60b1967c39
Add the scratch wave offset to the scratch buffer descriptor (SRSrc) in the entry function prologue. This allows us to removes the scratch wave offset register from the calling convention ABI. As part of this change, allow the use of an inline constant zero for the SOffset of MUBUF instructions accessing the stack in entry functions when a frame pointer is not requested/required. Entry functions with calls still need to set up the calling convention ABI stack pointer register, and reference it in order to address arguments of called functions. The ABI stack pointer register remains unswizzled, but is now wave-relative instead of queue-relative. Non-entry functions also use an inline constant zero SOffset for wave-relative scratch access, but continue to use the stack and frame pointers as before. When the stack or frame pointer is converted to a swizzled offset it is now scaled directly, as the scratch wave offset no longer needs to be subtracted first. Update llvm/docs/AMDGPUUsage.rst to reflect these changes to the calling convention. Tags: #llvm Differential Revision: https://reviews.llvm.org/D75138
77 lines
3.8 KiB
LLVM
77 lines
3.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs | FileCheck -check-prefix=GCN %s
|
|
|
|
; FIXME: The MUBUF loads in this test output are incorrect, their SOffset
|
|
; should use the frame offset register, not the ABI stack pointer register. We
|
|
; rely on the frame index argument of MUBUF stack accesses to survive until PEI
|
|
; so we can fix up the SOffset to use the correct frame register in
|
|
; eliminateFrameIndex. Some things like LocalStackSlotAllocation can lift the
|
|
; frame index up into something (e.g. `v_add_nc_u32`) that we cannot fold back
|
|
; into the MUBUF instruction, and so we end up emitting an incorrect offset.
|
|
; Fixing this may involve adding stack access pseudos so that we don't have to
|
|
; speculatively refer to the ABI stack pointer register at all.
|
|
|
|
; An assert was hit when frame offset register was used to address FrameIndex.
|
|
define amdgpu_kernel void @kernel_background_evaluate(float addrspace(5)* %kg, <4 x i32> addrspace(1)* %input, <4 x float> addrspace(1)* %output, i32 %i) {
|
|
; GCN-LABEL: kernel_background_evaluate:
|
|
; GCN: ; %bb.0: ; %entry
|
|
; GCN-NEXT: s_load_dword s6, s[0:1], 0x24
|
|
; GCN-NEXT: s_mov_b32 s36, SCRATCH_RSRC_DWORD0
|
|
; GCN-NEXT: s_mov_b32 s37, SCRATCH_RSRC_DWORD1
|
|
; GCN-NEXT: s_mov_b32 s38, -1
|
|
; GCN-NEXT: s_mov_b32 s39, 0x31c16000
|
|
; GCN-NEXT: s_add_u32 s36, s36, s3
|
|
; GCN-NEXT: s_addc_u32 s37, s37, 0
|
|
; GCN-NEXT: v_mov_b32_e32 v1, 0x2000
|
|
; GCN-NEXT: v_mov_b32_e32 v2, 0x4000
|
|
; GCN-NEXT: v_mov_b32_e32 v3, 0
|
|
; GCN-NEXT: v_mov_b32_e32 v4, 0x400000
|
|
; GCN-NEXT: s_mov_b64 s[0:1], s[36:37]
|
|
; GCN-NEXT: s_mov_b64 s[2:3], s[38:39]
|
|
; GCN-NEXT: s_mov_b32 s32, 0xc0000
|
|
; GCN-NEXT: v_add_nc_u32_e64 v32, 4, 0x4000
|
|
; GCN-NEXT: ; implicit-def: $vcc_hi
|
|
; GCN-NEXT: s_getpc_b64 s[4:5]
|
|
; GCN-NEXT: s_add_u32 s4, s4, svm_eval_nodes@rel32@lo+4
|
|
; GCN-NEXT: s_addc_u32 s5, s5, svm_eval_nodes@rel32@hi+4
|
|
; GCN-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GCN-NEXT: v_mov_b32_e32 v0, s6
|
|
; GCN-NEXT: s_swappc_b64 s[30:31], s[4:5]
|
|
; GCN-NEXT: v_cmp_ne_u32_e32 vcc_lo, 0, v0
|
|
; GCN-NEXT: s_and_saveexec_b32 s0, vcc_lo
|
|
; GCN-NEXT: s_cbranch_execz BB0_2
|
|
; GCN-NEXT: ; %bb.1: ; %if.then4.i
|
|
; GCN-NEXT: buffer_load_dword v0, v32, s[36:39], s32 offen
|
|
; GCN-NEXT: buffer_load_dword v1, v32, s[36:39], s32 offen offset:4
|
|
; GCN-NEXT: s_waitcnt vmcnt(0)
|
|
; GCN-NEXT: v_add_nc_u32_e32 v0, v1, v0
|
|
; GCN-NEXT: v_mul_lo_u32 v0, 0x41c64e6d, v0
|
|
; GCN-NEXT: v_add_nc_u32_e32 v0, 0x3039, v0
|
|
; GCN-NEXT: buffer_store_dword v0, v0, s[36:39], 0 offen
|
|
; GCN-NEXT: BB0_2: ; %shader_eval_surface.exit
|
|
; GCN-NEXT: s_endpgm
|
|
entry:
|
|
%sd = alloca < 1339 x i32>, align 16, addrspace(5)
|
|
%state = alloca <4 x i32>, align 4, addrspace(5)
|
|
%rslt = call i32 @svm_eval_nodes(float addrspace(5)* %kg, <1339 x i32> addrspace(5)* %sd, <4 x i32> addrspace(5)* %state, i32 0, i32 4194304)
|
|
%cmp = icmp eq i32 %rslt, 0
|
|
br i1 %cmp, label %shader_eval_surface.exit, label %if.then4.i
|
|
|
|
if.then4.i: ; preds = %entry
|
|
%rng_hash.i.i = getelementptr inbounds < 4 x i32>, <4 x i32> addrspace(5)* %state, i32 0, i32 1
|
|
%tmp0 = load i32, i32 addrspace(5)* %rng_hash.i.i, align 4
|
|
%rng_offset.i.i = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %state, i32 0, i32 2
|
|
%tmp1 = load i32, i32 addrspace(5)* %rng_offset.i.i, align 4
|
|
%add.i.i = add i32 %tmp1, %tmp0
|
|
%add1.i.i = add i32 %add.i.i, 0
|
|
%mul.i.i.i.i = mul i32 %add1.i.i, 1103515245
|
|
%add.i.i.i.i = add i32 %mul.i.i.i.i, 12345
|
|
store i32 %add.i.i.i.i, i32 addrspace(5)* undef, align 16
|
|
br label %shader_eval_surface.exit
|
|
|
|
shader_eval_surface.exit: ; preds = %entry
|
|
ret void
|
|
}
|
|
|
|
declare hidden i32 @svm_eval_nodes(float addrspace(5)*, <1339 x i32> addrspace(5)*, <4 x i32> addrspace(5)*, i32, i32) local_unnamed_addr
|