================
@@ -760,8 +787,10 @@ std::unique_ptr<llvm::Module> Interpreter::GenModule() {
return nullptr;
}
-CodeGenerator *Interpreter::getCodeGen() const {
- FrontendAction *WrappedAct = Act->getWrapped();
+CodeGenerator *Interpreter::getCodeGen(IncrementalAction *Action) const {
+ if (!Action)
+ Action = Act.get();
+ FrontendAction *WrappedAct = Action->getWrapped();
----------------
anutosh491 wrote:
Talking about changes with respect to codegen.
So errors like these
```
./bin/clang-repl --cuda
module flag identifiers must be unique (or of 'require' type)
!"wchar_size"
module flag identifiers must be unique (or of 'require' type)
!"nvvm-reflect-ftz"
module flag identifiers must be unique (or of 'require' type)
!"frame-pointer"
fatal error: error in backend: Broken module found, compilation aborted!
```
or even something like this
```
error: Added modules have incompatible data layouts:
e-p6:32:32-i64:64-i128:128-v16:16-v32:32-n16:32:64 (module) vs
e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128
(jit)
```
Occur because we use the `Act` (the host IncrementalAction) to come up with the
Module every time as of now.
But think about it when we are registering a PTU through registerPTU
```
if (!M)
M = GenModule();
```
And we have this block. We shouldn't be generating the same Module for both the
hostPTU (parsed through the incremental parser) and the devicePTU (parsed
through the incrementalCudaParser) .
For the first case the Module should revolve around the triple
`x86_64-unknown-linux-gnu` and for the device case the module should be framed
around the target triple `nvptx64-nvidia-cuda`.
This is taken care of based on which `IncrementalAction` we use. We should use
Act for hostPTU and DeviceACT for PTU.
Hence the changes here (as one trickles into the other)
```
CodeGenerator *getCodeGen(IncrementalAction *Action = nullptr) const;
std::unique_ptr<llvm::Module> GenModule(IncrementalAction *Action = nullptr);
PartialTranslationUnit &RegisterPTU(TranslationUnitDecl *TU,
std::unique_ptr<llvm::Module> M = {},
IncrementalAction *Action = nullptr);
```
So by default we use Act but we can take of it when required.
1) So while regiserting a devicePTU we use this
https://github.com/llvm/llvm-project/pull/136404/files#diff-b8484f1fc5b057f146ed5d9b6e2cd47c3f6f5ae879c7a0eee44f0a272581a88cR580
2) While registering a hostptu we use this
https://github.com/llvm/llvm-project/pull/136404/files#diff-b8484f1fc5b057f146ed5d9b6e2cd47c3f6f5ae879c7a0eee44f0a272581a88cR600
https://github.com/llvm/llvm-project/pull/136404
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
