This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Intentional or accidental backward-incompatibility w.r.t. process attributes?
- From: Dave Korn <dave dot korn dot cygwin at gmail dot com>
- To: Kai Tietz <ktietz70 at googlemail dot com>
- Cc: NightStrike <nightstrike at gmail dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Fri, 05 Apr 2013 17:06:35 +0100
- Subject: Re: Intentional or accidental backward-incompatibility w.r.t. process attributes?
- References: <515d3064 dot 870ecd0a dot 4726 dot ffff9d4b at mx dot google dot com>
On 04/04/2013 08:48, Kai Tietz wrote:
Hi Kai,
> That change is intentional and would be called __thiscall. To mix
> stdcall and regparam is no more supported AFAIK.
Why are stdcall and regparam not allowed together any more? They seem
entirely orthogonal to me, and the overall result is effectively to place an
arbitrary limit of 1 (the hidden this pointer) on the number of parameters
that can be passed in registers.
If I change __stdcall to __thiscall in the example, I get a clear error message:
> bar.h:3:57: error: regparam and thiscall attributes are not compatible
However in the original case the error message about "ambiguates old
declaration" is unclear: a) it doesn't say what the problem is, b) it goes
away if you repeat the regparm attribute on the function definition and
appears to compile correctly, generating what looks like valid assembly to me.
cheers,
DaveK