I recently discovered a nice feature in Eclipse CDT: the ability to show the return value of a function:
By default, the GNU compiler (gcc) optimizes each compilation unit (source file) separately. This is effective, but misses the opportunity to optimize across compilation units. Here is where the Link Time Optimization (LTO, option -flto) can help out: with a global view it can optimize one step further.
The other positive side effect is that the linker can flag possible issues like the one below which are not visible to the compiler alone:
type of '__SP_INIT' does not match original declaration [enabled by default]
Decisions, decisions! Such long weekends like Pentecost are a real challenge for a family with engineers:
- Should we join that record long traffic jam to Italy and be stuck for more than 4 hours and analyze it?
- Or: should we stay home, turn the BBQ smoker engine on fire, load it with baby back pork rib racks for a slow-and-low smoke treatment, while doing some on-the-side IDE and technology exploration?
Well, my family vote was kind of clear: they have chosen that second option. Not to mention that hidden technology piece in it, but that was part of the deal ;-).
And I’m sorry: this article is not about BBQ (for this see “Smoking BBQ Baby Back Ribs – Swiss Style“), it is about technology: I’m using the NXP MCUXpresso IDE and tools for many of my projects (see “Eclipse MCUXpresso IDE 10.1 with integrated MCUXpresso Configuration Tools“). Right before the this extended weekend, NXP has released the new v10.2.0 version, so here is where that technology exploration piece comes into play. Checking the release notes, this version number change includes so many cool stuff I decided to have a look and to check it out. Of course always having an electronic eye on the baby back ribs!
The map file produced by the GNU linker includes lots of information, however it is very cryptic to read. In “Listing Code and Data Size for each Source File with GNU and Eclipse” I showed how the GNU size utility can be used to report the code and data size for each object file. The Eclipse based MCUXpresso IDE comes with another nice view which shows detailed information about code and data allocation:
In “Flash-Resident USB-HID Bootloader with the NXP Kinetis K22 Microcontroller” I presented how I’m using the tinyK22 (or FRDM-K22F) with a flash resident USB HID bootloader. To make sure that the loaded application is not corrupted somehow, it is important to verify it with a Cyclic redundancy Checksum (CRC). The NXP KBOOT Bootloader can verify such a CRC, but how to generate one and how to use it is not really obvious (at least to me), so this article explains how to generate that CRC.
One of the great things with the FreeRTOS operating system is that it comes with free performance analysis: It shows me how much time is spent in each task. Best of all: it shows it in a graphical way inside Eclipse too:
To solve the real hard problem of Embedded Systems development, I usually need all the data I can get from the target. The Percepio Tracealizer is such a tool which can stream application and FreeRTOS trace from the target over a Segger J-Link connection using the Segger RTT protocol. I’m using that combination a lot.
Streaming trace data that way does not need a dedicated hardware like ETM Trace. Using RTT is usually not much intrusive and affects the performance of the target in the 1-2% range (of course depending on the amount of data).
But what worried me for several weeks is that after moving to FreeRTOS V10.0.0 and the same time updating the Segger libraries, the target performance was heavily affected:
Three years ago I published “Debugging Failure: Check List and Hints” and unfortunately this article is one of the most popular ones: obviously debugging problems are very common. Debugging with GDB works usually fine, but if things are failing, then it can be hard to find the cause for it. Recently I have been asked to check some failures, so here are two more hints about what could go wrong…
Binary files are just a binary blob without debug information. Most debug tools and flashers are able to deal (raw) binary (see “S-Record, Intel Hex and Binary Files“). But GDB or the P&E GDB server really needs a ELF/Dwarf file which usually has all the debug information in it. This is a problem if all what I have is a binary file.
This post is about transforming a raw binary (.bin) file into an ELF/Dwarf file with adding a header to it:
The Teensy boards are great, but as they are they are not really useful for real development, as they lack proper SWD debugging. In “Modifying the Teensy 3.5 and 3.6 for ARM SWD Debugging” I have found a way to get SWD debugging working, at that time with Kinetis Design Studio and the Segger J-Link. This article is about how debug the Teensy with free MCUXpresso IDE and the $20 NXP LPC-Link2 debug probe:
“Amazon FreeRTOS – IoT operating system for microcontrollers”: The announcement of FreeRTOS V10.0.0 was one of the biggest news last week for me. Not only is there now a Version 10, the bigger news is that FreeRTOS is now part of Amazon. Wow! Now this explains why Richard Barry (the founder behind FreeRTOS) was kind of hiding away for about a year: he joined Amazon as a principal engineer about a year ago. I think we all have to wait and see what it means for FreeRTOS.
Back in March 2017, NXP had rolled the MCUXpresso IDE starting with Version 10.0.0. With the intent to unify the SDK, LPCXpresso, CodeWarrior, Kinetis Design Studio and Processor Expert into one unified and integrated set of tools. V10.0.0 was a good start. The MCUXpresso IDE V10.0.2 in July was more of a smaller update, and the Pin and Clock configuration tools were not integrated, no added tool for peripheral configuration.
A week ago the MCUXpresso V10.1.0 has been released which shows where the journey is going: an free-of-charge and code size unlimited Eclipse based integrated set of tools to configure, build and debug Cortex-M (Kinetis, LPC and i.MX RT) microcontroller/processor based applications.
I have used it for a week, and although many things are still new, I thought I’m able to give an overview about what is new.
I have used E-Ink displays in projects three years ago, but from that time the technology has greatly evolved. That time displays were hard to get, expensive and difficult to use. Now things seem to change with e-ink displays available to the maker market :-). I’m able to get a 128×296 pixel e-paper display for $10! And for little more money I can have displays with black/white/red colors!
The ARM mbed USB MSD bootloader which is used on many silicon vendor boards has a big problem: it is vulnerable to operating systems like Windows 10 which can brick your board (see “Bricking and Recovering OpenSDA Boards in Windows 8 and 10“). To recover the board, typically a JTAG/SWD programmer has to be used. I have described in articles (see links section) how to recover from that situation, including using an inofficial new bootloader which (mostly) solves the problem. The good news is that ARM (mbed) has released an official and fixed bootloader. The bad news is that this bootloader does not work on every board because of a timing issue: the bootloader mostly enters bootloader mode instated executing the application.
For several projects I’m using library projects: I build a library and then use that library in the other project. If I change something in a library, I want to run make both on the referenced libraries and rebuild my application if needed. If you don’t know how to do this, then read on… 🙂
(… actually it means workign around known Eclipse CDT bug too….)
ARM Cortex-M microcontrollers can have multiple memory controllers. This is a good thing as it allows the hardware to do multiple parallel memory read/writes. However this makes the memory map more complicated for the software: it divides the memory into different regions and memory segments. This article is about how to enable FreeRTOS to use multiple memory blocks for a virtual combined memory heap: