Post #1286710
2025-07-30 02:48 UTC
Since people are doing the thing I explicitly asked them not to do: we accidentally case-analysed the timing through the tie cells that are read through that register. Changing bits 2 or 3 would cause a setup violation. For A3 we fixed the timing on bit 3 by reimplementing that part of the bus mux with an ECO cell and stealing a nearby buffer. It wasn't possible to fix bit 2 in the same way, so we can't use the value 0x4.
The same issue applies to the tie-cell-programmable IDAU watermark over the ROM, as described here in the bootrom source: https://github.com/raspberrypi/pico-bootrom-rp2350/blob/c6cdb1711f32c3e34faaebd58618a6d096dbd52e/src/main/arm/arm8_bootrom_rt0.S#L238-L264
Replies (1)
-
@ignaloidas@not.acu.lt 2025-07-30 03:22
@wren6991@types.pl that's both better and worse than what I thought