5.4. COFF Symbol Table

The Symbol Table described in this section is inherited from the traditional COFF format. It is distinct from CodeView® information. A file may contain both a COFF Symbol Table and CodeView debug information, and the two are kept separate. Some Microsoft tools use the Symbol Table for limited but important purposes, such as communicating COMDAT information to the linker. Section names and file names, as well as code and data symbols, are listed in the Symbol Table.

The location of the Symbol Table is indicated in the COFF Header.

The Symbol Table is an array of records, each 18 bytes long. Each record is either a standard or auxiliary symbol-table record. A standard record defines a symbol or name, and has the following format:

Zero or more auxiliary symbol-table records immediately follow each standard symbol-table record. However, typically not more than one auxiliary symbol-table record follows a standard symbol-table record (except for .file records with long file names). Each auxiliary record is the same size as a standard symbol-table record (18 bytes), but rather than define a new symbol, the auxiliary record gives additional information on the last symbol defined. The choice of which of several formats to use depends on the Storage Class field. Currently defined formats for auxiliary symbol table records are shown in “Auxiliary Symbol Records.”

Tools that read COFF symbol tables must ignore auxiliary symbol records whose interpretation is unknown. This allows the symbol table format to be extended to add new auxiliary records, without breaking existing tools.

5.4.1. Symbol Name Representation

The Name field in a symbol table consists of eight bytes that contain the name itself, if not too long, or else give an offset into the String Table. To determine whether the name itself or an offset is given, test the first four bytes for equality to zero.

5.4.2. Section Number Values

Normally, the Section Value field in a symbol table entry is a one-based index into the Section Table. However, this field is a signed integer and may take negative values. The following values, less than one, have special meanings:

5.4.3. Type Representation

The Type field of a symbol table entry contains two bytes, each byte representing type information. The least-significant byte represents simple (base) data type, and the most-significant byte represents complex type, if any:

The following values are defined for base type, although Microsoft tools generally do not use this field, setting the least-significant byte to 0. Instead, CodeView information is used to indicate types. However, the possible COFF values are listed here for completeness.

The most significant byte specifies whether the symbol is a pointer to, function returning, or array of the base type specified in the least significant byte. Microsoft tools use this field only to indicate whether or not the symbol is a function, so that the only two resulting values are 0x0 and 0x20 for the Type field. However, other tools can use this field to communicate more information.

It is very important to specify the function attribute correctly. This information is required for incremental linking to work correctly. For some architectures the information may be required for other purposes.

5.4.4. Storage Class

The Storage Class field of the Symbol Table indicates what kind of definition a symbol represents. The following table shows possible values. Note that the Storage Class field is an unsigned one-byte integer. The special value -1 should therefore be taken to mean its unsigned equivalent, 0xFF.

Although traditional COFF format makes use of many storage-class values, Microsoft tools rely on CodeView format for most symbolic information and generally use only four storage-class values: EXTERNAL (2), STATIC (3), FUNCTION (101), and STATIC (103). Except in the second column heading below, “Value” should be taken to mean the Value field of the symbol record (whose interpretation depends on the number found as the storage class).