Breakpad DWARF parser: Expand comments for ByteReader class.

This is preparation for adding support for reading Linux C++ exception
handling data's encoded pointers. The change should have no user-visible
effect; it simply expands the comments for dwarf2reader::ByteReader, and
regroups the member functions.

a=jimblandy, r=nealsid


git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@522 4c0a9323-5329-0410-9bdc-e9ce6186880e
This commit is contained in:
jimblandy 2010-02-11 22:44:01 +00:00
parent 71f7580891
commit 89f1396fd1

View file

@ -43,75 +43,154 @@ enum Endianness {
ENDIANNESS_LITTLE
};
// Class that knows how to read both big endian and little endian
// numbers, for use in DWARF2/3 reader.
// Takes an endianness argument.
// To read addresses and offsets, SetAddressSize and SetOffsetSize
// must be called first.
// A ByteReader knows how to read single- and multi-byte values of
// various endiannesses, sizes, and encodings, as used in DWARF
// debugging information.
class ByteReader {
public:
explicit ByteReader(enum Endianness endian);
// Construct a ByteReader capable of reading one-, two-, four-, and
// eight-byte values according to ENDIANNESS, absolute machine-sized
// addresses, DWARF-style "initial length" values, and signed and
// unsigned LEB128 numbers.
explicit ByteReader(enum Endianness endianness);
virtual ~ByteReader();
// Set the address size to SIZE, which sets up the ReadAddress member
// so that it works.
void SetAddressSize(uint8 size);
// Set the offset size to SIZE, which sets up the ReadOffset member
// so that it works.
void SetOffsetSize(uint8 size);
// Return the current offset size
uint8 OffsetSize() const { return offset_size_; }
// Return the current address size
uint8 AddressSize() const { return address_size_; }
// Read a single byte from BUFFER and return it as an unsigned 8 bit
// number.
uint8 ReadOneByte(const char* buffer) const;
// Read two bytes from BUFFER and return it as an unsigned 16 bit
// number.
// Read two bytes from BUFFER and return them as an unsigned 16 bit
// number, using this ByteReader's endianness.
uint16 ReadTwoBytes(const char* buffer) const;
// Read four bytes from BUFFER and return it as an unsigned 32 bit
// number. This function returns a uint64 so that it is compatible
// with ReadAddress and ReadOffset. The number it returns will
// never be outside the range of an unsigned 32 bit integer.
// Read four bytes from BUFFER and return them as an unsigned 32 bit
// number, using this ByteReader's endianness. This function returns
// a uint64 so that it is compatible with ReadAddress and
// ReadOffset. The number it returns will never be outside the range
// of an unsigned 32 bit integer.
uint64 ReadFourBytes(const char* buffer) const;
// Read eight bytes from BUFFER and return it as an unsigned 64 bit
// number
// Read eight bytes from BUFFER and return them as an unsigned 64
// bit number, using this ByteReader's endianness.
uint64 ReadEightBytes(const char* buffer) const;
// Read an unsigned LEB128 (Little Endian Base 128) number from
// BUFFER and return it as an unsigned 64 bit integer. LEN is set
// to the length read. Everybody seems to reinvent LEB128 as a
// variable size integer encoding, DWARF has had it for a long time.
// BUFFER and return it as an unsigned 64 bit integer. Set LEN to
// the number of bytes read.
//
// The unsigned LEB128 representation of an integer N is a variable
// number of bytes:
//
// - If N is between 0 and 0x7f, then its unsigned LEB128
// representation is a single byte whose value is N.
//
// - Otherwise, its unsigned LEB128 representation is (N & 0x7f) |
// 0x80, followed by the unsigned LEB128 representation of N /
// 128, rounded towards negative infinity.
//
// In other words, we break VALUE into groups of seven bits, put
// them in little-endian order, and then write them as eight-bit
// bytes with the high bit on all but the last.
uint64 ReadUnsignedLEB128(const char* buffer, size_t* len) const;
// Read a signed LEB128 number from BUFFER and return it as an
// signed 64 bit integer. LEN is set to the length read.
// signed 64 bit integer. Set LEN to the number of bytes read.
//
// The signed LEB128 representation of an integer N is a variable
// number of bytes:
//
// - If N is between -0x40 and 0x3f, then its signed LEB128
// representation is a single byte whose value is N in two's
// complement.
//
// - Otherwise, its signed LEB128 representation is (N & 0x7f) |
// 0x80, followed by the signed LEB128 representation of N / 128,
// rounded towards negative infinity.
//
// In other words, we break VALUE into groups of seven bits, put
// them in little-endian order, and then write them as eight-bit
// bytes with the high bit on all but the last.
int64 ReadSignedLEB128(const char* buffer, size_t* len) const;
// Read an offset from BUFFER and return it as an unsigned 64 bit
// integer. DWARF2/3 define offsets as either 4 or 8 bytes,
// generally depending on the amount of DWARF2/3 info present.
uint64 ReadOffset(const char* buffer) const;
// Indicate that addresses on this architecture are SIZE bytes long. SIZE
// must be either 4 or 8. (DWARF allows addresses to be any number of
// bytes in length from 1 to 255, but we only support 32- and 64-bit
// addresses at the moment.) You must call this before using the
// ReadAddress member function.
//
// For data in a .debug_info section, or something that .debug_info
// refers to like line number or macro data, the compilation unit
// header's address_size field indicates the address size to use. Call
// frame information doesn't indicate its address size (a shortcoming of
// the spec); you must supply the appropriate size based on the
// architecture of the target machine.
void SetAddressSize(uint8 size);
// Return the current address size, in bytes. This is either 4,
// indicating 32-bit addresses, or 8, indicating 64-bit addresses.
uint8 AddressSize() const { return address_size_; }
// Read an address from BUFFER and return it as an unsigned 64 bit
// integer. DWARF2/3 allow addresses to be any size from 0-255
// bytes currently. Internally we support 4 and 8 byte addresses,
// and will CHECK on anything else.
// integer, respecting this ByteReader's endianness and address size. You
// must call SetAddressSize before calling this function.
uint64 ReadAddress(const char* buffer) const;
// Read a DWARF2/3 initial length field from START, and report the
// length of the length field in LEN. Return the value of the length
// field. Set this reader's offset size as indicated by the length
// field's encoding.
// DWARF actually defines two slightly different formats: 32-bit DWARF
// and 64-bit DWARF. This is *not* related to the size of registers or
// addresses on the target machine; it refers only to the size of section
// offsets and data lengths appearing in the DWARF data. One only needs
// 64-bit DWARF when the debugging data itself is larger than 4GiB.
// 32-bit DWARF can handle x86_64 or PPC64 code just fine, unless the
// debugging data itself is very large.
//
// DWARF information identifies itself as 32-bit or 64-bit DWARF: each
// compilation unit and call frame information entry begins with an
// "initial length" field, which, in addition to giving the length of the
// data, also indicates the size of section offsets and lengths appearing
// in that data. The ReadInitialLength member function, below, reads an
// initial length and sets the ByteReader's offset size as a side effect.
// Thus, in the normal process of reading DWARF data, the appropriate
// offset size is set automatically. So, you should only need to call
// SetOffsetSize if you are using the same ByteReader to jump from the
// midst of one block of DWARF data into another.
// Read a DWARF "initial length" field from START, and return it as
// an unsigned 64 bit integer, respecting this ByteReader's
// endianness. Set *LEN to the length of the initial length in
// bytes, either four or twelve. As a side effect, set this
// ByteReader's offset size to either 4 (if we see a 32-bit DWARF
// initial length) or 8 (if we see a 64-bit DWARF initial length).
//
// A DWARF initial length is either:
//
// - a byte count stored as an unsigned 32-bit value less than
// 0xffffff00, indicating that the data whose length is being
// measured uses the 32-bit DWARF format, or
//
// - The 32-bit value 0xffffffff, followed by a 64-bit byte count,
// indicating that the data whose length is being measured uses
// the 64-bit DWARF format.
uint64 ReadInitialLength(const char* start, size_t* len);
// Read an offset from BUFFER and return it as an unsigned 64 bit
// integer, respecting the ByteReader's endianness. In 32-bit DWARF, the
// offset is 4 bytes long; in 64-bit DWARF, the offset is eight bytes
// long. You must call ReadInitialLength or SetOffsetSize before calling
// this function; see the comments above for details.
uint64 ReadOffset(const char* buffer) const;
// Return the current offset size, in bytes.
// A return value of 4 indicates that we are reading 32-bit DWARF.
// A return value of 8 indicates that we are reading 64-bit DWARF.
uint8 OffsetSize() const { return offset_size_; }
// Indicate that section offsets and lengths are SIZE bytes long. SIZE
// must be either 4 (meaning 32-bit DWARF) or 8 (meaning 64-bit DWARF).
// Usually, you should not call this function yourself; instead, let a
// call to ReadInitialLength establish the data's offset size
// automatically.
void SetOffsetSize(uint8 size);
private:
// Function pointer type for our address and offset readers.