2026-08-29 - The null jump was the shortcut coming due
Both payloads jumped to null inside their own find_pid,
and it blocked everything: no guest reached a socket call. Four
candidates, three eliminated by experiment (D359):
sysctl refusal answered zero-length success still null
signal answering zero answered a marker instead still exactly zero
zeroed data storage filled every page with markers still exactly zero
the guest's own .bss filled with 0xB5 THE FAULT CHANGED
.bss zeroed gives
instruction fetch from 0x0 in both; .bss
filled gives read of 0xffffffffffffffff at
image+0x28fc (klogsrv) and image+0x819c
(ftpsrv). Identical shape, so it is the shared runtime rather than one
program.
Entering at main skips __crt_start,
which is the program's initialisation. Globals hold what
.bss holds - zero - and zero is also what an uninitialised
function pointer looks like. The guest calls a global nothing set.
0x0 was never a mystery; it was the correct value of a
variable whose initialiser never ran.
Zeroed .bss stays right - C guarantees it.
ORBISTOUN_BSS_FILL is a diagnostic that breaks the contract
deliberately, the fourth of a family (stack, heap, direct, bss), off by
default including when the value fails to parse.
Worth keeping
A guest jumping to null and a guest reading an uninitialised global
produce the identical fault, and no amount of staring
at 0x0 separates them. The only thing that did was making
zero stop being the value.
And it revises the plan rather than advancing it:
PAYLOADS.md had entering at main as a way
round the handoff structure. It is a way past its first wall,
and the structure is back to being the problem - now clearly worth more
than the alternative, being one structure against an unknown number of
globals.