2026-08-30 - The hardware had already answered; the results were on disk
Prompted to check whether the sysctl probe had run, it had -
data/hardware/ps5-imports.txt, a full suite on a PS5 at
system software 12.400.009, with the
135-sysctl section in it. The answers to a question this
project had been refusing for good reason were sitting there.
kern.osrelease is
0.0-prototype. Not a version - a
development tag. zftpd reads it, cannot parse a version out
of it, and says detection failed. So the payload was correct against a
value nothing here knew, and D397's refusal to invent a plausible
9.00 was right: a made-up version would have sent it down a
different path and looked like it worked.
The console is 12.40, stated outright in
kern.version. The earlier 13.09 was the SDK version from a
different call; the platform keeps the two apart and now so does this.
machdep.tsc_freq came back 0x5f25_9b8e - the
same counter frequency the two clocks already reported, now by a third
independent route.
Implemented: sysctlbyname answers the measured strings
and integers, each at the platform's own width, pinned by a test. The
12.40 CEX profile is documented for the user to apply; the default still
refuses, because an unconfigured emulator does not get to claim it is a
particular console.
The lesson to keep: before building an instrument to ask,
check whether the instrument already ran. The probe was written
this session and the hardware run that used it happened in parallel; the
answer was a grep away from the whole time.