diff options
author | Greg Clayton <gclayton@apple.com> | 2016-06-24 23:48:00 +0000 |
---|---|---|
committer | Greg Clayton <gclayton@apple.com> | 2016-06-24 23:48:00 +0000 |
commit | d7f71add86c892ced324fc26cb0945911f3e26a5 (patch) | |
tree | 9ed27eb7686df44798786a1e5b80a0d9ec5c2e7f /lldb/packages/Python/lldbsuite/test/lang/cpp/rvalue-references | |
parent | 4d32eb6939d77f8059683c008bb807167110e930 (diff) | |
download | bcm5719-llvm-d7f71add86c892ced324fc26cb0945911f3e26a5.tar.gz bcm5719-llvm-d7f71add86c892ced324fc26cb0945911f3e26a5.zip |
Made templates that have Enumeration values as arguments work correctly.
We were checking for integer types only before this. So I added the ability for CompilerType objects to check for integer and enum types.
Then I searched for places that were using the CompilerType::IsIntegerType(...) function. Many of these places also wanted to be checking for enumeration types as well, so I have fixed those places. These are in the ABI plug-ins where we are figuring out which arguments would go in where in regisers/stack when making a function call, or determining where the return value would live. The real fix for this is to use clang to compiler a CGFunctionInfo and then modify the code to be able to take the IR and a calling convention and have the backend answer the questions correctly for us so we don't need to create a really bad copy of the ABI in each plug-in, but that is beyond the scope of this bug fix.
Also added a test case to ensure this doesn't regress in the future.
llvm-svn: 273750
Diffstat (limited to 'lldb/packages/Python/lldbsuite/test/lang/cpp/rvalue-references')
0 files changed, 0 insertions, 0 deletions