More missed changes, the response back to another system sending a
command that had no user to handle it wasn't formatted properly.
Signed-off-by: Corey Minyard <[email protected]>
A couple of issues:
The tested data sizes are wrong; during the design that changed and this
got missed.
The formatting of the reponse couldn't use the normal one, it has to be
an IPMB formatted response.
Reported-by: Jakub Kicinski <[email protected]>
Fixes: 059747c245 ("ipmi: Add support for IPMB direct messages")
Signed-off-by: Corey Minyard <[email protected]>
We're hitting OOB accesses in handle_ipmb_direct_rcv_rsp() (memcpy of
size -1) after user space generates a message. Looks like the message
is incorrectly assumed to be of the new IPMB type, because type is never
set and message is allocated with kmalloc() not kzalloc().
Fixes: 059747c245 ("ipmi: Add support for IPMB direct messages")
Signed-off-by: Jakub Kicinski <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
The sparse tool complains as follows:
drivers/char/ipmi/ipmi_msghandler.c:194:25: warning:
symbol 'remove_work_wq' was not declared. Should it be static?
This symbol is not used outside of ipmi_msghandler.c, so
marks it static.
Fixes: 1d49eb91e8 ("ipmi: Move remove_work to dedicated workqueue")
Reported-by: Hulk Robot <[email protected]>
Signed-off-by: Wei Yongjun <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Currently when removing an ipmi_user the removal is deferred as a work on
the system's workqueue. Although this guarantees the free operation will
occur in non atomic context, it can race with the ipmi_msghandler module
removal (see [1]) . In case a remove_user work is scheduled for removal
and shortly after ipmi_msghandler module is removed we can end up in a
situation where the module is removed fist and when the work is executed
the system crashes with :
BUG: unable to handle page fault for address: ffffffffc05c3450
PF: supervisor instruction fetch in kernel mode
PF: error_code(0x0010) - not-present page
because the pages of the module are gone. In cleanup_ipmi() there is no
easy way to detect if there are any pending works to flush them before
removing the module. This patch creates a separate workqueue and schedules
the remove_work works on it. When removing the module the workqueue is
drained when destroyed to avoid the race.
[1] https://bugs.launchpad.net/bugs/1950666
Cc: [email protected] # 5.1
Fixes: 3b9a907223 (ipmi: fix sleep-in-atomic in free_user at cleanup SRCU user->release_barrier)
Signed-off-by: Ioanna Alifieraki <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
coccicheck complains about the use of snprintf() in sysfs show
functions:
WARNING use scnprintf or sprintf
Use sysfs_emit instead of scnprintf, snprintf or sprintf makes more
sense.
Reported-by: Zeal Robot <[email protected]>
Signed-off-by: Ye Guojin <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
When CONFIG_I2C=m, CONFIG_I2C_SLAVE=y (bool), and CONFIG_IPMI_IPMB=y,
the build fails with:
ld: drivers/char/ipmi/ipmi_ipmb.o: in function `ipmi_ipmb_remove':
ipmi_ipmb.c:(.text+0x6b): undefined reference to `i2c_slave_unregister'
ld: drivers/char/ipmi/ipmi_ipmb.o: in function `ipmi_ipmb_thread':
ipmi_ipmb.c:(.text+0x2a4): undefined reference to `i2c_transfer'
ld: drivers/char/ipmi/ipmi_ipmb.o: in function `ipmi_ipmb_probe':
ipmi_ipmb.c:(.text+0x646): undefined reference to `i2c_slave_register'
ld: drivers/char/ipmi/ipmi_ipmb.o: in function `ipmi_ipmb_driver_init':
ipmi_ipmb.c:(.init.text+0xa): undefined reference to `i2c_register_driver'
ld: drivers/char/ipmi/ipmi_ipmb.o: in function `ipmi_ipmb_driver_exit':
ipmi_ipmb.c:(.exit.text+0x8): undefined reference to `i2c_del_driver'
This is due to having a tristate depending on a bool symbol.
By adding I2C (tristate) as a dependency, the desired dependencies
are met, causing IPMI_IPMB to be changed from =y to =m:
-CONFIG_IPMI_IPMB=y
+CONFIG_IPMI_IPMB=m
Fixes: 63c4eb3471 ("ipmi:ipmb: Add initial support for IPMI over IPMB")
Signed-off-by: Randy Dunlap <[email protected]>
Cc: Corey Minyard <[email protected]>
Cc: [email protected]
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Add direct OF support for fetching control parameters from the device
tree. Make it work like the device tree entries for the other IPMI
devices. Also add documentation for this.
Signed-off-by: Corey Minyard <[email protected]>
This driver was originally written to use the regmap abstraction with no
clear benefit. As the registers are always mmio and there is no sharing
of the region with other devices, we can safely read and write without
the locking that regmap provides.
This reduces the code size of the driver by about 25%.
Signed-off-by: Joel Stanley <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
There is an off-by-one bounds check on the rcvlen causing a potential
out of bounds write on iidev->rcvmsg. Fix this by using the >= operator
on the bounds check rather than the > operator.
Addresses-Coverity: ("Out-of-bounds write")
Fixes: 0ba0c3c5d1c1 ("ipmi:ipmb: Add initial support for IPMI over IPMB")
Signed-off-by: Colin Ian King <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
This provides access to the management controllers on an IPMB bus to a
device sitting on the IPMB bus. It also provides slave capability to
respond to received messages on the bus.
Signed-off-by: Corey Minyard <[email protected]>
Tested-by: Andrew Manley <[email protected]>
Reviewed-by: Andrew Manley <[email protected]>
An application has come up that has a device sitting right on the IPMB
that would like to communicate with the BMC on the IPMB using normal
IPMI commands.
Sending these commands and handling the responses is easy enough, no
modifications are needed to the IPMI infrastructure. But if this is an
application that also needs to receive IPMB commands and respond, some
way is needed to handle these incoming commands and send the responses.
Currently, the IPMI message handler only sends commands to the interface
and only receives responses from interface. This change extends the
interface to receive commands/responses and send commands/responses.
These are formatted differently in support of receiving/sending IPMB
messages directly.
Signed-off-by: Corey Minyard <[email protected]>
Tested-by: Andrew Manley <[email protected]>
Reviewed-by: Andrew Manley <[email protected]>
Don't do kfree or other risky things when oops_in_progress is set.
It's easy enough to avoid doing them
Signed-off-by: Corey Minyard <[email protected]>
You will get two decrements when the messages on a panic are sent, not
one, since commit 2033f68589 ("ipmi: Free receive messages when in an
oops") was added, but the watchdog code had a bug where it didn't set
the value properly.
Reported-by: Anton Lundin <[email protected]>
Cc: <[email protected]> # v5.4+
Fixes: 2033f68589 ("ipmi: Free receive messages when in an oops")
Signed-off-by: Corey Minyard <[email protected]>
Pull IPMI updates from Corey Minyard:
"A couple of very minor fixes for style and rate limiting.
Nothing big, but probably needs to go in"
* tag 'for-linus-5.15-1' of git://github.com/cminyard/linux-ipmi:
char: ipmi: use DEVICE_ATTR helper macro
ipmi: rate limit ipmi smi_event failure message
The caller of this function (parisc_driver_remove() in
arch/parisc/kernel/drivers.c) ignores the return value, so better don't
return any value at all to not wake wrong expectations in driver authors.
The only function that could return a non-zero value before was
ipmi_parisc_remove() which returns the return value of
ipmi_si_remove_by_dev(). Make this function return void, too, as for all
other callers the value is ignored, too.
Also fold in a small checkpatch fix for:
WARNING: Unnecessary space before function pointer arguments
+ void (*remove) (struct parisc_device *dev);
Acked-by: Dmitry Torokhov <[email protected]> (for drivers/input)
Signed-off-by: Uwe Kleine-König <[email protected]>
Acked-by: Sudip Mukherjee <[email protected]>
Acked-by: Jiri Slaby <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
Instead of open coding DEVICE_ATTR, use the helper macro
DEVICE_ATTR_RO to replace DEVICE_ATTR with 0444 octal
permissions.
This was detected as a part of checkpatch evaluation
investigating all reports of DEVICE_ATTR_RO warning
type.
Signed-off-by: Dwaipayan Ray <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Sometimes we can't get a valid si_sm_data, and we print an error
message accordingly. But the ipmi module seem to like retrying a lot,
in which case we flood the kernel log with a lot of messages, eg:
[46318019.164726] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318020.109700] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318021.158677] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318022.212598] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318023.258564] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318024.210455] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318025.260473] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318026.308445] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318027.356389] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318028.298288] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
[46318029.363302] ipmi_si IPI0001:00: Could not set the global enables: 0xc1.
Signed-off-by: Wen Yang <[email protected]>
Cc: Baoyou Xie <[email protected]>
Cc: Corey Minyard <[email protected]>
Cc: [email protected]
Cc: [email protected]
Message-Id: <[email protected]>
[Added a missing comma]
Signed-off-by: Corey Minyard <[email protected]>
Merge more updates from Andrew Morton:
"190 patches.
Subsystems affected by this patch series: mm (hugetlb, userfaultfd,
vmscan, kconfig, proc, z3fold, zbud, ras, mempolicy, memblock,
migration, thp, nommu, kconfig, madvise, memory-hotplug, zswap,
zsmalloc, zram, cleanups, kfence, and hmm), procfs, sysctl, misc,
core-kernel, lib, lz4, checkpatch, init, kprobes, nilfs2, hfs,
signals, exec, kcov, selftests, compress/decompress, and ipc"
* emailed patches from Andrew Morton <[email protected]>: (190 commits)
ipc/util.c: use binary search for max_idx
ipc/sem.c: use READ_ONCE()/WRITE_ONCE() for use_global_lock
ipc: use kmalloc for msg_queue and shmid_kernel
ipc sem: use kvmalloc for sem_undo allocation
lib/decompressors: remove set but not used variabled 'level'
selftests/vm/pkeys: exercise x86 XSAVE init state
selftests/vm/pkeys: refill shadow register after implicit kernel write
selftests/vm/pkeys: handle negative sys_pkey_alloc() return code
selftests/vm/pkeys: fix alloc_random_pkey() to make it really, really random
kcov: add __no_sanitize_coverage to fix noinstr for all architectures
exec: remove checks in __register_bimfmt()
x86: signal: don't do sas_ss_reset() until we are certain that sigframe won't be abandoned
hfsplus: report create_date to kstat.btime
hfsplus: remove unnecessary oom message
nilfs2: remove redundant continue statement in a while-loop
kprobes: remove duplicated strong free_insn_page in x86 and s390
init: print out unknown kernel parameters
checkpatch: do not complain about positive return values starting with EPOLL
checkpatch: improve the indented label test
checkpatch: scripts/spdxcheck.py now requires python3
...
The comparisons of the unsigned int hw_type to less than zero always
false because it is unsigned. Fix this by using an int for the
assignment and less than zero check.
Addresses-Coverity: ("Unsigned compared against 0")
Fixes: 9d2df9a0ad80 ("ipmi: kcs_bmc_aspeed: Implement KCS SerIRQ configuration")
Signed-off-by: Colin Ian King <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Some Aspeed KCS devices can derive the status register address from the
address of the data register. As such, the address of the status
register can be implicit in the configuration if desired. On the other
hand, sometimes address schemes might be requested that are incompatible
with the default addressing scheme. Allow these requests where possible
if the devicetree specifies the status register address.
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Chia-Wei Wang <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Input Buffer Full Interrupt Enable (IBFIE) is typoed as IBFIF for some
registers in the datasheet. Fix the driver to use the sensible acronym.
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Apply the SerIRQ ID and level/sense behaviours from the devicetree if
provided.
Signed-off-by: Andrew Jeffery <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
kcs_bmc_serio acts as a bridge between the KCS drivers in the IPMI
subsystem and the existing userspace interfaces available through the
serio subsystem. This is useful when userspace would like to make use of
the BMC KCS devices for purposes that aren't IPMI.
Signed-off-by: Andrew Jeffery <[email protected]>
Message-Id: <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Add a mechanism for controlling whether the client associated with a
KCS device will receive Input Buffer Full (IBF) and Output Buffer Empty
(OBE) events. This enables an abstract implementation of poll() for KCS
devices.
A wart in the implementation is that the ASPEED KCS devices don't
support an OBE interrupt for the BMC. Instead we pretend it has one by
polling the status register waiting for the Output Buffer Full (OBF) bit
to clear, and generating an event when OBE is observed.
Cc: CS20 KWLiu <[email protected]>
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Now that we have untangled the data-structures, split the userspace
interface out into its own module. Userspace interfaces and drivers are
registered to the KCS BMC core to support arbitrary binding of either.
Signed-off-by: Andrew Jeffery <[email protected]>
Message-Id: <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Move all client-private data out of `struct kcs_bmc` into the KCS client
implementation.
With this change the KCS BMC core code now only concerns itself with
abstract `struct kcs_bmc` and `struct kcs_bmc_client` types, achieving
expected separation of concerns. Further, the change clears the path for
implementation of alternative userspace interfaces.
The chardev data-structures are rearranged in the same manner applied to
the KCS device driver data-structures in an earlier patch - `struct
kcs_bmc_client` is embedded in the client's private data and we exploit
container_of() to translate as required.
Finally, now that it is free of client data, `struct kcs_bmc` is renamed
to `struct kcs_bmc_device` to contrast `struct kcs_bmc_client`.
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Strengthen the distinction between code that abstracts the
implementation of the KCS behaviours (device drivers) and code that
exploits KCS behaviours (clients). Neither needs to know about the APIs
required by the other, so provide separate headers.
Signed-off-by: Andrew Jeffery <[email protected]>
Message-Id: <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Make the KCS device drivers responsible for allocating their own memory.
Until now the private data for the device driver was allocated internal
to the private data for the chardev interface. This coupling required
the slightly awkward API of passing through the struct size for the
driver private data to the chardev constructor, and then retrieving a
pointer to the driver private data from the allocated chardev memory.
In addition to being awkward, the arrangement prevents the
implementation of alternative userspace interfaces as the device driver
private data is not independent.
Peel a layer off the onion and turn the data-structures inside out by
exploiting container_of() and embedding `struct kcs_device` in the
driver private data.
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Take steps towards defining a coherent API to separate the KCS device
drivers from the userspace interface. Decreasing the coupling will
improve the separation of concerns and enable the introduction of
alternative userspace interfaces.
For now, simply split the chardev logic out to a separate file. The code
continues to build into the same module.
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Enable more efficient implementation of read-modify-write sequences.
Both device drivers for the KCS BMC stack use regmaps. The new callback
allows us to exploit regmap_update_bits().
Signed-off-by: Andrew Jeffery <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Unpack and remove the aspeed_kcs_probe_of_v[12]() functions to aid
rearranging how the private device-driver memory is allocated.
Signed-off-by: Andrew Jeffery <[email protected]>
Message-Id: <[email protected]>
Reviewed-by: Zev Weiss <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
When an IPMI watchdog timer is being stopped in ipmi_close() or
ipmi_ioctl(WDIOS_DISABLECARD), the current watchdog action is updated to
WDOG_TIMEOUT_NONE and _ipmi_set_timeout(IPMI_SET_TIMEOUT_NO_HB) is called
to install this action. The latter function ends up invoking
__ipmi_set_timeout() which makes the actual 'Set Watchdog Timer' IPMI
request.
For IPMI 1.0, this operation results in fully stopping the watchdog timer.
For IPMI >= 1.5, function __ipmi_set_timeout() always specifies the "don't
stop" flag in the prepared 'Set Watchdog Timer' IPMI request. This causes
that the watchdog timer has its action correctly updated to 'none' but the
timer continues to run. A problem is that IPMI firmware can then still log
an expiration event when the configured timeout is reached, which is
unexpected because the watchdog timer was requested to be stopped.
The patch fixes this problem by not setting the "don't stop" flag in
__ipmi_set_timeout() when the current action is WDOG_TIMEOUT_NONE which
results in stopping the watchdog timer. This makes the behaviour for
IPMI >= 1.5 consistent with IPMI 1.0. It also matches the logic in
__ipmi_heartbeat() which does not allow to reset the watchdog if the
current action is WDOG_TIMEOUT_NONE as that would start the timer.
Signed-off-by: Petr Pavlu <[email protected]>
Message-Id: <[email protected]>
Cc: [email protected]
Signed-off-by: Corey Minyard <[email protected]>
Pull IPMI updates from Corey Minyard:
"A bunch of little cleanups
Nothing major, no functional changes"
* tag 'for-linus-5.13-1' of git://github.com/cminyard/linux-ipmi:
ipmi_si: Join string literals back
ipmi_si: Drop redundant check before calling put_device()
ipmi_si: Use strstrip() to remove surrounding spaces
ipmi_si: Get rid of ->addr_source_cleanup()
ipmi_si: Reuse si_to_str[] array in ipmi_hardcode_init_one()
ipmi_si: Introduce ipmi_panic_event_str[] array
ipmi_si: Use proper ACPI macros to check error code for failures
ipmi_si: Utilize temporary variable to hold device pointer
ipmi_si: Remove bogus err_free label
ipmi_si: Switch to use platform_get_mem_or_io()
ipmi: Handle device properties with software node API
ipmi:ssif: make ssif_i2c_send() void
ipmi: Refine retry conditions for getting device id
For easy grepping on debug purposes join string literals back in
the messages.
No functional change.
Signed-off-by: Andy Shevchenko <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
The ->addr_source_cleanup() callback is solely used by PCI driver
and only for one purpose, i.e. to disable device. Get rid of
->addr_source_cleanup() by switching to PCI managed API.
Signed-off-by: Andy Shevchenko <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Instead of making the comparison one by one, reuse si_to_str[] array
in ipmi_hardcode_init_one() in conjunction with match_string() API.
Signed-off-by: Andy Shevchenko <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Instead of repeating twice the constant literals, introduce
ipmi_panic_event_str[] array. It allows to simplify the code
with help of match_string() API.
Signed-off-by: Andy Shevchenko <[email protected]>
Message-Id: <[email protected]>
Signed-off-by: Corey Minyard <[email protected]>
Corey Minyard