rework kprintf

src/arch/amd64/paging.c

@@ -633,7 +633,6 @@ int kload_page(void *virt_addr) {
 	if (page->loaded)
 		return -1;
 	void *phys = alloc_phys_page();
-	kprintf("0x%p\n", phys);
 	if (phys == NULL)
 		return -2;
 	page->loaded = 1;

src/kmain.c

@@ -1,3 +1,4 @@
+#include <backtrace.h>
 #include <fpu.h>
 #include <acpi.h>
 #include <memory.h>
@@ -18,6 +19,8 @@ void kmain(struct boot_info *info) {
 	char *test = kalloc(5);
 	*test = 1;
 
+	//log_backtrace();
+
 	while (1) {
 		screen_redraw();
 		// loop so we dont halt

src/print.c

@@ -1,6 +1,7 @@
 #include <lib.h>
 #include <serial.h>
 #include <stdarg.h>
+#include <stdint.h>
 
 void kputc(char c) {
 	serial_out(c);
@@ -31,54 +32,110 @@ struct format_precision {
 };
 
 enum format_modifier {
-	MOD_NONE,
+	MOD_NONE = 0,
-	MOD_INVALID,
+	MOD_HALF = 1,
-	MOD_HALF_HALF,
+	MOD_HALF_HALF = 3,
-	MOD_HALF,
+	MOD_LONG = 3,
-	MOD_LONG_LONG,
+	MOD_LONG_LONG = 4,
-	MOD_LONG,
+	MOD_INVALID = 5,
 };
 
 enum format_conversion {
-	FMT_INT,
+	// decimal numbers
-	FMT_UINT,
+	FMT_CHAR = 0,
+	FMT_INT = 4,
+	FMT_UINT = 5,
+	// otehr numbers
 	FMT_OCT,
 	FMT_HEX,
 	FMT_HEX_UPPER,
-	FMT_CHAR,
+	// text
 	FMT_STR,
 	FMT_PTR,
-	FMT_PTR_UPPER,
+	// misc
-	FMT_PERCENT,
 	FMT_INVALID
 };
 
-static enum format_flag read_flag(const char *format, const char **end) {
+enum format_length {
+	LEN_UCHAR,
+	LEN_CHAR,
+	LEN_USHORT,
+	LEN_SHORT,
+	LEN_UINT,
+	LEN_INT,
+	LEN_ULONG,
+	LEN_LONG,
+	LEN_ULONGLONG,
+	LEN_LONGLONG,
+	LEN_SIZET,
+	LEN_INVALID
+};
 
-	enum format_flag flag = FLG_NONE;
+struct spacing {
+	unsigned left;
+	unsigned length;
+	unsigned right;
+	bool zero;
+};
+
+enum charcase {
+	UPPERCASE,
+	LOWERCASE
+};
+
+enum printtype {
+	NONE,
+	OCTAL,
+	HEX
+};
+
+static enum printtype conversion_to_printtype(enum format_conversion conversion) {
+	switch (conversion) {
+		case FMT_OCT:
+			return OCTAL;
+		case FMT_HEX:
+		case FMT_HEX_UPPER:
+			return HEX;
+		default:
+			return NONE;
+	}
+}
+
+static enum charcase conversion_to_charcase(enum format_conversion conversion) {
+	switch (conversion) {
+		case FMT_HEX_UPPER:
+			return UPPERCASE;
+		default:
+			return LOWERCASE;
+	}
+}
+
+static enum format_flag read_flags(const char *format, const char **end) {
+
+	enum format_flag flags = FLG_NONE;
 
 	for (; *format != '\0'; format++) {
 		switch (*format) {
 			case '#':
-				flag |= FLG_ALTERNATE;
+				flags |= FLG_ALTERNATE;
 				break;
 			case '0':
-				flag |= FLG_ZERO;
+				flags |= FLG_ZERO;
 				break;
 			case '-':
-				flag |= FLG_LEFT_ALIGN;
+				flags |= FLG_LEFT_ALIGN;
 				break;
 			case '+':
-				flag |= FLG_ADD_SIGN;
+				flags |= FLG_ADD_SIGN;
 				break;
 			default:
 				*end = format;
-				return flag;
+				return flags;
 		}
 	}
 
 	*end = format;
-	return flag;
+	return flags;
 }
 
 static struct format_width read_width(const char *format, const char **end) {
@@ -221,213 +278,344 @@ static enum format_conversion read_conversion(const char *format, const char **e
 			return FMT_STR;
 		case 'p':
 			return FMT_PTR;
-		case 'P':
-			return FMT_PTR_UPPER;
-		case '%':
-			return FMT_PERCENT;
 		default:
 			return FMT_INVALID;
 	}
 }
 
-static void print_string(
+static char ctoa_print(
-		const char *str,
+		char c,					 // the single digit to convert to ascii
-		enum format_flag flag,
+		enum charcase cc			// if this digit should be uppercase or lowercase
-		struct format_width width,
-		struct format_precision precision
-) {
-
-	size_t max_len = 0;
-	size_t min_len = 0;
-	size_t len = 0;
-
-	if (width.defined)
-		min_len = width.value;
-
-	if (precision.defined) {
-		max_len = precision.value;
-		len = max_len;
-		if (max_len < min_len)
-			min_len = max_len;
-	} else {
-		len = strlen(str);
-	}
-
-	if (!(flag & FLG_LEFT_ALIGN) && len < min_len) {
-		for (size_t i = 0; i < (min_len - len); i++) {
-			kputc(' ');
-		}
-	}
-
-	for (size_t i = 0; i < len; i++) {
-		kputc(str[i]);
-	}
-
-	if ((flag & FLG_LEFT_ALIGN) && len < min_len) {
-		for (size_t i = 0; i < (min_len - len); i++) {
-			kputc(' ');
-		}
-	}
-}
-
-static char get_letter(
-		char c,
-		char base
 ) {
+	char ascii_base = cc == UPPERCASE ? 'A' : 'a';
 	if (c >= 0 && c <= 9)
 		return c + '0';
 	c -= 10;
-	return c + base;
+	return c + ascii_base;
 }
 
-static char *get_decimal(
+static char *ulltoa_print(
-		long long num,
+		unsigned long long num,	 // the number to convert to ascii
-		char *buf,
+		unsigned radix,			 // the divisor for the number (base 10, octal, hex)
-		char sign,
+		char *buf,				  // the buffer to write to
-		int radix,
+		enum charcase cc,		   // if any a-z ascii should be uppercase or lowercase
-		char base
+		bool addsign				// add a sign even if its positive
 ) {
-
+	// we are printing the string backwards, add null byte
 	*buf = '\0';
 	buf--;
 
+	// if the number is zero, the while loop will never run
+	// so we have to add the zero digit ourselves
 	if (num == 0) {
 		*buf = '0';
 		buf--;
 	}
 
+	// for each digit, convert to ascii and add to buffer
 	while (num != 0) {
 		char i = num % radix;
-		char c = get_letter(i, base);
+		char c = ctoa_print(i, cc);
 		*buf = c;
 		buf--;
 		num /= radix;
 	}
 
-	if (sign) {
+	// since this is the unsigned function, add a plus sign if
-		*buf = sign;
+	// requested
+	if (addsign) {
+		*buf = '+';
 		buf--;
 	}
 
-	buf++;
+	buf++; // move forward one to be at the start of the string
 
 	return buf;
 }
 
-static void print_unum(
+static char *lltoa_print(
-		unsigned long long num,
+		signed long long num,	   // the number to convert to ascii
-		enum format_flag flag,
+		unsigned radix,			 // the divistor for the number (base 10, octal, hex)
-		struct format_width width,
+		char *buf,				  // the buffer to write to
-		struct format_precision precision,
+		enum charcase cc,		   // if any a-z ascii should be uppercase or lowercase
-		bool isneg,
+		bool addsign				// add a sign even if its positive
-		int radix,
-		char base
 ) {
+	bool isneg = num < 0;
 
-	size_t max_len = 0;
-	size_t min_len = 0;
-	size_t len = 0;
-
-	char sign = 0;
 	if (isneg)
-		sign = '-';
+		num = -num;
-	else if (flag & FLG_ADD_SIGN)
-		sign = '+';
 
-	char buf[1024];
+	buf = ulltoa_print(num, radix, buf, cc, !isneg && addsign);
-	char *str = get_decimal(
-			num,
-			buf + 1023,
-			sign,
-			radix,
-			base
-	);
 
-	bool space_pre = (flag & FLG_LEFT_ALIGN) || !(flag & FLG_ZERO);
+	if (isneg && num < 0) {
-
+		buf--;
-	if (!space_pre && radix == 16 && (flag & FLG_ALTERNATE)) {
+		*buf = '-';
-		char x = base + ('x' - 'a');
-		serial_out('0');
-		serial_out(x);
 	}
 
-	if (width.defined)
+	return buf;
-		min_len = width.value;
+}
 
-	if (precision.defined) {
+
-		max_len = precision.value;
+static void print_string_buffer(
-		len = max_len;
+		const char *buf,			// buffer containing the text we want to print
-		if (max_len < min_len)
+		struct spacing spacing	  // the spacing on the left right and middle
-			min_len = max_len;
+) {
-	} else {
+	for (unsigned i = 0; i < spacing.left; i++) {
-		len = strlen(str);
+		kputc(' ');
 	}
 
-	bool zero_padded = false;
+	for (unsigned i = 0; i < spacing.length; i++) {
-
+		kputc(*buf++);
-	if (!(flag & FLG_LEFT_ALIGN) && len < min_len) {
-		for (size_t i = 0; i < (min_len - len); i++) {
-			(flag & FLG_ZERO) ? kputc('0') : kputc(' ');
-		}
-		if (flag & FLG_ZERO)
-			zero_padded = true;
 	}
 
-	if (space_pre && radix == 16 && (flag & FLG_ALTERNATE)) {
+	for (unsigned i = 0; i < spacing.right; i++) {
-		char x = base + ('x' - 'a');
+		kputc(' ');
-		serial_out('0');
-		serial_out(x);
-	}
-
-	if (space_pre && radix == 16 && flag & FLG_ALTERNATE) {
-		char x = base + ('x' - 'a');
-		serial_out('0');
-		serial_out(x);
-	}
-
-	kputs(str);
-
-	if (!zero_padded && (flag & FLG_ALTERNATE) && radix == 8)
-		kputc('0');
-
-	if ((flag & FLG_LEFT_ALIGN) && len < min_len) {
-		for (size_t i = 0; i < (min_len - len); i++) {
-			(flag & FLG_ZERO) ? kputc('0') : kputc(' ');
-		}
 	}
 }
 
-static void print_num(
+static void print_number_buffer(
-		long long num,
+		const char *buf,			// buffer containing the text we want to print
-		enum format_flag flag,
+		struct spacing spacing,	 // the spacing on the left right and middle
-		struct format_width width,
+		enum printtype type,
-		struct format_precision precision,
+		enum charcase cc
-		int radix,
-		char base
 ) {
-	bool isneg = false;
 
-	if (num < 0) {
+	// put the 0x at the start of the string
-		num = ~num;
+	if (spacing.left && spacing.zero && type == HEX) {
-		isneg = true;
+		if (cc == UPPERCASE) {
+			kputs("0X");
+		} else {
+			kputs("0x");
+		}
 	}
 
-	print_unum(
+	if ((!spacing.left || !spacing.zero) && type == OCTAL) {
-			num,
+		kputc('0');
-			flag,
+	}
-			width,
+
-			precision,
+	for (unsigned i = 0; i < spacing.left; i++) {
-			isneg,
+		spacing.zero ? kputc('0') : kputc(' ');
-			radix,
+	}
-			base
+
-	);
+	if ((!spacing.left || !spacing.zero) && type == HEX) {
+		if (cc == UPPERCASE) {
+			kputs("0X");
+		} else {
+			kputs("0x");
+		}
+	}
+
+	for (unsigned i = 0; i < spacing.length; i++) {
+		kputc(*buf++);
+	}
+
+	for (unsigned i = 0; i < spacing.right; i++) {
+		kputc(' ');
+	}
+}
+
+static unsigned get_string_len(
+	const char *str,
+	struct format_precision precision
+) {
+	if (!precision.defined)
+		return strlen(str);
+	unsigned max = precision.value;
+	for (unsigned i = 0; i < max; i++) {
+		if (str[i] == '\0') {
+			// we broke early
+			// precision is greater then strlen
+			return i;
+		}
+	}
+	return max;
+}
+
+static struct spacing get_spacing(
+	struct format_width width,
+	struct format_precision precision,
+	enum format_flag flags,
+	unsigned length
+) {
+	struct spacing spacing;
+
+	unsigned min = 0,
+			 max = 0;
+
+	if (width.defined) {
+		min = width.value;
+	}
+
+	if (precision.defined) {
+		max = precision.value;
+		if (length > max) {
+			length = max;
+		}
+		if (min > max) {
+			min = max;
+		}
+	}
+
+	unsigned gap = 0;
+	if (max > length) {
+		gap = max - length;
+	}
+
+	spacing.length = length;
+
+	if (flags & FLG_LEFT_ALIGN) {
+		spacing.left = gap;
+		spacing.right = 0;
+	} else {
+		spacing.left = 0;
+		spacing.right = gap;
+	}
+
+	spacing.zero = (flags & FLG_ZERO);
+
+	return spacing;
+}
+
+static struct spacing get_string_spacing(
+	const char *str,
+	struct format_width width,
+	struct format_precision precision,
+	enum format_flag flags
+) {
+	unsigned length = get_string_len(str, precision);
+	return get_spacing(width, precision, flags, length);
+}
+
+static void print_string(
+		const char *str,
+		enum format_flag flags,
+		struct format_width width,
+		struct format_precision precision
+) {
+	struct spacing spacing =
+		get_string_spacing(str, width, precision, flags);
+	print_string_buffer(str, spacing);
+}
+
+static enum format_length apply_modifier(
+		enum format_conversion conversion,
+		enum format_modifier modifier
+) {
+	switch (conversion) {
+		case FMT_CHAR:
+			if (modifier == MOD_NONE)
+				return LEN_CHAR;
+			else
+				return LEN_INVALID;
+
+		case FMT_PTR:
+			if (modifier == MOD_NONE)
+				return LEN_SIZET;
+			else
+				return LEN_INVALID;
+
+		case FMT_INT:
+			if (modifier == MOD_LONG)
+				return LEN_LONG;
+			else if (modifier == MOD_LONG_LONG)
+				return LEN_LONGLONG;
+			else if (modifier == MOD_HALF)
+				return LEN_SHORT;
+			else if (modifier == MOD_HALF_HALF)
+				return LEN_CHAR;
+			else
+				return LEN_INT;
+
+		case FMT_UINT:
+		case FMT_OCT:
+		case FMT_HEX:
+		case FMT_HEX_UPPER:
+			if (modifier == MOD_LONG)
+				return LEN_ULONG;
+			else if (modifier == MOD_LONG_LONG)
+				return LEN_ULONGLONG;
+			else if (modifier == MOD_HALF)
+				return LEN_USHORT;
+			else if (modifier == MOD_HALF_HALF)
+				return LEN_UCHAR;
+			else
+				return LEN_UINT;
+
+		default:
+			return LEN_INVALID;
+	}
+}
+
+static unsigned printtype_to_radix(
+		enum printtype type
+) {
+	switch (type) {
+		case OCTAL:
+			return 8;
+		case HEX:
+			return 16;
+		case NONE:
+		default:
+			return 10;
+	}
+}
+
+static void print_signed_number(
+		signed long long num,
+		enum format_flag flags,
+		struct format_width width,
+		struct format_precision precision,
+		enum printtype type,
+		enum charcase cc
+) {
+	unsigned radix = printtype_to_radix(type);
+	char buf[256];
+	char *ptr = lltoa_print(num, radix, buf + 255, cc, flags & FLG_ADD_SIGN);
+	unsigned length = strlen(ptr);
+	struct spacing spacing =
+		get_spacing(width, precision, flags, length);
+	print_number_buffer(ptr, spacing, type, cc);
+}
+
+static void print_unsigned_number(
+		unsigned long long num,
+		enum format_flag flags,
+		struct format_width width,
+		struct format_precision precision,
+		enum printtype type,
+		enum charcase cc
+) {
+	unsigned radix = printtype_to_radix(type);
+	char buf[256];
+	char *ptr = ulltoa_print(num, radix, buf + 255, cc, flags & FLG_ADD_SIGN);
+	unsigned length = strlen(ptr);
+	struct spacing spacing =
+		get_spacing(width, precision, flags, length);
+	print_number_buffer(ptr, spacing, type, cc);
+}
+
+static bool is_conversion_number(enum format_conversion conversion) {
+	switch (conversion) {
+		case FMT_INT:
+		case FMT_UINT:
+		case FMT_OCT:
+		case FMT_HEX:
+		case FMT_HEX_UPPER:
+		case FMT_PTR:
+			return true;
+		case FMT_CHAR:
+		case FMT_STR:
+		case FMT_INVALID:
+		default:
+			return false;
+	}
 }
 
 void kvprintf(const char *format, va_list args) {
 	for (; *format != '\0'; format++) {
 		char c = *format;
 		if (c == '%') {
-			enum format_flag flag;
+			enum format_flag flags;
 			struct format_width width;
 			struct format_precision precision;
 			enum format_modifier modifier;
@@ -435,34 +623,24 @@ void kvprintf(const char *format, va_list args) {
 
 			const char *ptr = format + 1;
 
-			flag = read_flag(ptr, &ptr);
+			if (*(format + 1) == '%') {
+				kputc('%');
+				format++;
+				continue;
+			}
+
+			flags = read_flags(ptr, &ptr);
 			width = read_width(ptr, &ptr);
 			precision = read_precision(ptr, &ptr);
+
 			modifier = read_modifier(ptr, &ptr);
 
 			if (modifier == MOD_INVALID) {
-				kputc('%');
+				goto error;
-				continue;
 			}
 
 			conversion = read_conversion(ptr, &ptr);
 
-			if (conversion == FMT_INVALID) {
-				kputc('%');
-				continue;
-			}
-
-			union {
-				unsigned long long u;
-				long long l;
-				char c;
-				const char *str;
-				void *ptr;
-			} data = {0};
-
-			int radix = 0;
-			char base = 0;
-
 			if (width.varys) {
 				int len = va_arg(args, int);
 				width.value = len;
@@ -473,106 +651,80 @@ void kvprintf(const char *format, va_list args) {
 				precision.value = len;
 			}
 
-			switch (conversion) {
+			if (conversion == FMT_INVALID) {
-				case FMT_INT:
+				goto error;
-					if (modifier == MOD_NONE)
-						data.l = va_arg(args, int);
-					else if (modifier == MOD_HALF)
-						data.l = (short) va_arg(args, int);
-					else if (modifier == MOD_HALF_HALF)
-						data.l = (char) va_arg(args, int);
-					else if (modifier == MOD_LONG)
-						data.l = va_arg(args, long);
-					else if (modifier == MOD_LONG_LONG)
-						data.l = va_arg(args, long long);
-					radix = 10;
-					print_num(
-							data.l,
-							flag,
-							width,
-							precision,
-							radix,
-							base
-					);
-					break;
-				case FMT_UINT:
-				case FMT_OCT:
-				case FMT_HEX_UPPER:
-				case FMT_HEX:
-					if (modifier == MOD_NONE)
-						data.u = va_arg(args, unsigned int);
-					else if (modifier == MOD_HALF)
-						data.u = (unsigned short) va_arg(args, unsigned int);
-					else if (modifier == MOD_HALF_HALF)
-						data.u = (unsigned char) va_arg(args, unsigned int);
-					else if (modifier == MOD_LONG)
-						data.u = va_arg(args, unsigned long);
-					else if (modifier == MOD_LONG_LONG)
-						data.u = va_arg(args, unsigned long long);
-
-					if (conversion == FMT_UINT) {
-						radix = 10;
-					} else if (conversion == FMT_OCT) {
-						radix = 8;
-					} else if (conversion == FMT_HEX) {
-						radix = 16;
-						base = 'a';
-					} else if (conversion == FMT_HEX_UPPER) {
-						radix = 16;
-						base = 'A';
-					}
-					goto printunum;
-					break;
-				case FMT_PTR:
-				case FMT_PTR_UPPER:
-					flag |= FLG_ZERO;
-					data.u = va_arg(args, size_t);
-					radix = 16;
-					if (conversion == FMT_PTR)
-						base = 'a';
-					else
-						base = 'A';
-					goto printunum;
-					break;
-				printunum:
-					print_unum(
-							data.u,
-							flag,
-							width,
-							precision,
-							false,
-							radix,
-							base
-					);
-					break;
-				case FMT_CHAR: {
-					char buf[2];
-					buf[0] = (char) va_arg(args, int);
-					buf[1] = '\0';
-					print_string(
-							buf,
-							flag,
-							width,
-							precision
-					);
-					break;
-				}
-				case FMT_STR:
-					data.str = va_arg(args, const char*);
-					print_string(
-							data.str,
-							flag,
-							width,
-							precision
-					);
-					break;
-				case FMT_PERCENT:
-					kputc('%');
-					break;
-				case FMT_INVALID:
-					break;
 			}
+
+			if (conversion == FMT_INT) {
+				enum format_length length =
+					apply_modifier(conversion, modifier);
+				enum printtype type = NONE;
+				enum charcase cc = LOWERCASE;
+
+				signed long long num = 0;
+				switch (length) {
+					case LEN_CHAR:
+					case LEN_SHORT:
+					case LEN_INT:
+						num = va_arg(args, signed int);
+						break;
+					case LEN_LONG:
+						num = va_arg(args, signed long);
+						break;
+					case LEN_LONGLONG:
+						num = va_arg(args, signed long long);
+						break;
+					default:
+						goto error;
+				}
+
+				print_signed_number(num, flags, width, precision, type, cc);
+			} else if (is_conversion_number(conversion)) {
+				enum format_length length =
+					apply_modifier(conversion, modifier);
+				enum printtype type =
+					conversion_to_printtype(conversion);
+				enum charcase cc =
+					conversion_to_charcase(conversion);
+
+				unsigned long long num = 0;
+				switch (length) {
+					case LEN_UCHAR:
+					case LEN_USHORT:
+					case LEN_UINT:
+						num = va_arg(args, unsigned int);
+						break;
+					case LEN_ULONG:
+						num = va_arg(args, unsigned long);
+						break;
+					case LEN_ULONGLONG:
+						num = va_arg(args, unsigned long long);
+						break;
+					case LEN_SIZET:
+						num = va_arg(args, size_t);
+						break;
+					default:
+						goto error;
+				}
+
+				print_unsigned_number(num, flags, width, precision, type, cc);
+			} else if (conversion == FMT_STR) {
+				char *str = va_arg(args, char *);
+				print_string(str, flags, width, precision);
+			} else if (conversion == FMT_CHAR) {
+				char temp[2];
+				temp[0] = va_arg(args, int);
+				temp[1] = '\0';
+				print_string(temp, flags, width, precision);
+			}
+
 			format = ptr - 1;
+			continue;
+
+error:
+			kputc('%');
+			format++;
+			continue;
 		} else {
 			kputc(c);
 		}