perf hists: Fix recalculation of total_period when sorting entries
We were doing parts of it in hists__collapse_resort and parts of it in
hists__output_resort, leading to a bogus total_period.
Fix it by doing just the filtering operation when collapsing because
there we know that the Zoom operations adds filters just what is in
hists->entries, not to the new batch of entries being collapsed.
And move all the nr_entries + total_period recalculation to
hists__output_resort since we will traverse all entries anyway there.
Problem introduced when developing threaded addition of new batches
of hist_entries, i.e. post v3.1.
Reported-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-8xyh165h7hmwy0696hu25en6@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
CC /home/acme/git/build/perf/util/header.o
cc1: warnings being treated as errors
util/header.c: In function ‘write_cpudesc’:
util/header.c:281: warning: implicit declaration of function ‘getline’
util/header.c:281: warning: nested extern declaration of ‘getline’
make: *** [/home/acme/git/build/perf/util/header.o] Error 1
make: Leaving directory `/home/acme/git/linux/tools/perf'
[acme@localhost linux]$
This happens due to header ordering, in perf util.h sets _GNU_SOURCE, so
it must come first.
Reported-by: Ingo Molnar <mingo@elte.hu> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-alfra9wao63euguj7gr8jw7e@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf ui: Reimplement ui__popup_menu using ui__browser
Right now let it work just like the other browsers: in full screen, at
the top left corner. If people complain we can revisit, I found it OK
and the laziest/quickest approach at reusing the ui_browser ;-)
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-4bgeqizcxh04q0sk24cw43gk@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The __perf_evsel__open routing was grouping just the threads for that
specific events per cpu when we want to group all threads in all events
to the first fd opened on that cpu.
So pass the xyarray with the first event, where the other events will be
able to get that first per cpu fd.
At some point top and record will switch to using perf_evlist__open that
takes care of this detail and probably will also handle the fallback
from hw to soft counters, etc.
Reported-by: Deng-Cheng Zhu <dczhu@mips.com> Tested-by: Deng-Cheng Zhu <dczhu@mips.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-ebm34rh098i9y9v4cytfdp0x@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists browser: Refuse 'a' hotkey on non symbolic views
We don't allocate the histogram data structures for --sort lists without
"sym", so, just like was done for the menu, don't try to annotate when
'a' is pressed, just warn the user about it.
Reported-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-27mjg02s2mbw8lfxqv7jpzec@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Olsa [Thu, 20 Oct 2011 13:59:43 +0000 (15:59 +0200)]
perf tools: Fix tracing info recording
Fixing the way the tracing information is stored within record command.
The current implementation is causing issues for pipe output.
Following commands fail currently:
perf script syscall-counts ls
perf record -e syscalls:sys_exit_read ls | ./perf report -i -
The tracing information is part of the perf data file. It contains
several files from within the tracing debugfs and procs directories.
Beside some static header files, for each tracing event the format
file is added. The /proc/kallsyms file is also added.
The tracing data are stored with preceeding size. This is causing some
dificulties for pipe output, since there's no way to tell debugfs/proc
file size before reading it. So, for pipe output, all the debugfs files
were read twice. Once to get the overall size and once to store the
content itself. This can cause problem in case any of these file
changed, within the storage time.
To fix this behaviour and ensure the integrity of the tracing data, we:
- read debugfs/proc file into the temp file
- get temp file size and dump it to the pipe
- dump the temp file contents to the pipe
Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Neil Horman <nhorman@tuxdriver.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/r/20111020135943.GD2092@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists: Don't decay total_period for filtered entries
Following the 'perf report' model we don't zap hist_entry instances from
the rb tree, we just keep them with he->filtered set to a mask of the
filters applied to it (thread, parent, DSO so far).
In top we need to decay even filtered entries, but we better not touch
total_period for them...
Now everything seems to work when filters are applied on top as they
worked in 'report', i.e. both dynamic and static hist entry browsing
works with filters.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-yt4xsbq20u9x9ypuwwyw2kao@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
David Ahern [Wed, 19 Oct 2011 17:37:47 +0000 (11:37 -0600)]
perf hists browser: Do not exit on tab key with single event
TUI help states for multiple event sessions the TAB/UNTAB keys are used
to switch events. For single event sessions (e.g., the default) the tab
key currently causes the tui to exit. Change that to do nothing since
there is not no second event to switch to.
Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1319045867-12728-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
David Ahern [Tue, 18 Oct 2011 23:34:01 +0000 (17:34 -0600)]
perf tools: handle endianness of feature bitmap
Feature bitmap is declared as an array of unsigned longs -- not good
since its size can differ between the host that generated the data file
and the host analyzing the file.
We need to handle endianness, but we don't know the size of the unsigned
long where the file was generated. Take a best guess at determining it:
try 64-bit swap first (ie., file created on a 64-bit host), and check if
the hostname feature bit is set (this feature bit is forced on as of fbe96f2). If the bit is not, undo the 64-bit swap and try a 32-bit
swap. If the hostname bit is still not set (e.g., older data file), punt
and fallback to the original behavior -- clearing all feature bits and
setting buildid.
Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1318980841-12616-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
David Ahern [Wed, 19 Oct 2011 00:44:45 +0000 (18:44 -0600)]
perf tools: Add prelink suggestion to dso update message
Following a prelink run mapped files for long running processes can show
as deleted. The current message suggests restarting long running
processes. Add to that a suggestion that prelink might be the cause.
Old message:
/lib64/libc-2.14.so was updated, restart the long running
apps that use it!
New message:
/lib64/libc-2.14.so was updated (is prelink enabled?).
Restart the long running apps that use it!
Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1318985085-20776-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists browser: Apply the dso and thread filters when merging new batches
Now that we dynamicly add entries on the timer we need to not only
traverse all entries when the user zooms into threads and/or DSOs, but
as well after that apply it to the new batches of hist entries in
hists__collapse_resort.
Reported-by: Mike Galbraith <efault@gmx.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-zustn633c7hnrae94x6nld1p@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists: Move the dso and thread filters from hist_browser
Since with dynamic addition of new hist entries we need to apply those
filters as we merge new batches of hist_entry instances, for instance in
perf top.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-zjhhf8kh9w1buty9p10od6rz@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists: Don't format the percentage on hist_entry__snprintf
We can't have color correctly set there because in libslang (and in a future
GUI) the colors must be set on a separate function call, so move that part to a
separate function and make the stdio fprintf function call it.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-jpgy42438ce9tgbqppm397lq@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Josh Stone [Tue, 18 Oct 2011 01:00:45 +0000 (18:00 -0700)]
x86, perf, kprobes: Make kprobes's twobyte_is_boostable volatile
When compiling an i386_defconfig kernel with
gcc-4.6.1-9.fc15.i686, I noticed a warning about the asm operand
for test_bit in kprobes' can_boost. I discovered that this
caused only the first long of twobyte_is_boostable[] to be
output.
Jakub filed and fixed gcc PR50571 to correct the warning and
this output issue. But to solve it for less current gcc, we can
make kprobes' twobyte_is_boostable[] volatile, and it won't be
optimized out.
Before:
CC arch/x86/kernel/kprobes.o
In file included from include/linux/bitops.h:22:0,
from include/linux/kernel.h:17,
from [...]/arch/x86/include/asm/percpu.h:44,
from [...]/arch/x86/include/asm/current.h:5,
from [...]/arch/x86/include/asm/processor.h:15,
from [...]/arch/x86/include/asm/atomic.h:6,
from include/linux/atomic.h:4,
from include/linux/mutex.h:18,
from include/linux/notifier.h:13,
from include/linux/kprobes.h:34,
from arch/x86/kernel/kprobes.c:43:
[...]/arch/x86/include/asm/bitops.h: In function ‘can_boost.part.1’:
[...]/arch/x86/include/asm/bitops.h:319:2: warning: use of memory input without lvalue in asm operand 1 is deprecated [enabled by default]
Mike Galbraith [Mon, 17 Oct 2011 11:14:58 +0000 (09:14 -0200)]
perf hists browser: Add missing stdarg.h include
CC util/ui/browsers/annotate.o
In file included from util/ui/browsers/annotate.c:2:0:
util/ui/browsers/../helpline.h:9:42: error: expected declaration
specifiers or ‘...’ before ‘va_list’
CC util/ui/browsers/hists.o
make: *** [util/ui/browsers/annotate.o] Error 1
make: *** Waiting for unfinished jobs....
perf top: Fix the 'E' hotkey, select among multiple events
We were not recognizing 'E' as a hotkey due to a bug introduced when
switching to the new, hist_entry based top. Fix it by returning that 'E'
is mapped if evlist->nr_entries > 1.
Reported-by: Mike Galbraith <efault@gmx.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-zcx055vnhagddvqlaqxvdhtb@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Its becoming common to allow the user to filter out parts of the data
structure being browsed, like already done in the hists browser and in
the annotate browser in the next commit, so provide it directly in the
ui_browser class list_head helpers.
More work required to move the equivalent routines found now in the
hists browser to the rb_tree helpers.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-jk7danyt1d9ji4e3o2xuthpn@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists browser: Invalidate ui_browser->top after timer calls
With underlying dynamic data structures we need to invalidate pointers
to them after a timer, as that entry may have vanished (decayed in top,
for instance).
I forgot about browser_ui->top. Fix it by resetting it to null after a
timer. The seek operation from SEEK_SET will then set it to a valid
entry because it starts from rb_first(&hists->entries).
Reported-by: Ingo Molnar <mingo@elte.hu> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-2ssjm0ouh9tsz4dwkcu7c40n@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Steven Rostedt [Fri, 14 Oct 2011 14:44:25 +0000 (10:44 -0400)]
tracing: Fix returning of duplicate data after EOF in trace_pipe_raw
The trace_pipe_raw handler holds a cached page from the time the file
is opened to the time it is closed. The cached page is used to handle
the case of the user space buffer being smaller than what was read from
the ring buffer. The left over buffer is held in the cache so that the
next read will continue where the data left off.
After EOF is returned (no more data in the buffer), the index of
the cached page is set to zero. If a user app reads the page again
after EOF, the check in the buffer will see that the cached page
is less than page size and will return the cached page again. This
will cause reading the trace_pipe_raw again after EOF to return
duplicate data, making the output look like the time went backwards
but instead data is just repeated.
The fix is to not reset the index right after all data is read
from the cache, but to reset it after all data is read and more
data exists in the ring buffer.
Cc: stable <stable@kernel.org> Reported-by: Jeremy Eder <jeder@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Geunsik Lim [Fri, 12 Aug 2011 05:30:22 +0000 (14:30 +0900)]
ftrace: Fix README to state tracing_on to start/stop tracing
tracing_enabled option is deprecated.
To start/stop tracing, write to /sys/kernel/debug/tracing/tracing_on
without tracing_enabled. This patch is based on Linux 3.1.0-rc1
perf hists browser: Fix handling of TAB/UNTAB for multiple events
When using multiple events the 'top' and 'report' tools will first
present the user with a menu to choose the event to browse.
After that the user can either press <- to go back to the menu and
choose another event or instead press TAB to go the next event without
having to go back to the menu or shift-TAB (UNTAB) to go the previous
event, useful to quickly visually see if multiple events are correlated.
The handling of each hists browser return was broken by the ed7e566,
that combined both switches, the first that was for choosing the event
and the second that was for checking if switching to the next event
without passing thru the events menu.
Repeat with me: Don't be clever like that.
Fix it by moving the switch to right after the call to the hists
browser, making abundantly clear that the two switches are unrelated.
This also fixes a compiler warning about the 'pos' variable being
possibly used unitialized.
Reported-by: Ingo Molnar <mingo@elte.hu>
[ committer note: the line above is for the compiler warning ] Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-ujxkbvj9vy8w6xe2op5m51tb@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf top: Remove entries from entries_collapsed on decay
We were removing only when using a --sort order that needs collapsing,
while we also use it in the threaded case, causing memory corruption
because we were scribbling freed hist entries, oops.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-k16fb4jsulr7x0ixv43amb6d@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists browser: Recalculate browser pointers after resort/decay
In browsers that access dynamic underlying data structures, like in the
hists browser and its hist_entry rb_tree, we need to revalidate any
reference to the underlying data structure, because they can have gone
away, decayed.
This fixes a problem where after a while the top entries get behind the
top of the screen, i.e. the top_idx stays at 0, which means it is at the
first entry in the rb_tree when in fact it wasn't because the
browser->top didn't got revalidated after the timer ran and the
underlying data structure got updated.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-mhje66qssdko24q67a2lhlho@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Steven Rostedt [Thu, 22 Sep 2011 15:50:27 +0000 (11:50 -0400)]
tracing: Do not allocate buffer for trace_marker
When doing intense tracing, the kmalloc inside trace_marker can
introduce side effects to what is being traced.
As trace_marker() is used by userspace to inject data into the
kernel ring buffer, it needs to do so with the least amount
of intrusion to the operations of the kernel or the user space
application.
As the ring buffer is designed to write directly into the buffer
without the need to make a temporary buffer, and userspace already
went through the hassle of knowing how big the write will be,
we can simply pin the userspace pages and write the data directly
into the buffer. This improves the impact of tracing via trace_marker
tremendously!
Thanks to Peter Zijlstra and Thomas Gleixner for pointing out the
use of get_user_pages_fast() and kmap_atomic().
Suggested-by: Thomas Gleixner <tglx@linutronix.de> Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Fri, 30 Sep 2011 01:26:16 +0000 (21:26 -0400)]
tracing: Warn on output if the function tracer was found corrupted
As the function tracer is very intrusive, lots of self checks are
performed on the tracer and if something is found to be strange
it will shut itself down keeping it from corrupting the rest of the
kernel. This shutdown may still allow functions to be traced, as the
tracing only stops new modifications from happening. Trying to stop
the function tracer itself can cause more harm as it requires code
modification.
Although a WARN_ON() is executed, a user may not notice it. To help
the user see that something isn't right with the tracing of the system
a big warning is added to the output of the tracer that lets the user
know that their data may be incomplete.
Reported-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
I've also checked that this problem doesn't happen on
tracepoints when module removing because perf event
locks target module.
$ sudo ./perf record -e xfs:\* -aR sh
sh-4.2# rmmod xfs
ERROR: Module xfs is in use
sh-4.2# exit
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.203 MB perf.data (~8862 samples) ]
Ingo Molnar [Thu, 6 Oct 2011 12:20:27 +0000 (14:20 +0200)]
x86, nmi, drivers: Fix nmi splitup build bug
nmi.c needs an #include <linux/mca.h>:
arch/x86/kernel/nmi.c: In function ‘unknown_nmi_error’:
arch/x86/kernel/nmi.c:286:6: error: ‘MCA_bus’ undeclared (first use in this function)
arch/x86/kernel/nmi.c:286:6: note: each undeclared identifier is reported only once for each function it appears in
Another one is the hpwdt driver:
drivers/watchdog/hpwdt.c:507:9: error: ‘NMI_DONE’ undeclared (first use in this function)
Robert Richter [Wed, 21 Sep 2011 09:30:18 +0000 (11:30 +0200)]
perf, x86: Implement IBS initialization
This patch implements IBS feature detection and initialzation. The
code is shared between perf and oprofile. If IBS is available on the
system for perf, a pmu is setup.
Don Zickus [Fri, 30 Sep 2011 19:06:23 +0000 (15:06 -0400)]
x86, nmi: Track NMI usage stats
Now that the NMI handler are broken into lists, increment the appropriate
stats for each list. This allows us to see what is going on when they
get printed out in the next patch.
Don Zickus [Fri, 30 Sep 2011 19:06:22 +0000 (15:06 -0400)]
x86, nmi: Add in logic to handle multiple events and unknown NMIs
Previous patches allow the NMI subsystem to process multipe NMI events
in one NMI. As previously discussed this can cause issues when an event
triggered another NMI but is processed in the current NMI. This causes the
next NMI to go unprocessed and become an 'unknown' NMI.
To handle this, we first have to flag whether or not the NMI handler handled
more than one event or not. If it did, then there exists a chance that
the next NMI might be already processed. Once the NMI is flagged as a
candidate to be swallowed, we next look for a back-to-back NMI condition.
This is determined by looking at the %rip from pt_regs. If it is the same
as the previous NMI, it is assumed the cpu did not have a chance to jump
back into a non-NMI context and execute code and instead handled another NMI.
If both of those conditions are true then we will swallow any unknown NMI.
There still exists a chance that we accidentally swallow a real unknown NMI,
but for now things seem better.
An optimization has also been added to the nmi notifier rountine. Because x86
can latch up to one NMI while currently processing an NMI, we don't have to
worry about executing _all_ the handlers in a standalone NMI. The idea is
if multiple NMIs come in, the second NMI will represent them. For those
back-to-back NMI cases, we have the potentail to drop NMIs. Therefore only
execute all the handlers in the second half of a detected back-to-back NMI.
Don Zickus [Fri, 30 Sep 2011 19:06:21 +0000 (15:06 -0400)]
x86, nmi: Wire up NMI handlers to new routines
Just convert all the files that have an nmi handler to the new routines.
Most of it is straight forward conversion. A couple of places needed some
tweaking like kgdb which separates the debug notifier from the nmi handler
and mce removes a call to notify_die.
[Thanks to Ying for finding out the history behind that mce call
https://lkml.org/lkml/2010/5/27/114
And Boris responding that he would like to remove that call because of it
https://lkml.org/lkml/2011/9/21/163]
The things that get converted are the registeration/unregistration routines
and the nmi handler itself has its args changed along with code removal
to check which list it is on (most are on one NMI list except for kgdb
which has both an NMI routine and an NMI Unknown routine).
Signed-off-by: Don Zickus <dzickus@redhat.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Corey Minyard <minyard@acm.org> Cc: Jason Wessel <jason.wessel@windriver.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Robert Richter <robert.richter@amd.com> Cc: Huang Ying <ying.huang@intel.com> Cc: Corey Minyard <minyard@acm.org> Cc: Jack Steiner <steiner@sgi.com> Link: http://lkml.kernel.org/r/1317409584-23662-4-git-send-email-dzickus@redhat.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
Don Zickus [Fri, 30 Sep 2011 19:06:20 +0000 (15:06 -0400)]
x86, nmi: Create new NMI handler routines
The NMI handlers used to rely on the notifier infrastructure. This worked
great until we wanted to support handling multiple events better.
One of the key ideas to the nmi handling is to process _all_ the handlers for
each NMI. The reason behind this switch is because NMIs are edge triggered.
If enough NMIs are triggered, then they could be lost because the cpu can
only latch at most one NMI (besides the one currently being processed).
In order to deal with this we have decided to process all the NMI handlers
for each NMI. This allows the handlers to determine if they recieved an
event or not (the ones that can not determine this will be left to fend
for themselves on the unknown NMI list).
As a result of this change it is now possible to have an extra NMI that
was destined to be received for an already processed event. Because the
event was processed in the previous NMI, this NMI gets dropped and becomes
an 'unknown' NMI. This of course will cause printks that scare people.
However, we prefer to have extra NMIs as opposed to losing NMIs and as such
are have developed a basic mechanism to catch most of them. That will be
a later patch.
To accomplish this idea, I unhooked the nmi handlers from the notifier
routines and created a new mechanism loosely based on doIRQ. The reason
for this is the notifier routines have a couple of shortcomings. One we
could't guarantee all future NMI handlers used NOTIFY_OK instead of
NOTIFY_STOP. Second, we couldn't keep track of the number of events being
handled in each routine (most only handle one, perf can handle more than one).
Third, I wanted to eventually display which nmi handlers are registered in
the system in /proc/interrupts to help see who is generating NMIs.
The patch below just implements the new infrastructure but doesn't wire it up
yet (that is the next patch). Its design is based on doIRQ structs and the
atomic notifier routines. So the rcu stuff in the patch isn't entirely untested
(as the notifier routines have soaked it) but it should be double checked in
case I copied the code wrong.
Gleb Natapov [Wed, 5 Oct 2011 12:01:21 +0000 (14:01 +0200)]
perf, intel: Use GO/HO bits in perf-ctr
Intel does not have guest/host-only bit in perf counters like AMD
does. To support GO/HO bits KVM needs to switch EVENTSELn values
(or PERF_GLOBAL_CTRL if available) at a guest entry. If a counter is
configured to count only in a guest mode it stays disabled in a host,
but VMX is configured to switch it to enabled value during guest entry.
This patch adds GO/HO tracking to Intel perf code and provides interface
for KVM to get a list of MSRs that need to be switched on a guest entry.
Only cpus with architectural PMU (v1 or later) are supported with this
patch. To my knowledge there is not p6 models with VMX but without
architectural PMU and p4 with VMX are rare and the interface is general
enough to support them if need arise.
perf tools: Make perf.data more self-descriptive (v8)
The goal of this patch is to include more information about the host
environment into the perf.data so it is more self-descriptive. Overtime,
profiles are captured on various machines and it becomes hard to track
what was recorded, on what machine and when.
This patch provides a way to solve this by extending the perf.data file
with basic information about the host machine. To add those extensions,
we leverage the feature bits capabilities of the perf.data format. The
change is backward compatible with existing perf.data files.
We define the following useful new extensions:
- HEADER_HOSTNAME: the hostname
- HEADER_OSRELEASE: the kernel release number
- HEADER_ARCH: the hw architecture
- HEADER_CPUDESC: generic CPU description
- HEADER_NRCPUS: number of online/avail cpus
- HEADER_CMDLINE: perf command line
- HEADER_VERSION: perf version
- HEADER_TOPOLOGY: cpu topology
- HEADER_EVENT_DESC: full event description (attrs)
- HEADER_CPUID: easy-to-parse low level CPU identication
The small granularity for the entries is to make it easier to extend
without breaking backward compatiblity. Many entries are provided as
ASCII strings.
Perf report/script have been modified to print the basic information as
easy-to-parse ASCII strings. Extended information about CPU and NUMA
topology may be requested with the -I option.
Thanks to David Ahern for reviewing and testing the many versions of
this patch.
$ perf report --stdio
# ========
# captured on : Mon Sep 26 15:22:14 2011
# hostname : quad
# os release : 3.1.0-rc4-tip
# perf version : 3.1.0-rc4
# arch : x86_64
# nrcpus online : 4
# nrcpus avail : 4
# cpudesc : Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
# cpuid : GenuineIntel,6,15,11
# total memory : 8105360 kB
# cmdline : /home/eranian/perfmon/official/tip/build/tools/perf/perf record date
# event : name = cycles, type = 0, config = 0x0, config1 = 0x0, config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 29, 30, 31,
# HEADER_CPU_TOPOLOGY info available, use -I to display
# HEADER_NUMA_TOPOLOGY info available, use -I to display
# ========
#
...
Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Robert Richter <robert.richter@amd.com> Cc: Andi Kleen <ak@linux.intel.com> Link: http://lkml.kernel.org/r/20110930134040.GA5575@quad Signed-off-by: Stephane Eranian <eranian@google.com>
[ committer notes: Use --show-info in the tools as was in the docs, rename
perf_header_fprintf_info to perf_file_section__fprintf_info, fixup
conflict with f69b64f7 "perf: Support setting the disassembler style" ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists browser: Fix TAB/UNTAB use with multiple events
When requesting multiple events, say:
# perf top -e instructions -e cycles -e cache-misses
The first screen lets the user chose what to see first, then to switch
one can either use the left key to get back to the event menu or simply
use TAB to go the next and shift+TAB to go the prev.
When using TAB/UNTAB the call to perf_evlist__set_selected(event) was
missing, fix it.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-3xqqh3fwmt914gg43frey14y@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf tools: Fix broken number of samples for perf report -n
The perf report -n option was broken because it was not reporting the
correct number of samples depending on the sorting mode. By default,
samples are sorted by comm,dso,sym. That means that samples for the same
command (binary) get collapsed.
The hists__collapse_insert_entry() had a bug whereby it was aggregating
the number of events observed (periods) but not the number of samples.
Consequently, the number of samples reported could be below reality. The
percentage remained correct because based on the periods.
This patch fixes the problem by also aggregating the number of samples.
Here is an example:
1. Make sure newt-devel is not installed when building it
2. Use 'perf top --stdio' just like with report
3. Edit your ~/.perfconfig or system wide config and have this there:
[tui]
top = off
But you shouldn't, since the TUI is so much more powerful, has
integration with annotation and where lots more interesting features
will be developed, so if something annoys you (the colors?) just let me
know and I'll do my best to make it pleasant as a default.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-cy2tn4uj1t7c3aqss5l25of5@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf annotate browser: Allow navigation to called functions
I.e. when in the annotate TUI window, if Enter is pressed over an
assembly line with a 'callq' it will try to open another TUI window with
that symbol.
This is just a proof of concept and works only on x86_64, more work is
needed to support kernel modules, userland, other arches, etc, but
should already be useful as-is.
Suggested-by: Ingo Molnar <mingo@elte.hu> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-opyvskw5na3qdmkv8vxi3zbr@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
perf browsers: Add live mode to the hists, annotate browsers
This allows passing a timer to be run periodically, which will update
the hists tree that then gers refreshed on the screen, just like the
Live mode (symbol entries, annotation) we already have in 'perf top
--tui'.
Will be used by the new hist_entry/hists based 'top' tool.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-2r44qd8oe4sagzcgoikl8qzc@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf hists: Threaded addition and sorting of entries
By using a mutex just for inserting and rotating two hist_entry rb
trees, so that when sorting we can get the last batch of entries created
from the ring buffer, merge it with whatever we have processed so far
and show the output while new entries are being added.
The 'report' tool continues, for now, to do it without threading, but
will use this in the future to allow visualization of results in long
perf.data sessions while the entries are being processed.
The new 'top' tool will be the first user.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-9b05atsn0q6m7fqgrug8fk2i@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Joerg Roedel [Wed, 5 Oct 2011 12:01:16 +0000 (14:01 +0200)]
perf, core: Introduce attrs to count in either host or guest mode
The two new attributes exclude_guest and exclude_host can
bes used by user-space to tell the kernel to setup
performance counter to either only count while the CPU is in
guest or in host mode.
An additional check is also introduced to make sure
user-space does not try to exclude guest and host mode from
counting.
Linus Torvalds [Tue, 4 Oct 2011 17:37:06 +0000 (10:37 -0700)]
Merge git://github.com/davem330/net
* git://github.com/davem330/net:
pch_gbe: Fixed the issue on which a network freezes
pch_gbe: Fixed the issue on which PC was frozen when link was downed.
make PACKET_STATISTICS getsockopt report consistently between ring and non-ring
net: xen-netback: correctly restart Tx after a VM restore/migrate
bonding: properly stop queuing work when requested
can bcm: fix incomplete tx_setup fix
RDSRDMA: Fix cleanup of rds_iw_mr_pool
net: Documentation: Fix type of variables
ibmveth: Fix oops on request_irq failure
ipv6: nullify ipv6_ac_list and ipv6_fl_list when creating new socket
cxgb4: Fix EEH on IBM P7IOC
can bcm: fix tx_setup off-by-one errors
MAINTAINERS: tehuti: Alexander Indenbaum's address bounces
dp83640: reduce driver noise
ptp: fix L2 event message recognition
Linus Torvalds [Tue, 4 Oct 2011 16:59:22 +0000 (09:59 -0700)]
Merge branch 'fix/asoc' of git://github.com/tiwai/sound
* 'fix/asoc' of git://github.com/tiwai/sound:
ASoC: omap_mcpdm_remove cannot be __devexit
ASoC: Fix setting update bits for WM8753_LADC and WM8753_RADC
ASoC: use a valid device for dev_err() in Zylonite
projects / linux-beck.git / log