An cma optimization patch is used for cma_[alloc|free].

[Jichao Zou <zoujc@motorola.com>]

[-- Attachment #1: Type: text/plain, Size: 399 bytes --]

Pre-allocate CMA memory that configured in device
tree, this greatly improves the CMA memory
allocation efficiency, cma_[alloc|free] is less
than 1ms, old way is took a few ms to tens or
hundreds ms.

Thanks.

Best Regards,

Zou Jichao 邹纪超
Advisory Engineer, SW BSP
MBG ROW SW BJ PF BSP (CN)
Motorola Mobility, A Lenovo Company
motorola.com 





[-- Attachment #2: 0001-cma-optimize-cma-allocation.patch --]
[-- Type: application/octet-stream, Size: 4988 bytes --]

From a82dbc46e2343c394d3edcabcbc73ea6e9f403da Mon Sep 17 00:00:00 2001
From: Jichao Zou <zoujc@motorola.com>
Date: Thu, 12 Aug 2021 18:20:21 +0800
Subject: [PATCH] cma:optimize cma allocation.

Pre-allocate CMA memory that configured in device
tree, this greatly improves the CMA memory
allocation efficiency, cma_[alloc|free] is less
than 1ms, old way is took a few ms to tens or
hundreds ms.

Signed-off-by: Jichao Zou <zoujc@motorola.com>
---
 include/linux/cma.h     |  3 ++-
 kernel/dma/contiguous.c |  2 +-
 mm/cma.c                | 35 ++++++++++++++++++++++++++++++++---
 mm/cma.h                |  1 +
 4 files changed, 36 insertions(+), 5 deletions(-)

diff --git a/include/linux/cma.h b/include/linux/cma.h
index 53fd8c3cdbd0..68bc147a82a7 100644
--- a/include/linux/cma.h
+++ b/include/linux/cma.h
@@ -43,7 +43,8 @@ static inline int __init cma_declare_contiguous(phys_addr_t base,
 extern int cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
 					unsigned int order_per_bit,
 					const char *name,
-					struct cma **res_cma);
+					struct cma **res_cma,
+					unsigned long node);
 extern struct page *cma_alloc(struct cma *cma, unsigned long count, unsigned int align,
 			      bool no_warn);
 extern bool cma_release(struct cma *cma, const struct page *pages, unsigned long count);
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index 3d63d91cba5c..d77c2745244c 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -421,7 +421,7 @@ static int __init rmem_cma_setup(struct reserved_mem *rmem)
 		return -EINVAL;
 	}
 
-	err = cma_init_reserved_mem(rmem->base, rmem->size, 0, rmem->name, &cma);
+	err = cma_init_reserved_mem(rmem->base, rmem->size, 0, rmem->name, &cma, node);
 	if (err) {
 		pr_err("Reserved memory: unable to setup CMA region\n");
 		return err;
diff --git a/mm/cma.c b/mm/cma.c
index 995e15480937..c5682d03c5e9 100644
--- a/mm/cma.c
+++ b/mm/cma.c
@@ -32,6 +32,7 @@
 #include <linux/io.h>
 #include <linux/kmemleak.h>
 #include <trace/events/cma.h>
+#include <linux/of_fdt.h>
 
 #include "cma.h"
 
@@ -124,6 +125,17 @@ static void __init cma_activate_area(struct cma *cma)
 	INIT_HLIST_HEAD(&cma->mem_head);
 	spin_lock_init(&cma->mem_head_lock);
 #endif
+	if (cma->preallocated_cma) {
+		struct acr_info info = {0};
+
+		pfn = base_pfn;
+		if (!alloc_contig_range(pfn, pfn + cma->count, MIGRATE_CMA, GFP_KERNEL, &info)) {
+			pr_info("CMA area %s be pre-allocated successfully\n", cma->name);
+		} else {
+			cma->preallocated_cma = false;
+			pr_err("CMA area %s be pre-allocated failure\n", cma->name);
+		}
+	}
 
 	return;
 
@@ -159,13 +171,15 @@ core_initcall(cma_init_reserved_areas);
  *        the area will be set to "cmaN", where N is a running counter of
  *        used areas.
  * @res_cma: Pointer to store the created cma region.
+ * @node: CMA memory dtsi node.
  *
  * This function creates custom contiguous area from already reserved memory.
  */
 int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
 				 unsigned int order_per_bit,
 				 const char *name,
-				 struct cma **res_cma)
+				 struct cma **res_cma,
+				 unsigned long node)
 {
 	struct cma *cma;
 	phys_addr_t alignment;
@@ -204,6 +218,9 @@ int __init cma_init_reserved_mem(phys_addr_t base, phys_addr_t size,
 	cma->base_pfn = PFN_DOWN(base);
 	cma->count = size >> PAGE_SHIFT;
 	cma->order_per_bit = order_per_bit;
+	if (node)
+		cma->preallocated_cma = of_get_flat_dt_prop(node, "linux,preallocated-cma", NULL);
+
 	*res_cma = cma;
 	cma_area_count++;
 	totalcma_pages += (size / PAGE_SIZE);
@@ -369,7 +386,7 @@ int __init cma_declare_contiguous_nid(phys_addr_t base,
 		base = addr;
 	}
 
-	ret = cma_init_reserved_mem(base, size, order_per_bit, name, res_cma);
+	ret = cma_init_reserved_mem(base, size, order_per_bit, name, res_cma, 0);
 	if (ret)
 		goto free_mem;
 
@@ -471,6 +488,16 @@ struct page *cma_alloc(struct cma *cma, unsigned long count,
 		spin_unlock_irq(&cma->lock);
 
 		pfn = cma->base_pfn + (bitmap_no << cma->order_per_bit);
+
+		/*
+		 * cma bitmap should ensure that pfn is in the cma.
+		 */
+		if (cma->preallocated_cma) {
+			BUG_ON(pfn + count > cma->base_pfn + cma->count);
+			page = pfn_to_page(pfn);
+			ret = 0;
+			break;
+		}
 		ret = alloc_contig_range(pfn, pfn + count, MIGRATE_CMA,
 				     GFP_KERNEL | (no_warn ? __GFP_NOWARN : 0));
 
@@ -551,7 +578,9 @@ bool cma_release(struct cma *cma, const struct page *pages,
 
 	VM_BUG_ON(pfn + count > cma->base_pfn + cma->count);
 
-	free_contig_range(pfn, count);
+	if (!cma->preallocated_cma)
+		free_contig_range(pfn, count);
+
 	cma_clear_bitmap(cma, pfn, count);
 	trace_cma_release(cma->name, pfn, pages, count);
 
diff --git a/mm/cma.h b/mm/cma.h
index 2c775877eae2..1778cb0e68c4 100644
--- a/mm/cma.h
+++ b/mm/cma.h
@@ -30,6 +30,7 @@ struct cma {
 	/* kobject requires dynamic object */
 	struct cma_kobject *cma_kobj;
 #endif
+	bool preallocated_cma;
 };
 
 extern struct cma cma_areas[MAX_CMA_AREAS];
-- 
2.25.1