Skip to contentSkip to conversation

Revert "nvmem: add new config option"

PATCH In review Rafał Miłecki 2 participants 2 messages Oct 23, 2023 – Oct 27, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Revert "nvmem: add new config option"

From: Rafał Miłecki <[email protected]>

This reverts commit 517f14d9cf3533d5ab4fded195ab6f80a92e378f.

Config option "no_of_node" is no longer needed since adding a more explicit and targeted option "add_legacy_fixed_of_cells".

That "no_of_node" config option was needed *earlier* to help mtd's case.

DT nodes of MTD partitions (that are also NVMEM devices) may contain subnodes. Those SHOULD NOT be treated as NVMEM fixed cells.

To prevent NVMEM core code from parsing subnodes a "no_of_node" option was added (and set to true in mtd) to make for_each_child_of_node() in NVMEM a no-op. That was a bit hacky because it was messing with "of_node" pointer to achieve some side-effect.

With the introduction of "add_legacy_fixed_of_cells" config option things got more explicit. MTD subsystem simply tells NVMEM when to look for fixed cells and there is no need to hack "of_node" pointer anymore.

---
Greg: I reworded this PATCH's commit message. It's hopefully clear now
that this revert is OK only with the "add_legacy_fixed_of_cells".

---
 drivers/mtd/mtdcore.c          | 1 -
 drivers/nvmem/core.c           | 2 +-
 include/linux/nvmem-provider.h | 2 --
 3 files changed, 1 insertion(+), 4 deletions(-)
drivers/mtd/mtdcore.c +0 -1
index 320d8ad2babe..bb0759ca12f1 100644
@@ -560,7 +560,6 @@ static int mtd_nvmem_add(struct mtd_info *mtd)
 	config.read_only = true;
 	config.root_only = true;
 	config.ignore_wp = true;
-	config.no_of_node = !of_device_is_compatible(node, "nvmem-cells");
 	config.priv = mtd;
 
 	mtd->nvmem = nvmem_register(&config);
drivers/nvmem/core.c +1 -1
index 2710943f53c4..bf42b7e826db 100644
@@ -935,7 +935,7 @@ struct nvmem_device *nvmem_register(const struct nvmem_config *config)
 	nvmem->nkeepout = config->nkeepout;
 	if (config->of_node)
 		nvmem->dev.of_node = config->of_node;
-	else if (!config->no_of_node)
+	else
 		nvmem->dev.of_node = config->dev->of_node;
 
 	switch (config->id) {
include/linux/nvmem-provider.h +0 -2
index 1b81adebdb8b..e3930835235b 100644
@@ -89,7 +89,6 @@ struct nvmem_cell_info {
  * @read_only:	Device is read-only.
  * @root_only:	Device is accessibly to root only.
  * @of_node:	If given, this will be used instead of the parent's of_node.
- * @no_of_node:	Device should not use the parent's of_node even if it's !NULL.
  * @reg_read:	Callback to read data.
  * @reg_write:	Callback to write data.
  * @size:	Device size.
@@ -122,7 +121,6 @@ struct nvmem_config {
 	bool			ignore_wp;
 	struct nvmem_layout	*layout;
 	struct device_node	*of_node;
-	bool			no_of_node;
 	nvmem_reg_read_t	reg_read;
 	nvmem_reg_write_t	reg_write;
 	int	size;
Signature · 1 line
2.35.3
#1 lore raw
Quoted 21 lines · Rafał Miłecki

From: Rafał Miłecki <[email protected]>

This reverts commit 517f14d9cf3533d5ab4fded195ab6f80a92e378f.

Config option "no_of_node" is no longer needed since adding a more explicit and targeted option "add_legacy_fixed_of_cells".

That "no_of_node" config option was needed *earlier* to help mtd's case.

DT nodes of MTD partitions (that are also NVMEM devices) may contain subnodes. Those SHOULD NOT be treated as NVMEM fixed cells.

To prevent NVMEM core code from parsing subnodes a "no_of_node" option was added (and set to true in mtd) to make for_each_child_of_node() in NVMEM a no-op. That was a bit hacky because it was messing with "of_node" pointer to achieve some side-effect.

With the introduction of "add_legacy_fixed_of_cells" config option things got more explicit. MTD subsystem simply tells NVMEM when to look for fixed cells and there is no need to hack "of_node" pointer anymore.

Quoted 6 lines

---
Greg: I reworded this PATCH's commit message. It's hopefully clear now
that this revert is OK only with the "add_legacy_fixed_of_cells".

Much nicer, now applied, thanks!

greg k-h