Bugzilla – Bug 1047215
Cannot install Oracle Virtualbox 5.1.-5.1.22_115216 : Kernel installation is not valid
Last modified: 2017-08-01 14:42:40 UTC
Every things you need is in my thread to opensuse forum. PLease read https://forums.opensuse.org/showthread.php/525768-Cannot-install-oracle-virtualbox-5-1-5-1-22_115216. The full virtualbox install log is here : http://paste.opensuse.org/21080199 You will find some comments on the opensuse forum : https://forums.opensuse.org/showthread.php/525761-Safe-way-of-mixing-repos?p=2828720#5 My knowledge is so low that I have no opinion. Any help is welcome.
Suggest to close this bug in a 3rd party package as "WONTFIX" (mainly because there is no option called "CANTFIX"). AK
(In reply to Jean-Claude Dole from comment #0) > My knowledge is so low that I have no opinion. > Any help is welcome. Jean, I'm sorry but bugzilla is not the general escalation path for issues that failed to get fixed for a user in the forums. Having no other option or lack of knowledge is no reason for a bug by itself. We use bugs to document and track technical action to take. Just nothing that this is a third party package, and that we are shipping Virtualbox in the distribution. Quoting: > make[...] > test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \ > echo >&2; \ > echo >&2 " ERROR: Kernel configuration is invalid."; \ > echo >&2 " include/generated/autoconf.h or include/config/auto.conf are missing.";\ > echo >&2 " Run 'make oldconfig && make prepare' on kernel src to fix it."; \ > echo >&2 ; \ > /bin/false) and > In file included from /tmp/vbox.0/r0drv/linux/the-linux-kernel.h:98:0, > from /tmp/vbox.0/r0drv/linux/memobj-r0drv-linux.c:31: > /usr/src/linux-4.4.73-1/include/linux/mm.h:1264:6: note: expected ‘struct vm_area_struct **’ but argument is of type > ‘int’ > long get_user_pages(unsigned long start, unsigned long nr_pages, > ^ > /tmp/vbox.0/r0drv/linux/memobj-r0drv-linux.c:1100:33: error: too many arguments to function ‘get_user_pages’ > papVMAs); /* vmas */ > ^ > In file included from /tmp/vbox.0/r0drv/linux/the-linux-kernel.h:98:0, > from /tmp/vbox.0/r0drv/linux/memobj-r0drv-linux.c:31: > /usr/src/linux-4.4.73-1/include/linux/mm.h:1264:6: note: declared here > long get_user_pages(unsigned long start, unsigned long nr_pages, > ^ As noted in https://forums.opensuse.org/showthread.php/525768-Cannot-install-oracle-virtualbox-5-1-5-1-22_115216?p=2828727#post2828727 there is a patch for vbox, which our packages have but which are omitted from Oracle's. Larry, are you involved in Oracle's virtualbox at all? Can they fix their packages to work with our kernel?
Here a proposal for a temporary workaround from opensuse forum. PLease read : https://forums.opensuse.org/showthread.php/525768-Cannot-install-oracle-virtualbox-5-1-5-1-22_115216#4
(In reply to Andreas Stieger from comment #2) > (In reply to Jean-Claude Dole from comment #0) > > My knowledge is so low that I have no opinion. > > Any help is welcome. > > Jean, I'm sorry but bugzilla is not the general escalation path for issues > that failed to get fixed for a user in the forums. Having no other option or > lack of knowledge is no reason for a bug by itself. We use bugs to document > and track technical action to take. That what I mean Correct your modification from 42.2 to 42.3 I can install on 42.2 bit not on 42.3 > > Just nothing that this is a third party package, and that we are shipping > Virtualbox in the distribution. > That has nothing to do with the problem. > Quoting: > > > make[...] > > test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \ > > echo >&2; \ > > echo >&2 " ERROR: Kernel configuration is invalid."; \ > > echo >&2 " include/generated/autoconf.h or include/config/auto.conf are missing.";\ > > echo >&2 " Run 'make oldconfig && make prepare' on kernel src to fix it."; \ > > echo >&2 ; \ > > /bin/false) > That a very known error when there is a kernel problem.
(In reply to Jean-Claude Dole from comment #3) > Here a proposal for a temporary workaround from opensuse forum. > PLease read : > https://forums.opensuse.org/showthread.php/525768-Cannot-install-oracle- > virtualbox-5-1-5-1-22_115216#4 I think this workaround is for Oracle Virtualbox, not the kernel.
(In reply to Andreas Stieger from comment #2) > As noted in > https://forums.opensuse.org/showthread.php/525768-Cannot-install-oracle- > virtualbox-5-1-5-1-22_115216?p=2828727#post2828727 > there is a patch for vbox, which our packages have but which are omitted > from Oracle's. > > Larry, are you involved in Oracle's virtualbox at all? Can they fix their > packages to work with our kernel? Problem is, this patch only applies to the 4.4.x from 42.3 with a lot of backported code from newer versions and will break building the modules against the 4.4.x from 42.2. So the only way for Oracle to fix this would be to supply an extra package especially for 42.3 containing that change and I don't know if that is very likely to happen. AK
Yeah I do not think we can fix that for a third party package. Also we ship a working VirtualBox 5.1.22. Slightly better autoconf magic may replace the pure version number comparison that the code seems to be doing.
(In reply to Andreas Stieger from comment #7) > Yeah I do not think we can fix that for a third party package. Also we ship > a working VirtualBox 5.1.22. > > Slightly better autoconf magic may replace the pure version number > comparison that the code seems to be doing. I would not expect Oracle to handle any breakage caused by backporting of kernel features that break the API/version correspondence. I am interested in the "autoconf magic" mentioned above. Please elaborate.
After applying today update which contains kernel 4.4.74-1, I can install Virtualbox without problem. Thank you very much for correcting that bugs.
(In reply to Larry Finger from comment #8) > (In reply to Andreas Stieger from comment #7) > > Slightly better autoconf magic may replace the pure version number > > comparison that the code seems to be doing. > > I would not expect Oracle to handle any breakage caused by backporting of > kernel features that break the API/version correspondence. I am interested > in the "autoconf magic" mentioned above. Please elaborate. By that I mean that upstream may switch from comparing kernel version numbers to testing whether a particular API is available. But it seems this has already resolved itself.
Correction : Does not work. The virtualbox Gui opened, but vbox services does not run. I am sending this conversation to the Virtualbox team.
This bug is fixed in Factory and the fix has propagated to Tumbleweed. It should be in Leap 42.3 soon.
FWIW, https://www.virtualbox.org/changeset/68217/vbox/?old=68214&old_path=%2Fvbox Next upstream virtualbox release should have the above fix.