TRACE_EVENT is a more generic way to define tracepoints. Doing so adds
these new capabilities to this tracepoint:
- zero-copy and per-cpu splice() tracing
- binary tracing without printf overhead
- structured logging records exposed under /debug/tracing/events
- trace events embedded in function tracer output and other plugins
- user-defined, per tracepoint filter expressions
...
Cons:
- no dev_t info for the output of plug, unplug_timer and unplug_io events.
no dev_t info for getrq and sleeprq events if bio == NULL.
no dev_t info for rq_abort,...,rq_requeue events if rq->rq_disk == NULL.
This is mainly because we can't get the deivce from a request queue.
But this may change in the future.
- A packet command is converted to a string in TP_assign, not TP_print.
While blktrace do the convertion just before output.
Since pc requests should be rather rare, this is not a big issue.
- In blktrace, an event can have 2 different print formats, but a TRACE_EVENT
has a unique format, which means we have some unused data in a trace entry.
The overhead is minimized by using __dynamic_array() instead of __array().
I've benchmarked the ioctl blktrace vs the splice based TRACE_EVENT tracing:
dd dd + ioctl blktrace dd + TRACE_EVENT (splice)
1 7.36s, 42.7 MB/s 7.50s, 42.0 MB/s 7.41s, 42.5 MB/s
2 7.43s, 42.3 MB/s 7.48s, 42.1 MB/s 7.43s, 42.4 MB/s
3 7.38s, 42.6 MB/s 7.45s, 42.2 MB/s 7.41s, 42.5 MB/s
So the overhead of tracing is very small, and no regression when using
those trace events vs blktrace.
And the binary output of TRACE_EVENT is much smaller than blktrace:
# ls -l -h
-rw-r--r-- 1 root root 8.8M 06-09 13:24 sda.blktrace.0
-rw-r--r-- 1 root root 195K 06-09 13:24 sda.blktrace.1
-rw-r--r-- 1 root root 2.7M 06-09 13:25 trace_splice.out
Following are some comparisons between TRACE_EVENT and blktrace:
plug:
kjournald-480 [000] 303.084981: block_plug: [kjournald]
kjournald-480 [000] 303.084981: 8,0 P N [kjournald]
unplug_io:
kblockd/0-118 [000] 300.052973: block_unplug_io: [kblockd/0] 1
kblockd/0-118 [000] 300.052974: 8,0 U N [kblockd/0] 1
remap:
kjournald-480 [000] 303.085042: block_remap: 8,0 W 102736992 + 8 <- (8,8) 33384
kjournald-480 [000] 303.085043: 8,0 A W 102736992 + 8 <- (8,8) 33384
bio_backmerge:
kjournald-480 [000] 303.085086: block_bio_backmerge: 8,0 W 102737032 + 8 [kjournald]
kjournald-480 [000] 303.085086: 8,0 M W 102737032 + 8 [kjournald]
getrq:
kjournald-480 [000] 303.084974: block_getrq: 8,0 W 102736984 + 8 [kjournald]
kjournald-480 [000] 303.084975: 8,0 G W 102736984 + 8 [kjournald]
bash-2066 [001] 1072.953770: 8,0 G N [bash]
bash-2066 [001] 1072.953773: block_getrq: 0,0 N 0 + 0 [bash]
rq_complete:
konsole-2065 [001] 300.053184: block_rq_complete: 8,0 W () 103669040 + 16 [0]
konsole-2065 [001] 300.053191: 8,0 C W 103669040 + 16 [0]
ksoftirqd/1-7 [001] 1072.953811: 8,0 C N (5a 00 08 00 00 00 00 00 24 00) [0]
ksoftirqd/1-7 [001] 1072.953813: block_rq_complete: 0,0 N (5a 00 08 00 00 00 00 00 24 00) 0 + 0 [0]
rq_insert:
kjournald-480 [000] 303.084985: block_rq_insert: 8,0 W 0 () 102736984 + 8 [kjournald]
kjournald-480 [000] 303.084986: 8,0 I W 102736984 + 8 [kjournald]
Changelog from v2 -> v3:
- use the newly introduced __dynamic_array().
Changelog from v1 -> v2:
- use __string() instead of __array() to minimize the memory required
to store hex dump of rq->cmd().
- support large pc requests.
- add missing blk_fill_rwbs_rq() in block_rq_requeue TRACE_EVENT.
- some cleanups.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4A2DF669.5070905@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Merge reason: this topic is ready for upstream now. It passed
Oleg's review and Andrew had no further mm/*
objections/observations either.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds code that can benchmark the ring buffer as well as
test it. This code can be compiled into the kernel (not recommended)
or as a module.
A separate ring buffer is used to not interfer with other users, like
ftrace. It creates a producer and a consumer (option to disable creation
of the consumer) and will run for 10 seconds, then sleep for 10 seconds
and then repeat.
While running, the producer will write 10 byte loads into the ring
buffer with just putting in the current CPU number. The reader will
continually try to read the buffer. The reader will alternate from reading
the buffer via event by event, or by full pages.
The output is a pr_info, thus it will fill up the syslogs.
Starting ring buffer hammer
End ring buffer hammer
Time: 9000349 (usecs)
Overruns: 12578640
Read: 5358440 (by events)
Entries: 0
Total: 17937080
Missed: 0
Hit: 17937080
Entries per millisec: 1993
501 ns per entry
Sleeping for 10 secs
Starting ring buffer hammer
End ring buffer hammer
Time: 9936350 (usecs)
Overruns: 0
Read: 28146644 (by pages)
Entries: 74
Total: 28146718
Missed: 0
Hit: 28146718
Entries per millisec: 2832
353 ns per entry
Sleeping for 10 secs
Time: is the time the test ran
Overruns: the number of events that were overwritten and not read
Read: the number of events read (either by pages or events)
Entries: the number of entries left in the buffer
(the by pages will only read full pages)
Total: Entries + Read + Overruns
Missed: the number of entries that failed to write
Hit: the number of entries that were written
The above example shows that it takes ~353 nanosecs per entry when
there is a reader, reading by pages (and no overruns)
The event by event reader slowed the producer down to 501 nanosecs.
[ Impact: see how changes to the ring buffer affect stability and performance ]
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
This patch moves the ftrace creation into include/trace/ftrace.h and
simplifies the work of developers in adding new tracepoints.
Just the act of creating the trace points in include/trace and including
define_trace.h will create the events in the debugfs/tracing/events
directory.
This patch removes the need of include/trace/trace_events.h
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Jeremy Fitzhardinge reported this build failure:
LD .tmp_vmlinux1
arch/x86/kernel/built-in.o: In function `ds_take_timestamp':
git/linux/arch/x86/kernel/ds.c:1380: undefined reference to `trace_clock_global'
git/linux/arch/x86/kernel/ds.c:1380: undefined reference to `trace_clock_global'
Which is due to !CONFIG_TRACING && CONFIG_X86_DS=y.
Expose the trace clock code to CONFIG_X86_DS as well.
[ Unfortunately librarizing doesnt work well - ancient architectures
with no raw_local_irq_save() primitive break the build. ]
Reported-by: Jeremy Fitzhardinge <jeremy@goop.org>
LKML-Reference: <49E4413F.7070700@goop.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add a new config option, CONFIG_EVENT_TRACING that gets selected
when CONFIG_TRACING is selected and adds everything needed by the stuff
in trace_export - basically all the event tracing support needed by e.g.
bprint, minus the actual events, which are only included if
CONFIG_EVENT_TRACER is selected.
So CONFIG_EVENT_TRACER can be used to turn on or off the generated events
(what I think of as the 'event tracer'), while CONFIG_EVENT_TRACING turns
on or off the base event tracing support used by both the event tracer and
the other things such as bprint that can't be configured out.
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: fweisbec@gmail.com
LKML-Reference: <1239178441.10295.34.camel@tropicana>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds per-event filtering to the event tracing subsystem.
It adds a 'filter' debugfs file to each event directory. This file can
be written to to set filters; reading from it will display the current
set of filters set for that event.
Basically, any field listed in the 'format' file for an event can be
filtered on (including strings, but not yet other array types) using
either matching ('==') or non-matching ('!=') 'predicates'. A
'predicate' can be either a single expression:
# echo pid != 0 > filter
# cat filter
pid != 0
or a compound expression of up to 8 sub-expressions combined using '&&'
or '||':
# echo comm == Xorg > filter
# echo "&& sig != 29" > filter
# cat filter
comm == Xorg
&& sig != 29
Only events having field values matching an expression will be available
in the trace output; non-matching events are discarded.
Note that a compound expression is built up by echoing each
sub-expression separately - it's not the most efficient way to do
things, but it keeps the parser simple and assumes that compound
expressions will be relatively uncommon. In any case, a subsequent
patch introducing a way to set filters for entire subsystems should
mitigate any need to do this for lots of events.
Setting a filter without an '&&' or '||' clears the previous filter
completely and sets the filter to the new expression:
# cat filter
comm == Xorg
&& sig != 29
# echo comm != Xorg
# cat filter
comm != Xorg
To clear a filter, echo 0 to the filter file:
# echo 0 > filter
# cat filter
none
The limit of 8 predicates for a compound expression is arbitrary - for
efficiency, it's implemented as an array of pointers to predicates, and
8 seemed more than enough for any filter...
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <1237710665.7703.48.camel@charm-linux>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: faster and lighter tracing
Now that we have trace_bprintk() which is faster and consume lesser
memory than trace_printk() and has the same purpose, we can now drop
the old implementation in favour of the binary one from trace_bprintk(),
which means we move all the implementation of trace_bprintk() to
trace_printk(), so the Api doesn't change except that we must now use
trace_seq_bprintk() to print the TRACE_PRINT entries.
Some changes result of this:
- Previously, trace_bprintk depended of a single tracer and couldn't
work without. This tracer has been dropped and the whole implementation
of trace_printk() (like the module formats management) is now integrated
in the tracing core (comes with CONFIG_TRACING), though we keep the file
trace_printk (previously trace_bprintk.c) where we can find the module
management. Thus we don't overflow trace.c
- changes some parts to use trace_seq_bprintk() to print TRACE_PRINT entries.
- change a bit trace_printk/trace_vprintk macros to support non-builtin formats
constants, and fix 'const' qualifiers warnings. But this is all transparent for
developers.
- etc...
V2:
- Rebase against last changes
- Fix mispell on the changelog
V3:
- Rebase against last changes (moving trace_printk() to kernel.h)
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <1236356510-8381-5-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: save on memory for tracing
Current tracers are typically using a struct(like struct ftrace_entry,
struct ctx_switch_entry, struct special_entr etc...)to record a binary
event. These structs can only record a their own kind of events.
A new kind of tracer need a new struct and a lot of code too handle it.
So we need a generic binary record for events. This infrastructure
is for this purpose.
[fweisbec@gmail.com: rebase against latest -tip, make it safe while sched
tracing as reported by Steven Rostedt]
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <1236356510-8381-3-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: allow user apps to read binary format of basic ftrace entries
Currently, only defined raw events export their formats so a binary
reader can parse them. There's no reason that the default ftrace entries
can't export their formats.
This patch adds a subsystem called "ftrace" in the events directory
that includes the ftrace entries for basic ftrace recorded items.
These only have three files in the events directory:
type : printf
available_types : printf
format : format for the event entry
For example:
# cat /debug/tracing/events/ftrace/wakeup/format
name: wakeup
ID: 3
format:
field:unsigned char type; offset:0; size:1;
field:unsigned char flags; offset:1; size:1;
field:unsigned char preempt_count; offset:2; size:1;
field:int pid; offset:4; size:4;
field:int tgid; offset:8; size:4;
field:unsigned int prev_pid; offset:12; size:4;
field:unsigned char prev_prio; offset:16; size:1;
field:unsigned char prev_state; offset:17; size:1;
field:unsigned int next_pid; offset:20; size:4;
field:unsigned char next_prio; offset:24; size:1;
field:unsigned char next_state; offset:25; size:1;
field:unsigned int next_cpu; offset:28; size:4;
print fmt: "%u:%u:%u ==+ %u:%u:%u [%03u]"
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Impact: implement new tracing timestamp APIs
Add three trace clock variants, with differing scalability/precision
tradeoffs:
- local: CPU-local trace clock
- medium: scalable global clock with some jitter
- global: globally monotonic, serialized clock
Make the ring-buffer use the local trace clock internally.
Acked-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch changes the trace/sched.h to use the DECLARE_TRACE_FMT
such that they are automatically registered with the event tracer.
And it also adds the tracing sched headers to kernel/trace/events.c
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
This patch creates the event tracing infrastructure of ftrace.
It will create the files:
/debug/tracing/available_events
/debug/tracing/set_event
The available_events will list the trace points that have been
registered with the event tracer.
set_events will allow the user to enable or disable an event hook.
example:
# echo sched_wakeup > /debug/tracing/set_event
Will enable the sched_wakeup event (if it is registered).
# echo "!sched_wakeup" >> /debug/tracing/set_event
Will disable the sched_wakeup event (and only that event).
# echo > /debug/tracing/set_event
Will disable all events (notice the '>')
# cat /debug/tracing/available_events > /debug/tracing/set_event
Will enable all registered event hooks.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Impact: new tracer
The workqueue tracer provides some statistical informations
about each cpu workqueue thread such as the number of the
works inserted and executed since their creation. It can help
to evaluate the amount of work each of them have to perform.
For example it can help a developer to decide whether he should
choose a per cpu workqueue instead of a singlethreaded one.
It only traces statistical informations for now but it will probably later
provide event tracing too.
Such a tracer could help too, and be improved, to help rt priority sorted
workqueue development.
To have a snapshot of the workqueues state at any time, just do
cat /debugfs/tracing/trace_stat/workqueues
Ie:
1 125 125 reiserfs/1
1 0 0 scsi_tgtd/1
1 0 0 aio/1
1 0 0 ata/1
1 114 114 kblockd/1
1 0 0 kintegrityd/1
1 2147 2147 events/1
0 0 0 kpsmoused
0 105 105 reiserfs/0
0 0 0 scsi_tgtd/0
0 0 0 aio/0
0 0 0 ata_aux
0 0 0 ata/0
0 0 0 cqueue
0 0 0 kacpi_notify
0 0 0 kacpid
0 149 149 kblockd/0
0 0 0 kintegrityd/0
0 1000 1000 khelper
0 2270 2270 events/0
Changes in V2:
_ Drop the static array based on NR_CPU and dynamically allocate the stat array
with num_possible_cpus() and other cpu mask facilities....
_ Trace workqueue insertion at a bit lower level (insert_work instead of queue_work) to handle
even the workqueue barriers.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: new tracer plugin
This patch adapts kmemtrace raw events tracing to the unified tracing API.
To enable and use this tracer, just do the following:
echo kmemtrace > /debugfs/tracing/current_tracer
cat /debugfs/tracing/trace
You will have the following output:
# tracer: kmemtrace
#
#
# ALLOC TYPE REQ GIVEN FLAGS POINTER NODE CALLER
# FREE | | | | | | | |
# |
type_id 1 call_site 18446744071565527833 ptr 18446612134395152256
type_id 0 call_site 18446744071565585597 ptr 18446612134405955584 bytes_req 4096 bytes_alloc 4096 gfp_flags 208 node -1
type_id 1 call_site 18446744071565585534 ptr 18446612134405955584
type_id 0 call_site 18446744071565585597 ptr 18446612134405955584 bytes_req 4096 bytes_alloc 4096 gfp_flags 208 node -1
type_id 0 call_site 18446744071565636711 ptr 18446612134345164672 bytes_req 240 bytes_alloc 240 gfp_flags 208 node -1
type_id 1 call_site 18446744071565585534 ptr 18446612134405955584
type_id 0 call_site 18446744071565585597 ptr 18446612134405955584 bytes_req 4096 bytes_alloc 4096 gfp_flags 208 node -1
type_id 0 call_site 18446744071565636711 ptr 18446612134345164912 bytes_req 240 bytes_alloc 240 gfp_flags 208 node -1
type_id 1 call_site 18446744071565585534 ptr 18446612134405955584
type_id 0 call_site 18446744071565585597 ptr 18446612134405955584 bytes_req 4096 bytes_alloc 4096 gfp_flags 208 node -1
type_id 0 call_site 18446744071565636711 ptr 18446612134345165152 bytes_req 240 bytes_alloc 240 gfp_flags 208 node -1
type_id 0 call_site 18446744071566144042 ptr 18446612134346191680 bytes_req 1304 bytes_alloc 1312 gfp_flags 208 node -1
type_id 1 call_site 18446744071565585534 ptr 18446612134405955584
type_id 0 call_site 18446744071565585597 ptr 18446612134405955584 bytes_req 4096 bytes_alloc 4096 gfp_flags 208 node -1
type_id 1 call_site 18446744071565585534 ptr 18446612134405955584
That was to stay backward compatible with the format output produced in
inux/tracepoint.h.
This is the default ouput, but note that I tried something else.
If you change an option:
echo kmem_minimalistic > /debugfs/trace_options
and then cat /debugfs/trace, you will have the following output:
# tracer: kmemtrace
#
#
# ALLOC TYPE REQ GIVEN FLAGS POINTER NODE CALLER
# FREE | | | | | | | |
# |
- C 0xffff88007c088780 file_free_rcu
+ K 4096 4096 000000d0 0xffff88007cad6000 -1 getname
- C 0xffff88007cad6000 putname
+ K 4096 4096 000000d0 0xffff88007cad6000 -1 getname
+ K 240 240 000000d0 0xffff8800790dc780 -1 d_alloc
- C 0xffff88007cad6000 putname
+ K 4096 4096 000000d0 0xffff88007cad6000 -1 getname
+ K 240 240 000000d0 0xffff8800790dc870 -1 d_alloc
- C 0xffff88007cad6000 putname
+ K 4096 4096 000000d0 0xffff88007cad6000 -1 getname
+ K 240 240 000000d0 0xffff8800790dc960 -1 d_alloc
+ K 1304 1312 000000d0 0xffff8800791d7340 -1 reiserfs_alloc_inode
- C 0xffff88007cad6000 putname
+ K 4096 4096 000000d0 0xffff88007cad6000 -1 getname
- C 0xffff88007cad6000 putname
+ K 992 1000 000000d0 0xffff880079045b58 -1 alloc_inode
+ K 768 1024 000080d0 0xffff88007c096400 -1 alloc_pipe_info
+ K 240 240 000000d0 0xffff8800790dca50 -1 d_alloc
+ K 272 320 000080d0 0xffff88007c088780 -1 get_empty_filp
+ K 272 320 000080d0 0xffff88007c088000 -1 get_empty_filp
Yeah I shall confess kmem_minimalistic should be: kmem_alternative.
Whatever, I find it more readable but this a personal opinion of course.
We can drop it if you want.
On the ALLOC/FREE column, + means an allocation and - a free.
On the type column, you have K = kmalloc, C = cache, P = page
I would like the flags to be GFP_* strings but that would not be easy to not
break the column with strings....
About the node...it seems to always be -1. I don't know why but that shouldn't
be difficult to find.
I moved linux/tracepoint.h to trace/tracepoint.h as well. I think that would
be more easy to find the tracer headers if they are all in their common
directory.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: extend the tracing API
The goal of this patch is to normalize and make more easy the
implementation of statistical (histogram) tracing.
It implements a trace_stat file into the /debugfs/tracing directory where
one can print a one-shot output of statistics/histogram entries.
A tracer has to provide two basic iterator callbacks:
stat_start() => the first entry
stat_next(prev, idx) => the next one.
Note that it is adapted for arrays or hash tables or lists.... since it
provides a pointer to the previous entry and the current index of the
iterator.
These two callbacks are called to get a snapshot of the statistics at each
opening of the trace_stat file because. The values are so updated between
two "cat trace_stat". And the tracer is free to lock its datas during the
iteration to keep consistent values.
Since it is almost always interesting to sort statisticals values to
address the problems by priority, this infrastructure provides a "sorting"
of the stat entries too if desired. A tracer has just to provide a
stat_cmp callback to compare two entries and the stat tracing
infrastructure will build a sorted list of the given entries.
A last callback, called stat_headers, can be implemented by a tracer to
output headers on its trace.
If one of these callbacks is changed on runtime, it just have to signal it
to the stat tracing API by calling the init_tracer_stat() helper.
Changes in V2:
- Fix a memory leak if the user opens multiple times the trace_stat file
without closing it. Now we always free our list before rebuilding it.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: simplify/generalize/refactor trace.c
The trace.c file is becoming more difficult to maintain due to the
growing number of events. There is several formats that an event may
be printed. This patch sets up the infrastructure of an event hash to
allow for events to register how they should be printed.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: restructure code, cleanup
Remove BTS bits from the hw-branch-tracer (renamed from bts-tracer) and
use the ds interface.
Signed-off-by: Markus Metzger <markut.t.metzger@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: new "power-tracer" ftrace plugin
This patch adds a C/P-state ftrace plugin that will generate
detailed statistics about the C/P-states that are being used,
so that we can look at detailed decisions that the C/P-state
code is making, rather than the too high level "average"
that we have today.
An example way of using this is:
mount -t debugfs none /sys/kernel/debug
echo cstate > /sys/kernel/debug/tracing/current_tracer
echo 1 > /sys/kernel/debug/tracing/tracing_enabled
sleep 1
echo 0 > /sys/kernel/debug/tracing/tracing_enabled
cat /sys/kernel/debug/tracing/trace | perl scripts/trace/cstate.pl > out.svg
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: cleanup
This patch changes the name of the "return function tracer" into
function-graph-tracer which is a more suitable name for a tracing
which makes one able to retrieve the ordered call stack during
the code flow.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
