| Message ID | 20200731095540.29456-1-paolo.pisati@canonical.com |
|---|---|
| State | New |
| Headers | show |
| Series | [Groovy] UBUNTU: SAUCE: export __get_vm_area_caller() and map_kernel_range() | expand |
On 31/07/2020 10:55, Paolo Pisati wrote: > As explained here: > > https://www.virtualbox.org/ticket/19644 > > and in the recent virtualbox DKMS upload changelog: > > https://metadata.ftp-master.debian.org/changelogs/contrib/v/virtualbox/virtualbox_6.1.12-dfsg-8_changelog > > Virtualbox DKMS now requires access to these two kernel functions to fix a FTBFS > against Linux 5.8, so until they rewrite their VM code (something that is not > happening anytime soon, see comment 8 in the first link), export these two > symbols. > > Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com> > --- > mm/vmalloc.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c > index 317afdc7d247..dffac67ba7f7 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -326,6 +326,7 @@ int map_kernel_range(unsigned long start, unsigned long size, pgprot_t prot, > flush_cache_vmap(start, start + size); > return ret; > } > +EXPORT_SYMBOL(map_kernel_range); > > int is_vmalloc_or_module_addr(const void *x) > { > @@ -2130,6 +2131,7 @@ struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags, > return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE, > GFP_KERNEL, caller); > } > +EXPORT_SYMBOL(__get_vm_area_caller); > > /** > * get_vm_area - reserve a contiguous kernel virtual area > Sigh, I guess that needs to be done. Thanks for cornering this Paolo. Are you going to keep monitoring the upstream driver so we can rip these out once they fix their code? Acked-by: Colin Ian King <colin.king@canonical.com>
On Fri, Jul 31, 2020 at 11:55:40AM +0200, Paolo Pisati wrote: > As explained here: > > https://www.virtualbox.org/ticket/19644 > > and in the recent virtualbox DKMS upload changelog: > > https://metadata.ftp-master.debian.org/changelogs/contrib/v/virtualbox/virtualbox_6.1.12-dfsg-8_changelog > > Virtualbox DKMS now requires access to these two kernel functions to fix a FTBFS > against Linux 5.8, so until they rewrite their VM code (something that is not > happening anytime soon, see comment 8 in the first link), export these two > symbols. > > Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com> Makes sense to me, but, as mentioned by Colin, we should drop this patch once virtualbox fixes this upstream. Acked-by: Andrea Righi <andrea.righi@canonical.com> > --- > mm/vmalloc.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c > index 317afdc7d247..dffac67ba7f7 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -326,6 +326,7 @@ int map_kernel_range(unsigned long start, unsigned long size, pgprot_t prot, > flush_cache_vmap(start, start + size); > return ret; > } > +EXPORT_SYMBOL(map_kernel_range); > > int is_vmalloc_or_module_addr(const void *x) > { > @@ -2130,6 +2131,7 @@ struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags, > return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE, > GFP_KERNEL, caller); > } > +EXPORT_SYMBOL(__get_vm_area_caller); > > /** > * get_vm_area - reserve a contiguous kernel virtual area > -- > 2.25.1 > > > -- > kernel-team mailing list > kernel-team@lists.ubuntu.com > https://lists.ubuntu.com/mailman/listinfo/kernel-team
On Fri, Jul 31, 2020 at 12:21 PM Colin Ian King <colin.king@canonical.com> wrote: > > Sigh, I guess that needs to be done. Thanks for cornering this Paolo. > Are you going to keep monitoring the upstream driver so we can rip these > out once they fix their code? I added a card/reminder in the Groovy board to check back if there's any update to the DKMS/kernel Virtualbox patches. Thanks.
On 31/07/2020 15:50, Paolo Pisati wrote: > On Fri, Jul 31, 2020 at 12:21 PM Colin Ian King > <colin.king@canonical.com> wrote: >> >> Sigh, I guess that needs to be done. Thanks for cornering this Paolo. >> Are you going to keep monitoring the upstream driver so we can rip these >> out once they fix their code? > > I added a card/reminder in the Groovy board to check back if there's > any update to the DKMS/kernel Virtualbox patches. > > Thanks. > Thanks Paolo
On 31/07/2020 10:55, Paolo Pisati wrote: > As explained here: > > https://www.virtualbox.org/ticket/19644 > > and in the recent virtualbox DKMS upload changelog: > > https://metadata.ftp-master.debian.org/changelogs/contrib/v/virtualbox/virtualbox_6.1.12-dfsg-8_changelog > > Virtualbox DKMS now requires access to these two kernel functions to fix a FTBFS > against Linux 5.8, so until they rewrite their VM code (something that is not > happening anytime soon, see comment 8 in the first link), export these two > symbols. > > Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com> > --- > mm/vmalloc.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c > index 317afdc7d247..dffac67ba7f7 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -326,6 +326,7 @@ int map_kernel_range(unsigned long start, unsigned long size, pgprot_t prot, > flush_cache_vmap(start, start + size); > return ret; > } > +EXPORT_SYMBOL(map_kernel_range); > > int is_vmalloc_or_module_addr(const void *x) > { > @@ -2130,6 +2131,7 @@ struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags, > return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE, > GFP_KERNEL, caller); > } > +EXPORT_SYMBOL(__get_vm_area_caller); > > /** > * get_vm_area - reserve a contiguous kernel virtual area > I just spotted this: https://sources.debian.org/src/virtualbox/6.1.12-dfsg-8/debian/patches/ https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-1.patch https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-2.patch https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-3.patch Maybe that is worth looking at. Colin
On Fri, Jul 31, 2020 at 11:55:40AM +0200, Paolo Pisati wrote: > As explained here: > > https://www.virtualbox.org/ticket/19644 > > and in the recent virtualbox DKMS upload changelog: > > https://metadata.ftp-master.debian.org/changelogs/contrib/v/virtualbox/virtualbox_6.1.12-dfsg-8_changelog > > Virtualbox DKMS now requires access to these two kernel functions to fix a FTBFS > against Linux 5.8, so until they rewrite their VM code (something that is not > happening anytime soon, see comment 8 in the first link), export these two > symbols. > > Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com> > --- > mm/vmalloc.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c > index 317afdc7d247..dffac67ba7f7 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -326,6 +326,7 @@ int map_kernel_range(unsigned long start, unsigned long size, pgprot_t prot, > flush_cache_vmap(start, start + size); > return ret; > } > +EXPORT_SYMBOL(map_kernel_range); > > int is_vmalloc_or_module_addr(const void *x) > { > @@ -2130,6 +2131,7 @@ struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags, > return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE, > GFP_KERNEL, caller); > } > +EXPORT_SYMBOL(__get_vm_area_caller); Please use EXPORT_SYMBOL_GPL instead. Seth
On Fri, Jul 31, 2020 at 5:58 PM Colin Ian King <colin.king@canonical.com> wrote: > > > I just spotted this: > > > https://sources.debian.org/src/virtualbox/6.1.12-dfsg-8/debian/patches/ > > https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-1.patch > > https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-2.patch > > https://sources.debian.org/data/contrib/v/virtualbox/6.1.12-dfsg-8/debian/patches/kernel-5.8-3.patch > > Maybe that is worth looking at. Yes, those are the 3 patches applied to the DKMS deb pkg (that we already have in groovy since it's synced from debian/unstable) - the kernel EXPORT patch is on top of those: virtualbox (6.1.12-dfsg-8) unstable; urgency=medium * debian/patches/kernel-5.8-1.patch: * debian/patches/kernel-5.8-2.patch: * debian/patches/kernel-5.8-3.patch: - apply upstream proposed patches to fix a build failure with kernel 5.8 (note, this requires also a kernel patch) https://www.virtualbox.org/ticket/19644 -- Gianfranco Costamagna <locutusofborg@debian.org> Thu, 30 Jul 2020 18:58:32 +0200 $ rmadison virtualbox-dkms ... virtualbox-dkms | 6.1.12-dfsg-8 | groovy/multiverse | amd64
On Fri, Jul 31, 2020 at 10:50 PM <seth.forshee@canonical.com> wrote: > > > Please use EXPORT_SYMBOL_GPL instead. Ack, i'll apply the patch with EXPORT_SYMBOL_GPL() instead.
diff --git a/mm/vmalloc.c b/mm/vmalloc.c index 317afdc7d247..dffac67ba7f7 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -326,6 +326,7 @@ int map_kernel_range(unsigned long start, unsigned long size, pgprot_t prot, flush_cache_vmap(start, start + size); return ret; } +EXPORT_SYMBOL(map_kernel_range); int is_vmalloc_or_module_addr(const void *x) { @@ -2130,6 +2131,7 @@ struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags, return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE, GFP_KERNEL, caller); } +EXPORT_SYMBOL(__get_vm_area_caller); /** * get_vm_area - reserve a contiguous kernel virtual area
As explained here: https://www.virtualbox.org/ticket/19644 and in the recent virtualbox DKMS upload changelog: https://metadata.ftp-master.debian.org/changelogs/contrib/v/virtualbox/virtualbox_6.1.12-dfsg-8_changelog Virtualbox DKMS now requires access to these two kernel functions to fix a FTBFS against Linux 5.8, so until they rewrite their VM code (something that is not happening anytime soon, see comment 8 in the first link), export these two symbols. Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com> --- mm/vmalloc.c | 2 ++ 1 file changed, 2 insertions(+)