#618 Mouse does not work properly under DOS
The mouse performs strangely under both MS-DOS 6.2 and
Free DOS. The movement of the mouse seems to be
proportional to the distance from the center of the
bochs screen. That is if the mouse is below the center
the cursor will move down whenever I move the mouse and
vice versa.
I am using the standard MS-DOS mouse.com driver that
comes with DOS 6.2 on both systems, running bochs under
both 98 and Windows XP as the host (not at the same
time). I haven't tried it under windows, but it has the
problem using the mouse under edit as well as a few
other select applications I am using so it doesn't seem
to be confined to one app.
As far as I know all the applications use the standard
mouse interrupt to get the position of the mouse, which
I think is interrupt 30h, (you'll probably know more
than I do about it.)
It makes controlling the mouse under dos impossible. I
guess its something to do with the passing through the
mouse movement to the emulated hardware, as everyhting
works fine on a real box.
Logged In: YES
user_id=1557887
on my ms-dos 6.22 emulated sys, my session freezes whenever
i use MOUSE.COM!
Does this bug still exist?
I can't reproduce it with Bochs 2.6.9 running on Windows XP, unless I run Windows XP under QEMU on Linux.
This is an issue that I recall experiencing in the past, probably roughly 10 years ago on Windows XP. Some time ago I performed a physical-to-virtual migration of the machine I probably saw this on, and that's what I'm running under QEMU when I reproduce the issue, so perhaps the issue is somehow following that Windows installation. That machine certainly had a lot more software installed on it over its lifetime than the physical machine I tried to reproduce this on today, which has a fairly clean Windows XP install (aside from the bloatware that came with it).
In both cases, I was using pretty similar software inside Bochs - MS-DOS 6.22 with the CuteMouse v1.9 beta 4 driver from FreeDOS. I tried some different Microsoft Mouse driver versions and it didn't cause the problem to appear when running Windows XP on bare metal or go away when running Windows XP under QEMU.
My old Windows XP installation had Bochs 2.1.1 installed. Running that under QEMU, I have the same trouble with a Windows 95 guest, and I seem to recall hitting it with OS/2 2.x too.
I was able to reproduce this issue with Bochs 2.6.9 on Linux with a Windows 98 SE guest (with Bochs emulating a standard mouse, not a USB tablet) by running Bochs in a local VNC server. If I run the same Bochs version with the same guest, but not under the VNC server, there are no mouse problems. Presumably the issue in this case is that when Bochs tries to move the mouse cursor back to the middle of its window, this is not honored in some way - perhaps the X server provided by the VNC server honors it, but doesn't pass that request back to the client, or the client doesn't pass it to the X server the client is running on. Using a VNC viewer that always displays the local cursor might help to understand what is going on here.
I realise that this probably isn't the original issue, but perhaps the issue is similar to what I hit above when running under QEMU, and perhaps other instances of this bug have been due to issues with Bochs moving the mouse cursor to the center of the window too?