Jakub Chlanda

Results 10 issues of Jakub Chlanda

This PR brings `LEG` to 15 Oct 2017: [a5204d048729](https://github.com/frasercrmck/llvm-leg/commit/a5204d048729eef09b162dd12622c4afefb4978a) of llvm master. One thing to notice is that I discarded the original commits, made by @frasercrmck and @pasaulais and only...

This PR brings llvm-leg to release_39 of llvm (https://github.com/llvm-mirror/llvm/tree/release_39/).

It would appear that multiple namespaces in a single file are not handled correctly. Entries in a files dictionary are overriden. To recreate this issue: ``` c++ #ifndef __TEST_H #define...

GPU backends are not geared up to use stack protectors. Using the flag will result in miscompilation. This has been discovered in https://github.com/intel/llvm/issues/7269.

platform:gpu-amd

Please see https://www.khronos.org/registry/OpenCL/extensions/intel/cl_intel_subgroups.html for the details of the shuffles. This was uncovered when writing libclc's Intel subgroup shuffles, which use the same built-in `bpermute` (https://github.com/intel/llvm/pull/4664/files) and was failing tests from...

Customs dirs path was [asserting on the existence of `llvm-spirv` and `libclc-remangler`](https://github.com/jchlanda/llvm/blob/jakub/libclc_customs_fix/libclc/CMakeLists.txt#L104), however those tools have not been set up. Also only search for llvm-spirv if it has not been...

Emit metadata to describe number of dimensions specified in `reqd_work_group_size`. This is needed in order to be able to use that metadata correctly, since it was specified for OpenCL, and...

This patch provides the following: * support for JIT compilation of Nvidia and AMD kernels * materialization of specialization constants values, turning them to de-facto compile time constants * caching...

The runtime is able to recognised [banned platforms](https://intel.github.io/llvm-docs/doxygen/namespacesycl_1_1__V1_1_1detail.html#af47bb6df075ff6aa60fc6855d59f6fe6), this patch provides a way to list them using `sycl-ls`' `verbose` switch.

This patch extends RTC support to GPU (AMD and Nvidia) targets. Additionally: * reinstate __SYCL_PROGRAM_METADATA_TAG_NEED_FINALIZATION tag, * split sycl.cpp RTC file to exclude IMF from the body of the main...