Split the bpf_perf_event_output() helper as a preparation into
two parts. The new bpf_perf_event_output() will prepare the raw
record itself and test for unknown flags from BPF trace context,
where the __bpf_perf_event_output() does the core work. The
latter will be reused later on from bpf_event_output() directly.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds support for non-linear data on raw records. It
extends raw records to have one or multiple fragments that will
be written linearly into the ring slot, where each fragment can
optionally have a custom callback handler to walk and extract
complex, possibly non-linear data.

If a callback handler is provided for a fragment, then the new
__output_custom() will be used instead of __output_copy() for
the perf_output_sample() part. perf_prepare_sample() does all
the size calculation only once, so perf_output_sample() doesn't
need to redo the same work anymore, meaning real_size and padding
will be cached in the raw record. The raw record becomes 32 bytes
in size without holes; to not increase it further and to avoid
doing unnecessary recalculations in fast-path, we can reuse
next pointer of the last fragment, idea here is borrowed from
ZERO_OR_NULL_PTR(), which should keep the perf_output_sample()
path for PERF_SAMPLE_RAW minimal.

This facility is needed for BPF's event output helper as a first
user that will, in a follow-up, add an additional perf_raw_frag
to its perf_raw_record in order to be able to more efficiently
dump skb context after a linear head meta data related to it.
skbs can be non-linear and thus need a custom output function to
dump buffers. Currently, the skb data needs to be copied twice;
with the help of __output_custom() this work only needs to be
done once. Future users could be things like XDP/BPF programs
that work on different context though and would thus also have
a different callback function.

The few users of raw records are adapted to initialize their frag
data from the raw record itself, no change in behavior for them.
The code is based upon a PoC diff provided by Peter Zijlstra [1].

  [1] http://thread.gmane.org/gmane.linux.network/421294



Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

over time there were multiple requests to access different data
structures and fields of task_struct current, so finally add
the helper to access 'current' as-is. Tracing bpf programs will do
the rest of walking the pointers via bpf_probe_read().
Note that current can be null and bpf program has to deal it with,
but even dumb passing null into bpf_probe_read() is still safe.

Suggested-by: Brendan Gregg <brendan.d.gregg@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Follow-up commit to 1e33759c ("bpf, trace: add BPF_F_CURRENT_CPU
flag for bpf_perf_event_output") to add the same functionality into
bpf_perf_event_read() helper. The split of index into flags and index
component is also safe here, since such large maps are rejected during
map allocation time.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

We currently have two invocations, which is unnecessary. Fetch it only
once and use the smp_processor_id() variant, so we also get preemption
checks along with it when DEBUG_PREEMPT is set.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Some minor cleanups: i) Remove the unlikely() from fd array map lookups
and let the CPU branch predictor do its job, scenarios where there is not
always a map entry are very well valid. ii) Move the attribute type check
in the bpf_perf_event_read() helper a bit earlier so it's consistent wrt
checks with bpf_perf_event_output() helper as well. iii) remove some
comments that are self-documenting in kprobe_prog_is_valid_access() and
therefore make it consistent to tp_prog_is_valid_access() as well.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

If a task uses a non constant string for the format parameter in
trace_printk(), then the trace_printk_fmt variable is set to NULL. This
variable is then saved in the __trace_printk_fmt section.

The function hold_module_trace_bprintk_format() checks to see if duplicate
formats are used by modules, and reuses them if so (saves them to the list
if it is new). But this function calls lookup_format() that does a strcmp()
to the value (which is now NULL) and can cause a kernel oops.

This wasn't an issue till 3debb0a9 ("tracing: Fix trace_printk() to print
when not using bprintk()") which added "__used" to the trace_printk_fmt
variable, and before that, the kernel simply optimized it out (no NULL value
was saved).

The fix is simply to handle the NULL pointer in lookup_format() and have the
caller ignore the value if it was NULL.

Link: http://lkml.kernel.org/r/1464769870-18344-1-git-send-email-zhengjun.xing@intel.com



Reported-by: xingzhen <zhengjun.xing@intel.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Fixes: 3debb0a9 ("tracing: Fix trace_printk() to print when not using bprintk()")
Cc: stable@vger.kernel.org # v3.5+
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The behavior of perf event arrays are quite different from all
others as they are tightly coupled to perf event fds, f.e. shown
recently by commit e03e7ee3 ("perf/bpf: Convert perf_event_array
to use struct file") to make refcounting on perf event more robust.
A remaining issue that the current code still has is that since
additions to the perf event array take a reference on the struct
file via perf_event_get() and are only released via fput() (that
cleans up the perf event eventually via perf_event_release_kernel())
when the element is either manually removed from the map from user
space or automatically when the last reference on the perf event
map is dropped. However, this leads us to dangling struct file's
when the map gets pinned after the application owning the perf
event descriptor exits, and since the struct file reference will
in such case only be manually dropped or via pinned file removal,
it leads to the perf event living longer than necessary, consuming
needlessly resources for that time.

Relations between perf event fds and bpf perf event map fds can be
rather complex. F.e. maps can act as demuxers among different perf
event fds that can possibly be owned by different threads and based
on the index selection from the program, events get dispatched to
one of the per-cpu fd endpoints. One perf event fd (or, rather a
per-cpu set of them) can also live in multiple perf event maps at
the same time, listening for events. Also, another requirement is
that perf event fds can get closed from application side after they
have been attached to the perf event map, so that on exit perf event
map will take care of dropping their references eventually. Likewise,
when such maps are pinned, the intended behavior is that a user
application does bpf_obj_get(), puts its fds in there and on exit
when fd is released, they are dropped from the map again, so the map
acts rather as connector endpoint. This also makes perf event maps
inherently different from program arrays as described in more detail
in commit c9da161c ("bpf: fix clearing on persistent program
array maps").

To tackle this, map entries are marked by the map struct file that
added the element to the map. And when the last reference to that map
struct file is released from user space, then the tracked entries
are purged from the map. This is okay, because new map struct files
instances resp. frontends to the anon inode are provided via
bpf_map_new_fd() that is called when we invoke bpf_obj_get_user()
for retrieving a pinned map, but also when an initial instance is
created via map_create(). The rest is resolved by the vfs layer
automatically for us by keeping reference count on the map's struct
file. Any concurrent updates on the map slot are fine as well, it
just means that perf_event_fd_array_release() needs to delete less
of its own entires.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

similar to bpf_perf_event_output() the bpf_perf_event_read() helper
needs to check the type of the perf_event before reading the counter.

Fixes: a43eec30 ("bpf: introduce bpf_perf_event_output() helper")
Reported-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

The ctx structure passed into bpf programs is different depending on bpf
program type. The verifier incorrectly marked ctx->data and ctx->data_end
access based on ctx offset only. That caused loads in tracing programs
int bpf_prog(struct pt_regs *ctx) { .. ctx->ax .. }
to be incorrectly marked as PTR_TO_PACKET which later caused verifier
to reject the program that was actually valid in tracing context.
Fix this by doing program type specific matching of ctx offsets.

Fixes: 969bf05e ("bpf: direct packet access")
Reported-by: Sasha Goldshtein <goldshtn@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

In bpf_perf_event_read() and bpf_perf_event_output(), we must use
READ_ONCE() for fetching the struct file pointer, which could get
updated concurrently, so we must prevent the compiler from potential
refetching.

We already do this with tail calls for fetching the related bpf_prog,
but not so on stored perf events. Semantics for both are the same
with regards to updates.

Fixes: a43eec30 ("bpf: introduce bpf_perf_event_output() helper")
Fixes: 35578d79 ("bpf: Implement function bpf_perf_event_read() that get the selected hardware PMU conuter")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

In ftrace.c inside the function alloc_retstack_tasklist() (which will be
invoked when function_graph tracing is on) the tasklist_lock is being
held as reader while iterating through a list of threads. Here the lock
is being held as reader with irqs disabled. The tasklist_lock is never
write_locked in interrupt context so it is safe to not disable interrupts
for the duration of read_lock in this block which, can be significant,
given the block of code iterates through all threads. Hence changing the
code to call read_lock() and read_unlock() instead of read_lock_irqsave()
and read_unlock_irqrestore().

A similar change was made in commits: 8063e41d ("tracing: Change
syscall_*regfunc() to check PF_KTHREAD and use for_each_process_thread()")'
and 3472eaa1 ("sched: normalize_rt_tasks(): Don't use _irqsave for
tasklist_lock, use task_rq_lock()")'

Link: http://lkml.kernel.org/r/1463500874-77480-1-git-send-email-soumya.p.n@hpe.com



Signed-off-by: Soumya PN <soumya.p.n@hpe.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

If the size passed to ring_buffer_resize() is greater than MAX_LONG - BUF_PAGE_SIZE
then the DIV_ROUND_UP() will return zero.

Here's the details:

  # echo 18014398509481980 > /sys/kernel/debug/tracing/buffer_size_kb

tracing_entries_write() processes this and converts kb to bytes.

 18014398509481980 << 10 = 18446744073709547520

and this is passed to ring_buffer_resize() as unsigned long size.

 size = DIV_ROUND_UP(size, BUF_PAGE_SIZE);

Where DIV_ROUND_UP(a, b) is (a + b - 1)/b

BUF_PAGE_SIZE is 4080 and here

 18446744073709547520 + 4080 - 1 = 18446744073709551599

where 18446744073709551599 is still smaller than 2^64

 2^64 - 18446744073709551599 = 17

But now 18446744073709551599 / 4080 = 4521260802379792

and size = size * 4080 = 18446744073709551360

This is checked to make sure its still greater than 2 * 4080,
which it is.

Then we convert to the number of buffer pages needed.

 nr_page = DIV_ROUND_UP(size, BUF_PAGE_SIZE)

but this time size is 18446744073709551360 and

 2^64 - (18446744073709551360 + 4080 - 1) = -3823

Thus it overflows and the resulting number is less than 4080, which makes

  3823 / 4080 = 0

an nr_pages is set to this. As we already checked against the minimum that
nr_pages may be, this causes the logic to fail as well, and we crash the
kernel.

There's no reason to have the two DIV_ROUND_UP() (that's just result of
historical code changes), clean up the code and fix this bug.

Cc: stable@vger.kernel.org # 3.5+
Fixes: 83f40318 ("ring-buffer: Make removal of ring buffer pages atomic")
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The size variable to change the ring buffer in ftrace is a long. The
nr_pages used to update the ring buffer based on the size is int. On 64 bit
machines this can cause an overflow problem.

For example, the following will cause the ring buffer to crash:

 # cd /sys/kernel/debug/tracing
 # echo 10 > buffer_size_kb
 # echo 8556384240 > buffer_size_kb

Then you get the warning of:

 WARNING: CPU: 1 PID: 318 at kernel/trace/ring_buffer.c:1527 rb_update_pages+0x22f/0x260

Which is:

  RB_WARN_ON(cpu_buffer, nr_removed);

Note each ring buffer page holds 4080 bytes.

This is because:

 1) 10 causes the ring buffer to have 3 pages.
    (10kb requires 3 * 4080 pages to hold)

 2) (2^31 / 2^10  + 1) * 4080 = 8556384240
    The value written into buffer_size_kb is shifted by 10 and then passed
    to ring_buffer_resize(). 8556384240 * 2^10 = 8761737461760

 3) The size passed to ring_buffer_resize() is then divided by BUF_PAGE_SIZE
    which is 4080. 8761737461760 / 4080 = 2147484672

 4) nr_pages is subtracted from the current nr_pages (3) and we get:
    2147484669. This value is saved in a signed integer nr_pages_to_update

 5) 2147484669 is greater than 2^31 but smaller than 2^32, a signed int
    turns into the value of -2147482627

 6) As the value is a negative number, in update_pages_handler() it is
    negated and passed to rb_remove_pages() and 2147482627 pages will
    be removed, which is much larger than 3 and it causes the warning
    because not all the pages asked to be removed were removed.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=118001



Cc: stable@vger.kernel.org # 2.6.28+
Fixes: 7a8e76a3 ("tracing: unified trace buffer")
Reported-by: Hao Qin <QEver.cn@gmail.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

BLK_TC_NOTIFY is missed in mask_maps, so we can't print out notify or
set mask with 'notify' name.

Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

commit f4a1d08c introduces a regression. Originally for
BLK_TN_MESSAGE, we add message in trace and return. The commit ignores
the early return and add garbage info.

Signed-off-by: Shaohua Li <shli@fb.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

Filtering of events requires the data to be written to the ring buffer
before it can be decided to filter or not. This is because the parameters of
the filter are based on the result that is written to the ring buffer and
not on the parameters that are passed into the trace functions.

The ftrace ring buffer is optimized for writing into the ring buffer and
committing. The discard procedure used when filtering decides the event
should be discarded is much more heavy weight. Thus, using a temporary
filter when filtering events can speed things up drastically.

Without a temp buffer we have:

 # trace-cmd start -p nop
 # perf stat -r 10 hackbench 50
       0.790706626 seconds time elapsed ( +-  0.71% )

 # trace-cmd start -e all
 # perf stat -r 10 hackbench 50
       1.566904059 seconds time elapsed ( +-  0.27% )

 # trace-cmd start -e all -f 'common_preempt_count==20'
 # perf stat -r 10 hackbench 50
       1.690598511 seconds time elapsed ( +-  0.19% )

 # trace-cmd start -e all -f 'common_preempt_count!=20'
 # perf stat -r 10 hackbench 50
       1.707486364 seconds time elapsed ( +-  0.30% )

The first run above is without any tracing, just to get a based figure.
hackbench takes ~0.79 seconds to run on the system.

The second run enables tracing all events where nothing is filtered. This
increases the time by 100% and hackbench takes 1.57 seconds to run.

The third run filters all events where the preempt count will equal "20"
(this should never happen) thus all events are discarded. This takes 1.69
seconds to run. This is 10% slower than just committing the events!

The last run enables all events and filters where the filter will commit all
events, and this takes 1.70 seconds to run. The filtering overhead is
approximately 10%. Thus, the discard and commit of an event from the ring
buffer may be about the same time.

With this patch, the numbers change:

 # trace-cmd start -p nop
 # perf stat -r 10 hackbench 50
       0.778233033 seconds time elapsed ( +-  0.38% )

 # trace-cmd start -e all
 # perf stat -r 10 hackbench 50
       1.582102692 seconds time elapsed ( +-  0.28% )

 # trace-cmd start -e all -f 'common_preempt_count==20'
 # perf stat -r 10 hackbench 50
       1.309230710 seconds time elapsed ( +-  0.22% )

 # trace-cmd start -e all -f 'common_preempt_count!=20'
 # perf stat -r 10 hackbench 50
       1.786001924 seconds time elapsed ( +-  0.20% )

The first run is again the base with no tracing.

The second run is all tracing with no filtering. It is a little slower, but
that may be well within the noise.

The third run shows that discarding all events only took 1.3 seconds. This
is a speed up of 23%! The discard is much faster than even the commit.

The one downside is shown in the last run. Events that are not discarded by
the filter will take longer to add, this is due to the extra copy of the
event.

Cc: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Currently register functions for events will be called
through the 'reg' field of event class directly without
any check when seting up triggers.

Triggers for events that don't support register through
debug fs (events under events/ftrace are for trace-cmd to
read event format, and most of them don't have a register
function except events/ftrace/functionx) can't be enabled
at all, and an oops will be hit when setting up trigger
for those events, so just not creating them is an easy way
to avoid the oops.

Link: http://lkml.kernel.org/r/1462275274-3911-1-git-send-email-chuhu@redhat.com



Cc: stable@vger.kernel.org # 3.14+
Fixes: 85f2b082 ("tracing: Add basic event trigger framework")
Signed-off-by: Chunyu Hu <chuhu@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Nothing sets TRACE_EVENT_FL_USE_CALL_FILTER anymore. Remove it.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

trace_current_buffer_lock_reserve() has no more users. Remove it.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The only user of trace_current_buffer_lock_reserve() is in the boot up self
tests. Restructure the code a little to have that code use what everything
else uses: trace_event_buffer_lock_reserve().

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

There's no real difference between trace_buffer_unlock_commit() and
trace_buffer_unlock_commit_regs() except that the former passes NULL to
ftrace_stack_trace() instead of regs. Have the former be a static inline of
the latter which passes NULL for regs.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The function trace_current_buffer_discard_commit() has no callers, remove
it.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The functions trace_buffer_unlock_commit() and the _regs() version are only
used within the kernel/trace directory. Move them to the local header and
remove the export as well.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The function filter_check_discard() is small and only called by one user,
its code can be folded into that one caller and make the code a bit less
comlplex.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Nothing outside of the tracing directory calls filter_check_discard() or
check_filter_check_discard(). They should not be called by modules. Move
their prototypes into the local tracing header and remove their
EXPORT_SYMBOL() macros.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

The functions event_trigger_unlock_commit() and
event_trigger_unlock_commit_regs() are no longer used outside the tracing
system. Move them out of the generic headers and into the local one.

Along with __event_trigger_test_discard() that is only used by them.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

In the ppc64 big endian ABI, function symbols point to function
descriptors. The symbols which point to the function entry points
have a dot in front of the function name. Consequently, when the
ftrace filter mechanism searches for the symbol corresponding to
an entry point address, it gets the dot symbol.

As a result, ftrace filter users have to be aware of this ABI detail on
ppc64 and prepend a dot to the function name when setting the filter.

The perf probe command insulates the user from this by ignoring the dot
in front of the symbol name when matching function names to symbols,
but the sysfs interface does not. This patch makes the ftrace filter
mechanism do the same when searching symbols.

Fixes the following failure in ftracetest's kprobe_ftrace.tc:

  .../kprobe_ftrace.tc: line 9: echo: write error: Invalid argument

That failure is on this line of kprobe_ftrace.tc:

  echo _do_fork > set_ftrace_filter

This is because there's no _do_fork entry in the functions list:

  # cat available_filter_functions | grep _do_fork
  ._do_fork

This change introduces no regressions on the perf and ftracetest
testsuite results.

Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.vnet.ibm.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

With the following code snippet:

    ...
    char buf[64];
    ...
    if (copy_from_user(&buf, ubuf, cnt))
    ...

Even though the value of "&buf" equals "buf", but there is no need
to get the address of the "buf" again. Use "buf" instead of "&buf".

Link: http://lkml.kernel.org/r/20160418152329.18b72bea@debian



Signed-off-by: Wang Xiaoqiang <wangxq10@lzu.edu.cn>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

If tracing_map_elt_alloc() fails, it will return ERR_PTR() instead of
NULL, so change the check to IS_ERROR().  We also need to set the
failed entry in the map->elts array to NULL instead of ERR_PTR() so
tracing_map_free_elts() doesn't try freeing an ERR_PTR().

tracing_map_free_elts() should also zero out what it frees so a
reentrant call won't find previously freed elements.

Link: http://lkml.kernel.org/r/f29d03b00bce3aac8cf151a8a30e6c83e5fee66d.1461610073.git.tom.zanussi@linux.intel.com



Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Smatch flagged create_hist_field() as possibly being able to
dereference a NULL pointer, although the current code exits in all
cases where the event field could be NULL, so it's not actually a
problem.

Still, to prevent future changes to the code from overlooking new
cases, make the NULL pointer check explicit and warn once in that
case.

Link: http://lkml.kernel.org/r/cfbc003f534a3e441b4313272fd412310aba6336.1461610073.git.tom.zanussi@linux.intel.com



Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

tracing_map_elt_alloc() returns ERR_PTRs on error, never NULL.

Fixes: 08d43a5f ('tracing: Add lock-free tracing_map')
Link: http://lkml.kernel.org/r/20160423102347.GA11136@mwanda



Acked-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

As the event-fork option requires doing work when enabled and disabled, it
can not be passed down to created instances. The instance must clear this
flag when it is created, and must clear it when its removed.

As more options may be created with this need, a macro ZEROED_TRACE_FLAGS is
created that holds the flags that must not be inherited by the top level
instance, and must be cleared on removal of instances.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

This patch adds a new helper for cls/act programs that can push events
to user space applications. For networking, this can be f.e. for sampling,
debugging, logging purposes or pushing of arbitrary wake-up events. The
idea is similar to a43eec30 ("bpf: introduce bpf_perf_event_output()
helper") and 39111695 ("samples: bpf: add bpf_perf_event_output example").

The eBPF program utilizes a perf event array map that user space populates
with fds from perf_event_open(), the eBPF program calls into the helper
f.e. as skb_event_output(skb, &my_map, BPF_F_CURRENT_CPU, raw, sizeof(raw))
so that the raw data is pushed into the fd f.e. at the map index of the
current CPU.

User space can poll/mmap/etc on this and has a data channel for receiving
events that can be post-processed. The nice thing is that since the eBPF
program and user space application making use of it are tightly coupled,
they can define their own arbitrary raw data format and what/when they
want to push.

While f.e. packet headers could be one part of the meta data that is being
pushed, this is not a substitute for things like packet sockets as whole
packet is not being pushed and push is only done in a single direction.
Intention is more of a generically usable, efficient event pipe to applications.
Workflow is that tc can pin the map and applications can attach themselves
e.g. after cls/act setup to one or multiple map slots, demuxing is done by
the eBPF program.

Adding this facility is with minimal effort, it reuses the helper
introduced in a43eec30 ("bpf: introduce bpf_perf_event_output() helper")
and we get its functionality for free by overloading its BPF_FUNC_ identifier
for cls/act programs, ctx is currently unused, but will be made use of in
future. Example will be added to iproute2's BPF example files.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add a BPF_F_CURRENT_CPU flag to optimize the use-case where user space has
per-CPU ring buffers and the eBPF program pushes the data into the current
CPU's ring buffer which saves us an extra helper function call in eBPF.
Also, make sure to properly reserve the remaining flags which are not used.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fengguang Wu's bot found two comparisons of unsigned integers to zero. These
were real bugs, as it would miss error conditions returned to zero.

trace_events_hist.c:426:6-9: WARNING: Unsigned expression compared with zero: idx < 0
trace_events_hist.c:568:5-14: WARNING: Unsigned expression compared with zero: n_entries < 0

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Allow users to have numeric fields displayed as log2 values in case
value range is very wide by appending '.log2' to field names.

For example,

  # echo 'hist:key=bytes_req' > kmalloc/trigger
  # cat kmalloc/hist

  { bytes_req:        504 } hitcount:          1
  { bytes_req:         11 } hitcount:          1
  { bytes_req:        104 } hitcount:          1
  { bytes_req:         48 } hitcount:          1
  { bytes_req:       2048 } hitcount:          1
  { bytes_req:       4096 } hitcount:          1
  { bytes_req:        240 } hitcount:          1
  { bytes_req:        392 } hitcount:          1
  { bytes_req:         13 } hitcount:          1
  { bytes_req:         28 } hitcount:          1
  { bytes_req:         12 } hitcount:          1
  { bytes_req:         64 } hitcount:          2
  { bytes_req:        128 } hitcount:          2
  { bytes_req:         32 } hitcount:          2
  { bytes_req:          8 } hitcount:         11
  { bytes_req:         10 } hitcount:         13
  { bytes_req:         24 } hitcount:         25
  { bytes_req:        160 } hitcount:         29
  { bytes_req:         16 } hitcount:         33
  { bytes_req:         80 } hitcount:         36

When using '.log2' modifier, the output looks like:

  # echo 'hist:key=bytes_req.log2' > kmalloc/trigger
  # cat kmalloc/hist

  { bytes_req: ~ 2^12 } hitcount:          1
  { bytes_req: ~ 2^11 } hitcount:          1
  { bytes_req: ~ 2^9  } hitcount:          2
  { bytes_req: ~ 2^6  } hitcount:          3
  { bytes_req: ~ 2^3  } hitcount:         13
  { bytes_req: ~ 2^5  } hitcount:         19
  { bytes_req: ~ 2^8  } hitcount:         49
  { bytes_req: ~ 2^7  } hitcount:         57
  { bytes_req: ~ 2^4  } hitcount:         74

Link: http://lkml.kernel.org/r/7ff396b246c6a881f46b979735fddf05a0d6c71a.1457029949.git.tom.zanussi@linux.intel.com



Cc: Tom Zanussi <tom.zanussi@linux.intel.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Allow users to define 'named' hist triggers.  All triggers created
with the same 'name=xxx' option will update the same shared histogram
data.

This expands the hist trigger syntax from this:

    # echo hist:keys=xxx ... [ if filter] > event/trigger

to this:

    # echo hist:name=xxx:keys=xxx ... [ if filter] > event/trigger

Named histograms must use a 'compatible' set of keys and values, which
means each event added to a set of named triggers must have the same
names and types.

Reading the 'hist' file of any of the participating events will
produce the same output as any other participating event, which is to
be expected since they share the same data.

Link: http://lkml.kernel.org/r/1dbc84ee3322a75daaf5b3ef1d0cc0a2fb682fc7.1457029949.git.tom.zanussi@linux.intel.com



Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tested-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Named triggers are sets of triggers that share a common set of trigger
data.  An example of functionality that could benefit from this type
of capability would be a set of inlined probes that would each
contribute event counts, for example, to a shared counter data
structure.

The first named trigger registered with a given name owns the common
trigger data that the others subsequently registered with the same
name will reference.  The functions defined here allow users to add,
delete, and find named triggers.

It also adds functions to pause and unpause named triggers; since
named triggers act upon common data, they should also be paused and
unpaused as a group.

Link: http://lkml.kernel.org/r/c09ff648360f65b10a3e321eddafe18060b4a04f.1457029949.git.tom.zanussi@linux.intel.com



Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tested-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

Allow users to define any number of hist triggers per trace event.
Any number of hist triggers may be added for a given event, which may
differ by key, value, or filter.

Reading the event's 'hist' file will display the output of all the
hist triggers defined on an event concatenated in the order they were
defined.

Link: http://lkml.kernel.org/r/48a0c8dd34c344571de880fb35e211c6d9a28961.1457029949.git.tom.zanussi@linux.intel.com



Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tested-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>