Firefox crash recovery fails to re-populate some tabs???

[Dariusz Piatkowski]

OK, weird situation here, only because rarely, if ever, has this actually failed in the past. A couple of days ago I had a hard lock on my machine, only power off/on worked...subsequently if I kill Firefox (which I usually do if the sessions start running slow or I need to re-boot for some reason and I do not feel like properly shutting it downly only to have to re-start each tab afterwards, etc, etc) I now have firefox populating the tabs on the 1st and last browser window instantiations only...any browser windows in-between get tabs on them, but the tabs remain BLANK!

Sooo....my re-start no longer works, or to be more correct, only works in a limitted way.

I researched this, went through the clean-up of the session, restore, files...no go, it cleans up fine, but the very next time I happen to have to kill FF and re-start it, the problem comes right back. I even tried out the 'Session Master' add-on, great functionality, but it still did not help, so this leads me to believe the root cause lies somewhere deeper in the FF.

I have not pegged it yet to a certainly number of tabs and/or windows, however, the 1st and last windows always seems to hold true.

....yeah, weird...has anyone seen this before???

BTW: this is with the 45.9 pentium4 release and latest versions of libcx & pthread. I did backlevel them to the standard release, but that had no effect. FF xul.dll is set to standard load, so no highmem for code or data (tried with both scenarios, but that made no difference)....

[David Graser]

One of the things that has seemed to help me is that I lowered the default cached size from 350 MB to 100 MB.  It doesn't seem to bog down (slow down) as much after the change. 

Does ArcaOS with the mutiprocessor kernal still have that single input que (correct me with I have stated it wrong)?  In the past, this was attributed to a large amount of the lockups.

[Dariusz Piatkowski]

Hi David,

One of the things that has seemed to help me is that I lowered the default cached size from 350 MB to 100 MB.  It doesn't seem to bog down (slow down) as much after the change. 

Does ArcaOS with the mutiprocessor kernal still have that single input que (correct me with I have stated it wrong)?  In the past, this was attributed to a large amount of the lockups.

I am using HPFS386 here with a 64MB cache, what IFS are you using?

Re: SIQ, yes, that is the standard model on the OS/2 platform and re-working this, say along the lines of what MS put into NT for example, would require a significant effort that I believe can not be accomplished without having the source code available.

[David Graser]

I am using HPFS386 here with a 64MB cache, what IFS are you using?

[/quote]

Hi Dariusz

I have done several things to my system to try to help speed it up and make it possibly more stable.  I am always trying things.

I use the JFS file system.  I have always been impressed with its performance.

I created a RAM drive named "X" that uses both low and high memory using ArcaOS 5.02 in the Computer\setup folder. I have it set to HPFS.

I moved my temporary directories to my RAM drive in my config.sys.
SET TMP=x:\temp
SET TEMP=x:\temp
SET TMPDIR=x:\temp

I also operate my system with my OS2.INI and OS2SYS.INI moved to my RAM drive during operatiion using a program called UPDINI12a.zip.

http://hobbes.nmsu.edu/download/pub/os2/util/wps/updini12a.zip

rem SET USER_INI=D:\OS2\OS2.INI
SET USER_INI=X:\TEMP\OS2.INI
rem SET SYSTEM_INI=D:\OS2\OS2SYS.INI
SET SYSTEM_INI=X:\TEMP\OS2SYS.INI

RUN=D:\OS2\UPDINI.EXE D:\OS2\ 10

Because I use this program, I added the following line as the first line in my config.sys.

CALL=D:\OS2\XCOPY.EXE D:\OS2\OS*.INI D:\VAR\TEMP /T /R /O

This backs up both files during bootup to another location in cause something drastic were to happen.  I also archived my system before and after I made all these changes.

I also have made other changes such as

SWAPPATH=D:\OS2\SYSTEM 2048 24576
THREADS=1024
vme=no
SET SHELLHANDLESINC=40
Set queue_size_threshold=3000
set pm_async_focus_change=2000
SET DESKTOP=D:\DESKTOP
SET LIBCX_HIGHMEM=2

I also added the following patches

http://ca-arnold.de/download/newcalls104.zip

http://hobbes.nmsu.edu/download/pub/os2/system/patches/mixed/pipepatch.zip

My experience with my system is that, if it is going to hang, 99% percent of the time, it happens when the startup folder is executed.  Reboot and it would startup fine the next time.  It usually hangs when it says port is loading.  I use XWPs startup  folder to slow down the loading of the programs in the startup folder.  This has helped.  The other !% happens when I am using Firefox.  However, it is working great now.


I even experimented starting Firefox from the RAM drive.  This does not speed it up all all. This was just an experiment. 
This was not to be a permanent solution.

David

[David Graser]

I just found another problem.  I use the CPU temperature widget.  It normally stays around 120 degrees F.  I noticed at times that the CPU temperature takes off to around 170 to 180 degrees F.  When this happens, the computer locks up causing me to reboot or do a hard  boot. At first I thought it was the fan on my Lenovo Thinkpad X220.  However, it now appears that if my battery is fully charged and I am using my power supply, the temperature takes off.  If I unplug and operate off the battery, the temperature stays low.  The temperature stays low until the battery is fully charged and then it takes off again.  I ordered a new power supply hoping this will solve the problem.  Usually, I am in Firefox when this happens is why I originally attributed the hangs to Firefox.  This is a refurbished machine.  For being built around 2011, with it mutiprocessor core, it is fast.