Do not map crash kernel area. This mapping is not in use anymore.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
---
xen/arch/x86/setup.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c
index f07ee2b..4833ca3 100644
--- a/xen/arch/x86/setup.c
+++ b/xen/arch/x86/setup.c
@@ -1097,9 +1097,7 @@ void __init __start_xen(unsigned long mbi_p)
mod[i].mod_start,
PFN_UP(mod[i].mod_end), PAGE_HYPERVISOR);
}
- map_pages_to_xen((unsigned long)__va(kexec_crash_area.start),
- kexec_crash_area.start >> PAGE_SHIFT,
- PFN_UP(kexec_crash_area.size), PAGE_HYPERVISOR);
+
xen_virt_end = ((unsigned long)_end + (1UL << L2_PAGETABLE_SHIFT) -
1) &
~((1UL << L2_PAGETABLE_SHIFT) - 1);
destroy_xen_mappings(xen_virt_end, XEN_VIRT_START + BOOTSTRAP_MAP_BASE);
--
1.7.10.4
On 09/12/2013 18:01, Daniel Kiper wrote:> Do not map crash kernel area. This mapping is not in use anymore. > > Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>Acked-by: David Vrabel <david.vrabel@citrix.com> However, this isn''t a bug fix so I don''t think it''s suitable 4.4 now. David
>>> On 09.12.13 at 19:16, David Vrabel <dvrabel@cantab.net> wrote: > On 09/12/2013 18:01, Daniel Kiper wrote: >> Do not map crash kernel area. This mapping is not in use anymore. >> >> Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com> > > Acked-by: David Vrabel <david.vrabel@citrix.com> > > However, this isn''t a bug fix so I don''t think it''s suitable 4.4 now.I think this is quite clearly a bug fix, taking into consideration our earlier discussion about systems with more memory (16Tb if the physical address map is mostly contiguous) than Xen can handle. Jan
>>> On 09.12.13 at 19:01, Daniel Kiper <daniel.kiper@oracle.com> wrote: > Do not map crash kernel area. This mapping is not in use anymore.That description to me reads like it was used before. Was it really? I think the issue of it possibly being outside of mappable memory has been in existence for quite a while, and hence with the description being worded in that way this would be a suitable trigger for backporting to the stable trees. Jan> Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com> > --- > xen/arch/x86/setup.c | 4 +--- > 1 file changed, 1 insertion(+), 3 deletions(-) > > diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c > index f07ee2b..4833ca3 100644 > --- a/xen/arch/x86/setup.c > +++ b/xen/arch/x86/setup.c > @@ -1097,9 +1097,7 @@ void __init __start_xen(unsigned long mbi_p) > mod[i].mod_start, > PFN_UP(mod[i].mod_end), PAGE_HYPERVISOR); > } > - map_pages_to_xen((unsigned long)__va(kexec_crash_area.start), > - kexec_crash_area.start >> PAGE_SHIFT, > - PFN_UP(kexec_crash_area.size), PAGE_HYPERVISOR); > + > xen_virt_end = ((unsigned long)_end + (1UL << L2_PAGETABLE_SHIFT) - 1) & > ~((1UL << L2_PAGETABLE_SHIFT) - 1); > destroy_xen_mappings(xen_virt_end, XEN_VIRT_START + > BOOTSTRAP_MAP_BASE); > -- > 1.7.10.4
On Tue, Dec 10, 2013 at 08:22:29AM +0000, Jan Beulich wrote:> >>> On 09.12.13 at 19:01, Daniel Kiper <daniel.kiper@oracle.com> wrote: > > Do not map crash kernel area. This mapping is not in use anymore. > > That description to me reads like it was used before. Was it really?Probably it was done deliberately but I could not find any references to this mapping in old kexec code or description in commit messages why it was done. Or maybe it was mistake...> I think the issue of it possibly being outside of mappable memory > has been in existence for quite a while, and hence with the > description being worded in that way this would be a suitable > trigger for backporting to the stable trees.I have done kexec/kdump tests on Xen 4.3.1 without that mapping and it works without any issue. Hence, I think that we could backport it safely at least to Xen 4.3. However, I suppose that it should also work for Xen 4.2. Daniel
On 10/12/13 08:19, Jan Beulich wrote:>>>> On 09.12.13 at 19:16, David Vrabel <dvrabel@cantab.net> wrote: >> On 09/12/2013 18:01, Daniel Kiper wrote: >>> Do not map crash kernel area. This mapping is not in use anymore. >>> >>> Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com> >> >> Acked-by: David Vrabel <david.vrabel@citrix.com> >> >> However, this isn''t a bug fix so I don''t think it''s suitable 4.4 now. > > I think this is quite clearly a bug fix, taking into consideration our > earlier discussion about systems with more memory (16Tb if the > physical address map is mostly contiguous) than Xen can handle.Oh yes, of course. I''d forgotten about that. David
>>> On 10.12.13 at 13:50, Daniel Kiper <daniel.kiper@oracle.com> wrote: > On Tue, Dec 10, 2013 at 08:22:29AM +0000, Jan Beulich wrote: >> >>> On 09.12.13 at 19:01, Daniel Kiper <daniel.kiper@oracle.com> wrote: >> > Do not map crash kernel area. This mapping is not in use anymore. >> >> That description to me reads like it was used before. Was it really? > > Probably it was done deliberately but I could not find any references > to this mapping in old kexec code or description in commit messages > why it was done. Or maybe it was mistake...Could I ask you then to resend with a more appropriate description? Jan