RISC-V parser checks

4 files changed, 49 insertions, 19 deletions

diff --git a/risc-v/hgen/lexer.hgen b/risc-v/hgen/lexer.hgen
index 27df99f4..e42b8a62 100644
--- a/risc-v/hgen/lexer.hgen
+++ b/risc-v/hgen/lexer.hgen
@@ -184,3 +184,7 @@
 "amomin.d.aq.rl",   AMO {width=RISCVDOUBLE; aq=true; rl=true; op=RISCVAMOMIN};
 "amomaxu.d.aq.rl",  AMO {width=RISCVDOUBLE; aq=true; rl=true; op=RISCVAMOMAXU};
 "amominu.d.aq.rl",  AMO {width=RISCVDOUBLE; aq=true; rl=true; op=RISCVAMOMINU};
+
+(** pseudo instructions *********************************************)
+
+"li", LI ()
diff --git a/risc-v/hgen/parser.hgen b/risc-v/hgen/parser.hgen
index cf0ca80b..210e38fb 100644
--- a/risc-v/hgen/parser.hgen
+++ b/risc-v/hgen/parser.hgen
@@ -1,25 +1,41 @@
 | UTYPE reg COMMA NUM
-  { `RISCVUTYPE($4, $2, $1.op) }
+  { (* it's not clear if NUM here should be before or after filling the
+      lowest 12 bits with zeros, or if it should be signed or unsigned;
+      currently assuming: NUM does not include the 12 zeros, and is unsigned *)
+    if not (iskbituimm 20 $4) then failwith "immediate is not 20bit"
+    else `RISCVUTYPE ($4, $2, $1.op) }
 | JAL reg COMMA NUM
-  { `RISCVJAL($4, $2) }
+  { if not ($4 mod 2 = 0) then failwith "odd offset"
+    else if not (iskbitsimm 21 $4) then failwith "offset is not 21bit"
+    else `RISCVJAL ($4, $2) }
 | JALR reg COMMA reg COMMA NUM
-  { `RISCVJALR($6, $4, $2) }
+  { if not (iskbitsimm 12 $6) then failwith "offset is not 12bit"
+    else `RISCVJALR ($6, $4, $2) }
 | BTYPE reg COMMA reg COMMA NUM
-  { `RISCVBType($6, $4, $2, $1.op) }
+  { if not ($6 mod 2 = 0) then failwith "odd offset"
+    else if not (iskbitsimm 13 $6) then failwith "offset is not 13bit"
+    else `RISCVBType ($6, $4, $2, $1.op) }
 | ITYPE reg COMMA reg COMMA NUM
-  { `RISCVIType($6, $4, $2, $1.op) }
+  { if $1.op <> RISCVSLTIU && not (iskbitsimm 12 $6) then failwith "immediate is not 12bit"
+    else if $1.op = RISCVSLTIU && not (iskbituimm 12 $6) then failwith "unsigned immediate is not 12bit"
+    else `RISCVIType ($6, $4, $2, $1.op) }
+| ADDIW reg COMMA reg COMMA NUM
+  { if not (iskbitsimm 12 $6) then failwith "immediate is not 12bit"
+    else `RISCVADDIW ($6, $4, $2) }
 | SHIFTIOP reg COMMA reg COMMA NUM
-  { `RISCVShiftIop($6, $4, $2, $1.op) }
+  { if not (iskbituimm 6 $6) then failwith "unsigned immediate is not 6bit"
+    else `RISCVShiftIop ($6, $4, $2, $1.op) }
+| SHIFTW reg COMMA reg COMMA NUM
+  { if not (iskbituimm 5 $6) then failwith "unsigned immediate is not 5bit"
+    else `RISCVSHIFTW ($6, $4, $2, $1.op) }
 | RTYPE reg COMMA reg COMMA reg
   { `RISCVRType ($6, $4, $2, $1.op) }
 | LOAD reg COMMA NUM LPAR reg RPAR
-  { `RISCVLoad($4, $6, $2, $1.unsigned, $1.width, $1.aq, $1.rl) }
+  { if not (iskbitsimm 12 $4) then failwith "offset is not 12bit"
+    else `RISCVLoad ($4, $6, $2, $1.unsigned, $1.width, $1.aq, $1.rl) }
 | STORE reg COMMA NUM LPAR reg RPAR
-  { `RISCVStore($4, $2, $6, $1.width, $1.aq, $1.rl) }
-| ADDIW reg COMMA reg COMMA NUM
-  { `RISCVADDIW ($6, $4, $2) }
-| SHIFTW reg COMMA reg COMMA NUM
-  { `RISCVSHIFTW ($6, $4, $2, $1.op) }
+  { if not (iskbitsimm 12 $4) then failwith "offset is not 12bit"
+    else `RISCVStore ($4, $2, $6, $1.width, $1.aq, $1.rl) }
 | RTYPEW reg COMMA reg COMMA reg
   { `RISCVRTYPEW ($6, $4, $2, $1.op) }
 | FENCE FENCEOPTION COMMA FENCEOPTION
@@ -37,17 +53,22 @@
 | FENCEI
   { `RISCVFENCEI }
 | LOADRES reg COMMA LPAR reg RPAR
-  { `RISCVLoadRes($1.aq, $1.rl, $5, $1.width, $2) }
+  { `RISCVLoadRes ($1.aq, $1.rl, $5, $1.width, $2) }
 | LOADRES reg COMMA NUM LPAR reg RPAR
   { if $4 <> 0 then failwith "'lr' offset must be 0" else
-    `RISCVLoadRes($1.aq, $1.rl, $6, $1.width, $2) }
+    `RISCVLoadRes ($1.aq, $1.rl, $6, $1.width, $2) }
 | STORECON reg COMMA reg COMMA LPAR reg RPAR
-  { `RISCVStoreCon($1.aq, $1.rl, $4, $7, $1.width, $2) }
+  { `RISCVStoreCon ($1.aq, $1.rl, $4, $7, $1.width, $2) }
 | STORECON reg COMMA reg COMMA NUM LPAR reg RPAR
   { if $6 <> 0 then failwith "'sc' offset must be 0" else
-    `RISCVStoreCon($1.aq, $1.rl, $4, $8, $1.width, $2) }
+    `RISCVStoreCon ($1.aq, $1.rl, $4, $8, $1.width, $2) }
 | AMO reg COMMA reg COMMA LPAR reg RPAR
-  { `RISCVAMO($1.op, $1.aq, $1.rl, $4, $7, $1.width, $2) }
+  { `RISCVAMO ($1.op, $1.aq, $1.rl, $4, $7, $1.width, $2) }
 | AMO reg COMMA reg COMMA NUM LPAR reg RPAR
   { if $6 <> 0 then failwith "'amo<op>' offset must be 0" else
-    `RISCVAMO($1.op, $1.aq, $1.rl, $4, $8, $1.width, $2) }
+    `RISCVAMO ($1.op, $1.aq, $1.rl, $4, $8, $1.width, $2) }
+
+/* pseudo-ops */
+| LI reg COMMA NUM
+  { if not (iskbitsimm 12 $4) then failwith "immediate is not 12bit (li is currently implemented only with small immediate)"
+    else `RISCVIType ($4, IReg R0, $2, RISCVORI) }
diff --git a/risc-v/hgen/token_types.hgen b/risc-v/hgen/token_types.hgen
index d338d865..f29e318d 100644
--- a/risc-v/hgen/token_types.hgen
+++ b/risc-v/hgen/token_types.hgen
@@ -17,3 +17,7 @@ type token_StoreCon = {width : wordWidth; aq: bool; rl: bool }
 type token_AMO      = {width : wordWidth; aq: bool; rl: bool; op: riscvAmoop }
 
 type token_FENCEOPTION = Fence_R | Fence_W | Fence_RW
+
+(* pseudo-ops *)
+
+type token_LI = unit
diff --git a/risc-v/hgen/tokens.hgen b/risc-v/hgen/tokens.hgen
index b0cf1d88..f812adbd 100644
--- a/risc-v/hgen/tokens.hgen
+++ b/risc-v/hgen/tokens.hgen
@@ -15,4 +15,5 @@
 %token <RISCVHGenBase.token_FENCEI>   FENCEI
 %token <RISCVHGenBase.token_LoadRes>  LOADRES
 %token <RISCVHGenBase.token_StoreCon> STORECON
-%token <RISCVHGenBase.token_AMO> AMO
+%token <RISCVHGenBase.token_AMO>      AMO
+%token <RISCVHGenBase.token_LI>       LI