Because udlfb would be getting no notification of which pixels are changing, defio support in udlfb would have to be enabled (compile switch) to use page fault change detection.

Given that, it would depend on whether the DOS emulator could handle the fact that udlfb only supports a single framebuffer format: 16bpp RGB565. If the DOS emulator was using the standard fbdev interface, and could convert pixels to that format (or the DOS app was using that format natively), there should be nothing preventing it from working.

But I haven’t done this myself, and haven’t seen any reports from others yet.
.
Thanks!
Bernei

Is it possible to run the dos-application in fullscreen mode on seat ?

I tried but with no luck.

Have you tried that ?

Please let me know.

Regards,
Satish

Thanks for the update.
In that case, I will try to go with the distribution like Fedora 13 and/or RHEL-5.5
I hope, they will work like a charm.

For Ubuntu, I just followed the link, I have not updated the ubuntu’s any of package. So there must be some issues which need to be resolved.

Have you tested by booting the ubuntu ( jaunty ) with multiple USB Display-Link connected and getting all of them an individual screen, mouse, keyboard and sound ?

Also, I am still stuck at the process of maintaining the session of removed-device.

Please share your thought the same issue.

Thanking You
Satish

In terms of expertise, Alexander Todorov (http://otb.bg/) has done the most work with RedHat, although mostly 5.4+

So, sorry, focus is on getting things working on future distros, with no planned effort for older distros.

I started to migrate the project on RHEL-5.3 (32 bit).
But it seems that RHEL5.3 is not responding the same way Ubuntu does.

The kernel module ( udlfb ) making the kernel panic ( with no message in /var/log/messages ) and the X module ( displaylink_drv.so ) is not getting compiled at all, giving the error of fb_dev structure and missing xf86Modes.h header file.

I would like to know, Bernie, that Is there any tree maintained for RHEL-5.3 or any sort of support is available for the same ?

Thanks
Satish

No change on device removal behavior. I think it would be interesting to put udlfb in an (optional) mode where it doesn’t tear itself down on device removal if a client has the framebfufer open, rather it keeps a list of DL serial numbers, and reconnects when the same device comes back.

But it would be great to have a GDM-level temporary disconnect operation, rather than this low-level kind. No idea how to do it, though.

On your input device behavior – I’ve not seen anything like that. Suspect you’re using different version of xorg. Make sure to *not* use the sharevts option when starting X. And it make take the new Xorg InputClasses support in later distros to make this all clean for multiseat.

Please post any of your results and progress somewhere. I’m sure people will be interested.

Best wishes,
Bernie

[WORDPRESS HASHCASH] The poster sent us ’0 which is not a hashcash value.

Now I am getting very close to the desired project.

But the only problem, I am facing as of now is that, whenever I plug 2+ devices at the same time, then all of these devices share the keyboard and mouse. The mouse and keyboard of the *last* connected device remains in action.

Same things goes to happened when I connect the 2+ device and then boot the machine, Except that the mouse and keyboard of the HOST machine do not come up.

Also, If I plug-in and plug-out the device number of times then either that device do not come up and/or the Host machine and all the rest multi-seat sessions get hanged.

Can you shed some lights in theses issues ?

( P.S. Why the twitter account is not updated http://twitter.com/libdlo ? )

Thanks `n` Regards
Satish

[WORDPRESS HASHCASH] The poster sent us ’0 which is not a hashcash value.

Any progress for device-removal operation ?
I am still (re)searching on, the way to save the current session as soon as the device is get dis-connected and greeting the user the same session when he/she logs back.

Looking forward for your positive feedback.

Regards
Satish

[WORDPRESS HASHCASH] The poster sent us ’0 which is not a hashcash value.

Thanks for the kind words. Here’s the latest on multiseat and GDM: http://mail.gnome.org/archives/gdm-list/2010-April/msg00006.html

So the best help in that area would be to take those unapplied patches and help test and provide feedback on them, especially in combination with the instructions above.

I’m hoping there’ll be some news soon on Google Summer of Code, which might provide a good push for this work …

Thanks!
Bernie