182 lines · plain
1=====================================2The MSF File Format3=====================================4 5.. contents::6 :local:7 8.. _msf_layout:9 10File Layout11===========12 13The MSF file format consists of the following components:14 151. :ref:`msf_superblock`162. :ref:`msf_freeblockmap` (also know as Free Page Map, or FPM)173. Data18 19Each component is stored as an indexed block, the length of which is specified20in ``SuperBlock::BlockSize``. The file consists of 1 or more iterations of the21following pattern (sometimes referred to as an "interval"):22 231. 1 block of data242. Free Block Map 1 (corresponds to ``SuperBlock::FreeBlockMapBlock`` 1)253. Free Block Map 2 (corresponds to ``SuperBlock::FreeBlockMapBlock`` 2)264. ``SuperBlock::BlockSize - 3`` blocks of data27 28In the first interval, the first data block is used to store29:ref:`msf_superblock`.30 31The following diagram demonstrates the general layout of the file (\| denotes32the end of an interval, and is for visualization purposes only):33 34+-------------+-----------------------+------------------+------------------+----------+----+------+------+------+-------------+----+-----+35| Block Index | 0 | 1 | 2 | 3 - 4095 | \| | 4096 | 4097 | 4098 | 4099 - 8191 | \| | ... |36+=============+=======================+==================+==================+==========+====+======+======+======+=============+====+=====+37| Meaning | :ref:`msf_superblock` | Free Block Map 1 | Free Block Map 2 | Data | \| | Data | FPM1 | FPM2 | Data | \| | ... |38+-------------+-----------------------+------------------+------------------+----------+----+------+------+------+-------------+----+-----+39 40The file may end after any block, including immediately after a FPM1.41 42.. note::43 LLVM only supports 4096 byte blocks (sometimes referred to as the "BigMsf"44 variant), so the rest of this document will assume a block size of 4096.45 46.. _msf_superblock:47 48The Superblock49==============50At file offset 0 in an MSF file is the MSF *SuperBlock*, which is laid out as51follows:52 53.. code-block:: c++54 55 struct SuperBlock {56 char FileMagic[sizeof(Magic)];57 ulittle32_t BlockSize;58 ulittle32_t FreeBlockMapBlock;59 ulittle32_t NumBlocks;60 ulittle32_t NumDirectoryBytes;61 ulittle32_t Unknown;62 ulittle32_t BlockMapAddr;63 };64 65- **FileMagic** - Must be equal to ``"Microsoft C / C++ MSF 7.00\\r\\n"``66 followed by the bytes ``1A 44 53 00 00 00``.67- **BlockSize** - The block size of the internal file system. Valid values are68 512, 1024, 2048, and 4096 bytes. Certain aspects of the MSF file layout vary69 depending on the block sizes. For the purposes of LLVM, we handle only block70 sizes of 4KiB, and all further discussion assumes a block size of 4KiB.71- **FreeBlockMapBlock** - The index of a block within the file, at which begins72 a bitfield representing the set of all blocks within the file which are "free"73 (i.e. the data within that block is not used). See :ref:`msf_freeblockmap`74 for more information.75 **Important**: ``FreeBlockMapBlock`` can only be ``1`` or ``2``!76- **NumBlocks** - The total number of blocks in the file. ``NumBlocks *77 BlockSize`` should equal the size of the file on disk.78- **NumDirectoryBytes** - The size of the stream directory, in bytes. The79 stream directory contains information about each stream's size and the set of80 blocks that it occupies. It will be described in more detail later.81- **BlockMapAddr** - The index of a block within the MSF file. At this block is82 an array of ``ulittle32_t``'s listing the blocks that the stream directory83 resides on. For large MSF files, the stream directory (which describes the84 block layout of each stream) may not fit entirely on a single block. As a85 result, this extra layer of indirection is introduced, whereby this block86 contains the list of blocks that the stream directory occupies, and the stream87 directory itself can be stitched together accordingly. The number of88 ``ulittle32_t``'s in this array is given by ``ceil(NumDirectoryBytes /89 BlockSize)``.90 91.. _msf_freeblockmap:92 93The Free Block Map94==================95 96The Free Block Map (sometimes referred to as the Free Page Map, or FPM) is a97series of blocks which contains a bit flag for every block in the file. The98flag will be set to 0 if the block is in use, and 1 if the block is unused.99 100Each file contains two FPMs, one of which is active at any given time. This101feature is designed to support incremental and atomic updates of the underlying102MSF file. While writing to an MSF file, if the active FPM is FPM1, you can103write your new modified bitfield to FPM2, and vice versa. Only when you commit104the file to disk do you need to swap the value in the SuperBlock to point to105the new ``FreeBlockMapBlock``.106 107The Free Block Maps are stored as a series of single blocks throughout the file108at intervals of BlockSize. Because each FPM block is of size ``BlockSize``109bytes, it contains 8 times as many bits as an interval has blocks. This means110that the first block of each FPM refers to the first 8 intervals of the file111(the first 32768 blocks), the second block of each FPM refers to the next 8112blocks, and so on. This results in far more FPM blocks being present than are113required, but in order to maintain backwards compatibility the format must stay114this way.115 116The Stream Directory117====================118The Stream Directory is the root of all access to the other streams in an MSF119file. Beginning at byte 0 of the stream directory is the following structure:120 121.. code-block:: c++122 123 struct StreamDirectory {124 ulittle32_t NumStreams;125 ulittle32_t StreamSizes[NumStreams];126 ulittle32_t StreamBlocks[NumStreams][];127 };128 129And this structure occupies exactly ``SuperBlock->NumDirectoryBytes`` bytes.130Note that each of the last two arrays is of variable length, and in particular131that the second array is jagged.132 133**Example:** Suppose a hypothetical PDB file with a 4KiB block size, and 4134streams of lengths {1000 bytes, 8000 bytes, 16000 bytes, 9000 bytes}.135 136Stream 0: ceil(1000 / 4096) = 1 block137 138Stream 1: ceil(8000 / 4096) = 2 blocks139 140Stream 2: ceil(16000 / 4096) = 4 blocks141 142Stream 3: ceil(9000 / 4096) = 3 blocks143 144In total, 10 blocks are used. Let's see what the stream directory might look145like:146 147.. code-block:: c++148 149 struct StreamDirectory {150 ulittle32_t NumStreams = 4;151 ulittle32_t StreamSizes[] = {1000, 8000, 16000, 9000};152 ulittle32_t StreamBlocks[][] = {153 {4},154 {5, 6},155 {11, 9, 7, 8},156 {10, 15, 12}157 };158 };159 160In total, this occupies ``15 * 4 = 60`` bytes, so161``SuperBlock->NumDirectoryBytes`` would equal ``60``, and162``SuperBlock->BlockMapAddr`` would be an array of one ``ulittle32_t``, since163``60 <= SuperBlock->BlockSize``.164 165Note also that the streams are discontiguous, and that part of stream 3 is in the166middle of part of stream 2. You cannot assume anything about the layout of the167blocks!168 169Alignment and Block Boundaries170==============================171As may be clear by now, it is possible for a single field (whether it be a high172level record, a long string field, or even a single ``uint16``) to begin and173end in separate blocks. For example, if the block size is 4096 bytes, and a174``uint16`` field begins at the last byte of the current block, then it would175need to end on the first byte of the next block. Since blocks are not176necessarily contiguously laid out in the file, this means that both the consumer177and the producer of an MSF file must be prepared to split data apart178accordingly. In the aforementioned example, the high byte of the ``uint16``179would be written to the last byte of block N, and the low byte would be written180to the first byte of block N+1, which could be tens of thousands of bytes later181(or even earlier!) in the file, depending on what the stream directory says.182