[avr-libc-dev] [bug #41435] math library functions in uril/delay.h

[Jaakko Kantojärvi]

URL:
<http://savannah.nongnu.org/bugs/?41435>

Summary: math library functions in uril/delay.h
Project: AVR C Runtime Library
Submitted by: raphendyr
Submitted on: Sun 02 Feb 2014 04:13:34 PM GMT
Category: Header
Severity: 3 - Normal
Priority: 5 - Normal
Item Group: Header files
Status: None
Percent Complete: 0%
Assigned to: None
Originator Email:
Open/Closed: Open
Discussion Lock: Any
Release: Unknown
Fixed Release: None

_______________________________________________________

Details:

New version of _delay_ms (the one using __builtin_avr_delay_cycles) uses math
functions fabs and ceil.

Code expects that when run on hosted and optimized compilation these calls are
optimized away. As this is impossible in freestanding environment there is
macro test (__STDC_HOSTED__) and code fallback to old code.

This is kind of ok, but there is possibility to make this otherwise.

Following code works in freestanding environment:
```
static void _delay_ms(double ms) __attribute__ ((unused));
static void _delay_ms(double ms) {
extern void __builtin_avr_delay_cycles(unsigned long);
extern double __builtin_fabs(double);
extern double __builtin_ceil(double);

double tmp = ((F_CPU) / 1e3) * ms;
uint32_t ticks_dc = (uint32_t)(__builtin_ceil(__builtin_fabs(tmp)));

__builtin_avr_delay_cycles(ticks_dc);
}
```

As we can see from above, we could use __builtin_fabs and __builtin_ceil in
delay.h instead of expecting the compiler to use them.

I think this is better as we already use __buildin_avr_delay_cycles which
requires build time constant argument, thus we can use __builtin_fabs and
__builtin_ceil as we know their arguments are built time constant.




_______________________________________________________

Reply to this item at:

<http://savannah.nongnu.org/bugs/?41435>

_______________________________________________
Message sent via/by Savannah
http://savannah.nongnu.org/

[Cameron Tacklind]

Follow-up Comment #1, bug #41435 (project avr-libc):

This caused a bunch of confusion for me until I realized that ffreestanding
was causing _delay_ms to fail to compile for a ATmega32u4 device.

For reference, here are the errors I saw. It took a while to figure out that
_STDC_HOSTED_, set by ffreestanding, was making _delay_ms use old incompatible
assembly.


{standard input}: Assembler messages:
{standard input}:19: Error: constant value required
{standard input}:19: Error: register number above 15 required


The above was gcc error output from compiling this simple file:


#include <util/delay.h>
void foobar() {
_delay_ms(1);
}


For reference, this is the output if I tell g++ to just compile to assembly:


.file "test.cpp"
__SP_H__ = 0x3e
__SP_L__ = 0x3d
__SREG__ = 0x3f
__tmp_reg__ = 0
__zero_reg__ = 1
.text
.global _Z6foobarv
.type _Z6foobarv, @function
_Z6foobarv:
/* prologue: function */
/* frame size = 0 */
/* stack size = 0 */
.L__stack_usage = 0
ldi r24,lo8(-96)
ldi r25,lo8(15)
/* #APP */
; 110 "c:\program files
(x86)\avr8-gnu-toolchain\avr\include\util\delay_basic.h" 1
1: subi r24A,1
brne 1b
; 0 "" 2
/* #NOAPP */
ret
.size _Z6foobarv, .-_Z6foobarv
.ident "GCC: (AVR_8_bit_GNU_Toolchain_3.5.4_1709) 4.9.2"


_______________________________________________________

Reply to this item at:

<http://savannah.nongnu.org/bugs/?41435>

_______________________________________________
Message sent via/by Savannah
http://savannah.nongnu.org/

[Georg-Johann Lay]

Post by Cameron Tacklind
/* #APP */
; 110 "c:\program files

(x86)\avr8-gnu-toolchain\avr\include\util\delay_basic.h" 1

Post by Cameron Tacklind
1: subi r24A,1
brne 1b
; 0 "" 2
/* #NOAPP */

By no means I can see how __STC_HOSTED__ could get you in "A" after "r24".
The revision logs of delay_basic.h and delay.h[.in] don't mention such a bug,
and my builds of 4.9.2 from around that time also look sane. Also I never
came across a GCC bug which added additional "A" junk to operands.

The only issue with this is vor AVR_TINY where register class "w" is empty.

_______________________________________________________

Reply to this item at:

<http://savannah.nongnu.org/bugs/?41435>

_______________________________________________
Message sent via/by Savannah
http://savannah.nongnu.org/

[Cameron Tacklind]

Follow-up Comment #3, bug #41435 (project avr-libc):

My mistake. In trying to fix #36611 <https://savannah.nongnu.org/bugs/?36611>
for myself, I accidently screwed with my *delay_basic.h* and forgot I had done
so. That had the erroneous "A" that was causing my problems.

_______________________________________________________

Reply to this item at:

<http://savannah.nongnu.org/bugs/?41435>

_______________________________________________
Message sent via/by Savannah
http://savannah.nongnu.org/