CubeMX 4.15.1 (with F7 1.4.0) broke the UART

[kevin2399]

Posted on June 25, 2016 at 01:59

My UART interface was working great under CubeMx 4.14 with STM32F7 library v1.3.  There are a lot of changes to the UART HAL (not mentioned in the release notes for CubeMx, and there are no release notes for CubeF7) in the v1.4 release.  ST even changed a structure member name from State to gState.  I thought this might be the only issue, but my UART interface does not work at all under the new configuration.  I am really growing weary of 'upgrades' causing things to break.  Looking at the Hal code, there are significant changes between the two versions. Sure would be nice to know what they were trying to 'fix'. The CubeMx interface is suppoed to save developers time, not cause them to waste days trying to figure out why a working application no longer works simply from updating CubeMX.

[Tesla DeLorean]

Posted on June 25, 2016 at 17:28

What makes me sad is that if half the effort put behind HAL had been applied to building a middleware library around SPL we'd be miles ahead...

[Andrei Chichak]

Posted on June 30, 2016 at 00:04

I'm seeing this same effect after updating to firmware 1.12.0 on STM32F4.

The transmit section seems to be working properly, but receive generates noise errors (NF) and framing errors (FE).(whining about the SPL going away will not bring it back, move on)

[pdu-fr]

Posted on July 03, 2016 at 11:57

Same here, with STM32F4 and new FW 1.12.0

I'm using DMA mode for Rx and Tx. Code seens to run, but nothing is sent through UART. Nothing is receive too. I checked signal with logic analyzer.

[Tesla DeLorean]

Posted on July 03, 2016 at 15:41

Well you enjoy your support/regression nightmare, I've moved down a different path already.

[John F.]

Posted on July 04, 2016 at 09:30

''I've moved down a different path already.''

Is that anything the rest of us could benefit from? I've been persevering with the most current SPL for STM32F4 V1.7.0.

[AvaTar]

Posted on July 04, 2016 at 10:29

> I've been persevering with the most current SPL for STM32F4 V1.7.0.

Unfortunately, this is no such option for ''later'' MCUs, including the F7 line, where no SPL exists. From earlier posts, I conclude clive speaks about direct-register-access. I'm doing the same for F7, using parts of the awkward Cube SW as template. The hyped LL part seem but a sorry stub there - except I'm overlooking something. I can see it more relaxed, using STM32 only for hobby projects. My company had chosen a different Cortex M vendor years ago.