On 06/12/2015 07:16 PM, Sebastian wrote: > On Fri, 12 Jun 2015 09:36:54 +0100 > Andrew Haley <aph@xxxxxxxxxx> wrote: > >> Eh? The vriable access is a memory operation in [2]. It is >> surrounded by the cli/sei, as required. We cannot guarantee that >> non-memory operations will not be reordered with memory barriers, >> though. > You mean the assignment to "ivar" (only that is "surrounded")? Yes. It is the only memory operation. > Sorry that I wasn't clear about that. I meant the assignment to the local variable "val". It's moved across the barrier. No it isn't. > That's the whole point about the example. No it's not. The only thing moved across the barrier is the division. A memory op cannot be moved across a memory barrier. >>> Still the open question: Is access to array elements (or >>> dereferencing a pointer) always considered a "memory operation", >> >> Yes. > > So what's the difference between that variable "val" and my array elements, if I remove the "volatile"? val is not moved across any barrier. Here's the code: 00000112 <test2>: 112: bc 01 movw r22, r24 114: f8 94 cli 116: 8f ef ldi r24, 0xFF ; 255 118: 9f ef ldi r25, 0xFF ; 255 11a: 0e 94 96 00 call 0x12c ; 0x12c <__udivmodhi4> 11e: 70 93 01 02 sts 0x0201, r23 122: 60 93 00 02 sts 0x0200, r22 The store to val is here ^ 126: 78 94 sei 128: 08 95 ret > - Is it that I have array elements / pointer dereferencings? > (I think so) > - Or that it's a global variable? > (Probably not - shouldn't make a difference, at least when global but static) > - Or is it because I'm not concerned about timing, just about correctness? > (Probably not) Andrew.
