[OMPT] Drop LoongArch support

[Ami-zhang]

Summary

Currently, __builtin_frame_address() is required for OMPT. And
LoongArch only supports the current frame (i.e. depth = 0). So
this will fail if the depth != 0.

[xry111]

Is it possible to implement __builtin_frame_address for LoongArch then?

[MaskRay]

This change is only necessary and will be unneeded when __builtin_frame_address is implemented. OpenMP support is bringing up. LoongArch should not cause unnecessary churn to the upstream.
LoongArch users building openmp can disable LIBOMP_OMPT_SUPPORT for now.

[SixWeining]

In D137541#3911880, @xry111 wrote:

Is it possible to implement __builtin_frame_address for LoongArch then?

For __builtin_frame_address(depth) where depth > 0, can we return 0 directly if frame pointer elimination is enabled to avoid the issue mentioned here.
I wonder how gcc handle case?

[xry111]

GCC documentation claims:

On some machines it may be impossible to determine the frame
address of any function other than the current one; in such cases,
or when the top of the stack has been reached, this function
returns '0' if the first frame pointer is properly initialized by
the startup code.

Calling this function with a nonzero argument can have
unpredictable effects, including crashing the calling program.  As
a result, calls that are considered unsafe are diagnosed when the
'-Wframe-address' option is in effect.  Such calls should only be
made in debugging situations.

[Ami-zhang]

In D137541#3912741, @MaskRay wrote:

This change is only necessary and will be unneeded when __builtin_frame_address is implemented. OpenMP support is bringing up. LoongArch should not cause unnecessary churn to the upstream.
LoongArch users building openmp can disable LIBOMP_OMPT_SUPPORT for now.

I have abandoned this diff.
We'll try to support that condition where depth != 0 later .

[xry111]

If I read the code correctly, calling __builtin_frame_address with a depth > 0 is only used in tests. Can we disable those tests if __builtin_frame_address cannot handle depth > 0?

[SixWeining]

In D137541#3922921, @xry111 wrote:

If I read the code correctly, calling __builtin_frame_address with a depth > 0 is only used in tests. Can we disable those tests if __builtin_frame_address cannot handle depth > 0?

Good idea.

openmp/README.rst

 
 **LIBOMP_USE_INTERNODE_ALIGNMENT** = ``OFF|ON``
   Align certain data structures on 4096-byte.  This option is useful on
   multi-node systems where a small ``CACHE_LINE`` setting leads to false sharing.
 
 **LIBOMP_OMPT_SUPPORT** = ``ON|OFF``
   Include support for the OpenMP Tools Interface (OMPT).
   This option is supported and ``ON`` by default for x86, x86_64, AArch64,
-  PPC64, RISCV64 and LoongArch64 on Linux* and macOS*.
+  PPC64 and RISCV64 on Linux* and macOS*.
   This option is ``OFF`` if this feature is not supported for the platform.
 
 **LIBOMP_OMPT_OPTIONAL** = ``ON|OFF``
   Include support for optional OMPT functionality.  This option is ignored if
   **LIBOMP_OMPT_SUPPORT** is ``OFF``.
 
 **LIBOMP_STATS** = ``OFF|ON``
   Include stats-gathering code.

openmp/runtime/cmake/config-ix.cmake

else()
   if( # hardware architecture supported?
      ((LIBOMP_ARCH STREQUAL x86_64) OR
       (LIBOMP_ARCH STREQUAL i386) OR
 #      (LIBOMP_ARCH STREQUAL arm) OR
       (LIBOMP_ARCH STREQUAL aarch64) OR
       (LIBOMP_ARCH STREQUAL aarch64_a64fx) OR
       (LIBOMP_ARCH STREQUAL ppc64le) OR
       (LIBOMP_ARCH STREQUAL ppc64) OR
-      (LIBOMP_ARCH STREQUAL riscv64) OR
-      (LIBOMP_ARCH STREQUAL loongarch64))
+      (LIBOMP_ARCH STREQUAL riscv64))
      AND # OS supported?
      ((WIN32 AND LIBOMP_HAVE_PSAPI) OR APPLE OR (NOT WIN32 AND LIBOMP_HAVE_WEAK_ATTRIBUTE)))
     set(LIBOMP_HAVE_OMPT_SUPPORT TRUE)
   else()
     set(LIBOMP_HAVE_OMPT_SUPPORT FALSE)
   endif()
 endif()