linux/drivers/nvmem, branch v6.3

nvmem: core: return -ENOENT if nvmem cell is not found

2023-03-10T09:55:49Z

Prior to commit 5d8e6e6c10a3 ("nvmem: core: add an index parameter to the cell") of_nvmem_cell_get() would return -ENOENT if the cell wasn't found. Particularly, if of_property_match_string() returned -EINVAL, that return code was passed as the index to of_parse_phandle(), which then detected it as invalid and returned NULL. That led to an return code of -ENOENT. With the new code, the negative index will lead to an -EINVAL of of_parse_phandle_with_optional_args() which pass straight to the caller and break those who expect an -ENOENT. Fix it by always returning -ENOENT. Fixes: 5d8e6e6c10a3 ("nvmem: core: add an index parameter to the cell") Reported-by: Alexander Stein Link: https://lore.kernel.org/r/2143916.GUh0CODmnK@steina-w/ Signed-off-by: Michael Walle Tested-by: Alexander Stein Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230310094845.139400-1-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: stm32: fix OPTEE dependency

2023-02-06T18:07:00Z

The stm32 nvmem driver fails to link as built-in when OPTEE is a loadable module: aarch64-linux-ld: drivers/nvmem/stm32-bsec-optee-ta.o: in function `stm32_bsec: stm32-bsec-optee-ta.c:(.text+0xc8): undefined reference to `tee_client_open_session' aarch64-linux-ld: drivers/nvmem/stm32-bsec-optee-ta.o: in function `stm32_bsec: stm32-bsec-optee-ta.c:(.text+0x1fc): undefined reference to `tee_client_open_context' Change the CONFIG_NVMEM_STM32_ROMEM definition so it can only be built-in if OPTEE is either built-in or disabled, and make NVMEM_STM32_BSEC_OPTEE_TA a hidden symbol instead. Signed-off-by: Arnd Bergmann Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-23-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: qcom-spmi-sdam: register at device init time

2023-02-06T18:06:59Z

There are currently no in-tree users of the Qualcomm SDAM nvmem driver and there is generally no point in registering a driver that can be built as a module at subsys init time. Register the driver at the normal device init time instead and let driver core sort out the probe order. Signed-off-by: Johan Hovold Reviewed-by: Bjorn Andersson Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-21-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: rave-sp-eeprm: fix kernel-doc bad line warning

2023-02-06T18:06:59Z

Convert an empty line to " *" to avoid a kernel-doc warning: drivers/nvmem/rave-sp-eeprom.c:48: warning: bad line: Signed-off-by: Randy Dunlap Cc: Srinivas Kandagatla Cc: Andrey Vostrikov Cc: Nikita Yushchenko Cc: Andrey Smirnov Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-20-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: stm32: detect bsec pta presence for STM32MP15x

2023-02-06T18:06:59Z

On STM32MP15x SoC, the SMC backend is optional when OP-TEE is used; the PTA BSEC should be used as it is done on STM32MP13x platform, but the BSEC SMC can be also used: it is a legacy mode in OP-TEE, not recommended but used in previous OP-TEE firmware. The presence of OP-TEE is dynamically detected in STM32MP15x device tree and the supported NVMEM backend is dynamically detected: - PTA with stm32_bsec_pta_find - SMC with stm32_bsec_check With OP-TEE but without PTA and SMC detection, the probe is deferred for STM32MP15x devices. On STM32MP13x platform, only the PTA is supported with cfg->ta = true and this detection is skipped. Signed-off-by: Patrick Delaunay Reviewed-by: Etienne Carriere Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-19-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: stm32: add OP-TEE support for STM32MP13x

2023-02-06T18:06:59Z

For boot with OP-TEE on STM32MP13, the communication with the secure world no more use STMicroelectronics SMC but communication with the STM32MP BSEC TA, for data access (read/write) or lock operation: - all the request are sent to OP-TEE trusted application, - for upper OTP with ECC protection and with word programming only each OTP are permanently locked when programmed to avoid ECC error on the second write operation Signed-off-by: Patrick Delaunay Reviewed-by: Etienne Carriere Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-18-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: core: use nvmem_add_one_cell() in nvmem_add_cells_from_of()

2023-02-06T18:06:59Z

Convert nvmem_add_cells_from_of() to use the new nvmem_add_one_cell(). This will remove duplicate code and it will make it possible to add a hook to a nvmem layout in between, which can change fields before the cell is finally added. Signed-off-by: Michael Walle Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-17-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: core: add nvmem_add_one_cell()

2023-02-06T18:06:59Z

Add a new function to add exactly one cell. This will be used by the nvmem layout drivers to add custom cells. In contrast to the nvmem_add_cells(), this has the advantage that we don't have to assemble a list of cells on runtime. Signed-off-by: Michael Walle Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-16-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: core: drop the removal of the cells in nvmem_add_cells()

2023-02-06T18:06:59Z

If nvmem_add_cells() fails, the whole nvmem_register() will fail and the cells will then be removed anyway. This is a preparation to introduce a nvmem_add_one_cell() which can then be used by nvmem_add_cells(). This is then the same to what nvmem_add_cells_from_table() and nvmem_add_cells_from_of() do. Signed-off-by: Michael Walle Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-15-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman

nvmem: core: add an index parameter to the cell

2023-02-06T18:06:59Z

Sometimes a cell can represend multiple values. For example, a base ethernet address stored in the NVMEM can be expanded into multiple discreet ones by adding an offset. For this use case, introduce an index parameter which is then used to distiguish between values. This parameter will then be passed to the post process hook which can then use it to create different values during reading. At the moment, there is only support for the device tree path. You can add the index to the phandle, e.g. &net { nvmem-cells = <&base_mac_address 2>; nvmem-cell-names = "mac-address"; }; &nvmem_provider { base_mac_address: base-mac-address@0 { #nvmem-cell-cells = <1>; reg = <0 6>; }; }; Signed-off-by: Michael Walle Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20230206134356.839737-13-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman