BFD works by presenting a common abstract view of object files. An object file has a "header" with descriptive info; a variable number of "sections" that each have a name, some attributes, and a block of data; a symbol table; relocation entries; and so forth.
Internally, BFD translates the data from the abstract view into the details of the bit/byte layout required by the target processor and file format. Its key services include handling byte order differences, such as between a little-endian host and big-endian target, correct conversion between 32-bit and 64-bit data, and details of address arithmetic specified by relocation entries.
Although BFD was originally designed to be a generic library usable by a wide variety of tools, its licensing under the GPL, and the frequent need to tinker with the API to accommodate new systems' capabilities has tended to limit its use; BFD's main clients are the GNU Assembler (GAS), GNU Linker (GLD), and other GNU Binary Utilities ("binutils") tools, and the GNU Debugger (GDB). As a result, BFD is not distributed separately, but is always included with releases of binutils and GDB. Nevertheless, BFD is a critical component in the use of GNU tools for embedded systems development.
The BFD lib can be used to read the structured data out of a Core Dump.
When David Henkel-Wallace of Cygnus Support proposed developing the library, as a way to open up new business opportunities for the company, Richard Stallman said (correctly) that it would be difficult; David's response was "BFD" (big fucking deal). This became the library name, and "Binary File Descriptor" was invented later as the meaning of the letters.