[vbox-dev] need patch for kernel-4.16-rc1
sergio at serjux.com
Mon Mar 12 01:47:52 UTC 2018
On Sat, 2018-03-10 at 15:40 -0600, Larry Finger wrote:
> On 03/09/2018 09:55 PM, Sérgio Basto wrote:
> > On Fri, 2018-03-09 at 20:40 -0600, Larry Finger wrote:
> > > On 02/18/2018 07:36 PM, Sérgio Basto wrote:
> > > > On Mon, 2018-02-19 at 00:22 +0000, Sérgio Basto wrote:
> > > > > Hello
> > > > > vbox 5.2.6 fails to build kmods for kernel-4.16-rc1 error log
> > > > > in
> > > > > attach
> > > >
> > > > After remove vboxvideo from kmods build ( ernel-4.16-rc1 have
> > > > vboxvideo , I don't know if still as stage module ...) it
> > > > builds
> > > > fine .
> > > >
> > > > Also VirtualBox 5.2.6 builds fine with GCC8 !
> > >
> > > In case you want to build module with vboxvideo with kernel 4.16-
> > > rcX
> > > as I had to
> > > do, the patch is attached. This material is released under an MIT
> > > license.
> > Hi,
> > Cool thanks
> > I have some point to discus, I don't know from where to start .
> > Checking virtualbox-5.2.8-382.2.src.rpm from software.opensuse.org
> > .
> > and the virtualbox.spec I sse that you don't have the new 2 kmk
> > variables VBOX_USE_SYSTEM_GL_HEADERS=1 and
> > VBOX_NO_LEGACY_XORG_X11=1
> > With VBOX_NO_LEGACY_XORG_X11=1 you may drop
> > remove_vbox_video_build.patch and also IIRC
> > Fix_for_server_1.19.patch
> I do not doubt your conclusions. The only potential problem is that I
> have some
> users that port the VirtualBox packages back to earlier openSUSE
> versions. For
> example, when I removed the build for vboxvideo, I got a bug report
> that it be reinstated. All in all, if the build is not broken, I
> prefer not to
> change it.
OK , you may need "patch for kernel-4.16-rc1"
> > With VBOX_NO_LEGACY_XORG_X11=1 you may drop virtualbox-system-
> > x.patch
> > internal-headers.patch is also around vboxvideo so I guess you also
> > may
> > drop it .
> As explained above, I cannot.
> > and I'd like review other patches and diffs of src.rpm ... In other
> > day
> > we talked about one patch from src.rpm to boot with pxe IIRC and if
> > the
> > patch is useful, we should upstream it (we may open a ticket for
> > one
> > patch) or at least I'd like use it in "my" packages .
> > Other subject since kernel 4.16 we also have vboxguest in kernel
> > vanilla. The vboxguest driver lives under drivers/virt.
> > And we already had vboxvideo, vboxvideo is still in
> > drivers/staging.
> > These modules are enable in Fedora kernels (vboxguest only in
> > 4.16), so
> > I don't need build vboxvideo.ko and vboxsf is in review to enter in
> > kernel.
> I have to support kernels from 4.4 to 4.16. Thus my build must have
> and vboxsf.ko even if they are built into some kernels.
vboxvideo.ko was merged in 4.14 (I think), vboxguest.ko was merged in
4.16-rc1 and vboxsf.ko still in review, the goal is no need build kmods
in guest system .
With VBOX_NO_LEGACY_XORG_X11 we can enable or disable build of X11
vboxvideo_drv.so , we need that for kernels <= 3.11 , RHEL7 have
3.10.0-514.6.2.el7.x86_64 and code X11/kernel/drm was backported so you
also don't need vboxvideo_drv.so in RHEL 7.4+ but is an exception .
The patch you send is for vboxvideo.ko in kernel-4.16-rc1 , so it is
your choice build kmod or use staging drive , but patches that I
mention from "your" src.rpm, I think you can drop it because are
related in disable build of vboxvideo_drv.so
With VBOX_USE_SYSTEM_GL_HEADERS=1 , is almost the same of virtualbox-
system-x.patch , only include VBoxVideoGuest.h is not in virtualbox
Best regards and thanks,
Sérgio M. B.
More information about the vbox-dev