| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Tomer Maimon (4):
arm64: dts: nuvoton: npcm845: Split APB address ranges
arm64: dts: nuvoton: npcm845: Add peripheral nodes
arm64: dts: nuvoton: npcm845-evb: Fix DTS coding style
arm64: dts: nuvoton: npcm845-evb: Add peripheral nodes
Change-Id: I475b320ae660a69cfd6b89fdacb2355056aa0060
Signed-off-by: Andrew Jeffery <andrew@codeconstruct.com.au>
|
|
Tomer Maimon (4):
arm64: dts: nuvoton: npcm845: Split APB address ranges
arm64: dts: nuvoton: npcm845: Add peripheral nodes
arm64: dts: nuvoton: npcm845-evb: Fix DTS coding style
arm64: dts: nuvoton: npcm845-evb: Add peripheral nodes
Change-Id: Ic328f0c690c5b3fec9346b1c5103c2c96981776a
Signed-off-by: Andrew Jeffery <andrew@codeconstruct.com.au>
|
|
Eric Yang (1):
platform-mc: Lazily create FRU decorator interfaces
Change-Id: Ib4eb1cdfc7b7b4dc96e5863182b36c5632455697
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
With GBMC_AVOID_RWFS, gbmc_net_mask_or_rm() bind-mounts /dev/null over
files it can't remove. For files under /etc/systemd that is systemd's
marker for a masked file, which also hides the same-named copies in
/run/systemd and /usr/lib/systemd instead of falling back to them.
gbmc_ncsi_clear_ip_hook() hits this for
/etc/systemd/network/00-bmc-ncsi.network. On a read-only RWFS the vendor
/usr/lib/systemd/network/00-bmc-ncsi.network got masked as well, so ncsi
was reconfigured with 60-phosphor-networkd-default.network and lost all
of its 00-bmc-ncsi.network.d drop-ins: the RA gateway routes (metric 900),
the routing policy rules for the NCSI /76 and the proxy neighbors.
Without the "from <ncsi /76> lookup <ncsi table>" rule gbmc-br-gw-src
then also sourced the NCSI default routes from the front panel address
when the front panel went down.
Bind-mount the copy systemd would use if the file were removed instead,
falling back to /dev/null when there is none or it is masked itself.
Tested: With a read-only RWFS, ncsi matched the vendor default .network
and had no NCSI routes or rules. Unit tested the mask selection.
Change-Id: I602c544f76209f43d84b237d311801fa5f7ab16a
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
The BMC advertising a stateless /80 on gbmcbr only routed the static
/128 of each downstream BMC, and covered the rest of its allocation
(/76 in-band, /64 OOB) with an unreachable route. A tray that lost its
static address, e.g. after an RWFS purge removed /var/google/gbmc-br-ip,
was left with only its autoconfigured address in the /80, and all
traffic to it was dropped by the head's unreachable route even though
the tray was up and configured.
Install an on-link route for the advertised /80 over gbmcbr next to the
unreachable route in gbmc_br_set_runtime_ip(), and from the NCSI
deprecated IPs path through a gbmcbr drop-in (applied right away like its
rules and proxy neighbors), so autoconfigured downstream addresses are
reachable through the head.
On the tray, gbmc-br-from-ra now also writes an on-link route for each
RA prefix into its networkd drop-in. The kernel RA route for the prefix
is flushed on every gbmcbr reconfigure until the next RA, which left the
rest of the /80 (the head's CN facing addresses and other BMCs)
unreachable in the meantime. The route is also added with `ip` right
away, like the address.
Tested: A tray with only its autoconfigured /80 address was unreachable
through the head before and reachable after. Unit tested the drop-in.
Change-Id: Ie4d8b3b514a29a52677d26e3160cbac4f57f656e
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
The ULAs mirroring the gbmcbr link-local addresses were only added with
`ip addr`, so networkd considers them foreign and removes them every
time it reconfigures gbmcbr. Losing the ULAs changes the addresses on
the bridge and triggers more work from the ip monitor.
Write the ULAs we manage into a gbmcbr drop-in so networkd owns them
and keeps them across reloads. The drop-in is only rewritten, and
networkd reloaded, when the set of ULAs changes. The ULAs are still
added with `ip addr` so they are usable right away, and are put back
the same way without a reload if they go missing (e.g. networkd
dropping its addresses on carrier loss, which was the reason for moving
them out of networkd previously). ULAs configured statically, in the
.network file or the 60-ula.conf drop-in from gbmc-mac-config, are left
out of the drop-in so their settings aren't overridden. The drop-in gets
its own name so the two don't keep overwriting each other with
differently formatted addresses and forcing extra reloads.
Also fix the operator precedence in the stale ULA check at init, which
never matched.
Tested: Unit tested with stubs that the drop-in follows ULA adds and
removals and only reloads networkd when the set changes.
Change-Id: Ie404e924b45a1cf7c52de08b2d7800025fa28a05
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Passing interfaces to `gbmc_net_networkd_reload` does a
`networkctl reconfigure` on each of them after the reload. The forced
reconfigure was added as a workaround when networkd didn't reliably
pick up drop-in changes on a reload. Current networkd reconfigures a
link on reload whenever the stat info of its .network or any of its
drop-ins changes, so a plain reload is enough to apply changes.
A forced reconfigure removes and re-adds every address and route
networkd manages on the link and drops any address added outside of
networkd. For gbmcbr, every IP update was briefly taking away the BMC
addresses, the ULAs, and the RA addresses, which caused the source
address monitoring to trigger more reloads.
Every caller that forced a reconfigure had just written or removed a
drop-in for the link, except for:
- gbmc-nic-neigh on carrier up, where networkd already requests the
proxy NDP addresses from the drop-in itself on carrier gain.
- gbmc-br-ra waiting on a failed gbmcbr, which is kept since a failed
link is only retried by a reconfigure and has nothing configured.
gbmc-nic-cn was also reconfiguring the NICs when the drop-in it writes
is for gbmcbr. Make its failed_intfs local so failures don't
accumulate across calls and cause reloads forever.
Tested: gbmc-nic-neigh, gbmc-net-lib and gbmc-ip-monitor unit tests
pass, shellcheck unchanged.
Change-Id: Ibdcf86c9130516cfb83acf35e778b8e34548d5ab
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
gbmc-br-from-ra added the /80 RA addresses with `ip addr replace`, so
to networkd they were foreign and got removed every time it
reconfigured gbmcbr, i.e. on every reload. Each removal changed the
preferred source address, which made gbmc-br-gw-src rewrite
70-ip-hybrid-route.conf and reload networkd again, while
gbmc-br-from-ra saw the address missing and added it back. On a tray
whose head only advertised foreign prefixes this never converged and
reloaded networkd ~18 times a minute for hours.
Write the addresses to a gbmcbr networkd drop-in instead, so networkd
owns them and keeps them across reconfigures. networkd is only reloaded
when the set of addresses changes, not on every RA refresh, and a
missing address is no longer re-added by hand since networkd will
restore it. New addresses are still added with `ip addr` right away so
they are usable before the reload completes; networkd takes them over
when it applies the config.
Tested: Unit tested with stubs: new prefixes reload networkd once, RA
refreshes and route flushes don't, and prefixes expire on their timer.
No reload churn after boot on heads or trays in the RWFS test suite on
standalone, head+tray and two-tray machines.
Change-Id: I362b9f15cfea2f2aae2887cb6068823480001006
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Every reconfigure of gbmcbr restarts networkd's router discovery, which
flushes all of the RA routes until the next RA arrives. networkd
solicits a new RA as soon as it is done reconfiguring, so a flushed
route comes back within a second, but acting on the route deletion right
away briefly removed the address and addrlabel on every reconfigure.
Give a deleted RA route 5s to come back, using an ip monitor timer,
before dropping its prefix. networkd and the kernel remove the routes
for prefixes that are withdrawn with a zero lifetime or expire, so
those go away within 5s of the route being removed without needing to
track the RA lifetime ourselves.
Tested: Unit tested with stubs that route flushes coming back within the
grace period keep the address and prefixes whose route stays gone expire
on the timer.
Change-Id: I2a2f873f8645dda165a3bad3e4064accf7adad3f
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Hooks only had the 1s deferral to schedule work, which also holds back
reloads while it is pending. Add named one-shot timers that deliver a
`change=timer` event with `timer=<name>` to the hooks after a delay.
Timers are keyed on the uptime deadline so they aren't affected by the
wall clock jumping when time is synced. Arming a timer that already has
an earlier deadline pending keeps the earlier one, which avoids piling
up sleeping processes when a hook re-arms on every event; hooks re-arm
when the timer fires if they still need a later one. Stale or
cancelled timer events are dropped.
Tested: gbmc-ip-monitor unit tests pass, including new ones for parsing
and arming timers.
Change-Id: Ibe3e71aa6f15fe1b3de94869c2fa2f461e77f488
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Rashid MP (1):
docs: Update Network README for current D-Bus behavior
Change-Id: I077f95b80f9a9c6e88fbedd5338458c4475013d5
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (7):
test: flash_write: fix uninitialized values
vpnor: partition: remove move-assignment
vpnor: test: remove unused constant
transport_mbox: remove unused function
test: windows_equally_evictable: adjust main prototype
clang: add diagnostic pragmas
clang: fix modernize-use-nullptr errors
Change-Id: I0d1372d5530dd77c816999b16d249d5fc1f8596d
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Jayanth Othayoth (2):
amd-event-log: add runtime AFID integration
amd-event-log: add Redfish projection support
Sachin Shukla (1):
PEL: update RTCReadFailure message in registry json
Change-Id: Ic308f1dc40152a0b8a6b1f7e145ec8892353d6dc
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Alexander Hansen (1):
fix performance-prefer-single-char-overloads
Change-Id: I416d21324ecf5e1c55dbd16addc3492c099d613d
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: Iaa80d01fe5c2f4dc642509a5b7dfb8cc3c1c4c49
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Yuvakumar Selvamani (1):
redfish: Refactor FabricAdapter lambda
Change-Id: I106499f938da1116fc5a3d63bab73609e2613286
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Deepak Kodihalli (1):
fw-update: remove pkg process error path code dup
Dhruv Rathi (1):
fw-update: report 100% activation progress on update completion
Meghana Vangapandu (1):
libpldmresponder: Validate enum indices before indexing pvHandls
P Arun Kumar Reddy (1):
fw-update: initialize progress-tracking members
Patrick Williams (2):
clang-format: update latest spec and reformat with clang-23
clang-tidy: remove deprecated test
Vishnunithyasoundhar S (1):
pldmtool: fru: fix crashes on malformed FRU table responses
Change-Id: I09a7eca85ae87da4ed78d705b397cb5c9b98ad52
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Rekha Aparna (1):
Move Table class to tool_table files
Change-Id: Id44c4edc8a049cca907ec72bd8da77a1cbb0ed1f
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Rashid MP (1):
reject out-of-range IPv6 static address selectors
Change-Id: I559fcae6b9fb3853300cf5a492911900add8653c
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Metha, Mayank (1):
mctpreactor: support for AST2700 USB MCTP Reactor
Change-Id: Id6d92586d7315fd314e9ab5881a798ccc4748e5a
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Freddie Jheng (1):
i2c-vr: tps25990: fix checksum register list, ignoring E1h/E2h
Patrick Williams (2):
clang-format: update latest spec and reformat with clang-23
clang-tidy: remove deprecated test
Change-Id: I45c0004fc1686a08956a9da2f78e8605aae78272
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (2):
clang-format: update latest spec and reformat with clang-23
clang-tidy: remove deprecated test
Change-Id: I4c26631ba3c0ce785ad1e007a951378783102fd6
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I27d3c25b0a2cb38b84b267c77e6a56359c68023c
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Jagpal Singh Gill (3):
tools: hold the serial port in a session
tools: write a device's registers with modbus-tool
tools: test the modbus-tool write flow
Change-Id: I4734cebbd4eb31780380f61fa364e7ffbdce51c1
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
net names
On platforms with dual or multiple onboard USB controllers (such as
Server3/Basalt), systemd's predictable network interface naming does
not account for the USB root hub number and generates colliding names
(e.g., enu1u1u1, enu1u3u1), causing interface rename failures.
Add a udev rule to append the USB bus number to ID_NET_NAME_PATH
(e.g. enu1u1u1b3, enu1u3u1b4) to disambiguate interfaces across controllers.
Change-Id: Id097be8df20850a654cb028c1108404dda6d1faa
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I259d5f1ec5d082f2057eb48b15177f3bbeb63e1c
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Deepak Kodihalli (1):
docs: fix typos and stale interface names
Change-Id: I5e2b7806987ff5f98a87b7ce56581abbdbdad12a
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Mike Hsieh (1):
configurations: sanmiguel: add PVT PDB support for Infineon HSC
Change-Id: Ibc124543ecd5b81278d6fed51526eaab539f7c0d
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Potin Lai (1):
storagecommands: fix record_id_msb calculation in GetSDR
Change-Id: I9260bcf80ae6bc369d0aed2628215a33a0819844
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
leakdetector: sdbusplus: use automatic signal emission
Change-Id: I014f7493b500b28ed4d4771d6ecb59d5b05b885d
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I9715916cfda895907520ea6282f21731465f91e2
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I336d3c1fdfa014f8b47db727c9b2ab92200b4d21
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I4b4150b1ea4e4439a995478ae4e08a33c7124f70
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-23
Change-Id: I949b6c82ba473bc810735f9b80ccceddc43e7556
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Configure ncsi0 network interface:
1. Add link rule for BlueField USB CDC Ethernet interface (0955:cf11) to
name it ncsi0.
2. Add network config to enable DHCP and IPv6 on ncsi0.
3. Bring up Clover power when standby power enabled.
Change-Id: Idee855af8e7703a41b408a87e9e55a8db7d9ce35
Signed-off-by: Potin Lai <potin.lai@quantatw.com>
|
|
Deepak Kodihalli (2):
fw-update: remove activation error path duplication
fw-update: misc small cleanups in update managers
Change-Id: Ie6a8d5cb71ee0927225836fe73e1b9bf4ada912a
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Vinothkumar Shanmugavel (1):
Add RAII for OpenSSL key resources
William A. Kennington III (3):
authority: don't re-stage the authorities list at boot
authority: stage in /run and atomically commit the authorities list
authority: prefer RENAME_EXCHANGE when committing the authorities list
Change-Id: I7f2e18882eb99f3a9e5160787913c49ed8c30091
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Vinothkumar Shanmugavel (1):
OWNERS: add myself as a reviewer
Yuvakumar Selvamani (1):
Refactor chassis_utils getValidChassisPath lambda
Change-Id: I39146ceadd807ff8fa59266f020be1a31bdf8a19
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Kyle Hsieh (1):
configurations: ventura2a7: poll PSU status at 45s
Change-Id: I63bc90104feba3e2dcb64b054e3ebe7ebdd2817e
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (1):
clang-format: update latest spec and reformat with clang-21
Change-Id: I64183f0be64e972def6070c3230f8c1229f006c0
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Bart Van Assche (9):
test: Add Filesystem unit and integration tests
filesystem: Pass runFsck() arguments as an initializer list
filesystem: Pass runMkfs() arguments as an initializer list
Implement the runCommand() function
filesystem: Replace system() calls with runCommand()
filesystem: Pass the -y option instead of -p to fsck
filesystem: Fix the runFsck() return value documentation
estoraged: Do not treat successful fsck repairs as failures
util: Clarify sysfs read failure error messages
Change-Id: I9b9cf0b1cc0d4ba280cbb87d60395ca3a37e41d0
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Joel Pullokaran Jesin (1):
redfish/processor: add tests for helpers
Change-Id: Ibbbb5cb1f4b484ac33f5e25702f1865f71cbb3f0
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Alexander Hansen (1):
entity-manager: introduce struct EMConfig
Change-Id: I0b9b8f9eaafa9bf156d27ca76ba8e363aca360dd
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Yuvakumar Selvamani (1):
redfish: Refactor EventService GET handler lambda
Change-Id: I940801ce921692d8a080b5daee2da787f4dd27d6
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Potin Lai (1):
fru-device: defer D-Bus object registration to prevent index changes
Yu Chen (1):
fru_device: bound FRU area offset and length against buffer size
Change-Id: I9ce82822b960e46280ff8030101736f589b48867
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Description:
Power the host back on once every critical leak that took it down has
cleared.
Anacapa provides its own critical-leak-assert-handler-platform.sh and
critical-leak-deassert-handler, shadowing the common meta-facebook ones.
The assert side's hook_on_leak_detected records a stamp per detector
under /run when the leak took a running host down. The deassert handler
removes that detector's stamp and, once none remain and the host is off,
requests RequestedHostTransition On.
assert-power-good-drop now leaves a marker under /run when it syncs the
host to Off after an unexpected power loss, and deassert-power-good-drop
removes it when host power returns.
Motivation:
On a critical leak the tray CPLD drops host power by itself, and
assert-power-good-drop then syncs phosphor-state-manager to Off. When
the leak clears, nothing powers the host back on: the common deassert
handler takes no action, so the host stays off with the CPLD waiting in
WAIT_FOR_DC_ON_STS until someone powers it on by hand.
Asking for On as soon as the leak clears is safe even if the CPLD is
still holding the host in standby, because host-poweron already waits
for RSMRST_N before pressing the power button.
The stamp keeps the handler from turning on a host that was left off
deliberately. Deciding whether the host was up when the leak hit is
racy: the CPLD drops power before leakdetector reports the leak, and
assert-power-good-drop rewrites RequestedHostTransition to Off a few
seconds later, so the hook can run on either side of that write.
assert-power-good-drop already tells an unexpected power loss apart
from a requested power off, since it does nothing while a power off is
in progress. The marker records its verdict, so "the host was up" is
taken to mean that the requested state is still On or the marker is
present.
Tested:
- shellcheck on the changed scripts - PASS
- The stamp logic, exercised locally with busctl stubbed: one detector,
two overlapping detectors, a host already off at the leak, and a host
already running again at the clear - PASS
On an Anacapa tray, injecting a large leak through the tray CPLD with
the RMC in bring-up mode so that only this tray is affected:
- Without this change, the host stayed off after the leak cleared,
with the CPLD state at WAIT_FOR_DC_ON_STS.
- Host running when the leak hits:
+0s leak injected
+2s FM_PWRGD_CPU0_PWROK falls, host down
+5s Manifold_1 assert: "Host taken down by leak 'Manifold_1',
will power on once clear"
+11s assert-power-good-drop requests Off and leaves the marker,
host Off at +16s
+48s leak cleared
+50s Manifold_1 deassert: "Leak 'Manifold_1' cleared, powering
the host back on"
+55s FM_PWRGD_CPU0_PWROK rises, marker removed
+58s host Running
- PASS
- Host powered off on request, leak injected 30 seconds later: no
marker after the power off, "Host was already off at leak
'Manifold_1', leaving it off" on assert, "Leak 'Manifold_1' cleared,
host was not taken down by it" on deassert, and the host stayed off.
- PASS
Not tested on hardware: multiple detectors tripping together.
Change-Id: I5fff19c6de459156383eef2a62a3392d712a8d2c
Signed-off-by: Pete O_o <pmb@meta.com>
|
|
On Yosemite 4, FruDevice only probes OCP NIC cards during BMC boot.
When an OCP NIC is inserted or removed, the FRU data on D-Bus remains
stale because FruDevice does not re-evaluate the target I2C bus,
leaving the corresponding inventory objects out of sync.
Trigger FruDevice ReScanBus when NIC presence GPIOs (PRSNT_NIC[0-3]_N)
change state:
- Add rescan-fru-nic script to scan the target NIC I2C bus (24-27).
- Add rescan-fru-nic-add@ and rescan-fru-nic-remove@ oneshot services.
- Wait 2 seconds on insertion for hardware stabilization and debouncing
before rescanning.
- Configure phosphor-gpio-monitor targets for NIC presence transitions.
This ensures FruDevice and Entity-Manager dynamically maintain the
NIC FRU and inventory lifecycle upon hot-plug.
Tested:
- Unplugged NIC from slot 0:
Verified that FruDevice removed the FRU interface and Entity-Manager
removed the corresponding inventory object.
- Plugged NIC into slot 0:
Verified that FruDevice rescanned bus 24 after 2 seconds and
Entity-Manager re-created the NIC inventory object.
Change-Id: Ic8ce5fd843f9ac1789d2e04e54f9bf7d62974712
Signed-off-by: LeoWu wiwynn <leo.wu.wiwynn@gmail.com>
|
|
Alex Countryman (3):
chassis-availability: document conditionOverrides feature
chassis-availability: support per-chassis condition overrides
chassis-availability: add unit tests for JSON condition overrides
Janet Adkins (1):
Limit mapper get-service to known interfaces
Change-Id: I94fef70f4fb6c5b104cc3ea03ebbcfa7c8070af0
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Jagpal Singh Gill (1):
log_manager: do not reset entryId on eraseAll
Change-Id: Ibdf10210e35c4bb757e94395972a1fa1e1ce84e8
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Janet Adkins (1):
mapper: Add optional interfaces to get-service
Change-Id: I28b9339883ee498ef03e54af7e6ccb3b7114672f
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
