[RFC v2 0/2] powerpc/mm: enable memory hotplug on radix

[RFC v2 0/2] powerpc/mm: enable memory hotplug on radix

[Reza Arbab]

Memory hotplug is leading to hash page table calls, even on radix:

...
	arch_add_memory
		create_section_mapping
			htab_bolt_mapping
				BUG_ON(!ppc_md.hpte_insert);

Refactor {create,remove}_section_mapping() into hash__ and radix__ variants.

RFC/TODO:
I wasn't sure what to do in radix__remove_section_mapping(). Its vmemmap
counterpart radix__vmemmap_remove_mapping() is stubbed as a FIXME. I left it
empty for now.

v2:
* Do not simply fall through to vmemmap_{create,remove}_mapping(). As Aneesh 
  and Michael pointed out, they are tied to CONFIG_SPARSEMEM_VMEMMAP and only
  did what I needed by luck anyway.

v1:
* https://lkml.kernel.org/r/1466699962-22412-1-git-send-email-arbab@linux.vnet.ibm.com

Reza Arbab (2):
  powerpc/mm: refactor {create,remove}_section_mapping()
  powerpc/mm: add radix__{create,remove}_section_mapping()

 arch/powerpc/include/asm/book3s/64/hash.h  |  5 +++++
 arch/powerpc/include/asm/book3s/64/radix.h |  5 +++++
 arch/powerpc/include/asm/sparsemem.h       | 22 ++++++++++++++++++++--
 arch/powerpc/mm/hash_utils_64.c            |  4 ++--
 arch/powerpc/mm/pgtable-radix.c            | 23 +++++++++++++++++++++++
 5 files changed, 55 insertions(+), 4 deletions(-)

-- 
1.8.3.1


*** BLURB HERE ***

Reza Arbab (2):
  powerpc/mm: refactor {create,remove}_section_mapping()
  powerpc/mm: add radix__{create,remove}_section_mapping()

 arch/powerpc/include/asm/book3s/64/hash.h  |  5 +++++
 arch/powerpc/include/asm/book3s/64/radix.h |  5 +++++
 arch/powerpc/include/asm/sparsemem.h       | 22 ++++++++++++++++++++--
 arch/powerpc/mm/hash_utils_64.c            |  4 ++--
 arch/powerpc/mm/pgtable-radix.c            | 23 +++++++++++++++++++++++
 5 files changed, 55 insertions(+), 4 deletions(-)

-- 
1.8.3.1

[RFC v2 1/2] powerpc/mm: refactor {create,remove}_section_mapping()

[Reza Arbab]

Change {create,remove}_section_mapping() to be inline wrappers around
functions prefixed with "hash__".

This is preparation for the addition of their "radix__" variants. No
functional change.

Signed-off-by: Reza Arbab <arbab@linux.vnet.ibm.com>
---
 arch/powerpc/include/asm/book3s/64/hash.h |  5 +++++
 arch/powerpc/include/asm/sparsemem.h      | 14 ++++++++++++--
 arch/powerpc/mm/hash_utils_64.c           |  4 ++--
 3 files changed, 19 insertions(+), 4 deletions(-)

diff --git a/arch/powerpc/include/asm/book3s/64/hash.h b/arch/powerpc/include/asm/book3s/64/hash.h
index f61cad3..dd90574 100644
--- a/arch/powerpc/include/asm/book3s/64/hash.h
+++ b/arch/powerpc/include/asm/book3s/64/hash.h
@@ -201,6 +201,11 @@ extern int __meminit hash__vmemmap_create_mapping(unsigned long start,
 					      unsigned long phys);
 extern void hash__vmemmap_remove_mapping(unsigned long start,
 				     unsigned long page_size);
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+int hash__create_section_mapping(unsigned long start, unsigned long end);
+int hash__remove_section_mapping(unsigned long start, unsigned long end);
+#endif /* CONFIG_MEMORY_HOTPLUG */
 #endif /* !__ASSEMBLY__ */
 #endif /* __KERNEL__ */
 #endif /* _ASM_POWERPC_BOOK3S_64_HASH_H */
diff --git a/arch/powerpc/include/asm/sparsemem.h b/arch/powerpc/include/asm/sparsemem.h
index f6fc0ee..5014e0d 100644
--- a/arch/powerpc/include/asm/sparsemem.h
+++ b/arch/powerpc/include/asm/sparsemem.h
@@ -16,8 +16,18 @@
 #endif /* CONFIG_SPARSEMEM */
 
 #ifdef CONFIG_MEMORY_HOTPLUG
-extern int create_section_mapping(unsigned long start, unsigned long end);
-extern int remove_section_mapping(unsigned long start, unsigned long end);
+static inline int create_section_mapping(unsigned long start,
+					 unsigned long end)
+{
+	return hash__create_section_mapping(start, end);
+}
+
+static inline int remove_section_mapping(unsigned long start,
+					 unsigned long end)
+{
+	return hash__remove_section_mapping(start, end);
+}
+
 #ifdef CONFIG_NUMA
 extern int hot_add_scn_to_nid(unsigned long scn_addr);
 #else
diff --git a/arch/powerpc/mm/hash_utils_64.c b/arch/powerpc/mm/hash_utils_64.c
index 0821556..2a203cd 100644
--- a/arch/powerpc/mm/hash_utils_64.c
+++ b/arch/powerpc/mm/hash_utils_64.c
@@ -688,7 +688,7 @@ static unsigned long __init htab_get_table_size(void)
 }
 
 #ifdef CONFIG_MEMORY_HOTPLUG
-int create_section_mapping(unsigned long start, unsigned long end)
+int hash__create_section_mapping(unsigned long start, unsigned long end)
 {
 	int rc = htab_bolt_mapping(start, end, __pa(start),
 				   pgprot_val(PAGE_KERNEL), mmu_linear_psize,
@@ -702,7 +702,7 @@ int create_section_mapping(unsigned long start, unsigned long end)
 	return rc;
 }
 
-int remove_section_mapping(unsigned long start, unsigned long end)
+int hash__remove_section_mapping(unsigned long start, unsigned long end)
 {
 	int rc = htab_remove_mapping(start, end, mmu_linear_psize,
 				     mmu_kernel_ssize);
-- 
1.8.3.1

[RFC v2 2/2] powerpc/mm: add radix__{create, remove}_section_mapping()

[Reza Arbab]

Add radix variants of the memory hotplug mapping functions.

Signed-off-by: Reza Arbab <arbab@linux.vnet.ibm.com>
---
 arch/powerpc/include/asm/book3s/64/radix.h |  5 +++++
 arch/powerpc/include/asm/sparsemem.h       |  8 ++++++++
 arch/powerpc/mm/pgtable-radix.c            | 23 +++++++++++++++++++++++
 3 files changed, 36 insertions(+)

diff --git a/arch/powerpc/include/asm/book3s/64/radix.h b/arch/powerpc/include/asm/book3s/64/radix.h
index df29422..97e71cb 100644
--- a/arch/powerpc/include/asm/book3s/64/radix.h
+++ b/arch/powerpc/include/asm/book3s/64/radix.h
@@ -229,6 +229,11 @@ extern void radix__vmemmap_remove_mapping(unsigned long start,
 extern int radix__map_kernel_page(unsigned long ea, unsigned long pa,
 				 pgprot_t flags, unsigned int psz);
 
+#ifdef CONFIG_MEMORY_HOTPLUG
+int radix__create_section_mapping(unsigned long start, unsigned long end);
+void radix__remove_section_mapping(unsigned long start, unsigned long end);
+#endif /* CONFIG_MEMORY_HOTPLUG */
+
 static inline unsigned long radix__get_tree_size(void)
 {
 	unsigned long rts_field;
diff --git a/arch/powerpc/include/asm/sparsemem.h b/arch/powerpc/include/asm/sparsemem.h
index 5014e0d..42fe87e 100644
--- a/arch/powerpc/include/asm/sparsemem.h
+++ b/arch/powerpc/include/asm/sparsemem.h
@@ -19,12 +19,20 @@
 static inline int create_section_mapping(unsigned long start,
 					 unsigned long end)
 {
+	if (radix_enabled())
+		return radix__create_section_mapping(start, end);
+
 	return hash__create_section_mapping(start, end);
 }
 
 static inline int remove_section_mapping(unsigned long start,
 					 unsigned long end)
 {
+	if (radix_enabled()) {
+		radix__remove_section_mapping(start, end);
+		return 0;
+	}
+
 	return hash__remove_section_mapping(start, end);
 }
 
diff --git a/arch/powerpc/mm/pgtable-radix.c b/arch/powerpc/mm/pgtable-radix.c
index af897d9..ee0286c 100644
--- a/arch/powerpc/mm/pgtable-radix.c
+++ b/arch/powerpc/mm/pgtable-radix.c
@@ -527,3 +527,26 @@ int radix__has_transparent_hugepage(void)
 	return 0;
 }
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+int radix__create_section_mapping(unsigned long start, unsigned long end)
+{
+	unsigned long page_size = 1 << mmu_psize_defs[mmu_linear_psize].shift;
+
+	/* Align to the page size of the linear mapping. */
+	start = _ALIGN_DOWN(start, page_size);
+
+	for (; start < end; start += page_size) {
+		int rc = radix__map_kernel_page(start, __pa(start),
+						PAGE_KERNEL, page_size);
+	        if (rc)
+			return rc;
+	}
+
+	return 0;
+}
+
+void radix__remove_section_mapping(unsigned long start, unsigned long end)
+{
+}
+#endif /* CONFIG_MEMORY_HOTPLUG */
-- 
1.8.3.1

Re: [RFC v2 0/2] powerpc/mm: enable memory hotplug on radix

[Aneesh Kumar K.V]

Reza Arbab <arbab@linux.vnet.ibm.com> writes:

> Memory hotplug is leading to hash page table calls, even on radix:
>
> ...
> 	arch_add_memory
> 		create_section_mapping
> 			htab_bolt_mapping
> 				BUG_ON(!ppc_md.hpte_insert);
>
> Refactor {create,remove}_section_mapping() into hash__ and radix__ variants.
>
> RFC/TODO:
> I wasn't sure what to do in radix__remove_section_mapping(). Its vmemmap
> counterpart radix__vmemmap_remove_mapping() is stubbed as a FIXME. I left it
> empty for now.

For linear mapping and for vmemmap area we should do something equivalent of

static void __meminit
remove_pagetable(unsigned long start, unsigned long end, bool direct)

as done for x86.

-aneesh