Discussion:
Newbie questions
(too old to reply)
'Deininger, Andreas' andreas.deininger@agr.uni-goettingen.de [nuttx]
2018-02-27 10:47:49 UTC
Permalink
Hi,

recently I got aware of NuttX which looks very interesting to me.
Now, we are considering making use of NuttX in a commercial project (up to now, we are using FreeRTOS). In our project, we want to use STM32L4R7 as MCU, we ordered STM32L4R9I-DISCO for evaluation pruposes already.

Now, I have a few questions:

I started building the NuttX simulation on the Linux Subsystem of Windows 10 yesterday and I was able to compile and log into nsh. So far so good. When building Nuttx for STM32, I would probably start using the existing config for stm32l476g-disco, correct? What are the next steps? How much effort is needed to adapt this existing configuration to our board (STM32L4R9I-DISCO)?

In general: how mature is the port for STM32L4? Is it considered experimental, production ready or something in between? I'm asking since our schedule is tight, and we don't have resources and don't want to put a lot of energy in resolving potential issues me might faced with.

I know my question night be vague, however, with help of the list, I would like to get at least a feeling what might be expecting us. In our use case, we see clear advantages to having a POSIX compatible OS (opposed to the thin layer provided by FreeRTOS). Howvever, we don't have any experience with NuttX right now which makes the decision really difficult for us. Maybe someone can shed some light on this.

Best regards
Andreas
spudarnia@yahoo.com [nuttx]
2018-02-27 12:45:00 UTC
Permalink
Hi, Andreas,

First you should know that this forum has moved to https://groups.google.com/forum/#!pendingmsg/nuttx

I believe that the L4 port is very stable. There have been several products based on the part in the past and a few more under development now.

Greg
evgen@wolpike.com [nuttx]
2018-02-27 12:51:51 UTC
Permalink
Open source software rarely works problemless out of the box (especially GPL-infected, fortunately Nuttx is not). I can share the experience of our company, we are using Nuttx with STM32F7 MCUs and we had to debug and fix 3 drivers and port 1 driver (SD card, Ethernet, ADC and input capture). And there are still some problems remaining with SD cards (so far we have to use really fast SD cards to workaround it) and network (Ethernet TX stucks several times an hour - watchdog solves it by interface reinitialization, TCP listening sockets cannot be created any more after 1-2 days of continuous traffic flow - fortunately we need only UDP). Nuttx is cool, but according to our experience many drivers have not been stress-tested.
'Juha Niskanen (Haltian)' juha.niskanen@haltian.com [nuttx]
2018-02-27 12:55:00 UTC
Permalink
Hi Andreas,


STM32L4R7 is the new L4+ architecture for which there is no support at the moment. It should not be hard to adapt the existing STM32L4 code to it as differences seem not that bad. There is new SRAM3 for larger memories and a few new peripherals like octo-SPI and new graphics controllers (TTF, DSI) for which there is zero support in NuttX. Basics like GPIO, I2C, SPI, ADC, RTC etc. are likely compatible.


Your first task is to do the architecture support, followed by board configuration for the new L4+ eval boards.


"In general: how mature is the port for STM32L4? Is it considered experimental, production ready or something in between?"


I think it is something in between, depending on the peripherals used. Some peripherals are mature, others have never been really tested, some are missing. For example, I don't use USB at all so no idea if it works or not. OS itself is very stable. My current commercial STM32L4 project has volume estimates in tens of thousands of devices, so yes this is being used in real projects.


Best Regards,

Juha Niskanen / Haltian Ltd


________________________________
From: ***@yahoogroups.com <***@yahoogroups.com> on behalf of 'Deininger, Andreas' ***@agr.uni-goettingen.de [nuttx] <***@yahoogroups.com>
Sent: Tuesday, February 27, 2018 12:47:49 PM
To: ***@yahoogroups.com
Subject: [nuttx] Newbie questions



Hi,

recently I got aware of NuttX which looks very interesting to me.
Now, we are considering making use of NuttX in a commercial project (up to now, we are using FreeRTOS). In our project, we want to use STM32L4R7 as MCU, we ordered STM32L4R9I-DISCO for evaluation pruposes already.

Now, I have a few questions:

I started building the NuttX simulation on the Linux Subsystem of Windows 10 yesterday and I was able to compile and log into nsh. So far so good. When building Nuttx for STM32, I would probably start using the existing config for stm32l476g-disco, correct? What are the next steps? How much effort is needed to adapt this existing configuration to our board (STM32L4R9I-DISCO)?

In general: how mature is the port for STM32L4? Is it considered experimental, production ready or something in between? I'm asking since our schedule is tight, and we don't have resources and don't want to put a lot of energy in resolving potential issues me might faced with.

I know my question night be vague, however, with help of the list, I would like to get at least a feeling what might be expecting us. In our use case, we see clear advantages to having a POSIX compatible OS (opposed to the thin layer provided by FreeRTOS). Howvever, we don't have any experience with NuttX right now which makes the decision really difficult for us. Maybe someone can shed some light on this.

Best regards
Andreas
Sebastien Lorquet sebastien@lorquet.fr [nuttx]
2018-02-28 12:47:25 UTC
Permalink
hello,

shall we start a new port folder or add these new devices to the existing stm32l4 port?

sebatien
Post by 'Juha Niskanen (Haltian)' ***@haltian.com [nuttx]
Hi Andreas,
STM32L4R7 is the new L4+ architecture for which there is no support at
the moment. It should not be hard to adapt the existing STM32L4 code to
it as differences seem not that bad. There is new SRAM3 for larger
memories and a few new peripherals like octo-SPI and new graphics
controllers (TTF, DSI) for which there is zero support in NuttX. Basics
like GPIO, I2C, SPI, ADC, RTC etc. are likely compatible.
Your first task is to do the architecture support, followed by board
configuration for the new L4+ eval boards.
"In general: how mature is the port for STM32L4? Is it considered
experimental, production ready or something in between?"
I think it is something in between, depending on the peripherals used.
Some peripherals are mature, others have never been really tested, some
are missing. For example, I don't use USB at all so no idea if it works
or not. OS itself is very stable. My current commercial STM32L4 project
has volume estimates in tens of thousands of devices, so yes this is
being used in real projects.
Best Regards,
Juha Niskanen / Haltian Ltd
________________________________
Sent: Tuesday, February 27, 2018 12:47:49 PM
Subject: [nuttx] Newbie questions
Hi,
recently I got aware of NuttX which looks very interesting to me.
Now, we are considering making use of NuttX in a commercial project (up
to now, we are using FreeRTOS). In our project, we want to use
STM32L4R7 as MCU, we ordered STM32L4R9I-DISCO for evaluation pruposes
already.
I started building the NuttX simulation on the Linux Subsystem of
Windows 10 yesterday and I was able to compile and log into nsh. So far
so good. When building Nuttx for STM32, I would probably start using
the existing config for stm32l476g-disco, correct? What are the next
steps? How much effort is needed to adapt this existing configuration
to our board (STM32L4R9I-DISCO)?
In general: how mature is the port for STM32L4? Is it considered
experimental, production ready or something in between? I'm asking
since our schedule is tight, and we don't have resources and don't want
to put a lot of energy in resolving potential issues me might faced
with.
I know my question night be vague, however, with help of the list, I
would like to get at least a feeling what might be expecting us. In our
use case, we see clear advantages to having a POSIX compatible OS
(opposed to the thin layer provided by FreeRTOS). Howvever, we don't
have any experience with NuttX right now which makes the decision
really difficult for us. Maybe someone can shed some light on this.
Best regards
Andreas
--
Envoyé de mon appareil Android avec Courriel K-9 Mail. Veuillez excuser ma briÚveté.
'Juha Niskanen (Haltian)' juha.niskanen@haltian.com [nuttx]
2018-03-01 08:14:42 UTC
Permalink
Hi Sebastien,


Since memory map seems compatible, I would do this in existing L4 port and only introduce new subarch macro like CONFIG_STM32L4_STM32L4RXXX to separate the new RM0432 chips from existing ones. (The existing subarchs STM32L4X3, STM32L4X5, STM32L4X6 are also based on RMs documenting the chips, not strict parse of full chip name. The naming is somewhat unfortunate here). Then treat the L4+ chips as additional L4 chips and ignore the plus sign completely; do not encode it in macros or file names as it is most likely invented for marketing purposes anyway.


Danger here is if ST keeps heavily adding to the L4+-series and introducing new (mostly) compatible series, then we are repeating the sad experience of the legacy STM32 port that tries to support all of F1/F2/F3/F4 and L1. ST has a long history of arranging their product offering into overly complex matrix of different "families", "series", "lines", "categories", "densities" etc. so it is hard to predict how they will obfuscate it in future.


- Juha



________________________________
From: ***@yahoogroups.com <***@yahoogroups.com> on behalf of Sebastien Lorquet ***@lorquet.fr [nuttx] <***@yahoogroups.com>
Sent: Wednesday, February 28, 2018 2:47:25 PM
To: ***@yahoogroups.com
Subject: Re: [nuttx] Re: Newbie questions



hello,

shall we start a new port folder or add these new devices to the existing stm32l4 port?

sebatien


Le 2018 2 27 12:55:00 UTC, "'Juha Niskanen (Haltian)' ***@haltian..com [nuttx]" <***@yahoogroups.com> a écrit :


Hi Andreas,


STM32L4R7 is the new L4+ architecture for which there is no support at the moment. It should not be hard to adapt the existing STM32L4 code to it as differences seem not that bad. There is new SRAM3 for larger memories and a few new peripherals like octo-SPI and new graphics controllers (TTF, DSI) for which there is zero support in NuttX. Basics like GPIO, I2C, SPI, ADC, RTC etc. are likely compatible.


Your first task is to do the architecture support, followed by board configuration for the new L4+ eval boards.


"In general: how mature is the port for STM32L4? Is it considered experimental, production ready or something in between?"


I think it is something in between, depending on the peripherals used. Some peripherals are mature, others have never been really tested, some are missing. For example, I don't use USB at all so no idea if it works or not. OS itself is very stable. My current commercial STM32L4 project has volume estimates in tens of thousands of devices, so yes this is being used in real projects.


Best Regards,

Juha Niskanen / Haltian Ltd


________________________________
From: ***@yahoogroups.com <***@yahoogroups.com> on behalf of 'Deininger, Andreas' ***@agr.uni-goettingen.de [nuttx] <***@yahoogroups.com>
Sent: Tuesday, February 27, 2018 12:47:49 PM
To: ***@yahoogroups.com
Subject: [nuttx] Newbie questions



Hi,

recently I got aware of NuttX which looks very interesting to me.
Now, we are considering making use of NuttX in a commercial project (up to now, we are using FreeRTOS). In our project, we want to use STM32L4R7 as MCU, we ordered STM32L4R9I-DISCO for evaluation pruposes already.

Now, I have a few questions:

I started building the NuttX simulation on the Linux Subsystem of Windows 10 yesterday and I was able to compile and log into nsh. So far so good. When building Nuttx for STM32, I would probably start using the existing config for stm32l476g-disco, correct? What are the next steps? How much effort is needed to adapt this existing configuration to our board (STM32L4R9I-DISCO)?

In general: how mature is the port for STM32L4? Is it considered experimental, production ready or something in between? I'm asking since our schedule is tight, and we don't have resources and don't want to put a lot of energy in resolving potential issues me might faced with.

I know my question night be vague, however, with help of the list, I would like to get at least a feeling what might be expecting us. In our use case, we see clear advantages to having a POSIX compatible OS (opposed to the thin layer provided by FreeRTOS). Howvever, we don't have any experience with NuttX right now which makes the decision really difficult for us. Maybe someone can shed some light on this.

Best regards
Andreas

--
Envoyé de mon appareil Android avec Courriel K-9 Mail. Veuillez excuser ma brièveté.
Loading...