I\'ve been running GSt since it came out without problems, and always pitied the poor folks that had problems. I always assumed it was due to marginal hardware or misconfiguration or something, and I\'ve always used top quality hardware and I probably know PC hardware internals as well as anyone (I ported a Unix variant to PCs 10 years ago, done lots of driver work, etc).
Anyway, my dedicated GSt box started giving me fits last week. I needed a machine to scandisk a flaky drive before sending it out for replacement, and though I have a policy to never touch my sampler PC, it was the only one suitable for this task, so against my better judgement, I swapped out the second disk (the one with samples), tested the failing drive, and returned the machine to its old configuration. During this time, I only made 1 change to the system: I disabled com2, freeing up an IRQ, since this would allow me to reduce IRQ sharing on this box.
And this broke my reliable GSt machine! About half my coldboots failed to complete, I\'d get a Windows (win98se) alert panel:
This program has performed an illegal operation and will be shut down (...)
MSG32 caused in invalid page fault in module KERNEL32.DLL (...)
On a warm boot, the machine was always fine, and if it booted it continued to work perfectly, just as it always had. I called Tascam support and spoke to Chris. He\'d heard of the problem but had no good tips. He suggested full removal of GSt and installing the latest version. No change. I installed a registry from 2 days before the problems began. No change. I scandisk\'d all my drives. No problems.
At first I didn\'t think about the com port. I used the Asus Cusl2 bios to force all my cards onto unique IRQ\'s, and put my sound cards on the highest IRQ\'s (ie 9 10, I don\'t wanna explain IRQ priorities here...).
By doing this, I could change the behaviour, but I never really fixed it. Specifically, I could change the crash from an alert panel late in the Windows boot to a BSOD earlier in the boot, but the crash was at the exact same location (still in KERNEL32.DLL, same code segment, same instruction pointer).
To me this really smells like a timing related problem. I wonder if there are some critical sections (in my case in KERNEL32.DLL, searching the archives another likely candidate is WSTREAM or something like that) and if these are reentered one of the invocations is smashed and goes off into space.
For now, I have unforced all my IRQ\'s (returned to IRQ sharing) and reenabled my com ports and so far it seems I\'m back to normal, though its too soon to tell for certain. I wonder if lowering the priority of MSG.exe (the GST kernel module) affects the problem, is that what the (undocumented) priority button in the sequencer section does?
If I were Tascam and couldn\'t get to the root of this, I\'d chase it down with an in-circuit-emulator (I rented one before to chase down driver problems related to a bad hardware implementation, they\'re a pain to set up but easy to use). In the archives here, at Yahoo, and at tascam, there are dozens of people who have this problem, and it\'s horrible that no one knows the root of it. The general advice seems to be \"futz with everything on your system and pray for the best\" and near as I can tell, the success rate for this has been low.
I used GigaStudio since came out without a problem in Win 98. One day I couldn\'t boot up any longer. Always received the dreaded msg32 error that you are getting. No changes were made to my system and the hard drive hasn\'t had any errors. Did a full surface scan of the hard drive and found no problems.
Luckily, I already setup another partition with WinXP installed on it. I was in the process of testing this to make sure all of my music apps worked properly before killing the 98 partition.
I ended up uninstalling GigaStudio in the Win 98 and have not had any problems since. Why the program just went \'south\' one day I will never know.
This may not be relevent to your case but I was able to fix the msg32 problem thing on my system.
I had giga studio working fine under win xp but then used
some of the win XP optimizing tips for audio and then had problems with starting the program.
I traced it down to
Start/run/msconfig/ok/startup,then unchecking everything that was running in the background.
I had to go back and check msg32 for giga studio to work.
I\'ve been investigating this problem and not been able to solve it yet. In my case it\'s clearly exascerbated by plug n pray configuration. This experience is really starting to annoy me. It feels like GSt can be pretty fragile, and I think Tascam needs to step up and improve a bunch of things. When it goes south, you have to wait a good long time on a toll call to support, who is very aware there\'s a problem, but doesn\'t know what to do about it, other than trashing your system, starting over from scratch, and jumping through copy protection hoops. It\'s not clear to what extent support is in contact with engineering, it\'s not clear whether engineering is well aware that many people have this problem, or that it\'s a priority to find and fix.
Alberto, the backup idea is a good one, in fact I do have a year old system image (with an older GSt) made with DriveCopy, it would be a pain to go back from that, but that was the building block for a year of solid GSt\'ing. Its a last resort, but I\'ve read plenty of stories in the archives of folks that did complete reinstalls just to end up with the same problem on systems that once worked fine. I\'m not convinced yet that anything in the system is bad, I wonder if a functioning system is just luck of the draw, ie the timing of where GSt\'s msg32 threads are with respect to kernel critical sections?
I also find it discouraging that people have msg32 problems in both win98 and xp, which really suggests that the problem is tascam\'s to fix. Are they not on the ball here, or simply not forthcoming?
Sorry for the rant, but I\'ve invested a lot in what has been an ideal GS setup, and it has really derailed my work the last week. I\'m frustrated and don\'t know what I\'m going to do, only that it\'s going to derail another week before I throw it out the window if I don\'t find the fix...
There has seemed to be a heat related aspect to my failures (ie failing only on a coldboot in the morning, not later in the day or warmboots), and the system just failed to boot in safe mode with this error:
himem.sys has detected unreliable xms memory at address 00873886
Perhaps my micron memory has failed. I will post more when I know, but I would expect GSt to be more sensitive to bad RAM than anything else I have.
Thanks Lee, though when I said heat related I really meant temperature related (initially it only failed when my system was at its coldest). It\'s a very conservative system, already has extra fans, and runs very cool. I agree reseating is the most common fix, but I don\'t think that will be my problem here. I strongly suspect the RAM has gone flaky, perhaps a victim of static discharge. I will be testing this today and report my findings when I know.
I regret my rant now since I now suspect GS is not problematic, and in fact I\'ve not had any problems with any of my software for years. I will clarify when I know more, but won\'t edit my posts since I think a record of hardware and human error could also be useful.
I have had this msg32 message happen quite often and when it is persistent and does not load after reboot I uninstall gs and then reinstall it and it works fine. I am running on win98 se. It is a more bothersome problem then a really bad problem.
I think that this problem is originated by other software. I had gigastudio working on a win98, and one day it began to have that error. I tried everything and nothing worked.
Then i installed a second (fresh) win98 (in the same partition), and i never had that problem again (for several months).