> On Sep 23, 2022, at 1:38 PM, Christian Kennedy via cctalk
> <cctalk@classiccmp.org> wrote:
>
>
> On 23/09/22 10:22, Paul Koning via cctalk wrote:
>>
>> I view the deadstart panel as a type of boot ROM, different from other boot
>> ROMs only in that it's easy to change. It was tied to an I/O channel; the
>> deadstart operation would run an I/O read operation on that channel to load
>> the initial bits of code.
>
> <pedantic>It wasn't tied to a channel, it forced the instructions into PP
> zero, which in turn interacted with the channel.</pedantic>
Yes. I meant the panel is tied to channel 0, and deadstart runs an input from
that channel. To be fully precise:
The deadstart (master clear) signal forces a particular instruction state into
each PP, and it also sets the "deadstart synchronizer" -- the device that
connects the deadstart panel to channel zero -- into active state. The PPs
then execute the newly set instruction state. Since all other channels are
inactive after master clear, PPn for n != 0 would sit there waiting for channel
active. PP 0 reads words from the deadstart synchronizer: a zero word followed
by the 12 words on the panel, and then the synchronizer disconnects. That ends
the IAM instruction and PP 0 resumes at the address pointed to by location 0,
which is 0, so the panel contents is executed (preceded by a 0 opcode which is
a NOP).
paul