linux-stable.git/drivers/nvmem/core.c, branch v6.8

nvmem: include bit index in cell sysfs file name

2024-02-14T15:28:16+00:00

Creating sysfs files for all Cells caused a boot failure for linux-6.8-rc1 on
Apple M1, which (in downstream dts files) has multiple nvmem cells that use the
same byte address. This causes the device probe to fail with

[    0.605336] sysfs: cannot create duplicate filename '/devices/platform/soc@200000000/2922bc000.efuse/apple_efuses_nvmem0/cells/efuse@a10'
[    0.605347] CPU: 7 PID: 1 Comm: swapper/0 Tainted: G S                 6.8.0-rc1-arnd-5+ #133
[    0.605355] Hardware name: Apple Mac Studio (M1 Ultra, 2022) (DT)
[    0.605362] Call trace:
[    0.605365]  show_stack+0x18/0x2c
[    0.605374]  dump_stack_lvl+0x60/0x80
[    0.605383]  dump_stack+0x18/0x24
[    0.605388]  sysfs_warn_dup+0x64/0x80
[    0.605395]  sysfs_add_bin_file_mode_ns+0xb0/0xd4
[    0.605402]  internal_create_group+0x268/0x404
[    0.605409]  sysfs_create_groups+0x38/0x94
[    0.605415]  devm_device_add_groups+0x50/0x94
[    0.605572]  nvmem_populate_sysfs_cells+0x180/0x1b0
[    0.605682]  nvmem_register+0x38c/0x470
[    0.605789]  devm_nvmem_register+0x1c/0x6c
[    0.605895]  apple_efuses_probe+0xe4/0x120
[    0.606000]  platform_probe+0xa8/0xd0

As far as I can tell, this is a problem for any device with multiple cells on
different bits of the same address. Avoid the issue by changing the file name
to include the first bit number.

Fixes: 0331c611949f ("nvmem: core: Expose cells through sysfs")
Link: https://github.com/AsahiLinux/linux/blob/bd0a1a7d4/arch/arm64/boot/dts/apple/t600x-dieX.dtsi#L156
Cc:  
Cc: Miquel Raynal 
Cc: Rafał Miłecki 
Cc: Chen-Yu Tsai 
Cc: Srinivas Kandagatla 
Cc: Greg Kroah-Hartman 
Cc:  
Cc: Sven Peter 
Signed-off-by: Arnd Bergmann 
Signed-off-by: Srinivas Kandagatla 
Reviewed-by: Eric Curtin 
Reviewed-by: Miquel Raynal 
Link: https://lore.kernel.org/r/20240209163454.98051-1-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: core: add nvmem_dev_size() helper

2024-01-04T16:01:13+00:00

This is required by layouts that need to read whole NVMEM content. It's
especially useful for NVMEM devices without hardcoded layout (like
U-Boot environment data block).

Signed-off-by: Rafał Miłecki 
Reviewed-by: Miquel Raynal 
Link: https://lore.kernel.org/r/20231221173421.13737-2-zajec5@gmail.com
Signed-off-by: Greg Kroah-Hartman

nvmem: drop nvmem_layout_get_match_data()

2024-01-04T16:01:13+00:00

Thanks for layouts refactoring we now have "struct device" associated
with layout. Also its OF pointer points directly to the "nvmem-layout"
DT node.

All it takes to get match data is a generic of_device_get_match_data().

Signed-off-by: Rafał Miłecki 
Reviewed-by: Miquel Raynal 
Reviewed-by: Michael Walle 
Link: https://lore.kernel.org/r/20231219120104.3422-2-zajec5@gmail.com
Signed-off-by: Greg Kroah-Hartman

nvmem: layouts: refactor .add_cells() callback arguments

2024-01-04T16:01:13+00:00

Simply pass whole "struct nvmem_layout" instead of single variables.
There is nothing in "struct nvmem_layout" that we have to hide from
layout drivers. They also access it during .probe() and .remove().

Thanks to this change:

1. API gets more consistent
   All layouts drivers callbacks get the same argument

2. Layouts get correct device
   Before this change NVMEM core code was passing NVMEM device instead
   of layout device. That resulted in:
   * Confusing prints
   * Calling devm_*() helpers on wrong device
   * Helpers like of_device_get_match_data() dereferencing NULLs

3. It gets possible to get match data
   First of all nvmem_layout_get_match_data() requires passing "struct
   nvmem_layout" which .add_cells() callback didn't have before this. It
   doesn't matter much as it's rather useless now anyway (and will be
   dropped).
   What's more important however is that of_device_get_match_data() can
   be used now thanks to owning a proper device pointer.

Signed-off-by: Rafał Miłecki 
Reviewed-by: Miquel Raynal 
Reviewed-by: Michael Walle 
Link: https://lore.kernel.org/r/20231219120104.3422-1-zajec5@gmail.com
Signed-off-by: Greg Kroah-Hartman

nvmem: core: Expose cells through sysfs

2023-12-15T12:30:08+00:00

The binary content of nvmem devices is available to the user so in the
easiest cases, finding the content of a cell is rather easy as it is
just a matter of looking at a known and fixed offset. However, nvmem
layouts have been recently introduced to cope with more advanced
situations, where the offset and size of the cells is not known in
advance or is dynamic. When using layouts, more advanced parsers are
used by the kernel in order to give direct access to the content of each
cell, regardless of its position/size in the underlying
device. Unfortunately, these information are not accessible by users,
unless by fully re-implementing the parser logic in userland.

Let's expose the cells and their content through sysfs to avoid these
situations. Of course the relevant NVMEM sysfs Kconfig option must be
enabled for this support to be available.

Not all nvmem devices expose cells. Indeed, the .bin_attrs attribute
group member will be filled at runtime only when relevant and will
remain empty otherwise. In this case, as the cells attribute group will
be empty, it will not lead to any additional folder/file creation.

Exposed cells are read-only. There is, in practice, everything in the
core to support a write path, but as I don't see any need for that, I
prefer to keep the interface simple (and probably safer). The interface
is documented as being in the "testing" state which means we can later
add a write attribute if though relevant.

Signed-off-by: Miquel Raynal 
Tested-by: Rafał Miłecki 
Tested-by: Chen-Yu Tsai 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-9-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: core: Rework layouts to become regular devices

2023-12-15T12:30:07+00:00

Current layout support was initially written without modules support in
mind. When the requirement for module support rose, the existing base
was improved to adopt modularization support, but kind of a design flaw
was introduced. With the existing implementation, when a storage device
registers into NVMEM, the core tries to hook a layout (if any) and
populates its cells immediately. This means, if the hardware description
expects a layout to be hooked up, but no driver was provided for that,
the storage medium will fail to probe and try later from
scratch. Even if we consider that the hardware description shall be
correct, we could still probe the storage device (especially if it
contains the rootfs).

One way to overcome this situation is to consider the layouts as
devices, and leverage the native notifier mechanism. When a new NVMEM
device is registered, we can populate its nvmem-layout child, if any,
and wait for the matching to be done in order to get the cells (the
waiting can be easily done with the NVMEM notifiers). If the layout
driver is compiled as a module, it should automatically be loaded. This
way, there is no strong order to enforce, any NVMEM device creation
or NVMEM layout driver insertion will be observed as a new event which
may lead to the creation of additional cells, without disturbing the
probes with costly (and sometimes endless) deferrals.

In order to achieve that goal we create a new bus for the nvmem-layouts
with minimal logic to match nvmem-layout devices with nvmem-layout
drivers. All this infrastructure code is created in the layouts.c file.

Signed-off-by: Miquel Raynal 
Tested-by: Rafał Miłecki 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-7-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: Move and rename ->fixup_cell_info()

2023-12-15T12:30:07+00:00

This hook is meant to be used by any provider and instantiating a layout
just for this is useless. Let's instead move this hook to the nvmem
device and add it to the config structure to be easily shared by the
providers.

While at moving this hook, rename it ->fixup_dt_cell_info() to clarify
its main intended purpose.

Signed-off-by: Miquel Raynal 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-6-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: Simplify the ->add_cells() hook

2023-12-15T12:30:07+00:00

The layout entry is not used and will anyway be made useless by the new
layout bus infrastructure coming next, so drop it. While at it, clarify
the kdoc entry.

Signed-off-by: Miquel Raynal 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-5-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: Create a header for internal sharing

2023-12-15T12:30:07+00:00

Before adding all the NVMEM layout bus infrastructure to the core, let's
move the main nvmem_device structure in an internal header, only
available to the core. This way all the additional code can be added in
a dedicated file in order to keep the current core file tidy.

Signed-off-by: Miquel Raynal 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-4-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman

nvmem: Move of_nvmem_layout_get_container() in another header

2023-12-15T12:30:07+00:00

nvmem-consumer.h is included by consumer devices, extracting data from
NVMEM devices whereas nvmem-provider.h is included by devices providing
NVMEM content.

The only users of of_nvmem_layout_get_container() outside of the core
are layout drivers, so better move its prototype to nvmem-provider.h.

While we do so, we also move the kdoc associated with the function to
the header rather than the .c file.

Signed-off-by: Miquel Raynal 
Signed-off-by: Srinivas Kandagatla 
Link: https://lore.kernel.org/r/20231215111536.316972-3-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman