From b5ddc36d7f9c6d96ad1239977293631dbb487bec Mon Sep 17 00:00:00 2001
From: Christoffer Lerno <christoffer@aegik.com>
Date: Fri, 23 May 2025 21:40:14 +0200
Subject: [PATCH] Limit vector max size, default is 4096 bits, but may be
 increased using --max-vector-size.

---
 lib/std/core/env.c3                    |  1 +
 releasenotes.md                        |  1 +
 src/build/build.h                      |  3 +++
 src/build/build_options.c              | 18 +++++++++++++++++-
 src/build/builder.c                    |  3 +++
 src/build/project.c                    | 16 +++++++++++++++-
 src/compiler/compiler.c                |  1 +
 src/compiler/sema_types.c              | 13 +++++++++++--
 src/utils/common.h                     |  1 +
 test/test_suite/vector/vector_limit.c3 |  5 +++++
 test/unit/stdlib/core/builtintests.c3  | 16 ++++++++--------
 test/unit/stdlib/os/subprocess.c3      |  3 ++-
 12 files changed, 68 insertions(+), 13 deletions(-)
 create mode 100644 test/test_suite/vector/vector_limit.c3

diff --git a/lib/std/core/env.c3 b/lib/std/core/env.c3
index 0c9f7c3a8..81217f7c0 100644
--- a/lib/std/core/env.c3
+++ b/lib/std/core/env.c3
@@ -120,6 +120,7 @@ const String COMPILER_BUILD_HASH = $$BUILD_HASH;
 const String COMPILER_BUILD_DATE = $$BUILD_DATE;
 const OsType OS_TYPE = OsType.from_ordinal($$OS_TYPE);
 const ArchType ARCH_TYPE = ArchType.from_ordinal($$ARCH_TYPE);
+const usz MAX_VECTOR_SIZE = $$MAX_VECTOR_SIZE;
 const bool ARCH_32_BIT = $$REGISTER_SIZE == 32;
 const bool ARCH_64_BIT = $$REGISTER_SIZE == 64;
 const bool LIBC = $$COMPILER_LIBC_AVAILABLE;
diff --git a/releasenotes.md b/releasenotes.md
index fe506259e..0feb304bf 100644
--- a/releasenotes.md
+++ b/releasenotes.md
@@ -14,6 +14,7 @@
 - Deprecate `MyEnum.elements`.
 - Deprecate `SomeFn.params`.
 - Improve error message when encountering recursively defined structs. #2146
+- Limit vector max size, default is 4096 bits, but may be increased using --max-vector-size.
 
 ### Fixes
 - Assert triggered when casting from `int[2]` to `uint[2]` #2115
diff --git a/src/build/build.h b/src/build/build.h
index 416e4a0dd..a3bea7866 100644
--- a/src/build/build.h
+++ b/src/build/build.h
@@ -7,6 +7,7 @@
 #include "../version.h"
 #include <stdint.h>
 
+#define MAX_SYMTAB_SIZE (1024 * 1024)
 #define MAX_BUILD_LIB_DIRS 1024
 #define MAX_COMMAND_LINE_FILES 4096
 #define MAX_COMMAND_LINE_RUN_ARGS 2048
@@ -578,6 +579,7 @@ typedef struct BuildOptions_
 	RiscvFloatCapability riscv_float_capability;
 	MemoryEnvironment memory_environment;
 	SanitizeMode sanitize_mode;
+	uint32_t max_vector_size;
 	bool print_keywords;
 	bool print_attributes;
 	bool print_builtins;
@@ -702,6 +704,7 @@ typedef struct
 	CompilerBackend backend;
 	LinkerType linker_type;
 	uint32_t symtab_size;
+	uint32_t max_vector_size;
 	uint32_t switchrange_max_size;
 	uint32_t switchjump_max_size;
 	const char **args;
diff --git a/src/build/build_options.c b/src/build/build_options.c
index f0e70dde4..502d3be82 100644
--- a/src/build/build_options.c
+++ b/src/build/build_options.c
@@ -168,6 +168,7 @@ static void usage(bool full)
 		print_opt("--fp-math=<option>", "FP math behaviour: strict, relaxed, fast.");
 		print_opt("--win64-simd=<option>", "Win64 SIMD ABI: array, full.");
 		print_opt("--win-debug=<option>", "Select debug output on Windows: codeview or dwarf (default: codeview).");
+		print_opt("--max-vector-size <number>", "Set the maximum vector bit size to allow (default: 4096).");
 		PRINTF("");
 		print_opt("--print-linking", "Print linker arguments.");
 		PRINTF("");
@@ -771,7 +772,11 @@ static void parse_option(BuildOptions *options)
 				if (at_end() || next_is_opt()) error_exit("error: --symtab needs a valid integer.");
 				const char *symtab_string = next_arg();
 				int symtab = atoi(symtab_string);
-				if (symtab < 1024) PRINTF("Expected a valid positive integer >= 1024.");
+				if (symtab < 1024) error_exit("Expected the --symtab size to be valid positive integer >= 1024.");
+				if (symtab > MAX_SYMTAB_SIZE)
+				{
+					error_exit("The symtab size is too large. The maximum size is %d.", (int)MAX_SYMTAB_SIZE);
+				}
 				options->symtab_size = next_highest_power_of_2(symtab);
 				return;
 			}
@@ -887,6 +892,17 @@ static void parse_option(BuildOptions *options)
 				options->riscv_float_capability = parse_opt_select(RiscvFloatCapability, argopt, riscv_capability);
 				return;
 			}
+			if ((argopt = match_argopt("max-vector-size")))
+			{
+				int size = atoi(next_arg());
+				if (size < 128) error_exit("Expected a valid positive integer >= 128 or --max-vector-size.");
+				if (size > MAX_VECTOR_WIDTH) error_exit("Expected a valid positive integer <= %u for --max-vector-size.", (unsigned)MAX_VECTOR_WIDTH);
+				if (size != next_highest_power_of_2(size))
+				{
+					error_exit("The --max-vector-size value must be a power of 2, try using %u instead.", next_highest_power_of_2(size));
+				}
+				return;
+			}
 			if ((argopt = match_argopt("memory-env")))
 			{
 				options->memory_environment = parse_opt_select(MemoryEnvironment, argopt, memory_environment);
diff --git a/src/build/builder.c b/src/build/builder.c
index 36f241c03..4d117645a 100644
--- a/src/build/builder.c
+++ b/src/build/builder.c
@@ -403,6 +403,7 @@ static void update_build_target_from_options(BuildTarget *target, BuildOptions *
 	OVERRIDE_IF_SET(testfn);
 	OVERRIDE_IF_SET(benchfn);
 	OVERRIDE_IF_SET(symtab_size);
+	OVERRIDE_IF_SET(max_vector_size);
 	OVERRIDE_IF_SET(win.def);
 	OVERRIDE_IF_SET(no_entry);
 
@@ -416,6 +417,8 @@ static void update_build_target_from_options(BuildTarget *target, BuildOptions *
 	OVERRIDE_IF_SET(android.ndk_path);
 	OVERRIDE_IF_SET(android.api_version);
 
+	if (!target->max_vector_size) target->max_vector_size = DEFAULT_VECTOR_WIDTH;
+
 	if (options->silence_deprecation || options->verbosity_level < 0) target->silence_deprecation = options->silence_deprecation || options->verbosity_level < 0;
 	target->print_linking = options->print_linking || options->verbosity_level > 1;
 
diff --git a/src/build/project.c b/src/build/project.c
index 0a546cc06..12ec8c907 100644
--- a/src/build/project.c
+++ b/src/build/project.c
@@ -143,7 +143,6 @@ const char* project_target_keys[][2] = {
 const int project_target_keys_count = ELEMENTLEN(project_target_keys);
 const int project_deprecated_target_keys_count = ELEMENTLEN(project_deprecated_target_keys);
 
-#define MAX_SYMTAB_SIZE (1024 * 1024)
 #define GET_SETTING(type__, key__, strings__, comment__) \
   (type__)get_valid_string_setting(context, json, key__, strings__, 0, ELEMENTLEN(strings__), comment__)
 
@@ -276,6 +275,21 @@ static void load_into_build_target(BuildParseContext context, JSONObject *json,
 		target->symtab_size = (uint32_t)symtab_size;
 	}
 
+	// Vector size
+	long vector_size = get_valid_integer(context, json, "max-vector-size", false);
+	if (vector_size > 0)
+	{
+		if (vector_size < 128)
+		{
+			error_exit("Error reading %s: max vector size was less than 128.", context.file);
+		}
+		if (vector_size > MAX_VECTOR_WIDTH)
+		{
+			error_exit("Error reading %s: max vector width may not exceed %d.", context.file, MAX_VECTOR_WIDTH);
+		}
+		target->max_vector_size = (uint32_t)vector_size;
+	}
+
 	// Target
 	const char *arch_os_string = get_optional_string(context, json, "target");
 	if (arch_os_string)
diff --git a/src/compiler/compiler.c b/src/compiler/compiler.c
index 217612a0e..984a7a0e2 100644
--- a/src/compiler/compiler.c
+++ b/src/compiler/compiler.c
@@ -1456,6 +1456,7 @@ void compile()
 	setup_int_define("C_LONG_SIZE", compiler.platform.width_c_long, type_int);
 	setup_int_define("C_LONG_LONG_SIZE", compiler.platform.width_c_long_long, type_int);
 	setup_int_define("REGISTER_SIZE", compiler.platform.width_register, type_int);
+	setup_int_define("MAX_VECTOR_SIZE", compiler.build.max_vector_size, type_int);
 	setup_bool_define("C_CHAR_IS_SIGNED", compiler.platform.signed_c_char);
 	setup_bool_define("PLATFORM_BIG_ENDIAN", compiler.platform.big_endian);
 	setup_bool_define("PLATFORM_I128_SUPPORTED", compiler.platform.int128);
diff --git a/src/compiler/sema_types.c b/src/compiler/sema_types.c
index c7bdafe68..ed79af4ec 100644
--- a/src/compiler/sema_types.c
+++ b/src/compiler/sema_types.c
@@ -81,11 +81,11 @@ bool sema_resolve_array_like_len(SemaContext *context, TypeInfo *type_info, Arra
 	}
 
 	// Check max values.
-	if (int_icomp(len, is_vector ? MAX_VECTOR_WIDTH : MAX_ARRAY_SIZE, BINARYOP_GT))
+	if (int_icomp(len, is_vector ? compiler.build.max_vector_size / 8 : MAX_ARRAY_SIZE, BINARYOP_GT))
 	{
 		if (is_vector)
 		{
-			SEMA_ERROR(len_expr, "A vector may not exceed %d in width.", MAX_VECTOR_WIDTH);
+			SEMA_ERROR(len_expr, "A vector may not exceed %d in bit width.", compiler.build.max_vector_size);
 		}
 		else
 		{
@@ -190,6 +190,15 @@ static inline bool sema_resolve_array_type(SemaContext *context, TypeInfo *type,
 	if (!sema_check_array_type(context, type, type->array.base->type, kind, len, &type->type)) return type_info_poison(type);
 	ASSERT(!type->array.len || sema_cast_const(type->array.len));
 	type->resolve_status = RESOLVE_DONE;
+	if (kind == TYPE_INFO_VECTOR)
+	{
+		if (kind == TYPE_INFO_VECTOR && type_size(type->type) > compiler.build.max_vector_size / 8)
+		{
+			RETURN_SEMA_ERROR(type, "Vectors with bitsize over %u are not supported (this vector is %llu bits), "
+						   "but you can increase the maximum allowed using '--max-vector-size'.",
+						   compiler.build.max_vector_size, (unsigned long long)type_size(type->type) * 8);
+		}
+	}
 	return true;
 
 }
diff --git a/src/utils/common.h b/src/utils/common.h
index 0e9779b12..897b3ab61 100644
--- a/src/utils/common.h
+++ b/src/utils/common.h
@@ -22,6 +22,7 @@
 
 #define NO_ARENA 0
 #define MAX_VECTOR_WIDTH 65536
+#define DEFAULT_VECTOR_WIDTH 4096
 #define MAX_ARRAY_SIZE INT64_MAX
 #define MAX_SOURCE_LOCATION_LEN 255
 #define PROJECT_JSON "project.json"
diff --git a/test/test_suite/vector/vector_limit.c3 b/test/test_suite/vector/vector_limit.c3
new file mode 100644
index 000000000..f2740db03
--- /dev/null
+++ b/test/test_suite/vector/vector_limit.c3
@@ -0,0 +1,5 @@
+fn void main()
+{
+	double[<65>] x; // #error: Vectors with bitsize over
+	$echo $sizeof(x) * 8;
+}
diff --git a/test/unit/stdlib/core/builtintests.c3 b/test/unit/stdlib/core/builtintests.c3
index 917cf80c3..7bb39ba12 100644
--- a/test/unit/stdlib/core/builtintests.c3
+++ b/test/unit/stdlib/core/builtintests.c3
@@ -81,16 +81,16 @@ fn void test_hash()
 	(uint[<100>]){}.hash();
 	(uint[100]){}.hash();
 	(long){}.hash();
-	(long[<100>]){}.hash();
+	(long[<20>]){}.hash();
 	(long[100]){}.hash();
 	(ulong){}.hash();
-	(ulong[<100>]){}.hash();
+	(ulong[<20>]){}.hash();
 	(ulong[100]){}.hash();
 	(int128){}.hash();
-	(int128[<100>]){}.hash();
+	(int128[<20>]){}.hash();
 	(int128[100]){}.hash();
 	(uint128){}.hash();
-	(uint128[<100>]){}.hash();
+	(uint128[<20>]){}.hash();
 	(uint128[100]){}.hash();
 	(bool){}.hash();
 	(bool[<100>]){}.hash();
@@ -122,16 +122,16 @@ fn void test_hash_repeat()
 	assert((uint[<100>]){}.hash() == (uint[<100>]){}.hash());
 	assert((uint[100]){}.hash() == (uint[100]){}.hash());
 	assert((long){}.hash() == (long){}.hash());
-	assert((long[<100>]){}.hash() == (long[<100>]){}.hash());
+	assert((long[<20>]){}.hash() == (long[<20>]){}.hash());
 	assert((long[100]){}.hash() == (long[100]){}.hash());
 	assert((ulong){}.hash() == (ulong){}.hash());
-	assert((ulong[<100>]){}.hash() == (ulong[<100>]){}.hash());
+	assert((ulong[<20>]){}.hash() == (ulong[<20>]){}.hash());
 	assert((ulong[100]){}.hash() == (ulong[100]){}.hash());
 	assert((int128){}.hash() == (int128){}.hash());
-	assert((int128[<100>]){}.hash() == (int128[<100>]){}.hash());
+	assert((int128[<20>]){}.hash() == (int128[<20>]){}.hash());
 	assert((int128[100]){}.hash() == (int128[100]){}.hash());
 	assert((uint128){}.hash() == (uint128){}.hash());
-	assert((uint128[<100>]){}.hash() == (uint128[<100>]){}.hash());
+	assert((uint128[<20>]){}.hash() == (uint128[<20>]){}.hash());
 	assert((uint128[100]){}.hash() == (uint128[100]){}.hash());
 	assert((bool){}.hash() == (bool){}.hash());
 	assert((bool[<100>]){}.hash() == (bool[<100>]){}.hash());
diff --git a/test/unit/stdlib/os/subprocess.c3 b/test/unit/stdlib/os/subprocess.c3
index 59047d87b..3265a3f6d 100644
--- a/test/unit/stdlib/os/subprocess.c3
+++ b/test/unit/stdlib/os/subprocess.c3
@@ -1,4 +1,5 @@
-module std::os::process @test;
+// Currently disabled as they somewhat mess up the test output.
+module std::os::process @test @if(false);
 import std::os, std::io;
 
 fn void test_inherit_stdio_stdout()