MIPS64r6 LSA/DLSA instructions

[jkolek]

Summary

Implemented *_DESC and *_ENC classes for MIPS64r6 LSA and DLSA instructions.

[dsanders]

LSA/DLSA in MIPS32r6/MIPS64r6 are the same instructions as in the MSA ASE. Can you explain the need for a second copy rather than correcting the predicates of the pre-existing one?

[jkolek]

This patch implements support for LSA & DLSA instructions using predicates. This approach first includes all the MSA instructions in MIPS32r6/MIPS64r6 and then excludes all of them except LSA and DLSA. Maybe there is a better solution for this using predicates, but I didn't have a better idea. Zoran's suggestion is that we could reuse *_DESC and *_ENC classes from MSA, but to introduce new instruction definitions for MIPS32r6/MIPS64r6.

[dsanders]

To be clear, I was only requesting an explanation for the seemingly redundant definition. I was guessing that the 32 feature bit limitation was likely to be the reason you couldn't follow the pattern used elsewhere (see below for details) in which case I would have requested a FIXME comment and approved it.

Your previous patch is better than the current one. The current one makes it impossible to use MSA on MIPS32r6/MIPS64r6.

See FeatureMips3_32 for an example of how to do a 'FeatureX || FeatureY' predicate. You define a feature that is the set of instructions in both MIPS32r6 and MSA (likewise for MIPS64r6 and MSA) and make that feature an implied bit for both MIPS32r6 and MSA. The snag is that this requires extra feature bits and tablegen currently asserts when there's more than 32. I have some plans for fixing this (either lifting the 32 feature limit, or adding support for an or condition) but it won't be a quick fix.

lib/Target/Mips/MipsMSAInstrInfo.td

3252 ↗

(On Diff #9832)

It's a moot point since I've asked you to go back to the previous version of the patch but DLSA does not require the N64 ABI. It's requires 64-bit GPRs.

[jkolek]

Previous version of patch, rebased.

[dsanders]

You haven't answered the original question yet but I've just tried the FeatureMips3_32-style approach and we do hit the 32 feature bit limit. I know what needs to be done to fix that but I lack the time to do so at the moment. I therefore think it's best that we go ahead with this patch (with a minor change described below) and fix the 32 feature bit limit issue later.

Can you add a RUN line for mips32r6 and mips64r6 to test/CodeGen/Mips/msa/special.ll? E.g:

; RUN: llc -march=mips -mcpu=mips32r6 -mattr=+msa < %s | \
; RUN:   FileCheck %s --check-prefix=MIPS32
; RUN: llc -march=mips64 -mcpu=mips64r6 -mattr=+msa < %s | \
; RUN:   FileCheck %s --check-prefix=MIPS64

This will ensure that the two definitions don't cause problems for MSA on MIPS32r6/MIPS64r6.

With that change it will LGTM.

[jkolek]

I have added the RUN lines to test/CodeGen/Mips/msa/special.ll.

[dsanders]

LGTM

[jkolek]

Closed by commit rL211346 (authored by zjovanovic).

llvm/trunk/lib/Target/Mips/Mips32r6InstrFormats.td

 def OPCODE6_LL       : OPCODE6<0b110110>;
 def OPCODE6_LLD      : OPCODE6<0b110111>;
 def OPCODE6_SC       : OPCODE6<0b100110>;
 def OPCODE6_SCD      : OPCODE6<0b100111>;
 def OPCODE6_CLO      : OPCODE6<0b010001>;
 def OPCODE6_CLZ      : OPCODE6<0b010000>;
 def OPCODE6_DCLO     : OPCODE6<0b010011>;
 def OPCODE6_DCLZ     : OPCODE6<0b010010>;
+def OPCODE6_LSA      : OPCODE6<0b000101>;
+def OPCODE6_DLSA     : OPCODE6<0b010101>;
 
 class FIELD_FMT<bits<5> Val> {
   bits<5> Value = Val;
 }
 def FIELD_FMT_S : FIELD_FMT<0b10000>;
 def FIELD_FMT_D : FIELD_FMT<0b10001>;
 
 class FIELD_CMP_COND<bits<5> Val> {
class SPECIAL3_LL_SC_FM<OPCODE6 Operation> : MipsR6Inst {
   let Inst{25-21} = base;
   let Inst{20-16} = rt;
   let Inst{15-7} = offset;
   let Inst{5-0} = Operation.Value;
 
   string DecoderMethod = "DecodeSpecial3LlSc";
 }
 
+class SPECIAL_LSA_FM<OPCODE6 Operation> : MipsR6Inst {
+  bits<5> rd;
+  bits<5> rs;
+  bits<5> rt;
+  bits<2> imm2;
+
+  bits<32> Inst;
+
+  let Inst{31-26} = OPGROUP_SPECIAL.Value;
+  let Inst{25-21} = rs;
+  let Inst{20-16} = rt;
+  let Inst{15-11} = rd;
+  let Inst{10-8}  = 0b000;
+  let Inst{7-6}   = imm2;
+  let Inst{5-0}   = Operation.Value;
+}
+
 class REGIMM_FM<OPCODE5 Operation> : MipsR6Inst {
   bits<5> rs;
   bits<16> imm;
 
   bits<32> Inst;
 
   let Inst{31-26} = OPGROUP_REGIMM.Value;
   let Inst{25-21} = rs;

llvm/trunk/lib/Target/Mips/Mips32r6InstrInfo.td

 class CACHE_ENC : SPECIAL3_MEM_FM<OPCODE6_CACHE>;
 class PREF_ENC : SPECIAL3_MEM_FM<OPCODE6_PREF>;
 
 class LDC2_R6_ENC : COP2LDST_FM<OPCODE5_LDC2>;
 class LWC2_R6_ENC : COP2LDST_FM<OPCODE5_LWC2>;
 class SDC2_R6_ENC : COP2LDST_FM<OPCODE5_SDC2>;
 class SWC2_R6_ENC : COP2LDST_FM<OPCODE5_SWC2>;
 
+class LSA_R6_ENC : SPECIAL_LSA_FM<OPCODE6_LSA>;
+
 class LL_R6_ENC : SPECIAL3_LL_SC_FM<OPCODE6_LL>;
 class SC_R6_ENC : SPECIAL3_LL_SC_FM<OPCODE6_SC>;
 
 class CLO_R6_ENC : SPECIAL_2R_FM<OPCODE6_CLO>;
 class CLZ_R6_ENC : SPECIAL_2R_FM<OPCODE6_CLZ>;
 
 class CMP_CONDN_DESC_BASE<string CondStr, string Typestr,
                           RegisterOperand FGROpnd,
class COP2ST_DESC_BASE<string instr_asm, RegisterOperand COPOpnd> {
   string AsmString = !strconcat(instr_asm, "\t$rt, $addr");
   list<dag> Pattern = [];
   bit mayStore = 1;
 }
 
 class SDC2_R6_DESC : COP2ST_DESC_BASE<"sdc2", COP2Opnd>;
 class SWC2_R6_DESC : COP2ST_DESC_BASE<"swc2", COP2Opnd>;
 
+class LSA_R6_DESC_BASE<string instr_asm, RegisterOperand GPROpnd,
+                       Operand ImmOpnd> {
+  dag OutOperandList = (outs GPROpnd:$rd);
+  dag InOperandList = (ins GPROpnd:$rs, GPROpnd:$rt, ImmOpnd:$imm2);
+  string AsmString = !strconcat(instr_asm, "\t$rd, $rs, $rt, $imm2");
+  list<dag> Pattern = [];
+}
+
+class LSA_R6_DESC : LSA_R6_DESC_BASE<"lsa", GPR32Opnd, uimm2>;
+
 class LL_R6_DESC_BASE<string instr_asm, RegisterOperand GPROpnd> {
   dag OutOperandList = (outs GPROpnd:$rt);
   dag InOperandList = (ins mem_simm9:$addr);
   string AsmString = !strconcat(instr_asm, "\t$rt, $addr");
   list<dag> Pattern = [];
   bit mayLoad = 1;
 }
 

 defm D : CMP_CC_M<FIELD_CMP_FORMAT_D, "d", FGR64Opnd>;
 def DIV : DIV_ENC, DIV_DESC, ISA_MIPS32R6;
 def DIVU : DIVU_ENC, DIVU_DESC, ISA_MIPS32R6;
 def JIALC : JIALC_ENC, JIALC_DESC, ISA_MIPS32R6;
 def JIC : JIC_ENC, JIC_DESC, ISA_MIPS32R6;
 def JR_HB_R6 : JR_HB_R6_ENC, JR_HB_R6_DESC, ISA_MIPS32R6;
 def LDC2_R6 : LDC2_R6_ENC, LDC2_R6_DESC, ISA_MIPS32R6;
 def LL_R6 : LL_R6_ENC, LL_R6_DESC, ISA_MIPS32R6;
-// def LSA; // See MSA
+def LSA_R6 : LSA_R6_ENC, LSA_R6_DESC, ISA_MIPS32R6;
 def LWC2_R6 : LWC2_R6_ENC, LWC2_R6_DESC, ISA_MIPS32R6;
 def LWPC : LWPC_ENC, LWPC_DESC, ISA_MIPS32R6;
 def LWUPC : LWUPC_ENC, LWUPC_DESC, ISA_MIPS32R6;
 def MADDF_S : MADDF_S_ENC, MADDF_S_DESC, ISA_MIPS32R6;
 def MADDF_D : MADDF_D_ENC, MADDF_D_DESC, ISA_MIPS32R6;
 def MAXA_D : MAXA_D_ENC, MAXA_D_DESC, ISA_MIPS32R6;
 def MAXA_S : MAXA_S_ENC, MAXA_S_DESC, ISA_MIPS32R6;
 def MAX_D : MAX_D_ENC, MAX_D_DESC, ISA_MIPS32R6;

llvm/trunk/lib/Target/Mips/Mips64r6InstrInfo.td

 class DAUI_ENC    : DAUI_FM;
 class DAHI_ENC    : REGIMM_FM<OPCODE5_DAHI>;
 class DATI_ENC    : REGIMM_FM<OPCODE5_DATI>;
 class DBITSWAP_ENC : SPECIAL3_2R_FM<OPCODE6_DBITSWAP>;
 class DCLO_R6_ENC : SPECIAL_2R_FM<OPCODE6_DCLO>;
 class DCLZ_R6_ENC : SPECIAL_2R_FM<OPCODE6_DCLZ>;
 class DDIV_ENC    : SPECIAL_3R_FM<0b00010, 0b011110>;
 class DDIVU_ENC   : SPECIAL_3R_FM<0b00010, 0b011111>;
+class DLSA_R6_ENC : SPECIAL_LSA_FM<OPCODE6_DLSA>;
 class DMOD_ENC    : SPECIAL_3R_FM<0b00011, 0b011110>;
 class DMODU_ENC   : SPECIAL_3R_FM<0b00011, 0b011111>;
 class DMUH_ENC    : SPECIAL_3R_FM<0b00011, 0b111000>;
 class DMUHU_ENC   : SPECIAL_3R_FM<0b00011, 0b111001>;
 class DMUL_R6_ENC : SPECIAL_3R_FM<0b00010, 0b111000>;
 class DMULU_ENC   : SPECIAL_3R_FM<0b00010, 0b111001>;
 class LDPC_ENC    : PCREL18_FM<OPCODE3_LDPC>;
 class LLD_R6_ENC : SPECIAL3_LL_SC_FM<OPCODE6_LLD>;

 class DAHI_DESC    : AHI_ATI_DESC_BASE<"dahi", GPR64Opnd>;
 class DATI_DESC    : AHI_ATI_DESC_BASE<"dati", GPR64Opnd>;
 class DAUI_DESC    : AUI_DESC_BASE<"daui", GPR64Opnd>;
 class DBITSWAP_DESC : BITSWAP_DESC_BASE<"dbitswap", GPR64Opnd>;
 class DCLO_R6_DESC : CLO_R6_DESC_BASE<"dclo", GPR64Opnd>;
 class DCLZ_R6_DESC : CLZ_R6_DESC_BASE<"dclz", GPR64Opnd>;
 class DDIV_DESC    : DIVMOD_DESC_BASE<"ddiv", GPR64Opnd, sdiv>;
 class DDIVU_DESC   : DIVMOD_DESC_BASE<"ddivu", GPR64Opnd, udiv>;
+class DLSA_R6_DESC : LSA_R6_DESC_BASE<"dlsa", GPR64Opnd, uimm2>;
 class DMOD_DESC    : DIVMOD_DESC_BASE<"dmod", GPR64Opnd, srem>;
 class DMODU_DESC   : DIVMOD_DESC_BASE<"dmodu", GPR64Opnd, urem>;
 class DMUH_DESC    : MUL_R6_DESC_BASE<"dmuh", GPR64Opnd, mulhs>;
 class DMUHU_DESC   : MUL_R6_DESC_BASE<"dmuhu", GPR64Opnd, mulhu>;
 class DMUL_R6_DESC : MUL_R6_DESC_BASE<"dmul", GPR64Opnd, mul>;
 class DMULU_DESC   : MUL_R6_DESC_BASE<"dmulu", GPR64Opnd>;
 class LDPC_DESC    : PCREL_DESC_BASE<"ldpc", GPR64Opnd, simm18_lsl3>;
 class LLD_R6_DESC   : LL_R6_DESC_BASE<"lld", GPR64Opnd>;

 def DALIGN : DALIGN_ENC, DALIGN_DESC, ISA_MIPS64R6;
 def DATI : DATI_ENC, DATI_DESC, ISA_MIPS64R6;
 def DAUI : DAUI_ENC, DAUI_DESC, ISA_MIPS64R6;
 def DBITSWAP : DBITSWAP_ENC, DBITSWAP_DESC, ISA_MIPS64R6;
 def DCLO_R6 : DCLO_R6_ENC, DCLO_R6_DESC, ISA_MIPS64R6;
 def DCLZ_R6 : DCLZ_R6_ENC, DCLZ_R6_DESC, ISA_MIPS64R6;
 def DDIV : DDIV_ENC, DDIV_DESC, ISA_MIPS64R6;
 def DDIVU : DDIVU_ENC, DDIVU_DESC, ISA_MIPS64R6;
-// def DLSA; // See MSA
+def DLSA_R6 : DLSA_R6_ENC, DLSA_R6_DESC, ISA_MIPS64R6;
 def DMOD : DMOD_ENC, DMOD_DESC, ISA_MIPS64R6;
 def DMODU : DMODU_ENC, DMODU_DESC, ISA_MIPS64R6;
 def DMUH: DMUH_ENC, DMUH_DESC, ISA_MIPS64R6;
 def DMUHU: DMUHU_ENC, DMUHU_DESC, ISA_MIPS64R6;
 def DMUL_R6: DMUL_R6_ENC, DMUL_R6_DESC, ISA_MIPS64R6;
 def DMULU: DMULU_ENC, DMULU_DESC, ISA_MIPS64R6;
 def LDPC: LDPC_ENC, LDPC_DESC, ISA_MIPS64R6;
 def LLD_R6 : LLD_R6_ENC, LLD_R6_DESC, ISA_MIPS32R6;

llvm/trunk/test/CodeGen/Mips/msa/special.ll

 ; Test the MSA intrinsics that are encoded with the SPECIAL instruction format.
 
 ; RUN: llc -march=mips -mattr=+msa,+fp64 < %s | \
 ; RUN:   FileCheck %s --check-prefix=MIPS32
 ; RUN: llc -march=mips64 -mcpu=mips64r2 -mattr=+msa,+fp64 < %s | \
 ; RUN:   FileCheck %s --check-prefix=MIPS64
+; RUN: llc -march=mips -mcpu=mips32r6 -mattr=+msa < %s | \
+; RUN:   FileCheck %s --check-prefix=MIPS32
+; RUN: llc -march=mips64 -mcpu=mips64r6 -mattr=+msa < %s | \
+; RUN:   FileCheck %s --check-prefix=MIPS64
 
 define i32 @llvm_mips_lsa_test(i32 %a, i32 %b) nounwind {
 entry:
   %0 = tail call i32 @llvm.mips.lsa(i32 %a, i32 %b, i32 2)
   ret i32 %0
 }
 
 declare i32 @llvm.mips.lsa(i32, i32, i32) nounwind

llvm/trunk/test/MC/Mips/mips32r6/valid.s

# RUN: FileCheck %s -check-prefix=WARNING < %t0
         cmp.le.s   $f2,$f3,$f4      # CHECK: cmp.le.s $f2, $f3, $f4  # encoding: [0x46,0x84,0x18,0x8e]
         cmp.le.d   $f2,$f3,$f4      # CHECK: cmp.le.d $f2, $f3, $f4  # encoding: [0x46,0xa4,0x18,0x8e]
         cmp.ngt.s  $f2,$f3,$f4      # CHECK: cmp.ngt.s $f2, $f3, $f4  # encoding: [0x46,0x84,0x18,0x8f]
         cmp.ngt.d  $f2,$f3,$f4      # CHECK: cmp.ngt.d $f2, $f3, $f4  # encoding: [0x46,0xa4,0x18,0x8f]
         div     $2,$3,$4         # CHECK: div $2, $3, $4   # encoding: [0x00,0x64,0x10,0x9a]
         divu    $2,$3,$4         # CHECK: divu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x9b]
         jialc   $5, 256          # CHECK: jialc $5, 256    # encoding: [0xf8,0x05,0x01,0x00]
         jic     $5, 256          # CHECK: jic $5, 256      # encoding: [0xd8,0x05,0x01,0x00]
+        lsa     $2, $3, $4, 3    # CHECK: lsa  $2, $3, $4, 3 # encoding: [0x00,0x64,0x10,0xc5]
         lwpc    $2,268           # CHECK: lwpc $2, 268     # encoding: [0xec,0x48,0x00,0x43]
         lwupc   $2,268           # CHECK: lwupc $2, 268    # encoding: [0xec,0x50,0x00,0x43]
         mod     $2,$3,$4         # CHECK: mod $2, $3, $4   # encoding: [0x00,0x64,0x10,0xda]
         modu    $2,$3,$4         # CHECK: modu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xdb]
         mul     $2,$3,$4         # CHECK: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
         muh     $2,$3,$4         # CHECK: muh $2, $3, $4   # encoding: [0x00,0x64,0x10,0xd8]
         mulu    $2,$3,$4         # CHECK: mulu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x99]
         muhu    $2,$3,$4         # CHECK: muhu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xd9]

llvm/trunk/test/MC/Mips/mips64r6/valid.s

# RUN: FileCheck %s -check-prefix=WARNING < %t0
         jialc   $5, 256          # CHECK: jialc $5, 256    # encoding: [0xf8,0x05,0x01,0x00]
         jic     $5, 256          # CHECK: jic $5, 256      # encoding: [0xd8,0x05,0x01,0x00]
         mod     $2,$3,$4         # CHECK: mod $2, $3, $4   # encoding: [0x00,0x64,0x10,0xda]
         modu    $2,$3,$4         # CHECK: modu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xdb]
         ddiv    $2,$3,$4         # CHECK: ddiv $2, $3, $4  # encoding: [0x00,0x64,0x10,0x9e]
         ddivu   $2,$3,$4         # CHECK: ddivu $2, $3, $4 # encoding: [0x00,0x64,0x10,0x9f]
         dmod    $2,$3,$4         # CHECK: dmod $2, $3, $4  # encoding: [0x00,0x64,0x10,0xde]
         dmodu   $2,$3,$4         # CHECK: dmodu $2, $3, $4 # encoding: [0x00,0x64,0x10,0xdf]
+        lsa     $2, $3, $4, 3    # CHECK: lsa  $2, $3, $4, 3 # encoding: [0x00,0x64,0x10,0xc5]
+        dlsa    $2, $3, $4, 3    # CHECK: dlsa $2, $3, $4, 3 # encoding: [0x00,0x64,0x10,0xd5]
         ldpc    $2,123456        # CHECK: ldpc $2, 123456  # encoding: [0xec,0x58,0x3c,0x48]
         lwpc    $2,268           # CHECK: lwpc $2, 268     # encoding: [0xec,0x48,0x00,0x43]
         lwupc   $2,268           # CHECK: lwupc $2, 268    # encoding: [0xec,0x50,0x00,0x43]
         mul     $2,$3,$4         # CHECK: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
         muh     $2,$3,$4         # CHECK: muh $2, $3, $4   # encoding: [0x00,0x64,0x10,0xd8]
         mulu    $2,$3,$4         # CHECK: mulu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x99]
         muhu    $2,$3,$4         # CHECK: muhu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xd9]
         dmul    $2,$3,$4         # CHECK: dmul $2, $3, $4  # encoding: [0x00,0x64,0x10,0xb8]