intent/property-reset-polarity

Remedy ¶

Bias the reset net to its DE-asserted level. As drawn, the part is held in reset from power-up, which reads at bring-up as a device that never starts.

What it checks ¶

A net the design intent declares as an active-low (or active-high) reset must not be biased to its asserted level. Declared active-low with a pull-down, or active-high with a pull-up, is a contradiction between what the design says it wants and what it actually does.

What it does when it cannot tell ¶

A netlist states polarity nowhere. The only structural evidence is a bias resistor, and plenty of correct designs carry none: a supervisor or PMIC with an internal pull-up drives the reset line by itself and the schematic shows a bare net.

The rule now says so rather than staying quiet. A declared reset with no bias reports an INCONCLUSIVE finding, and a review item bound to it reads inconclusive, never pass. The message names what could not be resolved and what to check by hand.

This used to be silence, and a passing item then meant only “no contradiction found” rather than “polarity confirmed”, a distinction that had to be carried in prose here, in the declaration comment and in a test name, and that anyone reading a green report would never see. It is now in the report itself.

A divider reports inconclusive too, with a different message: two resistors hold the line at an intermediate level rather than at either rail, so which level the receiver reads depends on the ratio against its input thresholds. Telling you “no bias” on a board that visibly has two resistors would send you looking for the wrong thing.

Its sibling property-ac-coupled is decidable by looking and carries no inconclusive case.

For hardware engineers ¶

An active-low reset (usually drawn RESET_N, nRST, RST#) holds the part in reset while the line is LOW and lets it run when the line is HIGH. So the resting state must be high, and a pull-up to the rail provides it.

Put a pull-down on that line instead and the part is held in reset from the moment power comes up. The board looks dead, and the cause is one resistor to the wrong net, a bring-up failure that reads as a broken part or bad firmware for as long as it takes someone to meter the reset pin.

Active-high is the mirror image: it should rest low, so a pull-up holds it permanently asserted.

What counts as bias ¶

A resistor on the declared net whose other end reaches a power rail (pull-up) or a ground net (pull-down).

A net with both, a divider, reports neither. A divider sets an intermediate level, so it does not hold the line at either rail, and calling it a contradiction would be wrong.

Declaring it ¶

net_properties:
  - {net: SYS_RESET_N, property: reset-polarity, value: low}
  - {net: PHY_ENABLE,  property: reset-polarity, value: high}

The value is the assertion, and it is required: without it the rule has nothing to contradict, so an omitted or misspelled level is rejected at load rather than becoming a rule that silently never fires.

Fixing a finding ¶

The bias resistor goes to the wrong net, or the declared polarity is wrong. Check the part’s datasheet before moving the resistor, because if the declaration is what is wrong, moving the resistor would create the bug the rule was reporting.