mm: cma: NUMA node interface

I've noticed that there is no interfaces exposed by CMA which would let me to declare contigous memory on particular NUMA node.

This patchset adds the ability to try to allocate contiguous memory on specific node. It will fallback to other nodes if the specified one doesn't work.

Implement a new method for declaring contigous memory on particular node and keep cma_declare_contiguous() as a wrapper.

Signed-off-by Aslan Bakirov <[email protected]>

---
 include/linux/cma.h      | 13 +++++++++++--
 include/linux/memblock.h |  3 +++
 mm/cma.c                 | 16 +++++++++-------
 mm/memblock.c            |  2 +-
 4 files changed, 24 insertions(+), 10 deletions(-)

include/linux/cma.h +11 -2

index 190184b5ff32..eae834c2162f 100644
@@ -24,10 +24,19 @@ extern phys_addr_t cma_get_base(const struct cma *cma);
 extern unsigned long cma_get_size(const struct cma *cma);
 extern const char *cma_get_name(const struct cma *cma);
 
-extern int __init cma_declare_contiguous(phys_addr_t base,
+extern int __init cma_declare_contiguous_nid(phys_addr_t base,
 			phys_addr_t size, phys_addr_t limit,
 			phys_addr_t alignment, unsigned int order_per_bit,
-			bool fixed, const char *name, struct cma **res_cma);
+			bool fixed, const char *name, struct cma **res_cma,
+			int nid);
+static inline int __init cma_declare_contiguous(phys_addr_t base,
+			phys_addr_t size, phys_addr_t limit,
+			phys_addr_t alignment, unsigned int order_per_bit,
+			bool fixed, const char *name, struct cma **res_cma)
+{
+	return cma_declare_contiguous_nid(base, size, limit, alignment,
+			order_per_bit, fixed, name, res_cma, NUMA_NO_NODE);
+}
 extern int cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
 					unsigned int order_per_bit,
 					const char *name,

include/linux/memblock.h +3 -0

index 079d17d96410..6bc37a731d27 100644
@@ -348,6 +348,9 @@ static inline int memblock_get_region_node(const struct memblock_region *r)
 
 phys_addr_t memblock_phys_alloc_range(phys_addr_t size, phys_addr_t align,
 				      phys_addr_t start, phys_addr_t end);
+phys_addr_t memblock_alloc_range_nid(phys_addr_t size,
+				      phys_addr_t align, phys_addr_t start,
+				      phys_addr_t end, int nid, bool exact_nid);
 phys_addr_t memblock_phys_alloc_try_nid(phys_addr_t size, phys_addr_t align, int nid);
 
 static inline phys_addr_t memblock_phys_alloc(phys_addr_t size,

mm/cma.c +9 -7

index be55d1988c67..6405af3dc118 100644
@@ -220,7 +220,7 @@ int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
 }
 
 /**
- * cma_declare_contiguous() - reserve custom contiguous area
+ * cma_declare_contiguous_nid() - reserve custom contiguous area
  * @base: Base address of the reserved area optional, use 0 for any
  * @size: Size of the reserved area (in bytes),
  * @limit: End address of the reserved memory (optional, 0 for any).

@@ -229,6 +229,7 @@ int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
  * @fixed: hint about where to place the reserved area
  * @name: The name of the area. See function cma_init_reserved_mem()
  * @res_cma: Pointer to store the created cma region.
+ * @nid: nid of the free area to find, %NUMA_NO_NODE for any node
  *
  * This function reserves memory from early allocator. It should be
  * called by arch specific code once the early allocator (memblock or bootmem)

@@ -238,10 +239,11 @@ int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
  * If @fixed is true, reserve contiguous area at exactly @base.  If false,
  * reserve in range from @base to @limit.
  */
-int __init cma_declare_contiguous(phys_addr_t base,
+int __init cma_declare_contiguous_nid(phys_addr_t base,
 			phys_addr_t size, phys_addr_t limit,
 			phys_addr_t alignment, unsigned int order_per_bit,
-			bool fixed, const char *name, struct cma **res_cma)
+			bool fixed, const char *name, struct cma **res_cma,
+			int nid)
 {
 	phys_addr_t memblock_end = memblock_end_of_DRAM();
 	phys_addr_t highmem_start;

@@ -336,14 +338,14 @@ int __init cma_declare_contiguous(phys_addr_t base,
 		 * memory in case of failure.
 		 */
 		if (base < highmem_start && limit > highmem_start) {
-			addr = memblock_phys_alloc_range(size, alignment,
-							 highmem_start, limit);
+			addr = memblock_alloc_range_nid(size, alignment,
+					highmem_start, limit, nid, false);
 			limit = highmem_start;
 		}
 
 		if (!addr) {
-			addr = memblock_phys_alloc_range(size, alignment, base,
-							 limit);
+			addr = memblock_alloc_range_nid(size, alignment, base,
+					alimit, nid, false);
 			if (!addr) {
 				ret = -ENOMEM;
 				goto err;

mm/memblock.c +1 -1

index 4d06bbaded0f..c79ba6f9920c 100644
@@ -1349,7 +1349,7 @@ __next_mem_pfn_range_in_zone(u64 *idx, struct zone *zone,
  * Return:
  * Physical address of allocated memory block on success, %0 on failure.
  */
-static phys_addr_t __init memblock_alloc_range_nid(phys_addr_t size,
+phys_addr_t __init memblock_alloc_range_nid(phys_addr_t size,
 					phys_addr_t align, phys_addr_t start,
 					phys_addr_t end, int nid,
 					bool exact_nid)

Signature · 1 line

2.24.1

Quoted 1 line · Aslan Bakirov

I've noticed that there is no interfaces exposed by CMA which would let

me

Quoted 6 lines

to declare contigous memory on particular NUMA node.

This patchset adds the ability to try to allocate contiguous memory on specific node. It will fallback to other nodes if the specified one doesn't work.

Quoted 2 lines

Implement a new method for declaring contigous memory on particular node and keep cma_declare_contiguous() as a wrapper.

Is there an additional patch which uses this new interface?

Generally the patch seems reasonable but we should have a user.

include/linux/cma.h extern phys_addr_t cma_get_base(const struct cma

Quoted 15 lines

Signed-off-by Aslan Bakirov <[email protected]>

---
 include/linux/cma.h      | 13 +++++++++++--
 include/linux/memblock.h |  3 +++
 mm/cma.c                 | 16 +++++++++-------
 mm/memblock.c            |  2 +-
 4 files changed, 24 insertions(+), 10 deletions(-)

include/linux/cma.h +0 -0

index 190184b5ff32..eae834c2162f 100644
@@ -24,10 +24,19 @@ extern phys_addr_t cma_get_base(const struct cma

*cma);

include/linux/cma.h extern phys_addr_t cma_get_base(const struct cma *cma);

Quoted 8 lines

include/linux/cma.h +1 -2

 extern unsigned long cma_get_size(const struct cma *cma);
 extern const char *cma_get_name(const struct cma *cma);
 
-extern int __init cma_declare_contiguous(phys_addr_t base,
+extern int __init cma_declare_contiguous_nid(phys_addr_t base,
                      phys_addr_t size, phys_addr_t limit,
                      phys_addr_t alignment, unsigned int order_per_bit,
-                     bool fixed, const char *name, struct cma

**res_cma);

include/linux/memblock.h static inline int memblock_get_region_node(const

Quoted 18 lines

include/linux/cma.h +10 -0

+                     bool fixed, const char *name, struct cma **res_cma,
+                     int nid);
+static inline int __init cma_declare_contiguous(phys_addr_t base,
+                     phys_addr_t size, phys_addr_t limit,
+                     phys_addr_t alignment, unsigned int order_per_bit,
+                     bool fixed, const char *name, struct cma **res_cma)
+{
+     return cma_declare_contiguous_nid(base, size, limit, alignment,
+                     order_per_bit, fixed, name, res_cma, NUMA_NO_NODE);
+}
 extern int cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
                                      unsigned int order_per_bit,
                                      const char *name,

include/linux/memblock.h +0 -0

index 079d17d96410..6bc37a731d27 100644
@@ -348,6 +348,9 @@ static inline int memblock_get_region_node(const

struct memblock_region *r)

include/linux/memblock.h static inline int memblock_get_region_node(const · line 348

Quoted 2 lines

include/linux/memblock.h +0 -0

 phys_addr_t memblock_phys_alloc_range(phys_addr_t size, phys_addr_t

align,

include/linux/memblock.h static inline int memblock_get_region_node(const struct memblock_region *r)

Quoted 4 lines

include/linux/memblock.h +3 -0

                                    phys_addr_t start, phys_addr_t end);
+phys_addr_t memblock_alloc_range_nid(phys_addr_t size,
+                                   phys_addr_t align, phys_addr_t start,
+                                   phys_addr_t end, int nid, bool

exact_nid);

include/linux/memblock.h static inline int memblock_get_region_node(const · line 353

Quoted 1 line

include/linux/memblock.h +0 -0

 phys_addr_t memblock_phys_alloc_try_nid(phys_addr_t size, phys_addr_t

align, int nid);

mm/cma.c int __init cma_init_reserved_mem(phys_addr_t base,

Quoted 7 lines

include/linux/memblock.h +0 -0

 static inline phys_addr_t memblock_phys_alloc(phys_addr_t size,

mm/cma.c +0 -0

index be55d1988c67..6405af3dc118 100644
@@ -220,7 +220,7 @@ int __init cma_init_reserved_mem(phys_addr_t base,

phys_addr_t size,

mm/cma.c int __init cma_init_reserved_mem(phys_addr_t base,

Quoted 9 lines

mm/cma.c +1 -1

 }
 
 /**
- * cma_declare_contiguous() - reserve custom contiguous area
+ * cma_declare_contiguous_nid() - reserve custom contiguous area
  * @base: Base address of the reserved area optional, use 0 for any
  * @size: Size of the reserved area (in bytes),
  * @limit: End address of the reserved memory (optional, 0 for any).

@@ -229,6 +229,7 @@ int __init cma_init_reserved_mem(phys_addr_t base,

phys_addr_t size,

mm/cma.c int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,

Quoted 7 lines

mm/cma.c +1 -0

  * @fixed: hint about where to place the reserved area
  * @name: The name of the area. See function cma_init_reserved_mem()
  * @res_cma: Pointer to store the created cma region.
+ * @nid: nid of the free area to find, %NUMA_NO_NODE for any node
  *
  * This function reserves memory from early allocator. It should be
  * called by arch specific code once the early allocator (memblock or

bootmem)

mm/cma.c int __init cma_init_reserved_mem(phys_addr_t base,

Quoted 1 line

mm/cma.c +0 -0

@@ -238,10 +239,11 @@ int __init cma_init_reserved_mem(phys_addr_t base,

phys_addr_t size,

mm/cma.c int __init cma_init_reserved_mem(phys_addr_t base, · line 239

Quoted 1 line

mm/cma.c +0 -0

  * If @fixed is true, reserve contiguous area at exactly @base.  If

false,

Quoted 13 lines

mm/cma.c +3 -2

  * reserve in range from @base to @limit.
  */
-int __init cma_declare_contiguous(phys_addr_t base,
+int __init cma_declare_contiguous_nid(phys_addr_t base,
                      phys_addr_t size, phys_addr_t limit,
                      phys_addr_t alignment, unsigned int order_per_bit,
-                     bool fixed, const char *name, struct cma **res_cma)
+                     bool fixed, const char *name, struct cma **res_cma,
+                     int nid)
 {
      phys_addr_t memblock_end = memblock_end_of_DRAM();
      phys_addr_t highmem_start;

mm/cma.c int __init cma_declare_contiguous(phys_addr_t base, · line 340

Quoted 5 lines

mm/cma.c +0 -2

@@ -336,14 +338,14 @@ int __init cma_declare_contiguous(phys_addr_t base,
               * memory in case of failure.
               */
              if (base < highmem_start && limit > highmem_start) {
-                     addr = memblock_phys_alloc_range(size, alignment,
-                                                      highmem_start,

limit);

mm/cma.c int __init cma_declare_contiguous(phys_addr_t base,

Quoted 7 lines

mm/cma.c +2 -1

+                     addr = memblock_alloc_range_nid(size, alignment,
+                                     highmem_start, limit, nid, false);
                      limit = highmem_start;
              }
 
              if (!addr) {
-                     addr = memblock_phys_alloc_range(size, alignment,

base,

mm/cma.c int __init cma_declare_contiguous(phys_addr_t base, · line 347

Quoted 2 lines

mm/cma.c +1 -1

-                                                      limit);
+                     addr = memblock_alloc_range_nid(size, alignment,

base,

mm/memblock.c __next_mem_pfn_range_in_zone(u64 *idx, struct zone

Quoted 9 lines

mm/cma.c +1 -0

+                                     alimit, nid, false);
                      if (!addr) {
                              ret = -ENOMEM;
                              goto err;

mm/memblock.c +0 -0

index 4d06bbaded0f..c79ba6f9920c 100644
@@ -1349,7 +1349,7 @@ __next_mem_pfn_range_in_zone(u64 *idx, struct zone

*zone,

mm/memblock.c __next_mem_pfn_range_in_zone(u64 *idx, struct zone *zone,

Quoted 6 lines

mm/memblock.c +1 -1

  * Return:
  * Physical address of allocated memory block on success, %0 on failure.
  */
-static phys_addr_t __init memblock_alloc_range_nid(phys_addr_t size,
+phys_addr_t __init memblock_alloc_range_nid(phys_addr_t size,
                                      phys_addr_t align, phys_addr_t

start,

Quoted 6 lines

mm/memblock.c +0 -1

                                      phys_addr_t end, int nid,
                                      bool exact_nid)
--

2.24.1

Quoted 14 lines · Ira Weiny

Quoted 1 line · Aslan Bakirov

I've noticed that there is no interfaces exposed by CMA which would let

me

Quoted 8 lines

to declare contigous memory on particular NUMA node.

This patchset adds the ability to try to allocate contiguous memory on specific node. It will fallback to other nodes if the specified one doesn't work.

Implement a new method for declaring contigous memory on particular node and keep cma_declare_contiguous() as a wrapper.

Is there an additional patch which uses this new interface?

Quoted 1 line

Generally the patch seems reasonable but we should have a user.

Thanks for the comments. Yes, actually, this is the version 3 of first
patch ([PATCH 1/2] mm: cma: NUMA node interface)
of patchset. Second patch, which uses this interface is "[PATCH 2/2] mm:
hugetlb: Use node interface of cma"

Quoted 4 lines · Aslan Bakirov

Quoted 2 lines · Ira Weiny

Quoted 1 line · Aslan Bakirov

I've noticed that there is no interfaces exposed by CMA which would

let

Quoted 4 lines

Quoted 4 lines

me

Quoted 3 lines

to declare contigous memory on particular NUMA node.

This patchset adds the ability to try to allocate contiguous memory

on

Quoted 4 lines

Quoted 4 lines

Quoted 4 lines

specific node. It will fallback to other nodes if the specified one doesn't work.

Implement a new method for declaring contigous memory on particular

node

Quoted 7 lines

Quoted 5 lines

Quoted 1 line

and keep cma_declare_contiguous() as a wrapper.

Is there an additional patch which uses this new interface?

Generally the patch seems reasonable but we should have a user.

Quoted 3 lines

Thanks for the comments. Yes, actually, this is the version 3 of first
patch ([PATCH 1/2] mm: cma: NUMA node interface)
of patchset. Second patch, which uses this interface is "[PATCH 2/2]

mm:

Quoted 1 line

hugetlb: Use node interface of cma"

It would have been much more clear to send those two patches together as you can see.

--
Michal Hocko
SUSE Labs