motiejus/zig
1
Fork 0
Code Releases Activity

Merge remote-tracking branch 'origin/master' into llvm9

Browse Source
This commit is contained in:
Andrew Kelley
2019-09-19 17:02:32 -04:00
parent 5e34fb3597 925ffbce7f
commit 8a30edcde8
23 changed files with 1201 additions and 181 deletions
Download Patch File Download Diff File Expand all files Collapse all files

151
src/codegen.cpp
View File

@@ -4505,7 +4505,11 @@ static LLVMValueRef ir_render_optional_unwrap_ptr(CodeGen *g, IrExecutable *exec
}
}
static LLVMValueRef get_int_builtin_fn(CodeGen *g, ZigType *int_type, BuiltinFnId fn_id) {
static LLVMValueRef get_int_builtin_fn(CodeGen *g, ZigType *expr_type, BuiltinFnId fn_id) {
bool is_vector = expr_type->id == ZigTypeIdVector;
ZigType *int_type = is_vector ? expr_type->data.vector.elem_type : expr_type;
assert(int_type->id == ZigTypeIdInt);
uint32_t vector_len = is_vector ? expr_type->data.vector.len : 0;
ZigLLVMFnKey key = {};
const char *fn_name;
uint32_t n_args;
@@ -4529,6 +4533,7 @@ static LLVMValueRef get_int_builtin_fn(CodeGen *g, ZigType *int_type, BuiltinFnI
n_args = 1;
key.id = ZigLLVMFnIdBswap;
key.data.bswap.bit_count = (uint32_t)int_type->data.integral.bit_count;
key.data.bswap.vector_len = vector_len;
} else if (fn_id == BuiltinFnIdBitReverse) {
fn_name = "bitreverse";
n_args = 1;
@@ -4543,12 +4548,15 @@ static LLVMValueRef get_int_builtin_fn(CodeGen *g, ZigType *int_type, BuiltinFnI
return existing_entry->value;
char llvm_name[64];
sprintf(llvm_name, "llvm.%s.i%" PRIu32, fn_name, int_type->data.integral.bit_count);
if (is_vector)
sprintf(llvm_name, "llvm.%s.v%" PRIu32 "i%" PRIu32, fn_name, vector_len, int_type->data.integral.bit_count);
else
sprintf(llvm_name, "llvm.%s.i%" PRIu32, fn_name, int_type->data.integral.bit_count);
LLVMTypeRef param_types[] = {
get_llvm_type(g, int_type),
get_llvm_type(g, expr_type),
LLVMInt1Type(),
};
LLVMTypeRef fn_type = LLVMFunctionType(get_llvm_type(g, int_type), param_types, n_args, false);
LLVMTypeRef fn_type = LLVMFunctionType(get_llvm_type(g, expr_type), param_types, n_args, false);
LLVMValueRef fn_val = LLVMAddFunction(g->module, llvm_name, fn_type);
assert(LLVMGetIntrinsicID(fn_val));
@@ -4581,6 +4589,48 @@ static LLVMValueRef ir_render_ctz(CodeGen *g, IrExecutable *executable, IrInstru
return gen_widen_or_shorten(g, false, int_type, instruction->base.value.type, wrong_size_int);
}
static LLVMValueRef ir_render_shuffle_vector(CodeGen *g, IrExecutable *executable, IrInstructionShuffleVector *instruction) {
uint64_t len_a = instruction->a->value.type->data.vector.len;
uint64_t len_mask = instruction->mask->value.type->data.vector.len;
// LLVM uses integers larger than the length of the first array to
// index into the second array. This was deemed unnecessarily fragile
// when changing code, so Zig uses negative numbers to index the
// second vector. These start at -1 and go down, and are easiest to use
// with the ~ operator. Here we convert between the two formats.
IrInstruction *mask = instruction->mask;
LLVMValueRef *values = allocate<LLVMValueRef>(len_mask);
for (uint64_t i = 0; i < len_mask; i++) {
if (mask->value.data.x_array.data.s_none.elements[i].special == ConstValSpecialUndef) {
values[i] = LLVMGetUndef(LLVMInt32Type());
} else {
int32_t v = bigint_as_signed(&mask->value.data.x_array.data.s_none.elements[i].data.x_bigint);
uint32_t index_val = (v >= 0) ? (uint32_t)v : (uint32_t)~v + (uint32_t)len_a;
values[i] = LLVMConstInt(LLVMInt32Type(), index_val, false);
}
}
LLVMValueRef llvm_mask_value = LLVMConstVector(values, len_mask);
free(values);
return LLVMBuildShuffleVector(g->builder,
ir_llvm_value(g, instruction->a),
ir_llvm_value(g, instruction->b),
llvm_mask_value, "");
}
static LLVMValueRef ir_render_splat(CodeGen *g, IrExecutable *executable, IrInstructionSplatGen *instruction) {
ZigType *result_type = instruction->base.value.type;
src_assert(result_type->id == ZigTypeIdVector, instruction->base.source_node);
uint32_t len = result_type->data.vector.len;
LLVMTypeRef op_llvm_type = LLVMVectorType(get_llvm_type(g, instruction->scalar->value.type), 1);
LLVMTypeRef mask_llvm_type = LLVMVectorType(LLVMInt32Type(), len);
LLVMValueRef undef_vector = LLVMGetUndef(op_llvm_type);
LLVMValueRef op_vector = LLVMBuildInsertElement(g->builder, undef_vector,
ir_llvm_value(g, instruction->scalar), LLVMConstInt(LLVMInt32Type(), 0, false), "");
return LLVMBuildShuffleVector(g->builder, op_vector, undef_vector, LLVMConstNull(mask_llvm_type), "");
}
static LLVMValueRef ir_render_pop_count(CodeGen *g, IrExecutable *executable, IrInstructionPopCount *instruction) {
ZigType *int_type = instruction->op->value.type;
LLVMValueRef fn_val = get_int_builtin_fn(g, int_type, BuiltinFnIdPopCount);
@@ -5512,25 +5562,36 @@ static LLVMValueRef ir_render_mul_add(CodeGen *g, IrExecutable *executable, IrIn
static LLVMValueRef ir_render_bswap(CodeGen *g, IrExecutable *executable, IrInstructionBswap *instruction) {
LLVMValueRef op = ir_llvm_value(g, instruction->op);
ZigType *int_type = instruction->base.value.type;
ZigType *expr_type = instruction->base.value.type;
bool is_vector = expr_type->id == ZigTypeIdVector;
ZigType *int_type = is_vector ? expr_type->data.vector.elem_type : expr_type;
assert(int_type->id == ZigTypeIdInt);
if (int_type->data.integral.bit_count % 16 == 0) {
LLVMValueRef fn_val = get_int_builtin_fn(g, instruction->base.value.type, BuiltinFnIdBswap);
LLVMValueRef fn_val = get_int_builtin_fn(g, expr_type, BuiltinFnIdBswap);
return LLVMBuildCall(g->builder, fn_val, &op, 1, "");
}
// Not an even number of bytes, so we zext 1 byte, then bswap, shift right 1 byte, truncate
ZigType *extended_type = get_int_type(g, int_type->data.integral.is_signed,
int_type->data.integral.bit_count + 8);
LLVMValueRef shift_amt = LLVMConstInt(get_llvm_type(g, extended_type), 8, false);
if (is_vector) {
extended_type = get_vector_type(g, expr_type->data.vector.len, extended_type);
LLVMValueRef *values = allocate_nonzero<LLVMValueRef>(expr_type->data.vector.len);
for (uint32_t i = 0; i < expr_type->data.vector.len; i += 1) {
values[i] = shift_amt;
}
shift_amt = LLVMConstVector(values, expr_type->data.vector.len);
free(values);
}
// aabbcc
LLVMValueRef extended = LLVMBuildZExt(g->builder, op, get_llvm_type(g, extended_type), "");
// 00aabbcc
LLVMValueRef fn_val = get_int_builtin_fn(g, extended_type, BuiltinFnIdBswap);
LLVMValueRef swapped = LLVMBuildCall(g->builder, fn_val, &extended, 1, "");
// ccbbaa00
LLVMValueRef shifted = ZigLLVMBuildLShrExact(g->builder, swapped,
LLVMConstInt(get_llvm_type(g, extended_type), 8, false), "");
LLVMValueRef shifted = ZigLLVMBuildLShrExact(g->builder, swapped, shift_amt, "");
// 00ccbbaa
return LLVMBuildTrunc(g->builder, shifted, get_llvm_type(g, int_type), "");
return LLVMBuildTrunc(g->builder, shifted, get_llvm_type(g, expr_type), "");
}
static LLVMValueRef ir_render_bit_reverse(CodeGen *g, IrExecutable *executable, IrInstructionBitReverse *instruction) {
@@ -5549,10 +5610,29 @@ static LLVMValueRef ir_render_vector_to_array(CodeGen *g, IrExecutable *executab
assert(handle_is_ptr(array_type));
LLVMValueRef result_loc = ir_llvm_value(g, instruction->result_loc);
LLVMValueRef vector = ir_llvm_value(g, instruction->vector);
LLVMValueRef casted_ptr = LLVMBuildBitCast(g->builder, result_loc,
LLVMPointerType(get_llvm_type(g, instruction->vector->value.type), 0), "");
uint32_t alignment = get_ptr_align(g, instruction->result_loc->value.type);
gen_store_untyped(g, vector, casted_ptr, alignment, false);
ZigType *elem_type = array_type->data.array.child_type;
bool bitcast_ok = elem_type->size_in_bits == elem_type->abi_size * 8;
if (bitcast_ok) {
LLVMValueRef casted_ptr = LLVMBuildBitCast(g->builder, result_loc,
LLVMPointerType(get_llvm_type(g, instruction->vector->value.type), 0), "");
uint32_t alignment = get_ptr_align(g, instruction->result_loc->value.type);
gen_store_untyped(g, vector, casted_ptr, alignment, false);
} else {
// If the ABI size of the element type is not evenly divisible by size_in_bits, a simple bitcast
// will not work, and we fall back to extractelement.
LLVMTypeRef usize_type_ref = g->builtin_types.entry_usize->llvm_type;
LLVMTypeRef u32_type_ref = LLVMInt32Type();
LLVMValueRef zero = LLVMConstInt(usize_type_ref, 0, false);
for (uintptr_t i = 0; i < instruction->vector->value.type->data.vector.len; i++) {
LLVMValueRef index_usize = LLVMConstInt(usize_type_ref, i, false);
LLVMValueRef index_u32 = LLVMConstInt(u32_type_ref, i, false);
LLVMValueRef indexes[] = { zero, index_usize };
LLVMValueRef elem_ptr = LLVMBuildInBoundsGEP(g->builder, result_loc, indexes, 2, "");
LLVMValueRef elem = LLVMBuildExtractElement(g->builder, vector, index_u32, "");
LLVMBuildStore(g->builder, elem, elem_ptr);
}
}
return result_loc;
}
@@ -5563,12 +5643,34 @@ static LLVMValueRef ir_render_array_to_vector(CodeGen *g, IrExecutable *executab
assert(vector_type->id == ZigTypeIdVector);
assert(!handle_is_ptr(vector_type));
LLVMValueRef array_ptr = ir_llvm_value(g, instruction->array);
LLVMValueRef casted_ptr = LLVMBuildBitCast(g->builder, array_ptr,
LLVMPointerType(get_llvm_type(g, vector_type), 0), "");
ZigType *array_type = instruction->array->value.type;
assert(array_type->id == ZigTypeIdArray);
uint32_t alignment = get_abi_alignment(g, array_type->data.array.child_type);
return gen_load_untyped(g, casted_ptr, alignment, false, "");
LLVMTypeRef vector_type_ref = get_llvm_type(g, vector_type);
ZigType *elem_type = vector_type->data.vector.elem_type;
bool bitcast_ok = elem_type->size_in_bits == elem_type->abi_size * 8;
if (bitcast_ok) {
LLVMValueRef casted_ptr = LLVMBuildBitCast(g->builder, array_ptr,
LLVMPointerType(vector_type_ref, 0), "");
ZigType *array_type = instruction->array->value.type;
assert(array_type->id == ZigTypeIdArray);
uint32_t alignment = get_abi_alignment(g, array_type->data.array.child_type);
return gen_load_untyped(g, casted_ptr, alignment, false, "");
} else {
// If the ABI size of the element type is not evenly divisible by size_in_bits, a simple bitcast
// will not work, and we fall back to insertelement.
LLVMTypeRef usize_type_ref = g->builtin_types.entry_usize->llvm_type;
LLVMTypeRef u32_type_ref = LLVMInt32Type();
LLVMValueRef zero = LLVMConstInt(usize_type_ref, 0, false);
LLVMValueRef vector = LLVMGetUndef(vector_type_ref);
for (uintptr_t i = 0; i < instruction->base.value.type->data.vector.len; i++) {
LLVMValueRef index_usize = LLVMConstInt(usize_type_ref, i, false);
LLVMValueRef index_u32 = LLVMConstInt(u32_type_ref, i, false);
LLVMValueRef indexes[] = { zero, index_usize };
LLVMValueRef elem_ptr = LLVMBuildInBoundsGEP(g->builder, array_ptr, indexes, 2, "");
LLVMValueRef elem = LLVMBuildLoad(g->builder, elem_ptr, "");
vector = LLVMBuildInsertElement(g->builder, vector, elem, index_u32, "");
}
return vector;
}
}
static LLVMValueRef ir_render_assert_zero(CodeGen *g, IrExecutable *executable,
@@ -5896,6 +5998,7 @@ static LLVMValueRef ir_render_instruction(CodeGen *g, IrExecutable *executable,
case IrInstructionIdFrameSizeSrc:
case IrInstructionIdAllocaGen:
case IrInstructionIdAwaitSrc:
case IrInstructionIdSplatSrc:
zig_unreachable();
case IrInstructionIdDeclVarGen:
@@ -6054,6 +6157,10 @@ static LLVMValueRef ir_render_instruction(CodeGen *g, IrExecutable *executable,
return ir_render_spill_begin(g, executable, (IrInstructionSpillBegin *)instruction);
case IrInstructionIdSpillEnd:
return ir_render_spill_end(g, executable, (IrInstructionSpillEnd *)instruction);
case IrInstructionIdShuffleVector:
return ir_render_shuffle_vector(g, executable, (IrInstructionShuffleVector *) instruction);
case IrInstructionIdSplatGen:
return ir_render_splat(g, executable, (IrInstructionSplatGen *) instruction);
}
zig_unreachable();
}
@@ -7419,7 +7526,9 @@ static void do_code_gen(CodeGen *g) {
}
char *error = nullptr;
LLVMVerifyModule(g->module, LLVMAbortProcessAction, &error);
if (LLVMVerifyModule(g->module, LLVMReturnStatusAction, &error)) {
zig_panic("broken LLVM module found: %s", error);
}
}
static void zig_llvm_emit_output(CodeGen *g) {
@@ -7744,6 +7853,8 @@ static void define_builtin_fns(CodeGen *g) {
create_builtin_fn(g, BuiltinFnIdCompileLog, "compileLog", SIZE_MAX);
create_builtin_fn(g, BuiltinFnIdIntType, "IntType", 2); // TODO rename to Int
create_builtin_fn(g, BuiltinFnIdVectorType, "Vector", 2);
create_builtin_fn(g, BuiltinFnIdShuffle, "shuffle", 4);
create_builtin_fn(g, BuiltinFnIdSplat, "splat", 2);
create_builtin_fn(g, BuiltinFnIdSetCold, "setCold", 1);
create_builtin_fn(g, BuiltinFnIdSetRuntimeSafety, "setRuntimeSafety", 1);
create_builtin_fn(g, BuiltinFnIdSetFloatMode, "setFloatMode", 1);