pdp11-objcopy — user guide

pdp11-objcopy copies a PDP-11 a.out, or extracts its loadable image (text plus initialized data, with no header or symbols) or a single segment. PDP-11 is an a.out-only world — there is no ELF target — so, unlike the VAX tree's objcopy, there is no format conversion; the useful operations are the raw-image and segment extractions. Removing the symbol table and relocation is pdp11-strip's job. It runs on your host.

For how each operation works inside — the identity copy, the contiguous-layout slicing, the refused formats, and the bounds check — see the design document.

1. Synopsis

pdp11-objcopy [-O binary] [-j text|data] infile outfile

Both infile and outfile are required; there is no in-place default. The input format is detected. -O accepts only binary; -j accepts only text or data.

2. What it does

invocation	result
(no options)	identity copy: `infile` → `outfile`, byte for byte (any magic)
`-O binary`	the loadable image: text + initialized data, no header, no symbols
`-j text`	the text segment only, raw
`-j data`	the initialized-data segment only, raw

The plain copy is format-agnostic and works for every magic, including the First Edition (0405) and auto-overlay (0430/0431) forms. The extractions (-O binary, -j) need the contiguous on-disk layout of the ordinary magics 0407/0410/0411; 0405 and 0430/0431 are refused rather than mis-sliced. See design §3–§4.

3. Options

-O binary — the only -O value accepted. Writes the loadable text+data image with no header and no symbol table: the bytes the loader maps.
-j text / -j data — extract a single segment, raw. If both -O binary and -j are given, -j selects the slice.

An extraction validates the header's text and data sizes against the actual file length before writing anything, so a corrupt or hostile header is refused, not read out of bounds.

There is no option to convert between object formats and no option to strip symbols: objcopy here copies or extracts bytes only. The whole input is read into memory before the output is written, so naming the same path for infile and outfile is safe — but the write is a plain create, not an atomic rename, so a crash mid-write can still leave a partial output file.

4. Exit status

0 — success.
1 — any error: a bad option, the wrong number of file arguments, an input that is not a PDP-11 a.out (too short, or bad magic), a format that cannot be extracted (0405/0430/0431), header sizes that exceed the file, or an I/O error.

There is no separate usage exit code — every error, usage included, exits 1.

5. Examples

pdp11-objcopy prog prog.copy             # identity copy, byte for byte
pdp11-objcopy -O binary prog prog.bin    # the loadable text+data image
pdp11-objcopy -j text prog text.bin      # just the text segment, raw
pdp11-objcopy -j data prog data.bin      # just the initialized data, raw

To drop the symbol table and relocation instead of extracting bytes, use pdp11-strip.

Continue to the design document for the identity copy, the slicing, and the hostile-header bounds check.