Initial

3 files changed, 332 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..261e785
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+bfcomp
+*.o
+*.dSYM
+dst
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..73e948a
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,16 @@
+CC := gcc
+CFLAGS := -Wall -Wextra -std=c11 -O2 -fwrapv
+BIN := bfcomp
+
+.PHONY: all clean remake
+
+all: $(BIN)
+
+clean:
+	rm -rf $(BIN) *.dSYM
+
+remake: clean all
+
+
+$(BIN): $(wildcard *.c *.h)
+	$(CC) $(CFLAGS) -o $@ $(filter-out %.h,$^)
diff --git a/bfcomp.c b/bfcomp.c
new file mode 100644
index 0000000..9de8cf2
--- /dev/null
+++ b/bfcomp.c
@@ -0,0 +1,312 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <string.h>
+#include <unistd.h>
+#include <time.h>
+#include <assert.h>
+
+#define DBGTIME(tag,block) \
+	do { \
+		clock_t start; \
+		if(params.verbose){ \
+			start=clock(); \
+			fprintf(stderr,"%s",(tag)); \
+		} \
+		{block} \
+		if(params.verbose){ \
+			fprintf(stderr,"  (%lfs)\n",(double)(clock()-start)/CLOCKS_PER_SEC); \
+		} \
+	} while(0)
+
+
+int uniqid(void){
+	static int i=0;
+	return i++;
+}
+
+void usage(const char *argv0){
+	fprintf(stderr,
+		"Usage: %s [-h] [-m <memsize>] <srcfile> <dstfile>\n"
+		"Reads a Brainfuck program from <srcfile> and saves the compiled executable\n"
+		"in <dstfile>.\n"
+		"Uses nasm for assembling and gcc for linking.\n"
+		"  -h             Show help\n"
+		"  -m <memsize>   Specify amount of tape cells (bytes) for the BF code\n",
+		argv0);
+}
+
+__attribute__((noreturn)) void usage1(const char *argv0){
+	usage(argv0);
+	exit(1);
+}
+
+typedef struct Params{
+	int memsize;
+	const char *srcfname,*dstfname;
+	bool verbose;
+} Params;
+Params params={30000,NULL,NULL,false};
+
+void parseargs(int argc,char **argv){
+	if(argc<2)usage1(argv[0]);
+
+	for(int i=1;i<argc;i++){
+		bool skipnext=false;
+		if(argv[i][0]=='-'){
+			for(int j=1;argv[i][j];j++){
+				switch(argv[i][j]){
+					case 'h':
+						usage(argv[0]);
+						exit(0);
+
+					case 'm':{
+						if(i==argc-1)usage1(argv[0]);
+						char *endp;
+						params.memsize=strtol(argv[i+1],&endp,10);
+						if(!argv[i+1][0]||*endp||params.memsize<1)usage1(argv[0]);
+						skipnext=true;
+						break;
+					}
+
+					case 'v':
+						params.verbose=true;
+						break;
+				}
+			}
+		} else if(params.srcfname==NULL)params.srcfname=argv[i];
+		else if(params.dstfname==NULL)params.dstfname=argv[i];
+		else usage1(argv[0]);
+		if(skipnext)i++;
+	}
+
+	if(!params.srcfname||!params.dstfname)usage1(argv[0]);
+}
+
+int runcmd(const char *cmd/*,int in,int out*/){
+	pid_t pid=fork();
+	assert(pid>=0);
+	if(pid==0){
+		/*dup2(in,0);
+		dup2(out,1);*/
+		execl("/bin/sh","/bin/sh","-c",cmd,NULL);
+		__builtin_unreachable();
+	} else {
+		int statloc;
+		DBGTIME(cmd,{
+			while(true){
+				waitpid(pid,&statloc,0);
+				if(WIFEXITED(statloc))break;
+			}
+		});
+		return WEXITSTATUS(statloc);
+	}
+}
+
+FILE* gettempfilew(char name[20]){
+	memcpy(name,"/tmp/tmp.XXXXXXXXXX",20);
+	int fd=mkstemp(name);
+	assert(fd>=0);
+	FILE *f=fdopen(fd,"w");
+	assert(f);
+	return f;
+}
+
+int compilechain(const char *asmfname,const char *dstfname){
+	char ofname[20];
+	FILE *f=gettempfilew(ofname);
+	if(params.verbose)fprintf(stderr,"o -> %s\n",ofname);
+	assert(strlen(ofname)==19);
+	fclose(f);
+	char buf[25+19+1+strlen(asmfname)+1];
+	sprintf(buf,"nasm -f macho64 -w+all -o %s %s",ofname,asmfname);
+	int ret=runcmd(buf);
+	if(ret)return ret;
+	char buf2[4+19+4+strlen(dstfname)+1];
+	sprintf(buf2,"gcc %s -o %s",ofname,dstfname);
+	ret=runcmd(buf2);
+	unlink(ofname);
+	if(params.verbose)fprintf(stderr,"unlink(%s)  (ofile)\n",ofname);
+	return ret;
+}
+
+void writeprologue(FILE *asmf){
+	fprintf(asmf,
+		"global _main\n"
+		"\n"
+		"extern _malloc\n"
+		"extern _free\n"
+		"extern _printf\n"
+		"extern _putchar\n"
+		"extern _write\n"
+		"\n"
+		"default rel\n"
+		"\n"
+		"section .text\n"
+		"\n"
+		"_main:\n"
+		"\tpush rbp  ; prologue\n"
+		"\tmov rbp, rsp\n"
+		"\tadd rsp, 8\n"
+		"\tpush rbx\n"
+		"\tmov edi, %d\n"
+		"\tcall _malloc\n"
+		"\tcmp rax, 0\n"
+		"\tjz nomem\n"
+		"\tmov rbx, rax ;bufp\n"
+		"\tmov [buf], rax ;buf\n"
+		"\n",
+		params.memsize);
+}
+
+void writeprogram(FILE *srcf,FILE *asmf){
+	int addc=0,shiftc=0;
+	//int lastshiftc=0;
+	struct stackitem {int id; struct stackitem *next;} *loopstack=NULL;
+	bool eof=false;
+	while(!eof){
+		char c=fgetc(srcf);
+		if(feof(srcf))eof=1;
+		if(!eof&&strchr("+-><[].,",c)==NULL)continue;
+		if(addc&&c!='+'&&c!='-'){
+			addc=(addc%256)+256%256;
+			if(addc==1)fprintf(asmf,"\tinc byte [rbx]  ; 1 +\n");
+			else if(addc==-1)fprintf(asmf,"\tdec byte [rbx]  ; 1 -\n");
+			else fprintf(asmf,"\tadd byte [rbx], %d  ; %d %c\n",addc,addc<0?-addc:addc,addc<0?'-':'+');
+			addc=0;
+		}
+		if(shiftc&&c!='>'&&c!='<'){
+			shiftc=(shiftc%256)+256%256;
+			/*if(lastshiftc>0&&shiftc<0){
+				int id=uniqid();
+				fprintf(asmf,
+					"\tcmp [maxp], rbx\n"
+					"\tjge skip%d\n"
+					"\tmov [maxp], rbx\n"
+					"skip%d:\n",
+					id,id);
+			}
+			lastshiftc=shiftc;*/
+			if(shiftc==1)fprintf(asmf,"\tinc rbx  ; 1 >\n");
+			else if(shiftc==-1)fprintf(asmf,"\tdec rbx  ; 1 <\n");
+			else fprintf(asmf,"\tadd rbx, %d  ; %d %c\n",shiftc,shiftc<0?-shiftc:shiftc,shiftc<0?'<':'>');
+			shiftc=0;
+		}
+		switch(c){
+			case '+': addc++; break;
+			case '-': addc--; break;
+			case '>': shiftc++; break;
+			case '<': shiftc--; break;
+			case '[':{
+				struct stackitem *ls=malloc(sizeof(struct stackitem));
+				assert(ls);
+				ls->id=uniqid();
+				ls->next=loopstack;
+				loopstack=ls;
+				fprintf(asmf,
+					"loop%d:  ; [\n"
+					"\tcmp byte [rbx], 0\n"
+					"\tjz loop%d_end\n"
+					"loop%d_body:\n",
+					ls->id,ls->id,ls->id);
+				break;
+			}
+			case ']':{
+				assert(loopstack);
+				fprintf(asmf,
+					"\tcmp byte [rbx], 0  ; ]\n"
+					"\tjnz loop%d_body\n"
+					"loop%d_end:\n",
+					loopstack->id,loopstack->id);
+				struct stackitem *ls=loopstack->next;
+				free(loopstack);
+				loopstack=ls;
+				break;
+			}
+			case '.':
+				fprintf(asmf,
+					"\tmov edi, 0  ; .\n"
+					"\tmov dil, byte [rbx]\n"
+					"\tcall _putchar\n");
+				break;
+			case ',':
+				fprintf(asmf,
+					"\tcall _getchar  ; ,\n"
+					"\tmov byte [rbx], al\n");
+				break;
+			default:
+				if(!eof)assert(false);
+		}
+	}
+	assert(!loopstack);
+}
+
+void writeepilogue(FILE *asmf){
+	fprintf(asmf,
+		"\n"
+		"\tmov rdi, [buf]  ; epilogue\n"
+		"\tcall _free\n"
+		"\n"
+		/*"\tlea rdi, [formatstr] ;output last cell reached\n"
+		"\tmov rsi, [maxp]\n"
+		"\tsub rsi, [buf]\n"
+		"\txor eax, eax\n"
+		"\tcall _printf\n"
+		"\n"*/
+		"\txor eax, eax\n"
+		"\tjmp mainend\n"
+		"\n"
+		"nomem:\n"
+		"\tmov edi, 2\n"
+		"\tlea rsi, [nomemstr]\n"
+		"\tmov rdx, nomemstr.len\n"
+		"\tcall _write\n"
+		"\tmov eax, 1\n"
+		"\n"
+		"mainend:\n"
+		"\tpop rbx\n"
+		"\tmov rsp, rbp\n"
+		"\tpop rbp\n"
+		"\tret\n"
+		"\n"
+		"\n"
+		"section .data\n"
+		"sectalign 8\n"
+		"\n"
+		"nomemstr: db \"Out of memory!\"\n"
+		".len: equ $-nomemstr\n"
+		"\n"
+		"formatstr: db \"Last cell reached: %%d\", 10, 0\n"
+		"\n"
+		"align 8, db 0\n"
+		"maxp: dq 0\n"
+		"\n"
+		"\n"
+		"section .bss\n"
+		"sectalign 8\n"
+		"\n"
+		"buf: resq 1\n");
+}
+
+int main(int argc,char **argv){
+	parseargs(argc,argv);
+
+	FILE *srcf=fopen(params.srcfname,"r");
+	assert(srcf);
+
+	char asmfname[20];
+	FILE *asmf=gettempfilew(asmfname);
+	if(params.verbose)fprintf(stderr,"asm -> %s\n",asmfname);
+
+	DBGTIME("writeprologue",writeprologue(asmf););
+	DBGTIME("writeprogram",writeprogram(srcf,asmf););
+	DBGTIME("writeepilogue",writeepilogue(asmf););
+
+	fclose(asmf);
+	fclose(srcf);
+
+	int ret=compilechain(asmfname,params.dstfname);
+	unlink(asmfname);
+	if(params.verbose)fprintf(stderr,"unlink(%s)  (asmfname)\n",asmfname);
+	return ret;
+}