[PATCH V3 (RESEND) 0/3] arm64: Enable vmemmap mapping from device memory

This series enables vmemmap backing memory allocation from device memory
ranges on arm64. But before that, it enables vmemmap_populate_basepages()
and vmemmap_alloc_block_buf() to accommodate struct vmem_altmap based
alocation requests.

This series applies on 5.8-rc1.

Pending Question:

altmap_alloc_block_buf() does not have any other remaining users in the
tree after this change. Should it be converted into a static function and
it's declaration be dropped from the header (include/linux/mm.h). Avoided
doing so because I was not sure if there are any off-tree users or not.

Changes in V3:

- Dropped comment from free_hotplug_page_range() per Robin
- Modified comment in unmap_hotplug_range() per Robin
- Enabled altmap support in vmemmap_alloc_block_buf() per Robin

Changes in V2: (https://lkml.org/lkml/2020/3/4/475)

- Rebased on latest hot-remove series (v14) adding P4D page table support

Changes in V1: (https://lkml.org/lkml/2020/1/23/12)

- Added an WARN_ON() in unmap_hotplug_range() when altmap is
  provided without the page table backing memory being freed

Changes in RFC V2: (https://lkml.org/lkml/2019/10/21/11)

- Changed the commit message on 1/2 patch per Will
- Changed the commit message on 2/2 patch as well
- Rebased on arm64 memory hot remove series (v10)

RFC V1: (https://lkml.org/lkml/2019/6/28/32)

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Paul Walmsley <redacted>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <redacted>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <redacted>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: David Hildenbrand <redacted>
Cc: Mike Rapoport <redacted>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: "Kirill A. Shutemov" <redacted>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <redacted>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org

Anshuman Khandual (3):
  mm/sparsemem: Enable vmem_altmap support in vmemmap_populate_basepages()
  mm/sparsemem: Enable vmem_altmap support in vmemmap_alloc_block_buf()
  arm64/mm: Enable vmem_altmap support for vmemmap mappings

 arch/arm64/mm/mmu.c       | 59 ++++++++++++++++++++++++++-------------
 arch/ia64/mm/discontig.c  |  2 +-
 arch/powerpc/mm/init_64.c | 10 +++----
 arch/riscv/mm/init.c      |  2 +-
 arch/x86/mm/init_64.c     | 12 ++++----
 include/linux/mm.h        |  8 ++++--
 mm/sparse-vmemmap.c       | 38 ++++++++++++++++++++-----
 7 files changed, 87 insertions(+), 44 deletions(-)

-- 
2.20.1

There are many instances where vmemap allocation is often switched between
regular memory and device memory just based on whether altmap is available
or not. vmemmap_alloc_block_buf() is used in various platforms to allocate
vmemmap mappings. Lets also enable it to handle altmap based device memory
allocation along with existing regular memory allocations. This will help
in avoiding the altmap based allocation switch in many places.

While here also implement a regular memory allocation fallback mechanism
when the first preferred device memory allocation fails. This will ensure
preserving the existing semantics on powerpc platform. To summarize there
are three different methods to call vmemmap_alloc_block_buf().

(., NULL,   false) /* Allocate from system RAM */
(., altmap, false) /* Allocate from altmap without any fallback */
(., altmap, true)  /* Allocate from altmap with fallback (system RAM) */

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <redacted>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: x86@kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org
Tested-by: Jia He <redacted>
Suggested-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Anshuman Khandual <redacted>
---
 arch/arm64/mm/mmu.c       |  3 ++-
 arch/powerpc/mm/init_64.c | 10 +++++-----
 arch/x86/mm/init_64.c     |  6 ++----
 include/linux/mm.h        |  3 ++-
 mm/sparse-vmemmap.c       | 30 ++++++++++++++++++++++++------
 5 files changed, 35 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index 0adad8859393..7ca21adb4412 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c

@@ -1100,7 +1100,8 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
 		if (pmd_none(READ_ONCE(*pmdp))) {
 			void *p = NULL;
 
-			p = vmemmap_alloc_block_buf(PMD_SIZE, node);
+			p = vmemmap_alloc_block_buf(PMD_SIZE, node,
+						    NULL, false);
 			if (!p)
 				return -ENOMEM;
 

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index bc73abf0bc25..01e25b56eccb 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -225,12 +225,12 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
 		 * fall back to system memory if the altmap allocation fail.
 		 */
 		if (altmap && !altmap_cross_boundary(altmap, start, page_size)) {
-			p = altmap_alloc_block_buf(page_size, altmap);
-			if (!p)
-				pr_debug("altmap block allocation failed, falling back to system memory");
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    altmap, true);
+		} else {
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    NULL, false);
 		}
-		if (!p)
-			p = vmemmap_alloc_block_buf(page_size, node);
 		if (!p)
 			return -ENOMEM;
 

diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 19c0ed3271a3..4ae4f767c004 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c

@@ -1463,10 +1463,8 @@ static int __meminit vmemmap_populate_hugepages(unsigned long start,
 		if (pmd_none(*pmd)) {
 			void *p;
 
-			if (altmap)
-				p = altmap_alloc_block_buf(PMD_SIZE, altmap);
-			else
-				p = vmemmap_alloc_block_buf(PMD_SIZE, node);
+			p = vmemmap_alloc_block_buf(PMD_SIZE, node,
+						    altmap, false);
 			if (p) {
 				pte_t entry;
 

diff --git a/include/linux/mm.h b/include/linux/mm.h
index e40ac543d248..dade7c3f634d 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h

@@ -3015,7 +3015,8 @@ pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node,
 			    struct vmem_altmap *altmap);
 void *vmemmap_alloc_block(unsigned long size, int node);
 struct vmem_altmap;
-void *vmemmap_alloc_block_buf(unsigned long size, int node);
+void *vmemmap_alloc_block_buf(unsigned long size, int node,
+			      struct vmem_altmap *altmap, bool sysram_fallback);
 void *altmap_alloc_block_buf(unsigned long size, struct vmem_altmap *altmap);
 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
 int vmemmap_populate_basepages(unsigned long start, unsigned long end,

diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
index ceed10dec31e..388121c5bbcb 100644
--- a/mm/sparse-vmemmap.c
+++ b/mm/sparse-vmemmap.c

@@ -70,10 +70,31 @@ void * __meminit vmemmap_alloc_block(unsigned long size, int node)
 }
 
 /* need to make sure size is all the same during early stage */
-void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node)
+void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node,
+					 struct vmem_altmap *altmap,
+					 bool sysram_fallback)
 {
-	void *ptr = sparse_buffer_alloc(size);
+	void *ptr;
 
+	/*
+	 * There is no point in asking for sysram fallback
+	 * without an altmap request to begin with. So just
+	 * warn here to catch potential call sites that may
+	 * be violating this.
+	 */
+	WARN_ON(!altmap && sysram_fallback);
+
+	if (altmap) {
+		ptr = altmap_alloc_block_buf(size, altmap);
+		if (ptr)
+			return ptr;
+		pr_debug("altmap block allocation failed\n");
+		if (!sysram_fallback)
+			return NULL;
+		pr_debug("falling back to system memory\n");
+	}
+
+	ptr = sparse_buffer_alloc(size);
 	if (!ptr)
 		ptr = vmemmap_alloc_block(size, node);
 	return ptr;

@@ -147,10 +168,7 @@ pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node,
 		pte_t entry;
 		void *p;
 
-		if (altmap)
-			p = altmap_alloc_block_buf(PAGE_SIZE, altmap);
-		else
-			p = vmemmap_alloc_block_buf(PAGE_SIZE, node);
+		p = vmemmap_alloc_block_buf(PAGE_SIZE, node, altmap, false);
 		if (!p)
 			return NULL;
 		entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);

-- 
2.20.1

On Thu, Jun 18, 2020 at 06:45:27AM +0530, Anshuman Khandual wrote:

This series enables vmemmap backing memory allocation from device memory
ranges on arm64. But before that, it enables vmemmap_populate_basepages()
and vmemmap_alloc_block_buf() to accommodate struct vmem_altmap based
alocation requests.

This series applies on 5.8-rc1.

Pending Question:

altmap_alloc_block_buf() does not have any other remaining users in
the tree after this change. Should it be converted into a static
function and it's declaration be dropped from the header
(include/linux/mm.h). Avoided doing so because I was not sure if there
are any off-tree users or not.

Well, off-tree users probably have an active fork anyway so they could
switch to vmemmap_alloc_block_buf()...

Regardless, can you please update Documentation/vm/memory-model.rst to
keep it in sync with the code?

Changes in V3:

- Dropped comment from free_hotplug_page_range() per Robin
- Modified comment in unmap_hotplug_range() per Robin
- Enabled altmap support in vmemmap_alloc_block_buf() per Robin

Changes in V2: (https://lkml.org/lkml/2020/3/4/475)

- Rebased on latest hot-remove series (v14) adding P4D page table support

Changes in V1: (https://lkml.org/lkml/2020/1/23/12)

- Added an WARN_ON() in unmap_hotplug_range() when altmap is
  provided without the page table backing memory being freed

Changes in RFC V2: (https://lkml.org/lkml/2019/10/21/11)

- Changed the commit message on 1/2 patch per Will
- Changed the commit message on 2/2 patch as well
- Rebased on arm64 memory hot remove series (v10)

RFC V1: (https://lkml.org/lkml/2019/6/28/32)

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Paul Walmsley <redacted>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <redacted>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <redacted>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: David Hildenbrand <redacted>
Cc: Mike Rapoport <redacted>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: "Kirill A. Shutemov" <redacted>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <redacted>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org

Anshuman Khandual (3):
  mm/sparsemem: Enable vmem_altmap support in vmemmap_populate_basepages()
  mm/sparsemem: Enable vmem_altmap support in vmemmap_alloc_block_buf()
  arm64/mm: Enable vmem_altmap support for vmemmap mappings

 arch/arm64/mm/mmu.c       | 59 ++++++++++++++++++++++++++-------------
 arch/ia64/mm/discontig.c  |  2 +-
 arch/powerpc/mm/init_64.c | 10 +++----
 arch/riscv/mm/init.c      |  2 +-
 arch/x86/mm/init_64.c     | 12 ++++----
 include/linux/mm.h        |  8 ++++--
 mm/sparse-vmemmap.c       | 38 ++++++++++++++++++++-----
 7 files changed, 87 insertions(+), 44 deletions(-)

-- 
2.20.1

-- 
Sincerely yours,
Mike.

On 06/18/2020 02:26 PM, Mike Rapoport wrote:

On Thu, Jun 18, 2020 at 06:45:27AM +0530, Anshuman Khandual wrote:

quoted

This series enables vmemmap backing memory allocation from device memory
ranges on arm64. But before that, it enables vmemmap_populate_basepages()
and vmemmap_alloc_block_buf() to accommodate struct vmem_altmap based
alocation requests.

This series applies on 5.8-rc1.

Pending Question:

altmap_alloc_block_buf() does not have any other remaining users in
the tree after this change. Should it be converted into a static
function and it's declaration be dropped from the header
(include/linux/mm.h). Avoided doing so because I was not sure if there
are any off-tree users or not.

Well, off-tree users probably have an active fork anyway so they could
switch to vmemmap_alloc_block_buf()...

Sure, will make the function a static and remove it's declaration
from the header.

Regardless, can you please update Documentation/vm/memory-model.rst to
keep it in sync with the code?

Sure, will do.

Hi

I also tested the addional cases on arm64

1, 4k page size + devdax + --map=mem

2, 64k page size + devdax + --map=mem

3, 4k page size + devdax + --map=dev

4, 64k page size + devdax + --map=dev

case 4 is important to verify Anshuman's this series.

Host kernel: 5.7-rc3

guest kernel: 5.7-rc5 with this series

ndctl: https://github.com/pmem/ndctl/tree/c7767834871 
<https://github.com/pmem/ndctl/tree/c7767834871f7ce50a2abe1da946e9e16fb08eda>

On the guest:

1. ./ndctl/.libs/ndctl create-namespace -e namespace0.0 --mode=devdax --map=dev 
-s 1g -f -v -a 64K

  echo dax0.0 > /sys/bus/dax/drivers/device_dax/unbind
  echo dax0.0 > /sys/bus/dax/drivers/kmem/new_id

The 1g block was added

2. echo 0 > /sys/devices/system/memory/memory10/online

modprobe -r dax_pmem

The 1g block was removed


Some minor fix should be applied which is not relevant to this series itself. 
e.g numa id

---
Cheers,
Justin (Jia He)

On 2020/6/18 9:15, Anshuman Khandual wrote:

This series enables vmemmap backing memory allocation from device memory
ranges on arm64. But before that, it enables vmemmap_populate_basepages()
and vmemmap_alloc_block_buf() to accommodate struct vmem_altmap based
alocation requests.

This series applies on 5.8-rc1.

Pending Question:

altmap_alloc_block_buf() does not have any other remaining users in the
tree after this change. Should it be converted into a static function and
it's declaration be dropped from the header (include/linux/mm.h). Avoided
doing so because I was not sure if there are any off-tree users or not.

Changes in V3:

- Dropped comment from free_hotplug_page_range() per Robin
- Modified comment in unmap_hotplug_range() per Robin
- Enabled altmap support in vmemmap_alloc_block_buf() per Robin

Changes in V2: (https://lkml.org/lkml/2020/3/4/475)

- Rebased on latest hot-remove series (v14) adding P4D page table support

Changes in V1: (https://lkml.org/lkml/2020/1/23/12)

- Added an WARN_ON() in unmap_hotplug_range() when altmap is
   provided without the page table backing memory being freed

Changes in RFC V2: (https://lkml.org/lkml/2019/10/21/11)

- Changed the commit message on 1/2 patch per Will
- Changed the commit message on 2/2 patch as well
- Rebased on arm64 memory hot remove series (v10)

RFC V1: (https://lkml.org/lkml/2019/6/28/32)

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Paul Walmsley <redacted>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <redacted>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <redacted>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: David Hildenbrand <redacted>
Cc: Mike Rapoport <redacted>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: "Kirill A. Shutemov" <redacted>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <redacted>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Cc: x86@kernel.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org

Anshuman Khandual (3):
   mm/sparsemem: Enable vmem_altmap support in vmemmap_populate_basepages()
   mm/sparsemem: Enable vmem_altmap support in vmemmap_alloc_block_buf()
   arm64/mm: Enable vmem_altmap support for vmemmap mappings

  arch/arm64/mm/mmu.c       | 59 ++++++++++++++++++++++++++-------------
  arch/ia64/mm/discontig.c  |  2 +-
  arch/powerpc/mm/init_64.c | 10 +++----
  arch/riscv/mm/init.c      |  2 +-
  arch/x86/mm/init_64.c     | 12 ++++----
  include/linux/mm.h        |  8 ++++--
  mm/sparse-vmemmap.c       | 38 ++++++++++++++++++++-----
  7 files changed, 87 insertions(+), 44 deletions(-)

-- 

On Thu, Jun 18, 2020 at 06:45:29AM +0530, Anshuman Khandual wrote:

There are many instances where vmemap allocation is often switched between
regular memory and device memory just based on whether altmap is available
or not. vmemmap_alloc_block_buf() is used in various platforms to allocate
vmemmap mappings. Lets also enable it to handle altmap based device memory
allocation along with existing regular memory allocations. This will help
in avoiding the altmap based allocation switch in many places.

While here also implement a regular memory allocation fallback mechanism
when the first preferred device memory allocation fails. This will ensure
preserving the existing semantics on powerpc platform. To summarize there
are three different methods to call vmemmap_alloc_block_buf().

(., NULL,   false) /* Allocate from system RAM */
(., altmap, false) /* Allocate from altmap without any fallback */
(., altmap, true)  /* Allocate from altmap with fallback (system RAM) */

[...]

Is the fallback argument actually necessary. It may be cleaner to just
leave the code as is with the choice between altmap and NULL. If an arch
needs a fallback (only powerpc), they have the fallback in place
already. I don't see the powerpc code any better after this change.

I'm fine with the altmap argument though.

-- 
Catalin

On 07/02/2020 07:37 PM, Catalin Marinas wrote:

On Thu, Jun 18, 2020 at 06:45:29AM +0530, Anshuman Khandual wrote:

quoted

There are many instances where vmemap allocation is often switched between
regular memory and device memory just based on whether altmap is available
or not. vmemmap_alloc_block_buf() is used in various platforms to allocate
vmemmap mappings. Lets also enable it to handle altmap based device memory
allocation along with existing regular memory allocations. This will help
in avoiding the altmap based allocation switch in many places.

While here also implement a regular memory allocation fallback mechanism
when the first preferred device memory allocation fails. This will ensure
preserving the existing semantics on powerpc platform. To summarize there
are three different methods to call vmemmap_alloc_block_buf().

(., NULL,   false) /* Allocate from system RAM */
(., altmap, false) /* Allocate from altmap without any fallback */
(., altmap, true)  /* Allocate from altmap with fallback (system RAM) */

[...]

quoted

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index bc73abf0bc25..01e25b56eccb 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -225,12 +225,12 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
 		 * fall back to system memory if the altmap allocation fail.
 		 */
 		if (altmap && !altmap_cross_boundary(altmap, start, page_size)) {
-			p = altmap_alloc_block_buf(page_size, altmap);
-			if (!p)
-				pr_debug("altmap block allocation failed, falling back to system memory");
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    altmap, true);
+		} else {
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    NULL, false);
 		}
-		if (!p)
-			p = vmemmap_alloc_block_buf(page_size, node);
 		if (!p)
 			return -ENOMEM;

Is the fallback argument actually necessary. It may be cleaner to just
leave the code as is with the choice between altmap and NULL. If an arch
needs a fallback (only powerpc), they have the fallback in place
already. I don't see the powerpc code any better after this change.

I'm fine with the altmap argument though.

Okay. Will drop 'fallback' from vmemmap_alloc_block_buf() and update the
callers. There will also be a single change in the subsequent patch i.e
vmemmap_alloc_block_buf(PMD_SIZE, node, altmap).

Catalin Marinas [off-list ref] writes:

On Thu, Jun 18, 2020 at 06:45:29AM +0530, Anshuman Khandual wrote:

quoted

There are many instances where vmemap allocation is often switched between
regular memory and device memory just based on whether altmap is available
or not. vmemmap_alloc_block_buf() is used in various platforms to allocate
vmemmap mappings. Lets also enable it to handle altmap based device memory
allocation along with existing regular memory allocations. This will help
in avoiding the altmap based allocation switch in many places.

While here also implement a regular memory allocation fallback mechanism
when the first preferred device memory allocation fails. This will ensure
preserving the existing semantics on powerpc platform. To summarize there
are three different methods to call vmemmap_alloc_block_buf().

(., NULL,   false) /* Allocate from system RAM */
(., altmap, false) /* Allocate from altmap without any fallback */
(., altmap, true)  /* Allocate from altmap with fallback (system RAM) */

[...]

quoted

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index bc73abf0bc25..01e25b56eccb 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -225,12 +225,12 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
 		 * fall back to system memory if the altmap allocation fail.
 		 */
 		if (altmap && !altmap_cross_boundary(altmap, start, page_size)) {
-			p = altmap_alloc_block_buf(page_size, altmap);
-			if (!p)
-				pr_debug("altmap block allocation failed, falling back to system memory");
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    altmap, true);
+		} else {
+			p = vmemmap_alloc_block_buf(page_size, node,
+						    NULL, false);
 		}
-		if (!p)
-			p = vmemmap_alloc_block_buf(page_size, node);
 		if (!p)
 			return -ENOMEM;

Is the fallback argument actually necessary. It may be cleaner to just
leave the code as is with the choice between altmap and NULL. If an arch
needs a fallback (only powerpc), they have the fallback in place
already. I don't see the powerpc code any better after this change.

Yeah I agree.

cheers