From 209529e1829af1efe5cee0a64264e53c2ae8656c Mon Sep 17 00:00:00 2001
From: Lucas Clemente Vella <lvella@powdrlabs.com>
Date: Tue, 19 Sep 2023 20:46:19 +0100
Subject: [PATCH] Parsing .insn assembly directive.

---
 riscv/build.rs              |  5 ++-
 riscv/src/lib.rs            | 71 +++++++++++++++++++++++++++++++++++++
 riscv/src/riscv_asm.lalrpop | 10 +++++-
 3 files changed, 84 insertions(+), 2 deletions(-)

diff --git a/riscv/build.rs b/riscv/build.rs
index 0374a153e..c5977c1e9 100644
--- a/riscv/build.rs
+++ b/riscv/build.rs
@@ -12,7 +12,10 @@ fn main() {
 }
 
 fn build_lalrpop() {
-    lalrpop::process_root().unwrap();
+    lalrpop::Configuration::new()
+        .emit_rerun_directives(true)
+        .process_current_dir()
+        .unwrap();
 }
 
 fn build_instruction_tests() {
diff --git a/riscv/src/lib.rs b/riscv/src/lib.rs
index 31797c0bf..72be3acba 100644
--- a/riscv/src/lib.rs
+++ b/riscv/src/lib.rs
@@ -329,3 +329,74 @@ fn output_files_from_cargo_build_plan(
 
     assemblies
 }
+
+/// Maps an instruction in .insn syntax to Statement::Instruction() in the expected format.
+///
+/// See https://www.rowleydownload.co.uk/arm/documentation/gnu/as/RISC_002dV_002dFormats.html
+pub fn map_insn_i(
+    opcode6: Expression,
+    func3: Expression,
+    rd: Register,
+    rs1: Register,
+    simm12: Expression,
+) -> Statement {
+    let (Expression::Number(opcode6), Expression::Number(func3)) = (opcode6, func3) else {
+        panic!("Only literal opcode and function are supported in .insn syntax");
+    };
+
+    // These are almost all instructions in RISC-V Instruction Set Manual that
+    // we are supposed to implement and roughly fits the pattern of the I-type
+    // instruction. Only "csr*i" instructions are missing.
+
+    // First we try to match the instructions that uses the I-type encoding
+    // ordinarily, i.e. where all fields are what they are supposed to be:
+    let name = match (opcode6, func3) {
+        (0b1100111, 0b000) => "jalr",
+        (0b0000011, 0b000) => "lb",
+        (0b0000011, 0b001) => "lh",
+        (0b0000011, 0b010) => "lw",
+        (0b0000011, 0b100) => "lbu",
+        (0b0000011, 0b101) => "lhu",
+        (0b0010011, 0b000) => "addi",
+        (0b0010011, 0b010) => "slti",
+        (0b0010011, 0b011) => "sltiu",
+        (0b0010011, 0b100) => "xori",
+        (0b0010011, 0b110) => "ori",
+        (0b0010011, 0b111) => "andi",
+        (0b1110011, 0b001) => "csrrw",
+        (0b1110011, 0b010) => "csrrs",
+        (0b1110011, 0b011) => "csrrc",
+        // won't interpret "csr*i" instructions because it is too weird to
+        // encode an immediate as a register
+        opfunc => {
+            // We now try the instructions that take certain liberties with the
+            // I-type encoding, and don't use the standard arguments for it.
+            let name = match opfunc {
+                (0b0001111, 0b000) => "fence",
+                (0b0001111, 0b001) => "fence.i",
+                (0b1110011, 0b000) => {
+                    let Expression::Number(simm12) = simm12 else {
+                        panic!(
+                            "Only literal simm12 is supported for ecall and ebreak instructions"
+                        );
+                    };
+                    match simm12 {
+                        0 => "ecall",
+                        1 => "ebreak",
+                        _ => panic!("unknown instruction"),
+                    }
+                }
+                _ => panic!("unsupported .insn instruction"),
+            };
+            return Statement::Instruction(name.to_string(), Vec::new());
+        }
+    };
+
+    let args = vec![
+        Argument::Register(rd),
+        Argument::Register(rs1),
+        Argument::Expression(simm12),
+    ];
+
+    Statement::Instruction(name.to_string(), args)
+}
diff --git a/riscv/src/riscv_asm.lalrpop b/riscv/src/riscv_asm.lalrpop
index 34d15e58d..0ab997521 100644
--- a/riscv/src/riscv_asm.lalrpop
+++ b/riscv/src/riscv_asm.lalrpop
@@ -13,7 +13,7 @@
 use std::str::FromStr;
 use asm_utils::ast::{unescape_string, BinaryOpKind as BOp, UnaryOpKind as UOp,
     new_binary_op as bin_op, new_unary_op as un_op, new_function_op as fn_op};
-use crate::{Argument, Register, Statement, FunctionKind as FOp, Expression};
+use crate::{Argument, Register, Statement, FunctionKind as FOp, Expression, map_insn_i};
 
 grammar;
 
@@ -67,9 +67,17 @@ Directive: Statement = {
 }
 
 Instruction: Statement = {
+    <InsnDirective>,
     <DotlessSymbol> <Arguments> => Statement::Instruction(<>)
 }
 
+InsnDirective: Statement = {
+    ".insn" "i" <opcode6:Expression> "," <func3:Expression> "," <rd:Register> "," <rs1:Register> "," <simm12:Expression> => map_insn_i(opcode6, func3, rd, rs1, simm12),
+    ".insn" "i" <opcode6:Expression> "," <func3:Expression> "," <rd:Register> "," <simm12:Expression> "(" <rs1:Register> ")" => map_insn_i(opcode6, func3, rd, rs1, simm12)
+    // TODO: implement the other kinds of .insn instructions.
+    // See https://www.rowleydownload.co.uk/arm/documentation/gnu/as/RISC_002dV_002dFormats.html
+}
+
 Arguments: Vec<Argument> = {
     => vec![],
     <mut list:( <Argument> "," )*> <end:Argument>  => { list.push(end); list }