Support #pragma once and refine codebase

Previously in PR #158, #ifndef is not sufficient enough to address
optional file inclusion issue in shecc, since shecc currently includes
file before tokenization / parsing pass. Therefore, "#pragma once" is
the only way to prevent recursively file inclusion.

Although "#pragma once" is not standard anywhere in C standards, major
C compilers such as GCC and Clang do support this preprocessor
directive.

Author: ChAoSUnItY

mk/arm.mk

@@ -14,6 +14,7 @@ export ARM_EXEC
 arm-specific-defs = \
     $(Q)$(PRINTF) \
         "/* target: ARM */\n$\
+        \#pragma once\n$\
         \#define ARCH_PREDEFINED \"__arm__\" /* defined by GNU C and RealView */\n$\
         \#define ELF_MACHINE 0x28 /* up to ARMv7/Aarch32 */\n$\
         \#define ELF_FLAGS 0x5000200\n$\

mk/riscv.mk

@@ -10,6 +10,7 @@ export RISCV_EXEC
 riscv-specific-defs = \
     $(Q)$(PRINTF) \
         "/* target: RISCV */\n$\
+        \#pragma once\n$\
         \#define ARCH_PREDEFINED \"__riscv\" /* Older versions of the GCC toolchain defined __riscv__ */\n$\
         \#define ELF_MACHINE 0xf3\n$\
         \#define ELF_FLAGS 0\n$\

src/arm-codegen.c

@@ -8,6 +8,8 @@
 /* Translate IR to target machine code */
 
 #include "arm.c"
+#include "defs.h"
+#include "globals.c"
 
 void update_elf_offset(ph2_ir_t *ph2_ir)
 {

src/arm.c

@@ -26,6 +26,8 @@
  * the current instruction.
  */
 
+#include "defs.h"
+
 /* opcode */
 typedef enum {
     arm_and = 0,

src/defs.h

@@ -1,12 +1,12 @@
+#pragma once
 /*
  * shecc - Self-Hosting and Educational C Compiler.
  *
  * shecc is freely redistributable under the BSD 2 clause license. See the
  * file "LICENSE" for information on usage and redistribution of this file.
  */
 
-#ifndef SHECC_DEFS_H
-#define SHECC_DEFS_H
+#include <stdbool.h>
 
 /* definitions */
 
@@ -40,6 +40,7 @@
 #define MAX_NESTING 128
 #define MAX_OPERAND_STACK_SIZE 32
 #define MAX_ANALYSIS_STACK_SIZE 800
+#define MAX_INCLUSIONS 16
 
 /* Default capacities for common data structures */
 /* Default arena size is initialized with 256 KiB */
@@ -92,6 +93,89 @@ typedef struct {
     hashmap_node_t **buckets;
 } hashmap_t;
 
+/* lexer tokens */
+typedef enum {
+    T_start, /* FIXME: it was intended to start the state machine. */
+    T_numeric,
+    T_identifier,
+    T_comma,  /* , */
+    T_string, /* null-terminated string */
+    T_char,
+    T_open_bracket,  /* ( */
+    T_close_bracket, /* ) */
+    T_open_curly,    /* { */
+    T_close_curly,   /* } */
+    T_open_square,   /* [ */
+    T_close_square,  /* ] */
+    T_asterisk,      /* '*' */
+    T_divide,        /* / */
+    T_mod,           /* % */
+    T_bit_or,        /* | */
+    T_bit_xor,       /* ^ */
+    T_bit_not,       /* ~ */
+    T_log_and,       /* && */
+    T_log_or,        /* || */
+    T_log_not,       /* ! */
+    T_lt,            /* < */
+    T_gt,            /* > */
+    T_le,            /* <= */
+    T_ge,            /* >= */
+    T_lshift,        /* << */
+    T_rshift,        /* >> */
+    T_dot,           /* . */
+    T_arrow,         /* -> */
+    T_plus,          /* + */
+    T_minus,         /* - */
+    T_minuseq,       /* -= */
+    T_pluseq,        /* += */
+    T_asteriskeq,    /* *= */
+    T_divideeq,      /* /= */
+    T_modeq,         /* %= */
+    T_lshifteq,      /* <<= */
+    T_rshifteq,      /* >>= */
+    T_xoreq,         /* ^= */
+    T_oreq,          /* |= */
+    T_andeq,         /* &= */
+    T_eq,            /* == */
+    T_noteq,         /* != */
+    T_assign,        /* = */
+    T_increment,     /* ++ */
+    T_decrement,     /* -- */
+    T_question,      /* ? */
+    T_colon,         /* : */
+    T_semicolon,     /* ; */
+    T_eof,           /* end-of-file (EOF) */
+    T_ampersand,     /* & */
+    T_return,
+    T_if,
+    T_else,
+    T_while,
+    T_for,
+    T_do,
+    T_typedef,
+    T_enum,
+    T_struct,
+    T_sizeof,
+    T_elipsis, /* ... */
+    T_switch,
+    T_case,
+    T_break,
+    T_default,
+    T_continue,
+    /* C pre-processor directives */
+    T_cppd_include,
+    T_cppd_define,
+    T_cppd_undef,
+    T_cppd_error,
+    T_cppd_if,
+    T_cppd_elif,
+    T_cppd_else,
+    T_cppd_endif,
+    T_cppd_ifdef,
+    T_cppd_ifndef,
+    T_cppd_pragma
+} token_t;
+
 /* builtin types */
 typedef enum {
     TYPE_void = 0,
@@ -467,5 +551,3 @@ typedef struct {
     var_t *var;
     int polluted;
 } regfile_t;
-
-#endif

src/globals.c

@@ -1,3 +1,4 @@
+#pragma once
 /*
  * shecc - Self-Hosting and Educational C Compiler.
  *
@@ -8,6 +9,21 @@
 #include <stdbool.h>
 #include <stdlib.h>
 
+#include "defs.h"
+
+/* Lexer */
+char token_str[MAX_TOKEN_LEN];
+token_t next_token;
+char next_char;
+bool skip_newline = true;
+
+bool preproc_match;
+
+/* Point to the first character after where the macro has been called. It is
+ * needed when returning from the macro body.
+ */
+int macro_return_idx;
+
 /* Global objects */
 
 block_list_t BLOCKS;
@@ -58,6 +74,8 @@ int constants_idx = 0;
 
 source_t *SOURCE;
 
+hashmap_t *INCLUSION_MAP;
+
 /* ELF sections */
 
 char *elf_code;
@@ -968,6 +986,7 @@ void global_init()
     PH2_IR_FLATTEN = malloc(MAX_IR_INSTR * sizeof(ph2_ir_t *));
     LABEL_LUT = malloc(MAX_LABEL * sizeof(label_lut_t));
     SOURCE = create_source(MAX_SOURCE);
+    INCLUSION_MAP = hashmap_create(MAX_INCLUSIONS);
     ALIASES = malloc(MAX_ALIASES * sizeof(alias_t));
     CONSTANTS = malloc(MAX_CONSTANTS * sizeof(constant_t));
 
@@ -1000,6 +1019,7 @@ void global_release()
     free(PH2_IR_FLATTEN);
     free(LABEL_LUT);
     source_release(SOURCE);
+    hashmap_free(INCLUSION_MAP);
     free(ALIASES);
     free(CONSTANTS);
 

src/lexer.c

@@ -7,99 +7,8 @@
 
 #include <stdbool.h>
 
-/* lexer tokens */
-typedef enum {
-    T_start, /* FIXME: it was intended to start the state machine. */
-    T_numeric,
-    T_identifier,
-    T_comma,  /* , */
-    T_string, /* null-terminated string */
-    T_char,
-    T_open_bracket,  /* ( */
-    T_close_bracket, /* ) */
-    T_open_curly,    /* { */
-    T_close_curly,   /* } */
-    T_open_square,   /* [ */
-    T_close_square,  /* ] */
-    T_asterisk,      /* '*' */
-    T_divide,        /* / */
-    T_mod,           /* % */
-    T_bit_or,        /* | */
-    T_bit_xor,       /* ^ */
-    T_bit_not,       /* ~ */
-    T_log_and,       /* && */
-    T_log_or,        /* || */
-    T_log_not,       /* ! */
-    T_lt,            /* < */
-    T_gt,            /* > */
-    T_le,            /* <= */
-    T_ge,            /* >= */
-    T_lshift,        /* << */
-    T_rshift,        /* >> */
-    T_dot,           /* . */
-    T_arrow,         /* -> */
-    T_plus,          /* + */
-    T_minus,         /* - */
-    T_minuseq,       /* -= */
-    T_pluseq,        /* += */
-    T_asteriskeq,    /* *= */
-    T_divideeq,      /* /= */
-    T_modeq,         /* %= */
-    T_lshifteq,      /* <<= */
-    T_rshifteq,      /* >>= */
-    T_xoreq,         /* ^= */
-    T_oreq,          /* |= */
-    T_andeq,         /* &= */
-    T_eq,            /* == */
-    T_noteq,         /* != */
-    T_assign,        /* = */
-    T_increment,     /* ++ */
-    T_decrement,     /* -- */
-    T_question,      /* ? */
-    T_colon,         /* : */
-    T_semicolon,     /* ; */
-    T_eof,           /* end-of-file (EOF) */
-    T_ampersand,     /* & */
-    T_return,
-    T_if,
-    T_else,
-    T_while,
-    T_for,
-    T_do,
-    T_typedef,
-    T_enum,
-    T_struct,
-    T_sizeof,
-    T_elipsis, /* ... */
-    T_switch,
-    T_case,
-    T_break,
-    T_default,
-    T_continue,
-    /* C pre-processor directives */
-    T_cppd_include,
-    T_cppd_define,
-    T_cppd_undef,
-    T_cppd_error,
-    T_cppd_if,
-    T_cppd_elif,
-    T_cppd_else,
-    T_cppd_endif,
-    T_cppd_ifdef,
-    T_cppd_ifndef
-} token_t;
-
-char token_str[MAX_TOKEN_LEN];
-token_t next_token;
-char next_char;
-bool skip_newline = true;
-
-bool preproc_match;
-
-/* Point to the first character after where the macro has been called. It is
- * needed when returning from the macro body.
- */
-int macro_return_idx;
+#include "defs.h"
+#include "globals.c"
 
 bool is_whitespace(char c)
 {
@@ -223,6 +132,8 @@ token_t lex_token_internal(bool aliasing)
             return T_cppd_else;
         if (!strcmp(token_str, "#endif"))
             return T_cppd_endif;
+        if (!strcmp(token_str, "#pragma"))
+            return T_cppd_pragma;
         error("Unknown directive");
     }
 

src/parser.c

@@ -6,8 +6,13 @@
  */
 
 #include <stdbool.h>
+#include <stdio.h>
 #include <stdlib.h>
 
+#include "../config"
+#include "defs.h"
+#include "globals.c"
+
 /* C language syntactic analyzer */
 int global_var_idx = 0;
 int global_label_idx = 0;
@@ -513,6 +518,10 @@ bool read_preproc_directive()
         cppd_control_flow_skip_lines();
         return true;
     }
+    if (lex_accept_internal(T_cppd_pragma, false)) {
+        lex_expect(T_identifier);
+        return true;
+    }
 
     return false;
 }
@@ -3399,6 +3408,10 @@ void load_source_file(char *file)
 
     for (;;) {
         if (!fgets(buffer, MAX_LINE_LEN, f)) {
+            break;
+        }
+        if (!strncmp(buffer, "#pragma once", 12) &&
+            hashmap_contains(INCLUSION_MAP, file)) {
             fclose(f);
             return;
         }
@@ -3419,6 +3432,8 @@ void load_source_file(char *file)
             SOURCE->size += strlen(buffer);
         }
     }
+
+    hashmap_put(INCLUSION_MAP, file, NULL);
     fclose(f);
 }
 

src/peephole.c

@@ -5,6 +5,11 @@
  * file "LICENSE" for information on usage and redistribution of this file.
  */
 
+#include <stdbool.h>
+
+#include "defs.h"
+#include "globals.c"
+
 bool is_fusible_insn(ph2_ir_t *ph2_ir)
 {
     switch (ph2_ir->op) {

src/reg-alloc.c

@@ -12,6 +12,11 @@
  * dead variable and does NOT wrtie it back to the stack.
  */
 
+#include <stdbool.h>
+
+#include "defs.h"
+#include "globals.c"
+
 /* Aligns size to nearest multiple of 4, this meets
  * ARMv7's alignment requirement.
  *