<feed xmlns='http://www.w3.org/2005/Atom'>
<title>linux-toradex.git/tools/testing/memblock/tests, branch master</title>
<subtitle>Linux kernel for Apalis and Colibri modules</subtitle>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/'/>
<entry>
<title>Merge patch series "memblock: Remove unused exact nid allocation API"</title>
<updated>2026-10-13T09:45:55+00:00</updated>
<author>
<name>Mike Rapoport (Microsoft)</name>
<email>rppt@kernel.org</email>
</author>
<published>2026-10-13T09:45:55+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=80b1dad1d6f535d85b5b8dcea6f3d598847a7e63'/>
<id>80b1dad1d6f535d85b5b8dcea6f3d598847a7e63</id>
<content type='text'>
Kaitao Cheng &lt;kaitao.cheng@linux.dev&gt; says:

memblock: Remove unused exact nid allocation API

Commit ffe55393137c ("mm/sparse: remove sparse buffer pre-allocation
mechanism") removed the only memmap_alloc() caller that requested an
allocation from an exact NUMA node. All remaining callers allow fallback
to other nodes, leaving the exact_nid argument unused.

Remove the argument from memmap_alloc(). This leaves
memblock_alloc_exact_nid_raw() without any in-kernel callers, so move the
applicable exact-node allocation tests to memblock_alloc_range_nid() and
remove the unused API. Also remove the exact_nid argument from the internal
allocation helper, since all of its remaining callers allow fallback.

Exact-node allocation remains a required behavior of
memblock_alloc_range_nid(). CMA uses it when reserving per-node CMA areas;
allowing fallback could leave the area's recorded nid inconsistent with
the physical location of its memory. KHO also uses it for per-node scratch
areas that must remain on the requested node. Therefore, retain coverage
of this behavior by making the applicable NUMA tests call
memblock_alloc_range_nid() directly with exact_nid set.

The tests that exercise memblock_alloc_internal() retrying below min_addr
cannot be retained unchanged. Unlike memblock_alloc_internal(),
memblock_alloc_range_nid() treats its start address as a hard boundary and
does not retry from address zero. Remove those tests along with the now
unused TEST_F_EXACT dispatch from the shared allocation tests.

* patches from https://patch.msgid.link/20260919135327.74089-1-kaitao.cheng@linux.dev
  mm: Remove unused exact_nid parameter from memmap_alloc()
  memblock tests: Move exact nid tests to memblock_alloc_range_nid()
  memblock: Remove unused memblock_alloc_exact_nid_raw()

Link: https://patch.msgid.link/20260919135327.74089-1-kaitao.cheng@linux.dev
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Kaitao Cheng &lt;kaitao.cheng@linux.dev&gt; says:

memblock: Remove unused exact nid allocation API

Commit ffe55393137c ("mm/sparse: remove sparse buffer pre-allocation
mechanism") removed the only memmap_alloc() caller that requested an
allocation from an exact NUMA node. All remaining callers allow fallback
to other nodes, leaving the exact_nid argument unused.

Remove the argument from memmap_alloc(). This leaves
memblock_alloc_exact_nid_raw() without any in-kernel callers, so move the
applicable exact-node allocation tests to memblock_alloc_range_nid() and
remove the unused API. Also remove the exact_nid argument from the internal
allocation helper, since all of its remaining callers allow fallback.

Exact-node allocation remains a required behavior of
memblock_alloc_range_nid(). CMA uses it when reserving per-node CMA areas;
allowing fallback could leave the area's recorded nid inconsistent with
the physical location of its memory. KHO also uses it for per-node scratch
areas that must remain on the requested node. Therefore, retain coverage
of this behavior by making the applicable NUMA tests call
memblock_alloc_range_nid() directly with exact_nid set.

The tests that exercise memblock_alloc_internal() retrying below min_addr
cannot be retained unchanged. Unlike memblock_alloc_internal(),
memblock_alloc_range_nid() treats its start address as a hard boundary and
does not retry from address zero. Remove those tests along with the now
unused TEST_F_EXACT dispatch from the shared allocation tests.

* patches from https://patch.msgid.link/20260919135327.74089-1-kaitao.cheng@linux.dev
  mm: Remove unused exact_nid parameter from memmap_alloc()
  memblock tests: Move exact nid tests to memblock_alloc_range_nid()
  memblock: Remove unused memblock_alloc_exact_nid_raw()

Link: https://patch.msgid.link/20260919135327.74089-1-kaitao.cheng@linux.dev
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: Move exact nid tests to memblock_alloc_range_nid()</title>
<updated>2026-10-13T09:45:55+00:00</updated>
<author>
<name>Kaitao Cheng</name>
<email>chengkaitao@kylinos.cn</email>
</author>
<published>2026-09-19T13:53:26+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=fba3e86fb848f79c4573243affcd7d79d378958d'/>
<id>fba3e86fb848f79c4573243affcd7d79d378958d</id>
<content type='text'>
After memmap_alloc() stopped requesting allocations from an exact NUMA
node, memblock_alloc_exact_nid_raw() no longer has any in-kernel users
and is being removed.

Preserve coverage of exact-node allocation by converting the applicable
NUMA tests to call memblock_alloc_range_nid() with exact_nid set. Remove
the tests that depend on memblock_alloc_internal() retrying below
min_addr, as memblock_alloc_range_nid() treats its address range as a
hard constraint.

Remove the now-unused TEST_F_EXACT support from the shared allocation
tests.

Suggested-by: Lorenzo Stoakes &lt;ljs@kernel.org&gt;
Signed-off-by: Kaitao Cheng &lt;chengkaitao@kylinos.cn&gt;
Link: https://patch.msgid.link/20260919135327.74089-3-kaitao.cheng@linux.dev
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
After memmap_alloc() stopped requesting allocations from an exact NUMA
node, memblock_alloc_exact_nid_raw() no longer has any in-kernel users
and is being removed.

Preserve coverage of exact-node allocation by converting the applicable
NUMA tests to call memblock_alloc_range_nid() with exact_nid set. Remove
the tests that depend on memblock_alloc_internal() retrying below
min_addr, as memblock_alloc_range_nid() treats its address range as a
hard constraint.

Remove the now-unused TEST_F_EXACT support from the shared allocation
tests.

Suggested-by: Lorenzo Stoakes &lt;ljs@kernel.org&gt;
Signed-off-by: Kaitao Cheng &lt;chengkaitao@kylinos.cn&gt;
Link: https://patch.msgid.link/20260919135327.74089-3-kaitao.cheng@linux.dev
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: zero-extend simulated memory addresses</title>
<updated>2026-09-24T07:57:03+00:00</updated>
<author>
<name>Tianyi Chen</name>
<email>hi@tychen.cc</email>
</author>
<published>2026-09-23T02:31:24+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=6957416ae43de1830174f527b5b8cf737401a8dd'/>
<id>6957416ae43de1830174f527b5b8cf737401a8dd</id>
<content type='text'>
Converting a 32-bit pointer directly to the 64-bit phys_addr_t can
sign-extend it when BUILD=32 is used. A dummy allocation above 2 GiB
then gets registered at an address different from its unsigned pointer
value. The low-address tests compare that value against the registered
range and fail once the earlier maximum-address free test is fixed.

Convert through uintptr_t when registering and reporting dummy memory.
This preserves the pointer bits and agrees with the simulator __pa()
conversion and the allocation tests.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/20260923023126.1910311-3-hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Converting a 32-bit pointer directly to the 64-bit phys_addr_t can
sign-extend it when BUILD=32 is used. A dummy allocation above 2 GiB
then gets registered at an address different from its unsigned pointer
value. The low-address tests compare that value against the registered
range and fail once the earlier maximum-address free test is fixed.

Convert through uintptr_t when registering and reporting dummy memory.
This preserves the pointer bits and agrees with the simulator __pa()
conversion and the allocation tests.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/20260923023126.1910311-3-hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: use physical addresses for region removal</title>
<updated>2026-09-24T07:56:52+00:00</updated>
<author>
<name>Tianyi Chen</name>
<email>hi@tychen.cc</email>
</author>
<published>2026-09-23T02:31:23+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=9357596afac61154e7f54048c0add9f1293b9eba'/>
<id>9357596afac61154e7f54048c0add9f1293b9eba</id>
<content type='text'>
With BUILD=32, phys_addr_t remains 64-bit while pointers are 32-bit.
Casting the address near PHYS_ADDR_MAX to void * truncates it before
memblock_free() converts it back with __pa(). The reserved region is
therefore left unchanged and memblock_free_near_max_check() fails.

Use memblock_phys_free() for all synthetic physical ranges in the basic
API tests. This also removes the integer-to-pointer cast warnings from
the other removal checks. Retain the checks that freeing past
PHYS_ADDR_MAX clips the removed region.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/20260923023126.1910311-2-hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
With BUILD=32, phys_addr_t remains 64-bit while pointers are 32-bit.
Casting the address near PHYS_ADDR_MAX to void * truncates it before
memblock_free() converts it back with __pa(). The reserved region is
therefore left unchanged and memblock_free_near_max_check() fails.

Use memblock_phys_free() for all synthetic physical ranges in the basic
API tests. This also removes the integer-to-pointer cast warnings from
the other removal checks. Retain the checks that freeing past
PHYS_ADDR_MAX clips the removed region.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/20260923023126.1910311-2-hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: cover allocations below the low address limit</title>
<updated>2026-09-10T10:10:20+00:00</updated>
<author>
<name>Tianyi Chen</name>
<email>hi@tychen.cc</email>
</author>
<published>2026-09-09T23:55:33+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=2345d7e354e84c140de225022a38754eab499a94'/>
<id>2345d7e354e84c140de225022a38754eab499a94</id>
<content type='text'>
Add memblock_alloc_low() coverage using the simulator's low address
limit. Exercise aligned allocation, an allocation whose last byte is
immediately below the limit, an alignment constraint that prevents
fitting below it, and fully reserved low memory with high memory still
available.

Run each case with bottom-up and top-down allocation. Check zeroing
and reserved-region accounting as well as returned addresses. Verify
that an unrestricted allocation can use the free high memory after
the low allocation fails.

Document the simulated limit and remove the completed TODO.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/142ab1025fd1e4b5dcd2a1591b964626ebce9d23.1788997523.git.hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Add memblock_alloc_low() coverage using the simulator's low address
limit. Exercise aligned allocation, an allocation whose last byte is
immediately below the limit, an alignment constraint that prevents
fitting below it, and fully reserved low memory with high memory still
available.

Run each case with bottom-up and top-down allocation. Check zeroing
and reserved-region accounting as well as returned addresses. Verify
that an unrestricted allocation can use the free high memory after
the low allocation fails.

Document the simulated limit and remove the completed TODO.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/142ab1025fd1e4b5dcd2a1591b964626ebce9d23.1788997523.git.hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: model the low allocation limit within dummy memory</title>
<updated>2026-09-10T10:10:20+00:00</updated>
<author>
<name>Tianyi Chen</name>
<email>hi@tychen.cc</email>
</author>
<published>2026-09-09T23:55:28+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=f68c01000a6c3bddbf415b6e50a201196ec78358'/>
<id>f68c01000a6c3bddbf415b6e50a201196ec78358</id>
<content type='text'>
The simulator represents physical addresses using the address of its
allocated buffer. That buffer can lie above the default
ARCH_LOW_ADDRESS_LIMIT, preventing meaningful memblock_alloc_low()
tests.

Use the architecture override in asm/dma.h to place the limit halfway
through the memory registered by setup_memblock(). Resolve the limit
after allocating the buffer, leaving registered memory on both sides.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/f16c272cd52316b83c35d87987b4d8687c4a3182.1788997523.git.hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The simulator represents physical addresses using the address of its
allocated buffer. That buffer can lie above the default
ARCH_LOW_ADDRESS_LIMIT, preventing meaningful memblock_alloc_low()
tests.

Use the architecture override in asm/dma.h to place the limit halfway
through the memory registered by setup_memblock(). Resolve the limit
after allocating the buffer, leaving registered memory on both sides.

Assisted-by: LLM
Signed-off-by: Tianyi Chen &lt;hi@tychen.cc&gt;
Link: https://patch.msgid.link/f16c272cd52316b83c35d87987b4d8687c4a3182.1788997523.git.hi@tychen.cc
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>tools/testing/memblock: fix stale NUMA reservation tests</title>
<updated>2026-06-02T05:34:03+00:00</updated>
<author>
<name>Priyanshu Kumar</name>
<email>priyanshukumarpu@gmail.com</email>
</author>
<published>2026-04-15T12:27:31+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=9588076fb21992a5d14efeb99134ebf034c7b84f'/>
<id>9588076fb21992a5d14efeb99134ebf034c7b84f</id>
<content type='text'>
memblock allocations now reserve memory with MEMBLOCK_RSRV_KERN and,
on NUMA configurations, record the requested node on the reserved
region. Several memblock simulator NUMA tests still expected merges
that only worked before those reservation semantics changed, so the
suite aborted even though the allocator behavior was correct.

Update the NUMA merge expectations in the memblock_alloc_try_nid()
and memblock_alloc_exact_nid_raw() tests to match the current reserved
region metadata rules. For cases that should still merge, create the
pre-existing reservation with matching nid and MEMBLOCK_RSRV_KERN
metadata. Also strengthen the memblock_alloc_node() coverage by
checking the newly created reserved region directly instead of
re-reading the source memory node descriptor.

Finally, drop the stale README/TODO notes that still claimed
memblock_alloc_node() could not be tested.

The memblock simulator passes again with NUMA enabled after these
updates.

Signed-off-by: Priyanshu Kumar &lt;priyanshukumarpu@gmail.com&gt;
Link: https://patch.msgid.link/20260415122731.1768912-1-priyanshukumarpu@gmail.com
[rppt: dropped unrelated changes]
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
memblock allocations now reserve memory with MEMBLOCK_RSRV_KERN and,
on NUMA configurations, record the requested node on the reserved
region. Several memblock simulator NUMA tests still expected merges
that only worked before those reservation semantics changed, so the
suite aborted even though the allocator behavior was correct.

Update the NUMA merge expectations in the memblock_alloc_try_nid()
and memblock_alloc_exact_nid_raw() tests to match the current reserved
region metadata rules. For cases that should still merge, create the
pre-existing reservation with matching nid and MEMBLOCK_RSRV_KERN
metadata. Also strengthen the memblock_alloc_node() coverage by
checking the newly created reserved region directly instead of
re-reading the source memory node descriptor.

Finally, drop the stale README/TODO notes that still claimed
memblock_alloc_node() could not be tested.

The memblock simulator passes again with NUMA enabled after these
updates.

Signed-off-by: Priyanshu Kumar &lt;priyanshukumarpu@gmail.com&gt;
Link: https://patch.msgid.link/20260415122731.1768912-1-priyanshukumarpu@gmail.com
[rppt: dropped unrelated changes]
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock: add MEMBLOCK_RSRV_KERN flag</title>
<updated>2025-05-13T06:50:38+00:00</updated>
<author>
<name>Mike Rapoport (Microsoft)</name>
<email>rppt@kernel.org</email>
</author>
<published>2025-05-09T07:46:19+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=4c78cc596bb8d39532f059e0198eeabf370c50f5'/>
<id>4c78cc596bb8d39532f059e0198eeabf370c50f5</id>
<content type='text'>
Patch series "kexec: introduce Kexec HandOver (KHO)", v8.

Kexec today considers itself purely a boot loader: When we enter the new
kernel, any state the previous kernel left behind is irrelevant and the
new kernel reinitializes the system.

However, there are use cases where this mode of operation is not what we
actually want.  In virtualization hosts for example, we want to use kexec
to update the host kernel while virtual machine memory stays untouched. 
When we add device assignment to the mix, we also need to ensure that
IOMMU and VFIO states are untouched.  If we add PCIe peer to peer DMA, we
need to do the same for the PCI subsystem.  If we want to kexec while an
SEV-SNP enabled virtual machine is running, we need to preserve the VM
context pages and physical memory.  See "pkernfs: Persisting guest memory
and kernel/device state safely across kexec" Linux Plumbers Conference
2023 presentation for details:

  https://lpc.events/event/17/contributions/1485/

To start us on the journey to support all the use cases above, this patch
implements basic infrastructure to allow hand over of kernel state across
kexec (Kexec HandOver, aka KHO).  As a really simple example target, we
use memblock's reserve_mem.

With this patchset applied, memory that was reserved using "reserve_mem"
command line options remains intact after kexec and it is guaranteed to
reside at the same physical address.

== Alternatives ==

There are alternative approaches to (parts of) the problems above:

  * Memory Pools [1] - preallocated persistent memory region + allocator
  * PRMEM [2] - resizable persistent memory regions with fixed metadata
                pointer on the kernel command line + allocator
  * Pkernfs [3] - preallocated file system for in-kernel data with fixed
                  address location on the kernel command line
  * PKRAM [4] - handover of user space pages using a fixed metadata page
                specified via command line

All of the approaches above fundamentally have the same problem: They
require the administrator to explicitly carve out a physical memory
location because they have no mechanism outside of the kernel command line
to pass data (including memory reservations) between kexec'ing kernels.

KHO provides that base foundation.  We will determine later whether we
still need any of the approaches above for fast bulk memory handover of
for example IOMMU page tables.  But IMHO they would all be users of KHO,
with KHO providing the foundational primitive to pass metadata and bulk
memory reservations as well as provide easy versioning for data.

== Overview ==

We introduce a metadata file that the kernels pass between each other. 
How they pass it is architecture specific.  The file's format is a
Flattened Device Tree (fdt) which has a generator and parser already
included in Linux.  KHO is enabled in the kernel command line by `kho=on`.
When the root user enables KHO through
/sys/kernel/debug/kho/out/finalize, the kernel invokes callbacks to every
KHO users to register preserved memory regions, which contain drivers'
states.

When the actual kexec happens, the fdt is part of the image set that we
boot into.  In addition, we keep "scratch regions" available for kexec:
physically contiguous memory regions that are guaranteed to not have any
memory that KHO would preserve.  The new kernel bootstraps itself using
the scratch regions and sets all handed over memory as in use.  When
drivers initialize that support KHO, they introspect the fdt, restore
preserved memory regions, and retrieve their states stored in the
preserved memory.

== Limitations ==

Currently KHO is only implemented for file based kexec.  The kernel
interfaces in the patch set are already in place to support user space
kexec as well, but it is still not implemented it yet inside kexec tools.

== How to Use ==

To use the code, please boot the kernel with the "kho=on" command line
parameter.  KHO will automatically create scratch regions.  If you want to
set the scratch size explicitly you can use "kho_scratch=" command line
parameter.  For instance, "kho_scratch=16M,512M,256M" will reserve a 16
MiB low memory scratch area, a 512 MiB global scratch region, and 256 MiB
per NUMA node scratch regions on boot.

Make sure to have a reserved memory range requested with reserv_mem
command line option, for example, "reserve_mem=64m:4k:n1".

Then before you invoke file based "kexec -l", finalize KHO FDT:

  # echo 1 &gt; /sys/kernel/debug/kho/out/finalize

You can preview the generated FDT using `dtc`,

  # dtc /sys/kernel/debug/kho/out/fdt
  # dtc /sys/kernel/debug/kho/out/sub_fdts/memblock

`dtc` is available on ubuntu by `sudo apt-get install device-tree-compiler`.

Now kexec into the new kernel,

  # kexec -l Image --initrd=initrd -s
  # kexec -e

(The order of KHO finalization and "kexec -l" does not matter.)

The new kernel will boot up and contain the previous kernel's reserve_mem
contents at the same physical address as the first kernel.

You can also review the FDT passed from the old kernel,

  # dtc /sys/kernel/debug/kho/in/fdt
  # dtc /sys/kernel/debug/kho/in/sub_fdts/memblock


This patch (of 17):

To denote areas that were reserved for kernel use either directly with
memblock_reserve_kern() or via memblock allocations.

Link: https://lore.kernel.org/lkml/20250424083258.2228122-1-changyuanl@google.com/
Link: https://lore.kernel.org/lkml/aAeaJ2iqkrv_ffhT@kernel.org/
Link: https://lore.kernel.org/lkml/35c58191-f774-40cf-8d66-d1e2aaf11a62@intel.com/
Link: https://lore.kernel.org/lkml/20250424093302.3894961-1-arnd@kernel.org/
Link: https://lkml.kernel.org/r/20250509074635.3187114-1-changyuanl@google.com
Link: https://lkml.kernel.org/r/20250509074635.3187114-2-changyuanl@google.com
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
Co-developed-by: Changyuan Lyu &lt;changyuanl@google.com&gt;
Signed-off-by: Changyuan Lyu &lt;changyuanl@google.com&gt;
Cc: Alexander Graf &lt;graf@amazon.com&gt;
Cc: Andy Lutomirski &lt;luto@kernel.org&gt;
Cc: Anthony Yznaga &lt;anthony.yznaga@oracle.com&gt;
Cc: Arnd Bergmann &lt;arnd@arndb.de&gt;
Cc: Ashish Kalra &lt;ashish.kalra@amd.com&gt;
Cc: Ben Herrenschmidt &lt;benh@kernel.crashing.org&gt;
Cc: Borislav Betkov &lt;bp@alien8.de&gt;
Cc: Catalin Marinas &lt;catalin.marinas@arm.com&gt;
Cc: David Woodhouse &lt;dwmw2@infradead.org&gt;
Cc: Eric Biederman &lt;ebiederm@xmission.com&gt;
Cc: "H. Peter Anvin" &lt;hpa@zytor.com&gt;
Cc: Ingo Molnar &lt;mingo@redhat.com&gt;
Cc: James Gowans &lt;jgowans@amazon.com&gt;
Cc: Jonathan Corbet &lt;corbet@lwn.net&gt;
Cc: Krzysztof Kozlowski &lt;krzk@kernel.org&gt;
Cc: Marc Rutland &lt;mark.rutland@arm.com&gt;
Cc: Paolo Bonzini &lt;pbonzini@redhat.com&gt;
Cc: Pasha Tatashin &lt;pasha.tatashin@soleen.com&gt;
Cc: Peter Zijlstra &lt;peterz@infradead.org&gt;
Cc: Pratyush Yadav &lt;ptyadav@amazon.de&gt;
Cc: Rob Herring &lt;robh@kernel.org&gt;
Cc: Saravana Kannan &lt;saravanak@google.com&gt;
Cc: Stanislav Kinsburskii &lt;skinsburskii@linux.microsoft.com&gt;
Cc: Steven Rostedt &lt;rostedt@goodmis.org&gt;
Cc: Thomas Gleinxer &lt;tglx@linutronix.de&gt;
Cc: Thomas Lendacky &lt;thomas.lendacky@amd.com&gt;
Cc: Will Deacon &lt;will@kernel.org&gt;
Cc: Dave Hansen &lt;dave.hansen@linux.intel.com&gt;
Cc: Jason Gunthorpe &lt;jgg@nvidia.com&gt;
Signed-off-by: Andrew Morton &lt;akpm@linux-foundation.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Patch series "kexec: introduce Kexec HandOver (KHO)", v8.

Kexec today considers itself purely a boot loader: When we enter the new
kernel, any state the previous kernel left behind is irrelevant and the
new kernel reinitializes the system.

However, there are use cases where this mode of operation is not what we
actually want.  In virtualization hosts for example, we want to use kexec
to update the host kernel while virtual machine memory stays untouched. 
When we add device assignment to the mix, we also need to ensure that
IOMMU and VFIO states are untouched.  If we add PCIe peer to peer DMA, we
need to do the same for the PCI subsystem.  If we want to kexec while an
SEV-SNP enabled virtual machine is running, we need to preserve the VM
context pages and physical memory.  See "pkernfs: Persisting guest memory
and kernel/device state safely across kexec" Linux Plumbers Conference
2023 presentation for details:

  https://lpc.events/event/17/contributions/1485/

To start us on the journey to support all the use cases above, this patch
implements basic infrastructure to allow hand over of kernel state across
kexec (Kexec HandOver, aka KHO).  As a really simple example target, we
use memblock's reserve_mem.

With this patchset applied, memory that was reserved using "reserve_mem"
command line options remains intact after kexec and it is guaranteed to
reside at the same physical address.

== Alternatives ==

There are alternative approaches to (parts of) the problems above:

  * Memory Pools [1] - preallocated persistent memory region + allocator
  * PRMEM [2] - resizable persistent memory regions with fixed metadata
                pointer on the kernel command line + allocator
  * Pkernfs [3] - preallocated file system for in-kernel data with fixed
                  address location on the kernel command line
  * PKRAM [4] - handover of user space pages using a fixed metadata page
                specified via command line

All of the approaches above fundamentally have the same problem: They
require the administrator to explicitly carve out a physical memory
location because they have no mechanism outside of the kernel command line
to pass data (including memory reservations) between kexec'ing kernels.

KHO provides that base foundation.  We will determine later whether we
still need any of the approaches above for fast bulk memory handover of
for example IOMMU page tables.  But IMHO they would all be users of KHO,
with KHO providing the foundational primitive to pass metadata and bulk
memory reservations as well as provide easy versioning for data.

== Overview ==

We introduce a metadata file that the kernels pass between each other. 
How they pass it is architecture specific.  The file's format is a
Flattened Device Tree (fdt) which has a generator and parser already
included in Linux.  KHO is enabled in the kernel command line by `kho=on`.
When the root user enables KHO through
/sys/kernel/debug/kho/out/finalize, the kernel invokes callbacks to every
KHO users to register preserved memory regions, which contain drivers'
states.

When the actual kexec happens, the fdt is part of the image set that we
boot into.  In addition, we keep "scratch regions" available for kexec:
physically contiguous memory regions that are guaranteed to not have any
memory that KHO would preserve.  The new kernel bootstraps itself using
the scratch regions and sets all handed over memory as in use.  When
drivers initialize that support KHO, they introspect the fdt, restore
preserved memory regions, and retrieve their states stored in the
preserved memory.

== Limitations ==

Currently KHO is only implemented for file based kexec.  The kernel
interfaces in the patch set are already in place to support user space
kexec as well, but it is still not implemented it yet inside kexec tools.

== How to Use ==

To use the code, please boot the kernel with the "kho=on" command line
parameter.  KHO will automatically create scratch regions.  If you want to
set the scratch size explicitly you can use "kho_scratch=" command line
parameter.  For instance, "kho_scratch=16M,512M,256M" will reserve a 16
MiB low memory scratch area, a 512 MiB global scratch region, and 256 MiB
per NUMA node scratch regions on boot.

Make sure to have a reserved memory range requested with reserv_mem
command line option, for example, "reserve_mem=64m:4k:n1".

Then before you invoke file based "kexec -l", finalize KHO FDT:

  # echo 1 &gt; /sys/kernel/debug/kho/out/finalize

You can preview the generated FDT using `dtc`,

  # dtc /sys/kernel/debug/kho/out/fdt
  # dtc /sys/kernel/debug/kho/out/sub_fdts/memblock

`dtc` is available on ubuntu by `sudo apt-get install device-tree-compiler`.

Now kexec into the new kernel,

  # kexec -l Image --initrd=initrd -s
  # kexec -e

(The order of KHO finalization and "kexec -l" does not matter.)

The new kernel will boot up and contain the previous kernel's reserve_mem
contents at the same physical address as the first kernel.

You can also review the FDT passed from the old kernel,

  # dtc /sys/kernel/debug/kho/in/fdt
  # dtc /sys/kernel/debug/kho/in/sub_fdts/memblock


This patch (of 17):

To denote areas that were reserved for kernel use either directly with
memblock_reserve_kern() or via memblock allocations.

Link: https://lore.kernel.org/lkml/20250424083258.2228122-1-changyuanl@google.com/
Link: https://lore.kernel.org/lkml/aAeaJ2iqkrv_ffhT@kernel.org/
Link: https://lore.kernel.org/lkml/35c58191-f774-40cf-8d66-d1e2aaf11a62@intel.com/
Link: https://lore.kernel.org/lkml/20250424093302.3894961-1-arnd@kernel.org/
Link: https://lkml.kernel.org/r/20250509074635.3187114-1-changyuanl@google.com
Link: https://lkml.kernel.org/r/20250509074635.3187114-2-changyuanl@google.com
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
Co-developed-by: Changyuan Lyu &lt;changyuanl@google.com&gt;
Signed-off-by: Changyuan Lyu &lt;changyuanl@google.com&gt;
Cc: Alexander Graf &lt;graf@amazon.com&gt;
Cc: Andy Lutomirski &lt;luto@kernel.org&gt;
Cc: Anthony Yznaga &lt;anthony.yznaga@oracle.com&gt;
Cc: Arnd Bergmann &lt;arnd@arndb.de&gt;
Cc: Ashish Kalra &lt;ashish.kalra@amd.com&gt;
Cc: Ben Herrenschmidt &lt;benh@kernel.crashing.org&gt;
Cc: Borislav Betkov &lt;bp@alien8.de&gt;
Cc: Catalin Marinas &lt;catalin.marinas@arm.com&gt;
Cc: David Woodhouse &lt;dwmw2@infradead.org&gt;
Cc: Eric Biederman &lt;ebiederm@xmission.com&gt;
Cc: "H. Peter Anvin" &lt;hpa@zytor.com&gt;
Cc: Ingo Molnar &lt;mingo@redhat.com&gt;
Cc: James Gowans &lt;jgowans@amazon.com&gt;
Cc: Jonathan Corbet &lt;corbet@lwn.net&gt;
Cc: Krzysztof Kozlowski &lt;krzk@kernel.org&gt;
Cc: Marc Rutland &lt;mark.rutland@arm.com&gt;
Cc: Paolo Bonzini &lt;pbonzini@redhat.com&gt;
Cc: Pasha Tatashin &lt;pasha.tatashin@soleen.com&gt;
Cc: Peter Zijlstra &lt;peterz@infradead.org&gt;
Cc: Pratyush Yadav &lt;ptyadav@amazon.de&gt;
Cc: Rob Herring &lt;robh@kernel.org&gt;
Cc: Saravana Kannan &lt;saravanak@google.com&gt;
Cc: Stanislav Kinsburskii &lt;skinsburskii@linux.microsoft.com&gt;
Cc: Steven Rostedt &lt;rostedt@goodmis.org&gt;
Cc: Thomas Gleinxer &lt;tglx@linutronix.de&gt;
Cc: Thomas Lendacky &lt;thomas.lendacky@amd.com&gt;
Cc: Will Deacon &lt;will@kernel.org&gt;
Cc: Dave Hansen &lt;dave.hansen@linux.intel.com&gt;
Cc: Jason Gunthorpe &lt;jgg@nvidia.com&gt;
Signed-off-by: Andrew Morton &lt;akpm@linux-foundation.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: add test for memblock_set_node</title>
<updated>2025-04-07T06:28:01+00:00</updated>
<author>
<name>Wei Yang</name>
<email>richard.weiyang@gmail.com</email>
</author>
<published>2025-03-18T07:19:48+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=3b394dff15e14550a26b133fc7b556b5b526f6a5'/>
<id>3b394dff15e14550a26b133fc7b556b5b526f6a5</id>
<content type='text'>
Add a test to check memblock_set_node() behavior.

And create a corner case in which the memblock.reserved array is doubled
during memblock_set_node(). And finally make sure all regions in
memblock.reserved are with valid node id.

Signed-off-by: Wei Yang &lt;richard.weiyang@gmail.com&gt;
CC: Mike Rapoport &lt;rppt@kernel.org&gt;
CC: Yajun Deng &lt;yajun.deng@linux.dev&gt;
Link: https://lore.kernel.org/r/20250318071948.23854-4-richard.weiyang@gmail.com
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Add a test to check memblock_set_node() behavior.

And create a corner case in which the memblock.reserved array is doubled
during memblock_set_node(). And finally make sure all regions in
memblock.reserved are with valid node id.

Signed-off-by: Wei Yang &lt;richard.weiyang@gmail.com&gt;
CC: Mike Rapoport &lt;rppt@kernel.org&gt;
CC: Yajun Deng &lt;yajun.deng@linux.dev&gt;
Link: https://lore.kernel.org/r/20250318071948.23854-4-richard.weiyang@gmail.com
Signed-off-by: Mike Rapoport (Microsoft) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>memblock tests: add memblock_overlaps_region_checks</title>
<updated>2024-06-05T07:22:22+00:00</updated>
<author>
<name>Wei Yang</name>
<email>richard.weiyang@gmail.com</email>
</author>
<published>2024-05-07T07:58:30+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=1a879671bdfd14698a839f30de8e6d76e1e858fd'/>
<id>1a879671bdfd14698a839f30de8e6d76e1e858fd</id>
<content type='text'>
Add a test case for memblock_overlaps_region().

Signed-off-by: Wei Yang &lt;richard.weiyang@gmail.com&gt;
Link: https://lore.kernel.org/r/20240507075833.6346-5-richard.weiyang@gmail.com
Signed-off-by: Mike Rapoport (IBM) &lt;rppt@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Add a test case for memblock_overlaps_region().

Signed-off-by: Wei Yang &lt;richard.weiyang@gmail.com&gt;
Link: https://lore.kernel.org/r/20240507075833.6346-5-richard.weiyang@gmail.com
Signed-off-by: Mike Rapoport (IBM) &lt;rppt@kernel.org&gt;
</pre>
</div>
</content>
</entry>
</feed>
