[llvm-branch-commits] [AMDGPU] Set AS8 address width to 48 bits (PR #139419)

[Alexander Richardson via llvm-branch-commits]

================
@@ -145,79 +145,79 @@ define amdgpu_ps ptr addrspace(7) @s_ptrmask_buffer_fat_ptr_i32_neg8(ptr addrspa
   ret ptr addrspace(7) %masked
 }
 
-define ptr addrspace(8) @v_ptrmask_buffer_resource_variable_i128(ptr addrspace(8) %ptr, i128 %mask) {
-; GCN-LABEL: v_ptrmask_buffer_resource_variable_i128:
+define ptr addrspace(8) @v_ptrmask_buffer_resource_variable_i48(ptr addrspace(8) %ptr, i48 %mask) {
+; GCN-LABEL: v_ptrmask_buffer_resource_variable_i48:
 ; GCN:       ; %bb.0:
 ; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_or_b32_e32 v5, 0xffff0000, v5
 ; GCN-NEXT:    v_and_b32_e32 v1, v1, v5
 ; GCN-NEXT:    v_and_b32_e32 v0, v0, v4
-; GCN-NEXT:    v_and_b32_e32 v3, v3, v7
-; GCN-NEXT:    v_and_b32_e32 v2, v2, v6
 ; GCN-NEXT:    s_setpc_b64 s[30:31]
 ;
-; GFX10PLUS-LABEL: v_ptrmask_buffer_resource_variable_i128:
+; GFX10PLUS-LABEL: v_ptrmask_buffer_resource_variable_i48:
 ; GFX10PLUS:       ; %bb.0:
 ; GFX10PLUS-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10PLUS-NEXT:    v_or_b32_e32 v5, 0xffff0000, v5
 ; GFX10PLUS-NEXT:    v_and_b32_e32 v0, v0, v4
 ; GFX10PLUS-NEXT:    v_and_b32_e32 v1, v1, v5
-; GFX10PLUS-NEXT:    v_and_b32_e32 v2, v2, v6
-; GFX10PLUS-NEXT:    v_and_b32_e32 v3, v3, v7
 ; GFX10PLUS-NEXT:    s_setpc_b64 s[30:31]
-  %masked = call ptr addrspace(8) @llvm.ptrmask.p8.i128(ptr addrspace(8) %ptr, i128 %mask)
+  %masked = call ptr addrspace(8) @llvm.ptrmask.p8.i48(ptr addrspace(8) %ptr, i48 %mask)
   ret ptr addrspace(8) %masked
 }
 
-define ptr addrspace(8) @v_ptrmask_buffer_resource_variable_i128_neg8(ptr addrspace(8) %ptr) {
-; GCN-LABEL: v_ptrmask_buffer_resource_variable_i128_neg8:
+define ptr addrspace(8) @v_ptrmask_buffer_resource_variable_i48_neg8(ptr addrspace(8) %ptr) {
+; GCN-LABEL: v_ptrmask_buffer_resource_variable_i48_neg8:
 ; GCN:       ; %bb.0:
 ; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GCN-NEXT:    v_and_b32_e32 v0, -8, v0
 ; GCN-NEXT:    s_setpc_b64 s[30:31]
 ;
-; GFX10PLUS-LABEL: v_ptrmask_buffer_resource_variable_i128_neg8:
+; GFX10PLUS-LABEL: v_ptrmask_buffer_resource_variable_i48_neg8:
 ; GFX10PLUS:       ; %bb.0:
 ; GFX10PLUS-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10PLUS-NEXT:    v_and_b32_e32 v0, -8, v0
 ; GFX10PLUS-NEXT:    s_setpc_b64 s[30:31]
-  %masked = call ptr addrspace(8) @llvm.ptrmask.p8.i128(ptr addrspace(8) %ptr, i128 -8)
+  %masked = call ptr addrspace(8) @llvm.ptrmask.p8.i48(ptr addrspace(8) %ptr, i48 -8)
   ret ptr addrspace(8) %masked
 }
 
-define amdgpu_ps ptr addrspace(8) @s_ptrmask_buffer_resource_variable_i128(ptr addrspace(8) inreg %ptr, i128 inreg %mask) {
-; GCN-LABEL: s_ptrmask_buffer_resource_variable_i128:
+define amdgpu_ps ptr addrspace(8) @s_ptrmask_buffer_resource_variable_i48(ptr addrspace(8) inreg %ptr, i48 inreg %mask) {
+; GCN-LABEL: s_ptrmask_buffer_resource_variable_i48:
 ; GCN:       ; %bb.0:
-; GCN-NEXT:    s_and_b64 s[4:5], s[4:5], s[8:9]
+; GCN-NEXT:    s_or_b32 s7, s7, 0xffff0000
----------------
arichardson wrote:

I can't read AMDGPU assembly properly so would be good to double-check that this is correct.

https://github.com/llvm/llvm-project/pull/139419