| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
An issue is being pursued where the synchronization target,
obmc-host-start-pre@.target is not being properly stopped in some
instances. This results in the services which depend on it for
synchronization purposes to execute out of order.
The only critical interaction between services which utilize this target
is the fsi-scan and start_host services. You do not want to be starting
the host until the fsi-scan has completed. For now just define the
dependency directly between the services.
Issue openbmc/openbmc#3755 has been opened to further investigate the
target issue.
Tested:
- Verified when a system was in the invalid state that it correctly
booted without issue with this change.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I53dd2cb9976066a7d5a495bf7f2b0b061a8226f6
|
|
This service needs to be run in both the obmc-chassis-poweron target and
obmc-host-startmin target. This is so support is there for FSI
operations with just a "obmcutil chassison" as well as supporting the
warm reboot path where a chassison is not executed. The service will
only run once during a normal boot due to the way the service is
configured.
Similar to the CFAM reset, the FSI scan service should also not be run
on memory preserving reboots.
(From meta-openpower rev: b81b81a2dc37288396bdea38caff429d876f5740)
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I9447a374afa405197ca5ae9373f758c653ca8dc0
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
systemd recommends a wants vs. requires relationship so this is a small
step in that direction for OpenBMC.
This also alleviates a common pain point in bringup where FSI is not
working properly or a FSI device is not present in the system. With a
requires relationship, the chassis poweron target fails. What a user
really wants is for the power on to still succeed.
Other firmware, which requires FSI, will detect issues during the host
boot and take proper actions.
Tested:
Injected fail into fsi-scan service and verified with this change that
chassis power on still succeeded.
(From meta-openpower rev: 58b758a36c28544ee31b75a1aa7dddd5d0fe355c)
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I6615f81778430dd5d855149b97cba70fbba5f993
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Point launches of openpower-proc-control to bin instead of sbin.
openpower-proc-control: srcrev bump aa2030c2fb..b964c92815
Patrick Venture (1):
build: install into bin instead of sbin
(From meta-openpower rev: f4bf3e1ed9cb8de3030952b0e72843e41a943a03)
Signed-off-by: Patrick Venture <venture@google.com>
Change-Id: I06dd1a06b7d47fbe32854e66973dd1053ed58d26
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Adopt a more conventional directory hierarchy. meta-openpower is still
a _long_ way from suitable for hosting on yoctoproject.org but things
like this don't help.
(From meta-openpower rev: 2c8db5661cecd7e4cc20bcb96df76490e62cfb67)
Change-Id: I415caaa28ec728f24e755ddb549012f24874fefe
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
