Re: [avr-gcc-list] [Bug] Initialization for _delay_loop_2 OptimizedOut

[Dave Hansen]

Sorry Joerg, for the multiple copies: I sent the first directly to you 
accidently.

> As "Dave Hansen" <address@hidden> wrote:
>
> >Notice the brne and the rjmp target the same location.  So it doesn't
> >work for me.  What version are you using?
>
> I've tested it with both of the following, using various -mmcu=avrN and
> -Ox options:
>
> % avr-gcc -v
> Reading specs from /usr/local/lib/gcc-lib/avr/3.3/specs
> Configured with: ./configure --target=avr --prefix=/usr/local 
> i386-portbld-freebsd4.8
> Thread model: single
> gcc version 3.3 20030511 (prerelease)
>
> % avr-gcc -v
> Reading specs from /usr/local/lib/gcc-lib/avr/3.3.1/specs
> Configured with: ./configure --target=avr --prefix=/usr/local 
> i386-portbld-freebsd5.0
> Thread model: single
> gcc version 3.3.1 20030707 (prerelease)

Both newer than mine, I guess:

C:\Dave>avr-gcc -v
Reading specs from C:\WINAVR\BIN\..\lib\gcc-lib\avr\3.3\specs
Configured with: ../configure --prefix=/e/avrdev/install --target=avr 
--enable-l
anguages=c,c++ --disable-nls
Thread model: single
gcc version 3.3 20030421 (prerelease)

Perhaps it's been fixed already?  Just for completeness, my command line 
follows:

avr-gcc -c -mmcu=atmega32 -D OSC_FREQ=12000000 -I. -g -Os -funsigned-char 
-funsigned-bitfields -fpack-struct -fshort-enums -Wall -Wstrict-prototypes 
-Wa,-ahlms=delaytst.lst delaytst.c -o delaytst.o 2> delaytst.err

No, I don't type all that in, it's done in a makefile.  ;-)  I hope I 
transcribed it correctly...

FWIW, my "workaround" didn't fix my problem either.  The problem actually 
showed up in a much more complex routine, and the result of my workaround 
was the same as the library code: the value of an uninitialized register 
(r16) was copied (movw) into r30, and r30 was decremented (sbiw) loop.  I 
did finally get it to work by changing my_delay_loop as shown below:

  static inline void
  my_delay_loop(unsigned int count)
  {
      if (count)
      {
          asm volatile (
              "1: sbiw %0,1" "\n\t"
              "brne 1b"
              : //"=w" (count)
              : "w" (count)
          );
      }
  }

Note that the output list is commented out (so it is now empty), and the 
constraint string on the input list has been changed from "0" to "w".  I 
also changed the name of the parameter, because as you pointed out, it's the 
right thing to do, and this is production code.  Plus I find the leading 
underscores ugly.

The generated code is identical to that you posted earlier (ldi r24,3;  ldi 
r25,0; 1: sbiw r24,1; brne 1b).

Regards,
  -=Dave

_________________________________________________________________
STOP MORE SPAM with the new MSN 8 and get 2 months FREE*  
http://join.msn.com/?page=features/junkmail