Re: [avr-gcc-list] >4.5.1 better than this at register-structure (xmega) access?

[Weddington, Eric]

> -----Original Message-----
> From: Erik Walthinsen [mailto:address@hidden
> Sent: Wednesday, October 10, 2012 9:13 PM
> To: Weddington, Eric
> Cc: avr-gcc-list
> Subject: Re: [avr-gcc-list] >4.5.1 better than this at register-structure
> (xmega) access?
> 
> On 10/10/2012 02:24 PM, Weddington, Eric wrote:
> > But as you said: Does it do that with recent versions? I'm afraid that no
> one yet knows.
> 
> Nope: (stock binutils 2.22, gcc 4.72, avr-libc 1.80)

Well that's good to know! :-)

> 
> void structwrite(uint8_t v) {
>    TCC0.CTRLE = v;
>   a58:   e0 e0           ldi     r30, 0x00       ; 0
>   a5a:   f8 e0           ldi     r31, 0x08       ; 8
>   a5c:   84 83           std     Z+4, r24        ; 0x04
>   a5e:   08 95           ret
> }
> 
> void straightwrite(uint8_t v) {
>    TCC0_CTRLE = v;
>   a60:   80 93 04 08     sts     0x0804, r24
>   a64:   08 95           ret
> }
> 
> #define FS(s,m) (*(typeof(s.m) *)((void *)&s + offsetof(typeof(s),m)))
> void macrowrite(uint8_t v) {
>    FS(TCC0,CTRLE) = v;
>   a66:   80 93 04 08     sts     0x0804, r24
>   a6a:   08 95           ret
> }
> 
> (both -O2 and -O6 produce the same, compiled with -fno-inline)

AFAIK, there's no such thing as -O6. You've got -O0,-O1,-O2,-O3, and -Os.