This causes a link failure on ARM in certain configurations,
when we reference each atomic operation from .alt.smp.init in
order to patch out atomics on non-SMP systems:
`.exit.text' referenced in section `.alt.smp.init' of drivers/char/ipmi/ipmi_msghandler.o: defined in discarded section `.exit.text' of drivers/char/ipmi/ipmi_msghandler.o
In this case, we can trivially replace the atomic_inc() with
an atomic_set() that has the same effect and does not require
a fixup.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Message-Id: <20190415155509.3565087-1-arnd@arndb.de>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The "init_name" variable isn't used any more after commit 90b2d4f15f
("ipmi_si: Remove hacks for adding a dummy platform devices").
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Message-Id: <20190322065426.GB12551@kadam>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Fix sparse warning:
drivers/char/ipmi/ipmi_msghandler.c:635:20: warning:
symbol 'ipmi_interfaces_srcu' was not declared. Should it be static?
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Message-Id: <20190320133505.21984-1-yuehaibing@huawei.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The intended behavior of function ipmi_hardcode_init_one() is to default
to kcs interface when no type argument is presented when initializing
ipmi with hard coded addresses.
However, the array of char pointers allocated on the stack by function
ipmi_hardcode_init() was not inited to zeroes, so it contained stack
debris.
Consequently, passing the cruft stored in this array to function
ipmi_hardcode_init_one() caused a crash when it was unable to detect
that the char * being passed was nonsense and tried to access the
address specified by the bogus pointer.
The fix is simply to initialize the si_type array to zeroes, so if
there were no type argument given to at the command line, function
ipmi_hardcode_init_one() could properly default to the kcs interface.
Signed-off-by: Tony Camuso <tcamuso@redhat.com>
Message-Id: <1554837603-40299-1-git-send-email-tcamuso@redhat.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
An extra memset was put into a place that cleared the interface
type.
Reported-by: Tony Camuso <tcamuso@redhat.com>
Fixes: 3cd83bac48 ("ipmi: Consolidate the adding of platform devices")
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The "ival" variable needs to signed so that we don't read before the
start of the str[] array. This would only happen the user passed in a
module parameter that was just comprised of space characters.
Fixes: e80444ae4fc3 ("ipmi_si: Switch hotmod to use a platform device")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Message-Id: <20190222195530.GA306@kadam>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
When a hotmod-added device is removed or when the module is removed,
remove the platform devices that was created for it.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Instead of keeping track of each one, just scan the platform bus
for hardcode devices and remove them.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
It was being done in two different places now that hard-coded devices
use platform devices, and it's about to be three with hotmod switching
to platform devices. So put the code in one place.
This required some rework on some interfaces to make the type space
clean.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
When excuting a command like:
modprobe ipmi_si ports=0xffc0e3 type=bt
The system would get an oops.
The trouble here is that ipmi_si_hardcode_find_bmc() is called before
ipmi_si_platform_init(), but initialization of the hard-coded device
creates an IPMI platform device, which won't be initialized yet.
The real trouble is that hard-coded devices aren't created with
any device, and the fixup is done later. So do it right, create the
hard-coded devices as normal platform devices.
This required adding some new resource types to the IPMI platform
code for passing information required by the hard-coded device
and adding some code to remove the hard-coded platform devices
on module removal.
To enforce the "hard-coded devices passed by the user take priority
over firmware devices" rule, some special code was added to check
and see if a hard-coded device already exists.
Reported-by: Yang Yingliang <yangyingliang@huawei.com>
Cc: stable@vger.kernel.org # v4.15+
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Tested-by: Yang Yingliang <yangyingliang@huawei.com>
Use guid_copy() instead of memcpy() to hide guid_t implementation details and
to show we expect guid_t in a raw buffer.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Instead of magic number use pre-defined constant for UUID binary and
string representations.
While here, drop the implementation details of guid_t type.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
[Also converted a "17" in the error string to UUID_SIZE + 1]
Signed-off-by: Corey Minyard <cminyard@mvista.com>
devm_kasprintf() may return NULL if internal allocation failed so this
assignment is not safe. Moved the error exit path and added the !NULL
which then allows the devres manager to take care of cleanup.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Fixes: cd2315d471 ("ipmi: kcs_bmc: don't change device name")
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Haiyue Wang <haiyue.wang@linux.intel.com>
The user can ask the message to be returned even if it didn't supply
enough memory for the data, and it will return an error but still
fills in as much data as possible. However, the return value
wasn't being set to an error, it was being overwritten. Create a
second return value for that case.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Trivial fix to clean up an indentation issue, remove space
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The code to tell the lower layer to enable or disable watching for
certain things was lazy in disabling, it waited until a timer tick
to see if a disable was necessary. Not a really big deal, but it
could be improved.
Modify the code to enable and disable watching immediately and don't
do it from the background timer any more.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Tested-by: Kamlakant Patel <kamlakant.patel@cavium.com>
The IPMI driver has a mechanism to tell the lower layers it needs
to watch for messages, commands, and watchdogs (so it doesn't
needlessly poll). However, it needed some extensions, it needed
a way to tell what is being waited for so it could set the timeout
appropriately.
The update to the lower layer was also being done once a second
at best because it was done in the main timeout handler. However,
if a command is sent and a response message is coming back,
it needed to be started immediately. So modify the code to
update immediately if it needs to be enabled. Disable is still
lazy.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Tested-by: Kamlakant Patel <kamlakant.patel@cavium.com>
Commit 89986496de ("ipmi: Turn off all activity on an idle ipmi
interface") modified the IPMI code to only request events when the
driver had somethine waiting for events. The SSIF code, however,
was using the event fetch request to also fetch the flags.
Add a timer and the proper handling for the upper layer telling
whether flags fetches are required.
Reported-by: Kamlakant Patel <Kamlakant.Patel@cavium.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Tested-by: Kamlakant Patel <kamlakant.patel@cavium.com>
The IPMI driver was recently modified to use SRCU, but it turns out
this uses a chunk of percpu memory, even if IPMI is never used.
So modify thing to on initialize on the first use. There was already
code to sort of handle this for handling init races, so piggy back
on top of that, and simplify it in the process.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reported-by: Tejun Heo <tj@kernel.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org # 4.18
Some IPMI modules (e.g. ibmpex_msg_handler()) will have ipmi_usr_hdlr
handlers that call ipmi_free_recv_msg() directly. This will essentially
kfree(msg), leading to use-after-free.
This does not happen in the ipmi_devintf module, which will queue the
message and run ipmi_free_recv_msg() later.
BUG: KASAN: use-after-free in deliver_response+0x12f/0x1b0
Read of size 8 at addr ffff888a7bf20018 by task ksoftirqd/3/27
CPU: 3 PID: 27 Comm: ksoftirqd/3 Tainted: G O 4.19.11-amd64-ani99-debug #12.0.1.601133+pv
Hardware name: AppNeta r1000/X11SPW-TF, BIOS 2.1a-AP 09/17/2018
Call Trace:
dump_stack+0x92/0xeb
print_address_description+0x73/0x290
kasan_report+0x258/0x380
deliver_response+0x12f/0x1b0
? ipmi_free_recv_msg+0x50/0x50
deliver_local_response+0xe/0x50
handle_one_recv_msg+0x37a/0x21d0
handle_new_recv_msgs+0x1ce/0x440
...
Allocated by task 9885:
kasan_kmalloc+0xa0/0xd0
kmem_cache_alloc_trace+0x116/0x290
ipmi_alloc_recv_msg+0x28/0x70
i_ipmi_request+0xb4a/0x1640
ipmi_request_settime+0x1b8/0x1e0
...
Freed by task 27:
__kasan_slab_free+0x12e/0x180
kfree+0xe9/0x280
deliver_response+0x122/0x1b0
deliver_local_response+0xe/0x50
handle_one_recv_msg+0x37a/0x21d0
handle_new_recv_msgs+0x1ce/0x440
tasklet_action_common.isra.19+0xc4/0x250
__do_softirq+0x11f/0x51f
Fixes: e86ee2d44b ("ipmi: Rework locking and shutdown for hot remove")
Cc: stable@vger.kernel.org # 4.18
Signed-off-by: Fred Klassen <fklassen@appneta.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
channel and addr->channel are indirectly controlled by user-space,
hence leading to a potential exploitation of the Spectre variant 1
vulnerability.
These issues were detected with the help of Smatch:
drivers/char/ipmi/ipmi_msghandler.c:1381 ipmi_set_my_address() warn: potential spectre issue 'user->intf->addrinfo' [w] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1401 ipmi_get_my_address() warn: potential spectre issue 'user->intf->addrinfo' [r] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1421 ipmi_set_my_LUN() warn: potential spectre issue 'user->intf->addrinfo' [w] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1441 ipmi_get_my_LUN() warn: potential spectre issue 'user->intf->addrinfo' [r] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:2260 check_addr() warn: potential spectre issue 'intf->addrinfo' [r] (local cap)
Fix this by sanitizing channel and addr->channel before using them to
index user->intf->addrinfo and intf->addrinfo, correspondingly.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
[1] https://lore.kernel.org/lkml/20180423164740.GY17484@dhcp22.suse.cz/
Cc: stable@vger.kernel.org
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The block number was not being compared right, it was off by one
when checking the response.
Some statistics wouldn't be incremented properly in some cases.
Check to see if that middle-part messages always have 31 bytes of
data.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: stable@vger.kernel.org # 4.4
Now that synchronize_rcu() waits for preempt-disable regions of code
as well as RCU read-side critical sections, synchronize_sched() can be
replaced by synchronize_rcu(). This commit therefore makes this change.
Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: <openipmi-developer@lists.sourceforge.net>
Acked-by: Corey Minyard <cminyard@mvista.com>
The spec was fairly confusing about how multi-part transmit messages
worked, so the original implementation only added support for two
part messages. But after talking about it with others and finding
something I missed, I think it makes more sense.
The spec mentions smbus command 8 in a table at the end of the
section on SSIF support as the end transaction. If that works,
then all is good and as it should be. However, some implementations
seem to use a middle transaction <32 bytes tomark the end because of the
confusion in the spec, even though that is an SMBus violation if
the number of bytes is zero.
So this change adds some tests, if command=8 works, it uses that,
otherwise if an empty end transaction works, it uses a middle
transaction <32 bytes to mark the end. If neither works, then
it limits the size to 63 bytes as it is now.
Cc: Harri Hakkarainen <harri@cavium.com>
Cc: Bazhenov, Dmitry <dmitry.bazhenov@auriga.com>
Cc: Mach, Dat <Dat.Mach@cavium.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
The IPMI DMI code was adding platform overrides, which is not
really an ideal solution. Switch to using the id_table in
the drivers to identify the devices.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Fixes gcc '-Wunused-but-set-variable' warning:
drivers/char/ipmi/ipmi_msghandler.c: In function 'ipmi_set_my_LUN':
drivers/char/ipmi/ipmi_msghandler.c:1335:13: warning:
variable 'rv' set but not used [-Wunused-but-set-variable]
int index, rv = 0;
'rv' should be the correct return value.
Fixes: 048f7c3e35 ("ipmi: Properly release srcu locks on error conditions")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cleanups, do the replacement and change the levels to the proper
ones for the function they are serving, as many were wrong.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Looking at logs from systems all over the place, it looks like tons
of broken systems exist that set the base address to zero. I can
only guess that is some sort of non-standard idea to mark the
interface as not being present. It can't be zero, anyway, so just
complain and ignore it.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
getnstimeofday64() is deprecated because of the inconsistent naming,
it is only a wrapper around ktime_get_real_ts64() now, which could be
used as a direct replacement.
However, it is generally better to use CLOCK_MONOTONIC timestamps
where possible, to avoid glitches with a concurrent settimeofday()
or leap second.
The uses in ipmi are either for debugging prints or for comparing against
a prior timestamp, so using a monotonic ktime_get_ts64() is probably
best here.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Shifting unsigned char b by an int type can lead to sign-extension
overflow. For example, if b is 0xff and the shift is 24, then top
bit is sign-extended so the final value passed to writeq has all
the upper 32 bits set. Fix this by casting b to a 64 bit unsigned
before the shift.
Detected by CoverityScan, CID#1465246 ("Unintended sign extension")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
I noticed that 4.17.0 logs the follwing during ipmi_si setup:
ipmi_si 0000:01:04.6: probing via PCI
(NULL device *): Could not setup I/O space
ipmi_si 0000:01:04.6: [mem 0xf5ef0000-0xf5ef00ff] regsize 1 spacing 1 irq 21
Fix the "NULL device *) by moving io.dev assignment before its potential
use by ipmi_pci_probe_regspacing().
Result:
ipmi_si 0000:01:04.6: probing via PCI
ipmi_si 0000:01:04.6: Could not setup I/O space
ipmi_si 0000:01:04.6: [mem 0xf5ef0000-0xf5ef00ff] regsize 1 spacing 1 irq 21
Signed-off-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Use the more common logging style.
Miscellanea:
o Convert old style continuation printks without KERN_CONT to pr_cont
o Coalesce formats
o Realign arguments
o Remove unnecessary casts
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Add and use #define pr_fmt/dev_fmt, and remove #define PFX
This also prefixes some messages that were not previously prefixed.
Miscellanea:
o Convert printk(KERN_<level> to pr_<level>(
o Use %s, __func__ where appropriate
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Standardize the prefixing of output messages using the pr_fmt and dev_fmt
mechanisms instead of a separate #define PFX
Miscellanea:
o Because this message prefix is very long, use a non-standard define
of #define pr_fmt(fmt) "%s" fmt, "IPMI message handler: "
which removes ~170 bytes of object code in an x86-64 defconfig with ipmi
(with even more object code reduction on 32 bit compilations)
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
