jhuber6 added a comment.
This patch broke offloading on my machine. If I compile a basic offloading
application I get an invalid device image.
#include <omp.h>
int main() {
int isHost;
#pragma omp target map(from:isHost)
{ isHost = omp_is_initial_device(); }
return isHost;
}
Compiling and running with `clang -fopenmp -fopenmp-targets=nvptx64` gives me
the following results.
CUDA error: Error returned from cuModuleLoadDataEx
CUDA error: device kernel image is invalid
Libomptarget error: Unable to generate entries table for device id 0.
Libomptarget error: Failed to init globals on device 0
Libomptarget error: Run with LIBOMPTARGET_INFO=4 to dump host-target pointer
mappings.
Libomptarget fatal error 1: failure of target construct while offloading is
mandatory
Reverting `93d08acaacec951dbb302f77eeae51974985b6b2` fixes it.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D99551/new/
https://reviews.llvm.org/D99551
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
