Flummi/Unraid-SlackPack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
8489766033
Unraid-SlackPack/source/freeipmi/usr/share/doc/freeipmi/freeipmi-bugs-issues-and-workarounds.txt
dmacias72 bc10199d49 add freeipmi source
2016-03-24 14:02:30 -06:00

1903 lines
59 KiB
Plaintext
Raw Blame History

FreeIPMI Bugs, Issues, and Workarounds
by
Albert Chu
chu11@llnl.gov
Last Updated: December 29, 2014
The following are some short descriptions of the issues I've found
while developing FreeIPMI on various motherboards. The descriptions
below also describe workarounds I've had to implement to get around
issues. When possible, the motherboards these issues were found on
are also documented.
Each of the issues below is listed under one of the following
categories:
[COMPLIANCE] - IPMI implementations that are out of compliance with
the IPMI specification. These issues range from simple
bugs in the implementation, to misinterpretation of the
IPMI specification, to flag out illegal
implementations.
[INTERPRETATION] - Interpretation ambiguity between multiple vendors
and the IPMI specification.
[IMPLEMENTATION] - Implementations that are legal IPMI
implementations, but implementation choices makes
IPMI poorer, confusing, or more difficult to use.
For the record, I have nothing against the vendors and the products
that have these issues. I am just as prone to programming bugs as
they are. The IPMI specification is gigantic and mistakes will
happen. I expect some design decisions were made for time to release,
cost, customer needs, etc. too. This documentation is mostly for
historical purposes, people searching for information, and to document
the difficulty of developing portable IPMI code.
----------------------
| Inband IPMI Issues |
----------------------
----------------------------------------------------------------------
Inband IPMI Issue #1: [COMPLIANCE]
The SMBIOS/DMI tables do not properly indicate that KCS base address
is IO instead of memory-mapped.
Seen on: HP ProLiant DL145 G1
Resolution: Workaround option added to inform software to use
memory-mapped IO.
----------------------------------------------------------------------
-----------------------------
| LAN Authentication Issues |
-----------------------------
----------------------------------------------------------------------
LAN Authentication Issue #1: [IMPLEMENTATION]
BMCs would not respond to retransmissions of a Get Session Challenge
Request if a previous Get Session Challenge response was lost.
Seen on: Intel SR870BN4/Tiger 4
Resolution: Workaround added to send retransmitted Get Session
Challenge requests from a different source port.
LAN Authentication Issue #2: [COMPLIANCE]
The only authentication supported by the motherboard is an OEM
specific authentication. IPMI specification requires several
non-OEM authentication mechanisms.
Seen on: Undocumented Motherboard
Resolution: None
Notes: Undocumented motherboard was an early Supermicro motherboard
with IPMI support, but I never documented which one (achu).
LAN Authentication Issue #3: [COMPLIANCE]
The initial outbound sequence number on activate session response is
off by one. The activate session response packet is supposed to
contain the initial outbound sequence number passed during the
request.
Seen on: Intel SE7520JR2 with National Semiconductor PC87431M mBMC,
Intel SE7520AF2 with Intel Server Management Module (Professional
Edition)
Resolution: Initially, workaround was done to accept off by one
sequence number. Later when packet windows were implemented, the
situation was automatically handled.
LAN Authentication Issue #4: [COMPLIANCE]
The IPMI 2.0 packet responses for RAKP 2 have payload lengths that
are off by 1 (i.e. an invalid length).
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: Workaround option added to inform software to handle the
invalid length.
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
LAN Authentication Issue #5: [COMPLIANCE]
Username capabilities and/or K_g status are not reported properly by
the Get Authentication Capabilities response.
Seen on: Asus P5M2, Asus P5MT-R, Asus S162-E4, Asus RX4, Sun Fire
X4150, Sun Fire X4450, Intel SR1520ML, Intel X38ML
Resolution: Workaround option added to inform software to ignore
flags in the Get Authentication Capabilities response.
LAN Authentication Issue #6: [COMPLIANCE]
The IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL flag does not work in the
Open Session phase of IPMI 2.0 connections. The Open Session
response seems to simply return the privilege level passed in the
request (i.e. the IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL flag rather
than the actual highest level privilege that can be used).
Seen on: Intel SE7520AF2 with Intel Server Management Module
(Professional Edition)
Resolution: Workaround option added to inform software to not pass
IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL and pass a specific privilege
level.
LAN Authentication Issue #7: [COMPLIANCE]
The username field is incorrectly padded during IPMI 2.0
authentication, leading to invalid keys.
Seen on: Intel SE7520AF2 with Intel Server Management Module
(Professional Edition)
Resolution: Workaround option added to inform software to adjust the
username when authenticating.
LAN Authentication Issue #8: [COMPLIANCE]
When the authentication algorithm is HMAC-MD5-128 and the password
is greater than 16 bytes, the Intel BMC incorrectly truncates the
password to 16 bytes when generating keys, leading to invalid keys.
Seen on: Intel SE7520AF2 with Intel Server Management Module
(Professional Edition)
Resolution: Workaround option added to inform software to adjust the
password when authenticating.
LAN Authentication Issue #9: [COMPLIANCE]
During the RAKP4 response, the integrity check value is incorrectly
calculated based on the integrity algorithm instead of the
authentication algorithm.
Seen on: Intel SE7520AF2 with Intel Server Management Module
(Professional Edition)
Resolution: Workaround option added to inform software to look for
and handle this situation.
LAN Authentication Issue #10: [COMPLIANCE]
During the RAKP3 request, the name_only_lookup field must be
disabled, it cannot be enabled.
Seen on: Intel SE7520AF2 with Intel Server Management Module
(Professional Edition)
Resolution: Workaround option added to inform software to disable
the field.
LAN Authentication Issue #11: [COMPLIANCE]
Cipher suite IDs are are incorrectly attached to specific privilege
levels rather than a maximum privilege level limit. So you can only
authenticate at the configured privilege level rather than a
privilege level <= to it.
Seen on: Undocumented Motherboard
Resolution: By passing IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL during the
session establishment instead of the specific privilege level, the
problem is avoided.
Notes: Undocumented motherboard was an early Tyan motherboard with
IPMI 2.0 support, but never documented which one (achu).
LAN Authentication Issue #12: [COMPLIANCE]
During a RAKP2 response, the key exchange authentication code is the
wrong length.
Seen on: Sun Fire 4100 with ILOM
Resolution: Workaround option added to inform software to handle the
invalid length.
LAN Authentication Issue #13: [COMPLIANCE]
The remote BMC incorrectly calculates keys using the privilege
specified in the open session stage rather than the privilege used
during the RAKP1 stage. This can be problematic if you specify
IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL during that stage instead of a
specific privilege level (i.e. operator).
Seen on: Sun Fire X4150, Inventec 5441/Dell Xanadu II, Supermicro
X8DTH, Supermicro X8DTG, Supermicro X8DTU, Intel S5500WBV/Penguin
Relion 700, Intel S2600JF/Appro 512X, Quanta QSSC-S4R/Appro
GB812X-CN
Resolution: Workaround option added to inform software to pass a
specific privilege level instead of
IPMI_PRIVILEGE_LEVEL_HIGHEST_LEVEL.
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II. Due to legacy, vendor reports will not fix on Intel
S5500WBV/Penguin Relion 700 and Intel S2600JF/Appro 512X.
LAN Authentication Issue #14: [COMPLIANCE]
Given specific configuration of IPMI 1.5 authentication types,
certain user privileges are not specified correctly during Get
Authentication Capabilities and a CCh = "Invalid data field in
request" is returned during Activiate Session.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II.
LAN Authentication Issue #15: [COMPLIANCE]
Disabling a specific cipher suite ID does not prohibit a user from
continuing to use that cipher suite ID.
Seen on: Supermicro X8DTG, Supermicro X8DTU, MSI MS-91D2, Intel
Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: None
LAN Authentication Issue #16: [COMPLIANCE]
Cipher suites 0 fails to authenticate. During the RAKP4
communication, an integrity key is returned, when it in fact should
be 0 length.
Seen on: Supermicro X8DTG, Supermicro X8DTU, Intel S5500WBV/Penguin
Relion 700, Intel S2600JF/Appro 512X
Resolution: Workaround option added to inform software to ignore the
non-zero length field.
Notes: Confirmed fixed in newer firmware for Supermicro X8DTG,
Supermicro X8DTU, Intel S2600JF/Appro 512X.
LAN Authentication Issue #17: [COMPLIANCE]
Cipher suites 7 and 8 fail to establish a session. Authentication
codes are not returned for session packets, therefore packets are
dropped as they are viewed as malformed.
Seen on: Supermicro X8DTG, Supermicro X8DTU
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro X8DTG,
Supermicro X8DTU.
LAN Authentication Issue #18: [COMPLIANCE]
Cipher suites 11 and 12 fail to establish a session. Authentication
codes returned are incorrect.
Seen on: Supermicro X8DTG, Supermicro X8DTU
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro X8DTG,
Supermicro X8DTU.
LAN Authentication Issue #19: [IMPLEMENTATION]
When attempting to authenticate with a disabled Cipher Suite ID, a
0Ah status ("Unauthorized role or privelege level requested.") is
returned instead of 11h ("No Cipher Suite match with proposed
security algorithms."). The resulting error message is confusing.
Seen on: Supermicro X8DTG
Resolution: None
LAN Authentication Issue #20: [IMPLEMENTATION]
When attempting to authenticate with a disabled Cipher Suite ID, a
12h status ("Illegal or Unrecognized parameter.") is returned
instead of 11h ("No Cipher Suite match with proposed security
algorithms."). The resulting error message is confusing.
Seen on: MSI MS-91D2
Resolution: None
LAN Authentication Issue #21: [COMPLIANCE]
The tag bits for some of the cipher records are wrong.
Seen on: Sun Fire 4100 with ILOM
Resolution: Workaround added to parse unique cipher suite records.
LAN Authentication Issue #22: [COMPLIANCE]
The motherboard reports support of Cipher Suite ID 15, however when
attemption authenticate with it, it is reported as unsupported.
Seen on: Intel S2600JF/Appro 512X
Resolution: None
Notes: Confirmed fixed in newer firmware.
LAN Authentication Issue #23: [COMPLIANCE]
The motherboard allows IPMI 1.5 authentication when > 16 byte
passwords are configured. It appears the motherboard allows users
to authenticate using the 16 byte equivalent truncated password.
This violates the text in Table 22-35, which states "A password that
has been stored as a 20-byte password cannot be used for
establishing an IPMI v1.5 session."
Seen on: Supermicro X8DTG-D
LAN Authentication Issue #24: [COMPLIANCE]
When activating a session, the remote system requires that the
"maximum privilege requested" specified in the request be identical
to the maximum set for a user. It cannot be less than it. When
this occurs, an error of 0xD4 ("Cannot execute command due to
insufficient privilege level") is unexpectedly returned.
Seen on: Xyratex HB-F8-SRAY
Resolution: Specify specific privilege level to workaround this
issue.
LAN Authentication Issue #24: [COMPLIANCE]
When authenticating with the 'straight password key' authentication
type, all responses from a session contain an empty authentication
code.
Seen on: Xyratex HB-F8-SRAY
Resolution: Workaround added to ignore these invalid authentication
codes.
LAN Authentication Issue #25: [COMPLIANCE]
Cannot authenticate with a user without assigning privilege limit of
Administrator.
Seen on: Xyratex HB-F8-SRAY
Resolution: None
LAN Authentication Issue #26: [COMPLIANCE]
Cannot authenticate with IPMI 1.5, authentication code generated
incorrectly.
Seen on: Intel Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: Workaround added to ignore these invalid authentication
codes.
----------------------------------------------------------------------
----------------------
| LAN Session Issues |
----------------------
----------------------------------------------------------------------
LAN Session Issue #1: [COMPLIANCE]
There is no response from the IPMI close command if a RESET is
executed.
Seen on: Intel SR870BN4/Tiger 4
Resolution: Workaround added to close session without waiting for
response.
LAN Session Issue #2: [COMPLIANCE]
After the IPMI session is brought up, packet responses return empty
session IDs (i.e. 00000000h) to the client.
Seen on: Tyan S2882 with m3289 BMC
Resolution: Workaround option added to inform software to accept
empty session IDs.
Notes: Problem is reported to be fixed in newer firmware for Tyan
S2882 with m3289 BMC.
LAN Session Issue #3: [COMPLIANCE]
When Per-Message Authentication is disabled, packet responses
contain non-null authentication data when they should be null.
Seen on: Dell PowerEdge 2850, Dell Poweredge SC1425
Resolution: Workaround option added to verify unexpected non-null
authentication codes as though they were expected.
Notes: Confirmed fixed in newer firmware for Dell PowerEdge 2850,
Dell Poweredge SC1425.
LAN Session Issue #4: [COMPLIANCE]
The remote BMC will advertise that Per Message Authentication is
disabled, but actually require it for the protocol.
Seen on: IBM eServer 325
Resolution: Workaround option added to inform software to use Per
Message Authentication no matter what is advertised by the remote
BMC.
Notes: Differs from Session Issue #5 below in resolution.
LAN Session Issue #5: [COMPLIANCE]
The remote BMC will advertise that Per Message Authentication is
disabled, but actually require it for the protocol.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: Workaround added to notice this condition and re-enable
Per Message Authentication.
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
LAN Session Issue #6: [COMPLIANCE]
Session sequence numbers have the wrong endian during IPMI 1.5
sessions.
Seen on: Sun Fire 4100 with ILOM
Resolution: Workaround option added to inform software to handle
incorrect endian sequence numbers.
LAN Session Issue #7: [COMPLIANCE]
LAN session headers contain invalid checksums.
Seen on: Supermicro X9SCM-iiF, Supermicro X9DRi-F, Supermicro X9DRFR
Resolution: Workaround option added to inform software to ignore bad
checksums.
----------------------------------------------------------------------
--------------
| SDR Issues |
--------------
----------------------------------------------------------------------
SDR Issue #1: [COMPLIANCE]
The record count reported by a Get SDR Repository Info command is
not consistent to the number of records that exist on the
motherboard.
Seen on: Fujitsu RX 100, Fujitsu RX300/200-S8
Resolution: Workaround added to recognize situation and adjust count.
SDR Issue #2: [COMPLIANCE]
The record ID returned from a Get SDR Record command is sometimes
different from the record ID passed in.
Seen on: Supermicro X6-DHR-1G with BMC2.0 daughter card
Resolution: No resolution in FreeIPMI required. Resolved in
IPMItool with workaround to detect issue and use input record ID
instead of output record ID.
SDR Issue #3: [COMPLIANCE]
When reading a SDR record, sometimes records list an invalid excess
number of bytes. For example, the SDR record is 15 bytes long, but
a record of length 20 is indicated.
Seen on: HP Proliant DL585G7
Resolution: Workaround added to adjust and handle the correct length
of the record.
SDR Issue #4: [COMPLIANCE]
When reading SDR records, the SDR last record id record is not
implemented, leading to errors when reading the SDR.
Seen on: Unspecified Inspur motherboard
Resolution: Workaround added to stop reading records after
recognized max.
SDR Issue #5: [COMPLIANCE]
When reading an entire SDR record (with
IPMI_SDR_READ_ENTIRE_RECORD_BYTES_TO_READ) the response returns
fewer bytes than the actual length of the record. This leads to
issues later during SDR caching. When reading SDR records with
partial reads, things work correctly.
Seen on: Xyratex HB-F8-SRAY
Resolution: Workaround added to detect issue and do parital reads
instead.
SDR Issue #6: [COMPLIANCE]
Retrieving sensors results in a "IPMI Internal Error" b/c a "Get SDR
Repository Info" command is not supported on the motherboard
(completion code 0xC1).
Seen on: Artesyn ATCA-7367
Resolution: None
SDR Issue #7: [COMPLIANCE]
Retrieving sensors results in a "IPMI Internal Error" b/c a "Reserve
SDR Repository" command is not supported on the motherboard
(completion code 0xC1).
Seen on: Artesyn ATCA-7367
Resolution: None
----------------------------------------------------------------------
--------------------------
| Serial-Over-LAN Issues |
--------------------------
----------------------------------------------------------------------
Serial-Over-LAN Issue #1: [COMPLIANCE]
Serial breaks are not supported/do not work.
Seen on: Undocumented Motherboard
Resolution: None
Notes: Undocumented motherboard was an early Tyan motherboard with
IPMI support 2.0, but I never documented which one (achu).
Serial-Over-LAN Issue #2: [COMPLIANCE]
SOL payload sizes are reported incorrectly.
Seen on: Asus P5M2, Asus P5MT-R, Asus S162-E4, Asus RX4, Intel
SR1520ML, Intel X38ML, Inventec 5441/Dell Xanadu II, Supermicro
X8DTH, Supermicro X8DTH-iF, Supermicro X8DTG, Supermicro X8DTU,
Quanta QSSC-S4R/Appro GB812X-CN, Intel S5500WBV/Penguin Relion 700
Resolution: Workaround option added to inform software to ignore
payload size checks and assume reasonable maximums.
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II. Problem is reported to be fixed in newer firmware for
Supermicro X8DTH-iF.
Serial-Over-LAN Issue #3: [COMPLIANCE]
A non-default SOL port is specified but not functional.
Seen on: Asus P5M2, Asus P5MT-R
Resolution: Workaround option added to inform software to ignore the
non-default port and assume the default SOL port is still used.
Serial-Over-LAN Issue #4: [COMPLIANCE]
The Get Channel Payload Support command is not supported.
Seen on: Sun Fire 4100 with ILOM, Intel Windmill, Quanta Winterfell,
Wiwynn Windmill
Resolution: Workaround option added to inform software to skip this
point in the SOL state machine.
Notes: This command is optional in the IPMI spec, it is not optional
if SOL is supported.
Serial-Over-LAN Issue #5: [COMPLIANCE]
SOL sessions are not deactivated after a Deactivate Payload request,
despite the response indicating success. This could lead to a loop
that continually believes a SOL session is active, tries to
deactivate it, believes it is deactivated, then checks again if it
is active.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: Code added to error after a number of failed deactivations.
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card. This is similar to issue #17 below.
Serial-Over-LAN Issue #6: [COMPLIANCE]
When a new IPMI session is started and steals the SOL session from a
previous one, the initial SOL session is not sent a "SOL is
deactivating" flag in a "BMC to Remote Console" packet. The initial
IPMI session should eventually time out, but the closing is not
clean.
In some cases, the IPMI session may "stay alive" and it is only
serial data that may time out, leading to a situation that may be
very similar to Serial-Over-LAN Issue #8.
Seen on: Supermicro H8QME with SIMSO daughter card, Inventec
5441/Dell Xanadu II, Inventec 5442/Dell Xanadu III, Dell Poweredge
R610, Dell Poweredge R710, Quanta S99Q/Dell FS12-TY, Supermicro
X8DTG, Supermicro X8DTU, Intel S2600JF/Appro 512X, Quanta
QSSC-S4R/Appro GB812X-CN, Wistron / Dell Poweredge C6220.
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card, Inventec 5441/Dell Xanadu II, Inventec
5442/Dell Xanadu III.
Serial-Over-LAN Issue #7: [IMPLEMENTATION]
During a reboot, SOL packets appear to be temporarily internally
dropped. This leads to a large increase in sequence numbers once
the SOL session is "re-connected". While the SOL session is
technically alive, the inability to predict/handle the sequence
number jump makes the SOL session difficult to handle, as you cannot
determine if packets are proper retransmissions or not.
Seen on: Tyan S4811 with SMDC daughter card
Resolution: Code added to error after a number of consecutive
invalid packets are received.
Serial-Over-LAN Issue #8: [IMPLEMENTATION]
After a reboot, the SOL session appears to "disconnect" from the
motherboard's serial chip. While the IPMI/SOL session is
technically alive, and character data input from the user is
accepted by the remote BMC, no character data is sent back from the
remote motherboard. The SOL session is subsequently useless.
Seen on: Intel SR1520ML, Intel X38ML
Resolution: The SOL session must be closed and restarted. Code to
implement "serial NUL pings" may aid software in detecting this
situation more quickly.
Serial-Over-LAN Issue #9: [COMPLIANCE]
The Activate Payload requires the "BMC asserts CTS and DCD/DSR to
baseboard upon activation" flag to be set, it is not optional.
Seen on: Undocumented Motherboard
Resolution: None
Notes: Undocumented motherboard is an Intel motherboard.
Serial-Over-LAN Issue #10: [IMPLEMENTATION]
If the remote system requires encryption, but the user attempts to
connect without encryption, an error response other than "Cannot
activate payload without encryption" is returned. The resulting
error message is confusing.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II.
Serial-Over-LAN Issue #11: [COMPLIANCE]
The Get Payload Activation Status command is not implemented
correctly. The completion code CCh = "Invalid data field in
request" is returned when attempted.
Seen on: Supermicro X8SIL-F
Resolution: Workaround option added to inform software to skip this
point in the SOL state machine.
Notes: Problem is reported to be fixed in newer firmware for
Supermicro X8SIL-F.
Serial-Over-LAN Issue #12: [COMPLIANCE]
The reported SOL port is specified w/ the wrong endian.
Seen on: Supermicro X8DTH-iF
Resolution: Workaround option added to inform software to ignore the
non-default port and assume the default SOL port is still used.
Notes: Problem is reported to be fixed in newer firmware for
Supermicro X8DTH-iF.
Serial-Over-LAN Issue #13: [COMPLIANCE]
The Get Payload Activation Status command does not correctly report
if a previous SOL session was created.
Seen on: Quanta S99Q/Dell FS12-TY
Resolution: None
Serial-Over-LAN Issue #14: [COMPLIANCE]
SOL appears to be allowed regardless of SOL being enabled / disabled
or a user being enabled / disabled for SOL access.
Seen on: MSI MS-91D2
Resolution: None
Serial-Over-LAN Issue #15: [COMPLIANCE]
The Activate / Deactivate Payload commands appear to require Admin
privileges, regardless of how SOL support is configured. The IPMI
specification indicates privilege requirements for executing these
commands depends on configuration.
Seen on: MSI MS-91D2
Resolution: None
Serial-Over-LAN Issue #16: [COMPLIANCE]
The number of SOL payload instances that can simultaneously be
activated on a motherboard is always reported as 0, suggesting that
it cannot be activated.
Seen on: Dell Poweredge M605, Dell Poweredge M610, Dell Poweredge
M915
Resolution: Workaround added to assume the instance capacity is
really 1 when it is 0. The assumption is made that the vendor does
not support multiple instances, which is why it is reported as 0.
Serial-Over-LAN Issue #17: [COMPLIANCE] or [IMPLEMENTATION]
SOL sessions are not deactivated after a Deactivate Payload request.
Despite the SOL session being active, it appears the BMC does not
know it is activated, and returns "Payload already deactivated"
(0x80) completion code. This could lead to a loop that continually
believes a SOL session is activated when calling Activate Payload,
and deactivated when calling Deactivate Payload.
Seen on: Dell Poweredge M605, Dell Poweredge M610, Dell Poweredge
M915
Notes: It is possible the Dell BMC does not allow a SOL payload
deactivation except for the IPMI session that activated the SOL
payload. While deactivation of a payload from another IPMI session
is allowed in IPMI (see Table 15-2 [2]), I am not sure if it is
required. So I'm listing this as [COMPLIANCE] or [IMPLEMENTATION].
This is similar to issue #5 above.
Serial-Over-LAN Issue #18: [COMPLIANCE]
SOL is always reported to have an active session in progress, so no
SOL session can be created. This includes after a deactivate
payload completes successfully.
It is possible this motherboard does not support SOL. However, it
should indicate that early in the session setup (perhaps via the Get
Channel Payload Support request).
Seen on: Xyratex HB-F8-SRAY
Resolution: None
Serial-Over-LAN Issue #19: [COMPLIANCE]
SOL payload instances are not completely independent of each other.
Deactivate of SOL payload instance #X may deactivate a different SOL
payload instance #Y.
Seen on: Supermicro X8DTG-D, Inventec 5442/Dell Xanadu III
Resolution: None
Serial-Over-LAN Issue #20: [COMPLIANCE]
The Activate Payload requires Serial/modem alerts to be deferred
instead of failed even though motherboards do not have serial/modem
ports to communicate over.
Seen on: Intel Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: Workaround added to defer serial alerts.
Serial-Over-LAN Issue #21: [COMPLIANCE]
The BMC occassionally gets stuck and an increase in the SOL packet
sequence number is required to get the machine unstuck.
Seen on: Intel Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: Workaround added to increase sequence number on severe
situations.
----------------------------------------------------------------------
---------------------------------
| BMC/IPMI Configuration Issues |
---------------------------------
----------------------------------------------------------------------
BMC/IPMI Configuration Issue #1: [COMPLIANCE]
BMC requires that a password must be passed to the Set User Password
command even if you are just trying to enable/disable a user. The
IPMI specification explicitly states that the password is optional
when enabling/disabling a user.
Seen on: Sun X4140, Intel Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: Workaround added to detect this situation and retry with
a dummy password.
BMC/IPMI Configuration Issue #2: [COMPLIANCE]
Users cannot be enabled/disabled via Set User Password.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #3: [COMPLIANCE]
Test passwords via Set User Password not supported.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #4: [COMPLIANCE]
SOL privilege level, Character Send Threshold, SOL Retry Count, and
SOL Retry interval cannot be modified.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #5: [COMPLIANCE]
User access parameters "Enable IPMI Msgs" and "Privilege Limit" are
not changeable via Set User Access.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #6: [COMPLIANCE] [CONFIG-1]
Default Channel configuration has a privilege limit of OEM even
though motherboard does not support an OEM privilege. Attempts to
modify sub-fields "piecemeal" within the Set Channel Access command
fail a parameter out of range error (0xC9).
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #7: [COMPLIANCE]
Setting a channel privilege limit (e.g. LAN channel privilege limit)
does not affect authentication and/or execution of IPMI commands
over LAN.
Seen on: Supermicro H8QME with SIMSO daughter card, Inventec
5441/Dell Xanadu II
Resolution: None
Notes: Confirmed fixed in newer firmware for Supermicro H8QME with
SIMSO daughter card.
BMC/IPMI Configuration Issue #8: [IMPLEMENTATION]
When commiting/writing large amounts of configuration data, some
data does not get stored or written properly. It appears that the
BMC may lose data if overwhelmed with data written to the BMC.
Seen on: Supermicro H8QME with SIMSO daughter card, Quanta S99Q/Dell
FS12-TY
Resolution: Workaround option added to inform software to slow down
writing of data.
BMC/IPMI Configuration Issue #9: [COMPLIANCE]
The ability to authenticate under standard IPMI 1.5 authentication
types (i.e. MD5) OEM authentication to be enabled. It appears to be
an internal logic error on the motherboard.
Seen on: Tyan S4811 with SMDC daughter card
Resolution: OEM authentication must be enabled in the BMC
configuration.
BMC/IPMI Configuration Issue #10: [IMPLEMENTATION]
Get Username and Get User Payload commands fail with CCh = "Invalid
data field in request" if a username was not set previously.
Seen on: Sun X4140, Inventec 5441/Dell Xanadu II
Resolution: Workaround added to notify user to configure username
first.
BMC/IPMI Configuration Issue #11: [IMPLEMENTATION]
User access parameters "Enable IPMI Msgs" and "Privilege Limit" are
not changeable via Set User Access until a username has been set.
Seen on: Inventec 5441/Dell Xanadu II, Supermicro X8DTH, Intel
S5500WBV/Penguin Relion 700
Resolution: Workaround added to notify user to configure username
first.
BMC/IPMI Configuration Issue #12: [COMPLIANCE]
When disabling LAN IPMI via Set User Access "Enable IPMI Msgs", IPMI
over LAN still works for a user.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II.
BMC/IPMI Configuration Issue #13: [COMPLIANCE]
Setting Non-volatile settings also changes Volatile settings for
channel configuration via Set Channel Access.
Seen on: Inventec 5441/Dell Xanadu II, Intel S5500WBV/Penguin Relion
700, Supermicro X8DTH
Resolution: None
BMC/IPMI Configuration Issue #14: [COMPLIANCE]
ARP response configuration is not functional even though the BMC ARP
response is available.
Seen on: Inventec 5441/Dell Xanadu II, Dell Poweredge R610, Dell
Poweredge R710
Resolution: None
Notes: This command is optional in the IPMI spec, but it is not
optional if ARP response is supported (see Table 23-4 [2]).
Confirmed fixed in newer firmware for Inventec 5441/Dell Xanadu II.
BMC/IPMI Configuration Issue #15: [COMPLIANCE]
Setting a password using IPMI 2.0 extensions (i.e. up to 20 bytes in
length) does not work.
Seen on: Undocumented Motherboard
Resolution: User must configure password using IPMI 1.5 formatted
payloads.
BMC/IPMI Configuration Issue #16: [IMPLEMENTATION]
User cannot be enabled/disabled until after a non-null password has
been configured.
Seen on: Dell Poweredge R610, Dell Poweredge R710, Wistron / Dell
Poweredge C6220.
Resolution: Workaround added to re-try enable/disable user after
password set.
BMC/IPMI Configuration Issue #17: [COMPLIANCE] [CONFIG-1]
Default Authentication Type Enables configure several OEM
authentications on. However, the motherboard does not support these
OEM authentications, and any "piecemeal" configuration of the
remaining sub-fields results in a bad parameter completion code
(0xCC).
Seen on: Dell Poweredge R610, Dell Poweredge R710
Resolution: Workaround added to notice this circumstance and attempt
to configure fields simultaneously.
BMC/IPMI Configuration Issue #18: [IMPLEMENTATION]
Read only fields return an error code of 0xD5 (request parameter not
supported) instead of 0x82 (read only parameter). The resulting
error message is confusing.
Seen on: Dell Poweredge R610, Dell Poweredge R710
Resolution: None
BMC/IPMI Configuration Issue #19: [COMPLIANCE]
User IPMI messaging being enabled/disabled does not affect
authentication and/or execution of IPMI commands for the user.
Seen on: Dell Poweredge R610, Dell Poweredge R710
Resolution: None
BMC/IPMI Configuration Issue #20: [COMPLIANCE]
Configuring the Subnet Mask will also lead to the "Default Gateway
IP Address" to be configured. It appears to be some type of memory
corruption on the motherboard firmware.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
BMC/IPMI Configuration Issue #21: [IMPLEMENTATION]
Several configuration fields, most notably several SOL configuration
fields, are stored in volatile memory rather than non-volatile
memory. These particular configuration fields were not required by
IPMI to be stored in non-volatile memory, but ultimately is an issue
b/c it requires users to reconfigure IPMI on every system boot.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
Notes: Resolved/enhanced in newer firmware for Inventec 5441/Dell
Xanadu II.
BMC/IPMI Configuration Issue #22: [COMPLIANCE]
Gratuitous ARP configuration can be enabled, but is not available on
the system.
Seen on: Quanta S99Q/Dell FS12-TY
Resolution: None
BMC/IPMI Configuration Issue #23: [COMPLIANCE]
MAC address is read only, but motherboard does not return read-only
error when changing the MAC address. This leads the user to believe
MAC address can be configured.
Seen on: Quanta S99Q/Dell FS12-TY
Resolution: None
BMC/IPMI Configuration Issue #24: [COMPLIANCE]
IPMI over LAN functions only if ARPs are enabled. If ARPs are
disabled and you choose to store the MAC address locally in an ARP
cache, IPMI over LAN will still not function.
Seen on: Quanta S99Q/Dell FS12-TY
Resolution: None
BMC/IPMI Configuration Issue #25: [IMPLEMENTATION]
A default username cannot be changed.
Seen on: Supermicro X8DTG, Supermicro X8DTU, Intel S5500WBV/Penguin Relion 700
Resolution: None
BMC/IPMI Configuration Issue #26: [COMPLIANCE]
User access parameters configured via Set User Access do not work
for some users.
Seen on: Supermicro X8DTG, Supermicro X8DTU
Resolution: None
BMC/IPMI Configuration Issue #27: [COMPLIANCE]
Cipher Suites 6, 7, 8, 11, and 12 are not available on the board,
but their Cipher Suite's Maximum Privilege Level are configurable,
suggesting to the user they are available.
Seen on: Supermicro X8DTG, Supermicro X8DTU
Resolution: None
BMC/IPMI Configuration Issue #28: [COMPLIANCE]
Multiple LAN Channels exist, however when asking the BMC how many
SOL Channels there are, it indicates only 1. Generally speaking,
this need not be a bug. However, for this particular motherboard it
is.
Seen on: Intel S5500WBV/Penguin Relion 700
Resolution: None
BMC/IPMI Configuration Issue #29: [COMPLIANCE]
The number of entries returned from a RMCP+ Messaging Cipher Suite
Privilege Levels request is not valid. Not only is it not valid,
the number of entries does not even match the number of entries
specified by a RMCP+ Messaging Cipher Suite Entry Support Count
request.
Seen on: HP DL145
Resolution: Workaround added to not output a privilege if one is not
available.
BMC/IPMI Configuration Issue #30: [COMPLIANCE]
Cipher Suite 1-16 are listed as supported instead of cipher suites
0-15.
Seen on: Intel S2600JF/Appro 512X
Resolution: Workaround added to detect this specific situation and
adjust appropriately.
Notes: Confirmed fixed in newer firmware.
BMC/IPMI Configuration Issue #31: [IMPLEMENTATION]
Some new IPMI configuration are not accepted by the BMC until after
a system reboot is done.
Seen on: Dell Poweredge R610, Dell Poweredge R710, Inventec
5441/Dell Xanadu II
Resolution: None
Notes: A cold reset of the BMC is sufficient to resolve this issue
on Inventec 5441/Dell Xanadu II without rebooting the machine.
BMC/IPMI Configuration Issue #32: [COMPLIANCE]
Motherboard errors out with unspecified password when trying to set
a new user password (completion code 0x8F).
Seen on: Wistron / Dell Poweredge C6220
Resolution: None
Notes: Motherboard has internal password requirements leading to
error.
BMC/IPMI Configuration Issue #33: [COMPLIANCE]
When retrieving boot flags via Get System Boot Options, the response
payload under IPMI 2.0 contains an extra 12 bytes of
unknown/unspecified data.
Seen on: HP DL380 G5
Resolution: None
BMC/IPMI Configuration Issue #34: [COMPLIANCE]
Many configuration options return "Parameter Not Supported" (0x80)
when attempting to set a value, but do not when the parameter is
being read. It is likely the firmware writers meant that the
parameter was read only (0x82).
Seen on: Xyratex HB-F8-SRAY
Resolution: None
BMC/IPMI Configuration Issue #35: [COMPLIANCE]
User #1 cannot be enabled/disabled. It is permanently enabled.
Seen on: Supermicro X10DDW-i
Resolution: None
BMC/IPMI Configuration Issue #36: [COMPLIANCE] [CONFIG-1]
Default user privilege limit is RESERVED. Attempts to
modify sub-fields "piecemeal" within the Set User Access command
fail.
Seen on: Supermicro X10DDW-i
Resolution: Workaround added to work around this situation.
BMC/IPMI Configuration Issue #37: [COMPLIANCE]
IPMI LAN session w/ IPMI 1.5 times out after changing a user
password within the LAN session.
Seen on: Supermicro X9DRT-HF+
Resolution: It appeared that this motherboard was still accepting
request packets using the old Password for the authentication hash
but was responding with packets using the new Password for the hash.
Ignoring the authentication code of response packets was sufficient
to work around the problem. It was also discovered communicating
with IPMI 2.0 worked and did not exhibit this problem.
----------------------------------------------------------------------
-------------------------------------------------------
| Platform Event Filtering (PEF) Configuration Issues |
-------------------------------------------------------
----------------------------------------------------------------------
Platform Event Filtering Configuration Issue #1: [IMPLEMENTATION] [CONFIG-1]
Alert Policy configurations cannot be configured "piecemeal". If
all fields are not configured perfectly and simultaneously, every
configuration will result in a completion code of 0xCC (Invalid data
field in Request).
Seen on: Inventec 5441/Dell Xanadu II, Fujitsu RX 100
Resolution: Workaround added to notice this circumstance and attempt
to configure fields simultaneously.
----------------------------------------------------------------------
---------------------------------------
| Field Replaceable Unit (FRU) Issues |
---------------------------------------
----------------------------------------------------------------------
Field Replaceable Unit Issue #1: [COMPLIANCE]
Some FRU data has invalid checksums.
Seen on: Undocumented Motherboard, Inventec 5441/Dell Xanadu II,
Dell Poweredge R610, Dell Poweredge R710,
Resolution: Workaround option added to inform software to not check
checksums.
Notes: Undocumented motherboard was an Intel Tiger motherboard, but
never documented which one (achu).
Field Replaceable Unit Issue #2: [IMPLEMENTATION]
A FRU Chassis Info Area reports an unspecified chassis type.
Seen on: Quanta S99Q/Dell FS12-TY
Resolution: None
----------------------------------------------------------------------
------------------
| Sensors Issues |
------------------
----------------------------------------------------------------------
Sensors Issue #1: [COMPLIANCE]
Some sensors don't return a sensor state (16 bit field) although
only half of the sensor state (8 bit field) is optional.
Seen on: Dell 2950, Inventec 5441/Dell Xanadu II
Resolution: Workaround added to assume sensor state is 0 when this
occurs.
Sensors Issue #2: [COMPLIANCE]
Some sensors return sensor states that are illegal. For example, a
sensor state of 0x40 is returned, but 0x1 through 0x20 are the only
legal ones.
Seen on: Dell 2950, SuperMicro X9DRI-LN4F+
Resolution: Workaround added to ignore invalid sensor states.
Sensors Issue #3: [COMPLIANCE] [SENSORS-1]
Some sensors return
IPMI_COMP_CODE_REQUEST_SENSOR_DATA_OR_RECORD_NOT_PRESENT (0xCB)
completion codes for unknown reasons.
Seen on: Dell 2950
Resolution: Workaround added to assume sensor not available.
Sensors Issue #4: [COMPLIANCE] [SENSORS-1]
Some sensors return
IPMI_COMP_CODE_COMMAND_ILLEGAL_FOR_SENSOR_OR_RECORD_TYPE (0xCD) for
unknown reasons.
Seen on: Sun V20Z
Resolution: Workaround added to assume sensor not available.
Sensors Issue #5: [COMPLIANCE] [SENSORS-1]
Some sensors return IPMI_COMP_CODE_PARAMETER_OUT_OF_RANGE (0xC9)
completion codes for unknown reasons.
Seen on: Dell 2650
Resolution: Workaround added to assume sensor not available.
Sensors Issue #6: [COMPLIANCE] [SENSORS-1]
Some sensors return IPMI_COMP_CODE_COMMAND_CANNOT_RESPOND (0xCE)
completion codes for unknown reasons.
Seen on: HP ProLiant DL145 G2
Resolution: Workaround added to assume sensor not available.
Sensors Issue #7: [COMPLIANCE]
The Get Sensor Event Enable command always returns "sensor scanning
disabled" bit, even if sensor scanning is not disabled.
Seen on: Sun X4140
Resolution: None
Sensors Issue #8: [COMPLIANCE]
Several sensors have invalid Entity IDs. In other words, not in the
valid Entity ID range or OEM defined range.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None, generic unknown Entity ID type listed.
Sensors Issue #9: [COMPLIANCE]
Sensor bridging over IPMI 1.5 fails b/c IPMB respones have invalid
authentication codes.
Seen on: Inventec 5441/Dell Xanadu II, Inventec 5442/Dell Xanadu III
Resolution: Workaround option added to inform software to ignore
authentication codes in response packets.
Notes: Confirmed fixed in newer firmware for Inventec 5442/Dell
Xanadu III.
Sensors Issue #10: [COMPLIANCE] [SENSORS-1]
Some sensors return IPMI_COMP_CODE_REQUEST_PARAMETER_NOT_SUPPORTED
(0xD5) completion codes for unknown reasons.
Seen on: Sun Fire X4450 w/ ELOM 3.0.0
Resolution: Workaround added to assume sensor not available.
Sensors Issue #11: [COMPLIANCE]
Some sensors report invalid event bitmasks. For example, when the
lower critical threshold has not been crossed, the lower critical
threshold event bitmask is set.
Seen on: Intel SR2500, Intel SR1500SAS, Intel S5000PAL
Resolution: None
Sensors Issue #12: [COMPLIANCE]
Sensors on the motherboard are owned by the BMC, but the SDR reports
that it is not. Thus, sensor bridging fails.
Seen on: HP Proliant DL140 G1, Fujitsu RX300, Fujitsu RX300S2
Resolution: Workaround option added to inform software to assume the
sensor owner is the BMC no matter what.
Notes: Confirmed fixed in newer firmware for Fujitsu RX300, Fujitsu
RX300S2.
Sensors Issue #13: [INTERPRETATION] [IMPLEMENTATION]
Sensors do not assert state bits (i.e. sensor event bitmask)
consistently to indicate that a sensor is "doing good". For
example, for a "Power Supply" sensor, offset 00h (i.e. state 0 or
bitmask 0x0001) indicates "Presence Detected". On some motherboards
this is considered to be the "good" result to see from a "Power Supply"
sensor reading. On some other motherboards, the motherboards have
elected to not return any offset (i.e. no states set or bitmask
0x0000) when the power supply is fine, electing to only return an
offset when an actual failure occurs.
Seen on: Various, this is a general issue.
Resolution: None
Sensors Issue #14: [IMPLEMENTATION]
Event only sensors are listed in the SDR using a full or compact SDR
record. Software is very likely to believe these are sensors that
are readable, leading to non-optimal performance.
Seen on: Various, this is a general issue.
Resolution: None
Sensors Issue #15: [INTERPRETATION]
SDR records for some sensors list an event/reading type code for a
discrete sensor, but return an analog reading from a Get Sensor
Reading command. (The SDR's analog data format indicates an
analog/threshold based sensor, however no threshold limits for the
sensor are listed in the SDR.)
This was discovered when FreeIPMI and IPMItool sensor output
mismatched. In FreeIPMI, ipmi-sensors would report discrete states
(e.g. "At or Below (<=) Lower Non-Critical Threshold") without an
actual sensor reading (e.g. 100 Watts). In IPMItool, it would
output the sensor reading (e.g. 100 Watts) but invalid/incorrect
discrete states.
IMO, a SDR record that indicates an analog reading w/ a discrete
event/reading type code is illegal. At the minimum it makes no
sense. If you report an analog reading without any treshold
information, what is the use? While it is not explicitly stated in
the IPMI specification, this is strongly implied to be wrong. In
42.1, it states for threshold sensors, "The Get Sensor Reading
command for a threshold-based sensor contains the present 'analog'
reading for the sensor along in addition to the discrete threshold
comparison status bit field."
Seen on: HP Proliant DL380 G7, HP ProLiant ML310 G5
Resolution: Workaround option added to inform software to read the
sensor reading if there are enough hints it may be available.
Sensors Issue #16: [INTERPRETATION]
The analog data format lists 4 possible conditions: 00b = unsigned,
01b = 1's complement, 10b = 2's complement, 11b = does not return
analog reading. This strongly suggests that any SDR entry that does
not report an analog reading should set this field to 11b. However,
text below this section states, "If neither thresholds nor analog
reading are provided, this field should be written as 00h." Some
vendors set this field to 00b for non-analog sensors, while others
set it to 11b.
Seen on: Various, this is a general issue.
Resolution: None
Sensors Issue #17: [COMPLIANCE]
Some sensors report that their sensor scanning is disabled, even
though the sensor is actually not.
Seen on: Dell Poweredge 2900, Dell Poweredge 2950, Dell Poweredge
R410, Dell Poweredge R610, HP Integrity rx3600
Resolution: Workaround option added to inform software to ignore the
scanning bit and read a sensor reading regardless of the scanning
bits value.
Sensors Issue #18: [COMPLIANCE] [SENSORS-1]
Some sensors return IPMI_COMP_CODE_UNSPECIFIED_ERROR
(0xFF) completion codes for unknown reasons.
Seen on: Sun Blade x6250
Resolution: Workaround added to assume sensor not available.
Sensors Issue #19: [COMPLIANCE] [SENSORS-1]
Some sensors return IPMI_COMP_CODE_DESTINATION_UNAVAILABLE (0xD3)
completion codes for unknown reasons.
Seen on: Sun Blade 6000M2
Resolution: Workaround added to assume sensor not available.
Sensors Issue #20: [IMPLEMENTATION]
Some sensors are given a unit modifier (i.e. A / B, or A * B) but a
modifier unit of unspecified. This leads to sensors with units
like: "C / unspecified".
Seen on: Wistron / Dell Poweredge C6220
Resolution: Workaround added to handle corner case.
Sensors Issue #21: [IMPLEMENTATION]
Some SDR entries report RPM thresholds that are inconsistent
and illogical. For example, a upper critical threshold will be
reported that is below the nominal reading.
Seen on: HP DL140 G3 and HP ML160 G6
Resolution: None
----------------------------------------------------------------------
---------------------------------
| System Event Log (SEL) Issues |
---------------------------------
----------------------------------------------------------------------
System Event Log Issue #1: [COMPLIANCE]
SEL records incorrectly report the generator_id slave address.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: Workaround added to handle this as a special case and
adjust the generator ID appropriately.
System Event Log Issue #2: [COMPLIANCE]
Different record ids (e.g. 280 vs. 281) can be used to retrieve
identical SEL records.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: None
System Event Log Issue #3: [COMPLIANCE]
The Reserve SEL IPMI command incorrectly requires an Operator
privilege Level instead of a User privilege level.
Seen on: Supermicro H8QME with SIMSO daughter card
Resolution: Workaround added to simply continue without a
reservation ID.
System Event Log Issue #4: [COMPLIANCE]
The Get SEL Entry command is not supported.
Seen on: HP DL 585
Resolution: None
System Event Log Issue #5: [COMPLIANCE]
Some SEL records report illegal record types.
Seen on: HP DL 380 G5, Intel S2600JF/Appro 512X
Resolution: Workaround option added to inform software to assume the
record is a normal system event.
System Event Log Issue #6: [IMPLEMENTATION]
The SEL contains sensor numbers and sensor owner IDs which are not
listed in the SDR. This can lead to an output that contains less
useful information.
Seen on: Inventec 5441/Dell Xanadu II, Sun Fire X4150
Resolution: None
System Event Log Issue #7: [IMPLEMENTATION]
SEL events showing a threshold sensor do not properly report if they
are going high or going low, which can lead to confusion for
users. For example, the following set of events might occur.
Upper Non-critical - going high ; Sensor Reading = 242.00 W ; Threshold = 240.00 W
Upper Non-critical - going high ; Sensor Reading = 180.00 W ; Threshold = 240.00 W
The following would be preferred
Upper Non-critical - going high ; Sensor Reading = 242.00 W ; Threshold = 240.00 W
Upper Non-critical - going low ; Sensor Reading = 180.00 W ; Threshold = 240.00 W
It is very likely the implementors only thought of
asserting/deasserting identical events, without considerations of
the actual event messages and the resulting output that can occur.
Seen on: Dell Poweredge C410x
Resolution: None
Notes:
The following SEL output in ipmitool shows how it can lead to even more confusion.
Upper Non-critical going high | Reading 240 > Threshold 240 Watts
Upper Non-critical going high | Reading 88 > Threshold 240 Watts
It reads "88 > 240"
----------------------------------------------------------------------
-----------------------------
| BMC Watchdog Timer Issues |
-----------------------------
----------------------------------------------------------------------
BMC Watchdog Timer Issue #1: [COMPLIANCE]
The timer is running/stopped flag appears to be stuck on "stopped"
regardless if the timer is running or not. Functionality may be
limited due to this.
Seen on: Sun x4100, Sun x4200, Sun x4500
Resolution: Workaround option added to inform software to analyze
countdown rather than flag for running vs stopped state.
Notes: Confirmed fixed in newer firmware for Sun x4100, Sun x4200,
Sun x4500.
BMC Watchdog Timer Issue #2: [???]
A reply from a Get Watchdog Timer command returns a packet without
valid data.
Seen on: Unspecified Supermicro
Resolution: Handle special case and retry command.
----------------------------------------------------------------------
---------------
| DCMI Issues |
---------------
----------------------------------------------------------------------
DCMI Issue #1: [COMPLIANCE]
DCMI is reported as available, when it is not.
Seen on: Inventec 5441/Dell Xanadu II
Resolution: None
Notes: Confirmed fixed in newer firmware for Inventec 5441/Dell
Xanadu II.
DCMI Issue #2: [INTERPRETATION]
The DCMI spec indicates a potential completion code for the "Get
Power Limit" command as "No Set Power Limit" (0x80). FreeIPMI
originally interpreted this to mean the "Set Power Limit" command
was not available. Atleast one vendor interpreted this to mean "No
Power Limit Set". One can consider this an English interpretation
issue of 'No set POWER LIMIT' vs. 'No SET POWER LIMIT' (i.e. is
"set" a verb or part of a proper noun referencing the DCMI command).
Confounding this issue is the fact that the example implementation
in Intel's DCMItool implements the former, while the DCMI
Conformance test suite implements the latter. In addition to this,
with the latter interpretation, it need not be an indication of an
error, but rather a flag. So the rest of the packet can be
completely full of legitimate data.
Seen on: Various, this is a general issue.
Resolution: Code added to handle both interpretations.
DCMI Issue #3: [INTERPRETATION]
The DCMI 1.1 Specification includes new commands to set the asset
tag and management controller identifier string. The payload to
write these strings is:
Byte 1 - Group Extension (DCh)
Byte 2 - Offset to Write
Byte 3 - Bytes to Write (max 16)
Byte 4-N - Data
and the response is:
Byte 1 - Completion Code
Byte 2 - Group Extension (DCh)
Byte 3 - Total Length Written
The maximum size for both strings is listed at 64 bytes.
In the spec, it is unclear how this payload is to be used. It is
unclear if the "Total Length Written" field is similar to a posix
write, where <= the bytes passed in may be written. Or is it an
indication of the total length of bytes stored? The management
identifier string is stated to be NUL terminated, but the asset tag
is not stated with this requirement.
Seen on: Various, this is a general issue.
Resolution: Code assumes that all 64 bytes of the buffer must be
overwritten everytime the string is modified. This seems to be the
safest approach.
DCMI Issue #4: [COMPLIANCE]
The Set Asset Tag command does not return the length of the asset
tag written.
Seen on: Unspecified Supermicro Motherboard
Resolution: None
----------------------------------------------------------------------
------------------------------------
| Platform Event Trap (PET) Issues |
------------------------------------
----------------------------------------------------------------------
Platform Event Trap Issue #1: [COMPLIANCE]
Responses from a PET Acknowledge are invalid and malformed. The
payload is short by one byte (i.e. does not contain a completion
code).
Seen on: Dell Poweredge R610
Resolution: Workaround option added to inform software to accept
malformed packets.
----------------------------------------------------------------------
--------------------
| Misc IPMI Issues |
--------------------
----------------------------------------------------------------------
Misc IPMI Issue #1: [COMPLIANCE]
The Get System Info Parameters command incorrectly requires
administrator privileges rather than the defined user level
privileges.
Seen on: MSI MS-91D2
Resolution: None, user should attempt to authenticate at a higher
privilege level.
Misc IPMI Issue #2: [COMPLIANCE]
The Get System Info Parameters command returns insufficient
privilege level (0xD4) even when the privilege level is sufficient.
Seen on: Supermicro X8DTL-3F, Supermicro X9DRFR
Resolution: None, user should attempt to authenticate at a higher
privilege level.
Misc IPMI Issue #3: [COMPLIANCE]
The manufacturer ID specified by the motherboard was invalid
(i.e. not a valid IANA Enterprise number).
Seen on: Supermicro X8DTH, Supermicro X8DTG, Supermicro X8DTU
Resolution: Workaround added to recognize special case IANA number
as belonging to a specific manufacturer.
Misc IPMI Issue #4: [INTERPRETATION]
When retrieving the operating system name via Get System Info
Parameters, the 'set_selector', 'encoding', or 'string_length'
fields in the response is not set. It is not clear if this is an
error in their packet for indicating an unsupported field or an
empty field.
Seen on: Bull 510 Blade.
Resolution: Workaround added to bmc-info to handle special case.
Misc IPMI Issue #5: [COMPLIANCE]
The Get System Info Parameters command returns insufficient
privilege level (0xD4) regardless of privilege level.
Seen on: Intel Windmill, Quanta Winterfell, Wiwynn Windmill
Resolution: None
Misc IPMI Issue #6: [COMPLIANCE]
The Get SDR Repository Info command returns insufficient privilege
level (0xD4) regardless of privilege level.
Seen on: Wiwynn Windmill
Resolution: None, user should attempt to authenticate at a higher
privilege level.
Misc IPMI Issue #7: [COMPLIANCE]
System does not support RMCP ping/pong.
Seen on: Intel Windmill, Quanta Winterfell, Wiwynn Windmill
----------------------------------------------------------------------
------------
| Appendix |
------------
[CONFIG-1]
To see a discussion on why this implementation is poor, please see
"Configuration tool callback design" in the freeipmi-design.txt
document.
[SENSORS-1]
In all liklihood, these are all circumstances that the vendor wanted
to say "this sensor isn't available on this motherboard". It is
ironic that there are so many error codes vendors are returning for
this circumstance. It's also strange that so many vendors don't set
the "reading unavailable" or "sensor scanning disabled" bit when
returning the Get Sensor Reading command. But I digress.
I get the impression vendors try to put the same SDR on multiple
motherboards, regardless of whether a motherboard can even have a
sensor reading/event represented by that SDR entry, leading to the
strange range of error codes.
The best thing vendors could do is have an SDR for each motherboard
that is actually legit. The IPMI spec strongly suggests this as well:
"Sensor Data Records are data records that contain information about
the type and number of sensors in the platform ..." (1.7.7).
"The SDR Repository is intended to hold information indicating the set
of management controllers, sensors, and FRU Devices that is expected
to be in the system ... For example, suppose the baseboard had
connectors for five fans, but only the first four were supposed to be
populated. The SDRs for the system would report four fan sensors ..."
(33)
Reference in New Issue View Git Blame Copy Permalink