pdp11-size — user guide

pdp11-size prints the segment sizes — text, data, and bss — of a PDP-11 a.out file, together with their total in both decimal and octal. It is the 2.9BSD size (revision 2.5), reading the little-endian 16-bit objects this toolchain produces and meets. It runs on your host.

For how it works inside — the header it reads, the First Edition and overlay special cases — see the design document.

1. Synopsis

pdp11-size [object ...]

Each object is a PDP-11 a.out file. With no argument, size reads a.out. There are no options — size reads only header fields, and every one of them is always shown.

Output

A single header line, then one line per object:

text	data	bss	dec	oct
158 +	0 +	0 =	158 =	236	dsort.o

The columns are:

column	meaning
`text`	size of the text (code) segment
`data`	size of the initialised data segment
`bss`	size of the uninitialised data segment
`dec`	`text + data + bss`, in decimal
`oct`	the same total, in octal

The + and = signs are literal separators, and the object's name ends the line. Sizes are reported in bytes.

For an auto-overlay image (0430/0431), a second line reports the true core footprint and the size of each overlay:

1024 +	256 +	64 =	1344 =	2500	ovprog
2048 total text, overlays: (512,512)

Here total text is the base text plus every non-zero overlay, and the parenthesised list is the individual overlay sizes.

What it reads

The object format is taken from the header's magic number:

2.9BSD a.out, all six magics — 0407 (normal), 0410 (read-only text), 0411 (separated I&D), and the 0430/0431 auto-overlay images. The three segment words are reported directly, and overlay images get the extra overlay line.
First Edition a.out(V) (0405) — the Research V1 layout, whose header is counted inside the text size and which has no data segment. size corrects for the counted-in header (real text = a_text - 12) and reports the era's data area as the bss (see design §4).

A file whose first word is none of the recognised magics — or that is too short to hold a header — is reported as not in object file format and skipped.

Exit status

size always exits 0. There is no failure exit code: a file that cannot be opened produces a perror diagnostic on stderr, and a file that is not an object produces <name>: not in object file format — in both cases size moves on to the next argument and still exits 0.

Examples

pdp11-size prog                   # text/data/bss and totals for one object
pdp11-size                        # same, for a.out
pdp11-size *.o                    # a table, one line per object
pdp11-size ovprog                 # an overlay image: base sizes + overlay line
pdp11-size /usr/lib/lib*.a        # archives report "not in object file format"

size looks only at individual object/executable headers; point it at an archive and each .a is reported as not in object file format (use pdp11-nm to look inside an archive).

Continue to the design document for the header layout, the First Edition correction, and the overlay reporting.