B1 on-board closed loop surfaced a byte-level console interleave: before
the console writer (defaultTask) registers, con_append() routes to
dbg_console_write_raw() -> con_tx_bounded(), which is register-level
polling with no mutual exclusion. adcTask logged every 1 ms, so its
first-frame line collided with defaultTask's [NAND_FTL] init line and
the required marker spanned two interleaved lines -> false FAIL.
- console_log: add dbg_console_writer_ready() (returns s_con_writer_up)
- adc_task: gate first-frame and 10 s stats logs on writer readiness;
defer the first-frame report until the writer is up
- docs: close B1/P1-1 in the registry; flip stale "remaining @test"
notes; correct the console doc's obsolete "drop-only, no interleave"
claim (that held for HAL gState, not register-level raw send)
Bench (run_boot_test.py): VERDICT PASS, readback OK. adcTask
frames/seq +10000 per 10 s = 1000 Hz, idle=0, frst_err=0; stack
496/2048 (75%). @build 0/0.
wandering1
da66c500ea