[Libdlo] displaylink-mod status for kernel 2.6.18

Bernie Thompson bernie at berniethompson.com
Thu Nov 12 10:42:59 PST 2009

Hi Alexander,

Thanks for reporting this.  That problem comes up on kernels where the
kzalloc function fails for the largish framebuffer backing allocation.

The nice thing is we have so many darn branches floating around now,
that we have one for every occasion.  :) This branch of
displaylink-mod (which forgoes allocating a backing bufffer entirely,
with a tradeoff in performance) should work:

As will Jaya's displaylinkfb, or patching to use a different
allocation function (search the list for a patch from Dan Ellis).

The udlfb 0.4.0 release yesterday, however, will still have the same
behavior you're currently seeing on displaylink-mod 0.3.

On plans:

We need to get things moving forward with the driver that's already in
kernel (udlfb) so we can get all this stuff in one place.  From the
discussion with Greg, We'll try to keep up what was started yesterday
-- of doing new udlfb releases that bring things together and keep
them moving forward -- but also take on the work of doing it in
feature/topic branches that we can submit as somewhat functionally
independent email patches against the full kernel, to align with the
kernel & staging tree process.

Any offers to help or comments/suggestions on smarter ways of handling
things are always welcome.


On Thu, Nov 12, 2009 at 10:18 AM, Alexander Todorov <atodorov at otb.bg> wrote:
> Hello folks,
> I've managed to compile and install displaylink-mod-0.3.tar.gz without issues
> with kernel 2.6.18-164.6.1.el5. The driver loads but the attached monitor
> doesn't turn green but shows garbage.
> The relevant messages from /var/log/messages are:
> Nov 12 22:58:28 multiseat kernel: DisplayLink device attached (DU2781)
> Nov 12 22:58:29 multiseat kernel: found DisplayLink Chip unknown
> Nov 12 22:58:29 multiseat kernel: sent encryption null key: 16
> Nov 12 22:58:29 multiseat kernel: displaylink monitor info: W(1024) H(768)
> clock(6500)
> Nov 12 22:58:29 multiseat kernel: set video mode bulk message: 144 144
> Nov 12 22:58:29 multiseat kernel: displaylink register flush: 0
> Nov 12 22:58:29 multiseat kernel: non posso allocare il backing buffer
> Nov 12 22:58:29 multiseat kernel: displaylink: probe of 1-2.1:1.0 failed with
> error -12
> Nov 12 22:58:29 multiseat kernel: usbcore: registered new driver displaylink
> Nov 12 22:59:12 multiseat kernel: usbcore: deregistering driver displaylink
> This is with RHEL5.4 on an Asus EeePC with Intel graphics:
> 00:02.0 VGA compatible controller: Intel Corporation Mobile 915GM/GMS/910GML
> Express Graphics Controller (rev 04)
> 00:02.1 Display controller: Intel Corporation Mobile 915GM/GMS/910GML Express
> Graphics Controller (rev 04)
> Several questions:
> 1) How can I turn verbose output or enable debug mode so I can extract more info?
> 2) Is there a minimum kernel version for displaylink-mod or the other drivers?
> 3) Should I be testing with displayllink-mod or switch to udlfb 0.4.0 which is
> proposed upstream as Bernie said? What are the plans for upstreaming/merging
> other drivers?
> Thanks,
> Alexander.
> _______________________________________________
> Libdlo mailing list
> Libdlo at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/libdlo

More information about the Libdlo mailing list