NVS sample on STM32F4 fails even if the dts definition is correct #24357

akofoed · 2020-04-14T19:07:48Z

Describe the bug
Following @Laczen 's request and our investigation in #24237, tests on the STM32 show that the NVS driver in the current master fails even if the DTS is correct. The NVS driver works on zephyr-v2.2.0. I have no idea as to why.

To Reproduce

Correct the DTS storage_partition with this (taking the area to start at 1M):

               /* storage: 64KB for settings */
		storage_partition: partition@100000 {
			label = "storage";
			reg = <0x00100000 0x00010000>;
		};

Run the NVS sample on the nucleo_f429zi
compile and run
The following output occurs (I have added nvs_write rc line to main.c here):

Port /dev/ttyACM0, 18:17:13

Press CTRL-A Z for help on special keys

*** Booting Zephyr OS build zephyr-v2.2.0-1102-gf57ec57ce557  ***
No address found, adding 192.168.1.1 at id 1
nvs_write rc: -5
No key found, adding it at id 2
No Reboot counter found, adding it at id 3
Id: 4 not found, adding it
Longarray not found, adding it as id 5
[00:00:00.007,000] <inf> fs_nvs: 3 Sectors of 2048 bytes
[00:00:00.007,000] <inf> fs_nvs: alloc wra: 0, 7f0
[00:00:00.007,000] <inf> fs_nvs: data wra: 0, 0
Reboot counter history: 
 Error, no Reboot counter
*** Booting Zephyr OS build zephyr-v2.2.0-1102-gf57ec57ce557  ***
No address found, adding 192.168.1.1 at id 1
nvs_write rc: -5
No key found, adding it at id 2
No Reboot counter found, adding it at id 3
Id: 4 not found, adding it
Longarray not found, adding it as id 5
[00:00:00.007,000] <inf> fs_nvs: 3 Sectors of 2048 bytes
[00:00:00.007,000] <inf> fs_nvs: alloc wra: 0, 7f0
[00:00:00.007,000] <inf> fs_nvs: data wra: 0, 0

Expected behavior
Switch to zephyr-v2.2.0 with the above DTS still running on the nucleo_f429zi and observe the following output:

*** Booting Zephyr OS build zephyr-v2.2.0  ***
No address found, adding 192.168.1.1 at id 1
No key found, adding it at id 2
No Reboot counter found, adding it at id 3
Id: 4 not found, adding it
Longarray not found, adding it as id 5
[00:00:00.014,000] <inf> fs_nvs: 3 Sectors of 16384 bytes
[00:00:00.014,000] <inf> fs_nvs: alloc wra: 0, 3ff0
[00:00:00.014,000] <inf> fs_nvs: data wra: 0, 0
Reboot counter history: ...0
Oldest reboot counter: 0
Rebooting in ...5...4...3...2...1
*** Booting Zephyr OS build zephyr-v2.2.0  ***
Id: 1, Address: 192.168.1.1
Id: 2, Key: ff fe fd fc fb fa f9 f8 
Id: 3, Reboot_counter: 1
Id: 4, Data: DATA
Id: 5, Longarray: 0 1 2 3 4 5 6 7 8 9 a b c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 2 
[00:00:00.016,000] <inf> fs_nvs: 3 Sectors of 16384 bytes
[00:00:00.016,000] <inf> fs_nvs: alloc wra: 0, 3fc0
[00:00:00.016,000] <inf> fs_nvs: data wra: 0, a1
Reboot counter history: ...1...0
Oldest reboot counter: 0
Rebooting in ...5...4...3...2...1
*** Booting Zephyr OS build zephyr-v2.2.0  ***
Id: 1, Address: 192.168.1.1
Id: 2, Key: ff fe fd fc fb fa f9 f8

Impact
NVS not working on STM32F4

Screenshots or console output
N/A

Environment (please complete the following information):

OS: Linux (Ubuntu)
Toolchain:Zephyr SDK
Commit SHA: 071a986

Additional context
N/A

The text was updated successfully, but these errors were encountered:

FRASTM · 2020-04-16T16:09:44Z

I also see the error with nucleo_g431 but nucleo_l4r5zi is OK

FRASTM · 2020-04-17T10:16:57Z

actually for g431rb, error is :
[00:00:00.005,000] <err> flash_stm32: Read range invalid. Offset: 133112, len: 8
Reading the nvs_ate fails after the first nvs_flash_ate_rd(fs, *addr, ate);
where offset = 0x1f000 (which corresponds to the storage partition)
where addr = 0x7f0
because the / last ate in sector", do jump to previous sector */
gives an un-mapped memory address 133112 (0x207F8). This is the result of the *addr += ((fs->sector_count - 1) << ADDR_SECT_SHIFT);

Laczen · 2020-04-17T10:34:29Z

actually for g431rb, error is :
[00:00:00.005,000] <err> flash_stm32: Read range invalid. Offset: 133112, len: 8
Reading the nvs_ate fails after the first nvs_flash_ate_rd(fs, *addr, ate);
where offset = 0x1f000 (which corresponds to the storage partition)
where addr = 0x7f0 because the / last ate in sector", do jump to previous sector */
gives an un-mapped memory address 133112 (0x207F8). This is the result of the *addr += ((fs->sector_count - 1) << ADDR_SECT_SHIFT);

The definition for the g431rb does not allow 3 sectors to be used (only 2 sectors are defined as storage). Either increase the definition to 3 sectors or modify the nvs sample to use only 2 sectors.

akofoed · 2020-04-17T10:35:42Z

I have been looking at the changelog to find out what has changed between v2.2.0 and current-master but I can't find anything that explains why the flash-driver gives this -5 error.
I have done a test using flash_area_write() and that too gives a -5 (STM32F429ZI) using the 0x100000 128KB area (current-master).

FRASTM · 2020-04-17T13:29:43Z

@Laczen Thanks for your comment, which makes this sample test OK on G431, if I set the fs.sector_count = 2U; in the sample main.
BTW, even if it's a sample, this could be linked to a config or a macro.
Regarding the definition to 3 sectors of the G431, does it refer to the page size ?
storage_partition: partition@1f000 {
Because if I try to change the dts to (3 x 1024)

			label = "storage";
			reg = <0x00018000 0x00001800>

I can also make the sample pass with 3 sectors, but I get a compil warning : unit address and first address in 'reg' (0x18000) don't match for /soc/flash-controller@40022000/flash@8000000/partitions/partition@1f000

erwango · 2020-04-17T13:34:23Z

I have been looking at the changelog to find out what has changed between v2.2.0 and current-master but I can't find anything that explains why the flash-driver gives this -5 error.

@akofoed, can you rebase on top of latest master. There had been a few hickups on stm32 flash drivers few weeks ago. You might face an issue already solved. You can check samples/drivers/flash_shell: If not working you need to rebase.

Laczen · 2020-04-17T14:07:37Z

@Laczen Thanks for your comment, which makes this sample test OK on G431, if I set the fs.sector_count = 2U; in the sample main.
BTW, even if it's a sample, this could be linked to a config or a macro.
Regarding the definition to 3 sectors of the G431, does it refer to the page size ?
storage_partition: partition@1f000 {
Because if I try to change the dts to (3 x 1024)
			label = "storage";
			reg = <0x00018000 0x00001800>
I can also make the sample pass with 3 sectors, but I get a compil warning : unit address and first address in 'reg' (0x18000) don't match for /soc/flash-controller@40022000/flash@8000000/partitions/partition@1f000

The sector size should always be a multiple of the pagesize.

akofoed · 2020-04-17T15:44:07Z

@Laczen After updating master just now it works. This is NVS on current master running on nucleo_f429zi and working:

Booting Zephyr OS build zephyr-v2.2.0-1378-g071a986bb676  ***
No address found, adding 192.168.1.1 at id 1
No key found, adding it at id 2
No Reboot counter found, adding it at id 3
Id: 4 not found, adding it
Longarray not found, adding it as id 5
[00:00:00.016,000] <inf> fs_nvs: 3 Sectors of 16384 bytes
[00:00:00.016,000] <inf> fs_nvs: alloc wra: 0, 3ff0
[00:00:00.016,000] <inf> fs_nvs: data wra: 0, 0
Reboot counter history: ...0
Oldest reboot counter: 0
Rebooting in ...5...4...3...2...1
*** Booting Zephyr OS build zephyr-v2.2.0-1378-g071a986bb676  ***
Id: 1, Address: 192.168.1.1
Id: 2, Key: ff fe fd fc fb fa f9 f8 
Id: 3, Reboot_counter: 1
Id: 4, Data: DATA
Id: 5, Longarray: 0 1 2 3 4 5 6 7 8 9 a b c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 2 
[00:00:00.017,000] <inf> fs_nvs: 3 Sectors of 16384 bytes
[00:00:00.017,000] <inf> fs_nvs: alloc wra: 0, 3fc0
[00:00:00.017,000] <inf> fs_nvs: data wra: 0, a1
Reboot counter history: ...1...0
Oldest reboot counter: 0
Rebooting in ...5...4...3...2...1

Can we close this or do you need me to do more testing?

akofoed · 2020-04-18T22:24:18Z

I was testing the jedec,spi-nor version and my code says it cannot find the driver. So I tried the sample in samples/drivers/spi_flash/ and it too cannot find the driver. I have tested this on both current master and v2.2.0.

To see the error you don't need the spi-nor flash attached to your board. It is the spi-nor driver that it cannot find.
This is my addition to the DTS:

&spi1 {
	status = "okay";
	clock-frequency = <10000000>;
	cs-gpios = <&gpioe 11 0>;
	flash1: mx25r6435f@0 {
		compatible = "jedec,spi-nor";
		status = "okay";
		label = "MX25R64";
		spi-max-frequency = <10000000>;
		reg = <0>;
		jedec-id = [c2 28 17];
		size = <67108864>;
		wp-gpios = <&gpioe 9 0>;
		hold-gpios = <&gpioe 10 0>;
	};
};

&flash1 {
	partitions {
		compatible = "fixed-partitions";
		#address-cells = <1>;
		#size-cells = <1>;

		nvs_partition: partition@0 {
			label = "nvs-partition";
			reg = <0x00000000 0x00010000>;
		};
	};
};

Yes, I am using spi4 (gpioE11->E14) but I get the same error as in the sample that the driver cannot be found.
This is the output using the current master (just updated):

*** Booting Zephyr OS build zephyr-v2.2.0-1677-ge7c7f911a9cd  ***

JEDEC SPI-NOR SPI flash testing
==========================
SPI flash driver MX25R64 was not found!

Output on v2.2.0:

*** Booting Zephyr OS build zephyr-v2.2.0  ***

JEDEC SPI-NOR SPI flash testing
==========================
SPI flash driver MX25R64 was not found!

Am I missing something or can this be confirmed?

Laczen · 2020-04-19T06:50:11Z

@akofoed, make this a separate question and/or bug report. People will not find this problem in the current issue. As the present issue seems to be solved you can close it.

akofoed · 2020-04-19T08:22:57Z

I can confirm that this is working now. Closing issue.

akofoed added the bug The issue is a bug, or the PR is fixing a bug label Apr 14, 2020

akofoed closed this as completed Apr 19, 2020

carlescufi mentioned this issue Apr 21, 2020

Fail to pass samples/subsys/nvs #24237

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

NVS sample on STM32F4 fails even if the dts definition is correct #24357

NVS sample on STM32F4 fails even if the dts definition is correct #24357

akofoed commented Apr 14, 2020

FRASTM commented Apr 16, 2020

FRASTM commented Apr 17, 2020 •

edited

Loading

Laczen commented Apr 17, 2020

akofoed commented Apr 17, 2020

FRASTM commented Apr 17, 2020

erwango commented Apr 17, 2020

Laczen commented Apr 17, 2020

akofoed commented Apr 17, 2020

akofoed commented Apr 18, 2020

Laczen commented Apr 19, 2020

akofoed commented Apr 19, 2020

NVS sample on STM32F4 fails even if the dts definition is correct #24357

NVS sample on STM32F4 fails even if the dts definition is correct #24357

Comments

akofoed commented Apr 14, 2020

FRASTM commented Apr 16, 2020

FRASTM commented Apr 17, 2020 • edited Loading

Laczen commented Apr 17, 2020

akofoed commented Apr 17, 2020

FRASTM commented Apr 17, 2020

erwango commented Apr 17, 2020

Laczen commented Apr 17, 2020

akofoed commented Apr 17, 2020

akofoed commented Apr 18, 2020

Laczen commented Apr 19, 2020

akofoed commented Apr 19, 2020

FRASTM commented Apr 17, 2020 •

edited

Loading