EyesOfARaven

Just to bring this thread to a close, I wanted to say that the thing that fixed this issue for me was removing emm386 from config.sys - which strikes me as odd, but whatever: it worked.

It's a Rage 128 All-In-Wonder PCI, I was using a 4MB Ati Rage II DVD (somebody literally chopped the other 4MB chip off and sold it on Ebay to me [i didn't know while buying]) and it did well in everything except flash and Warzone 2100. I'm primarily doing DOS-based and 2D games, but a few 3d games are on my list of stuff to do with this machine, inc. Warzone 2100, Metal Fatigue, MDK.. If there's no solid consensus that the GF4 would kick this card's a** to the curb and back simply for flash and Warzone 2100 I'll do a test, but disassembling the machine, and installing new drivers is something I don't feel like doing, as I just changed TO the GF4, so I figured I'd ask. Correcting the above, I think it's a 32MB AIW, as the AMD site seems to indicate that the Rage Theater chip was found on the 32MB model and not the 16MB, and mine has it. That said, I think I'll give it a shot. I'll post back here and let ya know how it works out as search reference, so if someone comes across it in the future..

sp193: Might you tell me a bit about Windows codename Nashville? I hadn't heard of it. I've heard of Neptune, Chicago, etc... but never Nashville. Is it a Late Win95 build? Early 98 build? Or something inbetween? Sounds interesting! Edit: http://en.wikipedia.org/wiki/Windows_Nashville I gather it's somewhere inbetween. Got any screenshots - or can you describe it's UI? Edit2: Shouldn't have bothered asking before searching... http://www.pplware.com/2008/01/28/os-miste...2%2580%25A6-96/

I currently have a Geforce 4 440MX 64MB PCI card in my 98 box. It's running well, but I would like TV-in support. I have a 16MB ATI card that can do that, and was wondering if the performance hit would make flash lag more or have any effect on the range of games that would run on this system (1ghz p3, 512MB ram, 750GB HDD).. I think that the bottleneck at the processor would mean that anything requiring a 32MB or above card would likely be too slow to run anyway, and the 16MB would be fine. Thoughts?

Played a long game of SC again and didn't have any missing bitmap issues in explorer like last time. Don't know. No other problem/issues yet. Networking went wonky for a bit afterwards, but I think it was unrelated. Release/renew and running Opera fixed it. (Yes, running Opera helped - don't ask me why, but FF and IE didn't work after the renew until AFTER I ran Opera) Edit: spoke a bit too soon, the 98 logo is missing from the start menu now like before. I'm going to reboot - documenting the same problem isn't going to help. On the bright side, about 10 hours of uptime with no instability (as long as I stayed away from IE, and I did, lol)

I played StarCraft for a few hours, all worked fine, and I started to notice REALLY wierd graphical glitches afterwards. My guess is that the program is reclaiming GDI/USER resources from explorer since it determines them to not be in use while the game is up, and then they are just magically missing cuz explorer doesn't understand that they're gone or how to put them back. I took quite a few screenshots for you to troubleshoot with: http://rapidshare.de/files/47895770/bitmapfailures.rar.html

I don't think this is the sort of program you can pick and choose what it effects. I recommend if you want to continue using the alpha, you stick to your Firefox until then.

Those two tweaks affect the 32-bit heaps, while Tihiy's program (AFAIK) only messes with the 16-bit heaps. Edit: I closed DrWatson, not thinking the crashes were going to happen after testing for a while, and then IE crashes the second I close it. I open DrWatson, click close on the box, and it catches the subsequent crash (same ones I mentioned earlier with dibeng). I then tried to run IE again with DrWatson running and DrWatson disappeared, I tried to run it again, no go, I tried to run IE again, and it brought up the white error box with close/ignore about horrible problems, and the system locked when I hit close. One way to reproduce the DIBENG crash of IE is to navigate to http://www.microangelo.us/welcomet.asp and then click "enter".. Here's the watson log for GDI.EXE crashing after IE crashed from DIBENG: http://www.mediafire.com/?wjjwzqmm2nu I'm posting a watson log of IE crashing due to DIBENG and then explorer crashing shortly after while trying to copy the log for IE shortly after (then the system crashed with blue screen and I rebooted). IE Log: http://www.mediafire.com/?jnnyjjgb4qj Explorer Log: http://www.mediafire.com/?tq5m5iywk4d

IE6SP1 causes problems, I believe it's when it hits flash content loading. GDI.EXE is version 4.90.3002 After changing DIBENG.DLL back to stock, and not touching IE, and with some luck (one reboot crashed when rpxstart ran, no others have) it is working fine now. And aye, the IE6 crash is pretty reproducable, I'm just going to avoid IE until we get a new testing version from ya, as the performance increase for the rest of the system is too nice to trade out, heh. I'll reboot with drwatson in the start key and the patched dibeng, and see if I can get you a log. It was crashing almost everything, and I've had no issues with the DIBENG update PRE-RP10 alpha.

Yes but those have been almost completely resolved when I removed emm386 (still crashes after a few hours of browsing), and these are completely different, starting the second I browsed to youtube. Replacing the dibeng.dll from the cab (win98_22 I believe it was) fixed the dibeng problems. HOLY CRAP: This alpha version fixed the lag in flash for me using flash 10 on firefox 2! Excellent! Edit: (cept full screen, but it IS faster) same for flash 9, I just tested.

Here are my notes while testing, copypasted from a notepad session: 22 firefox tabs with firefox start up no problem at 84% gdi/user and 89% system free, resources freed afterwards too iexplore.exe caused a GPF in krnl386.exe 0001:000058ba after loading youtube.com, ie6sp1, flash 9 rebooting after that crash, startrpx caused an illegal operation, invalid page fault in kernel32.dll at 0187:bff7ce0 attempting to start startrpx by hand (i put it in the run key) results in "FAIL! Not mov and sub but 7f00809a there" The system then crashed, aside from cursor movement. Removing it from the run key on next stable startup and reinstalling RP9, as this seems to unstable on my system for daily testing. My system: P3 1ghz 512mb ram (510-11 seen by windows due to onboard graphics I dont use) 64mb geforce 4 440 mx pci 750gb fat32 single partition ide disk on promise 133tx2 ide controller pci windows 98 se sp3 2 me w/ multiple TSRs loaded high and a custom memory management setup (if you want I'll post autoexec and config and msdos files) Correction, I will be leaving it installed, after the next reboot I was able to use the system normally, iexplore caused a crash due to the DIBENG.DLL udpate from RP9 which i did NOT uninstall, a GPF at 0005:00003f2f ddhelp then did illegal operation of an invalid page fault in GDI.EXE at 0001:000073b0 No other ill effects from that crash. I will stop using IE6 for the remainder of the testing, as the other browsers should be more stable and capable now anyway. Edit 2: EVERYTHING is crashing due to dibeng.dll, you really should have warned us about that component specifically, as it isn't uninstalled with RP9. I will try to remove it (if I can)... Edit 3: Correction again, DIBENG.dll crashing seems to have only happened during that bootup, this is very odd. I will still attempt to remove the update to avoid further issue. Edit 4: Renaming dibeng.dll causes issues with explorer and apparently my color levels, I will try to expand one from a cab..

I tried to run the Boxee alpha on my 98 box with KernelEx, and whether or not KernelEx extensions (or which compatibility) are enabled, I get an error during setup at the end, and then upon trying to run the program get: The BOXEE.EXE file is linked to missing export SETUPAPI.DLL:CM_Request_Device_EjectA. Sounds like something that probably would work fine with just a stub in the DLL. Simultaneously with that error, I get the "A device attached to the system is not functioning" one, which I seem to get a lot for many programs. Thanks in advance for looking into it.

Good idea with the XPI.. I tried to run AeroSnap (.NET 2.0 app) on my 98SE2MESP3 box. It errors when you run it: "Application has generated an exception that could not be handled. Process ID=0xfffbfc9b (-263013), Thread ID=0xfffb6f17 (-299241). Click OK to terminate the application. Click CANCEL to debug the application." Upon hitting cancel, it informs me that not JIT debugger was specified. When I run it with KernelEx in Vista or XP SP2 mode, it says that it performed an illegal operation, module KERNEL32.DLL at 0187:bff87ede. I realize this program is far-fetched as far as trying to run on 98, but I thought I'd give it a go. Since data could always be used, I thought I'd post the info here.

I thought I tried that, but what the heck.. *tries again* Wow, you called it. Thanks Tihiy! xD It must have just been too late at night for me to think of that - it gets past that point in the install just fine now.

I had 95 running on my "retro box" just a few days ago, I have since upgraded the proc and changed a few parts, but at the time it had: 650mhz PIII ATI Rage II DVD 512MB RAM Intel 810e Chipset 750GB HDD (one partition FAT32) Kingston BT30 Network Card (The box is an upgraded Dell Dimension L550 - now VERY upgraded, heh...) I ran 95 over DOS 7.10 upgraded from WfW3.11. Worked great, drivers are definitely available (they are even available for Win3.x )...

When I run the .NET Framework 2.0 setup it goes about halfway and then errors, saying: Error 25015. Failed to install assembly 'c:\windows\microsoft.net\framework\v2.0.50727\microsoft.visualbasic.vsa.dll' There may be more data available. I googled about and found a few fixes, rename the assembly folder, delete the v2.0.50727 folder, etc etc.. I also tried running the setup in kernelex xp mode for the heck of it. All no good.

Yeah that's what I decided to do as well, and I'm glad to hear that 9x can throw EMS to apps in it's DOS boxes, as that should cover any EMS needs I would have anyway, and if I am in DOS I could just launch an EMM dynamically.

I am willing and eager to test any enhancements or utilities that you make as well, Tihiy. Thanks for all your work on Win9x, btw.

I do/do plan to run DOS apps, but even on my DOS machines I don't use EMM386, I run QEMM97 on those. AFAIK my games/progs don't use EMS though, just XMS, as I typically use NOEMS on QEMM or EMM386. Again, does anybody have any idea why Jemm isn't compatible with anything? Is it just me?

I put a mirror for RP9 and it's associated updates, as promised, on the Infotaxis Project website (mine) at: http://sites.google.com/site/infotaxisproject/Home/storage

I'm afraid that's the best I can do then, as when the system becomes any more unstable it's unlikely that it will function enough to even react to the crash itself. I took the snapshot after IE6 was non-responsive, so it wouldn't be much different were it to have been taken a bit later, as far as modules and such go. As for explorer, I have far less problems with that than IE, it's crashes are relegated to when I am copying lots of files with three dialogs open to various locations, and I'm practically inviting them to crash. Hopefully those will be reduced with my new tweaks, but if not, perhaps there is a way to get explorer windows open in a new process, such as in XP, on demand? This way if I know I'm going to be copying files around, I can open a new explorer process for each major operation, keeping them from crashing each other or the main shell instance. Thoughts? BTW: IE has been crashing less since I removed EMM386, though I couldn't imagine why. It took me 30-40 minutes of browsing with 4-5 windows open to get it to crash last time, and that's much more managable, though still not optimal. That's much closer to the experience with IE6 I remember when I was younger, so perhaps there's no fix for it? The crashes still occur with plenty of free heap though. Again, thanks for helping me figure this out guys.

Removing emm386 has certainly increased stability to a degree. I did a good multi-hour optimization session on my boot files, and switched over to umbpci.sys + himemx.exe in the end, adding a few new TSRs etc.. Went from 540K conventional free to 608K conventional free. Radical changes included renaming smartdrv and dblbuff so they wouldn't load, as I don't care to have my 750GB disk on a PCI controller buffered, and the rest of the system is fast enough for me without buffering as well. The conventional memory was worth it. I tried to do the hiram.exe trick, but it kept hanging on boot. I tried every combination involving hiram, himemx, himem, and umbpci, and no matter what I did it hung right before, or right after installing hiram. Anywho back to the topic at hand, I got a drwatson log right after the crashing began. Unfortunately I then closed drwatson, and couldn't get it back up as the crashing continued and brought the whole system to a state that needed a reboot. Here's the log file: http://rapidshare.com/files/256901245/wats...6crash.wlg.html

Not sure exactly when it started. I have been religiously running CCleaner as I do on all of my Windows boxes of all versions, which includes a registry cleaner. It doesn't massacre the registry afaik, but I haven't had a ton of personal experience with it on 98 yet. I have glanced through what it deletes and it's mostly unused file extensions, missing DLL file references, etc.. I've never used DrWatson but I assume it's straightforward. When I'm done copying these piles of files I'll browse around with it up on IE until it crashes again. Without emm386 there is little difference in my system. I am just copying files around and listening to music, which wouldnt cause instability WITH it, so I can't tell as far as that goes, but I only save like 20-30KB using UMB, so it's not too big a deal to keep it like this for now. Unrelated: Why do JemmEx and Jemm386 not work with... anything? They work great, take up like no RAM, but then when you go to run a DOS game in DOS with them, or launch Windows, they always exception out.. That's the only reason I don't use them..

Here's a pastebin of what curprocess had to say module-wise about explorer.exe: http://pastebin.org/2624 And iexplore.exe: http://pastebin.org/2625 I ran dependancy walker on Internet Explorer and it showed up with three missing DLL files: APPHELP.DLL INETCOM.DLL USERENV.DLL APPHELP and USERENV are under SHLWAPI.DLL, and USERENV is depended on directly by IEXPLORE.EXE..

(apologies for the length, but I wanted to be thorough) I'm experiencing regular crashes of explorer and IE6SP1, sometimes other apps, but most notably IE. I have installed the following in the listed order: 98SE 98SE SP3 beta 98SE2ME (without shell replacement) RP9 KernelEx 4.0RC2 (I am aware there is supposedly some incompatibility with RP9 and KernelEx, but I haven't experienced ill effects from that afaik. My first install of the two didn't work right, but after reinstalling KernelEx all was well..) I also applied a custom patch of a few hex edited files to make the version number in VER and any other references 4.95.2009, but this has not contributed to the problem afaik as it began prior to that. I have the resource meter running so I can keep an eye on the heaps, and they are usually around 80-90% free when the issues occur, which makes little sense to me, as I had been assuming they were to blame. RP9 did help the problems happen less often, so perhaps SOME of the crashes were due to that. What information can I provide you guys to help me troubleshoot? VXD list? Heapwalker? As well, I was having an oddball issue with EMS memory. I switched from himem.sys to himemx.exe, no change in this issue, which is that it can't find enough UMB to set up the page frame for EMS. I set NOEMS, and it tells me I have an invalid parameter, but it works, bah. (even with the error it lets me hit enter or whatev. to continue) The system is running great, with the exception of some newer games (newer in a 98SE sense.. Morrowind for example) refusing to work properly. Is my chimera of a system making itself incompatible with these apps? Youtube videos lag, especially fullscreen, whether flash9, flash10, Opera, IE6, Firefox 2, Firefox 3, anything. The same videos don't lag if I download them and play the FLV in VLC fullscreen. System specs: Hot-rodded Dell Dimension L550r (now boots and says L999r due to CPU) Pentium III @ 1ghz SB PCI64 (or one of the compatible variants, can't recall) Nvidia GeForce 4 440 MX 64MB 512MB RAM (half pc100, half pc133, for some reason it didn't like my other pc133 stick) 750GB HDD formatted in fat32 as one huge partition (did this in XP with third party utilities with the disk attached to a newer PC over USB)* *Note: The 750GB IDE disk is on a PCI Promise 133TX2 controller with proper cabling to run at Ultra 5, and the drivers for the controller are installed (as are those for everything else, even the chipset). This same box worked flawlessly in every imaginable way under DOS 7.10 with either no Windows, Win31, Win311FW, or Win95OSR2. I then decided on Win98 for it's compatibility in both directions, and after getting settled in and tweaking everything began to observe this problem. Thanks.