Re: [patch]: x64 windows about red-zone-area

[Kai Tietz <ktietz70 at googlemail dot com>]

2010/2/1 Joel Brobecker <brobecker@adacore.com>:
>> Question is, if for x64 windows target the red-zone size shouldn't be set
>> to zero. ?It is possible to use SYSV via attribute switching in x64
>> windows code, but by default x64 windows ABI doesn't have red-zone area at
>> all. So I think it makes sense to set it to zero in amd64-windows-tdep.c
>
> That makes sense, but... ?Just wondering, though - is creating a red zone
> harmful in GDB? As far as I can tell, this information is only used during
> inferior function calls (functions called from GDB), and just causes us
> to waste a few byes. Have you run out of stack space that saving these
> few bytes made a difference?
>
> I'm thinking of the case where a user tries to debug an application
> on x64 windows that does actually use a red zone. ?Wouldn't your change
> break inferior function calls in this case?

Well, I wasn't sure if it would have negative side-effects. This
doesn't seem to be, so maybe this patch isn't necessary. But that a
user tries to debug a function with red-zone, which is actual possible
if function was marked by sysv_abi, has other issues. AFAICS isn't the
current implementation (neither for linux, nor for windows) able to
switch calling convention here. This could get possibly sooner or
later an issue for the Wine project.

Regards,
Kai


-- 
|  (\_/) This is Bunny. Copy and paste
| (='.'=) Bunny into your signature to help
| (")_(") him gain world domination