Sign in Join

Live chat for Free Pascal, Lazarus, Delphi and every Pascal programmer.

Mailing lists / fpc-devel / 2012-09

fpc on AVR32

9 messages · 5 people · –

Reply by email Subscribe to this list Your subscription

Read-only archive. Use j and k to move between messages.

Florian Klämpfl
to Michel Catudal
Michel Catudal wrote
I would like to port fpc to the AVR32 uc3c devices. I need more details on the functions that need to be implemented.

Take a look at one of the other architectures like arm, fpc/compiler/arm contains the arm specific compiler files.

Reply by email

Jeppe Græsdal Johansen
to Florian Klämpfl

I have made a preliminary backend and RTL stub in branches/laksen/avr32new/

Some of the large problems is that the load instructions allow non-aligned loads in the ld.w forms. This proves to introduce many strange problems, and I don't have any debug equipment.

Regards, Jeppe
________________________________________
Fra: på vegne af Florian Klämpfl
Sendt: 13. september 2012 19:08
Til: FPC developers' list
Emne: Re: [fpc-devel] fpc on AVR32

Michel Catudal wrote
I would like to port fpc to the AVR32 uc3c devices. I need more details on the functions that need to be implemented.

Take a look at one of the other architectures like arm, fpc/compiler/arm contains the arm specific compiler files.

Reply by email

Florian Klämpfl
to Jeppe Græsdal Johansen
Jeppe Græsdal Johansen wrote 7 lines
I have made a preliminary backend and RTL stub in branches/laksen/avr32new/ Some of the large problems is that the load instructions allow non-aligned loads in the ld.w forms. This proves to introduce many strange problems,

Why is this a problem?

Reply by email

Jeppe Græsdal Johansen
to Florian Klämpfl
Florian Klämpfl wrote 9 lines
Am 13.09.2012 21:38, schrieb Jeppe Græsdal Johansen: > I have made a preliminary backend and RTL stub in > branches/laksen/avr32new/ > > Some of the large problems is that the load instructions allow > non-aligned loads in the ld.w forms. > This proves to introduce many > strange problems, Why is this a problem?

Because PC relative loads with the wide format aren't automatically aligned to a 4 byte boundary by the processor, and the instruction set is a variable length 16bit ISA like thumb-2.

There were a few other issues that were highly annoying too, but I can't remember what they were. Other than that the ISA was beautifully simple.

Reply by email

Michel Catudal
to Jeppe Græsdal Johansen: I have made a preliminary backend and RTL stub in branches/laksen/avr32new/ Some of the…
Jeppe Græsdal Johansen wrote 6 lines
I have made a preliminary backend and RTL stub in branches/laksen/avr32new/ Some of the large problems is that the load instructions allow non-aligned loads in the ld.w forms. This proves to introduce many strange problems, and I don't have any debug equipment. Regards, Jeppe ________________________________________

I bought JTAGICE mkII, it works nicely with Scientific Linux 6.2. The binaries from Fedora 12 and 13 work with Scientifc Linux. I will have no problem with debugging AVR32 code. What I find with AVR32 is that it makes nice compact code.

I have written code for a cluster with a Micronas (ARM7TDMI) and AVR32UC3C 144 pins and found a major reduction in code size and code is a lot faster. That is running at the same speed. I run will look at your code, nice to know that I don't have to start from scratch. I don't really like eclipse, do you know if anyone is working on a plugin for codeblocks? I will have to learn the AVR32 assembler. I have coded automotive applications with the 100 pins and 144 pins devices AVR32UC3C but only in C with IAR. I have a very good knowledge of the processors. I wrote a lot of assembler with the AVR8 but never with the AVR32.
At home I use gcc with eclipse since it is the only thing right now.

Just so you know, I started with microprocessors in the early 70s, thus not a newbie with microprocessors.

I will keep you posted of my progress.

Michel

For OS/2 and Linux Software visit http://home.comcast.net/~mcatudal

Reply by email

Martin Schreiber
to Michel Catudal
Michel Catudal wrote 10 lines
I bought JTAGICE mkII, it works nicely with Scientific Linux 6.2. The binaries from Fedora 12 and 13 work with Scientifc Linux. I will have no problem with debugging AVR32 code. What I find with AVR32 is that it makes nice compact code. I have written code for a cluster with a Micronas (ARM7TDMI) and AVR32UC3C 144 pins and found a major reduction in code size and code is a lot faster. That is running at the same speed. I run both at 48Mhz.

Agreed, it is a very good processor.
MSEide can be used to debug AVR32 programs with gdb and the ATMEL gdb server. I use it in combination with the "AVR ONE!" debug interface and the gnu tool chain.
http://sourceforge.net/projects/mseide-msegui/

Martin

Reply by email

Michael Schnell
to Michel Catudal: I bought JTAGICE mkII, it works nicely with Scientific Linux 6.2. The binaries from…
Michel Catudal wrote 5 lines
I have written code for a cluster with a Micronas (ARM7TDMI) and AVR32UC3C 144 pins and found a major reduction in code size and code is a lot faster.

Did you use the Thumb ISA for the ARM ? Same is said to be more compact than ARM32. With a slow memory interface (regarding CPU speed) Thumb might be faster, too.

-Michael

Reply by email

Michel Catudal
to Michael Schnell
Michael Schnell wrote 10 lines
On 09/14/2012 05:23 AM, Michel Catudal wrote: > > > I have written code for a cluster with a Micronas (ARM7TDMI) and AVR32UC3C 144 pins and found a major reduction in code size and code is a lot faster. Did you use the Thumb ISA for the ARM ? Same is said to be more compact than ARM32. With a slow memory interface (regarding CPU speed) Thumb might be faster, too. -Michael _______________________________________________

Yes because code would not fit otherwise and I had the largest flash processor available. The micronas device uses AMD/Spansion flash internally. We were in production for just a few years with Micronas and they decided to discontinue all the ARM devices that we were using. The redesign has cost us a fortune. A few months later at a show in Detroit a Micronas salesman asked me if I was interested in their processors. My response was that their name was a curse where I work.

For OS/2 and Linux Software visit http://home.comcast.net/~mcatudal

Reply by email