File Format¶

Files in this format start with a header record.

Then comes the Processor Information Record.

Then comes the data itself, organized into records.

Overall Structure¶

Header¶

The header consists of 8 words which are interpreted as a null terminated string:

Processor Information Record (PIR)¶

The PIR does not have a checksum; it is therefore possible that a corrupt file header will go undetected.

Data Record¶

Each data record consists of a byte count followed by the comensurate number of words. The data part of each record is raw byte values, no encoding.

The checksum is calculated by a simple sum of unsigned bytes into a 32‐bit accumulator.

Commentary¶

The implementation is based on two documents:

*

HP 64000 Logic Development System ‐ File Format Reference Manual , #64980‐90933, 1985

*

HP 64700 Operating Environment ‐ Absolute File Translator. HP Part No. B1471‐97008, 1992

These documents disagree on some points, and also do not quite match actual examples (firmware update files from early 90's equipment; probably generated from an unknown MC68k toolchain).

The main points of disagreement are:

*

Two first bytes of the file, before any header or the "Processor Information Record", are not officially documented anywhere yet found. (68k linker docs may be worth investigating if they can be found). In this implementation they are set to "82 04", where 04 matches HP doc's "type" value of 4 for Absolute files.

*

Header: Only mentioned in B1471. 16 bytes of user data? Present in this implementation.

*

"recsize" (author's terminology) : redundant field, before every record, giving length of its record minus 1, i.e. recsize == (DataRecord.numbytes + 6) ‐ 1