There are regression reports[1][2] that crashkernel region on x86_64 can't
be added into iomem tree sometime. This causes the later failure of kdump
loading.

This happened after commit 4a693ce65b18 ("kdump: defer the insertion of
crashkernel resources") was merged.

Even though, these reported issues are proved to be related to other
component, they are just exposed after above commmit applied, I still
would like to keep crashk_res and crashk_low_res being added into iomem
early as before because the early adding has been always there on x86_64
and working very well. For safety of kdump, Let's change it back.

Here, add a macro HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY to limit that
only ARCH defining the macro can have the early adding
crashk_res/_low_res into iomem. Then define
HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY on x86 to enable it.

Note: In reserve_crashkernel_low(), there's a remnant of crashk_low_res
hanlding which was mistakenly added back in commit 85fcde402db1 ("kexec:
split crashkernel reservation code out from crash_core.c").

[1]
[PATCH V2] x86/kexec: do not update E820 kexec table for setup_data
https://lore.kernel.org/all/Zfv8iCL6CT2JqLIC@darkstar.users.ipa.redhat.com/T/#u

[2]
Question about Address Range Validation in Crash Kernel Allocation
https://lore.kernel.org/all/4eeac1f733584855965a2ea62fa4da58@huawei.com/T/#u

Signed-off-by: Baoquan He <bhe@redhat.com>
---
v1->v2:
- Remove the stray space before 'define' when defining
  HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY in asm/crash_reserve.h of x86.
  This is not suggested for standalone macro defines. Thanks to Ingo.

 arch/x86/include/asm/crash_reserve.h | 2 ++
 kernel/crash_reserve.c               | 7 +++++++
 2 files changed, 9 insertions(+)

diff --git a/arch/x86/include/asm/crash_reserve.h b/arch/x86/include/asm/crash_reserve.h
index 152239f95541..7835b2cdff04 100644
--- a/arch/x86/include/asm/crash_reserve.h
+++ b/arch/x86/include/asm/crash_reserve.h
@@ -39,4 +39,6 @@ static inline unsigned long crash_low_size_default(void)
 #endif
 }
 
+#define HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY
+
 #endif /* _X86_CRASH_RESERVE_H */
diff --git a/kernel/crash_reserve.c b/kernel/crash_reserve.c
index bbb6c3cb00e4..066668799f75 100644
--- a/kernel/crash_reserve.c
+++ b/kernel/crash_reserve.c
@@ -366,7 +366,9 @@ static int __init reserve_crashkernel_low(unsigned long long low_size)
 
 	crashk_low_res.start = low_base;
 	crashk_low_res.end   = low_base + low_size - 1;
+#ifdef HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY
 	insert_resource(&iomem_resource, &crashk_low_res);
+#endif
 #endif
 	return 0;
 }
@@ -448,8 +450,12 @@ void __init reserve_crashkernel_generic(char *cmdline,
 
 	crashk_res.start = crash_base;
 	crashk_res.end = crash_base + crash_size - 1;
+#ifdef HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY
+	insert_resource(&iomem_resource, &crashk_res);
+#endif
 }
 
+#ifndef HAVE_ARCH_ADD_CRASH_RES_TO_IOMEM_EARLY
 static __init int insert_crashkernel_resources(void)
 {
 	if (crashk_res.start < crashk_res.end)
@@ -462,3 +468,4 @@ static __init int insert_crashkernel_resources(void)
 }
 early_initcall(insert_crashkernel_resources);
 #endif
+#endif
-- 
2.41.0

On Mon, 25 Mar 2024 09:50:50 +0800 Baoquan He <bhe@redhat.com> wrote:

> There are regression reports[1][2] that crashkernel region on x86_64 can't
> be added into iomem tree sometime. This causes the later failure of kdump
> loading.

So I think a cc:stable is needed.

> This happened after commit 4a693ce65b18 ("kdump: defer the insertion of
> crashkernel resources") was merged.
> 
> Even though, these reported issues are proved to be related to other
> component, they are just exposed after above commmit applied, I still
> would like to keep crashk_res and crashk_low_res being added into iomem
> early as before because the early adding has been always there on x86_64
> and working very well. For safety of kdump, Let's change it back.

I'll use 4a693ce65b18 as the Fixes: target, since there is no
"Exposed-by-non-buggy-patch:" tag.  To tell the -stable tree
maintainers how far back in history this should be backported.

On 03/25/24 at 01:29pm, Andrew Morton wrote:
> On Mon, 25 Mar 2024 09:50:50 +0800 Baoquan He <bhe@redhat.com> wrote:
> 
> > There are regression reports[1][2] that crashkernel region on x86_64 can't
> > be added into iomem tree sometime. This causes the later failure of kdump
> > loading.
> 
> So I think a cc:stable is needed.

Yeah, I forgot this.

> 
> > This happened after commit 4a693ce65b18 ("kdump: defer the insertion of
> > crashkernel resources") was merged.
> > 
> > Even though, these reported issues are proved to be related to other
> > component, they are just exposed after above commmit applied, I still
> > would like to keep crashk_res and crashk_low_res being added into iomem
> > early as before because the early adding has been always there on x86_64
> > and working very well. For safety of kdump, Let's change it back.
> 
> I'll use 4a693ce65b18 as the Fixes: target, since there is no
> "Exposed-by-non-buggy-patch:" tag.  To tell the -stable tree
> maintainers how far back in history this should be backported.

Thanks a lot.