Kernels built with PREEMPT_RCU=y and RCU_STRICT_GRACE_PERIOD=y trigger
irq-work from rcu_read_unlock(), and the resulting irq-work handler
invokes rcu_preempt_deferred_qs_handle(). The point of this triggering
is to force grace periods to end quickly in order to give tools like KASAN
a better chance of detecting RCU usage bugs such as leaking RCU-protected
pointers out of an RCU read-side critical section.
However, this irq-work triggering is unconditional. This works, but
there is no point in doing this irq-work unless the current grace period
is waiting on the running CPU or task, which is not the common case.
After all, in the common case there are many rcu_read_unlock() calls
per CPU per grace period.
This commit therefore triggers the irq-work only when the current grace
period is waiting on the running CPU or task.
This change was tested as follows on a four-CPU system:
This procedure produces results in this per-CPU set of files:
/sys/kernel/debug/tracing/trace_stat/function*
Sample output from one of these files is as follows:
Function Hit Time Avg s^2
-------- --- ---- --- ---
rcu_preempt_deferred_qs_handle 838746 182650.3 us 0.217 us 0.004 us
The baseline sum of the "Hit" values (the number of calls to this
function) was 3,319,015. With this commit, that sum was 1,140,359,
for a 2.9x reduction. The worst-case variance across the CPUs was less
than 25%, so this large effect size is statistically significant.
This patch fixes a race between queue_work() in
_dlm_lowcomms_commit_msg() and srcu_read_unlock(). The queue_work() can
take the final reference of a dlm_msg and so msg->idx can contain
garbage which is signaled by the following warning:
I reproduced this warning with dlm_locktorture test which is currently
not upstream. However this patch fix the issue by make a additional
refcount between dlm_lowcomms_new_msg() and dlm_lowcomms_commit_msg().
In case of the race the kref_put() in dlm_lowcomms_commit_msg() will be
the final put.
Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
An instance of Client does not implicitly close /dev/tpm* handle, once it
gets destroyed. Close the file handle in the class destructor
Client.__del__().
Fixes: 6ea3dfe1e0732 ("selftests: add TPM 2.0 tests") Cc: Shuah Khan <shuah@kernel.org> Cc: linux-kselftest@vger.kernel.org Cc: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Stefan Berger <stefanb@linux.ibm.com> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
f2fs_inode_info.cp_task was introduced for FS_CP_DATA_IO accounting
since commit b0af6d491a6b ("f2fs: add app/fs io stat").
However, cp_task usage coverage has been increased due to below
commits:
commit 040d2bb318d1 ("f2fs: fix to avoid deadloop if data_flush is on")
commit 186857c5a14a ("f2fs: fix potential recursive call when enabling data_flush")
So that, if data_flush mountoption is on, when data flush was
triggered from background, the IO from data flush will be accounted
as checkpoint IO type incorrectly.
In order to fix this issue, this patch splits cp_task into two:
a) cp_task: used for IO accounting
b) wb_task: used to avoid deadlock
Fixes: 040d2bb318d1 ("f2fs: fix to avoid deadloop if data_flush is on") Fixes: 186857c5a14a ("f2fs: fix potential recursive call when enabling data_flush") Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The following scenarios exist.
process A: process B:
->f2fs_drop_extent_tree ->f2fs_update_extent_cache_range
->f2fs_update_extent_tree_range
->write_lock
->set_inode_flag
->is_inode_flag_set
->__free_extent_tree // Shouldn't
// have been
// cleaned up
// here
->write_lock
In this case, the "FI_NO_EXTENT" flag is set between
f2fs_update_extent_tree_range and is_inode_flag_set
by other process. it leads to clearing the whole exten
tree which should not have happened. And we fix it by
move the setting it to the range of write_lock.
If an error is detected as a result of user-space process accessing a
corrupt memory location, the CPU may take an abort. Then the platform
firmware reports kernel via NMI like notifications, e.g. NOTIFY_SEA,
NOTIFY_SOFTWARE_DELEGATED, etc.
For NMI like notifications, commit 7f17b4a121d0 ("ACPI: APEI: Kick the
memory_failure() queue for synchronous errors") keep track of whether
memory_failure() work was queued, and make task_work pending to flush out
the queue so that the work is processed before return to user-space.
The code use init_mm to check whether the error occurs in user space:
if (current->mm != &init_mm)
The condition is always true, becase _nobody_ ever has "init_mm" as a real
VM any more.
In addition to abort, errors can also be signaled as asynchronous
exceptions, such as interrupt and SError. In such case, the interrupted
current process could be any kind of thread. When a kernel thread is
interrupted, the work ghes_kick_task_work deferred to task_work will never
be processed because entry_handler returns to call ret_to_kernel() instead
of ret_to_user(). Consequently, the estatus_node alloced from
ghes_estatus_pool in ghes_in_nmi_queue_one_entry() will not be freed.
After around 200 allocations in our platform, the ghes_estatus_pool will
run of memory and ghes_in_nmi_queue_one_entry() returns ENOMEM. As a
result, the event failed to be processed.
sdei: event 805 on CPU 113 failed with error: -2
Finally, a lot of unhandled events may cause platform firmware to exceed
some threshold and reboot.
The condition should generally just do
if (current->mm)
as described in active_mm.rst documentation.
Then if an asynchronous error is detected when a kernel thread is running,
(e.g. when detected by a background scrubber), do not add task_work to it
as the original patch intends to do.
Fixes: 7f17b4a121d0 ("ACPI: APEI: Kick the memory_failure() queue for synchronous errors") Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com> Reviewed-by: Tony Luck <tony.luck@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Counterintuitively, mod_timer(..., jiffies + 1) will cause the timer to
fire not in the next jiffy, but in two jiffies. The way to cause
the timer to fire in the next jiffy is with mod_timer(..., jiffies).
Doing so then lets us bump the upper bound back up again.
Fixes: 50ee7529ec45 ("random: try to actively add entropy rather than passively wait for it") Fixes: 829d680e82a9 ("random: cap jitter samples per bit to factor of HZ") Cc: Dominik Brodowski <linux@dominikbrodowski.net> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Sultan Alsawaf <sultan@kerneltoast.com> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The "code_length" value comes from the firmware file. If your firmware
is untrusted realistically there is probably very little you can do to
protect yourself. Still we try to limit the damage as much as possible.
Also Smatch marks any data read from the filesystem as untrusted and
prints warnings if it not capped correctly.
The "ntohl(ucode->code_length) * 2" multiplication can have an
integer overflow.
Fixes: 9e2c7d99941d ("crypto: cavium - Add Support for Octeon-tx CPT Engine") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
The "code_length" value comes from the firmware file. If your firmware
is untrusted realistically there is probably very little you can do to
protect yourself. Still we try to limit the damage as much as possible.
Also Smatch marks any data read from the filesystem as untrusted and
prints warnings if it not capped correctly.
The "code_length * 2" can overflow. The round_up(ucode_size, 16) +
sizeof() expression can overflow too. Prevent these overflows.
Fixes: d9110b0b01ff ("crypto: marvell - add support for OCTEON TX CPT engine") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since commit 7b4537199a4a ("kbuild: link symbol CRCs at final link,
removing CONFIG_MODULE_REL_CRCS"), the module versioning on the
(non-upstreamed-yet) kvx Linux port is broken due to unexpected padding
for __crc_* symbols. The kvx GCC adds padding so u32 gets 8-byte
alignment instead of 4.
I do not know if this happens for upstream architectures in general,
but any compiler has the freedom to insert padding for faster access.
Use the inline assembler to directly specify the wanted data layout.
This is how we previously did before the breakage.
When receiving some signal, GNU Make automatically deletes the target if
it has already been changed by the interrupted recipe.
If the target is possibly incomplete due to interruption, it must be
deleted so that it will be remade from scratch on the next run of make.
Otherwise, the target would remain corrupted permanently because its
timestamp had already been updated.
Thanks to this behavior of Make, you can stop the build any time by
pressing Ctrl-C, and just run 'make' to resume it.
Kbuild also relies on this feature, but it is equivalently important
for any build systems that make decisions based on timestamps (if you
want to support Ctrl-C reliably).
However, this does not always work as claimed; Make immediately dies
with Ctrl-C if its stderr goes into a pipe.
$ make 2>&1 | cat # hit Ctrl-C
echo hello > foo
sleep 3
^C$ # 'foo' is often left-over
The reason is because SIGINT is sent to the entire process group.
In this example, SIGINT kills 'cat', and 'make' writes the message to
the closed pipe, then dies with SIGPIPE before cleaning the target.
A typical bad scenario (as reported by [1], [2]) is to save build log
by using the 'tee' command:
$ make 2>&1 | tee log
This can be problematic for any build systems based on Make, so I hope
it will be fixed in GNU Make. The maintainer of GNU Make stated this is
a long-standing issue and difficult to fix [3]. It has not been fixed
yet as of writing.
So, we cannot rely on Make cleaning the target. We can do it by
ourselves, in signal traps.
As far as I understand, Make takes care of SIGHUP, SIGINT, SIGQUIT, and
SITERM for the target removal. I added the traps for them, and also for
SIGPIPE just in case cmd_* rule prints something to stdout or stderr
(but I did not observe an actual case where SIGPIPE was triggered).
[Note 1]
The trap handler might be worth explaining.
rm -f $@; trap - $(sig); kill -s $(sig) $$
This lets the shell kill itself by the signal it caught, so the parent
process can tell the child has exited on the signal. Generally, this is
a proper manner for handling signals, in case the calling program (like
Bash) may monitor WIFSIGNALED() and WTERMSIG() for WCE although this may
not be a big deal here because GNU Make handles SIGHUP, SIGINT, SIGQUIT
in WUE and SIGTERM in IUE.
IUE - Immediate Unconditional Exit
WUE - Wait and Unconditional Exit
WCE - Wait and Cooperative Exit
For details, see "Proper handling of SIGINT/SIGQUIT" [4].
[Note 2]
Reverting 392885ee82d3 ("kbuild: let fixdep directly write to .*.cmd
files") would directly address [1], but it only saves if_changed_dep.
As reported in [2], all commands that use redirection can potentially
leave an empty (i.e. broken) target.
[Note 3]
Another (even safer) approach might be to always write to a temporary
file, and rename it to $@ at the end of the recipe.
<command> > $(tmp-target)
mv $(tmp-target) $@
It would require a lot of Makefile changes, and result in ugly code,
so I did not take it.
[Note 4]
A little more thoughts about a pattern rule with multiple targets (or
a grouped target).
%.x %.y: %.z
<recipe>
When interrupted, GNU Make deletes both %.x and %.y, while this solution
only deletes $@. Probably, this is not a big deal. The next run of make
will execute the rule again to create $@ along with the other files.
There is a recursive lock on the cpu_hotplug_lock.
In kernel/trace/trace_osnoise.c:<start/stop>_per_cpu_kthreads:
- start_per_cpu_kthreads calls cpus_read_lock() and if
start_kthreads returns a error it will call stop_per_cpu_kthreads.
- stop_per_cpu_kthreads then calls cpus_read_lock() again causing
deadlock.
Fix this by calling cpus_read_unlock() before calling
stop_per_cpu_kthreads. This behavior can also be seen in commit f46b16520a08 ("trace/hwlat: Implement the per-cpu mode").
This error was noticed during the LTP ftrace-stress-test:
WARNING: possible recursive locking detected
--------------------------------------------
sh/275006 is trying to acquire lock: ffffffffb02f5400 (cpu_hotplug_lock){++++}-{0:0}, at: stop_per_cpu_kthreads
but task is already holding lock: ffffffffb02f5400 (cpu_hotplug_lock){++++}-{0:0}, at: start_per_cpu_kthreads
other info that might help us debug this:
Possible unsafe locking scenario:
For now, this selftest module can only work in x86 because of the
kprobe cmd was fixed use of x86 registers.
This patch adapted to register names under arm and riscv, So that
this module can be worked on those platform.
IOMMU_IOVA is intended to be an optional library for users to select as
and when they desire. Since it can be a module now, this means that
built-in code which has chosen not to select it should not fail to link
if it happens to have selected as a module by someone else. Replace
IS_ENABLED() with IS_REACHABLE() to do the right thing.
CC: Thierry Reding <thierry.reding@gmail.com> Reported-by: John Garry <john.garry@huawei.com> Fixes: 15bbdec3931e ("iommu: Make the iova library a module") Signed-off-by: Robin Murphy <robin.murphy@arm.com> Reviewed-by: Thierry Reding <treding@nvidia.com> Link: https://lore.kernel.org/r/548c2f683ca379aface59639a8f0cccc3a1ac050.1663069227.git.robin.murphy@arm.com Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
If an error happens while getting the key or session in the
->calc_signature implementations, 0 (success) is returned. Fix it by
returning a proper error code.
Since it seems to be highly unlikely to happen wrap the rc check in
unlikely() too.
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com> Fixes: 32811d242ff6 ("cifs: Start using per session key for smb2/3 for signature generation") Signed-off-by: Enzo Matsumiya <ematsumiya@suse.de> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit a38b71b0833e ("clocksource/drivers/arm_arch_timer:
Move system register timer programming over to CVAL") moves the
programming of the timers from the countdown timer (TVAL) over
to the comparator (CVAL). This makes it necessary to read the
counter when programming next event. However, the workaround of
Cortex-A73 erratum 858921 does not set the corresponding
set_next_event_phys and set_next_event_virt.
Add the appropriate hooks to apply the erratum mitigation when
programming the next timer event.
Fixes: a38b71b0833e ("clocksource/drivers/arm_arch_timer: Move system register timer programming over to CVAL") Signed-off-by: Kunkun Jiang <jiangkunkun@huawei.com> Acked-by: Marc Zyngier <maz@kernel.org> Reviewed-by: Oliver Upton <oliver.upton@linux.dev> Link: https://lore.kernel.org/r/20220914061424.1260-1-jiangkunkun@huawei.com Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When CONFIG_DMA_API_DEBUG is selected, while running the crypto self
test on the QAT crypto algorithms, the function add_dma_entry() reports
a warning similar to the one below, saying that overlapping mappings
are not supported. This occurs in tests where the input and the output
scatter list point to the same buffers (i.e. two different scatter lists
which point to the same chunks of memory).
The logic that implements the mapping uses the flag DMA_BIDIRECTIONAL
for both the input and the output scatter lists which leads to
overlapped write mappings. These are not supported by the DMA layer.
Fix by specifying the correct DMA transfer directions when mapping
buffers. For in-place operations where the input scatter list
matches the output scatter list, buffers are mapped once with
DMA_BIDIRECTIONAL, otherwise input buffers are mapped using the flag
DMA_TO_DEVICE and output buffers are mapped with DMA_FROM_DEVICE.
Overlapping a read mapping with a write mapping is a valid case in
dma-coherent devices like QAT.
The function that frees and unmaps the buffers, qat_alg_free_bufl()
has been changed accordingly to the changes to the mapping function.
The use of swab() is causing failures in 64-bit arch, as it
translates to __swab64() instead of the intended __swab32().
It eventually causes wrong results in xcbcmac & cmac algo.
Fixes: 78cf1c8bfcb8 ("crypto: inside-secure - Move ipad/opad into safexcel_context") Signed-off-by: Peter Harliman Liem <pliem@maxlinear.com> Acked-by: Antoine Tenart <atenart@kernel.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
__dma_async_device_channel_unregister called while 2 clients hold a reference
WARNING: CPU: 15 PID: 1 at drivers/dma/dmaengine.c:1110 __dma_async_device_channel_unregister+0xb7/0xc0
Call dma_release_channel for occupied channles before dma_async_device_unregister.
Fixes: 54cce8ecb925 ("crypto: ccp - ccp_dmaengine_unregister release dma channels") Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Koba Ko <koba.ko@canonical.com> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
Changes from v1:
* removed the default implementation from set_pub_key: it is assumed that
an implementation must always have this callback defined as there are
no use case for an algorithm, which doesn't need a public key
Many akcipher implementations (like ECDSA) support only signature
verifications, so they don't have all callbacks defined.
Commit 78a0324f4a53 ("crypto: akcipher - default implementations for
request callbacks") introduced default callbacks for sign/verify
operations, which just return an error code.
However, these are not enough, because before calling sign the caller would
likely call set_priv_key first on the instantiated transform (as the
in-kernel testmgr does). This function does not have a default stub, so the
kernel crashes, when trying to set a private key on an akcipher, which
doesn't support signature generation.
I've noticed this, when trying to add a KAT vector for ECDSA signature to
the testmgr.
With this patch the testmgr returns an error in dmesg (as it should)
instead of crashing the kernel NULL ptr dereference.
1) The "len" variable needs to be checked before the very first write.
Otherwise if omap2_iommu_dump_ctx() with "bytes" less than 32 it is a
buffer overflow.
2) The snprintf() function returns the number of bytes that *would* have
been copied if there were enough space. But we want to know the
number of bytes which were *actually* copied so use scnprintf()
instead.
Previously, update_tasks_cpumask() is not supposed to be called with
top cpuset. With cpuset partition that takes CPUs away from the top
cpuset, adjusting the cpus_mask of the tasks in the top cpuset is
necessary. Percpu kthreads, however, are ignored.
Fixes: ee8dde0cd2ce ("cpuset: Add new v2 cpuset.sched.partition flag") Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Issue:
While servicing interrupt, if the IRQ happens to be because of a SEED_DONE
due to a previous boot stage, you end up completing the completion
prematurely, hence causing kernel to crash while booting.
Fix:
Moving IRQ handler registering after imx_rngc_irq_mask_clear()
Fixes: 1d5449445bd0 (hwrng: mx-rngc - add a driver for Freescale RNGC) Signed-off-by: Kshitiz Varshney <kshitiz.varshney@nxp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
cgroup_get_from_path() is not widely used function. Its callers presume
the path is resolved under cgroup namespace. (There is one caller
currently and resolving in init NS won't make harm (netfilter). However,
future users may be subject to different effects when resolving
globally.)
Since, there's currently no use for the global resolution, modify the
existing function to take cgroup NS into account.
Currently the OS continues the PSP initialization when there is a write
failure to the init_ex_file. Therefore, the userspace would be told that
SEV is properly INIT'd even though the psp data file is not updated.
This is problematic because later when asked for the SEV data, the OS
won't be able to provide it.
Fixes: 3d725965f836 ("crypto: ccp - Add SEV_INIT_EX support") Reported-by: Peter Gonda <pgonda@google.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Jacky Li <jackyli@google.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
The SMCCC_RET_TRNG_NO_ENTROPY switch arm is never used because the
NO_ENTROPY return value is negative and negative values are handled
above the switch by immediately returning.
Fix by handling errors using a default arm in the switch.
Fixes: 0888d04b47a1 ("hwrng: Add Arm SMCCC TRNG based driver") Signed-off-by: James Cowgill <james.cowgill@blaize.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
[17619.659757] BUG: KASAN: global-out-of-bounds in param_get_int+0x34/0x60
[17619.673193] Read of size 4 at addr fffff01332d7ed00 by task read_all/1507958
...
[17619.698934] The buggy address belongs to the variable:
[17619.708371] sgl_sge_nr+0x0/0xffffffffffffa300 [hisi_zip]
There is a mismatch in hisi_zip when get/set the variable sgl_sge_nr.
The type of sgl_sge_nr is u16, and get/set sgl_sge_nr by
param_get/set_int.
Replacing param_get/set_int to param_get/set_ushort can fix this bug.
Fixes: f081fda293ffb ("crypto: hisilicon - add sgl_sge_nr module param for zip") Signed-off-by: Ye Weihua <yeweihua4@huawei.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
Function of sahara_aes_crypt maybe could be called by function
of crypto_skcipher_encrypt during the rx softirq, so it is not
allowed to use mutex lock.
Generally the hypervisor decides to allocate a window on different
VAS instances. But if user space wishes to allocate on the current VAS
instance where the process is executing, the kernel has to pass
associativity domain IDs to allocate VAS window HCALL.
To determine the associativity domain IDs for the current CPU,
smp_processor_id() is passed to node associativity HCALL which may
return H_P2 (-55) error during DLPAR CPU event. This is because Linux
CPU numbers (smp_processor_id()) are not the same as the hypervisor's
view of CPU numbers.
Fix the issue by passing hard_smp_processor_id() with
VPHN_FLAG_VCPU flag (PAPR 14.11.6.1 H_HOME_NODE_ASSOCIATIVITY).
Fixes: b22f2d88e435 ("powerpc/pseries/vas: Integrate API with open/close windows") Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com> Signed-off-by: Haren Myneni <haren@linux.ibm.com>
[mpe: Update change log to mention Linux vs HV CPU numbers] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/55380253ea0c11341824cd4c0fc6bbcfc5752689.camel@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
If prev is based on ftrace, 'ppc_inst_read(prev->ainsn.insn)' will occur
with a null pointer reference. At this point prev->addr will not be a
prefixed instruction, so the check can be skipped.
Check if prev is ftrace-based kprobe before reading 'prev->ainsn.insn'
to fix this problem.
Fixes: b4657f7650ba ("powerpc/kprobes: Don't allow breakpoints on suffixes") Signed-off-by: Li Huafei <lihuafei1@huawei.com>
[mpe: Trim oops] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220923093253.177298-1-lihuafei1@huawei.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 2eb28006431c ("powerpc/e500v2: Add Power ISA properties to comply
with ePAPR 1.1") introduced new include file e500v2_power_isa.dtsi and
should have used it for all e500v2 platforms. But apparently it was used
also for e500v1 platforms mpc8540, mpc8541, mpc8555 and mpc8560.
e500v1 cores compared to e500v2 do not support double precision floating
point SPE instructions. Hence power-isa-sp.fd should not be set on e500v1
platforms, which is in e500v2_power_isa.dtsi include file.
Fix this issue by introducing a new e500v1_power_isa.dtsi include file and
use it in all e500v1 device tree files.
Fixes: 2eb28006431c ("powerpc/e500v2: Add Power ISA properties to comply with ePAPR 1.1") Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220902212103.22534-1-pali@kernel.org Signed-off-by: Sasha Levin <sashal@kernel.org>
If a synchronous interrupt (e.g., hash fault) is taken inside an
irqs-disabled region which has MSR[EE]=1, then an asynchronous interrupt
that is PACA_IRQ_MUST_HARD_MASK (e.g., PMI) is taken inside the
synchronous interrupt handler, then the synchronous interrupt will
return with MSR[EE]=1 and the asynchronous interrupt fires again.
If the asynchronous interrupt is a PMI and the original context does not
have PMIs disabled (only Linux IRQs), the asynchronous interrupt will
fire despite having the PMI marked soft pending. This can confuse the
perf code and cause warnings.
This patch changes the interrupt return so that irqs-disabled MSR[EE]=1
contexts will be returned to with MSR[EE]=0 if a PACA_IRQ_MUST_HARD_MASK
interrupt has become pending in the meantime.
Fixes: 4423eb5ae32e ("powerpc/64/interrupt: make normal synchronous interrupts enable MSR[EE] if possible") Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220926054305.2671436-4-npiggin@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
This prevents interrupts in early boot (e.g., program check) from
enabling MSR[EE], potentially causing endian mismatch or other
crashes when reporting early boot traps.
Fixes: 4423eb5ae32ec ("powerpc/64/interrupt: make normal synchronous interrupts enable MSR[EE] if possible") Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220926054305.2671436-3-npiggin@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Big-endian GENERIC_CPU supports 970, but builds with -mcpu=power5.
POWER5 is ISA v2.02 whereas 970 is v2.01 plus Altivec. 2.02 added
the popcntb instruction which a compiler might use.
As reported[1] by Arnd, the arch-specific fadvise64_64 and fallocate
compatibility handlers assume parameters are passed with 32-bit
big-endian ABI. This affects the assignment of odd-even parameter pairs
to the high or low words of a 64-bit syscall parameter.
Fix fadvise64_64 fallocate compat handlers to correctly swap upper/lower
32 bits conditioned on endianness.
A future patch will replace the arch-specific compat fallocate with an
asm-generic implementation. This patch is intended for ease of
back-port.
Currently, we are using CPU_PM_CPU_IDLE_ENTER_PARAM() for all SBI HSM
suspend types so retentive suspend types are also treated non-retentive
and kernel will do redundant additional work for these states.
The BIT[31] of SBI HSM suspend types allows us to differentiate between
retentive and non-retentive suspend types so we should use this BIT
to call appropriate CPU_PM_CPU_IDLE_ENTER_xyz() macro.
Fixes: 6abf32f1d9c5 ("cpuidle: Add RISC-V SBI CPU idle driver") Signed-off-by: Anup Patel <apatel@ventanamicro.com> Link: https://lore.kernel.org/r/20220718084553.2056169-1-apatel@ventanamicro.com/ Reviewed-by: Andrew Jones <ajones@ventanamicro.com> Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In fsl_setup_msi_irqs(), use of_node_put() to drop the reference
returned by of_parse_phandle().
Fixes: 895d603f945ba ("powerpc/fsl_msi: add support for the fsl, msi property in PCI nodes") Co-authored-by: Miaoqian Lin <linmq006@gmail.com> Signed-off-by: Liang He <windhl@126.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220704145233.278539-1-windhl@126.com Signed-off-by: Sasha Levin <sashal@kernel.org>
When building with a recent version of clang, there are a couple of
errors around the call to module_init():
arch/powerpc/math-emu/math_efp.c:927:1: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
module_init(spe_mathemu_init);
^
int
arch/powerpc/math-emu/math_efp.c:927:13: error: a parameter list without types is only allowed in a function definition
module_init(spe_mathemu_init);
^
2 errors generated.
module_init() is a macro, which is not getting expanded because module.h
is not included in this file. Add the include so that the macro can
expand properly, clearing up the build failure.
commit db7cfc380900 ("ipc: Free mq_sysctls if ipc namespace creation
failed")
Here's a similar memory leak to the one fixed by the patch above.
retire_mq_sysctls need to be called when init_mqueue_fs fails after
setup_mq_sysctls.
Fixes: dc55e35f9e81 ("ipc: Store mqueue sysctls in the ipc namespace") Signed-off-by: Hangyu Hua <hbh25y@gmail.com> Link: https://lkml.kernel.org/r/20220715062301.19311-1-hbh25y@gmail.com Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The mailbox offset is not only used for receiving messages, but it is
also used by messages sent to the system controller by Linux that have a
payload, such as the "digital signature service". It is also overloaded
by certain other services (reprogramming of the FPGA fabric, see Link:)
to have a meaning other than the offset the system controller should
read from.
When the driver was written, no such services of the latter type were
in use & those of the former used an offset of zero so this has gone
un-noticed.
The "data" region of the PolarFire SoC's system controller mailbox is
not one continuous register space - the system controller's QSPI sits
between the control and data registers. Split the "data" reg into two
parts: "data" & "control". Optionally get the "data" register address
from the 3rd reg property in the devicetree & fall back to using the
old base + MAILBOX_REG_OFFSET that the current code uses.
pm_runtime_get_sync() will increment pm usage counter.
Forgetting to putting operation will result in reference leak.
Add missing pm_runtime_put_sync in some error paths.
Fixes: 9ac33b0ce81f ("CLK: TI: Driver for DRA7 ATL (Audio Tracking Logic)") Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Link: https://lore.kernel.org/r/20220602030838.52057-1-linmq006@gmail.com Reviewed-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In ti_find_clock_provider(), of_find_node_by_name() will call
of_node_put() for the 'from' argument, possibly putting the node one too
many times. Let's maintain the of_node_get() from the previous search
and only put when we're exiting the function early. This should avoid a
misbalanced reference count on the node.
Fixes: 51f661ef9a10 ("clk: ti: Add ti_find_clock_provider() to use clock-output-names") Signed-off-by: Liang He <windhl@126.com> Link: https://lore.kernel.org/r/20220915031121.4003589-1-windhl@126.com
[sboyd@kernel.org: Rewrite commit text, maintain reference instead of
get again] Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The return value of bcm2835_clock_rate_from_divisor is always unsigned
and also all caller expect this. So fix the declaration accordingly.
Fixes: 41691b8862e2 ("clk: bcm2835: Add support for programming the audio domain clocks") Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com> Link: https://lore.kernel.org/r/20220904141037.38816-1-stefan.wahren@i2se.com Reviewed-by: Ivan T. Ivanov <iivanov@suse.de> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
It turns out the internal SATA reference clock signal will stay
unavailable for the SATA interface consumer until the buffer on it's way
is ungated. So aside with having the actual clock divider enabled we need
to ungate a buffer placed on the signal way to the SATA controller (most
likely some rudiment from the initial SoC release). Seeing the switch flag
is placed in the same register as the SATA-ref clock divider at a
non-standard ffset, let's implement it as a separate clock controller with
the set-rate propagation to the parental clock divider wrapper. As such
we'll be able to disable/enable and still change the original clock source
rate.
Baikal-T1 CCU reference manual says that both xGMAC reference and xGMAC
PTP clocks are generated by two different wrappers with the same constant
divider thus each producing a 156.25 MHz signal. But for some reason both
of these clock sources are gated by a single switch-flag in the CCU
registers space - CCU_SYS_XGMAC_BASE.BIT(0). In order to make the clocks
handled independently we need to define a shared parental gate so the base
clock signal would be switched off only if both of the child-clocks are
disabled.
Note the ID is intentionally set to -2 since we are going to add a one
more internal clock identifier in the next commit.
Most likely due to copy-paste mistake the divider has been set to 10 while
according to the SoC reference manual it's supposed to be 8 thus having
PTP clock frequency of 156.25 MHz.
We have discovered random glitches during the system boot up procedure.
The problem investigation led us to the weird outcomes: when none of the
Renesas 5P49V6901 ports are explicitly enabled by the kernel driver, the
glitches disappeared. It was a mystery since the SoC external clock
domains were fed with different 5P49V6901 outputs. The driver code didn't
seem like bogus either. We almost despaired to find out a root cause when
the solution has been found for a more modern revision of the chip. It
turned out the 5P49V6901 clock generator stopped its output for a short
period of time during the VC5_OUT_DIV_CONTROL register writing. The same
problem was found for the 5P49V6965 revision of the chip and was
successfully fixed in commit fc336ae622df ("clk: vc5: fix output disabling
when enabling a FOD") by enabling the "bypass_sync" flag hidden inside
"Unused Factory Reserved Register". Even though the 5P49V6901 registers
description and programming guide doesn't provide any intel regarding that
flag, setting it up anyway in the officially unused register completely
eliminated the denoted glitches. Thus let's activate the functionality
submitted in commit fc336ae622df ("clk: vc5: fix output disabling when
enabling a FOD") for the Renesas 5P49V6901 chip too in order to remove the
ports implicit inter-dependency.
Fixes: dbf6b16f5683 ("clk: vc5: Add support for IDT VersaClock 5P49V6901") Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net> Link: https://lore.kernel.org/r/20220929225402.9696-2-Sergey.Semin@baikalelectronics.ru Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Correct the way that duplicate PPID mappings are handled for PMIC
arbiter v5. The final APID mapped to a given PPID should be the
one which has write owner = APPS EE, if it exists, or if not
that, then the first APID mapped to the PPID, if it exists.
When the dr_mode is "host", after the host enter runtime suspend,
the mtu3 can't do it, because the mtu3's device wakeup function is
not enabled, instead it's enabled in gadget init function, to fix
the issue, init wakeup early in mtu3's probe()
Fixes: 6b587394c65c ("usb: mtu3: support suspend/resume for dual-role mode") Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Reported-by: Tianping Fang <tianping.fang@mediatek.com> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com> Link: https://lore.kernel.org/r/20220929064459.32522-1-chunfeng.yun@mediatek.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
User reports observing timer event report channel halted but no error
observed in CHANERR register. The driver finished self-test and released
channel resources. Debug shows that __cleanup() can call
mod_timer() after the timer has been deleted and thus resurrect the
timer. While harmless, it causes suprious error message to be emitted.
Use mod_timer_pending() call to prevent deleted timer from being
resurrected.
During the previous |struct clk| to |struct clk_hw| clk provider API
migration in commit 6f691a586296 ("clk: mediatek: Switch to clk_hw
provider APIs"), a few clk_unregister_*() calls were missed.
Migrate the remaining ones to the |struct clk_hw| provider API, i.e.
change clk_unregister_*() to clk_hw_unregister_*().
Fixes: 6f691a586296 ("clk: mediatek: Switch to clk_hw provider APIs") Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20220926102523.2367530-3-wenst@chromium.org Signed-off-by: Chen-Yu Tsai <wenst@chromium.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the cleanup paths for the various clk register APIs in the MediaTek
clk library were added, the one in the dividers type used the wrong type
of unregister function. This would result in incorrect dereferencing of
the clk pointer and freeing of invalid pointers.
Fix this by switching to the correct type of clk unregistration call.
Fixes: 3c3ba2ab0226 ("clk: mediatek: mtk: Implement error handling in register APIs") Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20220926102523.2367530-2-wenst@chromium.org Signed-off-by: Chen-Yu Tsai <wenst@chromium.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The MFG_BG3D is a gate to enable/disable clock output to the GPU,
but the actual output is decided by multiple muxes; in particular:
mfg_ck_fast_ref muxes between "slow" (top_mfg_core_tmp) and
"fast" (MFGPLL) clock, while top_mfg_core_tmp muxes between the
26MHz clock and various system PLLs.
The clock gate comes after all the muxes, so its parent is
mfg_ck_fast_reg, not top_mfg_core_tmp.
Reparent MFG_BG3D to the latter to match the hardware and add the
CLK_SET_RATE_PARENT flag to it: this way we ensure propagating
rate changes that are requested on MFG_BG3D along its entire clock
tree.
In da9062_i2c_probe() regmap_clear_bits() tries to access CONFIG_J
register. As CONFIG_J is not present in da9061_aa_writeable_ranges[] probe
of da9061 fails:
da9062 2-0058: Entering I2C mode!
da9062 2-0058: Failed to set Two-Wire Bus Mode.
da9062: probe of 2-0058 failed with error -5
Add CONFIG_J register to da9061_aa_writeable_ranges[].
Fixes: 5c6f0f456351 ("mfd: da9062: Support SMBus and I2C mode") Signed-off-by: Jens Hillenstedt <jens.hillenstedt@ise.de> Reviewed-by: Adam Ward <DLG-Adam.Ward.opensource@dm.renesas.com> Signed-off-by: Lee Jones <lee@kernel.org> Link: https://lore.kernel.org/r/20220915092004.168744-1-jens.hillenstedt@ise.de Signed-off-by: Sasha Levin <sashal@kernel.org>
The mx25_tsadc_remove() function assumes all non-zero returns are success
but the platform_get_irq() function returns negative on error and
positive non-zero values on success. It never returns zero, but if it
did then treat that as a success.
Fixes: 18f773937968 ("mfd: fsl-imx25: Clean up irq settings during removal") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Martin Kaiser <martin@kaiser.cx> Signed-off-by: Lee Jones <lee@kernel.org> Link: https://lore.kernel.org/r/YvTfkbVQWYKMKS/t@kili Signed-off-by: Sasha Levin <sashal@kernel.org>
In lp8788_irq_init(), if an error occurs after a successful
irq_domain_add_linear() call, it must be undone by a corresponding
irq_domain_remove() call.
irq_domain_remove() should also be called in lp8788_irq_exit() for the same
reason.
If devm_of_platform_populate() fails, some resources need to be
released.
Introduce a mx25_tsadc_unset_irq() function that undoes
mx25_tsadc_setup_irq() and call it both from the new error handling path
of the probe and in the remove function.
The commit in Fixes: has added a pwm_add_table() call in the probe() and
a pwm_remove_table() call in the remove(), but forget to update the error
handling path of the probe.
Add the missing pwm_remove_table() call.
Fixes: a3aa9a93df9f ("mfd: intel_soc_pmic_core: ADD PWM lookup table for CRC PMIC based PWM") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Lee Jones <lee@kernel.org> Link: https://lore.kernel.org/r/20220801114211.36267-1-andriy.shevchenko@linux.intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
If allocation fails, the ida_simple_get() will return error number.
So master->idx could be error number and be used in dev_set_name().
Therefore, it should be better to check it and return error if fails,
like the ida_simple_get() in __fsi_get_new_minor().
Fixes: 09aecfab93b8 ("drivers/fsi: Add fsi master definition") Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn> Reviewed-by: Eddie James <eajames@linux.ibm.com> Link: https://lore.kernel.org/r/20220111073411.614138-1-jiasheng@iscas.ac.cn Signed-off-by: Joel Stanley <joel@jms.id.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently in resize_finish() in rxe_queue.c there is a loop which copies
the entries in the original queue into a newly allocated queue. The
termination logic for this loop is incorrect. The call to
queue_next_index() updates cons but has no effect on whether the queue is
empty. So if the queue starts out empty nothing is copied but if it is not
then the loop will run forever. This patch changes the loop to compare the
value of cons to the original producer index.
Fixes: ae6e843fe08d0 ("RDMA/rxe: Add memory barriers to kernel queues") Link: https://lore.kernel.org/r/20220825221446.6512-1-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Reviewed-by: Li Zhijian <lizhijian@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
While fixing up the driver I noticed that my IPQ8074 board was hanging
after CPUFreq switched the frequency during boot, WDT would eventually
reset it.
So mark apcs_alias0_core_clk as critical since its the clock feeding the
CPU cluster and must never be disabled.
Fix a NULL pointer crash that occurs when we are freeing the socket at the
same time we access it via sysfs.
The problem is that:
1. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() take
the frwd_lock and do sock_hold() then drop the frwd_lock. sock_hold()
does a get on the "struct sock".
2. iscsi_sw_tcp_release_conn() does sockfd_put() which does the last put
on the "struct socket" and that does __sock_release() which sets the
sock->ops to NULL.
3. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() then
call kernel_getpeername() which accesses the NULL sock->ops.
Above we do a get on the "struct sock", but we needed a get on the "struct
socket". Originally, we just held the frwd_lock the entire time but in
commit bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while
calling getpeername()") we switched to refcount based because the network
layer changed and started taking a mutex in that path, so we could no
longer hold the frwd_lock.
Instead of trying to maintain multiple refcounts, this just has us use a
mutex for accessing the socket in the interface code paths.
Link: https://lore.kernel.org/r/20220907221700.10302-1-michael.christie@oracle.com Fixes: bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while calling getpeername()") Signed-off-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
We don't always want to run the recv path from the network softirq because
when we have to have multiple sessions sharing the same CPUs, some sessions
can eat up the NAPI softirq budget and affect other sessions or users.
Allow us to queue the recv handling to the iscsi workqueue so we can have
the scheduler/wq code try to balance the work and CPU use across all
sessions' worker threads.
Note: It wasn't the original intent of the change but a nice side effect is
that for some workloads/configs we get a nice performance boost. For a
simple read heavy test:
where the iscsi threads, fio jobs, and rps_cpus share CPUs we see a 32%
throughput boost. We also see increases for small I/O IOPs tests but it's
not as high.
Link: https://lore.kernel.org/r/20220616224557.115234-4-michael.christie@oracle.com Reviewed-by: Lee Duncan <lduncan@suse.com> Signed-off-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Stable-dep-of: 57569c37f0ad ("scsi: iscsi: iscsi_tcp: Fix null-ptr-deref while calling getpeername()") Signed-off-by: Sasha Levin <sashal@kernel.org>
The issue is that the per-device running_req read in
pm8001_dev_gone_notify() never goes to zero and we never make progress.
This is caused by missing accounting for running_req for when an internal
abort command completes.
In commit 2cbbf489778e ("scsi: pm8001: Use libsas internal abort support")
we started to send internal abort commands as a proper sas_task. In this
when we deliver a sas_task to HW the per-device running_req is incremented
in pm8001_queue_command(). However it is never decremented for internal
abort commnds, so decrement in pm8001_mpi_task_abort_resp().
Link: https://lore.kernel.org/r/1663854664-76165-1-git-send-email-john.garry@huawei.com Fixes: 2cbbf489778e ("scsi: pm8001: Use libsas internal abort support") Acked-by: Jack Wang <jinpu.wang@ionos.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When executing SMP task failed, the smp_execute_task_sg() calls del_timer()
to delete "slow_task->timer". However, if the timer handler
sas_task_internal_timedout() is running, the del_timer() in
smp_execute_task_sg() will not stop it and a UAF will happen. The process
is shown below:
Fix by calling del_timer_sync() in smp_execute_task_sg(), which makes sure
the timer handler have finished before the "task->slow_task" is
deallocated.
Link: https://lore.kernel.org/r/20220920144213.10536-1-duoming@zju.edu.cn Fixes: 2908d778ab3e ("[SCSI] aic94xx: new driver") Reviewed-by: Jason Yan <yanaijie@huawei.com> Signed-off-by: Duoming Zhou <duoming@zju.edu.cn> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since commit edc6afc54968 ("tty: switch to ktermios and new framework")
termios speed is no longer stored only in c_cflag member but also in new
additional c_ispeed and c_ospeed members. If BOTHER flag is set in c_cflag
then termios speed is stored only in these new members.
Since commit 027b57170bf8 ("serial: core: Fix initializing and restoring
termios speed") termios speed is available also in struct console.
So properly restore also c_ispeed and c_ospeed members after suspend to fix
restoring termios speed which is not represented by Bnnn constant.
Fixes: 4516d50aabed ("serial: 8250: Use canary to restart console after suspend") Signed-off-by: Pali Rohár <pali@kernel.org> Link: https://lore.kernel.org/r/20220924104324.4035-1-pali@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently the gsmi driver registers a panic notifier as well as
reboot and die notifiers. The callbacks registered are called in
atomic and very limited context - for instance, panic disables
preemption and local IRQs, also all secondary CPUs (not executing
the panic path) are shutdown.
With that said, taking a spinlock in this scenario is a dangerous
invitation for lockup scenarios. So, fix that by checking if the
spinlock is free to acquire in the panic notifier callback - if not,
bail-out and avoid a potential hang.
Fixes: 74c5b31c6618 ("driver: Google EFI SMI") Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Ard Biesheuvel <ardb@kernel.org> Cc: David Gow <davidgow@google.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Julius Werner <jwerner@chromium.org> Cc: Petr Mladek <pmladek@suse.com> Reviewed-by: Evan Green <evgreen@chromium.org> Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com> Link: https://lore.kernel.org/r/20220909200755.189679-1-gpiccoli@igalia.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
No error handling is performed when platform_device_add()
return fails. Refer to the error handling of driver_set_override(),
add error handling for platform_device_add().
In some initialization functions of this driver, memory is allocated with
'i' acting as an index variable and increasing from 0. The commit in
"Fixes" introduces some clean-up codes in case of allocation failure,
which free memory in reverse order with 'i' decreasing to 0. However,
there are some problems:
- The case i=0 is left out. Thus memory is leaked.
- In case memory allocation fails right from the start, the memory
freeing loops will start with i=-1 and invalid memory locations will
be accessed.
One of these loops has been fixed in commit c8ff91535880 ("staging:
vt6655: fix potential memory leak"). Fix the remaining erroneous loops.
drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->cal_clk' from clk_prepare_enable() not released on lines:
58.
drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->cal_sleep_clk' from clk_prepare_enable() not released on
lines: 58.
drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->phy_clk' from clk_prepare_enable() not released on lines:
58.
Fix this by calling proper clk_disable_unprepare calls.
Fixes: 0b56e9a7e835 ("phy: Group vendor specific phy drivers") Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com> Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org> Link: https://lore.kernel.org/r/20220914051334.69282-1-dzm91@hust.edu.cn Signed-off-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
lpuart_dma_shutdown tears down lpuart dma, but lpuart_flush_buffer can
still occur which in turn tries to access dma apis if lpuart_dma_tx_use
flag is true. At this point since dma is torn down, these dma apis can
abort. Set lpuart_dma_tx_use and the corresponding rx flag
lpuart_dma_rx_use to false in lpuart_dma_shutdown so that dmas are not
accessed after they are relinquished.
Invoking TIOCVHANGUP on 8250_mid port on Ice Lake-D and then reopening
the port triggers these faults during serial8250_do_startup():
DMAR: DRHD: handling fault status reg 3
DMAR: [DMA Write NO_PASID] Request device [00:1a.0] fault addr 0x0 [fault reason 0x05] PTE Write access is not set
If the IRQ hasn't been set up yet, the UART will have zeroes in its MSI
address/data registers. Disabling the IRQ at the interrupt controller
won't stop the UART from performing a DMA write to the address programmed
in its MSI address register (zero) when it wants to signal an interrupt.
The UARTs (in Ice Lake-D) implement PCI 2.1 style MSI without masking
capability, so there is no way to mask the interrupt at the source PCI
function level, except disabling the MSI capability entirely, but that
would cause it to fall back to INTx# assertion, and the PCI specification
prohibits disabling the MSI capability as a way to mask a function's
interrupt service request.
The MSI address register is zeroed by the hangup as the irq is freed.
The interrupt is signalled during serial8250_do_startup() performing a
THRE test that temporarily toggles THRI in IER. The THRE test currently
occurs before UART's irq (and MSI address) is properly set up.
Refactor serial8250_do_startup() such that irq is set up before the
THRE test. The current irq setup code is intermixed with the timer
setup code. As THRE test must be performed prior to the timer setup,
extract it into own function and call it only after the THRE test.
The ->setup_timer() needs to be part of the struct uart_8250_ops in
order to not create circular dependency between 8250 and 8250_base
modules.
so there is some additional clean up required on these error paths.
Fixes: 6f0764b5adea ("usb: dwc3: add a power supply for current control") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Link: https://lore.kernel.org/r/YyxFYFnP53j9sCg+@kili Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When opts->pnp_string is changed with configfs, new memory is allocated for
the string. It does not, however, update dev->pnp_string, even though the
memory is freed. When rquesting the string, the host then gets old or
corrupted data rather than the new string. The ieee 1284 id string should
be allowed to change while the device is connected.
The bug was introduced in commit fdc01cc286be ("usb: gadget: printer:
Remove pnp_string static buffer"), which changed opts->pnp_string from a
char[] to a char*.
This patch changes dev->pnp_string from a char* to a char** pointing to
opts->pnp_string.
commit 8b328f8002bc ("xhci: re-initialize the HC during resume if HCE was
set") introduced a new warning message when the host controller error
was set and re-initializing.
This is expected behavior on some designs which already set
`xhci->broken_suspend` so the new warning is alarming to some users.
Modify the code to only show the warning if this was a surprising behavior
to the XHCI driver.
Set 'iova' and 'length' on ib_mr in ib_uverbs and ib_core layers to let all
drivers have the members filled. Also, this commit removes redundancy in
the respective drivers.
Previously, commit 04c0a5fcfcf65 ("IB/uverbs: Set IOVA on IB MR in uverbs
layer") changed to set 'iova', but seems to have missed 'length' and the
ib_core layer at that time.
The responder should always use WC's SLID as the dlid, to follow the
IB SPEC section "13.5.4.2 COMMON RESPONSE ACTIONS":
A responder always takes the following actions in constructing a
response packet:
- The SLID of the received packet is used as the DLID in the response
packet.
projects / thirdparty / kernel / stable.git / log
