GDC and leaf functions

Re: GDC and leaf functions

Mar 04, 2013

Iain Buclaw

Mar 04, 2013

David Nadlinger

Mar 04, 2013

Manu

Mar 04, 2013

Iain Buclaw

On Mar 4, 2013 3:27 PM, "Manu" <turkeyman@gmail.com> wrote: > > I'm doing a lot of experiments with GDC, and I'm noticing that leaf functions are still generating a stack frame and doing pointless work... > > int leaf() pure nothrow > { > return 0; > } > > compile with -frelease -finline -O3, etc > > I expect 2 opcodes, load immediate and return, but I get a whole pointless function including memory access O_O > > Any ideas? Is this normal? I'd expect so. Leaf functions in C require explicit attributes to mark them. Now that we've integrated UDA attributes into gcc attributes, we need to start building on what is needed to be implemented. :) import gcc.attributes; @attribute ("leaf") int leaf; Regards -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';

On Monday, 4 March 2013 at 17:35:45 UTC, Iain Buclaw wrote: > On Mar 4, 2013 3:27 PM, "Manu" <turkeyman@gmail.com> wrote: >> >> I'm doing a lot of experiments with GDC, and I'm noticing that leaf > functions are still generating a stack frame and doing pointless work... >> >> int leaf() pure nothrow >> { >> return 0; >> } >> >> compile with -frelease -finline -O3, etc >> >> I expect 2 opcodes, load immediate and return, but I get a whole > pointless function including memory access O_O >> >> Any ideas? Is this normal? > > I'd expect so. ?! I think you are seriously underestimating GCC here... ;) David

March 04, 2013

Re: GDC and leaf functions

Posted by Manu
in reply to David Nadlinger

Permalink

Manu

Posted in reply to David Nadlinger

Attachments:

text/html part

Permalink

Yeah, I've certainly never had to mark leaf's before...
Turns out it was my bad, wrong options, although SH4 is broken, which is
what started me scratching my head in the first place.


On 5 March 2013 03:40, David Nadlinger <see@klickverbot.at> wrote:

> On Monday, 4 March 2013 at 17:35:45 UTC, Iain Buclaw wrote:
>
>> On Mar 4, 2013 3:27 PM, "Manu" <turkeyman@gmail.com> wrote:
>>
>>>
>>> I'm doing a lot of experiments with GDC, and I'm noticing that leaf
>>>
>> functions are still generating a stack frame and doing pointless work...
>>
>>>
>>> int leaf() pure nothrow
>>> {
>>>   return 0;
>>> }
>>>
>>> compile with -frelease -finline -O3, etc
>>>
>>> I expect 2 opcodes, load immediate and return, but I get a whole
>>>
>> pointless function including memory access O_O
>>
>>>
>>> Any ideas? Is this normal?
>>>
>>
>> I'd expect so.
>>
>
> ?!
>
> I think you are seriously underestimating GCC here... ;)
>
> David
>

On Mar 4, 2013 5:46 PM, "David Nadlinger" <see@klickverbot.at> wrote: > > On Monday, 4 March 2013 at 17:35:45 UTC, Iain Buclaw wrote: >> >> On Mar 4, 2013 3:27 PM, "Manu" <turkeyman@gmail.com> wrote: >>> >>> >>> I'm doing a lot of experiments with GDC, and I'm noticing that leaf >> >> functions are still generating a stack frame and doing pointless work... >>> >>> >>> int leaf() pure nothrow >>> { >>> return 0; >>> } >>> >>> compile with -frelease -finline -O3, etc >>> >>> I expect 2 opcodes, load immediate and return, but I get a whole >> >> pointless function including memory access O_O >>> >>> >>> Any ideas? Is this normal? >> >> >> I'd expect so. > > > ?! > > I think you are seriously underestimating GCC here... ;) > I usually do. :p Regards -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';

Forums