Port-amd64 archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: VirtualBOX additions (again)



I went through the motions to find out the behaviour under 7.2-STABLE
amd64. I used the image from
https://nycdn.netbsd.org/pub/NetBSD-daily/netbsd-7/latest/images/NetBSD-7.2_STABLE-amd64.iso
for the system, downloading
https://nycdn.netbsd.org/pub/NetBSD-daily/netbsd-7/latest/source/sets/syssrc.tgz
as well. After locating the right PKG_PATH I installed pkgin and then
got the minimum required to build VirtualBox extensions (kbuild, yaml
and liubxslt). With the previous LocalConfig.kmk it compiles cleanly.
The right video driver is vboxvideo_drv_110.so (there is no
vboxvideo_drv.so yet, of course). As there is no ws_drv.so, a mouse
driver is also needed, which is not built by default; one can cd to
./src/VBox/Additions/x11/vboxmouse and run kmk there, which again
builds the driver cleanly.

I then placed the generated files from
out/netbsd.amd64/release/bin/additions into /usr/local/bin/additions
as before and made the requisite links (only copying the module where
it shoould be). 'Xorg -configure' generates almost the right xorg.conf
- again the mouse driver needs to be changed to vboxmouse; the device
- to /dev/vboxmouse0 or /dev/vboxmouse1 (it does not work with
/dev/vboxmouse at all, does not move the pointer).

This configuration generaly works, with the exception that when one
tries to move/resize stuff, the feedback during the operation is all
over the place - the window may show much more to the right when
moved, but on button release it comes around the place where it is
expected to be. The click registers properly.

I had however a few panics in the vboxdriver -
...
(gdb) target kvm netbsd.2.core
0xffffffff8064fa9f in cpu_reboot ()
(gdb) bt
#0  0xffffffff8064fa9f in cpu_reboot ()
#1  0xffffffff80878222 in vpanic ()
#2  0xffffffff808782dd in panic ()
#3  0xffffffff808bb656 in trap ()
#4  0xffffffff80100f26 in alltraps ()
#5  0xffffffff809a7681 in wsmousedoioctl ()
#6  0xffffffff8086aff1 in cdev_ioctl ()
#7  0xffffffff8098a686 in VOP_IOCTL ()
#8  0xffffffff809764ed in vn_ioctl ()
#9  0xffffffff80887918 in sys_fcntl ()
#10 0xffffffff8089329a in syscall ()
#11 0xffffffff80100691 in Xsyscall ()
...

As I said, I am not likely to use 7.2 much, but it was interesting to
me to compare with 8.0 and -current.

On Fri, 18 Jan 2019 at 09:39, Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
>
> On Fri, 18 Jan 2019 at 02:48, Valery Ushakov <uwe%stderr.spb.ru@localhost> wrote:
> >
> > On Thu, Jan 17, 2019 at 23:57:34 +0000, Chavdar Ivanov wrote:
> >
> > > on both 8.0 and -current, no patches needed. Apparently the problem
> > > with the click is also sorted, at least under 8.0 using the vboxvideo
> > > driver from the source distribution of VirtualBox; under -current I
> > > was still running the version I compiled some copuple of weeks ago
> > > from the Xorg sources.
> >
> > You compare old X server with VBox vboxvideo on -8 with new X server
> > with Xorg's vboxvideo on current, changing both variables.
>
> Under -8, the video driver is the one built as part of VirtualBox
> 6.0.2; previously it was the driver built under 6.0.0.
>
> Under -current, I tried VirtualBox 6.0.0, then -trunk (before 6.0.2
> was released), then finally 6.0.2 - in all these occasions using
> vboxvideo from either the system itself as added recently, or the
> version I built from Xorg sources as explained above.
>
>
> > Also,
> > vboxvideo doesn't handle the mouse, ws driver does :)
>
> Sure, in all occasions.
>
> >
> > So the only variable then is the new X server.
>
> I wrote what I observe; at the moment all NetBSD test spins seem to be
> working OK for me.
>
> I'll check -7 later.
>
>
> >
> > -uwe
>
>
>
> --
> ----



-- 
----


Home | Main Index | Thread Index | Old Index