OS2 World Community Forum
OS/2, eCS & ArcaOS - Technical => Applications => Topic started by: mauro on January 16, 2025, 02:07:42 pm
-
Hi all, have virtual machines both OS2 Warp 4.52 and ArcaOS 5 (virtualbox) , and recently discovered that now the email client of Seamonkey -assuming same as Thunderbird-, cause a system trap when calling the "email client" window from Seamonkey (no problem if I start a new message compose), only in Os2 W4.52, all ok in ArcaOS 5. Cannot understand the cause, SM version are same 2.42.9esr. dll issue in Warp4.52? It has been working fine since some time ago. Maybe caused by a QT5 update made for Dooble ?
Do not know.
Thank you for any help
-
Memory issue? Are you marking the DLL's to load high? Are you using smturbo? And of course, have you tried a different profile
-
Hi Dave
Memory issue? Are you marking the DLL's to load high?
believe not, I would need a refresh on how-to
Are you using smturbo?
don't know smturbo, can you advice the use?
And of course, have you tried a different profile
yes, seems not solving the issue.
Anyhow still do not understand why OK in ArcaOS5 and KO in WARP4.52 , unless ArcaOS5 itself applies highmem feature.
Thank you much
-
Easiest to get the highmem.exe and smturbo.exe from your ArcaOS install. X:\sys\apps\anturbo\smturbo.exe and x:\usr\bin\highmem.exe. Put highmem.exe on your path and smturbo.exe in your SM directory.
With SM closed, run highmem -c \path\to\seamonkey\*dll and then put smturbo.exe in your startup folder. You can run it too after the highmem command.
This will load the SM dlls into highmem and keep them there even after closing SM. This is important to work around kernel bugs in the 4.52 kernel.
And yes, ArcaOS has smturbo install (and after installing uninstall) in the SeaMonkey desktop folder. You do have to run it yourself and possibly you did.
-
made all what suggested, no improve.
Seamonkey.exe opens the browser, if start composing a new message (FILE menu) it let me do it, but goes in system trap as I try to open the Mail & Newsgroup window (WINDOWS menu) see message in the attached picture.
Repeat, it has worked OK since time ago, unfortunately I ignore after which software/installation has stopped to work properly.
I feel something related to QT5 stuff introduced for Dooble installation, but however made same on ArcaOS and no issue there
Thank you
-
Hello mauro
I'm not a SeaMonkey user, but I wanted to give it a try.
I have an ArcaOS 5.1 VM on Virtualbox. I got "SeaMonkey_2-42-9esr_P4_2022-06-12_en-US.zip (https://hobbesarchive.com/Home/Download?path=/Hobbes/pub/os2/apps/webbrowser/SeaMonkey_2-42-9esr_P4_2022-06-12_en-US.zip) ", unziped it on C:\Programs\SeaMonkey, and it run on the first try. I have Qt5 and Dooble running on it.
I'm missing to test Warp 4.52 on VirtualBox to try to replicate the issue. I will try to find the VM and test.
Regards
-
thank you Martin.
Also tried with that SM zip content, at first run makes me create a new user profile and browser starts, then nothing to do as try to open Mail & Messages folder.
The issue -guess- is directly related to Thunderbird execution as same software of Seamonkey email window
-
Hi mauro
Maybe you need to play around with some config.sys values. I use the following on an ArcaOS5.1 system to be able to run Seamonkey and OpenOffice without system memory problems:-
THREADS=512
VIRTUALADDRESSLIMIT=2560
PROCESSES=128
Have you had a read through the Seamonkey files README.OS2 and RPM_REQUIREMENTS?
Regards
Pete
-
hi Pete, applied the suggestion.
Just to resume what was the situation changed in config.sys
THREADS=3072 corrected to 512 ,
VIRTUALADDRESSLIMIT=2560 corrected to 3072 that I know being the maximum value possible
PROCESSES=... no such line found, which I created on your indication PROCESSES=128
then have given the command yum install for the prerequisites; after some packages found existing in their correct version, here what it did:
Installation:
bzip2 pentium4 1.0.8-1.oc00 netlabs-rel 46 k
libvpx-legacy-6 pentium4 1.8.2-1.oc00.L netlabs-exp 1.4 M
replacing libvpx.pentium4 1.8.2-1.oc00
Upgrade:
libcx pentium4 0.7.4-1.oc00 netlabs-rel 86 k
libjpeg-turbo pentium4 3.0.1-1.oc00 netlabs-exp 291 k
libpng pentium4 1.6.40-1.oc00 netlabs-exp 117 k
libvpx pentium4 1.13.1-1.oc00 netlabs-exp 1.5 M
zlib pentium4 1.3-1.oc00 netlabs-exp 91 k
Dependencies update:
minizip pentium4 1.3-1.oc00 netlabs-exp 25 k
================================================================================
Install 2 Packages
Upgrade 6 Packages
Furthermore, assuming that SM Mail & Messages is same as Thunderbird, I also installed the prerequisites stated for TB 45.x (effective changed mmap 0.5-1 386 to mmap 0.5-1 pentium4)
but SM Mail & Messages issue remained same as before, also with SM started in safe mode. It happens exactly as it should display the email pane folders
Thank you
-
While SeaMonkey and Thunderbird share much of the same code, strictly speaking they're different programs.
Long shot, I see you're using the W4 kernel. Perhaps try changing to the SMP kernel if it is in your 4.52 install.
Look in x:\os2\install, should be a SMP directory. You need to change the kernel, OS2LDR and DOSCALL1.DLL with the ones in the SMP directory.
DOSCALL1.DLL will need unlocking before swapping and a reboot after. For the kernel and os2ldr, you might have to change their attributes, then move them out of the way and replace both. These are all connected and need to be changed as a group before rebooting.
-
hi Dave x:\os2\install\SMP directory only contains one file called @smpdir .
Just considering -but probably I'm far from the right evaluation- that it used to work fine since some time ago, if it was a kernel issue I believe should not work from the beginning
-
Do you have (into config.sys) ?
SET NSPR_OS2_NO_HIRES_TIMER=1
-
hi Dave x:\os2\install\SMP directory only contains one file called @smpdir .
Just considering -but probably I'm far from the right evaluation- that it used to work fine since some time ago, if it was a kernel issue I believe should not work from the beginning
No UNI directory either?
The W4 kernel is missing a few things which hardly ever matter, as I said, a long shot but considering all you did was install some Qt stuff, it is weird.
Have you run chkdsk on that system? You'd have to boot to an alternative, perhaps the ArcaOS install ISO and run chkdsk x: /F:3.
What video driver are you using?
-
Do you have (into config.sys) ?
SET NSPR_OS2_NO_HIRES_TIMER=1
now tried this as well, no change.
Just to point, I have installed Thunderbird 45.8 pentium4 to see what happens, it behaves exactly the same: it let mi begin to configure a profile plus the account, then at the moment to display the folders pane, the system goes away (trap).
Also in Seamonkey browser, it perfectly works till the request to go in the email folders pane.
No problem, I will survive without.
Thanks all :)
-
Hello mauro,
...Just to point, I have installed Thunderbird 45.8 pentium4 to see what happens, it behaves exactly the same: it let mi begin to configure a profile plus the account, then at the moment to display the folders pane, the system goes away (trap).
Also in Seamonkey browser, it perfectly works till the request to go in the email folders pane.
No problem, I will survive without.
Can you provide a step-by-step instructions to replicate this?
Here is why I ask:
1) FF & TB installs here: they coexist very peacefully, and normally after a good 5 days of runtime I usually hard-kill FF because it grows to over 1G RAM consumption...re-starts brings it back and no issues
2) This last post of yours seems to indicate this issue can be replicated in my configuration as well (FF & TB), so I'd like to try this but w/o making any assumptions, thus the ask for 'step-by-step' instructions
I do have a 4.52 SMP install here, so the UNI vs SMP thing is certainly going to introduce a variable.
For what it's worth, here are the releases I'm running:
1) FF - 45.9.0
Compiler Version Compiler flags
gcc.exe 9.2.0 -Wall -Wempty-body -Wpointer-to-int-cast -Wsign-compare -Wtype-limits -Wno-unused -Wcast-align -Zomf -std=gnu99 -fgnu89-inline -fno-strict-aliasing -Zomf -fno-math-errno -pthread
c++ 9.2.0 -idirafter K:/usr/include/os2tk45 -Wall -Wempty-body -Woverloaded-virtual -Wsign-compare -Wwrite-strings -Wno-invalid-offsetof -Wcast-align -Zomf -fno-exceptions -fno-strict-aliasing -Zomf -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -fno-rtti -fno-exceptions -fno-math-errno -std=gnu++0x -pthread -DNDEBUG -DTRIMMED -g -march=pentium4 -mtune=generic -O2 -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -msse2 -fomit-frame-pointer
Configure arguments
--enable-application=browser '--enable-optimize=-march=pentium4 -mtune=generic -O2 -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -msse2' --disable-debug --enable-debug-symbols --disable-tests --enable-os2-high-mem --without-x --enable-official-branding --enable-system-cairo --enable-system-pixman --with-intl-api --with-system-icu --with-system-zlib --with-system-libvpx --prefix=/usr --disable-webrtc --disable-telemetry --disable-devtools --disable-crashreporter --disable-gamepad --disable-web-speech --disable-updater --disable-startupcache
2) TB - 45.8.0
Compiler Version Compiler flags
gcc.exe 9.2.0 -Wall -Wempty-body -Wpointer-to-int-cast -Wsign-compare -Wtype-limits -Wno-unused -Wcast-align -Zomf -std=gnu99 -fgnu89-inline -fno-strict-aliasing -Zomf -fno-math-errno -pthread
c++ 9.2.0 -idirafter K:/usr/include/os2tk45 -Wall -Wempty-body -Woverloaded-virtual -Wsign-compare -Wwrite-strings -Wno-invalid-offsetof -Wcast-align -Zomf -fno-exceptions -fno-strict-aliasing -Zomf -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -fno-rtti -fno-exceptions -fno-math-errno -std=gnu++0x -pthread -DNDEBUG -DTRIMMED -g -march=pentium4 -mtune=generic -O2 -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -msse2 -fomit-frame-pointer
Configure arguments
--enable-application=mail --prefix=/usr --libdir=/usr/lib --with-system-zlib --with-system-bz2 --enable-system-hunspell --disable-tests '--enable-optimize=-march=pentium4 -mtune=generic -O2 -flifetime-dse=1 -fno-delete-null-pointer-checks -Wclass-memaccess -msse2' --disable-mochitest --disable-installer --disable-updater --disable-debug --enable-debug-symbols --enable-system-cairo --enable-os2-high-mem --disable-crashreporter --enable-calendar --with-intl-api --with-system-icu --with-system-libvpx --with-system-png --disable-system-sqlite --with-system-jpeg --enable-release --enable-official-branding --prefix=/usr --disable-telemetry --disable-devtools --disable-crashreporter --disable-gamepad --disable-updater --disable-webrtc --enable-crypto --with-external-source-dir=Y:/work/cc45-git
-
made all what suggested, no improve.
Seamonkey.exe opens the browser, if start composing a new message (FILE menu) it let me do it, but goes in system trap as I try to open the Mail & Newsgroup window (WINDOWS menu) see message in the attached picture.
Repeat, it has worked OK since time ago, unfortunately I ignore after which software/installation has stopped to work properly.
I feel something related to QT5 stuff introduced for Dooble installation, but however made same on ArcaOS and no issue there
Thank you
If ANY DLL is marked to load high. ONLY do this on ArcaOS installaitons with recent kernels.
If a program exists when loaded high (by exiting or crashing) the change it high it will take your MCP 4.52 system down.
Another fix in the ArcaOS kernel is that when DLL are loaded high the memory is actually released.
Note to run any QT 5 or QT 6 based browser (dooble). You NEED to load the webskit DLL in high memory. Otherwhise the lower shared kernel memory is exhausted to quickly.
Roderick
-
Hi Roderick, I've been marking the Mozilla DLL's to load high for years, didn't have problems on freshly booted systems, after a few days things would go south.
-
Mauro, what color depth is your system using? Try changing it. Eg if using 16 million colours, chamge to 64K or the opposite.
It's really weird that both SM and TB are crashing the kernel at the same point. Are they both i686 builds? If not you should try the i686 builds.
-
I've read your suggestions, first of all thank you for spending your time in focusing my issue.
Will now try to apply what suggested and let you know.
Roderick please need also to know how to turn all dll eventually high-loaded in a "standard" load status , ignoring which are them.
Dariusz, not understood what you're asking me to do: having both FF and TB installed and then what action...?
-
Mauro, to remove the high memory thing from the dll's, highmem --unmark *dll
-
mauro,
...Dariusz, not understood what you're asking me to do: having both FF and TB installed and then what action...?
OK, in your original post you stated:
...recently discovered that now the email client of Seamonkey -assuming same as Thunderbird-, cause a system trap when calling the "email client" window from Seamonkey (no problem if I start a new message compose)...
I interpret that to mean:
1) you are in web page (browsing) where there is an email link
2) you click on that link and expect that the email client window would pop up, thus allowing you to create an email
3) instead of this outcome you experience a trap/failure
Is that correct?
Subsequently in your post I was responding to you stated:
...I have installed Thunderbird 45.8 pentium4 to see what happens...
...and I was commenting on the fact that I have FF & TB here, which have been pretty solid for me, so I could certainly try to reproduce your scenario, and if I fail to, we could try to chase the DLLs down given that I am on CP2 release as opposed to AOS install.
-
mauro,
I interpret that to mean:
1) you are in web page (browsing) where there is an email link
2) you click on that link and expect that the email client window would pop up, thus allowing you to create an email
3) instead of this outcome you experience a trap/failure
Is that correct?
ah, ok. more exactly:
I'm in SM window browser, I browse the web, then if I want to start a new mail message I can do it (compose message window selected from FILE menu opens, I write the message and send it). But I cannot touch the item Mail & Messages from WINDOW menu, the one which opens the email folders pane, at that point OS2 traps.
-
hi mauro,
ah, ok. more exactly:
I'm in SM window browser, I browse the web, then if I want to start a new mail message I can do it (compose message window selected from FILE menu opens, I write the message and send it). But I cannot touch the item Mail & Messages from WINDOW menu, the one which opens the email folders pane, at that point OS2 traps.
OK, I understand now...so this is all within the SM application, which I do not have installed here.
When you mentioned that you installed FF I thought I could try to help/test with my FF & TB combo, unfortunately I do not think this is possible. Maybe I misunderstood one of the posts though..sorry about the noise.
-
just discovered it is a VirtualBox (Windows 11 hosted) issue!!
In fact, have cloned same virtual machine on a Linux Mint hosted Virtualbox, and seen SM email pane working with no issue at all.
Now : what is the difference?
All settings between the two virtual machines are identical, checked one by one in all sections.
The only difference is:
Windows 11 VirtualBox is 7.1.4 r165100 (Qt6.5.3)
Linux Mint 22.1 Virtualbox is 7.0.16_Ubunbu r162802 (Qt5.15.13) .
The reason why it use to work well before, should refer to the time I upgraded the VirtulBox version in Windows. Also, cannot explain why all goes well in ArcaOS vm same Windows 11 Virtualbox.
Not easy for me finding a fix for this, unless succeed to downgrade VirtualBox in Windows.
Thank you
-
......, unless succeed to downgrade VirtualBox in Windows.
Thank you
no luck even downgrading Virtualbox x Windows, tried more than one downgrade: 7.0.22 , 7.0.16 , 6.1.50 with their extensions pack, the problem is still there
-
Mauro,
I don't have any good idea about what you changed that allowed this trap to occur. If you have a backup of the VM, you could try reapplying your recent changes until the trap returns.
The trap screen looks vaguely familiar but I've not yet located my work notes to verify this. My current suspect is a floating point exception that is mishandled by IBM's 14.106 W4 kernel. What happens is the kernel's trap 0x10 handler neglects to push a required parameter on the stack. This causes the common fault handler to get confused and decide that the exception occurred when the kernel was running in v8086 mode, which is incorrect. The result is the exception handler traps and the kernel panics with the 60000, 9084 panic id.
This defect has been corrected in the ArcaOS 14.201 W4 kernel.
-
Hi Steven, as noted here above, I see that does not seem to be a specific problem of the VM OS2 kernel itself, it works fine in a Virtualbox Linux hosted, viceversa it goes KO at that time on a Virtualbox Windows hosted. Of course I also tried to work whit a restored version of a previous OS2 VM, but it behaves the same. And have gone back with older Windows Virtualbox releases as well: same behavior. The issue could be caused by one of the Windows 11 update, which then affects OS2 VM but not ArcaOS VM.
Not really a serious situation, I've alternatives available not to stay learning IT engeneering which is away from my knoledges. It also may happen that the issue will disappears with future Windows 11 updates.
Thank you anyway
-
Hi/2.
Are you using OS2PCAT by Lars ? If not, I suggest to use it.
And if you are using PIIX4 as an ide controller, then try PIIX3.
-
Hi/2.
Are you using OS2PCAT by Lars ? If not, I suggest to use it.
And if you are using PIIX4 as an ide controller, then try PIIX3.
magic ! OS2PCAT4 was the care, so simple!
Thank you much
-
Yes, OS2PCAT.PSD fixes the exact error that Steven has described:
when the PSD loads, it changes the floating point exception handling to use the CPUs native exception handling (instead of invoking interrupt 13 via the old interrupt controller), replaces the original floating point exception handler (the handler for trap 0x10) with its own handler that subsequently pushes the missing parameter on the stack before invoking the original exception handler. That makes IBM's 14.106 W4 kernel happy (if a floating point exception occurs).
Just to be sure, it checks if the original exception handler is already pushing that additional parameter (the SMP kernels and obviously the latest AN W4 kernel already do that correctly) and skips pushing the additional parameter if it already works correctly.
In short: it's neither an error of your host OS nor of Virtualbox. It's all the W4 kernel's fault :-)
I was not aware that AN had fixed that directly in the W4 kernel for the later W4 kernels.
Lars
-
yes Lars, while it is obvious that your OS2PCAT.PSD has fixed the system trap (and thank you for the important job you did which is the only hope for people like me), I still cannot explain why the kernel failure does not happen in a linux host.
-
It is possible that your version of Virtualbox for Linux emulates treatment of floating point exceptions differently, that is, the old DOS style where a floating point exception leads to generation of an interrupt which is routed via the legacy interrupt controller and leads to invocation of an interrupt routine for interrupt 13 instead of an exception routine for exception 0x10.
And this works ok with the W4 kernel.
I seem to remember that most recent versions of Virtualbox for Windows also work this way but I am not sure. In any case, with the PSD, you are on the safe side as it enforces floating point exception handling via exception 0x10.
By the way: this is the most recent version of OS2PCAT.PSD:
https://hobbesarchive.com/Home/Download?path=/Hobbes/pub/os2/system/drivers/misc/OS2PCat_6-0.wpi