[cfe-commits] r166266 - in /cfe/trunk: lib/CodeGen/ABIInfo.h lib/CodeGen/CGCall.cpp lib/CodeGen/TargetInfo.cpp test/CodeGenCXX/regparm.cpp

[Rafael Espíndola]

working on it, thanks!

On 22 October 2012 19:58, Eli Friedman <eli.friedman at gmail.com> wrote:
> On Thu, Oct 18, 2012 at 10:04 PM, Rafael Espindola
> <rafael.espindola at gmail.com> wrote:
>> Author: rafael
>> Date: Fri Oct 19 00:04:37 2012
>> New Revision: 166266
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=166266&view=rev
>> Log:
>> Fix handling of the regparm attribute in the presence of classes with copy
>> constructors.
>>
>> When I first moved regparm support to TargetInfo.cpp I tried to isolate it
>> in classifyArgumentTypeWithReg, but it is actually a lot easier to flip the
>> code around and check for regparm at the end of the decision tree.
>>
>> Without this refactoring classifyArgumentTypeWithReg would have to duplicate
>> the logic about when to use non-byval indirect arguments.
>
> This commit is causing clang to eat memory on the following testcase,
> derived from the gcc testsuite (x86-32):
>
> struct x {
>   struct {
>     struct {} b[0];
>   } a;
> };
> __attribute((regparm(2))) void f(x a) {}
>
> The issue is that X86_32ABIInfo::classifyArgumentType is returning
> ABIArgInfo::getDirectInReg with a struct with zero members, which
> confuses CodeGenModule::ConstructAttributeList.
>
> -Eli