On 9.3.18 20:16, Richard Damon wrote:
> On 3/9/18 8:56 AM, Tauno Voipio wrote:
>> On 9.3.18 11:30, The Natural Philosopher wrote:
>>> On 09/03/18 09:17, Richard Kettlewell wrote:
>>>> The Natural Philosopher  writes:
>>>>> On 08/03/18 17:33, mm0fmf wrote:
>>>>>> On 08/03/2018 15:27, Martin Gregorie wrote:
>>>>>>> On Thu, 08 Mar 2018 13:34:57 +0000, The Natural Philosopher wrote:
>>>>>>>> I would at least put in a bit more
>>>>>>>>
>>>>>>>> int myfunction(int bufflth, char* buff)
>>>>>>>>       {
>>>>>>>>          return (int)buff[bufflth-1];
>>>>>>>>       }
>>>>>>>>
>>>>>>>> So you know where on the stack buff and bufflth are...
>>>>>>>
>>>>>>> Why would you want to know that? All the assembler operations
>>>>>>> using the
>>>>>>> arguments will be inside myfunction() and, unless the code is
>>>>>>> calculating
>>>>>>> a result to be returned and used elsewhere, the returned value is
>>>>>>> probably to report success or failure.
>>>>>>>
>>>>>>
>>>>>> If it returns success or failure, why return 1 of 4 billion possible
>>>>>> values?
>>>>>
>>>>> Another personm who has never coded assmebler for C..
>>>>>
>>>>> It not what you are rurining, its how to access (bufflth, char* buff)
>>>>> which are LOCAL stack based variables.
>>>>
>>>> The first few args are passed in registers on many modern systems. See
>>>> e.g.
>>>>
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0042f/IHI0042F_aapcs.pdf
>>>>
>>>> for instance.
>>>>
>>>
>>> Even so, the method still reveals WHICH registers
>>>
>>> But it is not a feature of the ARM per se, merely a convention.
>>>
>>> I assume gcc follows it, but there is no gurantee
>>
>> GCC follows the ARM EABI specification:
>>
>>   - stack on 8 byte boundaries,
>>   - arguments in r0 - r3
>>   - return value in r0
>>
>> I still suspect that the OP does not have a clue of the assembly
>> code he's attempting to link in.
>>
>
> The EABI goes a bit farther, as there are some register the called
> function must preserve (if it changes them, it must save them first) and
> other register the caller can freely change, and if the callee needs
> them preserved must save them before the call. This choice is actually
> enshrined in the interrupt processing (at least on some ARMs, I would
> have to check if true for all the variants), as an interrupt will save
> all the registers that a normal function is allowed to change without
> saving, thus making ISR routines normal void fun(void) functions.

If the ARM is a Cortex, the hardware takes care of preserving the
critical registers, so the exception service can usually be
written as a normal C function.

--

-TV

--- SoupGate-Win32 v1.05