drivers: smbus: stm32: add support for pcall api #97188

cfriedt · 2025-10-08T12:36:05Z

Add SMBus Process Call / Process Call Block APIs for STM32.

This implementation also supports PEC mode (packet error checking) and is dependent on PEC support already being supported in-tree.

For more details, please see Sections 6.5.6 and 6.5.8 of the SMBus 3.1 Specification.
https://smbus.org/specs/SMBus_3_1_20180319.pdf

etienne-lms

Commit "drivers: smbus: stm32: add support for pcall api" LGTM.

cfriedt · 2025-10-13T22:55:31Z

@erwango or @finikorg - PTAL when you have a moment.

erwango · 2025-10-15T06:56:39Z

@benediktibk

benediktibk

In the SMBus specification the process call seems to be specified differently:

There is no count for send and receive with a variable length, just a hard coded word exchange (write and read)?

There seems to be no difference between version 3.1 and 3.2 of the SMBus specification regarding the process call, btw.

benediktibk · 2025-10-15T08:43:04Z

drivers/smbus/smbus_stm32.c

+	 * This works because the STM I2C driver processes each message serially.
+	 * The addressing math assumes little-endian.
+	 */
+	msgs[3].buf = (uint8_t *)&msgs[4].len;


I think that came from @jgrowdenTT 😀

Thanks, but I can't claim credit :) it's a repeat of the idea in smbus_stm32_block_read

Probably @alewycky-tenstorrent in that case 😀

It seems a little unsafe to blindly trust a size from the wire 🙂

But since the driver already does it...

Indeed. If caller is expected to set the input value to the size available in the output buffer, it would be worth checking that the received data won't overflow memory. As you said @mathieuchopstm, it's seems something to address driver wide (possibly outside of this P-R).

etienne-lms · 2025-10-15T11:09:22Z

In the SMBus specification the process call seems to be specified differently:
(...)
There is no count for send and receive with a variable length, just a hard coded word exchange (write and read)?

I see this change implements the block write/read process call (Section "6.5.8 Block Write-Block Read Process Call" of the spec).

Indeed the "Process Call" (single word) this is already implemented by smbus_stm32_pcall() handler but currently does not support PEC. Commit message (and P-R description) should be fixed.

@cfriedt, @jgrowdenTT, did you plan to also implement single word "Process Call"?

cfriedt · 2025-10-15T15:19:06Z

did you plan to also implement single word "Process Call"?

@etienne-lms - I can take a look at it

cfriedt · 2025-10-15T15:22:10Z

In the SMBus specification the process call

Sorry - my bad, this was the pcall block API. I'll add the pcall API shortly.

Add SMBus Block Write-Block Read Process Call API for STM32. This implementation also supports PEC mode (packet error checking) and is dependent on PEC support already being supported in-tree. Signed-off-by: James Growden <jgrowden@tenstorrent.com> Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>

cfriedt · 2025-10-19T23:43:38Z

@etienne-lms @benediktibk @mathieuchopstm @erwango - please take another look

mathieuchopstm

Otherwise LGTM

mathieuchopstm · 2025-10-21T08:10:04Z

drivers/smbus/smbus_stm32.c


-	return result;
+	num_msgs = smbus_pec_num_msgs(data->config, ARRAY_SIZE(msgs));
+	smbus_write_prepare_pec(data->config, periph_addr, msgs, ARRAY_SIZE(msgs));


My understanding is that the PEC byte is read here. Is this correct?

(Also, am I misunderstanding or is the PEC byte completely ignored?? Shouldn't smbus_read_check_pec be called?)

That's correct. The transaction/PEC code for the block-write-block-read pcall should work here too.

mathieuchopstm · 2025-10-21T08:11:16Z

drivers/smbus/smbus_stm32.c

+	 * This works because the STM I2C driver processes each message serially.
+	 * The addressing math assumes little-endian.
+	 */
+	msgs[3].buf = (uint8_t *)&msgs[4].len;


It seems a little unsafe to blindly trust a size from the wire 🙂

But since the driver already does it...

Implement smbus_stm32_pcall() with packet error correction. For more information, please see chapter 6.5.6 of the SMBus specification. https://smbus.org/specs/SMBus_3_1_20180319.pdf Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>

sonarqubecloud · 2025-10-21T14:57:34Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

etienne-lms · 2025-10-21T16:02:59Z

drivers/smbus/smbus_stm32.c

+	 * This works because the STM I2C driver processes each message serially.
+	 * The addressing math assumes little-endian.
+	 */
+	msgs[3].buf = (uint8_t *)&msgs[4].len;


Indeed. If caller is expected to set the input value to the size available in the output buffer, it would be worth checking that the received data won't overflow memory. As you said @mathieuchopstm, it's seems something to address driver wide (possibly outside of this P-R).

cfriedt requested review from erwango and etienne-lms October 8, 2025 12:36

erwango requested a review from benediktibk October 8, 2025 12:38

etienne-lms reviewed Oct 8, 2025

View reviewed changes

cfriedt force-pushed the stm32-smbus-pcall branch from 021b6e2 to 59dbf70 Compare October 13, 2025 15:36

cfriedt marked this pull request as ready for review October 13, 2025 15:37

cfriedt requested a review from etienne-lms October 13, 2025 15:37

cfriedt added platform: STM32 ST Micro STM32 area: SMBus labels Oct 13, 2025

zephyrbot requested review from FRASTM, GeorgeCGV, djiatsaf-st, finikorg, gautierg-st, marwaiehm-st and mathieuchopstm October 13, 2025 15:39

zephyrbot assigned erwango Oct 13, 2025

etienne-lms previously approved these changes Oct 13, 2025

View reviewed changes

cfriedt assigned finikorg Oct 14, 2025

etienne-lms mentioned this pull request Oct 14, 2025

drivers: smbus: stm32: add block pcall. #97542

Closed

benediktibk suggested changes Oct 15, 2025

View reviewed changes

cfriedt marked this pull request as draft October 15, 2025 15:21

cfriedt dismissed etienne-lms’s stale review via e649a3d October 19, 2025 23:42

cfriedt force-pushed the stm32-smbus-pcall branch from 59dbf70 to e649a3d Compare October 19, 2025 23:42

cfriedt requested review from benediktibk and etienne-lms October 19, 2025 23:42

cfriedt marked this pull request as ready for review October 19, 2025 23:43

mathieuchopstm reviewed Oct 21, 2025

View reviewed changes

cfriedt force-pushed the stm32-smbus-pcall branch from e649a3d to 85ed3ab Compare October 21, 2025 13:56

cfriedt requested review from jgrowdenTT and mathieuchopstm October 21, 2025 13:56

etienne-lms approved these changes Oct 21, 2025

View reviewed changes

drivers: smbus: stm32: add support for pcall api #97188

Are you sure you want to change the base?

drivers: smbus: stm32: add support for pcall api #97188

Conversation

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

Uh oh!

Uh oh!

etienne-lms left a comment

Choose a reason for hiding this comment

Uh oh!

cfriedt commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

erwango commented Oct 15, 2025

Uh oh!

benediktibk left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cfriedt Oct 20, 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!

etienne-lms Oct 21, 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!

etienne-lms commented Oct 15, 2025

Uh oh!

cfriedt commented Oct 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cfriedt commented Oct 15, 2025

Uh oh!

cfriedt commented Oct 19, 2025

Uh oh!

mathieuchopstm left a comment

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!

Choose a reason for hiding this comment

Uh oh!

sonarqubecloud bot commented Oct 21, 2025

Quality Gate passed

Uh oh!

etienne-lms Oct 21, 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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

cfriedt commented Oct 8, 2025 •

edited

Loading

cfriedt commented Oct 13, 2025 •

edited

Loading

cfriedt Oct 20, 2025 •

edited

Loading

etienne-lms Oct 21, 2025 •

edited

Loading

cfriedt commented Oct 15, 2025 •

edited

Loading

etienne-lms Oct 21, 2025 •

edited

Loading