Re: Firefox52 broken on NetBSD9.3 / pkgsrc-2022Q4

[Alexander Schreiber <als%thangorodrim.ch@localhost>]

On Fri, Jan 20, 2023 at 05:10:52PM +0100, Connor McLaughlan wrote:
> On Tue, Jan 17, 2023 at 8:00 PM Alexander Schreiber <als%thangorodrim.ch@localhost> wrote:
> >
> > On Tue, Jan 17, 2023 at 05:06:49PM +0100, Connor McLaughlan wrote:
> > > Hi All,
> > >
> > > i reinstalled a sparc64 system with NetBSD9.3 and have built firefox52
> > > from pkgsrc-2022Q4:
> > > firefox52-52.9.0nb50.tgz
> > >
> > > It will bus error shortly after start, throwing this in the console:
> > >
> > > localhost$ firefox52
> > > [Child 4991] ###!!! ABORT: Aborting on channel error.: file
> > > /usr/pkgsrc/www/firefox52/work/firefox-52.9.0esr/ipc/glue/MessageChannel.cpp,
> > > line 2152
> > > [Child 4991] ###!!! ABORT: Aborting on channel error.: file
> > > /usr/pkgsrc/www/firefox52/work/firefox-52.9.0esr/ipc/glue/MessageChannel.cpp,
> > > line 2152
> > > [1]   Bus error (core dumped) firefox52
> > > localhost$
> > >
> > > Is it working for anyone else with the above system config?
> >
> > I can try, but the build is going to take a while (UltraSPARC-IIe @ 548 MHz
> > so no speed daemon despite the SSDs.
> >
> Thank you for your effort, but i think this will lead to the same outcome.

Yup.

It started up, I selected the profile to run but before the main app
window came up it died with SIGBUS too:

ERROR - ContextualIdentityService - Unknown version found in /home/users/als/.firefox/default/hv2xpsly.slt/containers.json
Bus error (core dumped)
als@telperion:~$ file ^C
als@telperion:~$ l *core*
-rw-------  1 als  users  86369896 Jan 20 22:48 firefox.core
als@telperion:~$ file firefox.core 
firefox.core: ELF 64-bit MSB core file, SPARC V9, total store ordering, version 1 (SYSV), NetBSD-style, from 'firefox', pid=26385, uid=1069, gid=100, nlwps=36, lwp=1 (signal 10/code 32767)
als@telperion:~$ find /usr/pkg -name firefox
/usr/pkg/lib/firefox52/firefox
als@telperion:~$ gdb /usr/pkg/lib/firefox52/firefox firefox.core 
GNU gdb (GDB) 8.3
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "sparc64--netbsd".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
    <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from /usr/pkg/lib/firefox52/firefox...
(No debugging symbols found in /usr/pkg/lib/firefox52/firefox)
[New process 1]
[New process 44]
[New process 43]
[New process 42]
[New process 41]
[New process 40]
[New process 39]
[New process 38]
[New process 37]
[New process 36]
[New process 35]
[New process 34]
[New process 33]
[New process 32]
[New process 31]
[New process 30]
[New process 29]
[New process 28]
[New process 27]
[New process 26]
[New process 25]
[New process 24]
[New process 23]
[New process 21]
[New process 20]
[New process 12]
[New process 11]
[New process 10]
[New process 9]
[New process 8]
[New process 7]
[New process 6]
[New process 5]
[New process 4]
[New process 3]
[New process 2]
Core was generated by `firefox'.
Program terminated with signal SIGBUS, Bus error.
#0  0x000000004cb09ec4 in ?? () from /usr/pkg/lib/firefox52/libxul.so
[Current thread is 1 (process 1)]
(gdb) bt
#0  0x000000004cb09ec4 in ?? () from /usr/pkg/lib/firefox52/libxul.so
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) 


> I have also tried the package from
> https://pkgsrc.zia.io/pub/pkgsrc/packages/NetBSD/sparc64/9.0_2022Q4/All/
> with the same crash result.
> 
> After that  i have bootstrapped pkgsrc-2020Q4 into a different
> directory and built the complete dependency chain to build
> firefox52-52.9.0nb32.tgz
> This firefox52 works and tells me that the base system of NetBSD 9.3
> is not the root cause.
> 
> Unfortunately i don't know how to debug and identify the problem further.

So SIGBUS on an architecture like sparc64 (which has _strong_ opinions
on load/store alignment) usually carries a very strong stink of unaligned
memory access (on x86, that will merely slow you down, on sparc64 the
CPU tells you "that ain't happening, mate").

Next step is to build firefox with actual debug symbols so one can check
exactly where that happens. Although I'm somewhat suspicious, given:

als@telperion:~$  ls -hal  /usr/pkg/lib/firefox52/firefox firefox.core 
-rwxr-xr-x  1 root  wheel  160K Jan 20 11:03 /usr/pkg/lib/firefox52/firefox
-rw-------  1 als   users   82M Jan 20 22:48 firefox.core
als@telperion:~$ file  /usr/pkg/lib/firefox52/firefox firefox.core 
/usr/pkg/lib/firefox52/firefox: ELF 64-bit MSB pie executable, SPARC V9, relaxed memory ordering, version 1 (SYSV), dynamically linked, interpreter /usr/libexec/ld.elf_so, for NetBSD 9.3, BuildID[sha1]=61ced02640d06298a232a68c8972babab9e0474d, compiler model: medmid, stripped
firefox.core:                   ELF 64-bit MSB core file, SPARC V9, total store ordering, version 1 (SYSV), NetBSD-style, from 'firefox', pid=26385, uid=1069, gid=100, nlwps=36, lwp=1 (signal 10/code 32767)

I have no idea about the system architecture of firefox, but I'm blindly
guessing that the small executable is merely a loading shell for a pile
of dynamic libraries that actually spin up firefox.

Kind regards,
            Alex.

-- 
"Opportunity is missed by most people because it is dressed in overalls and
 looks like work."                                      -- Thomas A. Edison