Age | Commit message (Collapse) | Author | Files | Lines |
|
Big cleanups to the phosphor-mapper/phosphor-objmgr recipe.
- Remove all "mapper-config-native" constructs as the underlying
repository has no such support.
- Remove all mapper-related bbclasses as they are no longer relevant.
- Remove environment support in service file as the mapper executable
no longer supports any arguments.
- Greatly simplify the 'libmapper' packaging to avoid stray python
in the recipe.
- Update all recipes to use `DEPENDS += "libmapper"` where appropriate
for clarity if they use libmapper.
- Rename the recipe from phosphor-mapper to phosphor-objmgr to match
the repository name, which allows simpler devtool invocations.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I5a64e2feddd1a7919813e007411f9c28b6c9a330
|
|
The close parenthesis was misplaced causing the first if statement to
always be skipped.
Tested: Built for 64-bit platform and inspected cr51-image-layout.json
Signed-off-by: Benjamin Fair <benjaminfair@google.com>
Change-Id: If3dacd917ae26908ef0d4c98866b58051176dea1
|
|
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I65febeeec11e6e5b40ee728f42cfbe77023dae1e
|
|
Move the update and image descriptor partitions to the end in images
targeting 64-bit machines since we may not have space in the current
location when the bootloader and kernel grow in size.
Tested: Built for 64-bit platform and inspected cr51-image-layout.json
Signed-off-by: Benjamin Fair <benjaminfair@google.com>
Change-Id: Ib66ebbb824319fd526a161bca9330b127f37fe6c
|
|
Any google image should able to support the RootOfTrust service
of BMCWeb. Adding the option will enable the feature of BMCWeb.
Signed-off-by: Hao Jiang <jianghao@google.com>
Change-Id: I13bf5f7ebef4ac907a02379a4b8c0cbaa87f3c51
|
|
Nikhil Namjoshi (2):
Fix the license comment format
Add ipmi OEM handler to get the BMC mode
Change-Id: I128898193033af562b43650b8cd10474af3fc530
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
The package "cracklib" is not needed. Ideally in the future we will
start using Google GLOME.
The "libpam" library depends on cracklib, so override the libpam
recipe also, so that it no longer depends on cracklib.
Tested: This has been tested locally on our local product, which uses
OpenBMC, and it appeared to work just fine for us.
Signed-off-by: Kasun Athukorala <kasunath@google.com>
Change-Id: I0d7714766a2e14151f00f6582abee78dee43614d
Signed-off-by: Josh Lehan <krellan@google.com>
|
|
glome-config only provides the config file for glome. It is more
appropriate to decouple them by adding glome-config to
OBMC_IMAGE_EXTRA_INSTALL in the meta-google layer. Then when glome is
migrated to another meta layer, it won't need this RDEPENDS.
(This is a follow-up of
https://gerrit.openbmc.org/c/openbmc/openbmc/+/56618/)
Also add glome-login to OBMC_IMAGE_EXTRA_INSTALL since it is also
needed.
Tested:
Built an image and run it on a real machine. Verified that glome is
enabled and the generated link gave a valid password.
Signed-off-by: Leo Tu <leotu@google.com>
Change-Id: I985670454f4749c5297261ec81466fed9cdc5c40
|
|
We have to use glome-config since the sample config is removed in
glome recipe. As a result, now glome depends on glome-config.
The platform also needs to overwrite these variables in their
bbappend files:
- glome-login.bb
* GLOME_FALLBACK_SERV
* GLOME_FALLBACK_OBJ
* GLOME_HOSTNAME_SUFFIX
* GLOME_BOARDSN_KEY
- glome-config.bb
* GLOME_PUBLIC_KEY
* GLOME_KEY_VERSION
* GLOME_URL_PREFIX
Tested:
1) `bitbake obmc-phosphor-image` build passes
2) inplace update the image on a real machine
3) connect to it
The glome is enabled and login is successful
Signed-off-by: Leo Tu <leotu@google.com>
Change-Id: I21a48cbdcfea67772f143b4fc115f717e6d7cbc5
|
|
We are blocking the old address internally to prevent large amounts of
DHCP spam from old builds. Increment this so new builds are not blocked.
Change-Id: Ib00e632faa9c79445a87c955ca62450379894923
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We can't do floating point math so just chop it off.
Change-Id: I7f78f05856d774236d6e9746b69b362b6468ca0a
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Upstream has changed the service name from `rngd.service` to
`rng-tools.service`. Change the name of the "nojitter" service
to match.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Ie7cef3f0b9106db38e6a399494a85d7e5fc5e3eb
|
|
Remove service dependency so that serial will not be delayed by
network.target.
Signed-off-by: David Wang <davidwang@quantatw.com>
Change-Id: I6febe89373a5e1c845b998905305aafe25744380
|
|
Michael Shen (2):
all: cleanup, prefer "ipmid/api-types.hpp" type to "ipmid/api.h" type
accel: rename `accelOobGetDbus` to `getBus`
Willy Tu (1):
fix typo: Prase -> Parse
Yunyun Lin (1):
google-ipmi-sys: Move Warning to debug statement
Change-Id: Ic83bd456d4e4f1da6a7544fe9f98f17c85dac237
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Patrick Williams (3):
sdbusplus: use shorter type aliases
OWNERS: fix syntax
MAINTAINERS: remove file
Change-Id: I840f81789852a998a1ced1ea0bd47c76ba96ea8c
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
We don't want to terminate just after 5 minutes, we want to make sure
the DHCP process has been idle for at least that long too.
Change-Id: I6311a6baf21c7bc10ece2d4994f225dbc8c06cc4
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We should complete the DHCP process even if we don't receive a bootfile
payload.
Change-Id: I77c45920fda2bcaa732b546f48c5d247c903268f
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This makes it possible to produce an error in the case that a hook sets
up state for something to trigger later.
Change-Id: Ied9f150153172213a98a73f6a556aa8fed87c75a
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Re-read the spec and realized the bit being used was off by 1.
Change-Id: I02d0e747f5e47cfbf3a8c4f4b9e45d16a6f8c8b1
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This does not affect correctness but it matches the file priority of
gbmc-upgrade.
Change-Id: Ie19fc91c5292e1588fe7a46763172273f0724e23
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We want to make sure this happens before any coalesced powercycle.
Change-Id: I12f161b4a69df49c4cf960badfba90d08737d7b9
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
libhoth is the google specific usb protocol implementation which is
required by hothd.
Signed-off-by: Hao Jiang <jianghao@google.com>
Change-Id: Ibe64304342681be390f717fcd62921d1cb1ffcb4
|
|
Copybara-Service (1):
Merge pull request #108 from l9i:l9i-fix-pylint
Markus Rudy (6):
Enforce some code standards when compiling C. (#110)
Merge pull request #112 from vvidic/ini-parser
Merge pull request #114 from vvidic/docker
Merge pull request #111 from vvidic/login-pam
Correctly free the message buffer in case of an snprintf error
Merge pull request #118 from vvidic/newline
Philipp Kern (10):
Merge pull request #106 from vvidic/cli-login
Merge pull request #113 from vvidic/addrinfo-free
Merge pull request #119 from vvidic/option-tests
Support compilation with both old and new pam_wrapper
Run the C compilation presubmit on Debian stable and testing
Merge pull request #123 from pkern/pamtest-fix
Merge pull request #124 from pkern/debian-presubmit
Merge pull request #128 from google/l9i/fakepassword
Merge pull request #129 from vvidic/getopt-long
Merge pull request #130 from vvidic/pam-options
Piotr Lewandowski (7):
pyglome: more consise exception messages
Fix the tag length check exception message
Use comma, not semicolon for intervals notation
pam_glome: support fake passwords from OpenSSH
Apply clang-format
Apply clang-format, Google style
Use an error message matching the rest of the file
Valentin Vidic (9):
Add login command to glome CLI (#91)
Free struct addrinfo allocated by getaddrinfo
Implement a simple INI parser for config files
Create a Docker container for testing glome-login and PAM module
Use the same auth function for PAM and login binary
Make tests optional using a global meson option
Fix handling of long authorization codes
Add support for long and config options
Sync PAM config options names with the rest of the code.
Change-Id: I07674f198f1a00ae7bff4feb99a01db940e7d7ad
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This is trivial to enable if desired, we mostly don't want a huge amount
of BIOS logspam to fill the journal and cause it to rotate out valuable
logs in normal situations.
Change-Id: I991e8f048e847ba081b69f755c48275b63d4af66
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This makes the override idempotent and correct. Otherwise, running the
script multiple times would produce broken unit files.
Change-Id: I351c0becc5555020ce9531cae164fa56f502cfdc
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
mkdir -p $work will fail if rwfs partition is full, and then it will
cause overlayfs mount fail as well. Finally it cause kernel panic and
bmc fails to boot:
rofs = mtd4 squashfs rwfs = mtd5 jffs2
[ 3.841846] jffs2: notice: (103) jffs2_build_xattr_subsystem: complete building xattr subsystem, 21 of xdatum (19 unchecked, 2 orphan) and 45 of xref (1 dead, 1 orphan) found.
mkdir: can't create directory 'run/initramfs/rw/work': No space left on device
[ 9.022304] overlayfs: failed to resolve 'run/initramfs/rw/work': -2
mount: mounting cow on /root failed: No such file or directory
chroot: can't execute '/bin/sh': No such file or directory
Unable to confirm /sbin/init is an executable non-empty file
in merged file system mounted at /root.
Change Root test failed!
Fatal error, triggering kernel panic!
[ 9.141777] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000100
[ 9.142489] CPU: 0 PID: 1 Comm: init Not tainted 5.10.36-60b3c9d #1
[ 9.142667] Hardware name: Generic DT based system
[ 9.142949] Backtrace:
[ 9.144308] [<80915100>] (dump_backtrace) from [<809153a4>] (show_stack+0x20/0x24)
[ 9.144623] r7:80b08ed4 r6:60000093 r5:00000000 r4:80d71c9c
[ 9.144816] [<80915384>] (show_stack) from [<80919b54>] (dump_stack+0x9c/0xb0)
[ 9.144982] [<80919ab8>] (dump_stack) from [<80915824>] (panic+0x114/0x33c)
[ 9.145136] r7:80b08ed4 r6:80d0a970 r5:00000000 r4:80dbe378
[ 9.145271] [<80915710>] (panic) from [<80126c14>] (do_exit+0x99c/0xa0c)
[ 9.145426] r3:00000001 r2:00000000 r1:00000100 r0:80b08ed4
[ 9.145552] r7:ffffe000
[ 9.145632] [<80126278>] (do_exit) from [<80127cf0>] (do_group_exit+0x50/0xc8)
[ 9.145787] r7:000000f8
[ 9.145863] [<80127ca0>] (do_group_exit) from [<80127d88>] (__wake_up_parent+0x0/0x30)
[ 9.146034] r7:000000f8 r6:00000004 r5:00000004 r4:010fd190
[ 9.146185] [<80127d68>] (sys_exit_group) from [<80100060>] (ret_fast_syscall+0x0/0x54)
[ 9.146425] Exception stack(0x810c5fa8 to 0x810c5ff0)
[ 9.146745] 5fa0: 010fd190 00000004 00000001 00000000 00000001 010fd190
[ 9.147049] 5fc0: 010fd190 00000004 00000004 000000f8 00000000 00000000 004e4a84 00000000
[ 9.147295] 5fe0: 004e4b40 7ec93a04 00474048 76df4a54
[ 9.147959] CPU1: stopping
[ 9.148650] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.10.36-60b3c9d #1
[ 9.148957] Hardware name: Generic DT based system
[ 9.149264] Backtrace:
[ 9.150182] [<80915100>] (dump_backtrace) from [<809153a4>] (show_stack+0x20/0x24)
[ 9.150598] r7:00000001 r6:60070193 r5:00000000 r4:80d71c9c
[ 9.150878] [<80915384>] (show_stack) from [<80919b54>] (dump_stack+0x9c/0xb0)
[ 9.151149] [<80919ab8>] (dump_stack) from [<8010f1f4>] (do_handle_IPI+0x2f8/0x32c)
[ 9.151430] r7:00000001 r6:0d15a000 r5:00000001 r4:80dbe278
[ 9.151649] [<8010eefc>] (do_handle_IPI) from [<8010f250>] (ipi_handler+0x28/0x30)
[ 9.151925] r9:810f6000 r8:81080800 r7:00000001 r6:0d15a000 r5:81084b80 r4:00000014
[ 9.152205] [<8010f228>] (ipi_handler) from [<80184754>] (handle_percpu_devid_fasteoi_ipi+0x80/0x154)
[ 9.152536] [<801846d4>] (handle_percpu_devid_fasteoi_ipi) from [<8017dce4>] (__handle_domain_irq+0x8c/0xe0)
[ 9.152911] r7:00000001 r6:00000000 r5:00000000 r4:80c6fbdc
[ 9.153147] [<8017dc58>] (__handle_domain_irq) from [<80101348>] (gic_handle_irq+0x7c/0x90)
[ 9.153508] r9:810f6000 r8:8f80200c r7:80c6fbe8 r6:8f802000 r5:810f7f38 r4:80d0541c
[ 9.153819] [<801012cc>] (gic_handle_irq) from [<80100b0c>] (__irq_svc+0x6c/0x90)
[ 9.154168] Exception stack(0x810f7f38 to 0x810f7f80)
[ 9.154541] 7f20: 00000000 000195e0
[ 9.154993] 7f40: 8ddcb144 8011ba00 810f6000 00000001 80d04d10 80d04d4c 80d9dfdc 80b0b4f0
[ 9.155402] 7f60: 00000000 810f7f94 810f7f98 810f7f88 80108fd4 80108fd8 60070013 ffffffff
[ 9.155762] r9:810f6000 r8:80d9dfdc r7:810f7f6c r6:ffffffff r5:60070013 r4:80108fd8
[ 9.156070] [<80108f90>] (arch_cpu_idle) from [<80926c68>] (default_idle_call+0x38/0x108)
[ 9.156382] [<80926c30>] (default_idle_call) from [<80159894>] (do_idle+0xdc/0x148)
[ 9.156676] [<801597b8>] (do_idle) from [<80159bd0>] (cpu_startup_entry+0x28/0x2c)
[ 9.156974] r9:410fc075 r8:8000406a r7:80dbe280 r6:10c0387d r5:00000001 r4:00000091
[ 9.157275] [<80159ba8>] (cpu_startup_entry) from [<8010f7c0>] (secondary_start_kernel+0x15c/0x180)
[ 9.157614] [<8010f664>] (secondary_start_kernel) from [<8010182c>] (__enable_mmu+0x0/0x14)
[ 9.157919] r5:00000051 r4:810e006a
Fix this by not removing workdir directly, but keeping the directory
and removing all entires under it. We will still get below warning,
but it can boot to shell and provide ssh access to give a chance of
recovery.
[ 9.519053] overlayfs: failed to create directory run/initramfs/rw/work/work (errno: 28); mounting read-only
Also update related patch file in meta-google to avoid build failure.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I69c640c70ee3e9d1133b7d61d391501616c229d6
|
|
We don't actually store the firmware at a top-level directory, we expect
it to be stored in a hierarchy of directories.
Change-Id: I04eebb166d34616b346b401d1201574939cab60c
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Change-Id: Ic2096e1f49697d837ff94c0a7d51c0b447eb831f
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We want to produce a meaningful error if we time out unpacking the
download tarball.
Change-Id: I4710e70925d8c4be93430739f8a2f7e4ece47ebf
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We have a few machines where mutliple images are served to the BMC. We
want to allow for both images to be stored without processing.
Change-Id: I20d342981a5139d26b71c2e10c4861ae9dfbe432
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Update the dev_id.json directly after the install and before the
do_package to make sure it always get install/packaged properly.
Tested:
```
MACHINE=$PLATFORM GBMC_VERSION=0.0.0.5 bitbake phosphor-ipmi-config
...
{
"addn_dev_support": 2,
"aux": 327680,
"firmware_revision": {
"major": 0,
"minor": 0
},
"id": 32,
"manuf_id": 1234,
"prod_id": 5678,
"revision": 1
}
```
Change-Id: Ic43ba03fddb972d2dbbda64a31237574fbd39a34
Signed-off-by: Willy Tu <wltu@google.com>
|
|
This reverts commit e5ffc52b405319038480c826d0ed8c76edef7338.
Reason for revert: The gbmc_version task does not always get triggered properly due to bitbake cache. This may cause failure in the version check and be problematic.
Change-Id: I783ad8563ad16a1e6ed3e7f4d92323655e97afac
Signed-off-by: Willy Tu <wltu@google.com>
|
|
Read GBMC_VERSION and configure the ipmi firmware revision with the
major/minor. The aux is used for the point and subpoint. This allow gBMC
to remove the internal patch that we used to support our versioning
format.
Requires the change in
https://gerrit.openbmc.org/c/openbmc/phosphor-host-ipmid/+/54147/
Tested:
With firmware_revision in dev_id.json
```
$ cat /usr/share/ipmi-providers/dev_id.json
$ ipmitool mc info
{
"id": 0,
"revision": 0,
"addn_dev_support": 0,
"firmware_revision": {
"major": 35,
"minor": 16
},
"manuf_id": 11129,
"prod_id": 14426,
"aux": 0
}
ipmitool mc info
Device ID : 0
Device Revision : 0
Firmware Revision : 35.10
IPMI Version : 2.0
Manufacturer ID : 11129
Manufacturer Name : Google, Inc.
Product ID : 14426 (0x385a)
Product Name : Unknown (0x385A)
Device Available : yes
Provides Device SDRs : no
Additional Device Support :
Aux Firmware Rev Info :
0x00
0x00
0x00
0x00
```
Change-Id: I64ed6f54612d45732e366a2b245d33db00540093
Signed-off-by: Willy Tu <wltu@google.com>
|
|
William A. Kennington III (2):
ncsid: Guarantee MAC override
ncsid: Fix interruptible RA
Change-Id: If227ed6c1f707808ad6c91a072939357325d0139
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
We want to remove wget from busybox so we need an alternate fetcher.
Tested: Ran locally against an installer URL to verify the behavior
Change-Id: Ib3a00002d7d2d02bd6b29e24f0dbe2c7c9243514
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
The stateless addr has some routing issues outside of the machine under
some ToR configs. Use the stateful address as a preference to avoid
these issues.
Change-Id: I6a30678fd95b5096ee6b3b8add1d1dd409f0db5f
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Only clear the route from the list if the change command succeeds. If we
always clear it, we sometimes race with addresses coming and going. When
this happens, we don't ever update the source route because the route is
gone from the table when the address gets re-added.
Change-Id: I1d8658d8f0e23d74a2bdaf96cab3e399abe2b53e
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Default routes should be preferred by external means before going over
the bridge. This is important for nodes with multiple paths.
Change-Id: If8e1afd57001d6eed36d5b5856c111598d325944
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We only want to advertise that we have a default route if we have an
external (non-bridge) interface that has a default route. This
disambiguates routes on the internal machine network to avoid
blackholes.
Change-Id: I4059ddf0f74a0b15ca3f93f096b79415a851d3f7
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This consolidates all of the mechanisms which write out a persistent IP
into a single place. It also transitions to writing a very simple
persistent file instead of systemd style network units.
Change-Id: Ib99d7646178d2c5383cf23b09248bf24544c1d9e
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This is the only case where we need to add the rule, any addresses that
are BMC specific have a different workflow for assigning the firewall
rules.
Change-Id: If15c0cddb23044f774b55b8585446f923c76d7e8
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
We want to answer all of the possible BMC addresses for neighbor
discovery in cases where the bridge is not present on the BMC image.
Change-Id: Id8f0f929bbac2c8526a5ab87b410124357e4bd3d
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
Change-Id: I5abbeb2bd49a0434cd51a98fc63e25b6901865eb
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This fixes parsing addresses that have a weight metric.
Change-Id: Ibe93523caea0dccc77bec04e494f46c3db85bbf9
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
This makes it possible to perform to coalesce multiple powercycles into
a single one.
Change-Id: I748219a037e38b617b8d662eff498fd3b37acbb6
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
William A. Kennington III (1):
ncsid: update-ra-gw: Always keep RA updated
Change-Id: I3801ae6de5b7b009279eddea7729b5ed40e2611f
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
After updates Gerrit is now hosted at https://gerrit.openbmc.org.
Urls in the documentation are now updated to reflect this change.
Signed-off-by: Corey Hardesty <corey.hardesty@icloud.com>
Change-Id: I3a7f7d338ca1ce600ccf7b41fa03bd62007925ba
|
|
Willy Tu (2):
owners: Add wltu as owner/maintainer
pcie_bifurcation: Fetch bifurcation at a PCIe slot with hardcoded values
Change-Id: Ia1200c4d5fcb521c9f0399ccec71b70336669931
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Willy Tu (1):
google-ipmi-sys: Add SysPCIeSlotBifurcation subcommand placeholder
Change-Id: If2a11c8b65fdc525d434820749bc8431fbc908d9
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
This is required for routing to work correctly between the impersonated
node and the BMC. We still respond to neighbor solicits for the
impersonated node but our routing is fixed.
Change-Id: I7db57233348532576b3e2a9cd71ae0406367198e
Signed-off-by: William A. Kennington III <wak@google.com>
|