[avr-gcc-list] BUG: asin (acos) & atan problem (64K boundary)

[Peter Bosscha]

Hello,
 
We're using the latest WinAvr release on Mega128, whenever asin or atan
is placed beyond the 64K boundary the results are (wildly) incorrect.
This affects asin, acos and atan.
 
I've tried to trace this but I'm a little lost, I had something similar
with the entire FPLib 2 years ago and that was fixed with the
PGM_SECTION define.
 
(In FPLIB.inc:)
 
    /* Put constant tables at low addresses in program memory, so they
are
       reachable for "lpm" without using RAMPZ on >64K devices.  */
 
    #define PGM_SECTION .section .progmem.gcc_fplib,"a",@progbits
 
I can see atan2 is wrong in that it has no PGM_SECTION for its table
but the asin does have that correct.
Still, as soon as asin goes across 64K, things go wrong.
 
Anyone any ideas ?
 
Regards,
 
Peter Bosscha
 

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
MailScanner thanks transtec Computers for their support.