rloew

My free KERNEL32.DLL Patch is for files larger than 2GiB. It should have no effect of any kind unless you were playing a video larger than 2GiB.

I am using K-Meleon 1.5.4 without KernelEx. I have to refresh the page with Javascript turned off to see a text entry Window. This is my first attempt to post this way. If you see this, it worked. Apparently it does. Now editing.

Confirmed. I removed my VXD from the Z87 and ran the test until the Mouse choked. The Graphics Card Interrupt was marked active. I sent an EOI to the Virtual PIC and the Mouse recovered. The MSI 970A, with the VXD, has been running for more than 24 Hours without any Disk or Mouse problems. When I release the VXD, I plan to add it to the NVSIZE Package. If there is enough interest I might offer it as a standalone package.

There may be some other Registers in the Chipset that are powered by the Standby Rail of the Power Supply. My MA785 remembers whether I last used a Mouse or a Keyboard on the single PS/2 Port, unless I lose Power. Of course there is also the CMOS which is Battery powered even if you pull the plug.

I used tools I wrote myself to find the location to patch. I have never used Ollydbg. First I used a binary search program to locate the Exception Code. Then I used a disassembler to identify the enclosing subroutine. Analysis showed that no return code was needed and the calling convention. Dropping a return instruction on the top of the subroutine was the appropriate thing to do.

Doubled Devices are not that unusual, even if there are no corrupt files. Removing both copies from Safe Mode and redetecting them usually works.

After several days of experimenting, I think I have found the problem. The NVidia Drivers appear to have an Interrupt handling problem. It appears that with a lot of activity, the Interrupt is being left in a masked state disabling further Interrupts. The Mouse hangups are the direct result of this as the Mouse Driver appears to use the Graphics Interrupt to poll the Mouse. Without Interrupts, the Mouse only moves when a timeout occurs in it's Driver. Disabling the Graphics Interrupt causes the same symptoms. I was testing with a CLI Math Program so it cannot be caused by Resource depletion. On my MSI 970A, the same Interrupt was also shared with an USB Controller. I will investigate this further but it may explain why USB crashes occur as well. The Hard Disk problem I have been working on would also be explained in the same manner, but I could find no solution by modifying it. This problem only appears with my SATA Patch as the Unpatched Driver cannot be used with these Controllers. I have written a VXD that cleans up the Interrupt Controller after each Interrupt has been processed. It has been working for several hours so far. No Disk Errors or Mouse Hangs. I should know by tomorrow if it works properly.

To Cold Boot, turning off Power for 1 to 2 Seconds is enough. If you need to erase the standby powered Registers or change hardware, and the Main Power is off, you might need to wait a Minute after pulling the plug for the Power Supply Capacitors to drain. If you are in a rush, press the Power Button to drain it faster or pull the plug with the Power On but not while running an OS.

On the MSI 970A, using two PCI-E Video cards did get separate Interrupts. Mixing a PCI-E Video Card and a PCI Video Card did not work. Both got the same Interrupt. The DMA Controller does not use Interrupts so that is not an issue. I have run my Test program on the M5A97 for 24 hours already without the BSOD. The Mouse choked up but I saw that on the Z87 as well. I am going to have to experiment with the Interrupt Handler in the ESDI_506.PDR Driver and maybe NVCORE.VXD as well. This particular issue you wouldn't see since you do not have my SATA Patch, so you would not be able to use the affected Ports.

I found a connection between the Graphics Card and the first SATA Controller on the M5A97 and the MSI 970A. They both share the same Interrupt. This may explain the Disk Write Failure BSODs. I moved my Hard Drive to the second Controller and am running another test. This sharing does not occur on my other Computers.

I finally did get a Mouse failure with my Z87 Motherboard with a 7950GX2, but no crashes or other lockups. I am retrying my Math program on the M5A97 without Multicore but it runs much slower and probably will take weeks to test. The XGA Driver ran without problems with the Math program. The MKV Video of course will not run with it. The MSI 970A froze with the XGA Driver but I have had enough thermal problems with it that I cannot be sure. I am thinking about a 760G Chipset Motherboard that promises to support the FX 8370 CPU I bought for the MSI 970A. Unfortunately it has only one PS/2 Port like the Gigabyte Motherboards. It doesn't say "Keyboard OR Mouse". Maybe there is hope for a splitter.

I tried a PS/2 splitter cable years ago. No luck. The manual says "Keyboard or Mouse". Apparently they really meant the "or". I have had problems with it thinking my KVMA Mouse connection is a keyboard after a power failure. I have to plug in a real mouse, boot the Computer, turn if off, reconnect the KVMA Mouse and turn it back on. There must be a hardware register that is maintained by standby power that selects the mode but I have not been able to find it.

Windows 95 does not have USB Keyboard or Mouse Drivers. Only the Hybrid that Lone Crusader and I developed can use them. Windows 98SE has USB 1 built in, but during installation or reinstallation, prompts are put up. Having one PS/2 Port does allow you to complete installation, but some people have none at all and cannot complete the install because their Mouse and Keyboard are disabled in the middle of the process.

I suspect these PS/2 Ports are interfaced through the USB Controller. You can get the same result using a PS/2 to USB Adapter. This would not work with Windows 95 nor solve the handover issue when installing to Windows 98. I haven't seen a PS/2 Card that has a true PS/2 Interface. I suspect that if they exist, they would be very old ISA Cards.

I use the ones from NUSB 3.3. Mice use USB 1 so the USB 2 Drivers are not needed unless there is no USB 1 Controller. Are you using a converter or a KVM Switch between the Mouse and Motherboard? Some have issues. The three unrecognized USB Devices are probably the USB 3 Controllers.

I don't have a 770 Motherboard so I have nothing to say about it. I do see a possible problem with the two AMD 970 Motherboards though. Neither one is able to reliably play a 1920x1080 MKV Video using Gom Player 1.5.4. The Video freezes after an hour or two and the Mouse becomes barely responsive. I have a Mathematics Program that runs all eight cores at near 100% as well using 64-Bit Memory. After 10-15 Hours I get a Disk Write Error BSOD. It occurs with every combination of a 6200, 7100, 7200 or 7950GT Video Card, with Versions 77.72, 81.98, or 82.69 Driver that I have tried. With no driver, it works fine. I am now testing the Z87 with a 7950GX2 Card and the M5A97 with a XVGA Driver.

The SATA Patch, like all of my Patches, can be Installed from DOS. It can be applied after the first pass of Installation before the Drivers become active. Slipstreaming it would require that a Patched ESDI_506.PDR File be added to the Installation CD, or update the File in the corresponding CAB File.

I have no idea why DMA would work with a SSD and not with a CD/DVD Drive. Nor how this would affect browsing. Xrayer's method is far more complicated than the one I suggested but handles more cases. The Tenda driver only needs some commented out lines restored in the INF File. Otherwise it installs without complexity. I use Iogear PS/2 KVMA Switches, which can be cascaded, along with Inland PS/2 to USB Converters. I have had some issues with proper operation of the Keyboard and/or Mouse with different Converters and Motherboards. UEFI BIOSes seem to have additional problems when in the BIOS Setup screen with both. I have never seen a problem with FireWire Controllers. Try my RAM Limitation Patch Demo with the /M Option. It handles a number of problems even with 1GiB or less of RAM.

I just started working with a second AM3+ Motherboard. A MSI 970A. The earlier one is the Asus M5A97. I don't have any 870 chipset Motherboards, at least with AM3+. I have not had problems with Disk DMA as the Hard Drive Controller has it's own Bus-Master DMA, so it is not dependent on the System DMA Controller. You probably will need my SATA Patch as neither of mine offered "Legacy" Mode. Both Motherboards use the RTL8111E NIC. The Tenda 9901 Driver will support this but only at 10/100MB/s and with a warning on every boot. What graphics card are you using with the GigaByte Motherboard? It might be causing the USB problem. I stopped buying GigaByte Motherboards because of the single PS/2 Port. A PS/2 splitting cable did not work with my MA785 as there is only one software defined port. I use a USB Keyboard and a PS/2 Mouse with it.

I am now experimenting with a MSI 970A Motherboard. It is definitely possible to run 98SE on this one. The VXD Error may be related to the lack of DMA Memory issue that the /M Option of my RAM Limitation Patch addresses. Setting MaxPhysPage has no effect on this. My Z87 required this and the SPLIT8MB Program in the Patch. Disabling ACPI is probably necessary but is not likely the cause of the VXD Error.

Although both 9x and NT use the WDM Framework, there are significant differences in the way they are implemented, even for functions that appear in both. I have tried to replace or stub out missing functions but have only had very limited success in porting Drivers to 9x.

The 1394 driver that comes with Windows 98SE is fairly generic, at least for Motherboard 1394 Controllers. I haven't found one that is not recognized. I don't have any 1394 peripherals to test with.

Is the DVD Driver installed properly. If not, are the Hard Disk Drivers working properly or are you in Compatability Mode?

That function was already in my to do list for WDMEX.

KernelEX is for User Mode Programs. Drivers are Kernel Mode Programs. I have been working on a Kernel Mode extender called WDMEX but results have been limited. It currently adds all Windows ME Functions to 98 and some 2K Functions. Without further analysis, I cannot say if this would help.