Re: firefox52 core dump on RPI2 NetBSD9.1

[Bodie <bodie%bodie.cz@localhost>]

On 19.11.2020 09:24, Mayuresh wrote:
> On Thu, Nov 19, 2020 at 08:27:22AM +0100, Bodie wrote:
> > Based on discussion on other list you were given tips what to do,
> > did you test them?
>
> If there are any tips that I missed, please do point out.
>
> [ If you are referring to a post that says my toolchain might be 
> invalid,
> I already pointed out that I did not build anything at my end. I have
> downloaded the binary packages and binary OS image. ]

Yes that is the post 
https://mail-index.netbsd.org/port-arm/2020/11/17/msg007061.html
It is pointing out that there may something happen during build of 
binary
package on NetBSD end and how to check this is to compile it on your end
on -stable or even -current. If compile on your end from pkgsrc2020Q3
create same result then it may be worth of PR

> > Besides based on your description you picked up totally wrong platform
> > for your job.
>
> I am using this device for several years and have used firefox on it. 
> It
> used to work fine till NetBSD 8 (or 7 I don't recollect). I can make do
> with 1 tab and when not needed, can switch off js.
>
> Of late I was using Raspbian on the device and wanted to switch back to
> the latest NetBSD version.

And did you notice during those years how browsers requirements raised
regarding HW and got thin regarding range of supported OSs?

> > If this is Severity - critical for you and you are stuck without 
> > browser
> > completely then you did something seriously wrong in planning phase of
> > whatever project you're in, because it means you do not have backup
> > platform or better to say more suited platform for the task at hands 
> > ;-)
>
> Not really, both the points. As I said I have been using NetBSD on this
> device and I have a fallback option to Raspbian.

So you have fallback, even more reasons to not report as critical. One
package issues is very different from eg. port not booting at all

> Basically, I myself had reported the .strtab issues long back in 8.99
> - I think that was in the context of squid probably. I got to know in 
> this
> thread that those have been fixed. So it's a bit surprising that they
> are recurring. And they are occurring with a binary download of both 
> the
> OS image and packages. Hence I thought they should be wider interest.

Easy answer. Lack of resources. It is easy to find something wrong, it 
is
harder to get hands dirty in code. For me eg. issues with freerdp2 on 
amd64.
I am fully aware there may not be quick fix and requires a lot of 
compile,
reading and such from my side, but it is part of the fun.

> I'd request inputs on the .strtab issue and why it occurs on all binary
> download with nothing built at my end. Whether NetBSD suits my device 
> and
> whether the device suits firefox - let me experiment. May be it proves
> to be unsuitable but let's take one issue at a time.

But do you know for sure that those binary downloads are correct? That
something from pkgsrc is able to compile and create binary package which
you can start does not mean it is running as it should ;-)

Eg. my Orange Pi One with 9.1 -stable

Memory: 211M Act, 134M Inact, 9784K Wired, 13M Exec, 320M File, 68M Free

with just tmux and 2 sessions (File part is occupied after some pkgin 
runs)
I know for 100% that this platform is not suitable for any modern FF or
Webkit/Webengine variant even if there will be more RAM. It is learning
or small server/IoT platform playground. For web with eg. links and 
such.

I know will be nice to have modern platform with available HW where one
can at least get close in responsiveness of GUI of 80's/90's non-i386
computers, but that approach will turn industry upside down :-)