AMDGPU: Add description for new atomicrmw metadata (#85052)

Add a spec for yet-to-be-implemented metadata to allow the backend to
fully handle atomicrmw lowering. This is the base of an alternative
to #69229, which inverts the direction to be correct by default, and
extends to cover the peer device case.

Author: arsenm

llvm/docs/AMDGPUUsage.rst

@@ -1394,6 +1394,84 @@ arguments.
 
   %val = load i32, ptr %in, align 4, !amdgpu.last.use !{}
 
+'``amdgpu.no.remote.memory``' Metadata
+---------------------------------------------
+
+Asserts a memory operation does not access bytes in host memory, or
+remote connected peer device memory (the address must be device
+local). This is intended for use with :ref:`atomicrmw <i_atomicrmw>`
+and other atomic instructions. This is required to emit a native
+hardware instruction for some :ref:`system scope
+<amdgpu-memory-scopes>` atomic operations on some subtargets. For most
+integer atomic operations, this is a sufficient restriction to emit a
+native atomic instruction.
+
+An :ref:`atomicrmw <i_atomicrmw>` without metadata will be treated
+conservatively as required to preserve the operation behavior in all
+cases. This will typically be used in conjunction with
+:ref:`\!amdgpu.no.fine.grained.memory<amdgpu_no_fine_grained_memory>`.
+
+
+.. code-block:: llvm
+
+  ; Indicates the atomic does not access fine-grained memory, or
+  ; remote device memory.
+  %old0 = atomicrmw sub ptr %ptr0, i32 1 acquire, !amdgpu.no.fine.grained.memory !0, !amdgpu.no.remote.memory !0
+
+  ; Indicates the atomic does not access peer device memory.
+  %old2 = atomicrmw sub ptr %ptr2, i32 1 acquire, !amdgpu.no.remote.memory !0
+
+  !0 = !{}
+
+.. _amdgpu_no_fine_grained_memory:
+
+'``amdgpu.no.fine.grained.memory``' Metadata
+-------------------------------------------------
+
+Asserts a memory access does not access bytes allocated in
+fine-grained allocated memory. This is intended for use with
+:ref:`atomicrmw <i_atomicrmw>` and other atomic instructions. This is
+required to emit a native hardware instruction for some :ref:`system
+scope <amdgpu-memory-scopes>` atomic operations on some subtargets. An
+:ref:`atomicrmw <i_atomicrmw>` without metadata will be treated
+conservatively as required to preserve the operation behavior in all
+cases. This will typically be used in conjunction with
+:ref:`\!amdgpu.no.remote.memory.access<amdgpu_no_remote_memory_access>`.
+
+.. code-block:: llvm
+
+  ; Indicates the access does not access fine-grained memory, or
+  ; remote device memory.
+  %old0 = atomicrmw sub ptr %ptr0, i32 1 acquire, !amdgpu.no.fine.grained.memory !0, !amdgpu.no.remote.memory.access !0
+
+  ; Indicates the access does not access fine-grained memory
+  %old2 = atomicrmw sub ptr %ptr2, i32 1 acquire, !amdgpu.no.fine.grained.memory !0
+
+  !0 = !{}
+
+.. _amdgpu_no_remote_memory_access:
+
+'``amdgpu.ignore.denormal.mode``' Metadata
+------------------------------------------
+
+For use with :ref:`atomicrmw <i_atomicrmw>` floating-point
+operations. Indicates the handling of denormal inputs and results is
+insignificant and may be inconsistent with the expected floating-point
+mode. This is necessary to emit a native atomic instruction on some
+targets for some address spaces where float denormals are
+unconditionally flushed. This is typically used in conjunction with
+:ref:`\!amdgpu.no.remote.memory.access<amdgpu_no_remote_memory_access>`
+and
+:ref:`\!amdgpu.no.fine.grained.memory<amdgpu_no_fine_grained_memory>`
+
+
+.. code-block:: llvm
+
+  %res0 = atomicrmw fadd ptr addrspace(1) %ptr, float %value seq_cst, align 4, !amdgpu.ignore.denormal.mode !0
+  %res1 = atomicrmw fadd ptr addrspace(1) %ptr, float %value seq_cst, align 4, !amdgpu.ignore.denormal.mode !0, !amdgpu.no.fine.grained.memory !0, !amdgpu.no.remote.memory.access !0
+
+  !0 = !{}
+
 
 LLVM IR Attributes
 ==================

llvm/docs/ReleaseNotes.rst

@@ -129,6 +129,8 @@ Changes to the AMDGPU Backend
 -----------------------------
 
 * Implemented the ``llvm.get.fpenv`` and ``llvm.set.fpenv`` intrinsics.
+* Added ``!amdgpu.no.fine.grained.memory`` and
+  ``!amdgpu.no.remote.memory`` metadata to control atomic behavior.
 
 * Implemented :ref:`llvm.get.rounding <int_get_rounding>` and :ref:`llvm.set.rounding <int_set_rounding>`