strange behavior of gecko-based browsers, AMD64, 3.9

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

strange behavior of gecko-based browsers, AMD64, 3.9

Travers Buda-2
I've been experiencing slow rendering (mostly with large images) and random
crashing with the gecko-based browsers on 3.9 on AMD64. Same browsers on 64
bit linux behave properly and are snappy. The network is not at fault for
the slow rendering. I've been unable to stick my finger on what is causing
the slow rendering (common) as compared to the crashing (about once every 25
minutes or so.) A page will load correctly in one instance, and then loaded
again can crash the browser. No errors are dumped in the term.

I suspect this is some conflict with Gecko-20060303 and mmap'ed malloc (only
grounds for this is a vanilla linux which is a whole different ball park.) I
don't have a 3.8 box handy to see what verion of Gecko those era browsers
were using. But its a fair guess to say it was not Gecko-20060303 =).

The offending browsers which I have used are mozilla-1.7.12p11,
galeon-1.3.21p3 (galeon only renders slowly, no crashes,)
mozilla-firefox-1.5.0.3, and mozilla-firefox-1.5.0.2, all of which are using
Gecko-20060303.

Has anyone else seen this or am I just crazy?

As always, lynx works like a charm, sans the fact there are no pretty
pictures. =(

Travers
Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Jim Capozzoli
Travers Buda wrote:

> I've been experiencing slow rendering (mostly with large images) and random
> crashing with the gecko-based browsers on 3.9 on AMD64. Same browsers on 64
> bit linux behave properly and are snappy. The network is not at fault for
> the slow rendering. I've been unable to stick my finger on what is causing
> the slow rendering (common) as compared to the crashing (about once
> every 25
> minutes or so.) A page will load correctly in one instance, and then loaded
> again can crash the browser. No errors are dumped in the term.
>
> I suspect this is some conflict with Gecko-20060303 and mmap'ed malloc
> (only
> grounds for this is a vanilla linux which is a whole different ball
> park.) I
> don't have a 3.8 box handy to see what verion of Gecko those era browsers
> were using. But its a fair guess to say it was not Gecko-20060303 =).
>
> The offending browsers which I have used are mozilla-1.7.12p11,
> galeon-1.3.21p3 (galeon only renders slowly, no crashes,)
> mozilla-firefox-1.5.0.3, and mozilla-firefox-1.5.0.2, all of which are
> using
> Gecko-20060303.
>
> Has anyone else seen this or am I just crazy?
>
> As always, lynx works like a charm, sans the fact there are no pretty
> pictures. =(
>
> Travers
>
I use openbsd/amd64 3.9 as a "desktop" box, and I don't experience any
of your problems with mozilla-firefox-1.5.0.1.

-Jim

Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Travers Buda-2
I know this sounds a bit nuts, but I've only gotten the crashing with fvwm,
2.2.5; the version that OpenBSD ships with and has not been updated in 5
years. (I'm not blaming anyone, I realize the scale and resources of this
project.) I noticed a more current verion is in ports, I'm going to look at
that and their changelogs.

Meanwhile, could you please test with 2.2.5 fvwm? If someone can duplicate
what I'm getting, then I'll see what I can do about tracking down atleast
this bug.



On 5/8/06, Jim Capozzoli <[hidden email]> wrote:

>
> Travers Buda wrote:
> > I've been experiencing slow rendering (mostly with large images) and
> random
> > crashing with the gecko-based browsers on 3.9 on AMD64. Same browsers on
> 64
> > bit linux behave properly and are snappy. The network is not at fault
> for
> > the slow rendering. I've been unable to stick my finger on what is
> causing
> > the slow rendering (common) as compared to the crashing (about once
> > every 25
> > minutes or so.) A page will load correctly in one instance, and then
> loaded
> > again can crash the browser. No errors are dumped in the term.
> >
> > I suspect this is some conflict with Gecko-20060303 and mmap'ed malloc
> > (only
> > grounds for this is a vanilla linux which is a whole different ball
> > park.) I
> > don't have a 3.8 box handy to see what verion of Gecko those era
> browsers
> > were using. But its a fair guess to say it was not Gecko-20060303 =).
> >
> > The offending browsers which I have used are mozilla-1.7.12p11,
> > galeon-1.3.21p3 (galeon only renders slowly, no crashes,)
> > mozilla-firefox-1.5.0.3, and mozilla-firefox-1.5.0.2, all of which are
> > using
> > Gecko-20060303.
> >
> > Has anyone else seen this or am I just crazy?
> >
> > As always, lynx works like a charm, sans the fact there are no pretty
> > pictures. =(
> >
> > Travers
> >
> I use openbsd/amd64 3.9 as a "desktop" box, and I don't experience any
> of your problems with mozilla-firefox-1.5.0.1.
>
> -Jim
>
>
Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Jim Capozzoli
Travers Buda wrote:

> I know this sounds a bit nuts, but I've only gotten the crashing with fvwm,
> 2.2.5; the version that OpenBSD ships with and has not been updated in 5
> years. (I'm not blaming anyone, I realize the scale and resources of this
> project.) I noticed a more current verion is in ports, I'm going to look at
> that and their changelogs.
>
> Meanwhile, could you please test with 2.2.5 fvwm? If someone can duplicate
> what I'm getting, then I'll see what I can do about tracking down atleast
> this bug.
>
I've tested mozilla-firefox-1.5.0.1 with fvwm 2.2.5 (as root), and I
didn't notice any major speed differences, nor any crashing.  I went
around to some sites, read an article on the bbc, it all seems to run
about the same speed.  So, idk what your problem could be..

-Jim

>
>
> On 5/8/06, Jim Capozzoli <[hidden email]> wrote:
>>
>> Travers Buda wrote:
>> > I've been experiencing slow rendering (mostly with large images) and
>> random
>> > crashing with the gecko-based browsers on 3.9 on AMD64. Same
>> browsers on
>> 64
>> > bit linux behave properly and are snappy. The network is not at fault
>> for
>> > the slow rendering. I've been unable to stick my finger on what is
>> causing
>> > the slow rendering (common) as compared to the crashing (about once
>> > every 25
>> > minutes or so.) A page will load correctly in one instance, and then
>> loaded
>> > again can crash the browser. No errors are dumped in the term.
>> >
>> > I suspect this is some conflict with Gecko-20060303 and mmap'ed malloc
>> > (only
>> > grounds for this is a vanilla linux which is a whole different ball
>> > park.) I
>> > don't have a 3.8 box handy to see what verion of Gecko those era
>> browsers
>> > were using. But its a fair guess to say it was not Gecko-20060303 =).
>> >
>> > The offending browsers which I have used are mozilla-1.7.12p11,
>> > galeon-1.3.21p3 (galeon only renders slowly, no crashes,)
>> > mozilla-firefox-1.5.0.3, and mozilla-firefox-1.5.0.2, all of which are
>> > using
>> > Gecko-20060303.
>> >
>> > Has anyone else seen this or am I just crazy?
>> >
>> > As always, lynx works like a charm, sans the fact there are no pretty
>> > pictures. =(
>> >
>> > Travers
>> >
>> I use openbsd/amd64 3.9 as a "desktop" box, and I don't experience any
>> of your problems with mozilla-firefox-1.5.0.1.
>>
>> -Jim
>>
>>
>

Reply | Threaded
Open this post in threaded view
|

fvwm 2.2.5 (Re: strange behavior of gecko-based browsers, AMD64, 3.9)

Matthieu Herrb-3
In reply to this post by Travers Buda-2
Travers Buda wrote:
> I know this sounds a bit nuts, but I've only gotten the crashing with fvwm,
> 2.2.5; the version that OpenBSD ships with and has not been updated in 5
> years. (I'm not blaming anyone, I realize the scale and resources of this
> project.)

More recent fvwm versions switched to a license that's not acceptable
for the base system. We're stuck with version 2.2.5. Patches against
2.2.5 (with  valid licenses -- don't just send me a diff against the
current fvwm :-) to fix problems are welcome.

We're still looking for a modern (with support for Xft/fontconfig and
the netwm protocol at least), actively maintained, window manager  with
no external dependencies, an acceptable license and not too many bugs
(especially on sparc64).

Last time I looked at this all window manager in ports where failing
some of the criteria above. But we're still open to suggestions.
--
Matthieu

Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Christian Weisgerber
In reply to this post by Travers Buda-2
Travers Buda <[hidden email]> wrote:

> I've been experiencing slow rendering (mostly with large images) and random
> crashing with the gecko-based browsers on 3.9 on AMD64.

mozilla-firefox 1.5.0.x has been working fine for me on amd64 and
sparc64.

--
Christian "naddy" Weisgerber                          [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Travers Buda-2
Ok, thanks for the testing and replies all, it must be isolated to me. Good!

Travers
Reply | Threaded
Open this post in threaded view
|

Re: strange behavior of gecko-based browsers, AMD64, 3.9

Zvezdan Petkovic
On Wed, May 10, 2006 at 11:36:32AM -0500, Travers Buda wrote:
> Ok, thanks for the testing and replies all, it must be isolated to me. Good!

I've just remembered that a couple of years ago I've encountered similar
problem.  After the site switched to a new version of Red Hat Linux, the
fvwm users had problems.  I traced it down to LANG=en_US.UTF-8 being set
by default starting with that version of Red Hat.  Once these users
unset LANG in their .profile (.login), or changed it to simply
LANG=en_US, fvwm worked fine again.

I'm not sure this is the case with your configuration, but it's worth
checking.

Reply | Threaded
Open this post in threaded view
|

Re: fvwm 2.2.5 (Re: strange behavior of gecko-based browsers, AMD64, 3.9)

Toni Mueller-7
In reply to this post by Matthieu Herrb-3

Hello,

On Tue, 09.05.2006 at 08:17:01 +0200, Matthieu Herrb <[hidden email]> wrote:
> More recent fvwm versions switched to a license that's not acceptable
> for the base system.

I didn't look up the license for 2.2.5, but the mentioned "unacceptable
license" for current versions is GPLv2.
Just for the record.


Best,
--Toni++