The full nohz tick is reprogrammed in irq_exit() only if the exit is not in
a nesting interrupt. This stands as an optimization: whether a hardirq or a
softirq is interrupted, the tick is going to be reprogrammed when necessary
at the end of the inner interrupt, with even potential new updates on the
timer queue.
When soft interrupts are interrupted, it's assumed that they are executing
on the tail of an interrupt return. In that case tick_nohz_irq_exit() is
called after softirq processing to take care of the tick reprogramming.
But the assumption is wrong: softirqs can be processed inline as well, ie:
outside of an interrupt, like in a call to local_bh_enable() or from
ksoftirqd.
Inline softirqs don't reprogram the tick once they are done, as opposed to
interrupt tail softirq processing. So if a tick interrupts an inline
softirq processing, the next timer will neither be reprogrammed from the
interrupting tick's irq_exit() nor after the interrupted softirq
processing. This situation may leave the tick unprogrammed while timers are
armed.
To fix this, simply keep reprogramming the tick even if a softirq has been
interrupted. That can be optimized further, but for now correctness is more
important.
Note that new timers enqueued in nohz_full mode after a softirq gets
interrupted will still be handled just fine through self-IPIs triggered by
the timer code.
Reported-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Cc: stable@vger.kernel.org # 4.14+
Link: https://lkml.kernel.org/r/1533303094-15855-1-git-send-email-frederic@kernel.org
The support of force threading interrupts which are set up with both a
primary and a threaded handler wreckaged the setup of regular requested
threaded interrupts (primary handler == NULL).
The reason is that it does not check whether the primary handler is set to
the default handler which wakes the handler thread. Instead it replaces the
thread handler with the primary handler as it would do with force threaded
interrupts which have been requested via request_irq(). So both the primary
and the thread handler become the same which then triggers the warnon that
the thread handler tries to wakeup a not configured secondary thread.
Fortunately this only happens when the driver omits the IRQF_ONESHOT flag
when requesting the threaded interrupt, which is normaly caught by the
sanity checks when force irq threading is disabled.
Fix it by skipping the force threading setup when a regular threaded
interrupt is requested. As a consequence the interrupt request which lacks
the IRQ_ONESHOT flag is rejected correctly instead of silently wreckaging
it.
Fixes: 2a1d3ab8986d ("genirq: Handle force threading of irqs with primary and thread handler")
Reported-by: Kurt Kanzenbach <kurt.kanzenbach@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Kurt Kanzenbach <kurt.kanzenbach@linutronix.de>
Cc: stable@vger.kernel.org
Code is emitting the following error message during boot on systems
without PMU hardware support while probing NMI capability.
NMI watchdog: Perf event create on CPU 0 failed with -2
This error is emitted as the perf subsystem returns -ENOENT due to lack of
PMUs in the system.
It is followed by the warning that NMI watchdog is disabled:
NMI watchdog: Perf NMI watchdog permanently disabled
While NMI disabled information is useful for ordinary users, seeing a PERF
event create failed with error code -2 is not.
Reduce the message severity to debug so that if debugging is still possible
in case the error code returned by perf is required for analysis.
Signed-off-by: Sinan Kaya <okaya@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Don Zickus <dzickus@redhat.com>
Cc: Kate Stewart <kstewart@linuxfoundation.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Colin Ian King <colin.king@canonical.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=599368
Link: https://lkml.kernel.org/r/20180803060943.2643-1-okaya@kernel.org
The bpf_get_local_storage() helper function is used
to get a pointer to the bpf local storage from a bpf program.
It takes a pointer to a storage map and flags as arguments.
Right now it accepts only cgroup storage maps, and flags
argument has to be 0. Further it can be extended to support
other types of local storage: e.g. thread local storage etc.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
As there is one-to-one relation between a bpf program
and cgroup local storage map, there is no sense in
creating a map of cgroup local storage maps.
Forbid it explicitly to avoid possible side effects.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
BPF_MAP_TYPE_CGROUP_STORAGE maps are special in a way
that the access from the bpf program side is lookup-free.
That means the result is guaranteed to be a valid
pointer to the cgroup storage; no NULL-check is required.
This patch introduces BPF_PTR_TO_MAP_VALUE return type,
which is required to cause the verifier accept programs,
which are not checking the map value pointer for being NULL.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This patch converts bpf_prog_array from an array of prog pointers
to the array of struct bpf_prog_array_item elements.
This allows to save a cgroup storage pointer for each bpf program
efficiently attached to a cgroup.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
If a bpf program is using cgroup local storage, allocate
a bpf_cgroup_storage structure automatically on attaching the program
to a cgroup and save the pointer into the corresponding bpf_prog_list
entry.
Analogically, release the cgroup local storage on detaching
of the bpf program.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit introduces the bpf_cgroup_storage_set() helper,
which will be used to pass a pointer to a cgroup storage
to the bpf helper.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit introduces BPF_MAP_TYPE_CGROUP_STORAGE maps:
a special type of maps which are implementing the cgroup storage.
>From the userspace point of view it's almost a generic
hash map with the (cgroup inode id, attachment type) pair
used as a key.
The only difference is that some operations are restricted:
1) a user can't create new entries,
2) a user can't remove existing entries.
The lookup from userspace is o(log(n)).
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commits extends existing bpf maps memory charging API
to support dynamic charging/uncharging.
This is required to account memory used by maps,
if all entries are created dynamically after
the map initialization.
Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
The BTF conflicts were simple overlapping changes.
The virtio_net conflict was an overlap of a fix of statistics counter,
happening alongisde a move over to a bonafide statistics structure
rather than counting value on the stack.
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix within_notrace_func() to check only notrace functions and to ignore the
kprobe-event which can not solve symbol addresses.
within_notrace_func() returns true if the given kprobe events probe point
seems to be out-of-range. But that is not the correct place to check for it,
it should be done in kprobes afterwards.
kprobe-events allow users to define a probe point on "currently unloaded
module" so that it can trace the event during module load. In this case, the
user will put a probe on a symbol which is not in kallsyms yet and it hits
the within_notrace_func(). As a result, kprobe-events always refuses if
user defines a probe on a "currenly unloaded module".
Fixes: commit 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function")
Link: http://lkml.kernel.org/r/153319624799.29007.13604430345640129926.stgit@devbox
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
we prefer to the kmemdup rather than kmalloc+memcpy. so just
replace them.
Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: Jessica Yu <jeyu@kernel.org>
Using cpu_all_mask in clockevents cpumask may result in issues while
comparing multiple clockevent devices to choose the preferred one.
On one of the platforms with 2 system (i.e. non per-CPU) timers with
different ratings, having cpu_all_mask for one of the device resulted in a
boot hang due to a endless loop in clockevents_notify_released() as both
were clocksources were selected as preferred.
In order to prevent such issues in the future, warn if any clockevent
driver sets cpu_all_mask as it's cpumask and just override it to use
cpu_possible_mask. All the existing occurrences of cpu_all_mask are already
replaced with cpu_possible_mask.
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lkml.kernel.org/r/1531308264-24220-3-git-send-email-sudeep.holla@arm.com
This is the last instance of cpu_all_mask usage in the core framework.
Replace it with cpu_possible_mask like all other instances in the
clockevent drivers. This makes it possible to add a warning in the core
clockevents_register_device on usage of cpu_all_mask from any clockevent
drivers in the future.
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lkml.kernel.org/r/1531308264-24220-2-git-send-email-sudeep.holla@arm.com
timer_base::must_forward_clock is indicating that the base clock might be
stale due to a long idle sleep.
The forwarding of the base clock takes place in the timer softirq or when a
timer is enqueued to a base which is idle. If the enqueue of timer to an
idle base happens from a remote CPU, then the following race can happen:
CPU0 CPU1
run_timer_softirq mod_timer
base = lock_timer_base(timer);
base->must_forward_clk = false
if (base->must_forward_clk)
forward(base); -> skipped
enqueue_timer(base, timer, idx);
-> idx is calculated high due to
stale base
unlock_timer_base(timer);
base = lock_timer_base(timer);
forward(base);
The root cause is that timer_base::must_forward_clk is cleared outside the
timer_base::lock held region, so the remote queuing CPU observes it as
cleared, but the base clock is still stale. This can cause large
granularity values for timers, i.e. the accuracy of the expiry time
suffers.
Prevent this by clearing the flag with timer_base::lock held, so that the
forwarding takes place before the cleared flag is observable by a remote
CPU.
Signed-off-by: Gaurav Kohli <gkohli@codeaurora.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: john.stultz@linaro.org
Cc: sboyd@kernel.org
Cc: linux-arm-msm@vger.kernel.org
Link: https://lkml.kernel.org/r/1533199863-22748-1-git-send-email-gkohli@codeaurora.org
Return statements in functions returning bool should use true or false
instead of an integer value.
This code was detected with the help of Coccinelle.
Link: http://lkml.kernel.org/r/20180802010056.GA31012@embeddedor.com
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The value of ring_buffer_record_is_set_on() is either true or false, so have
its return value be bool.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The value of ring_buffer_record_is_on() is either true or false, so have its
return value be bool.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Almost all architectures include it. Add a ARCH_NO_PREEMPT symbol to
disable preempt support for alpha, hexagon, non-coldfire m68k and
user mode Linux.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
There's code that expects tracer_tracing_is_on() to be either true or false,
not some random number. Currently, it should only return one or zero, but
just in case, change its return value to bool, to enforce it.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The start function of the hwlat tracer should never be called when the hwlat
thread already exists. If it is called, do a WARN_ON().
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The hwlat tracer uses a kernel thread to measure latencies. The function
that creates this kernel thread, start_kthread(), can be called when the
tracer is initialized and when the tracer is explicitly enabled.
start_kthread() does not check if there is an existing hwlat kernel
thread and will create a new one each time it is called.
This causes the reference to the previous thread to be lost. Without the
thread reference, the old kernel thread becomes unstoppable and
continues to use CPU time even after the hwlat tracer has been disabled.
This problem can be observed when a system is booted with tracing
enabled and the hwlat tracer is configured like this:
echo hwlat > current_tracer; echo 1 > tracing_on
Add the missing check for an existing kernel thread in start_kthread()
to prevent this problem. This function and the rest of the hwlat kernel
thread setup and teardown are already serialized because they are called
through the tracer core code with trace_type_lock held.
[
Note, this only fixes the symptom. The real fix was not to call
this function when tracing_on was already one. But this still makes
the code more robust, so we'll add it.
]
Link: http://lkml.kernel.org/r/1533120354-22923-1-git-send-email-erica.bugden@linutronix.de
Signed-off-by: Erica Bugden <erica.bugden@linutronix.de>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Currently, when one echo's in 1 into tracing_on, the current tracer's
"start()" function is executed, even if tracing_on was already one. This can
lead to strange side effects. One being that if the hwlat tracer is enabled,
and someone does "echo 1 > tracing_on" into tracing_on, the hwlat tracer's
start() function is called again which will recreate another kernel thread,
and make it unable to remove the old one.
Link: http://lkml.kernel.org/r/1533120354-22923-1-git-send-email-erica.bugden@linutronix.de
Cc: stable@vger.kernel.org
Fixes: 2df8f8a6a897e ("tracing: Fix regression with irqsoff tracer and tracing_on file")
Reported-by: Erica Bugden <erica.bugden@linutronix.de>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
We were hitting a panic in production where we put too many times on the
request queue. This is because we'd get the throttle_queue of the
parent if we fork()'ed while we needed to be throttled, but we didn't
have a reference on it. Instead just clear these flags on fork so the
child doesn't pay for the sins of its father.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
-----BEGIN PGP SIGNATURE-----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=MN3z
-----END PGP SIGNATURE-----
Merge tag 'audit-pr-20180731' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit
Pull audit fix from Paul Moore:
"A single small audit fix to guard against memory allocation failures
when logging information about a kernel module load.
It's small, easy to understand, and self-contained; while nothing is
zero risk, this should be pretty low"
* tag 'audit-pr-20180731' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit:
audit: fix potential null dereference 'context->module.name'
When check_alu_op() handles a BPF_MOV64 between two registers,
it calls check_reg_arg(DST_OP) on the dst register, marking it
as unbounded. If the src and dst register are the same, this
marks the src as unbounded, which can lead to unexpected errors
for further checks that rely on bounds info. For example:
BPF_MOV64_IMM(BPF_REG_2, 0),
BPF_MOV64_REG(BPF_REG_2, BPF_REG_2),
BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_2),
BPF_MOV64_IMM(BPF_REG_0, 0),
BPF_EXIT_INSN(),
Results in:
"math between ctx pointer and register with unbounded
min value is not allowed"
check_alu_op() now uses check_reg_arg(DST_OP_NO_MARK), and MOVs
that need to mark the dst register (MOVIMM, MOV32) do so.
Added a test case for MOV64 dst == src, and dst != src.
Signed-off-by: Arthur Fabre <afabre@cloudflare.com>
Acked-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
local_timer_softirq_pending() checks whether the timer softirq is
pending with: local_softirq_pending() & TIMER_SOFTIRQ.
This is wrong because TIMER_SOFTIRQ is the softirq number and not a
bitmask. So the test checks for the wrong bit.
Use BIT(TIMER_SOFTIRQ) instead.
Fixes: 5d62c183f9e9 ("nohz: Prevent a timer interrupt storm in tick_nohz_stop_sched_tick()")
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Daniel Bristot de Oliveira <bristot@redhat.com>
Acked-by: Frederic Weisbecker <frederic@kernel.org>
Cc: bigeasy@linutronix.de
Cc: peterz@infradead.org
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20180731161358.29472-1-anna-maria@linutronix.de
This patch detaches the preemptirq tracepoints from the tracers and
keeps it separate.
Advantages:
* Lockdep and irqsoff event can now run in parallel since they no longer
have their own calls.
* This unifies the usecase of adding hooks to an irqsoff and irqson
event, and a preemptoff and preempton event.
3 users of the events exist:
- Lockdep
- irqsoff and preemptoff tracers
- irqs and preempt trace events
The unification cleans up several ifdefs and makes the code in preempt
tracer and irqsoff tracers simpler. It gets rid of all the horrific
ifdeferry around PROVE_LOCKING and makes configuration of the different
users of the tracepoints more easy and understandable. It also gets rid
of the time_* function calls from the lockdep hooks used to call into
the preemptirq tracer which is not needed anymore. The negative delta in
lines of code in this patch is quite large too.
In the patch we introduce a new CONFIG option PREEMPTIRQ_TRACEPOINTS
as a single point for registering probes onto the tracepoints. With
this,
the web of config options for preempt/irq toggle tracepoints and its
users becomes:
PREEMPT_TRACER PREEMPTIRQ_EVENTS IRQSOFF_TRACER PROVE_LOCKING
| | \ | |
\ (selects) / \ \ (selects) /
TRACE_PREEMPT_TOGGLE ----> TRACE_IRQFLAGS
\ /
\ (depends on) /
PREEMPTIRQ_TRACEPOINTS
Other than the performance tests mentioned in the previous patch, I also
ran the locking API test suite. I verified that all tests cases are
passing.
I also injected issues by not registering lockdep probes onto the
tracepoints and I see failures to confirm that the probes are indeed
working.
This series + lockdep probes not registered (just to inject errors):
[ 0.000000] hard-irqs-on + irq-safe-A/21: ok | ok | ok |
[ 0.000000] soft-irqs-on + irq-safe-A/21: ok | ok | ok |
[ 0.000000] sirq-safe-A => hirqs-on/12:FAILED|FAILED| ok |
[ 0.000000] sirq-safe-A => hirqs-on/21:FAILED|FAILED| ok |
[ 0.000000] hard-safe-A + irqs-on/12:FAILED|FAILED| ok |
[ 0.000000] soft-safe-A + irqs-on/12:FAILED|FAILED| ok |
[ 0.000000] hard-safe-A + irqs-on/21:FAILED|FAILED| ok |
[ 0.000000] soft-safe-A + irqs-on/21:FAILED|FAILED| ok |
[ 0.000000] hard-safe-A + unsafe-B #1/123: ok | ok | ok |
[ 0.000000] soft-safe-A + unsafe-B #1/123: ok | ok | ok |
With this series + lockdep probes registered, all locking tests pass:
[ 0.000000] hard-irqs-on + irq-safe-A/21: ok | ok | ok |
[ 0.000000] soft-irqs-on + irq-safe-A/21: ok | ok | ok |
[ 0.000000] sirq-safe-A => hirqs-on/12: ok | ok | ok |
[ 0.000000] sirq-safe-A => hirqs-on/21: ok | ok | ok |
[ 0.000000] hard-safe-A + irqs-on/12: ok | ok | ok |
[ 0.000000] soft-safe-A + irqs-on/12: ok | ok | ok |
[ 0.000000] hard-safe-A + irqs-on/21: ok | ok | ok |
[ 0.000000] soft-safe-A + irqs-on/21: ok | ok | ok |
[ 0.000000] hard-safe-A + unsafe-B #1/123: ok | ok | ok |
[ 0.000000] soft-safe-A + unsafe-B #1/123: ok | ok | ok |
Link: http://lkml.kernel.org/r/20180730222423.196630-4-joel@joelfernandes.org
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
It is useful to get the running time of a thread. Doing so in an
efficient manner can be important for performance of user applications.
Avoiding system calls in `clock_gettime` when handling
CLOCK_THREAD_CPUTIME_ID is important. Other clocks are handled in the
VDSO, but CLOCK_THREAD_CPUTIME_ID falls back on the system call.
CLOCK_THREAD_CPUTIME_ID is not handled in the VDSO since it would have
costs associated with maintaining updated user space accessible time
offsets. These offsets have to be updated everytime the a thread is
scheduled/descheduled. However, for programs regularly checking the
running time of a thread, this is a performance improvement.
This patch takes a middle ground, and adds support for cap_user_time an
optional feature of the perf_event API. This way costs are only
incurred when the perf_event api is enabled. This is done the same way
as it is in x86.
Ultimately this allows calculating the thread running time in userspace
on aarch64 as follows (adapted from perf_event_open manpage):
u32 seq, time_mult, time_shift;
u64 running, count, time_offset, quot, rem, delta;
struct perf_event_mmap_page *pc;
pc = buf; // buf is the perf event mmaped page as documented in the API.
if (pc->cap_usr_time) {
do {
seq = pc->lock;
barrier();
running = pc->time_running;
count = readCNTVCT_EL0(); // Read ARM hardware clock.
time_offset = pc->time_offset;
time_mult = pc->time_mult;
time_shift = pc->time_shift;
barrier();
} while (pc->lock != seq);
quot = (count >> time_shift);
rem = count & (((u64)1 << time_shift) - 1);
delta = time_offset + quot * time_mult +
((rem * time_mult) >> time_shift);
running += delta;
// running now has the current nanosecond level thread time.
}
Summary of changes in the patch:
For aarch64 systems, make arch_perf_update_userpage update the timing
information stored in the perf_event page. Requiring the following
calculations:
- Calculate the appropriate time_mult, and time_shift factors to convert
ticks to nano seconds for the current clock frequency.
- Adjust the mult and shift factors to avoid shift factors of 32 bits.
(possibly unnecessary)
- The time_offset userspace should apply when doing calculations:
negative the current sched time (now), because time_running and
time_enabled fields of the perf_event page have just been updated.
Toggle bits to appropriate values:
- Enable cap_user_time
Signed-off-by: Michael O'Farrell <micpof@gmail.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Pull networking fixes from David Miller:
"Several smallish fixes, I don't think any of this requires another -rc
but I'll leave that up to you:
1) Don't leak uninitialzed bytes to userspace in xfrm_user, from Eric
Dumazet.
2) Route leak in xfrm_lookup_route(), from Tommi Rantala.
3) Premature poll() returns in AF_XDP, from Björn Töpel.
4) devlink leak in netdevsim, from Jakub Kicinski.
5) Don't BUG_ON in fib_compute_spec_dst, the condition can
legitimately happen. From Lorenzo Bianconi.
6) Fix some spectre v1 gadgets in generic socket code, from Jeremy
Cline.
7) Don't allow user to bind to out of range multicast groups, from
Dmitry Safonov with a follow-up by Dmitry Safonov.
8) Fix metrics leak in fib6_drop_pcpu_from(), from Sabrina Dubroca"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
netlink: Don't shift with UB on nlk->ngroups
net/ipv6: fix metrics leak
xen-netfront: wait xenbus state change when load module manually
can: ems_usb: Fix memory leak on ems_usb_disconnect()
openvswitch: meter: Fix setting meter id for new entries
netlink: Do not subscribe to non-existent groups
NET: stmmac: align DMA stuff to largest cache line length
tcp_bbr: fix bw probing to raise in-flight data for very small BDPs
net: socket: Fix potential spectre v1 gadget in sock_is_registered
net: socket: fix potential spectre v1 gadget in socketcall
net: mdio-mux: bcm-iproc: fix wrong getter and setter pair
ipv4: remove BUG_ON() from fib_compute_spec_dst
enic: handle mtu change for vf properly
net: lan78xx: fix rx handling before first packet is send
nfp: flower: fix port metadata conversion bug
bpf: use GFP_ATOMIC instead of GFP_KERNEL in bpf_parse_prog()
bpf: fix bpf_skb_load_bytes_relative pkt length check
perf build: Build error in libbpf missing initialization
net: ena: Fix use of uninitialized DMA address bits field
bpf: btf: Use exact btf value_size match in map_check_btf()
...
In recent tests with IRQ on/off tracepoints, a large performance
overhead ~10% is noticed when running hackbench. This is root caused to
calls to rcu_irq_enter_irqson and rcu_irq_exit_irqson from the
tracepoint code. Following a long discussion on the list [1] about this,
we concluded that srcu is a better alternative for use during rcu idle.
Although it does involve extra barriers, its lighter than the sched-rcu
version which has to do additional RCU calls to notify RCU idle about
entry into RCU sections.
In this patch, we change the underlying implementation of the
trace_*_rcuidle API to use SRCU. This has shown to improve performance
alot for the high frequency irq enable/disable tracepoints.
Test: Tested idle and preempt/irq tracepoints.
Here are some performance numbers:
With a run of the following 30 times on a single core x86 Qemu instance
with 1GB memory:
hackbench -g 4 -f 2 -l 3000
Completion times in seconds. CONFIG_PROVE_LOCKING=y.
No patches (without this series)
Mean: 3.048
Median: 3.025
Std Dev: 0.064
With Lockdep using irq tracepoints with RCU implementation:
Mean: 3.451 (-11.66 %)
Median: 3.447 (-12.22%)
Std Dev: 0.049
With Lockdep using irq tracepoints with SRCU implementation (this series):
Mean: 3.020 (I would consider the improvement against the "without
this series" case as just noise).
Median: 3.013
Std Dev: 0.033
[1] https://patchwork.kernel.org/patch/10344297/
[remove rcu_read_lock_sched_notrace as its the equivalent of
preempt_disable_notrace and is unnecessary to call in tracepoint code]
Link: http://lkml.kernel.org/r/20180730222423.196630-3-joel@joelfernandes.org
Cleaned-up-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
[ Simplified WARN_ON_ONCE() ]
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
get_cpu_var disables preemption which has the potential to call into the
preemption disable trace points causing some complications. There's also
no need to disable preemption in uses of get_lock_stats anyway since
preempt is already disabled. So lets simplify the code.
Link: http://lkml.kernel.org/r/20180730222423.196630-2-joel@joelfernandes.org
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Move selftest function to its own compile unit so it can be compiled
with the ftrace cflags (CC_FLAGS_FTRACE) allowing it to be probed
during the ftrace startup tests.
Link: http://lkml.kernel.org/r/153294604271.32740.16490677128630177030.stgit@devbox
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Prohibit kprobe-events probing on notrace functions. Since probing on a
notrace function can cause a recursive event call. In most cases those are just
skipped, but in some cases it falls into an infinite recursive call.
This protection can be disabled by the kconfig
CONFIG_KPROBE_EVENTS_ON_NOTRACE=y, but it is highly recommended to keep it
"n" for normal kernel builds. Note that this is only available if "kprobes on
ftrace" has been implemented on the target arch and CONFIG_KPROBES_ON_FTRACE=y.
Link: http://lkml.kernel.org/r/153294601436.32740.10557881188933661239.stgit@devbox
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Francis Deslauriers <francis.deslauriers@efficios.com>
[ Slight grammar and spelling fixes ]
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The variable 'context->module.name' may be null pointer when
kmalloc return null, so it's better to check it before using
to avoid null dereference.
Another one more thing this patch does is using kstrdup instead
of (kmalloc + strcpy), and signal a lost record via audit_log_lost.
Cc: stable@vger.kernel.org # 4.11
Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
Reviewed-by: Jiang Biao <jiang.biao2@zte.com.cn>
Reviewed-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
After commit 249d4a9b3246 ("timers: Reinitialize per cpu bases on hotplug")
i.e. the introduction of state CPUHP_TIMERS_PREPARE instead of
CPUHP_TIMERS_DEAD the step name "timers:dead" is not longer accurate.
Rename it to "timers:prepare".
[ tglx: Massaged changelog ]
Signed-off-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: gkohli@codeaurora.org
Cc: neeraju@codeaurora.org
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Brendan Jackman <brendan.jackman@arm.com>
Cc: Mathieu Malaterre <malat@debian.org>
Link: https://lkml.kernel.org/r/1532443668-26810-1-git-send-email-mojha@codeaurora.org
Pull scheduler fixes from Ingo Molnar:
"Misc fixes:
- a deadline scheduler related bug fix which triggered a kernel
warning
- an RT_RUNTIME_SHARE fix
- a stop_machine preemption fix
- a potential NULL dereference fix in sched_domain_debug_one()"
* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/rt: Restore rt_runtime after disabling RT_RUNTIME_SHARE
sched/deadline: Update rq_clock of later_rq when pushing a task
stop_machine: Disable preemption after queueing stopper threads
sched/topology: Check variable group before dereferencing it
Pull locking fixes from Ingo Molnar:
"A paravirt UP-patching fix, and an I2C MUX driver lockdep warning fix"
* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/pvqspinlock/x86: Use LOCK_PREFIX in __pv_queued_spin_unlock() assembly code
i2c/mux, locking/core: Annotate the nested rt_mutex usage
locking/rtmutex: Allow specifying a subclass for nested locking
sched_clock_init() used be called early during boot when interrupts were
still disabled. After the recent changes to utilize sched clock early the
sched_clock_init() call happens when interrupts are already enabled, which
triggers the following warning:
WARNING: CPU: 0 PID: 0 at kernel/time/sched_clock.c:180 sched_clock_register+0x44/0x278
[<c001a13c>] (warn_slowpath_null) from [<c052367c>] (sched_clock_register+0x44/0x278)
[<c052367c>] (sched_clock_register) from [<c05238d8>] (generic_sched_clock_init+0x28/0x88)
[<c05238d8>] (generic_sched_clock_init) from [<c0521a00>] (sched_clock_init+0x54/0x74)
[<c0521a00>] (sched_clock_init) from [<c0519c18>] (start_kernel+0x310/0x3e4)
[<c0519c18>] (start_kernel) from [<00000000>] ( (null))
Disable IRQs for the duration of generic_sched_clock_init().
Fixes: 857baa87b642 ("sched/clock: Enable sched clock early")
Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: steven.sistare@oracle.com
Cc: daniel.m.jordan@oracle.com
Link: https://lkml.kernel.org/r/20180730135252.24599-1-pasha.tatashin@oracle.com
On arches with no persistent clock a message like this is printed during
boot:
[ 0.000000] Persistent clock returned invalid value
The value is not invalid: Zero means that no persistent clock is available
and the absence of persistent clock should be quietly accepted.
Fixes: 3eca993740b8 ("timekeeping: Replace read_boot_clock64() with read_persistent_wall_and_boot_offset()")
Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: steven.sistare@oracle.com
Cc: daniel.m.jordan@oracle.com
Cc: sboyd@kernel.org
Cc: john.stultz@linaro.org
Link: https://lkml.kernel.org/r/20180725200018.23722-1-pasha.tatashin@oracle.com
Daniel Borkmann says:
====================
pull-request: bpf 2018-07-28
The following pull-request contains BPF updates for your *net* tree.
The main changes are:
1) API fixes for libbpf's BTF mapping of map key/value types in order
to make them compatible with iproute2's BPF_ANNOTATE_KV_PAIR()
markings, from Martin.
2) Fix AF_XDP to not report POLLIN prematurely by using the non-cached
consumer pointer of the RX queue, from Björn.
3) Fix __xdp_return() to check for NULL pointer after the rhashtable
lookup that retrieves the allocator object, from Taehee.
4) Fix x86-32 JIT to adjust ebp register in prologue and epilogue
by 4 bytes which got removed from overall stack usage, from Wang.
5) Fix bpf_skb_load_bytes_relative() length check to use actual
packet length, from Daniel.
6) Fix uninitialized return code in libbpf bpf_perf_event_read_simple()
handler, from Thomas.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
