2022-08-25 21:55:47

by Michael Walle

Subject: [RFC PATCH v1 13/14] nvmem: layouts: u-boot-env: add device node

Register the device node so we can actually make use of the cells from
within the device tree.

This obviously only works if the environment variable name can be mapped
to the device node, which isn't always the case. Think of "_" vs "-".
But for simple things like ethaddr, this will work.

Signed-off-by: Michael Walle <[email protected]>
---
drivers/nvmem/layouts/u-boot-env.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/nvmem/layouts/u-boot-env.c b/drivers/nvmem/layouts/u-boot-env.c
index 40b7e9a778c4..0901cc1f5acd 100644
--- a/drivers/nvmem/layouts/u-boot-env.c
+++ b/drivers/nvmem/layouts/u-boot-env.c
@@ -5,9 +5,11 @@

#include <linux/crc32.h>
#include <linux/dev_printk.h>
+#include <linux/device.h>
#include <linux/mod_devicetable.h>
#include <linux/nvmem-consumer.h>
#include <linux/nvmem-provider.h>
+#include <linux/of.h>
#include <linux/slab.h>

enum u_boot_env_format {
@@ -26,7 +28,8 @@ struct u_boot_env_image_redundant {
u8 data[];
} __packed;

-static int u_boot_env_add_cells(struct nvmem_device *nvmem, uint8_t *buf,
+static int u_boot_env_add_cells(struct device *dev,
+ struct nvmem_device *nvmem, uint8_t *buf,
size_t data_offset, size_t data_len)
{
struct nvmem_cell_info info = {0};
@@ -46,6 +49,7 @@ static int u_boot_env_add_cells(struct nvmem_device *nvmem, uint8_t *buf,
info.name = var;
info.offset = data_offset + value - data;
info.bytes = strlen(value);
+ info.np = of_get_child_by_name(dev->of_node, var);

err = nvmem_add_one_cell(nvmem, &info);
if (err)
@@ -113,7 +117,7 @@ static int u_boot_add_cells(struct device *dev, struct nvmem_device *nvmem,
}

buf[size - 1] = '\0';
- err = u_boot_env_add_cells(nvmem, buf, data_offset, data_len);
+ err = u_boot_env_add_cells(dev, nvmem, buf, data_offset, data_len);
if (err)
dev_err(dev, "Failed to add cells: %d\n", err);

--
2.30.2

2022-08-28 14:41:48

by Michael Walle

Subject: Re: [RFC PATCH v1 13/14] nvmem: layouts: u-boot-env: add device node

Am 2022-08-28 15:55, schrieb Rafał Miłecki:
> On 25.08.2022 23:44, Michael Walle wrote:
>> Register the device node so we can actually make use of the cells from
>> within the device tree.
>>
>> This obviously only works if the environment variable name can be
>> mapped
>> to the device node, which isn't always the case. Think of "_" vs "-".
>> But for simple things like ethaddr, this will work.
>
> We probably shouldn't support undocumented syntax (bindings).

If we only support a predefined list of variables, we can make them
device tree compatible anyway. E.g. we could have a mapping
"serial-number" <-> "serial#"

-michael

> I've identical local patch that waits for
> [PATCH] dt-bindings: nvmem: u-boot,env: add basic NVMEM cells
> https://patchwork.ozlabs.org/project/devicetree-bindings/patch/[email protected]/
> to be accepted.

2022-08-28 15:23:23

by Rafał Miłecki

Subject: Re: [RFC PATCH v1 13/14] nvmem: layouts: u-boot-env: add device node

On 25.08.2022 23:44, Michael Walle wrote:
> Register the device node so we can actually make use of the cells from
> within the device tree.
>
> This obviously only works if the environment variable name can be mapped
> to the device node, which isn't always the case. Think of "_" vs "-".
> But for simple things like ethaddr, this will work.

We probably shouldn't support undocumented syntax (bindings).

I've identical local patch that waits for
[PATCH] dt-bindings: nvmem: u-boot,env: add basic NVMEM cells
https://patchwork.ozlabs.org/project/devicetree-bindings/patch/[email protected]/
to be accepted.