[STM32] I2C HAL driver problems and fixes

[barthess]

P.S.: I assume you have defined STM32F1XX and neutralized the chDbgCheck() calls in i2c_lld_master_receive_timeout() and in i2c_lld_master_transmit_timeout()

Correct. I stably get I2C_LOCKED state on second read from bus.

[barthess]

Can you join us in IRC?

[MartinP]

Can you join us in IRC?

Sorry to answer that late
during day I'm @ work and can only check sporadically the forum and my Developer-Environment is far away @ home

Martin

[MartinP]

Barthess: Here is my test environment:

My quick and dirty test environment

Untitled.jpg (237.42 KiB) Viewed 5965 times

here is the source I used

chibios.svn-based on trunk r4564.rar

(60.65 KiB) Downloaded 651 times

I'm using usb-serial. type in shell command "i2c" to start the test

Martin

[barthess]

Martin,
I really can not understand why your code works. In addition to TMP75 and LIS3 I have tested 24AA eeprom IC. In all cases I get errors.

Giovanni,
Most (if not all) users start their projects by coping one from haltest or from demo. Most of them does not touch default settings in mcuconf.h because can not understand them. For normal functioning on STM32 platform I2C must have highest priority but default confings does not correspond this rule. I propose to rebalance priorities in all mcuconfs shipped with ChibiOS.

[Giovanni]

It is OK if there is no other workaround for this issue. I am not particularly happy to put such a complex ISR at highest priority.

Giovanni

[barthess]

Giovanni,
I am not happy too but this is minimal evil. Job done at trunk with small reworking shell script and manually rechecked. Script added to repo (.tools/edit_aid) too.

Martin,
We both have the same STM32F103RBT6 MCUs but with different revisions (your revision is "1", my is "Y"). Probably this is reason.