Check for DIMM failures in cosmo #2174
Conversation
task/cosmo-spd/src/main.rs
Outdated
| ($dev:ident, $index:expr) => { | ||
| let $dev = self.dimms.dimm_pcamp.$dev(); | ||
| if !$dev { | ||
| ringbuf_entry!(Trace::DimmFailure { index: $index }); |
There was a problem hiding this comment.
Eventually, we'll definitely want to record an ereport here as well --- I opened an issue for that: #2173
check the various DIMM PCAMP signals
labbott
force-pushed
the
cosmo_check_dimm_failure
branch
from
fae58fa to
433155c
Compare
|
Is there an assumption that something else has already taken corrective action on the box here? If we lose a DIMM regulator power good, which I realize is only one part of the PCAMP situation, but is the default behavior based on the PMIC situation, then that implies that we need to MAPO, capture information, and transition to A2. Is this merely logging here because the FPGA has issued the MAPO? Running with a failed DIMM here is not helpful to anyone and the host is going to panic in mere moment after this if we actually have this failure. Related, if we do, is the normal hot-swap logic that we have in A2 going to actually handle clearing faults and related? |
|
Keeping this as a draft until we finalize how we want to handle our response/reporting |
check the various DIMM PCAMP signals