diff options
Diffstat (limited to 'io_buffer.c')
| -rw-r--r-- | io_buffer.c | 519 |
1 files changed, 416 insertions, 103 deletions
diff --git a/io_buffer.c b/io_buffer.c index 96f13c364a..dc8f547a32 100644 --- a/io_buffer.c +++ b/io_buffer.c @@ -39,6 +39,7 @@ size_t RUBY_IO_BUFFER_DEFAULT_SIZE; enum { RB_IO_BUFFER_HEXDUMP_DEFAULT_WIDTH = 16, + RB_IO_BUFFER_HEXDUMP_MAXIMUM_WIDTH = 1024, RB_IO_BUFFER_INSPECT_HEXDUMP_MAXIMUM_SIZE = 256, RB_IO_BUFFER_INSPECT_HEXDUMP_WIDTH = 16, @@ -273,7 +274,7 @@ io_buffer_free(struct rb_io_buffer *buffer) } static void -rb_io_buffer_type_mark_and_move(void *_buffer) +rb_io_buffer_type_mark(void *_buffer) { struct rb_io_buffer *buffer = _buffer; if (buffer->source != Qnil) { @@ -282,7 +283,21 @@ rb_io_buffer_type_mark_and_move(void *_buffer) // which can be otherwise moved by GC compaction. rb_gc_mark(buffer->source); } else { - rb_gc_mark_and_move(&buffer->source); + rb_gc_mark_movable(buffer->source); + } + } +} + +static void +rb_io_buffer_type_compact(void *_buffer) +{ + struct rb_io_buffer *buffer = _buffer; + if (buffer->source != Qnil) { + if (RB_TYPE_P(buffer->source, T_STRING)) { + // The `source` String has to be pinned, because the `base` may point to the embedded String content, + // which can be otherwise moved by GC compaction. + } else { + buffer->source = rb_gc_location(buffer->source); } } } @@ -311,10 +326,10 @@ rb_io_buffer_type_size(const void *_buffer) static const rb_data_type_t rb_io_buffer_type = { .wrap_struct_name = "IO::Buffer", .function = { - .dmark = rb_io_buffer_type_mark_and_move, + .dmark = rb_io_buffer_type_mark, .dfree = rb_io_buffer_type_free, .dsize = rb_io_buffer_type_size, - .dcompact = rb_io_buffer_type_mark_and_move, + .dcompact = rb_io_buffer_type_compact, }, .data = NULL, .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_EMBEDDABLE, @@ -370,7 +385,7 @@ io_buffer_extract_size(VALUE argument) } // Extract a width argument, which must be a non-negative integer, and must be -// at least the given minimum. +// at least the given minimum and at most RB_IO_BUFFER_HEXDUMP_MAXIMUM_WIDTH. static inline size_t io_buffer_extract_width(VALUE argument, size_t minimum) { @@ -384,6 +399,10 @@ io_buffer_extract_width(VALUE argument, size_t minimum) rb_raise(rb_eArgError, "Width must be at least %" PRIuSIZE "!", minimum); } + if (width > RB_IO_BUFFER_HEXDUMP_MAXIMUM_WIDTH) { + rb_raise(rb_eArgError, "Width must be at most %" PRIuSIZE "!", (size_t)RB_IO_BUFFER_HEXDUMP_MAXIMUM_WIDTH); + } + return width; } @@ -465,6 +484,8 @@ io_buffer_extract_offset_length(VALUE self, int argc, VALUE argv[], size_t *offs VALUE rb_io_buffer_type_allocate(VALUE self) { + io_buffer_experimental(); + struct rb_io_buffer *buffer = NULL; VALUE instance = TypedData_Make_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, buffer); @@ -555,7 +576,7 @@ io_buffer_for_yield_instance_ensure(VALUE _arguments) * buffer.get_string(0, 1) * # => "t" * string - * # => "best" + * # => "test" * * buffer.resize(100) * # in `resize': Cannot resize external buffer! (IO::Buffer::AccessError) @@ -635,8 +656,6 @@ rb_io_buffer_new(void *base, size_t size, enum rb_io_buffer_flags flags) VALUE rb_io_buffer_map(VALUE io, size_t size, rb_off_t offset, enum rb_io_buffer_flags flags) { - io_buffer_experimental(); - VALUE instance = rb_io_buffer_type_allocate(rb_cIOBuffer); struct rb_io_buffer *buffer = NULL; @@ -653,18 +672,25 @@ rb_io_buffer_map(VALUE io, size_t size, rb_off_t offset, enum rb_io_buffer_flags * call-seq: IO::Buffer.map(file, [size, [offset, [flags]]]) -> io_buffer * * Create an IO::Buffer for reading from +file+ by memory-mapping the file. - * +file_io+ should be a +File+ instance, opened for reading. + * +file+ should be a +File+ instance, opened for reading or reading and writing. * * Optional +size+ and +offset+ of mapping can be specified. + * Trying to map an empty file or specify +size+ of 0 will raise an error. + * Valid values for +offset+ are system-dependent. + * + * By default, the buffer is writable and expects the file to be writable. + * It is also shared, so several processes can use the same mapping. * - * By default, the buffer would be immutable (read only); to create a writable - * mapping, you need to open a file in read-write mode, and explicitly pass - * +flags+ argument without IO::Buffer::IMMUTABLE. + * You can pass IO::Buffer::READONLY in +flags+ argument to make a read-only buffer; + * this allows to work with files opened only for reading. + * Specifying IO::Buffer::PRIVATE in +flags+ creates a private mapping, + * which will not impact other processes or the underlying file. + * It also allows updating a buffer created from a read-only file. * * File.write('test.txt', 'test') * * buffer = IO::Buffer.map(File.open('test.txt'), nil, 0, IO::Buffer::READONLY) - * # => #<IO::Buffer 0x00000001014a0000+4 MAPPED READONLY> + * # => #<IO::Buffer 0x00000001014a0000+4 EXTERNAL MAPPED FILE SHARED READONLY> * * buffer.readonly? # => true * @@ -672,7 +698,7 @@ rb_io_buffer_map(VALUE io, size_t size, rb_off_t offset, enum rb_io_buffer_flags * # => "test" * * buffer.set_string('b', 0) - * # `set_string': Buffer is not writable! (IO::Buffer::AccessError) + * # 'IO::Buffer#set_string': Buffer is not writable! (IO::Buffer::AccessError) * * # create read/write mapping: length 4 bytes, offset 0, flags 0 * buffer = IO::Buffer.map(File.open('test.txt', 'r+'), 4, 0) @@ -694,31 +720,48 @@ io_buffer_map(int argc, VALUE *argv, VALUE klass) // We might like to handle a string path? VALUE io = argv[0]; + rb_off_t file_size = rb_file_size(io); + // Compiler can confirm that we handled file_size <= 0 case: + if (UNLIKELY(file_size <= 0)) { + rb_raise(rb_eArgError, "Invalid negative or zero file size!"); + } + // Here, we assume that file_size is positive: + else if (UNLIKELY((uintmax_t)file_size > SIZE_MAX)) { + rb_raise(rb_eArgError, "File larger than address space!"); + } + size_t size; if (argc >= 2 && !RB_NIL_P(argv[1])) { size = io_buffer_extract_size(argv[1]); - } - else { - rb_off_t file_size = rb_file_size(io); - - // Compiler can confirm that we handled file_size < 0 case: - if (file_size < 0) { - rb_raise(rb_eArgError, "Invalid negative file size!"); + if (UNLIKELY(size == 0)) { + rb_raise(rb_eArgError, "Size can't be zero!"); } - // Here, we assume that file_size is positive: - else if ((uintmax_t)file_size > SIZE_MAX) { - rb_raise(rb_eArgError, "File larger than address space!"); - } - else { - // This conversion should be safe: - size = (size_t)file_size; + if (UNLIKELY(size > (size_t)file_size)) { + rb_raise(rb_eArgError, "Size can't be larger than file size!"); } } + else { + // This conversion should be safe: + size = (size_t)file_size; + } // This is the file offset, not the buffer offset: rb_off_t offset = 0; if (argc >= 3) { offset = NUM2OFFT(argv[2]); + if (UNLIKELY(offset < 0)) { + rb_raise(rb_eArgError, "Offset can't be negative!"); + } + if (UNLIKELY(offset >= file_size)) { + rb_raise(rb_eArgError, "Offset too large!"); + } + if (RB_NIL_P(argv[1])) { + // Decrease size if it's set from the actual file size: + size = (size_t)(file_size - offset); + } + else if (UNLIKELY((size_t)(file_size - offset) < size)) { + rb_raise(rb_eArgError, "Offset too large!"); + } } enum rb_io_buffer_flags flags = 0; @@ -767,8 +810,6 @@ io_flags_for_size(size_t size) VALUE rb_io_buffer_initialize(int argc, VALUE *argv, VALUE self) { - io_buffer_experimental(); - rb_check_arity(argc, 0, 2); struct rb_io_buffer *buffer = NULL; @@ -858,8 +899,8 @@ rb_io_buffer_get_bytes(VALUE self, void **base, size_t *size) } // Internal function for accessing bytes for writing, wil -static inline void -io_buffer_get_bytes_for_writing(struct rb_io_buffer *buffer, void **base, size_t *size) +static void +io_buffer_validate_for_writing(struct rb_io_buffer *buffer) { if (buffer->flags & RB_IO_BUFFER_READONLY || (!NIL_P(buffer->source) && OBJ_FROZEN(buffer->source))) { @@ -869,6 +910,21 @@ io_buffer_get_bytes_for_writing(struct rb_io_buffer *buffer, void **base, size_t if (!io_buffer_validate(buffer)) { rb_raise(rb_eIOBufferInvalidatedError, "Buffer is invalid!"); } +} + +static struct rb_io_buffer * +get_io_buffer_for_writing(VALUE self) +{ + struct rb_io_buffer *buffer = NULL; + TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, buffer); + io_buffer_validate_for_writing(buffer); + return buffer; +} + +static inline void +io_buffer_get_bytes_for_writing(struct rb_io_buffer *buffer, void **base, size_t *size) +{ + io_buffer_validate_for_writing(buffer); if (buffer->base) { *base = buffer->base; @@ -889,11 +945,17 @@ rb_io_buffer_get_bytes_for_writing(VALUE self, void **base, size_t *size) } static void -io_buffer_get_bytes_for_reading(struct rb_io_buffer *buffer, const void **base, size_t *size) +io_buffer_validate_for_reading(struct rb_io_buffer *buffer) { if (!io_buffer_validate(buffer)) { rb_raise(rb_eIOBufferInvalidatedError, "Buffer has been invalidated!"); } +} + +static void +io_buffer_get_bytes_for_reading(struct rb_io_buffer *buffer, const void **base, size_t *size) +{ + io_buffer_validate_for_reading(buffer); if (buffer->base) { *base = buffer->base; @@ -1332,14 +1394,23 @@ io_buffer_readonly_p(VALUE self) return RBOOL(rb_io_buffer_readonly_p(self)); } -static void -io_buffer_lock(struct rb_io_buffer *buffer) +static int +io_buffer_try_lock(struct rb_io_buffer *buffer) { if (buffer->flags & RB_IO_BUFFER_LOCKED) { - rb_raise(rb_eIOBufferLockedError, "Buffer already locked!"); + return 0; } buffer->flags |= RB_IO_BUFFER_LOCKED; + return 1; +} + +static void +io_buffer_lock(struct rb_io_buffer *buffer) +{ + if (!io_buffer_try_lock(buffer)) { + rb_raise(rb_eIOBufferLockedError, "Buffer already locked!"); + } } VALUE @@ -1388,6 +1459,17 @@ rb_io_buffer_try_unlock(VALUE self) return 0; } +static VALUE +rb_io_buffer_locked_ensure(VALUE self) +{ + struct rb_io_buffer *buffer = NULL; + TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, buffer); + + buffer->flags &= ~RB_IO_BUFFER_LOCKED; + + return Qnil; +} + /* * call-seq: locked { ... } * @@ -1430,11 +1512,7 @@ rb_io_buffer_locked(VALUE self) buffer->flags |= RB_IO_BUFFER_LOCKED; - VALUE result = rb_yield(self); - - buffer->flags &= ~RB_IO_BUFFER_LOCKED; - - return result; + return rb_ensure(rb_yield, self, rb_io_buffer_locked_ensure, self); } /* @@ -1445,7 +1523,7 @@ rb_io_buffer_locked(VALUE self) * * for a buffer created from scratch: free memory. * * for a buffer created from string: undo the association. * - * After the buffer is freed, no further operations can't be performed on it. + * After the buffer is freed, no further operations can be performed on it. * * You can resize a freed buffer to re-allocate it. * @@ -1488,13 +1566,21 @@ VALUE rb_io_buffer_free_locked(VALUE self) return self; } +static bool +size_sum_is_bigger_than(size_t a, size_t b, size_t x) +{ + struct rbimpl_size_overflow_tag size = rbimpl_size_add_overflow(a, b); + return size.overflowed || size.result > x; +} + // Validate that access to the buffer is within bounds, assuming you want to // access length bytes from the specified offset. static inline void io_buffer_validate_range(struct rb_io_buffer *buffer, size_t offset, size_t length) { - // We assume here that offset + length won't overflow: - if (offset + length > buffer->size) { + io_buffer_validate_for_reading(buffer); + + if (size_sum_is_bigger_than(offset, length, buffer->size)) { rb_raise(rb_eArgError, "Specified offset+length is bigger than the buffer size!"); } } @@ -1698,6 +1784,8 @@ rb_io_buffer_resize(VALUE self, size_t size) struct rb_io_buffer *buffer = NULL; TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, buffer); + io_buffer_validate_for_reading(buffer); + if (buffer->flags & RB_IO_BUFFER_LOCKED) { rb_raise(rb_eIOBufferLockedError, "Cannot resize locked buffer!"); } @@ -1804,9 +1892,9 @@ rb_io_buffer_compare(VALUE self, VALUE other) } static void -io_buffer_validate_type(size_t size, size_t offset) +io_buffer_validate_type(size_t size, size_t offset, size_t extend) { - if (offset > size) { + if (size_sum_is_bigger_than(offset, extend, size)) { rb_raise(rb_eArgError, "Type extends beyond end of buffer! (offset=%"PRIdSIZE" > size=%"PRIdSIZE")", offset, size); } } @@ -1826,6 +1914,9 @@ io_buffer_validate_type(size_t size, size_t offset) // :u64, :U64 | unsigned 64-bit integer. // :s64, :S64 | signed 64-bit integer. // +// :u128, :U128 | unsigned 128-bit integer. +// :s128, :S128 | signed 128-bit integer. +// // :f32, :F32 | 32-bit floating point number. // :f64, :F64 | 64-bit floating point number. @@ -1857,13 +1948,57 @@ ruby_swapf64(double value) return swap.value; } +// Structures and conversion functions are now in numeric.h/numeric.c +// Unified swap function for 128-bit integers (works with both signed and unsigned) +// Since both rb_uint128_t and rb_int128_t have the same memory layout, +// we can use a union to make the swap function work with both types +static inline rb_uint128_t +ruby_swap128_uint(rb_uint128_t x) +{ + rb_uint128_t result; +#ifdef HAVE_UINT128_T +#if __has_builtin(__builtin_bswap128) + result.value = __builtin_bswap128(x.value); +#else + // Manual byte swap for 128-bit integers + uint64_t low = (uint64_t)x.value; + uint64_t high = (uint64_t)(x.value >> 64); + low = ruby_swap64(low); + high = ruby_swap64(high); + result.value = ((uint128_t)low << 64) | high; +#endif +#else + // Fallback swap function using two 64-bit integers + // For big-endian data on little-endian host (or vice versa): + // 1. Swap bytes within each 64-bit part + // 2. Swap the order of the parts (since big-endian stores high first, little-endian stores low first) + result.parts.low = ruby_swap64(x.parts.high); + result.parts.high = ruby_swap64(x.parts.low); +#endif + return result; +} + +static inline rb_int128_t +ruby_swap128_int(rb_int128_t x) +{ + union uint128_int128_conversion conversion = { + .int128 = x + }; + conversion.uint128 = ruby_swap128_uint(conversion.uint128); + return conversion.int128; +} + +#define IO_BUFFER_VALIDATE_TYPE_FOR_WRITING(buffer, base, size, offset, type) \ + (io_buffer_get_bytes_for_writing(buffer, &(base), &(size)), \ + io_buffer_validate_type(size, offset, sizeof(type))) + #define IO_BUFFER_DECLARE_TYPE(name, type, endian, wrap, unwrap, swap) \ static ID RB_IO_BUFFER_DATA_TYPE_##name; \ \ static VALUE \ io_buffer_read_##name(const void* base, size_t size, size_t *offset) \ { \ - io_buffer_validate_type(size, *offset + sizeof(type)); \ + io_buffer_validate_type(size, *offset, sizeof(type)); \ type value; \ memcpy(&value, (char*)base + *offset, sizeof(type)); \ if (endian != RB_IO_BUFFER_HOST_ENDIAN) value = swap(value); \ @@ -1872,10 +2007,12 @@ io_buffer_read_##name(const void* base, size_t size, size_t *offset) \ } \ \ static void \ -io_buffer_write_##name(const void* base, size_t size, size_t *offset, VALUE _value) \ +io_buffer_write_##name(struct rb_io_buffer* buffer, size_t *offset, VALUE _value) \ { \ - io_buffer_validate_type(size, *offset + sizeof(type)); \ + void* base; size_t size; \ + IO_BUFFER_VALIDATE_TYPE_FOR_WRITING(buffer, base, size, *offset, type); \ type value = unwrap(_value); \ + IO_BUFFER_VALIDATE_TYPE_FOR_WRITING(buffer, base, size, *offset, type); \ if (endian != RB_IO_BUFFER_HOST_ENDIAN) value = swap(value); \ memcpy((char*)base + *offset, &value, sizeof(type)); \ *offset += sizeof(type); \ @@ -1903,6 +2040,11 @@ IO_BUFFER_DECLARE_TYPE(U64, uint64_t, RB_IO_BUFFER_BIG_ENDIAN, RB_ULL2NUM, RB_NU IO_BUFFER_DECLARE_TYPE(s64, int64_t, RB_IO_BUFFER_LITTLE_ENDIAN, RB_LL2NUM, RB_NUM2LL, ruby_swap64) IO_BUFFER_DECLARE_TYPE(S64, int64_t, RB_IO_BUFFER_BIG_ENDIAN, RB_LL2NUM, RB_NUM2LL, ruby_swap64) +IO_BUFFER_DECLARE_TYPE(u128, rb_uint128_t, RB_IO_BUFFER_LITTLE_ENDIAN, rb_uint128_to_numeric, rb_numeric_to_uint128, ruby_swap128_uint) +IO_BUFFER_DECLARE_TYPE(U128, rb_uint128_t, RB_IO_BUFFER_BIG_ENDIAN, rb_uint128_to_numeric, rb_numeric_to_uint128, ruby_swap128_uint) +IO_BUFFER_DECLARE_TYPE(s128, rb_int128_t, RB_IO_BUFFER_LITTLE_ENDIAN, rb_int128_to_numeric, rb_numeric_to_int128, ruby_swap128_int) +IO_BUFFER_DECLARE_TYPE(S128, rb_int128_t, RB_IO_BUFFER_BIG_ENDIAN, rb_int128_to_numeric, rb_numeric_to_int128, ruby_swap128_int) + IO_BUFFER_DECLARE_TYPE(f32, float, RB_IO_BUFFER_LITTLE_ENDIAN, DBL2NUM, NUM2DBL, ruby_swapf32) IO_BUFFER_DECLARE_TYPE(F32, float, RB_IO_BUFFER_BIG_ENDIAN, DBL2NUM, NUM2DBL, ruby_swapf32) IO_BUFFER_DECLARE_TYPE(f64, double, RB_IO_BUFFER_LITTLE_ENDIAN, DBL2NUM, NUM2DBL, ruby_swapf64) @@ -1927,6 +2069,10 @@ io_buffer_buffer_type_size(ID buffer_type) IO_BUFFER_DATA_TYPE_SIZE(U64) IO_BUFFER_DATA_TYPE_SIZE(s64) IO_BUFFER_DATA_TYPE_SIZE(S64) + IO_BUFFER_DATA_TYPE_SIZE(u128) + IO_BUFFER_DATA_TYPE_SIZE(U128) + IO_BUFFER_DATA_TYPE_SIZE(s128) + IO_BUFFER_DATA_TYPE_SIZE(S128) IO_BUFFER_DATA_TYPE_SIZE(f32) IO_BUFFER_DATA_TYPE_SIZE(F32) IO_BUFFER_DATA_TYPE_SIZE(f64) @@ -1936,6 +2082,15 @@ io_buffer_buffer_type_size(ID buffer_type) rb_raise(rb_eArgError, "Invalid type name!"); } +static inline ID +io_buffer_type_id(VALUE name) +{ + Check_Type(name, T_SYMBOL); + if (!STATIC_SYM_P(name)) return 0; + return rb_sym2id(name); +} +#define TYPE_ID(name) io_buffer_type_id(name) + /* * call-seq: * size_of(buffer_type) -> byte size @@ -1952,12 +2107,12 @@ io_buffer_size_of(VALUE klass, VALUE buffer_type) if (RB_TYPE_P(buffer_type, T_ARRAY)) { size_t total = 0; for (long i = 0; i < RARRAY_LEN(buffer_type); i++) { - total += io_buffer_buffer_type_size(RB_SYM2ID(RARRAY_AREF(buffer_type, i))); + total += io_buffer_buffer_type_size(TYPE_ID(RARRAY_AREF(buffer_type, i))); } return SIZET2NUM(total); } else { - return SIZET2NUM(io_buffer_buffer_type_size(RB_SYM2ID(buffer_type))); + return SIZET2NUM(io_buffer_buffer_type_size(TYPE_ID(buffer_type))); } } @@ -1983,6 +2138,11 @@ rb_io_buffer_get_value(const void* base, size_t size, ID buffer_type, size_t *of IO_BUFFER_GET_VALUE(s64) IO_BUFFER_GET_VALUE(S64) + IO_BUFFER_GET_VALUE(u128) + IO_BUFFER_GET_VALUE(U128) + IO_BUFFER_GET_VALUE(s128) + IO_BUFFER_GET_VALUE(S128) + IO_BUFFER_GET_VALUE(f32) IO_BUFFER_GET_VALUE(F32) IO_BUFFER_GET_VALUE(f64) @@ -2012,6 +2172,10 @@ rb_io_buffer_get_value(const void* base, size_t size, ID buffer_type, size_t *of * * +:U64+: unsigned integer, 8 bytes, big-endian * * +:s64+: signed integer, 8 bytes, little-endian * * +:S64+: signed integer, 8 bytes, big-endian + * * +:u128+: unsigned integer, 16 bytes, little-endian + * * +:U128+: unsigned integer, 16 bytes, big-endian + * * +:s128+: signed integer, 16 bytes, little-endian + * * +:S128+: signed integer, 16 bytes, big-endian * * +:f32+: float, 4 bytes, little-endian * * +:F32+: float, 4 bytes, big-endian * * +:f64+: double, 8 bytes, little-endian @@ -2035,7 +2199,7 @@ io_buffer_get_value(VALUE self, VALUE type, VALUE _offset) rb_io_buffer_get_bytes_for_reading(self, &base, &size); - return rb_io_buffer_get_value(base, size, RB_SYM2ID(type), &offset); + return rb_io_buffer_get_value(base, size, TYPE_ID(type), &offset); } /* @@ -2065,7 +2229,7 @@ io_buffer_get_values(VALUE self, VALUE buffer_types, VALUE _offset) for (long i = 0; i < RARRAY_LEN(buffer_types); i++) { VALUE type = rb_ary_entry(buffer_types, i); - VALUE value = rb_io_buffer_get_value(base, size, RB_SYM2ID(type), &offset); + VALUE value = rb_io_buffer_get_value(base, size, TYPE_ID(type), &offset); rb_ary_push(array, value); } @@ -2134,7 +2298,7 @@ io_buffer_each(int argc, VALUE *argv, VALUE self) ID buffer_type; if (argc >= 1) { - buffer_type = RB_SYM2ID(argv[0]); + buffer_type = TYPE_ID(argv[0]); } else { buffer_type = RB_IO_BUFFER_DATA_TYPE_U8; @@ -2172,7 +2336,7 @@ io_buffer_values(int argc, VALUE *argv, VALUE self) ID buffer_type; if (argc >= 1) { - buffer_type = RB_SYM2ID(argv[0]); + buffer_type = TYPE_ID(argv[0]); } else { buffer_type = RB_IO_BUFFER_DATA_TYPE_U8; @@ -2193,7 +2357,7 @@ io_buffer_values(int argc, VALUE *argv, VALUE self) /* * call-seq: - * each_byte([offset, [count]]) {|offset, byte| ...} -> self + * each_byte([offset, [count]]) {|byte| ...} -> self * each_byte([offset, [count]]) -> enumerator * * Iterates over the buffer, yielding each byte starting from +offset+. @@ -2217,7 +2381,11 @@ io_buffer_each_byte(int argc, VALUE *argv, VALUE self) rb_io_buffer_get_bytes_for_reading(self, &base, &size); size_t offset, count; - io_buffer_extract_offset_count(RB_IO_BUFFER_DATA_TYPE_U8, size, argc-1, argv+1, &offset, &count); + io_buffer_extract_offset_count(RB_IO_BUFFER_DATA_TYPE_U8, size, argc, argv, &offset, &count); + + if (size_sum_is_bigger_than(offset, count, size)) { + rb_raise(rb_eArgError, "Specified offset+count is bigger than the buffer size!"); + } for (size_t i = 0; i < count; i++) { unsigned char *value = (unsigned char *)base + i + offset; @@ -2228,9 +2396,10 @@ io_buffer_each_byte(int argc, VALUE *argv, VALUE self) } static inline void -rb_io_buffer_set_value(const void* base, size_t size, ID buffer_type, size_t *offset, VALUE value) +rb_io_buffer_set_value(struct rb_io_buffer *buffer, VALUE buffer_type, size_t *offset, VALUE value) { -#define IO_BUFFER_SET_VALUE(name) if (buffer_type == RB_IO_BUFFER_DATA_TYPE_##name) {io_buffer_write_##name(base, size, offset, value); return;} + ID type = TYPE_ID(buffer_type); +#define IO_BUFFER_SET_VALUE(name) if (type == RB_IO_BUFFER_DATA_TYPE_##name) {io_buffer_write_##name(buffer, offset, value); return;} IO_BUFFER_SET_VALUE(U8); IO_BUFFER_SET_VALUE(S8); @@ -2249,6 +2418,11 @@ rb_io_buffer_set_value(const void* base, size_t size, ID buffer_type, size_t *of IO_BUFFER_SET_VALUE(s64); IO_BUFFER_SET_VALUE(S64); + IO_BUFFER_SET_VALUE(u128); + IO_BUFFER_SET_VALUE(U128); + IO_BUFFER_SET_VALUE(s128); + IO_BUFFER_SET_VALUE(S128); + IO_BUFFER_SET_VALUE(f32); IO_BUFFER_SET_VALUE(F32); IO_BUFFER_SET_VALUE(f64); @@ -2258,6 +2432,21 @@ rb_io_buffer_set_value(const void* base, size_t size, ID buffer_type, size_t *of rb_raise(rb_eArgError, "Invalid type name!"); } +struct io_buffer_set_value_arguments { + struct rb_io_buffer *buffer; + size_t offset; + VALUE type, value; +}; + +static VALUE +io_buffer_set_value_try(VALUE arguments) +{ + struct io_buffer_set_value_arguments *args = (void *)arguments; + size_t offset = args->offset; + rb_io_buffer_set_value(args->buffer, args->type, &offset, args->value); + return SIZET2NUM(offset); +} + /* * call-seq: set_value(type, offset, value) -> offset * @@ -2291,13 +2480,30 @@ rb_io_buffer_set_value(const void* base, size_t size, ID buffer_type, size_t *of static VALUE io_buffer_set_value(VALUE self, VALUE type, VALUE _offset, VALUE value) { - void *base; - size_t size; - size_t offset = io_buffer_extract_offset(_offset); + struct io_buffer_set_value_arguments arguments = { + .buffer = get_io_buffer_for_writing(self), + .offset = io_buffer_extract_offset(_offset), + .type = type, + .value = value, + }; + + if (!io_buffer_try_lock(arguments.buffer)) { + return io_buffer_set_value_try((VALUE)&arguments); + } + return rb_ensure(io_buffer_set_value_try, (VALUE)&arguments, rb_io_buffer_locked_ensure, self); +} - rb_io_buffer_get_bytes_for_writing(self, &base, &size); +static VALUE +io_buffer_set_values_try(VALUE arguments) +{ + struct io_buffer_set_value_arguments *args = (void *)arguments; + size_t offset = args->offset; - rb_io_buffer_set_value(base, size, RB_SYM2ID(type), &offset, value); + for (long i = 0; i < RARRAY_LEN(args->type); i++) { + VALUE type = rb_ary_entry(args->type, i); + VALUE value = rb_ary_entry(args->value, i); + rb_io_buffer_set_value(args->buffer, type, &offset, value); + } return SIZET2NUM(offset); } @@ -2319,31 +2525,31 @@ io_buffer_set_value(VALUE self, VALUE type, VALUE _offset, VALUE value) static VALUE io_buffer_set_values(VALUE self, VALUE buffer_types, VALUE _offset, VALUE values) { + struct io_buffer_set_value_arguments arguments = { + .buffer = get_io_buffer_for_writing(self), + }; + if (!RB_TYPE_P(buffer_types, T_ARRAY)) { rb_raise(rb_eArgError, "Argument buffer_types should be an array!"); } + arguments.type = buffer_types; + + arguments.offset = io_buffer_extract_offset(_offset); if (!RB_TYPE_P(values, T_ARRAY)) { rb_raise(rb_eArgError, "Argument values should be an array!"); } + arguments.value = values; if (RARRAY_LEN(buffer_types) != RARRAY_LEN(values)) { rb_raise(rb_eArgError, "Argument buffer_types and values should have the same length!"); } - size_t offset = io_buffer_extract_offset(_offset); - - void *base; - size_t size; - rb_io_buffer_get_bytes_for_writing(self, &base, &size); - - for (long i = 0; i < RARRAY_LEN(buffer_types); i++) { - VALUE type = rb_ary_entry(buffer_types, i); - VALUE value = rb_ary_entry(values, i); - rb_io_buffer_set_value(base, size, RB_SYM2ID(type), &offset, value); + if (!io_buffer_try_lock(arguments.buffer)) { + return io_buffer_set_values_try((VALUE)&arguments); } + return rb_ensure(io_buffer_set_values_try, (VALUE)&arguments, rb_io_buffer_locked_ensure, self); - return SIZET2NUM(offset); } static size_t IO_BUFFER_BLOCKING_SIZE = 1024*1024; @@ -2379,7 +2585,7 @@ io_buffer_memmove(struct rb_io_buffer *buffer, size_t offset, const void *source io_buffer_validate_range(buffer, offset, length); - if (source_offset + length > source_size) { + if (size_sum_is_bigger_than(source_offset, length, source_size)) { rb_raise(rb_eArgError, "The computed source range exceeds the size of the source buffer!"); } @@ -2465,7 +2671,9 @@ rb_io_buffer_initialize_copy(VALUE self, VALUE source) io_buffer_initialize(self, buffer, NULL, source_size, io_flags_for_size(source_size), Qnil); - return io_buffer_copy_from(buffer, source_base, source_size, 0, NULL); + VALUE result = io_buffer_copy_from(buffer, source_base, source_size, 0, NULL); + RB_GC_GUARD(source); + return result; } /* @@ -2550,7 +2758,9 @@ io_buffer_copy(int argc, VALUE *argv, VALUE self) rb_io_buffer_get_bytes_for_reading(source, &source_base, &source_size); - return io_buffer_copy_from(buffer, source_base, source_size, argc-1, argv+1); + VALUE result = io_buffer_copy_from(buffer, source_base, source_size, argc-1, argv+1); + RB_GC_GUARD(source); + return result; } /* @@ -2628,7 +2838,9 @@ io_buffer_set_string(int argc, VALUE *argv, VALUE self) const void *source_base = RSTRING_PTR(string); size_t source_size = RSTRING_LEN(string); - return io_buffer_copy_from(buffer, source_base, source_size, argc-1, argv+1); + VALUE result = io_buffer_copy_from(buffer, source_base, source_size, argc-1, argv+1); + RB_GC_GUARD(string); + return result; } void @@ -3248,14 +3460,14 @@ io_buffer_pwrite(int argc, VALUE *argv, VALUE self) } static inline void -io_buffer_check_mask(const struct rb_io_buffer *buffer) +io_buffer_check_mask_size(size_t size) { - if (buffer->size == 0) + if (size == 0) rb_raise(rb_eIOBufferMaskError, "Zero-length mask given!"); } static void -memory_and(unsigned char * restrict output, unsigned char * restrict base, size_t size, unsigned char * restrict mask, size_t mask_size) +memory_and(unsigned char * restrict output, const unsigned char * restrict base, size_t size, const unsigned char * restrict mask, size_t mask_size) { for (size_t offset = 0; offset < size; offset += 1) { output[offset] = base[offset] & mask[offset % mask_size]; @@ -3283,19 +3495,27 @@ io_buffer_and(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + const void *base; + size_t size; + io_buffer_get_bytes_for_reading(buffer, &base, &size); + + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); - VALUE output = rb_io_buffer_new(NULL, buffer->size, io_flags_for_size(buffer->size)); + io_buffer_check_mask_size(mask_size); + + VALUE output = rb_io_buffer_new(NULL, size, io_flags_for_size(size)); struct rb_io_buffer *output_buffer = NULL; TypedData_Get_Struct(output, struct rb_io_buffer, &rb_io_buffer_type, output_buffer); - memory_and(output_buffer->base, buffer->base, buffer->size, mask_buffer->base, mask_buffer->size); + memory_and(output_buffer->base, base, size, mask_base, mask_size); return output; } static void -memory_or(unsigned char * restrict output, unsigned char * restrict base, size_t size, unsigned char * restrict mask, size_t mask_size) +memory_or(unsigned char * restrict output, const unsigned char * restrict base, size_t size, const unsigned char * restrict mask, size_t mask_size) { for (size_t offset = 0; offset < size; offset += 1) { output[offset] = base[offset] | mask[offset % mask_size]; @@ -3323,19 +3543,27 @@ io_buffer_or(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + const void *base; + size_t size; + io_buffer_get_bytes_for_reading(buffer, &base, &size); + + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); - VALUE output = rb_io_buffer_new(NULL, buffer->size, io_flags_for_size(buffer->size)); + io_buffer_check_mask_size(mask_size); + + VALUE output = rb_io_buffer_new(NULL, size, io_flags_for_size(size)); struct rb_io_buffer *output_buffer = NULL; TypedData_Get_Struct(output, struct rb_io_buffer, &rb_io_buffer_type, output_buffer); - memory_or(output_buffer->base, buffer->base, buffer->size, mask_buffer->base, mask_buffer->size); + memory_or(output_buffer->base, base, size, mask_base, mask_size); return output; } static void -memory_xor(unsigned char * restrict output, unsigned char * restrict base, size_t size, unsigned char * restrict mask, size_t mask_size) +memory_xor(unsigned char * restrict output, const unsigned char * restrict base, size_t size, const unsigned char * restrict mask, size_t mask_size) { for (size_t offset = 0; offset < size; offset += 1) { output[offset] = base[offset] ^ mask[offset % mask_size]; @@ -3363,19 +3591,27 @@ io_buffer_xor(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + const void *base; + size_t size; + io_buffer_get_bytes_for_reading(buffer, &base, &size); + + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); - VALUE output = rb_io_buffer_new(NULL, buffer->size, io_flags_for_size(buffer->size)); + io_buffer_check_mask_size(mask_size); + + VALUE output = rb_io_buffer_new(NULL, size, io_flags_for_size(size)); struct rb_io_buffer *output_buffer = NULL; TypedData_Get_Struct(output, struct rb_io_buffer, &rb_io_buffer_type, output_buffer); - memory_xor(output_buffer->base, buffer->base, buffer->size, mask_buffer->base, mask_buffer->size); + memory_xor(output_buffer->base, base, size, mask_base, mask_size); return output; } static void -memory_not(unsigned char * restrict output, unsigned char * restrict base, size_t size) +memory_not(unsigned char * restrict output, const unsigned char * restrict base, size_t size) { for (size_t offset = 0; offset < size; offset += 1) { output[offset] = ~base[offset]; @@ -3386,7 +3622,7 @@ memory_not(unsigned char * restrict output, unsigned char * restrict base, size_ * call-seq: * ~source -> io_buffer * - * Generate a new buffer the same size as the source by applying the binary NOT + * Generate a new buffer the same size as the source by applying the unary NOT * operation to the source. * * ~IO::Buffer.for("1234567890") @@ -3400,11 +3636,15 @@ io_buffer_not(VALUE self) struct rb_io_buffer *buffer = NULL; TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, buffer); - VALUE output = rb_io_buffer_new(NULL, buffer->size, io_flags_for_size(buffer->size)); + const void *base; + size_t size; + io_buffer_get_bytes_for_reading(buffer, &base, &size); + + VALUE output = rb_io_buffer_new(NULL, size, io_flags_for_size(size)); struct rb_io_buffer *output_buffer = NULL; TypedData_Get_Struct(output, struct rb_io_buffer, &rb_io_buffer_type, output_buffer); - memory_not(output_buffer->base, buffer->base, buffer->size); + memory_not(output_buffer->base, base, size); return output; } @@ -3460,13 +3700,17 @@ io_buffer_and_inplace(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + io_buffer_check_mask_size(mask_buffer->size); io_buffer_check_overlaps(buffer, mask_buffer); void *base; size_t size; io_buffer_get_bytes_for_writing(buffer, &base, &size); + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); + memory_and_inplace(base, size, mask_buffer->base, mask_buffer->size); return self; @@ -3506,13 +3750,17 @@ io_buffer_or_inplace(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + io_buffer_check_mask_size(mask_buffer->size); io_buffer_check_overlaps(buffer, mask_buffer); void *base; size_t size; io_buffer_get_bytes_for_writing(buffer, &base, &size); + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); + memory_or_inplace(base, size, mask_buffer->base, mask_buffer->size); return self; @@ -3552,13 +3800,17 @@ io_buffer_xor_inplace(VALUE self, VALUE mask) struct rb_io_buffer *mask_buffer = NULL; TypedData_Get_Struct(mask, struct rb_io_buffer, &rb_io_buffer_type, mask_buffer); - io_buffer_check_mask(mask_buffer); + io_buffer_check_mask_size(mask_buffer->size); io_buffer_check_overlaps(buffer, mask_buffer); void *base; size_t size; io_buffer_get_bytes_for_writing(buffer, &base, &size); + const void *mask_base; + size_t mask_size; + io_buffer_get_bytes_for_reading(mask_buffer, &mask_base, &mask_size); + memory_xor_inplace(base, size, mask_buffer->base, mask_buffer->size); return self; @@ -3576,7 +3828,7 @@ memory_not_inplace(unsigned char * restrict base, size_t size) * call-seq: * source.not! -> io_buffer * - * Modify the source buffer in place by applying the binary NOT + * Modify the source buffer in place by applying the unary NOT * operation to the source. * * source = IO::Buffer.for("1234567890").dup # Make a read/write copy. @@ -3604,6 +3856,60 @@ io_buffer_not_inplace(VALUE self) return self; } +static size_t +memory_bit_count(const unsigned char *base, size_t size) +{ + size_t count = 0; + + // Process 8 bytes at a time for efficiency: + const uint64_t *base64 = (const uint64_t *)base; + size_t count64 = size / 8; + for (size_t i = 0; i < count64; i += 1) { + count += rb_popcount64(base64[i]); + } + + // Process any remaining bytes: + size_t remaining = size % 8; + const unsigned char *tail = base + (count64 * 8); + for (size_t i = 0; i < remaining; i += 1) { + count += rb_popcount32(tail[i]); + } + + return count; +} + +/* + * call-seq: bit_count([offset, [length]]) -> integer + * + * Returns the number of set bits (1s) in the buffer, also known as the + * Hamming weight or population count. An optional +offset+ and +length+ + * can be provided to count bits in a subrange of the buffer. + * + * IO::Buffer.for("\xFF\x00\x0F").bit_count + * # => 12 + * + * IO::Buffer.for("\xFF\x00\x0F").bit_count(1, 2) + * # => 4 + */ +static VALUE +io_buffer_bit_count(int argc, VALUE *argv, VALUE self) +{ + rb_check_arity(argc, 0, 2); + + size_t offset, length; + struct rb_io_buffer *buffer = io_buffer_extract_offset_length(self, argc, argv, &offset, &length); + + io_buffer_validate_range(buffer, offset, length); + + const void *base; + size_t size; + io_buffer_get_bytes_for_reading(buffer, &base, &size); + + size_t count = memory_bit_count((const unsigned char *)base + offset, length); + + return SIZET2NUM(count); +} + /* * Document-class: IO::Buffer * @@ -3670,9 +3976,9 @@ io_buffer_not_inplace(VALUE self) * * File.write('test.txt', 'test data') * # => 9 - * buffer = IO::Buffer.map(File.open('test.txt')) + * buffer = IO::Buffer.map(File.open('test.txt'), nil, 0, IO::Buffer::READONLY) * # => - * # #<IO::Buffer 0x00007f3f0768c000+9 MAPPED IMMUTABLE> + * # #<IO::Buffer 0x00007f3f0768c000+9 EXTERNAL MAPPED FILE SHARED READONLY> * # ... * buffer.get_string(5, 2) # read 2 bytes, starting from offset 5 * # => "da" @@ -3821,6 +4127,11 @@ Init_IO_Buffer(void) IO_BUFFER_DEFINE_DATA_TYPE(s64); IO_BUFFER_DEFINE_DATA_TYPE(S64); + IO_BUFFER_DEFINE_DATA_TYPE(u128); + IO_BUFFER_DEFINE_DATA_TYPE(U128); + IO_BUFFER_DEFINE_DATA_TYPE(s128); + IO_BUFFER_DEFINE_DATA_TYPE(S128); + IO_BUFFER_DEFINE_DATA_TYPE(f32); IO_BUFFER_DEFINE_DATA_TYPE(F32); IO_BUFFER_DEFINE_DATA_TYPE(f64); @@ -3854,6 +4165,8 @@ Init_IO_Buffer(void) rb_define_method(rb_cIOBuffer, "xor!", io_buffer_xor_inplace, 1); rb_define_method(rb_cIOBuffer, "not!", io_buffer_not_inplace, 0); + rb_define_method(rb_cIOBuffer, "bit_count", io_buffer_bit_count, -1); + // IO operations: rb_define_method(rb_cIOBuffer, "read", io_buffer_read, -1); rb_define_method(rb_cIOBuffer, "pread", io_buffer_pread, -1); |