.

        New cvs -r netbsd, rebuild, atf-run | atf-report with hard copy
        to file, no unexpected errors.

        Did a reboot, and running a fresh X, ran x11perf with no problems.

        Firefox36 tests:
        
        Rebuilt, lintpkgsrc -i, back to xulrunner192, nspr deinstall
        clean.  Built with CFLAGS += -g3 and anew, firefox36 works on
        http://www.haaretz.co.il/ w/o js until an article is selected
        for a new window and then the same hang/kill/no restart problem.
        ktrace.out is huge and seems to cycle in lines as it is run
        through kdump.  Other rebuilds of firefox36 working for a time
        w/o js enabled, including a referred site of Shlomi Fish from
        the www.hamakor.org.il site, an org.il site, which died in the same
        way with firefox36 w/o js enabled.

        Tried other browsers, lynx, links -g, dillo, emacs w3.
        All but w3 do not have java(script), all display some sort of
        text, with links -g giving he, utf-8 chars and newspaper
        pictures L2R not R2L , same for dillo, and w3 working to an
        extent if I went in and disabled w3-java, and rebuilt emacs with
        or without image files (its an editor, but to test this
        problem).

        I18n, bidi worked with some difficulty in rendering their page,
        images a problem, and with images or with js, on 53% parse of
        the page it would hang and need to C-g on the buffer or w3 in
        emacs.  Have notes from *Messages*, and without images/java last
        night, DOM parse trees from the www page.  Other sites such as
        www.acri.org.il and cs.haifa.ac.il, cs.technion.ac.il,
        www.huji.ac.il worked with bidi and emacs w3 



        program crashes due thus to js.

        Someone on the MCTO on Sat. aft. who knew about my family
        through marriage now divorced in the Jewish comm. here from
        years ago said something about how they running Windows and
        their browsers (firefox or some other) could not access Israeli
        co.il sites just as I was having trouble with www.haaretz.co.il,
        Was odd, never seen the woman before.  Reported common contexts
        not generally known for some 14 years and earlier.  Just added
        the remark in getting off the bus, picking me out of a crowd,
        about 50ish, female 3/4 mile - 1 mi. up the road from where I
        need to get off the bus.  Reports are rumored but MS with IE or
        something and such has no problem under some set of politically
        motivated community relations with computers, www sites, and
        choice of tools. ?!, given US v Microsoft, etc., not a surprise
        in a way, but helping this, and however this can be passed on to
        mozilla developers network or X11 fixes or etc. helps others
        albeit indirectly.  How true about her software/computer I don't
        know, but, a useful environmental improvement in such things in
        the end or sooner --- no control over what tools she uses.

        Seems to be a problem of the javascript on this and a couple of
        other www sites, noted where occurred, with firefox36 and
        others, is a problem with the Java VM for js on firefox36 and
        emacs w3 (defaults to hotjava, not on my systems, kaffe is
        installed, also fails), and X11, seteuid etc. and such processes
        --- the firefox still running error message when there is none,
        and no ps -augx for the firefox/xulrunner process, and refusal
        to restart anew.

        Looked into ECMAscript spec, js articles, json, etc. for
        parallel lgs with open source tools in process or such and a
        better locally secure execution possibility, and extending
        existing browsers cited above to bidi, utf8, he il especially
        but generally.  Blocked for the moment due to various conditions
        including housing, a debt repayment and other issues locally.

        Don't know how to get at what in firefox36 in its code base is
        messing up X11 and how its (spidermonkey) js, js-jit
        etc. subsystems affect this.  Someone might want to have a look.
        Don't need firefox(36 or other) per se if some other frontend as
        open source browser does the jobs I need.  Libs etc. have become
        design dependent on it or its correspondence to IE in their
        problems.  Nonetheless.



        John R. Towler
        jtowler%soncom.com@localhost
        Mon Jan  7, 2013




        peano.jtcl.org:

bash-4.2$ Xorg -version

X.Org X Server 1.10.3
Release Date: 2011-07-08
X Protocol Version 11, Revision 0
Build Operating System: NetBSD/i386  -
Current Operating System: NetBSD peano.jtcl.org 6.0_STABLE NetBSD 6.0_STABLE 
(Peano7c -r netbsd-6 1/2/2012) #0: Wed Jan  2 19:54:14 CST 2013  
root%peano.jtcl.org@localhost:/usr/obj/sys/arch/i386/compile/Peano7c i386
Build Date: 01 August 2011  01:01:00AM

Current version of pixman: 0.23.2
    Before reporting problems, check http://wiki.X.Org
    to make sure that you have the latest version.

bash-4.2$ uname -a
NetBSD peano.jtcl.org 6.0_STABLE NetBSD 6.0_STABLE (Peano7c -r netbsd-6
1/2/2012) #0: Wed Jan  2 19:54:14 CST 2013
root%peano.jtcl.org@localhost:/usr/obj/sys/arch/i386/compile/Peano7c i386




        



Home | Main Index | Thread Index | Old Index