diff --git a/src/intel/compiler/brw_asm.c b/src/intel/compiler/brw_asm.c
index 2bbac833eb2..4a5e3e76516 100644
--- a/src/intel/compiler/brw_asm.c
+++ b/src/intel/compiler/brw_asm.c
@@ -44,6 +44,8 @@ i965_postprocess_labels()
                case BRW_OPCODE_ELSE:
                case BRW_OPCODE_ENDIF:
                case BRW_OPCODE_WHILE:
+               case BRW_OPCODE_GOTO:
+               case BRW_OPCODE_JOIN:
                   brw_eu_inst_set_jip(p->devinfo, inst, relative_offset);
                   break;
                case BRW_OPCODE_BREAK:
@@ -59,6 +61,7 @@ i965_postprocess_labels()
                switch (opcode) {
                case BRW_OPCODE_IF:
                case BRW_OPCODE_ELSE:
+               case BRW_OPCODE_GOTO:
                   brw_eu_inst_set_uip(p->devinfo, inst, relative_offset);
                   break;
                case BRW_OPCODE_WHILE:
diff --git a/src/intel/compiler/brw_disasm.c b/src/intel/compiler/brw_disasm.c
index 9d123d93c1c..63d006fb48e 100644
--- a/src/intel/compiler/brw_disasm.c
+++ b/src/intel/compiler/brw_disasm.c
@@ -45,7 +45,9 @@ brw_has_jip(const struct intel_device_info *devinfo, enum opcode opcode)
           opcode == BRW_OPCODE_WHILE ||
           opcode == BRW_OPCODE_BREAK ||
           opcode == BRW_OPCODE_CONTINUE ||
-          opcode == BRW_OPCODE_HALT;
+          opcode == BRW_OPCODE_HALT ||
+          opcode == BRW_OPCODE_GOTO ||
+          opcode == BRW_OPCODE_JOIN;
 }
 
 bool
@@ -55,7 +57,8 @@ brw_has_uip(const struct intel_device_info *devinfo, enum opcode opcode)
           opcode == BRW_OPCODE_ELSE ||
           opcode == BRW_OPCODE_BREAK ||
           opcode == BRW_OPCODE_CONTINUE ||
-          opcode == BRW_OPCODE_HALT;
+          opcode == BRW_OPCODE_HALT ||
+          opcode == BRW_OPCODE_GOTO;
 }
 
 bool
diff --git a/src/intel/compiler/brw_eu.c b/src/intel/compiler/brw_eu.c
index b8300d3a4c0..6fbd4efd3c3 100644
--- a/src/intel/compiler/brw_eu.c
+++ b/src/intel/compiler/brw_eu.c
@@ -617,6 +617,7 @@ static const struct opcode_desc opcode_descs[] = {
    { BRW_OPCODE_CALL,     44,  "call",    0,    0,    GFX_ALL },
    { BRW_OPCODE_RET,      45,  "ret",     0,    0,    GFX_ALL },
    { BRW_OPCODE_GOTO,     46,  "goto",    0,    0,    GFX_ALL },
+   { BRW_OPCODE_JOIN,     47,  "join",    0,    0,    GFX_ALL },
    { BRW_OPCODE_WAIT,     48,  "wait",    0,    1,    GFX_LT(GFX12) },
    { BRW_OPCODE_SEND,     49,  "send",    1,    1,    GFX_LT(GFX12) },
    { BRW_OPCODE_SENDC,    50,  "sendc",   1,    1,    GFX_LT(GFX12) },
diff --git a/src/intel/compiler/brw_eu_compact.c b/src/intel/compiler/brw_eu_compact.c
index ac1dfadd102..332f2e2b12b 100644
--- a/src/intel/compiler/brw_eu_compact.c
+++ b/src/intel/compiler/brw_eu_compact.c
@@ -2250,7 +2250,8 @@ update_uip_jip(const struct brw_isa_info *isa, brw_eu_inst *insn,
    brw_eu_inst_set_jip(devinfo, insn, (uint32_t)jip_compacted << shift);
 
    if (brw_eu_inst_opcode(isa, insn) == BRW_OPCODE_ENDIF ||
-       brw_eu_inst_opcode(isa, insn) == BRW_OPCODE_WHILE)
+       brw_eu_inst_opcode(isa, insn) == BRW_OPCODE_WHILE ||
+       brw_eu_inst_opcode(isa, insn) == BRW_OPCODE_JOIN)
       return;
 
    int32_t uip_compacted = brw_eu_inst_uip(devinfo, insn) >> shift;
@@ -2417,6 +2418,8 @@ brw_compact_instructions(struct brw_codegen *p, int start_offset,
       case BRW_OPCODE_ELSE:
       case BRW_OPCODE_ENDIF:
       case BRW_OPCODE_WHILE:
+      case BRW_OPCODE_GOTO:
+      case BRW_OPCODE_JOIN:
          if (brw_eu_inst_cmpt_control(devinfo, insn)) {
             brw_eu_inst uncompacted;
             uncompact_instruction(&c, &uncompacted,
diff --git a/src/intel/compiler/brw_eu_defines.h b/src/intel/compiler/brw_eu_defines.h
index 7149f75497d..f1f2bc92a0b 100644
--- a/src/intel/compiler/brw_eu_defines.h
+++ b/src/intel/compiler/brw_eu_defines.h
@@ -193,6 +193,7 @@ enum opcode {
    BRW_OPCODE_CALL,
    BRW_OPCODE_RET,
    BRW_OPCODE_GOTO,
+   BRW_OPCODE_JOIN,
    BRW_OPCODE_WAIT,
    BRW_OPCODE_SEND,
    BRW_OPCODE_SENDC,
diff --git a/src/intel/compiler/brw_eu_validate.c b/src/intel/compiler/brw_eu_validate.c
index 2f4385d7468..6a3f48928b1 100644
--- a/src/intel/compiler/brw_eu_validate.c
+++ b/src/intel/compiler/brw_eu_validate.c
@@ -2500,6 +2500,7 @@ brw_hw_decode_inst(const struct brw_isa_info *isa,
    case BRW_OPCODE_CALLA:
    case BRW_OPCODE_CALL:
    case BRW_OPCODE_GOTO:
+   case BRW_OPCODE_JOIN:
       inst->format = FORMAT_BRANCH;
       break;
 
diff --git a/src/intel/compiler/brw_gram.y b/src/intel/compiler/brw_gram.y
index bfa30c58310..5ced965dffa 100644
--- a/src/intel/compiler/brw_gram.y
+++ b/src/intel/compiler/brw_gram.y
@@ -664,6 +664,7 @@ relocatableinstruction:
    | branchinstruction
    | breakinstruction
    | loopinstruction
+   | joininstruction
    ;
 
 illegalinstruction:
@@ -1138,6 +1139,30 @@ branchinstruction:
 
       brw_pop_insn_state(p);
    }
+   | predicate GOTO execsize JIP JUMP_LABEL UIP JUMP_LABEL instoptions
+   {
+      add_label(p, $5, INSTR_LABEL_JIP);
+      add_label(p, $7, INSTR_LABEL_UIP);
+
+      brw_next_insn(p, $2);
+      i965_asm_set_instruction_options(p, $8);
+      brw_eu_inst_set_exec_size(p->devinfo, brw_last_inst, $3);
+
+      brw_pop_insn_state(p);
+   }
+   ;
+
+joininstruction:
+   predicate JOIN execsize JIP JUMP_LABEL instoptions
+   {
+      add_label(p, $5, INSTR_LABEL_JIP);
+
+      brw_next_insn(p, $2);
+      i965_asm_set_instruction_options(p, $6);
+      brw_eu_inst_set_exec_size(p->devinfo, brw_last_inst, $3);
+
+      brw_pop_insn_state(p);
+   }
    ;
 
 /* break instruction */
diff --git a/src/intel/compiler/brw_lex.l b/src/intel/compiler/brw_lex.l
index b521f16b165..80ffd2b0f80 100644
--- a/src/intel/compiler/brw_lex.l
+++ b/src/intel/compiler/brw_lex.l
@@ -86,6 +86,7 @@ goto            { yylval.integer = BRW_OPCODE_GOTO; return GOTO; }
 halt            { yylval.integer = BRW_OPCODE_HALT; return HALT; }
 if              { yylval.integer = BRW_OPCODE_IF; return IF; }
 illegal         { yylval.integer = BRW_OPCODE_ILLEGAL; return ILLEGAL; }
+join            { yylval.integer = BRW_OPCODE_JOIN; return JOIN; }
 jmpi            { yylval.integer = BRW_OPCODE_JMPI; return JMPI; }
 line            { yylval.integer = BRW_OPCODE_LINE; return LINE; }
 lrp             { yylval.integer = BRW_OPCODE_LRP; return LRP; }