[PATCH v3 0/7] Refactor reserved memory regions handling code

[PATCH v3 0/7] Refactor reserved memory regions handling code

[Marek Szyprowski]

Hello,

The reserved memory regions handling code was reworked to handle
unlimited so called "static" memory nodes in commit 00c9a452a235 ("of:
reserved_mem: Add code to dynamically allocate reserved_mem array").

The side effect of this rework was a set of bugs fixed later by commits
0fd17e598333 ("of: reserved_mem: Allow reserved_mem framework detect
"cma=" kernel param") and 2c223f7239f3 ("of: reserved_mem: Restructure
call site for dma_contiguous_early_fixup()"). As a result, the code in
drivers/of/of_reserved_mem.c became a mix of generic code and CMA
specific fixups.

In this patchset I try to untangle this spaghetti and perform some code
cleanup. I hope nothing breaks this time.

Best regards
Marek Szyprowski, PhD
Samsung R&D Institute Poland


Changelog:

v3:
- fixed more issues pointed by Sashiko in
  https://sashiko.dev/#/patchset/20260323100901.4079171-1-m.szyprowski@samsung.com
  (restored use of _OF_DECLARE macro, extended some comments and commit
  descriptions, the remaining items I consider not relevant)

v2: https://lore.kernel.org/all/20260323100901.4079171-1-m.szyprowski@samsung.com/
- added missing ops assignment removal in tegra210-emc-table and swiotlb
  drivers
- fixed issues pointed by kernel test robot and Sashiko: removed typos,
  improved comments
- fixed incorrect node passed to fdt_validate_reserved_mem_node() in
  fdt_scan_reserved_mem_reg_nodes()

v1: https://lore.kernel.org/all/20260313150802.1121442-1-m.szyprowski@samsung.com/
- initial version


Patch summary:

Marek Szyprowski (7):
  of: reserved_mem: remove fdt node from the structure
  of: reserved_mem: use -ENODEV instead of -ENOENT
  of: reserved_mem: switch to ops based OF_DECLARE()
  of: reserved_mem: replace CMA quirks by generic methods
  of: reserved_mem: rearrange code a bit
  of: reserved_mem: clarify fdt_scan_reserved_mem*() functions
  of: reserved_mem: rework fdt_init_reserved_mem_node()

 drivers/memory/tegra/tegra210-emc-table.c |  19 +-
 drivers/of/fdt.c                          |   2 +-
 drivers/of/of_private.h                   |   2 +-
 drivers/of/of_reserved_mem.c              | 320 +++++++++++++---------
 include/linux/cma.h                       |  10 -
 include/linux/dma-map-ops.h               |   3 -
 include/linux/of_reserved_mem.h           |  16 +-
 kernel/dma/coherent.c                     |  19 +-
 kernel/dma/contiguous.c                   |  86 ++++--
 kernel/dma/swiotlb.c                      |  19 +-
 10 files changed, 285 insertions(+), 211 deletions(-)

-- 
2.34.1

[PATCH v3 1/7] of: reserved_mem: remove fdt node from the structure

[Marek Szyprowski]

FDT node is not needed for anything besides the initialization, so it can
be simply passed as an argument to the reserved memory region init
function.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/memory/tegra/tegra210-emc-table.c |  3 ++-
 drivers/of/of_reserved_mem.c              | 25 ++++++++++-------------
 include/linux/of_reserved_mem.h           |  4 ++--
 kernel/dma/coherent.c                     |  4 +---
 kernel/dma/contiguous.c                   |  3 +--
 kernel/dma/swiotlb.c                      |  5 ++---
 6 files changed, 19 insertions(+), 25 deletions(-)

diff --git a/drivers/memory/tegra/tegra210-emc-table.c b/drivers/memory/tegra/tegra210-emc-table.c
index 34a8785d2861..ac1d1e13482a 100644
--- a/drivers/memory/tegra/tegra210-emc-table.c
+++ b/drivers/memory/tegra/tegra210-emc-table.c
@@ -75,7 +75,8 @@ static const struct reserved_mem_ops tegra210_emc_table_ops = {
 	.device_release = tegra210_emc_table_device_release,
 };
 
-static int tegra210_emc_table_init(struct reserved_mem *rmem)
+static int tegra210_emc_table_init(unsigned long node,
+				   struct reserved_mem *rmem)
 {
 	pr_debug("Tegra210 EMC table at %pa, size %lu bytes\n", &rmem->base,
 		 (unsigned long)rmem->size);
diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 1fd28f805610..6705b7afebf0 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -104,7 +104,8 @@ static void __init alloc_reserved_mem_array(void)
 	reserved_mem = new_array;
 }
 
-static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem);
+static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
+					      unsigned long node);
 /*
  * fdt_reserved_mem_save_node() - save fdt node for second pass initialization
  */
@@ -118,13 +119,12 @@ static void __init fdt_reserved_mem_save_node(unsigned long node, const char *un
 		return;
 	}
 
-	rmem->fdt_node = node;
 	rmem->name = uname;
 	rmem->base = base;
 	rmem->size = size;
 
 	/* Call the region specific initialization function */
-	fdt_init_reserved_mem_node(rmem);
+	fdt_init_reserved_mem_node(rmem, node);
 
 	reserved_mem_count++;
 }
@@ -483,7 +483,8 @@ static const struct of_device_id __rmem_of_table_sentinel
 /*
  * __reserved_mem_init_node() - call region specific reserved memory init code
  */
-static int __init __reserved_mem_init_node(struct reserved_mem *rmem)
+static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
+					   unsigned long node)
 {
 	extern const struct of_device_id __reservedmem_of_table[];
 	const struct of_device_id *i;
@@ -493,10 +494,10 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem)
 		reservedmem_of_init_fn initfn = i->data;
 		const char *compat = i->compatible;
 
-		if (!of_flat_dt_is_compatible(rmem->fdt_node, compat))
+		if (!of_flat_dt_is_compatible(node, compat))
 			continue;
 
-		ret = initfn(rmem);
+		ret = initfn(node, rmem);
 		if (ret == 0) {
 			pr_info("initialized node %s, compatible id %s\n",
 				rmem->name, compat);
@@ -526,11 +527,6 @@ static int __init __rmem_cmp(const void *a, const void *b)
 	if (ra->size > rb->size)
 		return 1;
 
-	if (ra->fdt_node < rb->fdt_node)
-		return -1;
-	if (ra->fdt_node > rb->fdt_node)
-		return 1;
-
 	return 0;
 }
 
@@ -564,19 +560,20 @@ static void __init __rmem_check_for_overlap(void)
 /**
  * fdt_init_reserved_mem_node() - Initialize a reserved memory region
  * @rmem: reserved_mem struct of the memory region to be initialized.
+ * @node: fdt node of the initialized region
  *
  * This function is used to call the region specific initialization
  * function for a reserved memory region.
  */
-static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem)
+static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
+					      unsigned long node)
 {
-	unsigned long node = rmem->fdt_node;
 	int err = 0;
 	bool nomap;
 
 	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
 
-	err = __reserved_mem_init_node(rmem);
+	err = __reserved_mem_init_node(rmem, node);
 	if (err != 0 && err != -ENOENT) {
 		pr_info("node %s compatible matching fail\n", rmem->name);
 		if (nomap)
diff --git a/include/linux/of_reserved_mem.h b/include/linux/of_reserved_mem.h
index f573423359f4..5159938bfe03 100644
--- a/include/linux/of_reserved_mem.h
+++ b/include/linux/of_reserved_mem.h
@@ -11,7 +11,6 @@ struct resource;
 
 struct reserved_mem {
 	const char			*name;
-	unsigned long			fdt_node;
 	const struct reserved_mem_ops	*ops;
 	phys_addr_t			base;
 	phys_addr_t			size;
@@ -25,7 +24,8 @@ struct reserved_mem_ops {
 				  struct device *dev);
 };
 
-typedef int (*reservedmem_of_init_fn)(struct reserved_mem *rmem);
+typedef int (*reservedmem_of_init_fn)(unsigned long node,
+				      struct reserved_mem *rmem);
 
 #ifdef CONFIG_OF_RESERVED_MEM
 
diff --git a/kernel/dma/coherent.c b/kernel/dma/coherent.c
index 1147497bc512..34621acbd3c5 100644
--- a/kernel/dma/coherent.c
+++ b/kernel/dma/coherent.c
@@ -367,10 +367,8 @@ static const struct reserved_mem_ops rmem_dma_ops = {
 	.device_release	= rmem_dma_device_release,
 };
 
-static int __init rmem_dma_setup(struct reserved_mem *rmem)
+static int __init rmem_dma_setup(unsigned long node, struct reserved_mem *rmem)
 {
-	unsigned long node = rmem->fdt_node;
-
 	if (of_get_flat_dt_prop(node, "reusable", NULL))
 		return -EINVAL;
 
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index c56004d314dc..81a2fa4971ee 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -475,9 +475,8 @@ static const struct reserved_mem_ops rmem_cma_ops = {
 	.device_release = rmem_cma_device_release,
 };
 
-static int __init rmem_cma_setup(struct reserved_mem *rmem)
+static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 {
-	unsigned long node = rmem->fdt_node;
 	bool default_cma = of_get_flat_dt_prop(node, "linux,cma-default", NULL);
 	struct cma *cma;
 	int err;
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 487dcf73b88d..2f53d8c599d1 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -1882,10 +1882,9 @@ static const struct reserved_mem_ops rmem_swiotlb_ops = {
 	.device_release = rmem_swiotlb_device_release,
 };
 
-static int __init rmem_swiotlb_setup(struct reserved_mem *rmem)
+static int __init rmem_swiotlb_setup(unsigned long node,
+				     struct reserved_mem *rmem)
 {
-	unsigned long node = rmem->fdt_node;
-
 	if (of_get_flat_dt_prop(node, "reusable", NULL) ||
 	    of_get_flat_dt_prop(node, "linux,cma-default", NULL) ||
 	    of_get_flat_dt_prop(node, "linux,dma-default", NULL) ||
-- 
2.34.1

[PATCH v3 2/7] of: reserved_mem: use -ENODEV instead of -ENOENT

[Marek Szyprowski]

When given reserved memory region doesn't really support given node,
return -ENODEV instead of -ENOENT. Then fix __reserved_mem_init_node()
function to properly propagate error code different from -ENODEV instead
of silently ignoring it.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/of/of_reserved_mem.c | 7 ++++---
 kernel/dma/coherent.c        | 2 +-
 kernel/dma/contiguous.c      | 2 +-
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 6705b7afebf0..9aff460a0420 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -488,9 +488,10 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 {
 	extern const struct of_device_id __reservedmem_of_table[];
 	const struct of_device_id *i;
-	int ret = -ENOENT;
+	int ret = -ENODEV;
 
-	for (i = __reservedmem_of_table; i < &__rmem_of_table_sentinel; i++) {
+	for (i = __reservedmem_of_table; ret == -ENODEV &&
+	     i < &__rmem_of_table_sentinel; i++) {
 		reservedmem_of_init_fn initfn = i->data;
 		const char *compat = i->compatible;
 
@@ -574,7 +575,7 @@ static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
 	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
 
 	err = __reserved_mem_init_node(rmem, node);
-	if (err != 0 && err != -ENOENT) {
+	if (err != 0 && err != -ENODEV) {
 		pr_info("node %s compatible matching fail\n", rmem->name);
 		if (nomap)
 			memblock_clear_nomap(rmem->base, rmem->size);
diff --git a/kernel/dma/coherent.c b/kernel/dma/coherent.c
index 34621acbd3c5..64f9ba618e19 100644
--- a/kernel/dma/coherent.c
+++ b/kernel/dma/coherent.c
@@ -370,7 +370,7 @@ static const struct reserved_mem_ops rmem_dma_ops = {
 static int __init rmem_dma_setup(unsigned long node, struct reserved_mem *rmem)
 {
 	if (of_get_flat_dt_prop(node, "reusable", NULL))
-		return -EINVAL;
+		return -ENODEV;
 
 #ifdef CONFIG_ARM
 	if (!of_get_flat_dt_prop(node, "no-map", NULL)) {
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index 81a2fa4971ee..e6fc6906b5c0 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -483,7 +483,7 @@ static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 
 	if (!of_get_flat_dt_prop(node, "reusable", NULL) ||
 	    of_get_flat_dt_prop(node, "no-map", NULL))
-		return -EINVAL;
+		return -ENODEV;
 
 	if (!IS_ALIGNED(rmem->base | rmem->size, CMA_MIN_ALIGNMENT_BYTES)) {
 		pr_err("Reserved memory: incorrect alignment of CMA region\n");
-- 
2.34.1

[PATCH v3 3/7] of: reserved_mem: switch to ops based OF_DECLARE()

[Marek Szyprowski]

Move init function from OF_DECLARE() argument to the given reserved
memory region ops structure and then pass that structure to the
OF_DECLARE() initializer. This node_init callback is mandatory for the
reserved mem driver. Such change makes it possible in the future to add
more functions called by the generic code before given memory region is
initialized and rmem object is created.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/memory/tegra/tegra210-emc-table.c | 16 ++++++++--------
 drivers/of/of_reserved_mem.c              | 17 +++++++++++++----
 include/linux/of_reserved_mem.h           | 13 ++++++-------
 kernel/dma/coherent.c                     | 13 +++++++------
 kernel/dma/contiguous.c                   | 15 ++++++++-------
 kernel/dma/swiotlb.c                      | 14 +++++++-------
 6 files changed, 49 insertions(+), 39 deletions(-)

diff --git a/drivers/memory/tegra/tegra210-emc-table.c b/drivers/memory/tegra/tegra210-emc-table.c
index ac1d1e13482a..4b3c478b2743 100644
--- a/drivers/memory/tegra/tegra210-emc-table.c
+++ b/drivers/memory/tegra/tegra210-emc-table.c
@@ -70,20 +70,20 @@ static void tegra210_emc_table_device_release(struct reserved_mem *rmem,
 	memunmap(timings);
 }
 
-static const struct reserved_mem_ops tegra210_emc_table_ops = {
-	.device_init = tegra210_emc_table_device_init,
-	.device_release = tegra210_emc_table_device_release,
-};
-
 static int tegra210_emc_table_init(unsigned long node,
 				   struct reserved_mem *rmem)
 {
 	pr_debug("Tegra210 EMC table at %pa, size %lu bytes\n", &rmem->base,
 		 (unsigned long)rmem->size);
 
-	rmem->ops = &tegra210_emc_table_ops;
-
 	return 0;
 }
+
+static const struct reserved_mem_ops tegra210_emc_table_ops = {
+	.node_init = tegra210_emc_table_init,
+	.device_init = tegra210_emc_table_device_init,
+	.device_release = tegra210_emc_table_device_release,
+};
+
 RESERVEDMEM_OF_DECLARE(tegra210_emc_table, "nvidia,tegra210-emc-table",
-		       tegra210_emc_table_init);
+		       &tegra210_emc_table_ops);
diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 9aff460a0420..4dd0d6f6a4b0 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -480,8 +480,16 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
 static const struct of_device_id __rmem_of_table_sentinel
 	__used __section("__reservedmem_of_table_end");
 
-/*
- * __reserved_mem_init_node() - call region specific reserved memory init code
+/**
+ * __reserved_mem_init_node() - initialize a reserved memory region
+ * @rmem: reserved_mem structure to initialize
+ * @node: FDT node describing the reserved memory region
+ *
+ * This function iterates through the reserved memory drivers and calls the
+ * node_init callback for the compatible entry matching the node. On success,
+ * the operations pointer is stored in the reserved_mem structure.
+ *
+ * Return: 0 on success, -ENODEV if no compatible match found
  */
 static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 					   unsigned long node)
@@ -492,14 +500,15 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 
 	for (i = __reservedmem_of_table; ret == -ENODEV &&
 	     i < &__rmem_of_table_sentinel; i++) {
-		reservedmem_of_init_fn initfn = i->data;
+		const struct reserved_mem_ops *ops = i->data;
 		const char *compat = i->compatible;
 
 		if (!of_flat_dt_is_compatible(node, compat))
 			continue;
 
-		ret = initfn(node, rmem);
+		ret = ops->node_init(node, rmem);
 		if (ret == 0) {
+			rmem->ops = ops;
 			pr_info("initialized node %s, compatible id %s\n",
 				rmem->name, compat);
 			break;
diff --git a/include/linux/of_reserved_mem.h b/include/linux/of_reserved_mem.h
index 5159938bfe03..747a1e73d5dd 100644
--- a/include/linux/of_reserved_mem.h
+++ b/include/linux/of_reserved_mem.h
@@ -18,19 +18,17 @@ struct reserved_mem {
 };
 
 struct reserved_mem_ops {
+	int	(*node_init)(unsigned long fdt_node, struct reserved_mem *rmem);
 	int	(*device_init)(struct reserved_mem *rmem,
 			       struct device *dev);
 	void	(*device_release)(struct reserved_mem *rmem,
 				  struct device *dev);
 };
 
-typedef int (*reservedmem_of_init_fn)(unsigned long node,
-				      struct reserved_mem *rmem);
-
 #ifdef CONFIG_OF_RESERVED_MEM
 
-#define RESERVEDMEM_OF_DECLARE(name, compat, init)			\
-	_OF_DECLARE(reservedmem, name, compat, init, reservedmem_of_init_fn)
+#define RESERVEDMEM_OF_DECLARE(name, compat, ops)			\
+	_OF_DECLARE(reservedmem, name, compat, ops, struct reserved_mem_ops *)
 
 int of_reserved_mem_device_init_by_idx(struct device *dev,
 				       struct device_node *np, int idx);
@@ -48,8 +46,9 @@ int of_reserved_mem_region_count(const struct device_node *np);
 
 #else
 
-#define RESERVEDMEM_OF_DECLARE(name, compat, init)			\
-	_OF_DECLARE_STUB(reservedmem, name, compat, init, reservedmem_of_init_fn)
+#define RESERVEDMEM_OF_DECLARE(name, compat, ops)			\
+	_OF_DECLARE_STUB(reservedmem, name, compat, ops,		\
+			 struct reserved_mem_ops *)
 
 static inline int of_reserved_mem_device_init_by_idx(struct device *dev,
 					struct device_node *np, int idx)
diff --git a/kernel/dma/coherent.c b/kernel/dma/coherent.c
index 64f9ba618e19..bcdc0f76d2e8 100644
--- a/kernel/dma/coherent.c
+++ b/kernel/dma/coherent.c
@@ -362,10 +362,6 @@ static void rmem_dma_device_release(struct reserved_mem *rmem,
 		dev->dma_mem = NULL;
 }
 
-static const struct reserved_mem_ops rmem_dma_ops = {
-	.device_init	= rmem_dma_device_init,
-	.device_release	= rmem_dma_device_release,
-};
 
 static int __init rmem_dma_setup(unsigned long node, struct reserved_mem *rmem)
 {
@@ -388,7 +384,6 @@ static int __init rmem_dma_setup(unsigned long node, struct reserved_mem *rmem)
 	}
 #endif
 
-	rmem->ops = &rmem_dma_ops;
 	pr_info("Reserved memory: created DMA memory pool at %pa, size %ld MiB\n",
 		&rmem->base, (unsigned long)rmem->size / SZ_1M);
 	return 0;
@@ -405,5 +400,11 @@ static int __init dma_init_reserved_memory(void)
 core_initcall(dma_init_reserved_memory);
 #endif /* CONFIG_DMA_GLOBAL_POOL */
 
-RESERVEDMEM_OF_DECLARE(dma, "shared-dma-pool", rmem_dma_setup);
+static const struct reserved_mem_ops rmem_dma_ops = {
+	.node_init	= rmem_dma_setup,
+	.device_init	= rmem_dma_device_init,
+	.device_release	= rmem_dma_device_release,
+};
+
+RESERVEDMEM_OF_DECLARE(dma, "shared-dma-pool", &rmem_dma_ops);
 #endif
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index e6fc6906b5c0..efeebda92537 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -470,11 +470,6 @@ static void rmem_cma_device_release(struct reserved_mem *rmem,
 	dev->cma_area = NULL;
 }
 
-static const struct reserved_mem_ops rmem_cma_ops = {
-	.device_init	= rmem_cma_device_init,
-	.device_release = rmem_cma_device_release,
-};
-
 static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 {
 	bool default_cma = of_get_flat_dt_prop(node, "linux,cma-default", NULL);
@@ -499,7 +494,6 @@ static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 	if (default_cma)
 		dma_contiguous_default_area = cma;
 
-	rmem->ops = &rmem_cma_ops;
 	rmem->priv = cma;
 
 	pr_info("Reserved memory: created CMA memory pool at %pa, size %ld MiB\n",
@@ -511,5 +505,12 @@ static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 
 	return 0;
 }
-RESERVEDMEM_OF_DECLARE(cma, "shared-dma-pool", rmem_cma_setup);
+
+static const struct reserved_mem_ops rmem_cma_ops = {
+	.node_init	= rmem_cma_setup,
+	.device_init	= rmem_cma_device_init,
+	.device_release = rmem_cma_device_release,
+};
+
+RESERVEDMEM_OF_DECLARE(cma, "shared-dma-pool", &rmem_cma_ops);
 #endif
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 2f53d8c599d1..9a15e7231e39 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -1877,11 +1877,6 @@ static void rmem_swiotlb_device_release(struct reserved_mem *rmem,
 	dev->dma_io_tlb_mem = &io_tlb_default_mem;
 }
 
-static const struct reserved_mem_ops rmem_swiotlb_ops = {
-	.device_init = rmem_swiotlb_device_init,
-	.device_release = rmem_swiotlb_device_release,
-};
-
 static int __init rmem_swiotlb_setup(unsigned long node,
 				     struct reserved_mem *rmem)
 {
@@ -1891,11 +1886,16 @@ static int __init rmem_swiotlb_setup(unsigned long node,
 	    of_get_flat_dt_prop(node, "no-map", NULL))
 		return -EINVAL;
 
-	rmem->ops = &rmem_swiotlb_ops;
 	pr_info("Reserved memory: created restricted DMA pool at %pa, size %ld MiB\n",
 		&rmem->base, (unsigned long)rmem->size / SZ_1M);
 	return 0;
 }
 
-RESERVEDMEM_OF_DECLARE(dma, "restricted-dma-pool", rmem_swiotlb_setup);
+static const struct reserved_mem_ops rmem_swiotlb_ops = {
+	.node_init = rmem_swiotlb_setup,
+	.device_init = rmem_swiotlb_device_init,
+	.device_release = rmem_swiotlb_device_release,
+};
+
+RESERVEDMEM_OF_DECLARE(dma, "restricted-dma-pool", &rmem_swiotlb_ops);
 #endif /* CONFIG_DMA_RESTRICTED_POOL */
-- 
2.34.1

[PATCH v3 4/7] of: reserved_mem: replace CMA quirks by generic methods

[Marek Szyprowski]

Add optional reserved memory callbacks to perform region verification and
early fixup, then move all CMA related code in of_reserved_mem.c to them.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/of/of_reserved_mem.c    | 118 ++++++++++++++++++++++----------
 include/linux/cma.h             |  10 ---
 include/linux/dma-map-ops.h     |   3 -
 include/linux/of_reserved_mem.h |   3 +
 kernel/dma/contiguous.c         |  70 ++++++++++++++-----
 5 files changed, 137 insertions(+), 67 deletions(-)

diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 4dd0d6f6a4b0..5dd585bcf8a8 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -24,8 +24,6 @@
 #include <linux/slab.h>
 #include <linux/memblock.h>
 #include <linux/kmemleak.h>
-#include <linux/cma.h>
-#include <linux/dma-map-ops.h>
 
 #include "of_private.h"
 
@@ -106,6 +104,11 @@ static void __init alloc_reserved_mem_array(void)
 
 static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
 					      unsigned long node);
+static int fdt_validate_reserved_mem_node(unsigned long node,
+					  phys_addr_t *align);
+static int fdt_fixup_reserved_mem_node(unsigned long node,
+				       phys_addr_t base, phys_addr_t size);
+
 /*
  * fdt_reserved_mem_save_node() - save fdt node for second pass initialization
  */
@@ -154,21 +157,19 @@ static int __init __reserved_mem_reserve_reg(unsigned long node,
 					     const char *uname)
 {
 	phys_addr_t base, size;
-	int i, len;
+	int i, len, err;
 	const __be32 *prop;
-	bool nomap, default_cma;
+	bool nomap;
 
 	prop = of_flat_dt_get_addr_size_prop(node, "reg", &len);
 	if (!prop)
 		return -ENOENT;
 
 	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
-	default_cma = of_get_flat_dt_prop(node, "linux,cma-default", NULL);
 
-	if (default_cma && cma_skip_dt_default_reserved_mem()) {
-		pr_err("Skipping dt linux,cma-default for \"cma=\" kernel param.\n");
-		return -EINVAL;
-	}
+	err = fdt_validate_reserved_mem_node(node, NULL);
+	if (err && err != -ENODEV)
+		return err;
 
 	for (i = 0; i < len; i++) {
 		u64 b, s;
@@ -179,10 +180,7 @@ static int __init __reserved_mem_reserve_reg(unsigned long node,
 		size = s;
 
 		if (size && early_init_dt_reserve_memory(base, size, nomap) == 0) {
-			/* Architecture specific contiguous memory fixup. */
-			if (of_flat_dt_is_compatible(node, "shared-dma-pool") &&
-			    of_get_flat_dt_prop(node, "reusable", NULL))
-				dma_contiguous_early_fixup(base, size);
+			fdt_fixup_reserved_mem_node(node, base, size);
 			pr_debug("Reserved memory: reserved region for node '%s': base %pa, size %lu MiB\n",
 				uname, &base, (unsigned long)(size / SZ_1M));
 		} else {
@@ -253,17 +251,19 @@ void __init fdt_scan_reserved_mem_reg_nodes(void)
 
 	fdt_for_each_subnode(child, fdt, node) {
 		const char *uname;
-		bool default_cma = of_get_flat_dt_prop(child, "linux,cma-default", NULL);
 		u64 b, s;
+		int ret;
 
 		if (!of_fdt_device_is_available(fdt, child))
 			continue;
-		if (default_cma && cma_skip_dt_default_reserved_mem())
-			continue;
 
 		if (!of_flat_dt_get_addr_size(child, "reg", &b, &s))
 			continue;
 
+		ret = fdt_validate_reserved_mem_node(child, NULL);
+		if (ret && ret != -ENODEV)
+			continue;
+
 		base = b;
 		size = s;
 
@@ -397,7 +397,7 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
 	phys_addr_t base = 0, align = 0, size;
 	int i, len;
 	const __be32 *prop;
-	bool nomap, default_cma;
+	bool nomap;
 	int ret;
 
 	prop = of_get_flat_dt_prop(node, "size", &len);
@@ -421,19 +421,10 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
 	}
 
 	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
-	default_cma = of_get_flat_dt_prop(node, "linux,cma-default", NULL);
-
-	if (default_cma && cma_skip_dt_default_reserved_mem()) {
-		pr_err("Skipping dt linux,cma-default for \"cma=\" kernel param.\n");
-		return -EINVAL;
-	}
 
-	/* Need adjust the alignment to satisfy the CMA requirement */
-	if (IS_ENABLED(CONFIG_CMA)
-	    && of_flat_dt_is_compatible(node, "shared-dma-pool")
-	    && of_get_flat_dt_prop(node, "reusable", NULL)
-	    && !nomap)
-		align = max_t(phys_addr_t, align, CMA_MIN_ALIGNMENT_BYTES);
+	ret = fdt_validate_reserved_mem_node(node, &align);
+	if (ret && ret != -ENODEV)
+		return ret;
 
 	prop = of_flat_dt_get_addr_size_prop(node, "alloc-ranges", &len);
 	if (prop) {
@@ -468,18 +459,76 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
 		       uname, (unsigned long)(size / SZ_1M));
 		return -ENOMEM;
 	}
-	/* Architecture specific contiguous memory fixup. */
-	if (of_flat_dt_is_compatible(node, "shared-dma-pool") &&
-	    of_get_flat_dt_prop(node, "reusable", NULL))
-		dma_contiguous_early_fixup(base, size);
+
+	fdt_fixup_reserved_mem_node(node, base, size);
+
 	/* Save region in the reserved_mem array */
 	fdt_reserved_mem_save_node(node, uname, base, size);
 	return 0;
 }
 
+extern const struct of_device_id __reservedmem_of_table[];
 static const struct of_device_id __rmem_of_table_sentinel
 	__used __section("__reservedmem_of_table_end");
 
+/**
+ * fdt_fixup_reserved_mem_node() - call fixup function for a reserved memory node
+ * @node: FDT node to fixup
+ * @base: base address of the reserved memory region
+ * @size: size of the reserved memory region
+ *
+ * This function iterates through the reserved memory drivers and calls
+ * the node_fixup callback for the compatible entry matching the node.
+ *
+ * Return: 0 on success, -ENODEV if no compatible match found
+ */
+static int __init fdt_fixup_reserved_mem_node(unsigned long node,
+					phys_addr_t base, phys_addr_t size)
+{
+	const struct of_device_id *i;
+	int ret = -ENODEV;
+
+	for (i = __reservedmem_of_table; ret == -ENODEV &&
+	     i < &__rmem_of_table_sentinel; i++) {
+		const struct reserved_mem_ops *ops = i->data;
+
+		if (!of_flat_dt_is_compatible(node, i->compatible))
+			continue;
+
+		if (ops->node_fixup)
+			ret = ops->node_fixup(node, base, size);
+	}
+	return ret;
+}
+
+/**
+ * fdt_validate_reserved_mem_node() - validate a reserved memory node
+ * @node: FDT node to validate
+ * @align: pointer to store the validated alignment (may be modified by callback)
+ *
+ * This function iterates through the reserved memory drivers and calls
+ * the node_validate callback for the compatible entry matching the node.
+ *
+ * Return: 0 on success, -ENODEV if no compatible match found
+ */
+static int __init fdt_validate_reserved_mem_node(unsigned long node, phys_addr_t *align)
+{
+	const struct of_device_id *i;
+	int ret = -ENODEV;
+
+	for (i = __reservedmem_of_table; ret == -ENODEV &&
+	     i < &__rmem_of_table_sentinel; i++) {
+		const struct reserved_mem_ops *ops = i->data;
+
+		if (!of_flat_dt_is_compatible(node, i->compatible))
+			continue;
+
+		if (ops->node_validate)
+			ret = ops->node_validate(node, align);
+	}
+	return ret;
+}
+
 /**
  * __reserved_mem_init_node() - initialize a reserved memory region
  * @rmem: reserved_mem structure to initialize
@@ -494,7 +543,6 @@ static const struct of_device_id __rmem_of_table_sentinel
 static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 					   unsigned long node)
 {
-	extern const struct of_device_id __reservedmem_of_table[];
 	const struct of_device_id *i;
 	int ret = -ENODEV;
 
@@ -511,7 +559,7 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 			rmem->ops = ops;
 			pr_info("initialized node %s, compatible id %s\n",
 				rmem->name, compat);
-			break;
+			return ret;
 		}
 	}
 	return ret;
diff --git a/include/linux/cma.h b/include/linux/cma.h
index d0793eaaadaa..8555d38a97b1 100644
--- a/include/linux/cma.h
+++ b/include/linux/cma.h
@@ -61,14 +61,4 @@ extern int cma_for_each_area(int (*it)(struct cma *cma, void *data), void *data)
 extern bool cma_intersects(struct cma *cma, unsigned long start, unsigned long end);
 
 extern void cma_reserve_pages_on_error(struct cma *cma);
-
-#ifdef CONFIG_DMA_CMA
-extern bool cma_skip_dt_default_reserved_mem(void);
-#else
-static inline bool cma_skip_dt_default_reserved_mem(void)
-{
-	return false;
-}
-#endif
-
 #endif
diff --git a/include/linux/dma-map-ops.h b/include/linux/dma-map-ops.h
index 8a07df5a9ef6..d0642a6bc985 100644
--- a/include/linux/dma-map-ops.h
+++ b/include/linux/dma-map-ops.h
@@ -147,9 +147,6 @@ static inline void dma_free_contiguous(struct device *dev, struct page *page,
 {
 	__free_pages(page, get_order(size));
 }
-static inline void dma_contiguous_early_fixup(phys_addr_t base, unsigned long size)
-{
-}
 #endif /* CONFIG_DMA_CMA*/
 
 #ifdef CONFIG_DMA_DECLARE_COHERENT
diff --git a/include/linux/of_reserved_mem.h b/include/linux/of_reserved_mem.h
index 747a1e73d5dd..e8b20b29fa68 100644
--- a/include/linux/of_reserved_mem.h
+++ b/include/linux/of_reserved_mem.h
@@ -18,6 +18,9 @@ struct reserved_mem {
 };
 
 struct reserved_mem_ops {
+	int	(*node_validate)(unsigned long fdt_node, phys_addr_t *align);
+	int	(*node_fixup)(unsigned long fdt_node, phys_addr_t base,
+			      phys_addr_t size);
 	int	(*node_init)(unsigned long fdt_node, struct reserved_mem *rmem);
 	int	(*device_init)(struct reserved_mem *rmem,
 			       struct device *dev);
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index efeebda92537..65d216663e81 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -91,16 +91,6 @@ static int __init early_cma(char *p)
 }
 early_param("cma", early_cma);
 
-/*
- * cma_skip_dt_default_reserved_mem - This is called from the
- * reserved_mem framework to detect if the default cma region is being
- * set by the "cma=" kernel parameter.
- */
-bool __init cma_skip_dt_default_reserved_mem(void)
-{
-	return size_cmdline != -1;
-}
-
 #ifdef CONFIG_DMA_NUMA_CMA
 
 static struct cma *dma_contiguous_numa_area[MAX_NUMNODES];
@@ -470,25 +460,65 @@ static void rmem_cma_device_release(struct reserved_mem *rmem,
 	dev->cma_area = NULL;
 }
 
+static int __init __rmem_cma_verify_node(unsigned long node)
+{
+	if (!of_get_flat_dt_prop(node, "reusable", NULL) ||
+	    of_get_flat_dt_prop(node, "no-map", NULL))
+		return -ENODEV;
+
+	if (size_cmdline != -1 &&
+	    of_get_flat_dt_prop(node, "linux,cma-default", NULL)) {
+		pr_err("Skipping dt linux,cma-default node in favor for \"cma=\" kernel param.\n");
+		return -EBUSY;
+	}
+	return 0;
+}
+
+static int __init rmem_cma_validate(unsigned long node, phys_addr_t *align)
+{
+	int ret = __rmem_cma_verify_node(node);
+
+	if (ret)
+		return ret;
+
+	if (align)
+		*align = max_t(phys_addr_t, *align, CMA_MIN_ALIGNMENT_BYTES);
+
+	return 0;
+}
+
+static int __init rmem_cma_fixup(unsigned long node, phys_addr_t base,
+				    phys_addr_t size)
+{
+	int ret = __rmem_cma_verify_node(node);
+
+	if (ret)
+		return ret;
+
+	/* Architecture specific contiguous memory fixup. */
+	dma_contiguous_early_fixup(base, size);
+	return 0;
+}
+
 static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 {
 	bool default_cma = of_get_flat_dt_prop(node, "linux,cma-default", NULL);
 	struct cma *cma;
-	int err;
+	int ret;
 
-	if (!of_get_flat_dt_prop(node, "reusable", NULL) ||
-	    of_get_flat_dt_prop(node, "no-map", NULL))
-		return -ENODEV;
+	ret = __rmem_cma_verify_node(node);
+	if (ret)
+		return ret;
 
 	if (!IS_ALIGNED(rmem->base | rmem->size, CMA_MIN_ALIGNMENT_BYTES)) {
 		pr_err("Reserved memory: incorrect alignment of CMA region\n");
 		return -EINVAL;
 	}
 
-	err = cma_init_reserved_mem(rmem->base, rmem->size, 0, rmem->name, &cma);
-	if (err) {
+	ret = cma_init_reserved_mem(rmem->base, rmem->size, 0, rmem->name, &cma);
+	if (ret) {
 		pr_err("Reserved memory: unable to setup CMA region\n");
-		return err;
+		return ret;
 	}
 
 	if (default_cma)
@@ -499,14 +529,16 @@ static int __init rmem_cma_setup(unsigned long node, struct reserved_mem *rmem)
 	pr_info("Reserved memory: created CMA memory pool at %pa, size %ld MiB\n",
 		&rmem->base, (unsigned long)rmem->size / SZ_1M);
 
-	err = dma_heap_cma_register_heap(cma);
-	if (err)
+	ret = dma_heap_cma_register_heap(cma);
+	if (ret)
 		pr_warn("Couldn't register CMA heap.");
 
 	return 0;
 }
 
 static const struct reserved_mem_ops rmem_cma_ops = {
+	.node_validate  = rmem_cma_validate,
+	.node_fixup	= rmem_cma_fixup,
 	.node_init	= rmem_cma_setup,
 	.device_init	= rmem_cma_device_init,
 	.device_release = rmem_cma_device_release,
-- 
2.34.1

[PATCH v3 5/7] of: reserved_mem: rearrange code a bit

[Marek Szyprowski]

Move __rmem_check_for_overlap() and __rmem_cmp() functions before
fdt_scan_reserved_mem_reg_nodes() to avoid forward declaration and keep
related code close together.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/of/of_reserved_mem.c | 99 ++++++++++++++++++------------------
 1 file changed, 49 insertions(+), 50 deletions(-)

diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 5dd585bcf8a8..f9b6d3ebcc20 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -214,7 +214,55 @@ static int __init __reserved_mem_check_root(unsigned long node)
 	return 0;
 }
 
-static void __init __rmem_check_for_overlap(void);
+static int __init __rmem_cmp(const void *a, const void *b)
+{
+	const struct reserved_mem *ra = a, *rb = b;
+
+	if (ra->base < rb->base)
+		return -1;
+
+	if (ra->base > rb->base)
+		return 1;
+
+	/*
+	 * Put the dynamic allocations (address == 0, size == 0) before static
+	 * allocations at address 0x0 so that overlap detection works
+	 * correctly.
+	 */
+	if (ra->size < rb->size)
+		return -1;
+	if (ra->size > rb->size)
+		return 1;
+
+	return 0;
+}
+
+static void __init __rmem_check_for_overlap(void)
+{
+	int i;
+
+	if (reserved_mem_count < 2)
+		return;
+
+	sort(reserved_mem, reserved_mem_count, sizeof(reserved_mem[0]),
+	     __rmem_cmp, NULL);
+	for (i = 0; i < reserved_mem_count - 1; i++) {
+		struct reserved_mem *this, *next;
+
+		this = &reserved_mem[i];
+		next = &reserved_mem[i + 1];
+
+		if (this->base + this->size > next->base) {
+			phys_addr_t this_end, next_end;
+
+			this_end = this->base + this->size;
+			next_end = next->base + next->size;
+			pr_err("OVERLAP DETECTED!\n%s (%pa--%pa) overlaps with %s (%pa--%pa)\n",
+			       this->name, &this->base, &this_end,
+			       next->name, &next->base, &next_end);
+		}
+	}
+}
 
 /**
  * fdt_scan_reserved_mem_reg_nodes() - Store info for the "reg" defined
@@ -565,55 +613,6 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 	return ret;
 }
 
-static int __init __rmem_cmp(const void *a, const void *b)
-{
-	const struct reserved_mem *ra = a, *rb = b;
-
-	if (ra->base < rb->base)
-		return -1;
-
-	if (ra->base > rb->base)
-		return 1;
-
-	/*
-	 * Put the dynamic allocations (address == 0, size == 0) before static
-	 * allocations at address 0x0 so that overlap detection works
-	 * correctly.
-	 */
-	if (ra->size < rb->size)
-		return -1;
-	if (ra->size > rb->size)
-		return 1;
-
-	return 0;
-}
-
-static void __init __rmem_check_for_overlap(void)
-{
-	int i;
-
-	if (reserved_mem_count < 2)
-		return;
-
-	sort(reserved_mem, reserved_mem_count, sizeof(reserved_mem[0]),
-	     __rmem_cmp, NULL);
-	for (i = 0; i < reserved_mem_count - 1; i++) {
-		struct reserved_mem *this, *next;
-
-		this = &reserved_mem[i];
-		next = &reserved_mem[i + 1];
-
-		if (this->base + this->size > next->base) {
-			phys_addr_t this_end, next_end;
-
-			this_end = this->base + this->size;
-			next_end = next->base + next->size;
-			pr_err("OVERLAP DETECTED!\n%s (%pa--%pa) overlaps with %s (%pa--%pa)\n",
-			       this->name, &this->base, &this_end,
-			       next->name, &next->base, &next_end);
-		}
-	}
-}
 
 /**
  * fdt_init_reserved_mem_node() - Initialize a reserved memory region
-- 
2.34.1

[PATCH v3 6/7] of: reserved_mem: clarify fdt_scan_reserved_mem*() functions

[Marek Szyprowski]

Rename fdt_scan_reserved_mem_reg_nodes() to fdt_scan_reserved_mem_late()
to clearly show how it differs from fdt_scan_reserved_mem() and update
description of both functions.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/of/fdt.c             |  2 +-
 drivers/of/of_private.h      |  2 +-
 drivers/of/of_reserved_mem.c | 24 +++++++++++++++---------
 3 files changed, 17 insertions(+), 11 deletions(-)

diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
index 2967e4aff807..104e697bee7b 100644
--- a/drivers/of/fdt.c
+++ b/drivers/of/fdt.c
@@ -1295,7 +1295,7 @@ void __init unflatten_device_tree(void)
 	void *fdt = initial_boot_params;
 
 	/* Save the statically-placed regions in the reserved_mem array */
-	fdt_scan_reserved_mem_reg_nodes();
+	fdt_scan_reserved_mem_late();
 
 	/* Populate an empty root node when bootloader doesn't provide one */
 	if (!fdt) {
diff --git a/drivers/of/of_private.h b/drivers/of/of_private.h
index df0bb00349e0..0ae16da066e2 100644
--- a/drivers/of/of_private.h
+++ b/drivers/of/of_private.h
@@ -186,7 +186,7 @@ static inline struct device_node *__of_get_dma_parent(const struct device_node *
 #endif
 
 int fdt_scan_reserved_mem(void);
-void __init fdt_scan_reserved_mem_reg_nodes(void);
+void __init fdt_scan_reserved_mem_late(void);
 
 bool of_fdt_device_is_available(const void *blob, unsigned long node);
 
diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index f9b6d3ebcc20..037e3d74dde1 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -265,16 +265,15 @@ static void __init __rmem_check_for_overlap(void)
 }
 
 /**
- * fdt_scan_reserved_mem_reg_nodes() - Store info for the "reg" defined
- * reserved memory regions.
+ * fdt_scan_reserved_mem_late() - Scan FDT and initialize remaining reserved
+ * memory regions.
  *
- * This function is used to scan through the DT and store the
- * information for the reserved memory regions that are defined using
- * the "reg" property. The region node number, name, base address, and
- * size are all stored in the reserved_mem array by calling the
- * fdt_reserved_mem_save_node() function.
+ * This function is used to scan again through the DT and initialize the
+ * "static" reserved memory regions, that are defined using the "reg"
+ * property. Each such region is then initialized with its specific init
+ * function and stored in the global reserved_mem array.
  */
-void __init fdt_scan_reserved_mem_reg_nodes(void)
+void __init fdt_scan_reserved_mem_late(void)
 {
 	const void *fdt = initial_boot_params;
 	phys_addr_t base, size;
@@ -328,7 +327,14 @@ void __init fdt_scan_reserved_mem_reg_nodes(void)
 static int __init __reserved_mem_alloc_size(unsigned long node, const char *uname);
 
 /*
- * fdt_scan_reserved_mem() - scan a single FDT node for reserved memory
+ * fdt_scan_reserved_mem() - reserve and allocate memory occupied by
+ * reserved memory regions.
+ *
+ * This function is used to scan through the FDT and mark memory occupied
+ * by all static (defined by the "reg" property) reserved memory regions.
+ * Then memory for all dynamic regions (defined by size & alignment) is
+ * allocated, a region specific init function is called and region information
+ * is stored in the reserved_mem array.
  */
 int __init fdt_scan_reserved_mem(void)
 {
-- 
2.34.1

[PATCH v3 7/7] of: reserved_mem: rework fdt_init_reserved_mem_node()

[Marek Szyprowski]

Move the content of fdt_reserved_mem_save_node() to
fdt_init_reserved_mem_node() function. Initialization is no longer
performed in two steps as it was initially, so
fdt_reserved_mem_save_node() name is a bit misleading and that function
now performs full initialization of the reserved memory region.

This also fixes the problem of keeping pointers to the regions, which
failed to initialize, what might cause issues when such region is
assigned to the device.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/of/of_reserved_mem.c | 62 ++++++++++++++++--------------------
 1 file changed, 28 insertions(+), 34 deletions(-)

diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
index 037e3d74dde1..8d5777cb5d1b 100644
--- a/drivers/of/of_reserved_mem.c
+++ b/drivers/of/of_reserved_mem.c
@@ -102,36 +102,13 @@ static void __init alloc_reserved_mem_array(void)
 	reserved_mem = new_array;
 }
 
-static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
-					      unsigned long node);
+static void fdt_init_reserved_mem_node(unsigned long node, const char *uname,
+				       phys_addr_t base, phys_addr_t size);
 static int fdt_validate_reserved_mem_node(unsigned long node,
 					  phys_addr_t *align);
 static int fdt_fixup_reserved_mem_node(unsigned long node,
 				       phys_addr_t base, phys_addr_t size);
 
-/*
- * fdt_reserved_mem_save_node() - save fdt node for second pass initialization
- */
-static void __init fdt_reserved_mem_save_node(unsigned long node, const char *uname,
-					      phys_addr_t base, phys_addr_t size)
-{
-	struct reserved_mem *rmem = &reserved_mem[reserved_mem_count];
-
-	if (reserved_mem_count == total_reserved_mem_cnt) {
-		pr_err("not enough space for all defined regions.\n");
-		return;
-	}
-
-	rmem->name = uname;
-	rmem->base = base;
-	rmem->size = size;
-
-	/* Call the region specific initialization function */
-	fdt_init_reserved_mem_node(rmem, node);
-
-	reserved_mem_count++;
-}
-
 static int __init early_init_dt_reserve_memory(phys_addr_t base,
 					       phys_addr_t size, bool nomap)
 {
@@ -316,7 +293,7 @@ void __init fdt_scan_reserved_mem_late(void)
 
 		if (size) {
 			uname = fdt_get_name(fdt, child, NULL);
-			fdt_reserved_mem_save_node(child, uname, base, size);
+			fdt_init_reserved_mem_node(child, uname, base, size);
 		}
 	}
 
@@ -515,9 +492,8 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
 	}
 
 	fdt_fixup_reserved_mem_node(node, base, size);
+	fdt_init_reserved_mem_node(node, uname, base, size);
 
-	/* Save region in the reserved_mem array */
-	fdt_reserved_mem_save_node(node, uname, base, size);
 	return 0;
 }
 
@@ -619,30 +595,46 @@ static int __init __reserved_mem_init_node(struct reserved_mem *rmem,
 	return ret;
 }
 
-
 /**
  * fdt_init_reserved_mem_node() - Initialize a reserved memory region
- * @rmem: reserved_mem struct of the memory region to be initialized.
  * @node: fdt node of the initialized region
+ * @uname: name of the reserved memory node
+ * @base: base address of the reserved memory region
+ * @size: size of the reserved memory region
  *
- * This function is used to call the region specific initialization
- * function for a reserved memory region.
+ * This function calls the region-specific initialization function for a
+ * reserved memory region and saves all region-specific data to the
+ * reserved_mem array to allow of_reserved_mem_lookup() to find it.
  */
-static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
-					      unsigned long node)
+static void __init fdt_init_reserved_mem_node(unsigned long node, const char *uname,
+					      phys_addr_t base, phys_addr_t size)
 {
 	int err = 0;
 	bool nomap;
 
+	struct reserved_mem *rmem = &reserved_mem[reserved_mem_count];
+
+	if (reserved_mem_count == total_reserved_mem_cnt) {
+		pr_err("not enough space for all defined regions.\n");
+		return;
+	}
+
+	rmem->name = uname;
+	rmem->base = base;
+	rmem->size = size;
+
 	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
 
 	err = __reserved_mem_init_node(rmem, node);
 	if (err != 0 && err != -ENODEV) {
 		pr_info("node %s compatible matching fail\n", rmem->name);
+		rmem->name = NULL;
+
 		if (nomap)
 			memblock_clear_nomap(rmem->base, rmem->size);
 		else
 			memblock_phys_free(rmem->base, rmem->size);
+		return;
 	} else {
 		phys_addr_t end = rmem->base + rmem->size - 1;
 		bool reusable =
@@ -654,6 +646,8 @@ static void __init fdt_init_reserved_mem_node(struct reserved_mem *rmem,
 			reusable ? "reusable" : "non-reusable",
 			rmem->name ? rmem->name : "unknown");
 	}
+
+	reserved_mem_count++;
 }
 
 struct rmem_assigned_device {
-- 
2.34.1

Re: [PATCH v3 0/7] Refactor reserved memory regions handling code

[Rob Herring]

On Wed, Mar 25, 2026 at 10:00:16AM +0100, Marek Szyprowski wrote:
> Hello,
> 
> The reserved memory regions handling code was reworked to handle
> unlimited so called "static" memory nodes in commit 00c9a452a235 ("of:
> reserved_mem: Add code to dynamically allocate reserved_mem array").
> 
> The side effect of this rework was a set of bugs fixed later by commits
> 0fd17e598333 ("of: reserved_mem: Allow reserved_mem framework detect
> "cma=" kernel param") and 2c223f7239f3 ("of: reserved_mem: Restructure
> call site for dma_contiguous_early_fixup()"). As a result, the code in
> drivers/of/of_reserved_mem.c became a mix of generic code and CMA
> specific fixups.
> 
> In this patchset I try to untangle this spaghetti and perform some code
> cleanup. I hope nothing breaks this time.
> 
> Best regards
> Marek Szyprowski, PhD
> Samsung R&D Institute Poland
> 
> 
> Changelog:
> 
> v3:
> - fixed more issues pointed by Sashiko in
>   https://sashiko.dev/#/patchset/20260323100901.4079171-1-m.szyprowski@samsung.com
>   (restored use of _OF_DECLARE macro, extended some comments and commit
>   descriptions, the remaining items I consider not relevant)
> 
> v2: https://lore.kernel.org/all/20260323100901.4079171-1-m.szyprowski@samsung.com/
> - added missing ops assignment removal in tegra210-emc-table and swiotlb
>   drivers
> - fixed issues pointed by kernel test robot and Sashiko: removed typos,
>   improved comments
> - fixed incorrect node passed to fdt_validate_reserved_mem_node() in
>   fdt_scan_reserved_mem_reg_nodes()
> 
> v1: https://lore.kernel.org/all/20260313150802.1121442-1-m.szyprowski@samsung.com/
> - initial version
> 
> 
> Patch summary:
> 
> Marek Szyprowski (7):
>   of: reserved_mem: remove fdt node from the structure
>   of: reserved_mem: use -ENODEV instead of -ENOENT
>   of: reserved_mem: switch to ops based OF_DECLARE()
>   of: reserved_mem: replace CMA quirks by generic methods
>   of: reserved_mem: rearrange code a bit
>   of: reserved_mem: clarify fdt_scan_reserved_mem*() functions
>   of: reserved_mem: rework fdt_init_reserved_mem_node()

I've applied the series, thanks!

Rob

Re: [PATCH v3 0/7] Refactor reserved memory regions handling code

[Marek Szyprowski]

Hi Rob

On 25.03.2026 15:16, Rob Herring wrote:
> On Wed, Mar 25, 2026 at 10:00:16AM +0100, Marek Szyprowski wrote:
>> The reserved memory regions handling code was reworked to handle
>> unlimited so called "static" memory nodes in commit 00c9a452a235 ("of:
>> reserved_mem: Add code to dynamically allocate reserved_mem array").
>>
>> The side effect of this rework was a set of bugs fixed later by commits
>> 0fd17e598333 ("of: reserved_mem: Allow reserved_mem framework detect
>> "cma=" kernel param") and 2c223f7239f3 ("of: reserved_mem: Restructure
>> call site for dma_contiguous_early_fixup()"). As a result, the code in
>> drivers/of/of_reserved_mem.c became a mix of generic code and CMA
>> specific fixups.
>>
>> In this patchset I try to untangle this spaghetti and perform some code
>> cleanup. I hope nothing breaks this time.
>>
>> Best regards
>> Marek Szyprowski, PhD
>> Samsung R&D Institute Poland
>>
>>
>> Changelog:
>>
>> v3:
>> - fixed more issues pointed by Sashiko in
>>    https://protect2.fireeye.com/v1/url?k=c0975fdb-a11c4aed-c096d494-74fe485cbff1-5d0a8d4ece172e7b&q=1&e=abd3a151-11c5-430e-a6ac-b04cc3b34ab7&u=https%3A%2F%2Fsashiko.dev%2F%23%2Fpatchset%2F20260323100901.4079171-1-m.szyprowski%40samsung.com
>>    (restored use of _OF_DECLARE macro, extended some comments and commit
>>    descriptions, the remaining items I consider not relevant)
>>
>> v2: https://lore.kernel.org/all/20260323100901.4079171-1-m.szyprowski@samsung.com/
>> - added missing ops assignment removal in tegra210-emc-table and swiotlb
>>    drivers
>> - fixed issues pointed by kernel test robot and Sashiko: removed typos,
>>    improved comments
>> - fixed incorrect node passed to fdt_validate_reserved_mem_node() in
>>    fdt_scan_reserved_mem_reg_nodes()
>>
>> v1: https://lore.kernel.org/all/20260313150802.1121442-1-m.szyprowski@samsung.com/
>> - initial version
>>
>>
>> Patch summary:
>>
>> Marek Szyprowski (7):
>>    of: reserved_mem: remove fdt node from the structure
>>    of: reserved_mem: use -ENODEV instead of -ENOENT
>>    of: reserved_mem: switch to ops based OF_DECLARE()
>>    of: reserved_mem: replace CMA quirks by generic methods
>>    of: reserved_mem: rearrange code a bit
>>    of: reserved_mem: clarify fdt_scan_reserved_mem*() functions
>>    of: reserved_mem: rework fdt_init_reserved_mem_node()
> I've applied the series, thanks!

I forgot to mention this earlier. There are other pending changes 
tokernel/dma/contiguous.c like 
https://lore.kernel.org/all/20260303-dma-buf-heaps-as-modules-v3-0-24344812c707@kernel.org/ 
which I would like to merge to -next. This conflicts with this patchset. 
Could You provide a stable branch with those changes to let me resolve 
conflicts in kernel/dma/contiguous.c on top of it?

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

Re: [PATCH v3 0/7] Refactor reserved memory regions handling code

[Rob Herring]

On Thu, Mar 26, 2026 at 4:58 AM Marek Szyprowski
<m.szyprowski@samsung.com> wrote:
>
> Hi Rob
>
> On 25.03.2026 15:16, Rob Herring wrote:
> > On Wed, Mar 25, 2026 at 10:00:16AM +0100, Marek Szyprowski wrote:
> >> The reserved memory regions handling code was reworked to handle
> >> unlimited so called "static" memory nodes in commit 00c9a452a235 ("of:
> >> reserved_mem: Add code to dynamically allocate reserved_mem array").
> >>
> >> The side effect of this rework was a set of bugs fixed later by commits
> >> 0fd17e598333 ("of: reserved_mem: Allow reserved_mem framework detect
> >> "cma=" kernel param") and 2c223f7239f3 ("of: reserved_mem: Restructure
> >> call site for dma_contiguous_early_fixup()"). As a result, the code in
> >> drivers/of/of_reserved_mem.c became a mix of generic code and CMA
> >> specific fixups.
> >>
> >> In this patchset I try to untangle this spaghetti and perform some code
> >> cleanup. I hope nothing breaks this time.
> >>
> >> Best regards
> >> Marek Szyprowski, PhD
> >> Samsung R&D Institute Poland
> >>
> >>
> >> Changelog:
> >>
> >> v3:
> >> - fixed more issues pointed by Sashiko in
> >>    https://protect2.fireeye.com/v1/url?k=c0975fdb-a11c4aed-c096d494-74fe485cbff1-5d0a8d4ece172e7b&q=1&e=abd3a151-11c5-430e-a6ac-b04cc3b34ab7&u=https%3A%2F%2Fsashiko.dev%2F%23%2Fpatchset%2F20260323100901.4079171-1-m.szyprowski%40samsung.com
> >>    (restored use of _OF_DECLARE macro, extended some comments and commit
> >>    descriptions, the remaining items I consider not relevant)
> >>
> >> v2: https://lore.kernel.org/all/20260323100901.4079171-1-m.szyprowski@samsung.com/
> >> - added missing ops assignment removal in tegra210-emc-table and swiotlb
> >>    drivers
> >> - fixed issues pointed by kernel test robot and Sashiko: removed typos,
> >>    improved comments
> >> - fixed incorrect node passed to fdt_validate_reserved_mem_node() in
> >>    fdt_scan_reserved_mem_reg_nodes()
> >>
> >> v1: https://lore.kernel.org/all/20260313150802.1121442-1-m.szyprowski@samsung.com/
> >> - initial version
> >>
> >>
> >> Patch summary:
> >>
> >> Marek Szyprowski (7):
> >>    of: reserved_mem: remove fdt node from the structure
> >>    of: reserved_mem: use -ENODEV instead of -ENOENT
> >>    of: reserved_mem: switch to ops based OF_DECLARE()
> >>    of: reserved_mem: replace CMA quirks by generic methods
> >>    of: reserved_mem: rearrange code a bit
> >>    of: reserved_mem: clarify fdt_scan_reserved_mem*() functions
> >>    of: reserved_mem: rework fdt_init_reserved_mem_node()
> > I've applied the series, thanks!
>
> I forgot to mention this earlier. There are other pending changes
> tokernel/dma/contiguous.c like
> https://lore.kernel.org/all/20260303-dma-buf-heaps-as-modules-v3-0-24344812c707@kernel.org/
> which I would like to merge to -next. This conflicts with this patchset.
> Could You provide a stable branch with those changes to let me resolve
> conflicts in kernel/dma/contiguous.c on top of it?

Here you go:

The following changes since commit 6de23f81a5e08be8fbf5e8d7e9febc72a5b5f27f:

  Linux 7.0-rc1 (2026-02-22 13:18:59 -0800)

are available in the Git repository at:

  ssh://git@gitolite.kernel.org/pub/scm/linux/kernel/git/robh/linux.git
dt-reserved-mem-cleanups

for you to fetch changes up to 34e0e2a8ea9e9e4f4dceb33072103dffaa1366b3:

  of: reserved_mem: rework fdt_init_reserved_mem_node() (2026-03-26
14:12:02 -0500)

----------------------------------------------------------------
Marek Szyprowski (7):
      of: reserved_mem: remove fdt node from the structure
      of: reserved_mem: use -ENODEV instead of -ENOENT
      of: reserved_mem: switch to ops based OF_DECLARE()
      of: reserved_mem: replace CMA quirks by generic methods
      of: reserved_mem: rearrange code a bit
      of: reserved_mem: clarify fdt_scan_reserved_mem*() functions
      of: reserved_mem: rework fdt_init_reserved_mem_node()

 drivers/memory/tegra/tegra210-emc-table.c |  19 +++++-----
 drivers/of/fdt.c                          |   2 +-
 drivers/of/of_private.h                   |   2 +-
 drivers/of/of_reserved_mem.c              | 320
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------------------------------------------------------------
 include/linux/cma.h                       |  10 ------
 include/linux/dma-map-ops.h               |   3 --
 include/linux/of_reserved_mem.h           |  16 +++++----
 kernel/dma/coherent.c                     |  19 +++++-----
 kernel/dma/contiguous.c                   |  86
+++++++++++++++++++++++++++++++--------------
 kernel/dma/swiotlb.c                      |  19 +++++-----
 10 files changed, 285 insertions(+), 211 deletions(-)