[ECOS] question about stack assign for interrupt service routine.

[ECOS] question about stack assign for interrupt service routine.

[Tony Ko]

hi.
i understand new stack is assigned per thread.
then, what about interrupt?
is new stack  assigned per interrupt?

thanks in advance.

tony.

[ECOS] Question about Hello World eCos example

[Andre Sebastien]

Hi everybody

I build Hello example et run it on an PID board !!!

I use GCC 2.95.3 for arm-lib and eCos

For load the ELF i use a LAUTERBACH and JTAG link !!!
The software is T32

I connect the serial port A on a PC with an HYPER-TERMINAL
configured with 38400 bauds, 8bits,NO PARITY,NO CONTROL

i execute the program i receive on the HYPER-TERMINAL this :
$O48656C6C6F2C2065436F7320776F726C64212035200A#BB

if we interprets it as HEX CODE it be :
          $0
           0x48 0x65 0x6C 0x6C 0x6F 0x2C 0x20  = Hello,  (+ a backspace)

           0x65 0x43 0x6F 0x73 0x20 = eCos     (+ a backspace)
           0x77 0x6F  0x72 0x6C 0x64 0x21 0x20 = world!  (+ a backspace)

           0x35 0x20 0x0A  = 5 \n
           #BB

It s exactly that i try to print : printf("Hello, eCos world! %d \n",5);

PS: what is $0 and #BB at the beginning and the end of output ?

But never the program go out the function "printf", it seems to go in a
infinite loop !!!!!
What does it wrong ?

best regards.

Seb.

Re: [ECOS] Question about Hello World eCos example

[Robin Farine]

Andre Sebastien <sebastien.andre@sxb.bsf.alcatel.fr> writes:

[...]

> It s exactly that i try to print : printf("Hello, eCos world! %d \n",5);
> 
> PS: what is $0 and #BB at the beginning and the end of output ?

Not $0, but $O (O packet). The GDB protocol uses such packets to output text
non-related to GDB itself (e.g. your printf). The protocol also appends a
checksum to every packet, the #BB in your case.

> But never the program go out the function "printf", it seems to go in a
> infinite loop !!!!!  What does it wrong ?

It's waiting for an ack from the remote side of a GDB connection !!!!!

Try to set the option below in your ecos.ecc:

    cdl_component CYGDBG_HAL_DIAG_TO_DEBUG_CHAN {
        user_value 0
    };

[...]

Robin

Re: [ECOS] Question about Hello World eCos example

[Andre Sebastien]

Thanks a lot Robin Farin and Rob Jansen !!

So now it s work nice !!! but Robin Farin what is exactly this component
CYDBG_HAL_DIAG_TO_DEBUG_CHAN ???
it seems to disable the GDB PROTOCOLS for the SERIAL ports ?


How can i know the behaviour of these components ?

thx.


Sebastien.


Robin Farine wrote:

> Andre Sebastien <sebastien.andre@sxb.bsf.alcatel.fr> writes:
>
> [...]
>
> > It s exactly that i try to print : printf("Hello, eCos world! %d \n",5);
> >
> > PS: what is $0 and #BB at the beginning and the end of output ?
>
> Not $0, but $O (O packet). The GDB protocol uses such packets to output text
> non-related to GDB itself (e.g. your printf). The protocol also appends a
> checksum to every packet, the #BB in your case.
>
> > But never the program go out the function "printf", it seems to go in a
> > infinite loop !!!!!  What does it wrong ?
>
> It's waiting for an ack from the remote side of a GDB connection !!!!!
>
> Try to set the option below in your ecos.ecc:
>
>     cdl_component CYGDBG_HAL_DIAG_TO_DEBUG_CHAN {
>         user_value 0
>     };
>
> [...]
>
> Robin

Re: [ECOS] Question about Hello World eCos example

[Robin Farine]

Andre Sebastien <sebastien.andre@sxb.bsf.alcatel.fr> writes:

[...]

> How can i know the behaviour of these components ?

Start from your generated ecos.ecc, search for keywords, read the comments and
when you find something interesting, search the eCos CDL files in the "packages"
directory until you find the CDL file that define the interesting
component/option. Then look at the header & source files of the corresponding
package to see how the component/option parameterizes the package.

[...]

Robin

Re: [ECOS] Question about Hello World eCos example

[David Airlie]

GDB debugging ... there is an option for mangling output in the .ecc, set
this to none..

Dave.

On Wed, 25 Jul 2001, Andre Sebastien wrote:

> Hi everybody
> 
> I build Hello example et run it on an PID board !!!
> 
> I use GCC 2.95.3 for arm-lib and eCos
> 
> For load the ELF i use a LAUTERBACH and JTAG link !!!
> The software is T32
> 
> I connect the serial port A on a PC with an HYPER-TERMINAL
> configured with 38400 bauds, 8bits,NO PARITY,NO CONTROL
> 
> i execute the program i receive on the HYPER-TERMINAL this :
> $O48656C6C6F2C2065436F7320776F726C64212035200A#BB
> 
> if we interprets it as HEX CODE it be :
>           $0
>            0x48 0x65 0x6C 0x6C 0x6F 0x2C 0x20  = Hello,  (+ a backspace)
> 
>            0x65 0x43 0x6F 0x73 0x20 = eCos     (+ a backspace)
>            0x77 0x6F  0x72 0x6C 0x64 0x21 0x20 = world!  (+ a backspace)
> 
>            0x35 0x20 0x0A  = 5 \n
>            #BB
> 
> It s exactly that i try to print : printf("Hello, eCos world! %d \n",5);
> 
> PS: what is $0 and #BB at the beginning and the end of output ?
> 
> But never the program go out the function "printf", it seems to go in a
> infinite loop !!!!!
> What does it wrong ?
> 
> best regards.
> 
> Seb.
> 

-- 
      David Airlie, Software Engineer, Parthus Technologies plc.,
       Mary Rosse Centre, National Tech Park, Limerick, Ireland.
   t: +353-61-508116 / f: +353-61-508101 / David.Airlie@parthus.com

Re: [ECOS] question about stack assign for interrupt service routine.

[Jonathan Larmour]

Tony Ko wrote:
> 
> hi.
> i understand new stack is assigned per thread.
> then, what about interrupt?
> is new stack  assigned per interrupt?

On most targets there is a separate interrupt stack. But it is used for all
interrupts.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
Come to the Red Hat TechWorld open source conference in Brussels!
Keynotes, techie talks and exhibitions    http://www.redhat-techworld.com/

[ECOS] question about GCC for Arm Installation

[Grant Whitacre]

I recently downloaded and built the dev tools for Arm development -- GCC
(2.95.2), binary utilities (2.10.1) and insight (5.0).  Everything seems to
get built just fine but when I try to do a minimum eCos build with default
settings using the Configuration tool (1.3.1), I get error messages:

make[1]: Entering directory
`/c/Coldfire/eCos/MyBuilds/MinimalForArm3_build/hal/arm/arch/v1_3_1'
arm-elf-gcc -mcpu=arm7tdmi -Wall -Wpointer-arith -Wstrict-prototypes -Winlin
e -Wundef -Woverloaded-virtual -g -O2 -ffunction-sections -fdata-sections -f
no-rtti -fno-exceptions -fvtable-gc -finit-priority  -I//C/Coldfire/eCos/MyB
uilds/MinimalForArm3_install/include -I//c/coldfire/ecos/packages/hal/arm/ar
ch/v1_3_1 -I//c/coldfire/ecos/packages/hal/arm/arch/v1_3_1/src -I//c/coldfir
e/ecos/packages/hal/arm/arch/v1_3_1/tests -I. -Wp,-MD,arm.tmp -o
hal_mk_defs.tmp -S
//c/coldfire/ecos/packages/hal/arm/arch/v1_3_1/src/hal_mk_defs.c
cc1.exe: Invalid option `-Woverloaded-virtual'
cc1.exe: Invalid option `-fno-rtti'
cc1.exe: Invalid option `-fvtable-gc'
cc1.exe: Invalid option `-finit-priority'
//c/coldfire/ecos/packages/hal/arm/arch/v1_3_1/src/hal_mk_defs.c: In
function `main':
//c/coldfire/ecos/packages/hal/arm/arch/v1_3_1/src/hal_mk_defs.c:105:
warning: control reaches end of non-void function
make[1]: *** [arm.inc] Error 1
make[1]: Leaving directory
`/c/Coldfire/eCos/MyBuilds/MinimalForArm3_build/hal/arm/arch/v1_3_1'
make: *** [build] Error 2


I've followed the installation procedure exactly as described on the
"Building the ARM development tools for Windows" site.  I've also tried
upgrading to the latest versions of the tools (GCC-v3.0 and
binutil-v2.11.2), but I get the same results.  Does anyone have any ideas?
Do I need anything else - like newlib or gcc-g++?  (I tried to download the
latter but this folder on all the mirror sites was unavailable).

Thanks,
Grant

Re: [ECOS] question about GCC for Arm Installation

[Jonathan Larmour]

Grant Whitacre wrote:
> 
> I recently downloaded and built the dev tools for Arm development -- GCC
> (2.95.2), binary utilities (2.10.1) and insight (5.0).  Everything seems [snip]
> //c/coldfire/ecos/packages/hal/arm/arch/v1_3_1/src/hal_mk_defs.c
> cc1.exe: Invalid option `-Woverloaded-virtual'
[snip]
> 
> I've followed the installation procedure exactly as described on the
> "Building the ARM development tools for Windows" site.  I've also tried
> upgrading to the latest versions of the tools (GCC-v3.0 and
> binutil-v2.11.2), but I get the same results.

Those versions of tools should definitely not complain about those options.
It seems most likely you are somehow picking up other tools. Are you sure
that arm-elf-gcc is pointing to the right tools? Use "type -p arm-elf-gcc"
to check. Also make sure GCC_EXEC_PREFIX is unset in your environment.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
Come to the Red Hat TechWorld open source conference in Brussels!
Keynotes, techie talks and exhibitions    http://www.redhat-techworld.com/

Re: [ECOS] Question about Hello World eCos example

[rob.wj.jansen]

Andre,

> i execute the program i receive on the HYPER-TERMINAL this :
> $O48656C6C6F2C2065436F7320776F726C64212035200A#BB

This is clearly GDB output.
During build it is decided that your output will always go via your GDB channel.
The reason why printf does not return is because you need to acknowledge this
packet (with a "+").

Same thing happened on my target. Try to remove the GDB stub and rebuild eCos.
If this fails, then search for the point where it is decided that your diagnostics output
has to be via GDB packets. I forgot where this was defined but I somewhere found
that GDB packetized output is done if you have either GDB stubs OR if you use
the RAM memory map (they forgot about JTAG debugging pods).

The diagnostics output is done from packages/hal/arm/pid/current/src/hal_diag.c.
Just place a breakpoint on the hal_diag_write_char() function and you will see.

Regards,

     Rob Jansen

Software Engineer
Competence Center Platforms
BU Mobile Communications
Meijhorst 60-10, 6537 KT Nijmegen, The Netherlands
Tel: +31-24-353-6329
Fax: +31-24-353-3613
mailto:Rob.WJ.Jansen@philips.com