This is the mail archive of the newlib@sourceware.org mailing list for the newlib project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Problem with ARM port and pow() function


Alle 17:57, martedì 23 ottobre 2007, Nick Clifton ha scritto:

> Are you still compiling for thumb ?  If so then the problem might be 
> connected to how the first newlib math function calls the second newlib 
> function (or how the second one returns).  Could you try compiling (and 
> linking) in arm mode, to see if this fixes the problem ?

I've tried thumb, arm and interwork modes.
Always the same result: crash.
But this is a good point: maybe the linker uses the incorrect library version 
(but I checked the linker reports, and it refers to the correct one) or that 
the library is not compiled in the correct mode...

Now I'm going to recompile newlib with -O0 option, and I want also to try 
adding in makefiles CFLAGS_TO_TARGET the --mcpu=arm7tdmi-s -gdwarf-2 options.

Let's see the results...
Maybe it would be faster to write my own math library!!!!!!

;-)


> The other possibility is stack corruption.  If the return address (from 
> the second function to the first function) is held on the stack, and 
> some bug in the second function corrupts the stack, then this can cause 
> all kinds of mischief.  To look for this, step into the second function 
> (in the debugger) and then look on the stack for the return address. 

I will do that as soon as I have compiled the newlib with -O0 optimization.

Thank you all for your suggestions.
I can't believe to be the only one who really tries to use GNU ARM Toolchain 
floating point math functions with ARM7...

Anyone using a real hardware (not simulator) has experienced them?


Fausto


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]