OSDev.org

Hi,

i did search the articles in this forum, but not found the answer.

I'm using windows7 64bit system with an optical mouse with left button, right button and a wheel. I'm using winring0 driver for the direct hardware port access. My binary project is written in c# and only built for x86 (32bit binary).

I've verified to simulate the keyboard input in my project with following code:
uint status = (uint)ReadIoPortByte(0x64);
while ((status & 0x03) > 0)
{
Thread.Sleep(20);
status = (uint)ReadIoPortByte(0x64);
}

WriteIoPortByte(0x64, 0xd2);
Thread.Sleep(1); //if no sleep is called here, it does not work
WriteIoPortByte(0x60, scanCode);

uint status = (uint)ReadIoPortByte(0x64);
while ((status & 0x03) > 0)
{
Thread.Sleep(20);
status = (uint)ReadIoPortByte(0x64);
}

WriteIoPortByte(0x64, 0xd2);
Thread.Sleep(1);
WriteIoPortByte(0x60, (Byte)((uint)scanCode | 0x80));


But it does not work to simulate the mouse left click with the following key code:
WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x09);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x08);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

WriteIoPortByte(0x64, 0xd3);
WriteIoPortByte(0x60, 0x00);

I also inserted some sleep(n) functions, check the bit0 and/or bit1 of status register read from 0x64, check the ACK(0xfa) read from 0x60, send extra/less commands with byte 0x00 to mouse, tried all the possible combinations. I did see it once works, very stable, but, the next day, after reboot the system, it no longer works. I'm unable to reproduce the good case any more these days.

another thing need to mention is: i tried the code in win7 32bit system in VirtualBox on the same machine, it works, very stable. I have not verify if it relative to 32bit VS 64bit OS.

thanks a lot

Hi,

Some minor notes...

a) Most modern systems use USB keyboard and mouse, and therefore this code will fail in most cases.

b) This code has severe race conditions (e.g. nothing prevents the real driver screwing up your hack at the wrong time). With the approach you're using these severe race conditions are unsolvable.

c) The correct approach would be to put something between the real driver/s and the GUI, and not to slap a hideous hackfest between the real driver and the hardware. I'm fairly certain Windows has some sort of API/interface that's actually intended for this (e.g. for IME support).

d) This isn't a Windows development forum; it's a forum for people developing new/alternative OSs


Cheers,

Brendan

thanks a lot Brendan.

a) yes, i'm using usb keyboard and usb mouse. The code works for usb keyboard, and works for mouse in the virtual machine.

b) i'm running the code during the time i do not touch keyboard and mouse, just guess the race conditions do not happen. For the worst case, even it sometimes happens, i just guess it should work for some cases that race conditions do not happen.

c) yes, there are APIs that windows provides to simulate mouse/key, such as mouse_event and keybd_event, but they are 'disabled'/shield by my target application, so i have to simulate at direct port access level.

d) sorry for the wrong posting. I just thought it is not relative to windows GUI, but the underlying hardware port direct access method. I saw something similar topics in the wiki and so asked here.

thanks again, and sorry.

Obviously you're trying to hack something.

The guaranteed course of action is to write your own virtual mouse driver so that the app can't detect differences in events between the two.

hi Combuster, do you mean to write a filter driver for the mouse? it looks like too big ...

i tried win7 64bit in VirtaulBox on the same machine, the mouse click also works ...

Hi,

guoguo wrote:i tried win7 64bit in VirtaulBox on the same machine, the mouse click also works ...

So what you're saying is that, when running inside VirtualBox (which emulates PS/2 keyboard and mouse by default) your code (which assumes PS/2 keyboard and mouse) "works"; and when running on real hardware (that most likely uses USB keyboard and mouse) your code (which doesn't support USB) doesn't work?


Cheers,

Brendan

Hi Brendan,

yes, exactly as you said.

some more information, in 'Computer Management' -- 'Device Manager'

Inside the virtual machine, there are:
'Keyboards'
--Standard PS/2 Keyboard
'Mice and other pointing devices'
--HID-compliant mouse
--Microsoft PS/2 Mouse

Inside the host OS running on the real hardware, there are:
'Keyboards'
--Standard PS/2 Keyboard
'Mice and other pointing devices'
--HID-compliant mouse
--ThinkPad UltraNav Pointing Device

It explains that why the keyboard press works on both the host machine and the virtual machine, also explains why the mouse click works on the virtual machine. But, as mentioned in my original post, i did see the mouse click works on my host machine in a late night, very stable, but, it no longer works the next day....

Looks like that the direction is to add a 'virtual' PS/2 mouse in my system, is there such a way ...