From 351a895d695cb2a709e7d6c9768d9ee8fe6be9fc Mon Sep 17 00:00:00 2001 From: Konstantin Nazarov Date: Sat, 14 Dec 2024 22:56:28 +0000 Subject: [PATCH] Add support for printf, UART, and iron out a few bugs --- example/Makefile | 6 +- example/boot.s | 2 +- example/example.c | 7 +- example/linker.ld | 6 +- example/printf.c | 914 ++++++++++++++++++++++++++++++++++++++++++++++ example/printf.h | 117 ++++++ example/putchar.c | 4 + src/debug.cpp | 9 +- src/elf.cpp | 6 +- src/rve.cpp | 6 +- src/vm.cpp | 229 ++++++++---- src/vm.hpp | 33 ++ 12 files changed, 1257 insertions(+), 82 deletions(-) create mode 100644 example/printf.c create mode 100644 example/printf.h create mode 100644 example/putchar.c diff --git a/example/Makefile b/example/Makefile index 76531bc..001e78d 100644 --- a/example/Makefile +++ b/example/Makefile @@ -1,4 +1,6 @@ -example: example.c Makefile boot.s linker.ld +example: example.c Makefile boot.s linker.ld printf.h printf.c putchar.c riscv32-none-elf-as -march=rv32i -mabi=ilp32 boot.s -o boot.o + riscv32-none-elf-gcc -fno-builtin -fvisibility=hidden -nostdlib -nostartfiles -march=rv32im -mabi=ilp32 -D PRINTF_DISABLE_SUPPORT_FLOAT=1 -D PRINTF_DISABLE_SUPPORT_LONG_LONG=1 -c printf.c -o printf.o -g + riscv32-none-elf-gcc -fno-builtin -fvisibility=hidden -nostdlib -nostartfiles -march=rv32im -mabi=ilp32 -c putchar.c -o putchar.o -g riscv32-none-elf-gcc -fno-builtin -fvisibility=hidden -nostdlib -nostartfiles -march=rv32im -mabi=ilp32 -c example.c -o example.o -g - riscv32-none-elf-ld boot.o example.o -T linker.ld -o example -g + riscv32-none-elf-ld boot.o example.o printf.o putchar.o -T linker.ld -o example -g diff --git a/example/boot.s b/example/boot.s index af4f484..edaff0d 100644 --- a/example/boot.s +++ b/example/boot.s @@ -1,6 +1,6 @@ .globl _boot _boot: - li x2, 0x8000 + li x2, 0x80000 call main sbreak j . diff --git a/example/example.c b/example/example.c index 7830293..0de0bb8 100644 --- a/example/example.c +++ b/example/example.c @@ -1,4 +1,4 @@ -static int mem = 1; +#include "printf.h" int fact(int n) { if (n == 0) @@ -8,7 +8,10 @@ int fact(int n) { } int main() { - mem = fact(8); + int n = 8; + int res = fact(8); + + printf("%d! = %d\n", n, res); return 0; } diff --git a/example/linker.ld b/example/linker.ld index aac741b..3cf558e 100644 --- a/example/linker.ld +++ b/example/linker.ld @@ -7,19 +7,19 @@ SECTIONS { *(.text) /* Place all .text sections (code) here */ } - . = 0x1000; + . = 0x10000; .data : { *(.data) /* Place all .data sections (initialized data) here */ } - . = 0x2000; + . = 0x20000; .bss : { *(.bss) /* Place all .bss sections (uninitialized data) here */ } - . = 0x3000; + . = 0x30000; .stack : { *(.stack) diff --git a/example/printf.c b/example/printf.c new file mode 100644 index 0000000..8a700ad --- /dev/null +++ b/example/printf.c @@ -0,0 +1,914 @@ +/////////////////////////////////////////////////////////////////////////////// +// \author (c) Marco Paland (info@paland.com) +// 2014-2019, PALANDesign Hannover, Germany +// +// \license The MIT License (MIT) +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. +// +// \brief Tiny printf, sprintf and (v)snprintf implementation, optimized for speed on +// embedded systems with a very limited resources. These routines are thread +// safe and reentrant! +// Use this instead of the bloated standard/newlib printf cause these use +// malloc for printf (and may not be thread safe). +// +/////////////////////////////////////////////////////////////////////////////// + +#include +#include + +#include "printf.h" + + +// define this globally (e.g. gcc -DPRINTF_INCLUDE_CONFIG_H ...) to include the +// printf_config.h header file +// default: undefined +#ifdef PRINTF_INCLUDE_CONFIG_H +#include "printf_config.h" +#endif + + +// 'ntoa' conversion buffer size, this must be big enough to hold one converted +// numeric number including padded zeros (dynamically created on stack) +// default: 32 byte +#ifndef PRINTF_NTOA_BUFFER_SIZE +#define PRINTF_NTOA_BUFFER_SIZE 32U +#endif + +// 'ftoa' conversion buffer size, this must be big enough to hold one converted +// float number including padded zeros (dynamically created on stack) +// default: 32 byte +#ifndef PRINTF_FTOA_BUFFER_SIZE +#define PRINTF_FTOA_BUFFER_SIZE 32U +#endif + +// support for the floating point type (%f) +// default: activated +#ifndef PRINTF_DISABLE_SUPPORT_FLOAT +#define PRINTF_SUPPORT_FLOAT +#endif + +// support for exponential floating point notation (%e/%g) +// default: activated +#ifndef PRINTF_DISABLE_SUPPORT_EXPONENTIAL +#define PRINTF_SUPPORT_EXPONENTIAL +#endif + +// define the default floating point precision +// default: 6 digits +#ifndef PRINTF_DEFAULT_FLOAT_PRECISION +#define PRINTF_DEFAULT_FLOAT_PRECISION 6U +#endif + +// define the largest float suitable to print with %f +// default: 1e9 +#ifndef PRINTF_MAX_FLOAT +#define PRINTF_MAX_FLOAT 1e9 +#endif + +// support for the long long types (%llu or %p) +// default: activated +#ifndef PRINTF_DISABLE_SUPPORT_LONG_LONG +#define PRINTF_SUPPORT_LONG_LONG +#endif + +// support for the ptrdiff_t type (%t) +// ptrdiff_t is normally defined in as long or long long type +// default: activated +#ifndef PRINTF_DISABLE_SUPPORT_PTRDIFF_T +#define PRINTF_SUPPORT_PTRDIFF_T +#endif + +/////////////////////////////////////////////////////////////////////////////// + +// internal flag definitions +#define FLAGS_ZEROPAD (1U << 0U) +#define FLAGS_LEFT (1U << 1U) +#define FLAGS_PLUS (1U << 2U) +#define FLAGS_SPACE (1U << 3U) +#define FLAGS_HASH (1U << 4U) +#define FLAGS_UPPERCASE (1U << 5U) +#define FLAGS_CHAR (1U << 6U) +#define FLAGS_SHORT (1U << 7U) +#define FLAGS_LONG (1U << 8U) +#define FLAGS_LONG_LONG (1U << 9U) +#define FLAGS_PRECISION (1U << 10U) +#define FLAGS_ADAPT_EXP (1U << 11U) + + +// import float.h for DBL_MAX +#if defined(PRINTF_SUPPORT_FLOAT) +#include +#endif + + +// output function type +typedef void (*out_fct_type)(char character, void* buffer, size_t idx, size_t maxlen); + + +// wrapper (used as buffer) for output function type +typedef struct { + void (*fct)(char character, void* arg); + void* arg; +} out_fct_wrap_type; + + +// internal buffer output +static inline void _out_buffer(char character, void* buffer, size_t idx, size_t maxlen) +{ + if (idx < maxlen) { + ((char*)buffer)[idx] = character; + } +} + + +// internal null output +static inline void _out_null(char character, void* buffer, size_t idx, size_t maxlen) +{ + (void)character; (void)buffer; (void)idx; (void)maxlen; +} + + +// internal _putchar wrapper +static inline void _out_char(char character, void* buffer, size_t idx, size_t maxlen) +{ + (void)buffer; (void)idx; (void)maxlen; + if (character) { + _putchar(character); + } +} + + +// internal output function wrapper +static inline void _out_fct(char character, void* buffer, size_t idx, size_t maxlen) +{ + (void)idx; (void)maxlen; + if (character) { + // buffer is the output fct pointer + ((out_fct_wrap_type*)buffer)->fct(character, ((out_fct_wrap_type*)buffer)->arg); + } +} + + +// internal secure strlen +// \return The length of the string (excluding the terminating 0) limited by 'maxsize' +static inline unsigned int _strnlen_s(const char* str, size_t maxsize) +{ + const char* s; + for (s = str; *s && maxsize--; ++s); + return (unsigned int)(s - str); +} + + +// internal test if char is a digit (0-9) +// \return true if char is a digit +static inline bool _is_digit(char ch) +{ + return (ch >= '0') && (ch <= '9'); +} + + +// internal ASCII string to unsigned int conversion +static unsigned int _atoi(const char** str) +{ + unsigned int i = 0U; + while (_is_digit(**str)) { + i = i * 10U + (unsigned int)(*((*str)++) - '0'); + } + return i; +} + + +// output the specified string in reverse, taking care of any zero-padding +static size_t _out_rev(out_fct_type out, char* buffer, size_t idx, size_t maxlen, const char* buf, size_t len, unsigned int width, unsigned int flags) +{ + const size_t start_idx = idx; + + // pad spaces up to given width + if (!(flags & FLAGS_LEFT) && !(flags & FLAGS_ZEROPAD)) { + for (size_t i = len; i < width; i++) { + out(' ', buffer, idx++, maxlen); + } + } + + // reverse string + while (len) { + out(buf[--len], buffer, idx++, maxlen); + } + + // append pad spaces up to given width + if (flags & FLAGS_LEFT) { + while (idx - start_idx < width) { + out(' ', buffer, idx++, maxlen); + } + } + + return idx; +} + + +// internal itoa format +static size_t _ntoa_format(out_fct_type out, char* buffer, size_t idx, size_t maxlen, char* buf, size_t len, bool negative, unsigned int base, unsigned int prec, unsigned int width, unsigned int flags) +{ + // pad leading zeros + if (!(flags & FLAGS_LEFT)) { + if (width && (flags & FLAGS_ZEROPAD) && (negative || (flags & (FLAGS_PLUS | FLAGS_SPACE)))) { + width--; + } + while ((len < prec) && (len < PRINTF_NTOA_BUFFER_SIZE)) { + buf[len++] = '0'; + } + while ((flags & FLAGS_ZEROPAD) && (len < width) && (len < PRINTF_NTOA_BUFFER_SIZE)) { + buf[len++] = '0'; + } + } + + // handle hash + if (flags & FLAGS_HASH) { + if (!(flags & FLAGS_PRECISION) && len && ((len == prec) || (len == width))) { + len--; + if (len && (base == 16U)) { + len--; + } + } + if ((base == 16U) && !(flags & FLAGS_UPPERCASE) && (len < PRINTF_NTOA_BUFFER_SIZE)) { + buf[len++] = 'x'; + } + else if ((base == 16U) && (flags & FLAGS_UPPERCASE) && (len < PRINTF_NTOA_BUFFER_SIZE)) { + buf[len++] = 'X'; + } + else if ((base == 2U) && (len < PRINTF_NTOA_BUFFER_SIZE)) { + buf[len++] = 'b'; + } + if (len < PRINTF_NTOA_BUFFER_SIZE) { + buf[len++] = '0'; + } + } + + if (len < PRINTF_NTOA_BUFFER_SIZE) { + if (negative) { + buf[len++] = '-'; + } + else if (flags & FLAGS_PLUS) { + buf[len++] = '+'; // ignore the space if the '+' exists + } + else if (flags & FLAGS_SPACE) { + buf[len++] = ' '; + } + } + + return _out_rev(out, buffer, idx, maxlen, buf, len, width, flags); +} + + +// internal itoa for 'long' type +static size_t _ntoa_long(out_fct_type out, char* buffer, size_t idx, size_t maxlen, unsigned long value, bool negative, unsigned long base, unsigned int prec, unsigned int width, unsigned int flags) +{ + char buf[PRINTF_NTOA_BUFFER_SIZE]; + size_t len = 0U; + + // no hash for 0 values + if (!value) { + flags &= ~FLAGS_HASH; + } + + // write if precision != 0 and value is != 0 + if (!(flags & FLAGS_PRECISION) || value) { + do { + const char digit = (char)(value % base); + buf[len++] = digit < 10 ? '0' + digit : (flags & FLAGS_UPPERCASE ? 'A' : 'a') + digit - 10; + value /= base; + } while (value && (len < PRINTF_NTOA_BUFFER_SIZE)); + } + + return _ntoa_format(out, buffer, idx, maxlen, buf, len, negative, (unsigned int)base, prec, width, flags); +} + + +// internal itoa for 'long long' type +#if defined(PRINTF_SUPPORT_LONG_LONG) +static size_t _ntoa_long_long(out_fct_type out, char* buffer, size_t idx, size_t maxlen, unsigned long long value, bool negative, unsigned long long base, unsigned int prec, unsigned int width, unsigned int flags) +{ + char buf[PRINTF_NTOA_BUFFER_SIZE]; + size_t len = 0U; + + // no hash for 0 values + if (!value) { + flags &= ~FLAGS_HASH; + } + + // write if precision != 0 and value is != 0 + if (!(flags & FLAGS_PRECISION) || value) { + do { + const char digit = (char)(value % base); + buf[len++] = digit < 10 ? '0' + digit : (flags & FLAGS_UPPERCASE ? 'A' : 'a') + digit - 10; + value /= base; + } while (value && (len < PRINTF_NTOA_BUFFER_SIZE)); + } + + return _ntoa_format(out, buffer, idx, maxlen, buf, len, negative, (unsigned int)base, prec, width, flags); +} +#endif // PRINTF_SUPPORT_LONG_LONG + + +#if defined(PRINTF_SUPPORT_FLOAT) + +#if defined(PRINTF_SUPPORT_EXPONENTIAL) +// forward declaration so that _ftoa can switch to exp notation for values > PRINTF_MAX_FLOAT +static size_t _etoa(out_fct_type out, char* buffer, size_t idx, size_t maxlen, double value, unsigned int prec, unsigned int width, unsigned int flags); +#endif + + +// internal ftoa for fixed decimal floating point +static size_t _ftoa(out_fct_type out, char* buffer, size_t idx, size_t maxlen, double value, unsigned int prec, unsigned int width, unsigned int flags) +{ + char buf[PRINTF_FTOA_BUFFER_SIZE]; + size_t len = 0U; + double diff = 0.0; + + // powers of 10 + static const double pow10[] = { 1, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000, 1000000000 }; + + // test for special values + if (value != value) + return _out_rev(out, buffer, idx, maxlen, "nan", 3, width, flags); + if (value < -DBL_MAX) + return _out_rev(out, buffer, idx, maxlen, "fni-", 4, width, flags); + if (value > DBL_MAX) + return _out_rev(out, buffer, idx, maxlen, (flags & FLAGS_PLUS) ? "fni+" : "fni", (flags & FLAGS_PLUS) ? 4U : 3U, width, flags); + + // test for very large values + // standard printf behavior is to print EVERY whole number digit -- which could be 100s of characters overflowing your buffers == bad + if ((value > PRINTF_MAX_FLOAT) || (value < -PRINTF_MAX_FLOAT)) { +#if defined(PRINTF_SUPPORT_EXPONENTIAL) + return _etoa(out, buffer, idx, maxlen, value, prec, width, flags); +#else + return 0U; +#endif + } + + // test for negative + bool negative = false; + if (value < 0) { + negative = true; + value = 0 - value; + } + + // set default precision, if not set explicitly + if (!(flags & FLAGS_PRECISION)) { + prec = PRINTF_DEFAULT_FLOAT_PRECISION; + } + // limit precision to 9, cause a prec >= 10 can lead to overflow errors + while ((len < PRINTF_FTOA_BUFFER_SIZE) && (prec > 9U)) { + buf[len++] = '0'; + prec--; + } + + int whole = (int)value; + double tmp = (value - whole) * pow10[prec]; + unsigned long frac = (unsigned long)tmp; + diff = tmp - frac; + + if (diff > 0.5) { + ++frac; + // handle rollover, e.g. case 0.99 with prec 1 is 1.0 + if (frac >= pow10[prec]) { + frac = 0; + ++whole; + } + } + else if (diff < 0.5) { + } + else if ((frac == 0U) || (frac & 1U)) { + // if halfway, round up if odd OR if last digit is 0 + ++frac; + } + + if (prec == 0U) { + diff = value - (double)whole; + if ((!(diff < 0.5) || (diff > 0.5)) && (whole & 1)) { + // exactly 0.5 and ODD, then round up + // 1.5 -> 2, but 2.5 -> 2 + ++whole; + } + } + else { + unsigned int count = prec; + // now do fractional part, as an unsigned number + while (len < PRINTF_FTOA_BUFFER_SIZE) { + --count; + buf[len++] = (char)(48U + (frac % 10U)); + if (!(frac /= 10U)) { + break; + } + } + // add extra 0s + while ((len < PRINTF_FTOA_BUFFER_SIZE) && (count-- > 0U)) { + buf[len++] = '0'; + } + if (len < PRINTF_FTOA_BUFFER_SIZE) { + // add decimal + buf[len++] = '.'; + } + } + + // do whole part, number is reversed + while (len < PRINTF_FTOA_BUFFER_SIZE) { + buf[len++] = (char)(48 + (whole % 10)); + if (!(whole /= 10)) { + break; + } + } + + // pad leading zeros + if (!(flags & FLAGS_LEFT) && (flags & FLAGS_ZEROPAD)) { + if (width && (negative || (flags & (FLAGS_PLUS | FLAGS_SPACE)))) { + width--; + } + while ((len < width) && (len < PRINTF_FTOA_BUFFER_SIZE)) { + buf[len++] = '0'; + } + } + + if (len < PRINTF_FTOA_BUFFER_SIZE) { + if (negative) { + buf[len++] = '-'; + } + else if (flags & FLAGS_PLUS) { + buf[len++] = '+'; // ignore the space if the '+' exists + } + else if (flags & FLAGS_SPACE) { + buf[len++] = ' '; + } + } + + return _out_rev(out, buffer, idx, maxlen, buf, len, width, flags); +} + + +#if defined(PRINTF_SUPPORT_EXPONENTIAL) +// internal ftoa variant for exponential floating-point type, contributed by Martijn Jasperse +static size_t _etoa(out_fct_type out, char* buffer, size_t idx, size_t maxlen, double value, unsigned int prec, unsigned int width, unsigned int flags) +{ + // check for NaN and special values + if ((value != value) || (value > DBL_MAX) || (value < -DBL_MAX)) { + return _ftoa(out, buffer, idx, maxlen, value, prec, width, flags); + } + + // determine the sign + const bool negative = value < 0; + if (negative) { + value = -value; + } + + // default precision + if (!(flags & FLAGS_PRECISION)) { + prec = PRINTF_DEFAULT_FLOAT_PRECISION; + } + + // determine the decimal exponent + // based on the algorithm by David Gay (https://www.ampl.com/netlib/fp/dtoa.c) + union { + uint64_t U; + double F; + } conv; + + conv.F = value; + int exp2 = (int)((conv.U >> 52U) & 0x07FFU) - 1023; // effectively log2 + conv.U = (conv.U & ((1ULL << 52U) - 1U)) | (1023ULL << 52U); // drop the exponent so conv.F is now in [1,2) + // now approximate log10 from the log2 integer part and an expansion of ln around 1.5 + int expval = (int)(0.1760912590558 + exp2 * 0.301029995663981 + (conv.F - 1.5) * 0.289529654602168); + // now we want to compute 10^expval but we want to be sure it won't overflow + exp2 = (int)(expval * 3.321928094887362 + 0.5); + const double z = expval * 2.302585092994046 - exp2 * 0.6931471805599453; + const double z2 = z * z; + conv.U = (uint64_t)(exp2 + 1023) << 52U; + // compute exp(z) using continued fractions, see https://en.wikipedia.org/wiki/Exponential_function#Continued_fractions_for_ex + conv.F *= 1 + 2 * z / (2 - z + (z2 / (6 + (z2 / (10 + z2 / 14))))); + // correct for rounding errors + if (value < conv.F) { + expval--; + conv.F /= 10; + } + + // the exponent format is "%+03d" and largest value is "307", so set aside 4-5 characters + unsigned int minwidth = ((expval < 100) && (expval > -100)) ? 4U : 5U; + + // in "%g" mode, "prec" is the number of *significant figures* not decimals + if (flags & FLAGS_ADAPT_EXP) { + // do we want to fall-back to "%f" mode? + if ((value >= 1e-4) && (value < 1e6)) { + if ((int)prec > expval) { + prec = (unsigned)((int)prec - expval - 1); + } + else { + prec = 0; + } + flags |= FLAGS_PRECISION; // make sure _ftoa respects precision + // no characters in exponent + minwidth = 0U; + expval = 0; + } + else { + // we use one sigfig for the whole part + if ((prec > 0) && (flags & FLAGS_PRECISION)) { + --prec; + } + } + } + + // will everything fit? + unsigned int fwidth = width; + if (width > minwidth) { + // we didn't fall-back so subtract the characters required for the exponent + fwidth -= minwidth; + } else { + // not enough characters, so go back to default sizing + fwidth = 0U; + } + if ((flags & FLAGS_LEFT) && minwidth) { + // if we're padding on the right, DON'T pad the floating part + fwidth = 0U; + } + + // rescale the float value + if (expval) { + value /= conv.F; + } + + // output the floating part + const size_t start_idx = idx; + idx = _ftoa(out, buffer, idx, maxlen, negative ? -value : value, prec, fwidth, flags & ~FLAGS_ADAPT_EXP); + + // output the exponent part + if (minwidth) { + // output the exponential symbol + out((flags & FLAGS_UPPERCASE) ? 'E' : 'e', buffer, idx++, maxlen); + // output the exponent value + idx = _ntoa_long(out, buffer, idx, maxlen, (expval < 0) ? -expval : expval, expval < 0, 10, 0, minwidth-1, FLAGS_ZEROPAD | FLAGS_PLUS); + // might need to right-pad spaces + if (flags & FLAGS_LEFT) { + while (idx - start_idx < width) out(' ', buffer, idx++, maxlen); + } + } + return idx; +} +#endif // PRINTF_SUPPORT_EXPONENTIAL +#endif // PRINTF_SUPPORT_FLOAT + + +// internal vsnprintf +static int _vsnprintf(out_fct_type out, char* buffer, const size_t maxlen, const char* format, va_list va) +{ + unsigned int flags, width, precision, n; + size_t idx = 0U; + + if (!buffer) { + // use null output function + out = _out_null; + } + + while (*format) + { + // format specifier? %[flags][width][.precision][length] + if (*format != '%') { + // no + out(*format, buffer, idx++, maxlen); + format++; + continue; + } + else { + // yes, evaluate it + format++; + } + + // evaluate flags + flags = 0U; + do { + switch (*format) { + case '0': flags |= FLAGS_ZEROPAD; format++; n = 1U; break; + case '-': flags |= FLAGS_LEFT; format++; n = 1U; break; + case '+': flags |= FLAGS_PLUS; format++; n = 1U; break; + case ' ': flags |= FLAGS_SPACE; format++; n = 1U; break; + case '#': flags |= FLAGS_HASH; format++; n = 1U; break; + default : n = 0U; break; + } + } while (n); + + // evaluate width field + width = 0U; + if (_is_digit(*format)) { + width = _atoi(&format); + } + else if (*format == '*') { + const int w = va_arg(va, int); + if (w < 0) { + flags |= FLAGS_LEFT; // reverse padding + width = (unsigned int)-w; + } + else { + width = (unsigned int)w; + } + format++; + } + + // evaluate precision field + precision = 0U; + if (*format == '.') { + flags |= FLAGS_PRECISION; + format++; + if (_is_digit(*format)) { + precision = _atoi(&format); + } + else if (*format == '*') { + const int prec = (int)va_arg(va, int); + precision = prec > 0 ? (unsigned int)prec : 0U; + format++; + } + } + + // evaluate length field + switch (*format) { + case 'l' : + flags |= FLAGS_LONG; + format++; + if (*format == 'l') { + flags |= FLAGS_LONG_LONG; + format++; + } + break; + case 'h' : + flags |= FLAGS_SHORT; + format++; + if (*format == 'h') { + flags |= FLAGS_CHAR; + format++; + } + break; +#if defined(PRINTF_SUPPORT_PTRDIFF_T) + case 't' : + flags |= (sizeof(ptrdiff_t) == sizeof(long) ? FLAGS_LONG : FLAGS_LONG_LONG); + format++; + break; +#endif + case 'j' : + flags |= (sizeof(intmax_t) == sizeof(long) ? FLAGS_LONG : FLAGS_LONG_LONG); + format++; + break; + case 'z' : + flags |= (sizeof(size_t) == sizeof(long) ? FLAGS_LONG : FLAGS_LONG_LONG); + format++; + break; + default : + break; + } + + // evaluate specifier + switch (*format) { + case 'd' : + case 'i' : + case 'u' : + case 'x' : + case 'X' : + case 'o' : + case 'b' : { + // set the base + unsigned int base; + if (*format == 'x' || *format == 'X') { + base = 16U; + } + else if (*format == 'o') { + base = 8U; + } + else if (*format == 'b') { + base = 2U; + } + else { + base = 10U; + flags &= ~FLAGS_HASH; // no hash for dec format + } + // uppercase + if (*format == 'X') { + flags |= FLAGS_UPPERCASE; + } + + // no plus or space flag for u, x, X, o, b + if ((*format != 'i') && (*format != 'd')) { + flags &= ~(FLAGS_PLUS | FLAGS_SPACE); + } + + // ignore '0' flag when precision is given + if (flags & FLAGS_PRECISION) { + flags &= ~FLAGS_ZEROPAD; + } + + // convert the integer + if ((*format == 'i') || (*format == 'd')) { + // signed + if (flags & FLAGS_LONG_LONG) { +#if defined(PRINTF_SUPPORT_LONG_LONG) + const long long value = va_arg(va, long long); + idx = _ntoa_long_long(out, buffer, idx, maxlen, (unsigned long long)(value > 0 ? value : 0 - value), value < 0, base, precision, width, flags); +#endif + } + else if (flags & FLAGS_LONG) { + const long value = va_arg(va, long); + idx = _ntoa_long(out, buffer, idx, maxlen, (unsigned long)(value > 0 ? value : 0 - value), value < 0, base, precision, width, flags); + } + else { + const int value = (flags & FLAGS_CHAR) ? (char)va_arg(va, int) : (flags & FLAGS_SHORT) ? (short int)va_arg(va, int) : va_arg(va, int); + idx = _ntoa_long(out, buffer, idx, maxlen, (unsigned int)(value > 0 ? value : 0 - value), value < 0, base, precision, width, flags); + } + } + else { + // unsigned + if (flags & FLAGS_LONG_LONG) { +#if defined(PRINTF_SUPPORT_LONG_LONG) + idx = _ntoa_long_long(out, buffer, idx, maxlen, va_arg(va, unsigned long long), false, base, precision, width, flags); +#endif + } + else if (flags & FLAGS_LONG) { + idx = _ntoa_long(out, buffer, idx, maxlen, va_arg(va, unsigned long), false, base, precision, width, flags); + } + else { + const unsigned int value = (flags & FLAGS_CHAR) ? (unsigned char)va_arg(va, unsigned int) : (flags & FLAGS_SHORT) ? (unsigned short int)va_arg(va, unsigned int) : va_arg(va, unsigned int); + idx = _ntoa_long(out, buffer, idx, maxlen, value, false, base, precision, width, flags); + } + } + format++; + break; + } +#if defined(PRINTF_SUPPORT_FLOAT) + case 'f' : + case 'F' : + if (*format == 'F') flags |= FLAGS_UPPERCASE; + idx = _ftoa(out, buffer, idx, maxlen, va_arg(va, double), precision, width, flags); + format++; + break; +#if defined(PRINTF_SUPPORT_EXPONENTIAL) + case 'e': + case 'E': + case 'g': + case 'G': + if ((*format == 'g')||(*format == 'G')) flags |= FLAGS_ADAPT_EXP; + if ((*format == 'E')||(*format == 'G')) flags |= FLAGS_UPPERCASE; + idx = _etoa(out, buffer, idx, maxlen, va_arg(va, double), precision, width, flags); + format++; + break; +#endif // PRINTF_SUPPORT_EXPONENTIAL +#endif // PRINTF_SUPPORT_FLOAT + case 'c' : { + unsigned int l = 1U; + // pre padding + if (!(flags & FLAGS_LEFT)) { + while (l++ < width) { + out(' ', buffer, idx++, maxlen); + } + } + // char output + out((char)va_arg(va, int), buffer, idx++, maxlen); + // post padding + if (flags & FLAGS_LEFT) { + while (l++ < width) { + out(' ', buffer, idx++, maxlen); + } + } + format++; + break; + } + + case 's' : { + const char* p = va_arg(va, char*); + unsigned int l = _strnlen_s(p, precision ? precision : (size_t)-1); + // pre padding + if (flags & FLAGS_PRECISION) { + l = (l < precision ? l : precision); + } + if (!(flags & FLAGS_LEFT)) { + while (l++ < width) { + out(' ', buffer, idx++, maxlen); + } + } + // string output + while ((*p != 0) && (!(flags & FLAGS_PRECISION) || precision--)) { + out(*(p++), buffer, idx++, maxlen); + } + // post padding + if (flags & FLAGS_LEFT) { + while (l++ < width) { + out(' ', buffer, idx++, maxlen); + } + } + format++; + break; + } + + case 'p' : { + width = sizeof(void*) * 2U; + flags |= FLAGS_ZEROPAD | FLAGS_UPPERCASE; +#if defined(PRINTF_SUPPORT_LONG_LONG) + const bool is_ll = sizeof(uintptr_t) == sizeof(long long); + if (is_ll) { + idx = _ntoa_long_long(out, buffer, idx, maxlen, (uintptr_t)va_arg(va, void*), false, 16U, precision, width, flags); + } + else { +#endif + idx = _ntoa_long(out, buffer, idx, maxlen, (unsigned long)((uintptr_t)va_arg(va, void*)), false, 16U, precision, width, flags); +#if defined(PRINTF_SUPPORT_LONG_LONG) + } +#endif + format++; + break; + } + + case '%' : + out('%', buffer, idx++, maxlen); + format++; + break; + + default : + out(*format, buffer, idx++, maxlen); + format++; + break; + } + } + + // termination + out((char)0, buffer, idx < maxlen ? idx : maxlen - 1U, maxlen); + + // return written chars without terminating \0 + return (int)idx; +} + + +/////////////////////////////////////////////////////////////////////////////// + +int printf_(const char* format, ...) +{ + va_list va; + va_start(va, format); + char buffer[1]; + const int ret = _vsnprintf(_out_char, buffer, (size_t)-1, format, va); + va_end(va); + return ret; +} + + +int sprintf_(char* buffer, const char* format, ...) +{ + va_list va; + va_start(va, format); + const int ret = _vsnprintf(_out_buffer, buffer, (size_t)-1, format, va); + va_end(va); + return ret; +} + + +int snprintf_(char* buffer, size_t count, const char* format, ...) +{ + va_list va; + va_start(va, format); + const int ret = _vsnprintf(_out_buffer, buffer, count, format, va); + va_end(va); + return ret; +} + + +int vprintf_(const char* format, va_list va) +{ + char buffer[1]; + return _vsnprintf(_out_char, buffer, (size_t)-1, format, va); +} + + +int vsnprintf_(char* buffer, size_t count, const char* format, va_list va) +{ + return _vsnprintf(_out_buffer, buffer, count, format, va); +} + + +int fctprintf(void (*out)(char character, void* arg), void* arg, const char* format, ...) +{ + va_list va; + va_start(va, format); + const out_fct_wrap_type out_fct_wrap = { out, arg }; + const int ret = _vsnprintf(_out_fct, (char*)(uintptr_t)&out_fct_wrap, (size_t)-1, format, va); + va_end(va); + return ret; +} diff --git a/example/printf.h b/example/printf.h new file mode 100644 index 0000000..6104ccf --- /dev/null +++ b/example/printf.h @@ -0,0 +1,117 @@ +/////////////////////////////////////////////////////////////////////////////// +// \author (c) Marco Paland (info@paland.com) +// 2014-2019, PALANDesign Hannover, Germany +// +// \license The MIT License (MIT) +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. +// +// \brief Tiny printf, sprintf and snprintf implementation, optimized for speed on +// embedded systems with a very limited resources. +// Use this instead of bloated standard/newlib printf. +// These routines are thread safe and reentrant. +// +/////////////////////////////////////////////////////////////////////////////// + +#ifndef _PRINTF_H_ +#define _PRINTF_H_ + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** + * Output a character to a custom device like UART, used by the printf() function + * This function is declared here only. You have to write your custom implementation somewhere + * \param character Character to output + */ +void _putchar(char character); + + +/** + * Tiny printf implementation + * You have to implement _putchar if you use printf() + * To avoid conflicts with the regular printf() API it is overridden by macro defines + * and internal underscore-appended functions like printf_() are used + * \param format A string that specifies the format of the output + * \return The number of characters that are written into the array, not counting the terminating null character + */ +#define printf printf_ +int printf_(const char* format, ...); + + +/** + * Tiny sprintf implementation + * Due to security reasons (buffer overflow) YOU SHOULD CONSIDER USING (V)SNPRINTF INSTEAD! + * \param buffer A pointer to the buffer where to store the formatted string. MUST be big enough to store the output! + * \param format A string that specifies the format of the output + * \return The number of characters that are WRITTEN into the buffer, not counting the terminating null character + */ +#define sprintf sprintf_ +int sprintf_(char* buffer, const char* format, ...); + + +/** + * Tiny snprintf/vsnprintf implementation + * \param buffer A pointer to the buffer where to store the formatted string + * \param count The maximum number of characters to store in the buffer, including a terminating null character + * \param format A string that specifies the format of the output + * \param va A value identifying a variable arguments list + * \return The number of characters that COULD have been written into the buffer, not counting the terminating + * null character. A value equal or larger than count indicates truncation. Only when the returned value + * is non-negative and less than count, the string has been completely written. + */ +#define snprintf snprintf_ +#define vsnprintf vsnprintf_ +int snprintf_(char* buffer, size_t count, const char* format, ...); +int vsnprintf_(char* buffer, size_t count, const char* format, va_list va); + + +/** + * Tiny vprintf implementation + * \param format A string that specifies the format of the output + * \param va A value identifying a variable arguments list + * \return The number of characters that are WRITTEN into the buffer, not counting the terminating null character + */ +#define vprintf vprintf_ +int vprintf_(const char* format, va_list va); + + +/** + * printf with output function + * You may use this as dynamic alternative to printf() with its fixed _putchar() output + * \param out An output function which takes one character and an argument pointer + * \param arg An argument pointer for user data passed to output function + * \param format A string that specifies the format of the output + * \return The number of characters that are sent to the output function, not counting the terminating null character + */ +int fctprintf(void (*out)(char character, void* arg), void* arg, const char* format, ...); + + +#ifdef __cplusplus +} +#endif + + +#endif // _PRINTF_H_ diff --git a/example/putchar.c b/example/putchar.c new file mode 100644 index 0000000..7753ccb --- /dev/null +++ b/example/putchar.c @@ -0,0 +1,4 @@ +void _putchar(char character) { + char* uart_thr = (char*)0x10000000; + *uart_thr = character; +} diff --git a/src/debug.cpp b/src/debug.cpp index 47cb85b..8fda93c 100644 --- a/src/debug.cpp +++ b/src/debug.cpp @@ -190,7 +190,10 @@ void GDBStub::handle_packet(const std::string &packet) { break; case 's': - vm.step(); + try { + vm.step(); + } catch (const EbreakException &ex) { + } send_packet("S05"); // TODO: step execution break; @@ -222,7 +225,7 @@ void GDBStub::handle_packet(const std::string &packet) { // Insert breakpoint if (packet[1] == '0') { uint32_t addr = - std::stoul(packet.substr(3, packet.find(',')), nullptr, 16); + std::stoul(packet.substr(3, packet.find(',', 3)), nullptr, 16); if (breakpoints.count(addr) == 0) { uint32_t original_instr = vm.read_memory_word(addr); @@ -237,7 +240,7 @@ void GDBStub::handle_packet(const std::string &packet) { // Delete breakpoint if (packet[1] == '0') { uint32_t addr = - std::stoul(packet.substr(3, packet.find(',')), nullptr, 16); + std::stoul(packet.substr(3, packet.find(',', 3)), nullptr, 16); if (breakpoints.count(addr) > 0) { // Restore the original instruction diff --git a/src/elf.cpp b/src/elf.cpp index ea65438..3062f41 100644 --- a/src/elf.cpp +++ b/src/elf.cpp @@ -95,7 +95,8 @@ std::vector load_elf(const std::string& filename, size_t memory_size) { for (const Elf32Section& shdr : sectionHeaders) { const char* sectionName = §ionStrTable[shdr.sh_name]; if (std::strcmp(sectionName, ".text") == 0 || - std::strcmp(sectionName, ".sdata") == 0) { + std::strcmp(sectionName, ".sdata") == 0 || + std::strcmp(sectionName, ".rodata") == 0) { memoryEnd = std::max(memoryEnd, shdr.sh_addr + shdr.sh_size); } } @@ -110,7 +111,8 @@ std::vector load_elf(const std::string& filename, size_t memory_size) { for (const Elf32Section& shdr : sectionHeaders) { const char* sectionName = §ionStrTable[shdr.sh_name]; if (std::strcmp(sectionName, ".text") == 0 || - std::strcmp(sectionName, ".sdata") == 0) { + std::strcmp(sectionName, ".sdata") == 0 || + std::strcmp(sectionName, ".rodata") == 0) { std::vector sectionData(shdr.sh_size); file.seekg(shdr.sh_offset); file.read(reinterpret_cast(&loadedData[shdr.sh_addr]), diff --git a/src/rve.cpp b/src/rve.cpp index 677218c..a8c1abe 100644 --- a/src/rve.cpp +++ b/src/rve.cpp @@ -10,7 +10,7 @@ #include "vm.hpp" int main(int argc, char *argv[]) { - const size_t MEMORY_SIZE = 128 * 1024; + const size_t MEMORY_SIZE = 512 * 1024; bool debug = false; std::string program_filename = ""; @@ -46,10 +46,6 @@ int main(int argc, char *argv[]) { std::cerr << "Emulator error: " << e.what() << std::endl; return 1; } - std::vector res_mem = vm.read_memory(0x1000, 4); - uint32_t *res = (uint32_t *)&res_mem[0]; - - std::cout << "result: " << *res << std::endl; } else { // to debug, do: "set debug remote 1" in gdb // and then "target remote :1234" diff --git a/src/vm.cpp b/src/vm.cpp index 8e34c17..b5f86f4 100644 --- a/src/vm.cpp +++ b/src/vm.cpp @@ -12,9 +12,39 @@ inline int32_t sign_extend(int32_t value, int bits) { return (value ^ mask) - mask; } +uint8_t UART::read_register(uint32_t address) { + switch (address) { + case UART_LSR: + // Always ready to transmit + return LSR_TRANSMITTER_EMPTY; + default: + return 0; + } +} + +void UART::write_register(uint32_t address, uint8_t value) { + switch (address) { + case UART_THR: + std::cout.put(static_cast(value)); + break; + } +} + +bool UART::is_transmitter_ready() { + return read_register(UART_LSR) & LSR_TRANSMITTER_EMPTY; +} + VM::VM(const std::vector& memory, const std::string& file_path) : memory_(memory), file_path(file_path) {} +void VM::setreg(int regnum, uint32_t value) { + if (regnum == 0) { + return; + } + + registers[regnum] = value; +} + std::vector VM::read_memory(size_t start, size_t size) { if (start + size > memory_.size()) { return std::vector(size, 0); @@ -23,12 +53,64 @@ std::vector VM::read_memory(size_t start, size_t size) { memory_.begin() + start + size); } -uint32_t VM::read_memory_word(size_t pos) { return *(uint32_t*)&memory_[pos]; } +uint32_t VM::read_memory_word(size_t pos) { + if (pos + 3 >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } + return *(uint32_t*)&memory_[pos]; +} + +uint16_t VM::read_memory_half_word(size_t pos) { + if (pos + 1 >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } + + return *(uint16_t*)&memory_[pos]; +} + +uint8_t VM::read_memory_byte(size_t pos) { + if (pos >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } + + return memory_[pos]; +} void VM::write_memory_word(size_t pos, uint32_t value) { + if (pos + 1 >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } *(uint32_t*)&memory_[pos] = value; } +void VM::write_memory_half_word(size_t pos, uint16_t value) { + if (pos + 3 >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } + *(uint16_t*)&memory_[pos] = value; +} + +void VM::write_memory_byte(size_t pos, uint8_t value) { + if (is_mmap(pos, 1)) { + if (pos >= UART_ADDR && pos < UART_ADDR + 8) { + uart.write_register(pos - UART_ADDR, value); + } + return; + } + + if (pos >= memory_.size()) { + throw std::runtime_error("Memory access out of bounds"); + } + memory_[pos] = value; +} + +bool VM::is_mmap(size_t pos, size_t size) { + if (pos + size < UART_ADDR) return false; + if (pos >= UART_ADDR + 8) return false; + + return (pos < UART_ADDR + 8) && (pos + size >= UART_ADDR - 1); +} + uint32_t VM::read_register(size_t regnum) { if (regnum == 32) return pc; @@ -42,12 +124,9 @@ uint32_t VM::read_register(size_t regnum) { const std::string& VM::get_file_path() { return file_path; } void VM::step() { - size_t memory_size = memory_.size(); - uint8_t* memory = &memory_[0]; - - uint32_t instr = *(uint32_t*)&memory[pc]; + uint32_t instr = *(uint32_t*)&memory_[pc]; // std::cout << "pc: " << std::hex << pc << std::dec << "\n"; - // std::cout << "instr: " << std::hex << instr << "\n"; + // std::cout << "instr: " << std::hex << instr << "\n"; pc += 4; // Decode instruction @@ -63,26 +142,26 @@ void VM::step() { case 0x33: { // R-type if (funct7 == 0x00) { if (funct3 == 0x0) { // ADD - registers[rd] = registers[rs1] + registers[rs2]; + setreg(rd, registers[rs1] + registers[rs2]); } else if (funct3 == 0x04) { // XOR - registers[rd] = registers[rs1] ^ registers[rs2]; + setreg(rd, registers[rs1] ^ registers[rs2]); } else if (funct3 == 0x06) { // OR - registers[rd] = registers[rs1] | registers[rs2]; + setreg(rd, registers[rd] = registers[rs1] | registers[rs2]); } else if (funct3 == 0x07) { // AND - registers[rd] = registers[rs1] & registers[rs2]; + setreg(rd, registers[rs1] & registers[rs2]); } else if (funct3 == 0x01) { // SLL - registers[rd] = registers[rs1] << registers[rs2]; + setreg(rd, registers[rs1] << registers[rs2]); } else if (funct3 == 0x05) { // SRL uint32_t value = registers[rs1]; uint32_t shift_amount = registers[rs2] & 0x1F; - registers[rd] = value << shift_amount; + setreg(rd, value >> shift_amount); } else if (funct3 == 0x02) { // SLT registers[rd] = (static_cast(registers[rs1]) < static_cast(registers[rs2])) ? 0 : 1; } else if (funct3 == 0x03) { // SLTU - registers[rd] = (registers[rs1] < registers[rs2]) ? 1 : 0; + setreg(rd, (registers[rs1] < registers[rs2]) ? 1 : 0); } else { throw std::runtime_error("Unknown R-type instruction"); } @@ -93,7 +172,7 @@ void VM::step() { // Only the lower 5 bits are used for shift int32_t value = static_cast(registers[rs1]); int32_t shift_amount = registers[rs2] & 0x1F; - registers[rd] = value >> shift_amount; + setreg(rd, value >> shift_amount); } else { throw std::runtime_error("Unknown R-type instruction"); } @@ -102,50 +181,50 @@ void VM::step() { int64_t result = static_cast(static_cast(registers[rs1])) * static_cast(static_cast(registers[rs2])); - registers[rd] = static_cast(result); + setreg(rd, static_cast(result)); } else if (funct3 == 0x1) { // MULH int64_t result = static_cast(static_cast(registers[rs1])) * static_cast(static_cast(registers[rs2])); - registers[rd] = static_cast(result >> 32); + setreg(rd, static_cast(result >> 32)); } else if (funct3 == 0x2) { // MULSU int64_t result = static_cast(static_cast(registers[rs1])) * static_cast(registers[rs2]); - registers[rd] = static_cast(result >> 32); + setreg(rd, static_cast(result >> 32)); } else if (funct3 == 0x3) { // MULU uint64_t result = static_cast(registers[rs1]) * static_cast(registers[rs2]); - registers[rd] = static_cast(result >> 32); // Upper 32 bits - } else if (funct3 == 0x4) { // DIV + setreg(rd, static_cast(result >> 32)); // Upper 32 bits + } else if (funct3 == 0x4) { // DIV int32_t dividend = static_cast(registers[rs1]); int32_t divisor = static_cast(registers[rs2]); if (divisor == 0) { - registers[rd] = -1; // Division by zero result + setreg(rd, -1); // Division by zero result } else if (dividend == INT32_MIN && divisor == -1) { - registers[rd] = dividend; // Overflow case + setreg(rd, dividend); // Overflow case } else { - registers[rd] = dividend / divisor; + setreg(rd, dividend / divisor); } } else if (funct3 == 0x5) { // DIVU uint32_t dividend = registers[rs1]; uint32_t divisor = registers[rs2]; - registers[rd] = (divisor == 0) ? UINT32_MAX : dividend / divisor; + setreg(rd, (divisor == 0) ? UINT32_MAX : dividend / divisor); } else if (funct3 == 0x6) { // REM int32_t dividend = static_cast(registers[rs1]); int32_t divisor = static_cast(registers[rs2]); if (divisor == 0) { - registers[rd] = - dividend; // Remainder with zero divisor is the dividend + setreg(rd, + dividend); // Remainder with zero divisor is the dividend } else if (dividend == INT32_MIN && divisor == -1) { - registers[rd] = 0; // Overflow case + setreg(rd, 0); // Overflow case } else { - registers[rd] = dividend % divisor; + setreg(rd, dividend % divisor); } } else if (funct3 == 0x7) { // REMU uint32_t dividend = registers[rs1]; uint32_t divisor = registers[rs2]; - registers[rd] = (divisor == 0) ? dividend : dividend % divisor; + setreg(rd, (divisor == 0) ? dividend : dividend % divisor); } else { throw std::runtime_error("Unknown R-type instruction"); } @@ -154,20 +233,48 @@ void VM::step() { } break; } - case 0x13: { // I-type (ADDI) + case 0x13: { // I-type (ADDI and friends) imm = sign_extend(instr >> 20, 12); // Extract 12-bit immediate if (funct3 == 0x0) { // ADDI - registers[rd] = registers[rs1] + imm; + setreg(rd, registers[rs1] + imm); + } else if (funct3 == 0x4) { // XORI + setreg(rd, registers[rs1] ^ imm); + } else if (funct3 == 0x6) { // ORI + setreg(rd, registers[rs1] | imm); + } else if (funct3 == 0x07) { // ANDI + setreg(rd, registers[rs1] & imm); + } else if (funct3 == 0x01) { + if (((imm >> 5) & 0x7f) == 0x0) { // SLLI + uint32_t value = registers[rs1]; + uint32_t shift_amount = imm & 0x1F; + setreg(rd, value << shift_amount); + } else { + throw std::runtime_error("Unknown I-type instruction"); + } + } else if (funct3 == 0x05) { + if (((imm >> 5) & 0x7f) == 0x20) { // SRAI + int32_t value = static_cast(imm & 0x1f); + int32_t shift_amount = imm & 0x1F; + setreg(rd, value >> shift_amount); + } else if (((imm >> 5) & 0x7f) == 0x0) { // SRLI + uint32_t value = registers[rs1]; + uint32_t shift_amount = imm & 0x1F; + setreg(rd, value >> shift_amount); + } else { + throw std::runtime_error("Unknown I-type instruction"); + } } else { throw std::runtime_error("Unknown I-type instruction"); } break; } case 0x63: { // B-type (branches) - imm = ((instr >> 7) & 0x1E) | ((instr >> 20) & 0x7E0) | - ((instr >> 19) & 0x800) | ((instr >> 31) << 12); - imm = sign_extend(imm, 13); // Sign-extend 13-bit immediate - if (funct3 == 0x0) { // BEQ + imm = ((int64_t)(int32_t)(instr & 0x80000000) >> 19) | + ((instr & 0x80) << 4) // imm[11] + | ((instr >> 20) & 0x7e0) // imm[10:5] + | ((instr >> 7) & 0x1e); + + if (funct3 == 0x0) { // BEQ if (registers[rs1] == registers[rs2]) { pc += imm - 4; // Offset PC (adjust for pre-increment) } @@ -198,24 +305,21 @@ void VM::step() { imm = sign_extend(instr >> 20, 12); // Extract 12-bit immediate if (funct3 == 0x00) { // LB uint32_t addr = registers[rs1] + imm; - if (addr + 1 > memory_size) { - throw std::runtime_error("Memory access out of bounds"); - } - registers[rd] = 0; - std::memcpy(®isters[rd], memory + addr, sizeof(uint8_t)); + // registers[rd] = sign_extend(read_memory_byte(addr), 8); + setreg(rd, read_memory_byte(addr)); } else if (funct3 == 0x01) { // LH uint32_t addr = registers[rs1] + imm; - if (addr + 2 > memory_size) { - throw std::runtime_error("Memory access out of bounds"); - } - registers[rd] = 0; - std::memcpy(®isters[rd], memory + addr, sizeof(uint16_t)); + // registers[rd] = sign_extend(read_memory_half_word(addr), 16); + setreg(rd, read_memory_half_word(addr)); } else if (funct3 == 0x2) { // LW uint32_t addr = registers[rs1] + imm; - if (addr + 4 > memory_size) { - throw std::runtime_error("Memory access out of bounds"); - } - std::memcpy(®isters[rd], memory + addr, sizeof(uint32_t)); + setreg(rd, read_memory_word(addr)); + } else if (funct3 == 0x4) { // LBU + uint32_t addr = registers[rs1] + imm; + setreg(rd, read_memory_byte(addr)); + } else if (funct3 == 0x5) { // LHU + uint32_t addr = registers[rs1] + imm; + setreg(rd, read_memory_half_word(addr)); } else { throw std::runtime_error("Unknown load instruction"); } @@ -226,22 +330,19 @@ void VM::step() { imm = sign_extend(imm, 12); // Sign-extend 12-bit immediate if (funct3 == 0x0) { // SB uint32_t addr = registers[rs1] + imm; - if (addr + 1 > memory_size) { - throw std::runtime_error("Memory access out of bounds"); - } - std::memcpy(memory + addr, ®isters[rs2], sizeof(uint8_t)); + write_memory_byte(addr, registers[rs2]); } else if (funct3 == 0x1) { // SH uint32_t addr = registers[rs1] + imm; - if (addr + 2 > memory_size) { + if (addr + 2 > memory_.size()) { throw std::runtime_error("Memory access out of bounds"); } - std::memcpy(memory + addr, ®isters[rs2], sizeof(uint16_t)); + std::memcpy(&memory_[addr], ®isters[rs2], sizeof(uint16_t)); } else if (funct3 == 0x2) { // SW uint32_t addr = registers[rs1] + imm; - if (addr + 4 > memory_size) { + if (addr + 4 > memory_.size()) { throw std::runtime_error("Memory access out of bounds"); } - std::memcpy(memory + addr, ®isters[rs2], sizeof(uint32_t)); + std::memcpy(&memory_[addr], ®isters[rs2], sizeof(uint32_t)); } else { throw std::runtime_error("Unknown store instruction"); } @@ -255,7 +356,7 @@ void VM::step() { ((instr >> 20) & 0x1) << 11 | // imm[11] ((instr & 0xFF000)); // imm[19:12] - registers[rd] = pc; // Save return address + setreg(rd, pc); // Save return address pc += offset - 4; break; } @@ -264,22 +365,22 @@ void VM::step() { uint32_t target = (registers[rs1] + offset) & ~1; // Target address (LSB cleared) - registers[rd] = pc; // Save return address + setreg(rd, pc); // Save return address pc = target; break; } case 0x37: { // LUI uint32_t imm = (instr >> 12) & 0xFFFFF; // Extract 20-bit immediate - registers[rd] = - imm - << 12; // Shift the immediate to the upper 20 bits of the register + setreg(rd, + imm << 12); // Shift the immediate to the upper 20 bits of the + // register break; } case 0x17: { // AUIPC uint32_t imm = (instr >> 12) & 0xFFFFF; // Extract 20-bit immediate - registers[rd] = - pc + - (imm << 12); // Add the immediate (shifted left) to the current PC + setreg(rd, + pc - 4 + (imm << 12)); // Add the immediate (shifted left) to + // the current PC break; } case 0x73: { // EBREAK diff --git a/src/vm.hpp b/src/vm.hpp index b81652d..75c62cd 100644 --- a/src/vm.hpp +++ b/src/vm.hpp @@ -8,6 +8,27 @@ class EbreakException : std::exception {}; const int NUM_REGISTERS = 32; // Standard RISC-V has 32 registers +const int UART_ADDR = 0x10000000; + +class UART { + public: + uint8_t read_register(uint32_t address); + void write_register(uint32_t address, uint8_t value); + + bool is_transmitter_ready(); + + private: + enum Registers { + UART_RBR = 0x00, // Receiver Buffer Register + UART_THR = 0x00, // Transmitter Holding Register + UART_LSR = 0x05 // Line Status Register + }; + + // Line Status Register bits + enum LSRBits { LSR_TRANSMITTER_EMPTY = 0x20 }; + + uint8_t registers[8] = {0}; +}; class VM { public: @@ -17,17 +38,29 @@ class VM { void eval(); std::vector read_memory(size_t start, size_t size); + uint32_t read_memory_word(size_t pos); + uint16_t read_memory_half_word(size_t pos); + uint8_t read_memory_byte(size_t pos); + void write_memory_word(size_t pos, uint32_t value); + void write_memory_half_word(size_t pos, uint16_t value); + void write_memory_byte(size_t pos, uint8_t value); + + bool is_mmap(size_t pos, size_t size); uint32_t read_register(size_t regnum); const std::string &get_file_path(); + void setreg(int regnum, uint32_t value); + private: std::vector memory_; uint32_t registers[NUM_REGISTERS] = {0}; uint32_t pc = 0; std::string file_path; + + UART uart; };