drivers: flash: flash_hp_ra: perform blank check before reading #88113

jeremydick · 2025-04-03T13:13:38Z

The value read from unwritten areas of Renesas RAxxx SoCs data flash is undefined. To prevent reading unwritten areas a blank check command is performed first. If the area is blank, we return dummy data so it behaves the same as other flash devices.

github-actions · 2025-04-03T13:14:15Z

Hello @jeremydick, and thank you very much for your first pull request to the Zephyr project!
Our Continuous Integration pipeline will execute a series of checks on your Pull Request commit messages and code, and you are expected to address any failures by updating the PR. Please take a look at our commit message guidelines to find out how to format your commit messages, and at our contribution workflow to understand how to update your Pull Request. If you haven't already, please make sure to review the project's Contributor Expectations and update (by amending and force-pushing the commits) your pull request if necessary.
If you are stuck or need help please join us on Discord and ask your question there. Additionally, you can escalate the review when applicable. 😊

de-nordic · 2025-04-03T13:42:41Z

@KhiemNguyenT re-assigning to you, your IC.

quytranpzz · 2025-04-04T06:05:17Z

drivers/flash/Kconfig.renesas_ra


+config CHECK_BEFORE_READING
+	bool "Verify area before reading it"
+	default y


Not all Renesas RA MCU families with flash hp have undefined value after erasing. Can we add erased-undefined properties in dts and enable this config based on it? like default $(dt_nodelabel_bool_prop,flash1,erased-undefined)

Yes, it looks like the RA4E1, RA4E2, RA4M2, RA4M3, RA6E1, RA6E2, RA6M1, RA6M2, RA6M3, RA6M4, RA6M5, RA8D1, RA8M1, and RA8T1 families have the data flash the will readback an undefined value when erased. Is that all of them, or are there any others I have missed?

Sorry for late response. There is currently a PR to support the RA4L1 #88036, and the erase value for it is 0xff (RA4L1). Thanks for the classification.

Thanks. Since the erase value for the RA4L1 family is 0xff nothing needs to be done for it, only families that have an undefined erase value need to have the erased-undefined property added

quytranpzz · 2025-04-04T06:05:37Z

drivers/flash/flash_hp_ra.c

 	LOG_DBG("flash: read 0x%lx, len: %u", (long)(offset + flash_data->area_address), len);

-	memcpy(data, (uint8_t *)(offset + flash_data->area_address), len);
+#ifdef CONFIG_CHECK_BEFORE_READING


Can we move the data flash region check here instead of inside the is_area_readable. Then, with the code flash region, we won't need to call is_area_readable

The flash region check has been moved

quytranpzz · 2025-04-21T04:58:50Z

@KhiemNguyenT , @thaoluonguw , @duynguyenxa , @thenguyenyf, @khoa-nguyen-18
Performing a blank check when the erase value is undefined would improve the current flash driver. Please take a look.

khoa-nguyen-18

LGTM

drivers/flash/flash_hp_ra.c

drivers/flash/flash_hp_ra.h

dts/bindings/mtd/renesas,ra-nv-flash.yaml

de-nordic · 2025-04-23T15:35:52Z

drivers/flash/flash_hp_ra.c

+#ifdef CONFIG_CHECK_BEFORE_READING
+	} else if (rc == -ENODATA) {
+		/* Erased area, return dummy data as an erased page. */
+		memset(data, 0xFF, len);
+		rc = 0;
+#endif
+	}


Not in this PR, nevertheless: @jeremydick @khoa-nguyen-18 @thenguyenyf We need to have a discussion on defining -ENODATA return for flash_read API call.
As far as I can see here, your memory has a erasea value (0xff) as a valid write data and erased is separate state.
I think that it may be useful for users to know that, in case of your memory, you can actually detect that the memory has not yet been written at specified location and in such case function should just return without providing bogus data (erase-value filled buffer).

Hi @de-nordic, thank you for your feedback.

In my opinion, returning -ENODATA isn’t a typical way for users to check if a flash page has been erased. Most applications, including the flash/common test, simply check for (ret == 0) && (read_buffer_value == 0xFF) to confirm that the page is erased.

The issue we're facing is that the Renesas flash in some SoC series didn't have the 0xFF after erase, which causes compatibility problems with this common check. We’ve considered two options to address this:

Update the Zephyr's flash test/application condition to accommodate different erased values, and make a broad announcement to inform users of Renesas flash drivers that they need to update their condition checks.

Provide a configuration option in the Renesas flash driver that allows users to choose whether they want the erased buffer to be 0xFF (as proposed in this PR). If this configuration is not selected, we should return -ENODATA (need to update). We believe this makes sense, especially for users porting applications from other MCU vendors to Renesas MCUs who want to avoid making changes to their existing code.

As I have stated, we will not be able to resolve this in this PR, but we need a discussion on -ENODATA as valid return from flash_read.

Hi @de-nordic, thank you for your feedback.

In my opinion, returning -ENODATA isn’t a typical way for users to check if a flash page has been erased. Most applications, including the flash/common test, simply check for (ret == 0) && (read_buffer_value == 0xFF) to confirm that the page is erased.

It is not for checking whether flash page is erased, it is rather for checking whether user has placed data in specified location.
Note that there is code in Zephyr that reads page and compares it against erase-value to figure out whether page has been erased (looping through buffer), so it can now be written. In case of your device the code will not be able to recognize a difference between erased area and area that has been written with erase-value.

The issue we're facing is that the Renesas flash in some SoC series didn't have the 0xFF after erase, which causes compatibility problems with this common check. We’ve considered two options to address this:

1. Update the Zephyr's flash test/application condition to accommodate different erased values, and make a broad announcement to inform users of Renesas flash drivers that they need to update their condition checks. 2. Provide a configuration option in the Renesas flash driver that allows users to choose whether they want the erased buffer to be `0xFF` (as proposed in this PR). If this configuration is not selected, we should return `-ENODATA` (need to update). We believe this makes sense, especially for users porting applications from other MCU vendors to Renesas MCUs who want to avoid making changes to their existing code.

I agree with what you did; what I mean is that we need to clarify definition of flash_read on the matter of returning -ENODATA, modify in tree code (tests, etc) and announce the change, because it is API breaking change. This will take more effort and I do not think it should happen in this PR.

@de-nordic I agree, longer term it makes the most sense to update the API to allow -ENODATA as a valid return for flash_read, and update all usages of the flash driver within the tree handle that.

thenguyenyf

Please rebase to solve conflict

jeremydick · 2025-05-05T15:18:30Z

Rebased to resolve conflicts

Copilot

Pull Request Overview

This PR enhances the Renesas RAxxx SoC data flash driver by performing a blank check prior to reading, returning dummy data when an area is erased. Key changes include:

Adding the new "erased-undefined" property in the DTS binding for configuring blank checks.
Defining new flash flags (FLASH_FLAG_BLANK and FLASH_FLAG_NOT_BLANK) and updating the flash callback accordingly.
Implementing a new is_area_readable function and modifying flash_ra_read to utilize blank checks before reading.

Reviewed Changes

Copilot reviewed 3 out of 19 changed files in this pull request and generated 1 comment.

File	Description
dts/bindings/mtd/renesas,ra-nv-data-flash.yaml	Added a new boolean property to trigger blank check behavior
drivers/flash/soc_flash_renesas_ra_hp.h	Introduced new flash flag macros under a configuration option
drivers/flash/soc_flash_renesas_ra_hp.c	Integrated a blank check into the read path via is_area_readable

Files not reviewed (16)

drivers/flash/Kconfig.renesas_ra: Language not supported
dts/arm/renesas/ra/ra4/r7fa4e10d2cfm.dtsi: Language not supported
dts/arm/renesas/ra/ra4/r7fa4e10d2cne.dtsi: Language not supported
dts/arm/renesas/ra/ra4/r7fa4e2b93cfm.dtsi: Language not supported
dts/arm/renesas/ra/ra4/r7fa4m2ad3cfp.dtsi: Language not supported
dts/arm/renesas/ra/ra4/r7fa4m3af3cfb.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6e10f2cfp.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6e2bb3cfm.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6m1ad3cfp.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6m2af3cfb.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6m3ah3cfc.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6m4af3cfb.dtsi: Language not supported
dts/arm/renesas/ra/ra6/r7fa6m5bh3cfc.dtsi: Language not supported
dts/arm/renesas/ra/ra8/r7fa8d1bhecbd.dtsi: Language not supported
dts/arm/renesas/ra/ra8/r7fa8m1ahecbd.dtsi: Language not supported
dts/arm/renesas/ra/ra8/r7fa8t1ahecbd.dtsi: Language not supported

Comments suppressed due to low confidence (1)

dts/bindings/mtd/renesas,ra-nv-data-flash.yaml:17

[nitpick] The property name 'erased-undefined' may be ambiguous given that the feature performs a blank check; consider renaming it to 'blank-check' for clarity and consistency.

erased-undefined:

drivers/flash/soc_flash_renesas_ra_hp.c

khoa-nguyen-18

I have some minor suggestions regarding your recent update

drivers/flash/Kconfig.renesas_ra

dts/bindings/mtd/renesas,ra-nv-data-flash.yaml

…flash Add an erased-undefined property for Renesas RA series MCUS data flash that will read back undefined values when erased Signed-off-by: Jeremy Dick <jdick@pivotint.com>

The value read from unwritten areas of Renesas RAxxx SoCs data flash is undefined. To prevent reading unwritten areas a blank check command is performed first. If the area is blank, we return dummy data so it behaves the same as other flash devices. Signed-off-by: Jeremy Dick <jdick@pivotint.com>

jeremydick · 2025-05-08T14:02:56Z

@KhiemNguyenT could you please review this change to perform a blank check before reading flash with undefined erase values?

KhiemNguyenT

LGTM

github-actions bot added area: Flash platform: Renesas RA Renesas Electronics Corporation, RA labels Apr 3, 2025

github-actions bot requested review from KhiemNguyenT, de-nordic, duynguyenxa, soburi and thaoluonguw April 3, 2025 13:14

github-actions bot assigned de-nordic Apr 3, 2025

de-nordic assigned KhiemNguyenT and unassigned de-nordic Apr 3, 2025

quytranpzz reviewed Apr 4, 2025

View reviewed changes

jeremydick force-pushed the renesas_flash_read_fix branch from 67461ac to a1b4517 Compare April 4, 2025 13:37

github-actions bot requested a review from rghaddab April 4, 2025 13:37

jeremydick requested a review from quytranpzz April 4, 2025 13:41

jeremydick force-pushed the renesas_flash_read_fix branch 2 times, most recently from 271089f to 52abfff Compare April 7, 2025 19:51

github-actions bot requested a review from thenguyenyf April 7, 2025 19:52

quytranpzz approved these changes Apr 8, 2025

View reviewed changes

khoa-nguyen-18 approved these changes Apr 21, 2025

View reviewed changes

thenguyenyf previously approved these changes Apr 21, 2025

View reviewed changes

drivers/flash/flash_hp_ra.c Outdated Show resolved Hide resolved

drivers/flash/flash_hp_ra.c Outdated Show resolved Hide resolved

drivers/flash/flash_hp_ra.h Outdated Show resolved Hide resolved

jeremydick dismissed thenguyenyf’s stale review via 64d6ce4 April 21, 2025 13:31

jeremydick force-pushed the renesas_flash_read_fix branch from 52abfff to 64d6ce4 Compare April 21, 2025 13:31

jeremydick requested a review from thenguyenyf April 21, 2025 13:33

thenguyenyf approved these changes Apr 22, 2025

View reviewed changes

soburi reviewed Apr 22, 2025

View reviewed changes

dts/bindings/mtd/renesas,ra-nv-flash.yaml Outdated Show resolved Hide resolved

de-nordic reviewed Apr 23, 2025

View reviewed changes

de-nordic previously approved these changes Apr 25, 2025

View reviewed changes

soburi previously approved these changes Apr 25, 2025

View reviewed changes

thenguyenyf previously approved these changes May 5, 2025

View reviewed changes

jeremydick dismissed stale reviews from thenguyenyf, soburi, and de-nordic via 8eee469 May 5, 2025 15:13

jeremydick force-pushed the renesas_flash_read_fix branch from 64d6ce4 to 8eee469 Compare May 5, 2025 15:13

jeremydick requested review from de-nordic, soburi and thenguyenyf May 5, 2025 15:18

kartben requested a review from Copilot May 6, 2025 01:07

Copilot AI reviewed May 6, 2025

View reviewed changes

drivers/flash/soc_flash_renesas_ra_hp.c Show resolved Hide resolved

khoa-nguyen-18 reviewed May 6, 2025

View reviewed changes

drivers/flash/Kconfig.renesas_ra Outdated Show resolved Hide resolved

dts/bindings/mtd/renesas,ra-nv-data-flash.yaml Outdated Show resolved Hide resolved

thenguyenyf previously approved these changes May 6, 2025

View reviewed changes

jeremydick dismissed thenguyenyf’s stale review via 7902426 May 6, 2025 12:34

jeremydick force-pushed the renesas_flash_read_fix branch from 8eee469 to 7902426 Compare May 6, 2025 12:34

de-nordic previously approved these changes May 6, 2025

View reviewed changes

jeremydick requested a review from thenguyenyf May 6, 2025 15:58

jeremydick added 2 commits May 6, 2025 14:31

dts: arm: renesas: Add erase-value-undefined property for Renesas RA …

5180710

…flash Add an erased-undefined property for Renesas RA series MCUS data flash that will read back undefined values when erased Signed-off-by: Jeremy Dick <jdick@pivotint.com>

jeremydick dismissed de-nordic’s stale review via f12a693 May 6, 2025 19:32

jeremydick force-pushed the renesas_flash_read_fix branch from 7902426 to f12a693 Compare May 6, 2025 19:32

soburi approved these changes May 6, 2025

View reviewed changes

thenguyenyf approved these changes May 7, 2025

View reviewed changes

KhiemNguyenT approved these changes May 13, 2025

View reviewed changes

KhiemNguyenT added this to the v4.2.0 milestone May 13, 2025

fabiobaltieri merged commit b5c0984 into zephyrproject-rtos:main May 13, 2025
24 checks passed

jeremydick deleted the renesas_flash_read_fix branch September 3, 2025 12:18

drivers: flash: flash_hp_ra: perform blank check before reading #88113

drivers: flash: flash_hp_ra: perform blank check before reading #88113

Uh oh!

Conversation

jeremydick commented Apr 3, 2025

Uh oh!

github-actions bot commented Apr 3, 2025

Uh oh!

de-nordic commented Apr 3, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

quytranpzz Apr 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

quytranpzz commented Apr 21, 2025

Uh oh!

khoa-nguyen-18 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

khoa-nguyen-18 Apr 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

thenguyenyf left a comment

Choose a reason for hiding this comment

Uh oh!

jeremydick commented May 5, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Uh oh!

khoa-nguyen-18 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

jeremydick commented May 8, 2025

Uh oh!

KhiemNguyenT left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

quytranpzz Apr 8, 2025 •

edited

Loading

khoa-nguyen-18 Apr 24, 2025 •

edited

Loading