I'm just starting on the development on the arm7 processor (SAMSUNG
S3C44BOX). I want to use linux as my devel platform (my laptop has
ubuntu installed on it). I've few questions:
1) is the arm project manager (APM) available for linux (I googled for
it but coudnt find any useful info, I think its not available
2) I've installed the arm toolchain (arm-elf-gcc and others version
2.95.3), my question is can we only use this toolchain to develop
applications. I heard that in the APM we can use, for example, for GPP
port D rPCOND, can we use the same variables as in the APM.
Please let me know where I should start. There is no OS installed on the
arm board. I would like to write some programs in C (on my laptop,
compile it ) and then download them onto the board.
I am currently developing on Ubuntu for ARM7 (LPC2124) core and i am
certainly finding it easier than under Windows.
1. Have never used it, but doubt it.
I have been mainly using gedit text editor for source file. I build
using terminal using a well laid out and commented makefile. I have
found this setup prevents confusion with no unknown/hidden build options
As far as IDEs go I have tried to use Eclipse several time for C or ARM
development but it is just too clunky, slow and inflexible imho.
Recently i found a new IDE called Codelite, that i am currently testing,
it has a few rough edges but has a promising outlook and is really
flexible and quick.
2. Have you really installed arm-elf-gcc v2.95.3?
I have read many times that gcc versions of at least v4 are recommended.
My toolchain consists of arm-elf-gcc 4.3.4.
I do not know your level of experience but you do not need an OS as such
to work on the board, look for some demo apps that flash LEDs etc.
There are a lot of RTOSs if that is what you mean and one i would
recommend and currently implementing is ChibiOS/RT. A caveat though, get
the basics of ARM7 sorted before attempting any RTOS.
Cas, Thanks for your reply.
I have installed the arm 4.3.3 earlier (using the build details here-
http://www.gnuarm.com/). I was having some trouble to compile the
uClinux kernel(2.4 I obtained along with the distribution cd (SAMSUNG
Also I'm required to write device drivers (for a course at school), so I
had to compile the kernel. It was giving me error "arm/clone.S:34:Error:
undefined symbol 'EINVAL' in operation". So I had to use the arm v2.95.3
tool chain which has been tested. Can you please tell me if arm v4.3.3
can be used to compile uClinux 2.4.
Sorry I forgot to ask this in the previous post... Can you please point
me to some sample codes where we use the registers in the arm processor
(like the GPP control reg's ), or suggest some books which help us
understand the intricacies of the arm processor.
I have no experience of Samsung ARM7 cores nor using uClinux so the best
i can suggest is to search for documents and technical manual for the
S3C44B0 processor, manufacturer docs should detail the register
addresses and functions.
a useful link for porting:
> It was giving me error "arm/clone.S:34:Error:
> undefined symbol 'EINVAL' in operation". So I had to use the arm v2.95.3
> tool chain which has been tested.
That was probably a large backward step for what is no doubt a very
simple problem. You should apply the same debugging approach that you
would if an error appeared in your own code. This is not a toolchain
On the face of it line 34 of the assembler file clone.S references a
symbol EINVAL without a definition. The obvious solution is to provide
EINVAL is an errno.h macro for "invalid argument". By convention the
GNU toolchain treats a file with externsion .S (capital S) as an
assembler file that is first passed through the C pro-processor. This
means that to resolve this problem, you should simply #include
It may work with 2.95 because of some other header indirectly including
errno.h, which 4.x does not.
Clifford, Thanks for the suggestion. I did not pay attention to the
I tried to recompile the uClinux (2.4) again using the 4.3.3 tool chain.
Now, I used the <linux/errno.h> header in the clone.S file, and I passed
through that stage with no errors I had to edit quite a few file for
such kind of errors.
I'm now struck at this point where it shows me the following error:
make: Entering directory `/home/bhargava/Desktop/uClinux-dist/linux-2.4.x'
unset GCC_EXEC_PREFIX; gcc -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -o scripts/split-include scripts/split-include.c
scripts/split-include.c: In function ‘main’:
scripts/split-include.c:133: warning: ignoring return value of ‘fgets’, declared with attribute warn_unused_result
scripts/split-include include/linux/autoconf.h include/config
arm-elf-gcc -D__KERNEL__ -I/home/bhargava/Desktop/uClinux-dist/linux-2.4.x/include -gstabs+ -Wall -Wstrict-prototypes -Wno-trigraphs -O1 -g -fno-strict-aliasing -fno-common -pipe -fno-builtin -D__linux__ -DNO_MM -mapcs-32 -mtune=arm7tdmi -mshort-load-bytes -msoft-float -gstabs+ -O2 -c -o init/main.o init/main.c
cc1: error: unrecognized command line option "-mapcs-32"
cc1: error: unrecognized command line option "-mshort-load-bytes"
make: *** [init/main.o] Error 1
make: Leaving directory `/home/bhargava/Desktop/uClinux-dist/linux-2.4.x'
make: *** [linux] Error 1
I googled for answers but with little luck. I think -mapcs-32 and
-mshort-load-bytes command line options are obsolete now.This is the
first major piece of software that I'm building (make). So I've no clue
how the options have to be configured.
Is there any way to remove these options and proceed through the make
Bhargava Yammanuru wrote:
> cc1: error: unrecognized command line option "-mapcs-32"
> cc1: error: unrecognized command line option "-mshort-load-bytes"
You should check the manuals for the two different versions you used to
see what these options did, and if they are supported in the new
version. Manuals for all versions of GCC since 2.95 are at
-mapcs-32 forces use of a 32bit program counter. Old obsolete (i.e. no
longer manufactured) ARM parts has a 26 bit program counter. GCC
nolonger supports these obsolete parts, so a 32bit program counter is
always used. You can safely remove this option.
Similarly -mshort-load-bytes is obsolete. I believe also that it is
only relevant to older ARM architectures.
Of course there is never anything to prevent you from simply removing
options that the compiler rejects and trying it. It won't explode!
Clifford Slocombe wrote:
> Manuals for all versions of GCC since 2.95 are at
> Of course there is never anything to prevent you from simply removing
> options that the compiler rejects and trying it. It won't explode!
Of course it won't explode, but how do I remove these options. Which
file (or what are the most probable files) has to be edited to remove
these command line options.
make: Entering directory `/home/bhargava/Desktop/uClinux-dist/linux-2.4.x'
arm-elf-gcc -D__KERNEL__ -I/home/bhargava/Desktop/uClinux-dist/linux-\ 2.4.x/include -gstabs+ -Wall -Wstrict-prototypes -Wno-trigraphs \
-O1 -g -fno-strict-aliasing -fno-common -pipe -fno-builtin -D__linux__ \
-DNO_MM -mapcs-32 -mtune=arm7tdmi -mshort-load-bytes \
-msoft-float -gstabs+ -O2 -c -o init/main.o init/main.c
I tried greping in the uClinux-dist/linux-2.4.x directory, but I did not
find any references (yet), but because of deadlines at school I had to
stop greping process.
If you can tell me the files to look for, it will be immense help to me.
> If you can tell me the files to look for, it will be immense help to me.
Normally gcc options are defined within the makefile but uClinux may be
I was just re-reading your posts and I have to ask, does your school
project require you use uClinux? It seems, from my perspective, a very
complicated project to begin embedded development with.
The reason I mention this, is that I am working on a project dealing
with small RTOS's and they are complex on an ARM processor. I have spent
a lot of time learning, with non-OS code, about the ARM7 architecture,
much more than actually implementing any RTOS theory.
Clifford do you have any thoughts on this?
Calum Lind wrote:
> Normally gcc options are defined within the makefile but uClinux may be
Thanks, Calum (cas). I'll look at the Makefile and post updates here.
> I was just re-reading your posts and I have to ask, does your school
> project require you use uClinux?
yes, we need to write few drivers and other things, a huge list to
things to do :(
Do you use minicom to transfer executables to your arm board. I was
able to transfer the kernel image to the board using the Ethernet.
As suggested by you I want to learn working on the ARM board without the
ROTS whenever I've time. So I started with the ubiquitous hello world.
I'm having trouble to transfer the hello world executable to the arm
board using minicom. The problem is all the people (who are working on
the ARM7 board) I know are working on windows (use hyper-terminal, APM).
So I've to rely on forums like these. So I'm sorry if these questions
I set minicom and then used loadb
Then the following is displayed
## Ready for binary (kermit) download ...
then I used ctrl+A S and selected kermit and then selected the file to
be transferred, nothing happens after that. It just shows the above
What am I doing wrong? Any ideas ???
Im afraid this is not the same development route as myself, as it
suggests uboot is installed, so I am not much help. However minicom
should work fine as a Hyperterminal replacement so the issue might be
the kernel you loaded.
The basics I talked about are bare-metal development where I use startup
code, linker script, source code all compiled into a binary which I
flash to my target using JTAG or Serial(ISP).
Simplest guide I could find is here:
While looking for something else i came across this page that could be
of some help to you:
Thanks for the link Cas, I was able to complete till this part I was
having question about transferring the binary onto the board and making
the arm cpu execute the code.
I am new to this post. I am interested also in programming a s3c44box
chip of samsung (contains an arm7tdmi) from a linux host. I have an
Olimex ARM-USB-TINY connector which I plan to connect to my development
board, which is a S3CEV40 (supplied by embest).
My question is, can you help me in setting up a gdb/openocd to programm
and debug code on chip? Can you point me out on a starting point for it?
Thanks in advance.