Open matutem opened 1 year ago
CC @GregAC @rswarbrick
@antmarzam: Would you be able to look into this as part of M4?
@antmarzam: Would you be able to look into this as part of M4?
@andreaskurth Sure! I'll look into this!
Unable to get this done for M4. Moving to M5 with P2.
@vogelpi - you created an issue for CDC coverage for Xcelium tools - unsure if at the same time this could be covered. Whoever looks at one of the issues might be able to get this one as a related issue? https://github.com/lowRISC/opentitan/issues/23663
Description
The aon_timer_scoreboard makes many assumptions about the exact number of cycles it will take for interrupts or resets will take, based on the a CDC crossing taking exactly one cycle. Foe example, see https://cs.opensource.google/opentitan/opentitan/+/master:hw/ip/aon_timer/dv/env/aon_timer_scoreboard.sv;drc=3d455cd4914bfe6ce235de1a3e2d44359a9aa6df;l=298.
This leads to failures when CDC takes an extra cycle. At least one such failure in aon_timer_stress_all with seed 1407603594 was traced to this problem. Unfortunately this will take some extra work to untangle.