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 ¶

Suppose the design intent declares a net as an active-low (or active-high) reset. That net must not be biased to its asserted level. If a declared active-low reset has a pull-down, or a declared active-high reset has a pull-up, the design does the opposite of what it says it wants.

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, because 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 saying nothing. 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.

The rule used to say nothing here. A passing item then meant only “no contradiction found” rather than “polarity confirmed”. That distinction had to be written in prose here, in the declaration comment and in a test name, and nobody reading a green report would ever see it. 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. The level the receiver reads then depends on the ratio against its input thresholds. If the rule said “no bias” on a board that visibly has two resistors, you would go looking for the wrong thing.

Its sibling property-ac-coupled is decidable by looking and has 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. At bring-up the failure reads as a broken part or bad firmware until someone meters the reset pin.

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

What counts as bias ¶

A resistor on the declared net counts as bias when its other end reaches a power rail (pull-up) or a ground net (pull-down).

A net with both is a divider. It counts as neither direction and reports inconclusive, as above. 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 ¶

intent:
  nets:
    SYS_RESET_N: {reset: low}
    PHY_ENABLE:  {reset: high}

The level 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. If the declaration is wrong, moving the resistor would make the board match it, and the board would then have the bug the rule was reporting.