diff options
| author | Matt Spinler <spinler@us.ibm.com> | 2022-02-04 19:48:46 +0300 |
|---|---|---|
| committer | Andrew Geissler <geissonator@yahoo.com> | 2022-02-08 22:58:43 +0300 |
| commit | 2d29c6d216e9803917c8060809040d042906279c (patch) | |
| tree | 8b9db5d11e0828507e1d7d1bf7eba7dd70a4f045 /meta-openpower | |
| parent | 1dcc03c336761634dc726825caf5c2f929268474 (diff) | |
| download | openbmc-2d29c6d216e9803917c8060809040d042906279c.tar.xz | |
openpower:occ: Ensure OCC disabled before poweroff
There previously wasn't a hard dependency to not remove power until the
OCCs were all disabled, causing failures with the code that communicates
to the OCCs as the OCCs didn't have power but the BMC code didn't know
that.
This would show up, sometimes at least, as a very long hang in the SBE
driver, which in turned prevented the chassis power state D-Bus property
from getting set to off during that time, which caused other pieces of
code to still think the power was on even though it was really off.
This would then show up as read failures on sensors that were only
accessible with power on.
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Iac77a69766c197705adcf756da98f5a005f78c22
Diffstat (limited to 'meta-openpower')
| -rw-r--r-- | meta-openpower/recipes-phosphor/occ/openpower-occ-control/op-occ-disable@.service | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/meta-openpower/recipes-phosphor/occ/openpower-occ-control/op-occ-disable@.service b/meta-openpower/recipes-phosphor/occ/openpower-occ-control/op-occ-disable@.service index e18d91c154..8a6f2bb61d 100644 --- a/meta-openpower/recipes-phosphor/occ/openpower-occ-control/op-occ-disable@.service +++ b/meta-openpower/recipes-phosphor/occ/openpower-occ-control/op-occ-disable@.service @@ -1,5 +1,6 @@ [Unit] Description=OpenPOWER OCC Active Disable +Wants=obmc-host-stop-pre@%i.target Before=obmc-host-stop-pre@%i.target Before=op-enter-mpreboot@%i.service Conflicts=op-occ-enable@%i.service |