[PATCH] D28058: [OpenCL] Correct ndrange_t implementation

[Yaxun Liu via Phabricator via cfe-commits]

yaxunl added a comment.

I am thinking whether we should restrict ndrange_t to be struct type only, otherwise a user could typedef anything as ndrange_t and we may lose the ndrange_t type in the IR, which will cause issue if we want to translate the IR to SPIRV.


https://reviews.llvm.org/D28058