Well, Przemek noted the warnign about the vector. So maybe HIS makefile was for 
the wrong processor.
Note that you need to define the processor type twice: once when including its 
definies into the code (often for a complete processor group) and once for the 
compiler and linker in the makefile.
Of course 114 is correct for the 5438 where the vector segment starts at 
0xff80. But if compiler/linker are given a processor where the vector segmens 
start at 0xffe0, it gives a warning and won't work.

You shouldn't use classes or library functions inside an ISR. ISRs should be as 
short and quick as possible and accessing classes is usually not. It's better 
to set a flag inside the ISR and handle the job outside.
If you absolutely must, you should implement a simple task scheduler and 
trigger a task switch from inside the ISR to handle the subject. A simple task 
scheduler takes up only a few hundred bytes and the 5438 has 
plenty of them (compared to the 1232 where we implemented one anyways)
That won't solve your actual code problem but maybe when re-considering the 
program flow it may disappear anyway.

JMGross

----- Ursprüngliche Nachricht -----
Von: Piotr Romaniuk
An: [email protected]
Gesendet am: 28 Sep 2009 23:22:33
Betreff: Re: [Mspgcc-users] cannot compile __bic_SR_register_on_exit( ) from 
mspgcc v.3.2.3

I compiled the code for msp430f5438, hence the vector equal to 114 (it is
located at 0xFFF2).
The C program can be compiled when I removed unnecessary "extern C" - thank
you Przemek for suggestion.

I also experimented with C++ but there is a problem with __FUNCTION__ (or
__func__).
In C++ they are not a string literals but "const char *", so it cannot be
concatenated with other
literals. Please try to compile with GCC the .cpp file containing following
code:

int main( int argc, char *argv[] )
{
char *txt = __FUNCTION__ "something";
const char *txt = __FUNCTION__ "something";
}

My intention was to use ISR routine in .cpp file, I need there an access to
classes and need to use _bic_SR.... 
I saw that "extern C" simplifies function names in listing, so the names are
compatible with __FUNCTION__,
but mentioned above problem with concatenation remains.

Piotr Romaniuk 


-----Original Message-----
From: JMGross [mailto:[email protected]] 
Sent: Monday, September 28, 2009 6:13 PM
To: MSPGCC mailing list,
Subject: Re: [Mspgcc-users] cannot compile __bic_SR_register_on_exit( ) from
mspgcc v.3.2.3



The valid vectors are all below 0x10000, so 0x10052 does not point into the
vector table.
This is most likely caused by a wrong start value for the vectors.
Unfortunately, all vectors are defined up from a defined start address
instead of down from 0xfffe. So the start address differs with the number of
vectors. 
And I remember a wrong start address being discovered in one of the header
files several months ago (together with a patch posted here in the list and
merged into CVS).

Anyway, a second look into the source file shows USCI_A0_VECTOR being
defined in the very same file as 114 (which is 0x72). Give the default start
address of 0xffe0 for the vector tables in older devices, this becomes 
0x10052 which is outside the vector segment.
For the newer devices (with more interrupt siources) the start address for
vectors is 0xff80. With this, the given vector offset will result in the
proper address of 0xfff2.
So most likely the makefile is wrong (stating the wrong target device) or
the linker script has a wrong segment definition table.

JMGross


----- Ursprüngliche Nachricht -----
Von: Przemek Klosowski
An: GCC for MSP430 - http://mspgcc.sf.net
Gesendet am: 27 Sep 2009 15:47:13
Betreff: Re: [Mspgcc-users] cannot compile __bic_SR_register_on_exit( ) from
mspgcc v.3.2.3


Per Steve Underwood's MSPGCC manual,  you have to #include <signal.h> in the
interrupt service code modules (1) . Also, extern "C" is not necessary for
compiling C, and I believe is not even alowed in C code: it prevents
mangling of C++ externals so it's only needed for C++ code (i.e.
.cpp/.C/.cxx files). With those two changes, your code compiles. with
warnings about __FUNCTION__ being deprecated, and an invalid Interrupt
vector being assigned to your ISR.

I understand that the current GCC replaces  __FUNCTION__ with __func__
so the iomacros.h file should be patched accordingly. I am not sure about
the ISR vector number: the warning is about vector 0x10052 which isn't even
mentioned in the file. Perhaps someone familiar with writing ISRs for MSPGCC
can explain what's going on.


----------------------------------------------------------------------------
--
Come build with us! The BlackBerry&reg; Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9&#45;12, 2009. Register now&#33;
http://p.sf.net/sfu/devconf
_______________________________________________
Mspgcc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mspgcc-users


------------------------------------------------------------------------------
Come build with us! The BlackBerry&reg; Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9&#45;12, 2009. Register now&#33;
http://p.sf.net/sfu/devconf
_______________________________________________
Mspgcc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mspgcc-users


Reply via email to