The transactional API patch between the generic and model-specific
code introduced several important bugs with event scheduling, at
least on X86. If you had pinned events, e.g., watchdog, and were
over-committing the PMU, you would get bogus counts. The bug was
showing up on Intel CPU because events would move around more
often that on AMD. But the problem also existed on AMD, though
harder to expose.
The issues were:
- group_sched_in() was missing a cancel_txn() in the error path
- cpuc->n_added was not properly maintained, leading to missing
actions in hw_perf_enable(), i.e., n_running being 0. You cannot
update n_added until you know the transaction has succeeded. In
case of failed transaction n_added was not adjusted back.
- in case of failed transactions, event_sched_out() was called
and eventually invoked x86_disable_event() to touch the HW reg.
But with transactions, on X86, event_sched_in() does not touch
HW registers, it simply collects events into a list. Thus, you
could end up calling x86_disable_event() on a counter which
did not correspond to the current event when idx != -1.
The patch modifies the generic and X86 code to avoid all those problems.
First, we keep track of the number of events added last. In case the
transaction fails, we substract them from n_added. This approach is
necessary (as opposed to delaying updates to n_added) because not all
event updates use the transaction API, e.g., single events.
Second, we encapsulate the event_sched_in() and event_sched_out() in
group_sched_in() inside the transaction. That makes the operations
symmetrical and you can also detect that you are inside a transaction
and skip the HW reg access by checking cpuc->group_flag.
With this patch, you can now overcommit the PMU even with pinned
system-wide events present and still get valid counts.
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1274796225.5882.1389.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Patch b7e2ecef92 (perf, trace: Optimize tracepoints by removing
IRQ-disable from perf/tracepoint interaction) made the
unfortunate mistake of assuming the world is x86 only, correct
this.
The problem was that perf_fetch_caller_regs() did
local_save_flags() into regs->flags, and I re-used that to
remove another local_save_flags(), forgetting !x86 doesn't have
regs->flags.
Do the reverse, remove the local_save_flags() from
perf_fetch_caller_regs() and let the ftrace site do the
local_save_flags() instead.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Paul Mackerras <paulus@samba.org>
Cc: acme@redhat.com
Cc: efault@gmx.de
Cc: fweisbec@gmail.com
Cc: rostedt@goodmis.org
LKML-Reference: <1274778175.5882.623.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Convert to the transactional PMU API and remove the duplication of
group_sched_in().
Reviewed-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: David Miller <davem@davemloft.net>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1272002172.5707.61.camel@minggr.sh.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Rename perf_event_attr::precise to perf_event_attr::precise_ip and
widen it to 2 bits. This new field describes the required precision of
the PERF_SAMPLE_IP field:
0 - SAMPLE_IP can have arbitrary skid
1 - SAMPLE_IP must have constant skid
2 - SAMPLE_IP requested to have 0 skid
3 - SAMPLE_IP must have 0 skid
And modify the Intel PEBS code accordingly. The PEBS implementation
now supports up to precise_ip == 2, where we perform the IP fixup.
Also s/PERF_RECORD_MISC_EXACT/&_IP/ to clarify its meaning, this bit
should be set for each PERF_SAMPLE_IP field known to match the actual
instruction triggering the event.
This new scheme allows for a PEBS mode that uses the buffer for more
than a single event.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There may exist constraints with a cmask set to zero. In this case
for_each_event_constraint() will not work properly. Now weight is used
instead of the cmask for loop exit detection. Weight is always a value
other than zero since the default contains the HWEIGHT from the
counter mask and in other cases a value of zero does not fit too.
This is in preparation of ibs event constraints that wont have a
cmask.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271190201-25705-7-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
To reuse this function for events with different enable bit masks,
this mask is part of the function's argument list now.
The function will be used later to control ibs events too.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271190201-25705-6-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The perfctr setup calls are in the corresponding .hw_config()
functions now. This makes it possible to introduce config functions
for other pmu events that are not perfctr specific.
Also, all of a sudden the code looks much nicer.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271190201-25705-4-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Move x86_setup_perfctr(), no other changes made.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271190201-25705-3-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Split __hw_perf_event_init() to configure pmu events other than
perfctrs. Perfctr code is moved to a separate function
x86_setup_perfctr(). This and the following patches refactor the code.
Split in multiple patches for better review.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271190201-25705-2-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Below patch introduces perf_guest_info_callbacks and related
register/unregister functions. Add more PERF_RECORD_MISC_XXX bits
meaning guest kernel and guest user space.
Signed-off-by: Zhang Yanmin <yanmin_zhang@linux.intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
It is useless now that we have a pure stack frame
walker, as given addr are always reliable.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Stephane noticed that the ANY flag was in generic arch code, and Cyrill
reported that it broke the P4 code.
Solve this by merging x86_pmu::raw_event into x86_pmu::hw_config and
provide intel_pmu and amd_pmu specific versions of this callback.
The intel_pmu one deals with the ANY flag, the amd_pmu adds the few extra
event bits AMD64 has.
Reported-by: Stephane Eranian <eranian@google.com>
Reported-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Robert Richter <robert.richter@amd.com>
Acked-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1269968113.5258.442.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
ARCH_PERFMON_EVENTSEL bit masks are often used in the kernel. This
patch adds macros for the bit masks and removes local defines. The
function intel_pmu_raw_event() becomes x86_pmu_raw_event() which is
generic for x86 models and same also for p6. Duplicate code is
removed.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20100330092821.GH11907@erda.amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The big rename:
cdd6c48 perf: Do the big rename: Performance Counters -> Performance Events
accidentally renamed some members of stucts that were named after
registers in the spec. To avoid confusion this patch reverts some
changes. The related specs are MSR descriptions in AMD's BKDGs and the
ARCHITECTURAL PERFORMANCE MONITORING section in the Intel 64 and IA-32
Architectures Software Developer's Manuals.
This patch does:
$ sed -i -e 's:num_events:num_counters:g' \
arch/x86/include/asm/perf_event.h \
arch/x86/kernel/cpu/perf_event_amd.c \
arch/x86/kernel/cpu/perf_event.c \
arch/x86/kernel/cpu/perf_event_intel.c \
arch/x86/kernel/cpu/perf_event_p6.c \
arch/x86/kernel/cpu/perf_event_p4.c \
arch/x86/oprofile/op_model_ppro.c
$ sed -i -e 's:event_bits:cntval_bits:g' -e 's:event_mask:cntval_mask:g' \
arch/x86/kernel/cpu/perf_event_amd.c \
arch/x86/kernel/cpu/perf_event.c \
arch/x86/kernel/cpu/perf_event_intel.c \
arch/x86/kernel/cpu/perf_event_p6.c \
arch/x86/kernel/cpu/perf_event_p4.c
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1269880612-25800-2-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When profiling a 32-bit process on a 64-bit kernel, callgraph tracing
stopped after the first function, because it has seen a garbage memory
address (tried to interpret the frame pointer, and return address as a
64-bit pointer).
Fix this by using a struct stack_frame with 32-bit pointers when the
TIF_IA32 flag is set.
Note that TIF_IA32 flag must be used, and not is_compat_task(), because
the latter is only set when the 32-bit process is executing a syscall,
which may not always be the case (when tracing page fault events for
example).
Signed-off-by: Török Edwin <edwintorok@gmail.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
LKML-Reference: <1268820436-13145-1-git-send-email-edwintorok@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Commit 3f6da39 ("perf: Rework and fix the arch CPU-hotplug hooks") moved
the amd northbridge allocation from CPUS_ONLINE to CPUS_PREPARE_UP
however amd_nb_id() doesn't work yet on prepare so it would simply bail
basically reverting to a state where we do not properly track node wide
constraints - causing weird perf results.
Fix up the AMD NorthBridge initialization code by allocating from
CPU_UP_PREPARE and installing it from CPU_STARTING once we have the
proper nb_id. It also properly deals with the allocation failing.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
[ robustify using amd_has_nb() ]
Signed-off-by: Stephane Eranian <eranian@google.com>
LKML-Reference: <1269353485.5109.48.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Scheduler's task migration events don't work because they always
pass NULL regs perf_sw_event(). The event hence gets filtered
in perf_swevent_add().
Scheduler's context switches events use task_pt_regs() to get
the context when the event occured which is a wrong thing to
do as this won't give us the place in the kernel where we went
to sleep but the place where we left userspace. The result is
even more wrong if we switch from a kernel thread.
Use the hot regs snapshot for both events as they belong to the
non-interrupt/exception based events family. Unlike page faults
or so that provide the regs matching the exact origin of the event,
we need to save the current context.
This makes the task migration event working and fix the context
switch callchains and origin ip.
Example: perf record -a -e cs
Before:
10.91% ksoftirqd/0 0 [k] 0000000000000000
|
--- (nil)
perf_callchain
perf_prepare_sample
__perf_event_overflow
perf_swevent_overflow
perf_swevent_add
perf_swevent_ctx_event
do_perf_sw_event
__perf_sw_event
perf_event_task_sched_out
schedule
run_ksoftirqd
kthread
kernel_thread_helper
After:
23.77% hald-addon-stor [kernel.kallsyms] [k] schedule
|
--- schedule
|
|--60.00%-- schedule_timeout
| wait_for_common
| wait_for_completion
| blk_execute_rq
| scsi_execute
| scsi_execute_req
| sr_test_unit_ready
| |
| |--66.67%-- sr_media_change
| | media_changed
| | cdrom_media_changed
| | sr_block_media_changed
| | check_disk_change
| | cdrom_open
v2: Always build perf_arch_fetch_caller_regs() now that software
events need that too. They don't need it from modules, unlike trace
events, so we keep the EXPORT_SYMBOL in trace_event_perf.c
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: David Miller <davem@davemloft.net>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Implement the workaround for Intel Errata AAK100 and AAP53.
Also, remove the Core-i7 name for Nehalem events since there are
also Westmere based i7 chips.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1269608924.12097.147.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If reserve_pmc_hardware() succeeds but reserve_ds_buffers()
fails, then we need to release_pmc_hardware. It won't be done
by the destroy() callback because we return before setting it
in case of error.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: <stable@kernel.org>
Cc: peterz@infradead.org
Cc: paulus@samba.org
Cc: davem@davemloft.net
Cc: fweisbec@gmail.com
Cc: robert.richter@amd.com
Cc: perfmon2-devel@lists.sf.net
LKML-Reference: <4ba1568b.15185e0a.182a.7802@mx.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
--
arch/x86/kernel/cpu/perf_event.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
Since apic_write() maps to a plain noop in the !CONFIG_X86_LOCAL_APIC
case we're safe to remove this conditional compilation and clean up
the code a bit.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: fweisbec@gmail.com
Cc: acme@redhat.com
Cc: eranian@google.com
Cc: peterz@infradead.org
LKML-Reference: <20100317104356.232371479@openvz.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The same information is stored also in x86_pmu.intel_ctrl. This
patch removes perf_event_mask and instead uses
x86_pmu.intel_ctrl directly.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1268826553-19518-5-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This member in the struct is not used anymore and can be
removed.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1268826553-19518-4-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The function reserve_pmc_hardware() and release_pmc_hardware()
were hard to read. This patch improves readability of the code by
removing most of the CONFIG_X86_LOCAL_APIC macros.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1268826553-19518-2-git-send-email-robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
perf_arch_fetch_caller_regs() is exported for the overriden x86
version, but not for the generic weak version.
As a general rule, weak functions should not have their symbol
exported in the same file they are defined.
So let's export it on trace_event_perf.c as it is used by trace
events only.
This fixes:
ERROR: ".perf_arch_fetch_caller_regs" [fs/xfs/xfs.ko] undefined!
ERROR: ".perf_arch_fetch_caller_regs" [arch/powerpc/platforms/cell/spufs/spufs.ko] undefined!
-v2: And also only build it if trace events are enabled.
-v3: Fix changelog mistake
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1268697902-9518-1-git-send-regression-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If x86_pmu.hw_config() fails a fixed error code (-EOPNOTSUPP) is
returned even if a different error was reported. This patch fixes
this.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Acked-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Lin Ming <ming.m.lin@intel.com>
Cc: acme@redhat.com
Cc: eranian@google.com
Cc: gorcunov@openvz.org
Cc: peterz@infradead.org
Cc: fweisbec@gmail.com
LKML-Reference: <20100316160733.GR1585@erda.amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
perf_arch_fetch_caller_regs() is exported for the overriden x86
version, but not for the generic weak version.
As a general rule, weak functions should not have their symbol
exported in the same file they are defined.
So let's export it on trace_event_perf.c as it is used by trace
events only.
This fixes:
ERROR: ".perf_arch_fetch_caller_regs" [fs/xfs/xfs.ko] undefined!
ERROR: ".perf_arch_fetch_caller_regs" [arch/powerpc/platforms/cell/spufs/spufs.ko] undefined!
-v2: And also only build it if trace events are enabled.
-v3: Fix changelog mistake
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1268697902-9518-1-git-send-regression-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
In case of not assigned x86_pmu and software events NULL dereference may
being hit via x86_pmu::schedule_events method.
Fix it by checking if x86_pmu is initialized at all.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Lin Ming <ming.m.lin@intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <20100311215016.GG25162@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The netburst PMU is way different from the "architectural
perfomance monitoring" specification that current CPUs use.
P4 uses a tuple of ESCR+CCCR+COUNTER MSR registers to handle
perfomance monitoring events.
A few implementational details:
1) We need a separate x86_pmu::hw_config helper in struct
x86_pmu since register bit-fields are quite different from P6,
Core and later cpu series.
2) For the same reason is a x86_pmu::schedule_events helper
introduced.
3) hw_perf_event::config consists of packed ESCR+CCCR values.
It's allowed since in reality both registers only use a half
of their size. Of course before making a real write into a
particular MSR we need to unpack the value and extend it to
a proper size.
4) The tuple of packed ESCR+CCCR in hw_perf_event::config
doesn't describe the memory address of ESCR MSR register
so that we need to keep a mapping between these tuples
used and available ESCR (various P4 events may use same
ESCRs but not simultaneously), for this sake every active
event has a per-cpu map of hw_perf_event::idx <--> ESCR
addresses.
5) Since hw_perf_event::idx is an offset to counter/control register
we need to lift X86_PMC_MAX_GENERIC up, otherwise kernel
strips it down to 8 registers and event armed may never be turned
off (ie the bit in active_mask is set but the loop never reaches
this index to check), thanks to Peter Zijlstra
Restrictions:
- No cascaded counters support (do we ever need them?)
- No dependent events support (so PERF_COUNT_HW_INSTRUCTIONS
doesn't work for now)
- There are events with same counters which can't work simultaneously
(need to use intersected ones due to broken counter 1)
- No PERF_COUNT_HW_CACHE_ events yet
Todo:
- Implement dependent events
- Need proper hashing for event opcodes (no linear search, good for
debugging stage but not in real loads)
- Some events counted during a clock cycle -- need to set threshold
for them and count every clock cycle just to get summary statistics
(ie to behave the same way as other PMUs do)
- Need to swicth to use event_constraints
- To support RAW events we need to encode a global list of P4 events
into p4_templates
- Cache events need to be added
Event support status matrix:
Event status
-----------------------------
cycles works
cache-references works
cache-misses works
branch-misses works
bus-cycles partially (does not work on 64bit cpu with HT enabled)
instruction doesnt work (needs dependent event [mop tagging])
branches doesnt work
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20100311165439.GB5129@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Export perf_trace_regs and perf_arch_fetch_caller_regs since module will
use these.
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
[ use EXPORT_PER_CPU_SYMBOL_GPL() ]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <4B989C1B.2090407@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
What happens is that we schedule badly like:
<...>-1987 [019] 280.252808: x86_pmu_start: event-46/1300c0: idx: 0
<...>-1987 [019] 280.252811: x86_pmu_start: event-47/1300c0: idx: 1
<...>-1987 [019] 280.252812: x86_pmu_start: event-48/1300c0: idx: 2
<...>-1987 [019] 280.252813: x86_pmu_start: event-49/1300c0: idx: 3
<...>-1987 [019] 280.252814: x86_pmu_start: event-50/1300c0: idx: 32
<...>-1987 [019] 280.252825: x86_pmu_stop: event-46/1300c0: idx: 0
<...>-1987 [019] 280.252826: x86_pmu_stop: event-47/1300c0: idx: 1
<...>-1987 [019] 280.252827: x86_pmu_stop: event-48/1300c0: idx: 2
<...>-1987 [019] 280.252828: x86_pmu_stop: event-49/1300c0: idx: 3
<...>-1987 [019] 280.252829: x86_pmu_stop: event-50/1300c0: idx: 32
<...>-1987 [019] 280.252834: x86_pmu_start: event-47/1300c0: idx: 1
<...>-1987 [019] 280.252834: x86_pmu_start: event-48/1300c0: idx: 2
<...>-1987 [019] 280.252835: x86_pmu_start: event-49/1300c0: idx: 3
<...>-1987 [019] 280.252836: x86_pmu_start: event-50/1300c0: idx: 32
<...>-1987 [019] 280.252837: x86_pmu_start: event-51/1300c0: idx: 32 *FAIL*
This happens because we only iterate the n_running events in the first
pass, and reset their index to -1 if they don't match to force a
re-assignment.
Now, in our RR example, n_running == 0 because we fully unscheduled, so
event-50 will retain its idx==32, even though in scheduling it will have
gotten idx=0, and we don't trigger the re-assign path.
The easiest way to fix this is the below patch, which simply validates
the full assignment in the second pass.
Reported-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1268311069.5037.31.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Events that trigger overflows by interrupting a context can
use get_irq_regs() or task_pt_regs() to retrieve the state
when the event triggered. But this is not the case for some
other class of events like trace events as tracepoints are
executed in the same context than the code that triggered
the event.
It means we need a different api to capture the regs there,
namely we need a hot snapshot to get the most important
informations for perf: the instruction pointer to get the
event origin, the frame pointer for the callchain, the code
segment for user_mode() tests (we always use __KERNEL_CS as
trace events always occur from the kernel) and the eflags
for further purposes.
v2: rename perf_save_regs to perf_fetch_caller_regs as per
Masami's suggestion.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Archs <linux-arch@vger.kernel.org>
We don't need checking_{wr,rd}msr() calls, since we should know what cpu
we're running on and not use blindly poke at msrs.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This CPU has just too many handycaps to be really useful.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <20100305154128.890278662@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Saner PERF_CAPABILITIES support, which also exposes pebs_trap. Use that
latter to make PEBS's use of LBR conditional since a fault-like pebs
should already report the correct IP.
( As of this writing there is no known hardware that implements
!pebs_trap )
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <20100304140100.770650663@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Use the LBR to fix up the PEBS IP+1 issue.
As said, PEBS reports the next instruction, here we use the LBR to find
the last branch and from that construct the actual IP. If the IP matches
the LBR-TO, we use LBR-FROM, otherwise we use the LBR-TO address as the
beginning of the last basic block and decode forward.
Once we find a match to the current IP, we use the previous location.
This patch introduces a new ABI element: PERF_RECORD_MISC_EXACT, which
conveys that the reported IP (PERF_SAMPLE_IP) is the exact instruction
that caused the event (barring CPU errata).
The fixup can fail due to various reasons:
1) LBR contains invalid data (quite possible)
2) part of the basic block got paged out
3) the reported IP isn't part of the basic block (see 1)
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Masami Hiramatsu <mhiramat@redhat.com>
Cc: "Zhang, Yanmin" <yanmin_zhang@linux.intel.com>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <20100304140100.619375431@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Implement simple suport Intel Last-Branch-Record, it supports all
hardware that implements FREEZE_LBRS_ON_PMI, but does not (yet) implement
the LBR config register.
The Intel LBR is a FIFO of From,To addresses describing the last few
branches the hardware took.
This patch does not add perf interface to the LBR, but merely provides an
interface for internal use.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <20100304140100.544191154@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch implements support for Intel Precise Event Based Sampling,
which is an alternative counter mode in which the counter triggers a
hardware assist to collect information on events. The hardware assist
takes a trap like snapshot of a subset of the machine registers.
This data is written to the Intel Debug-Store, which can be programmed
with a data threshold at which to raise a PMI.
With the PEBS hardware assist being trap like, the reported IP is always
one instruction after the actual instruction that triggered the event.
This implements a simple PEBS model that always takes a single PEBS event
at a time. This is done so that the interaction with the rest of the
system is as expected (freq adjust, period randomization, lbr,
callchains, etc.).
It adds an ABI element: perf_event_attr::precise, which indicates that we
wish to use this (constrained, but precise) mode.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <20100304140100.392111285@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
hw_perf_enable() would enable already enabled events.
This causes problems with code that assumes that ->enable/->disable calls
are balanced (like the LBR code does).
What happens is that events that were already running and left in place
would get enabled again.
Avoid this by only enabling new events that match their previous
assignment.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
hw_perf_enable() would disable events that were not yet enabled.
This causes problems with code that assumes that ->enable/->disable calls
are balanced (like the LBR code does).
What happens is that we disable newly added counters that match their
previous assignment, even though they are not yet programmed on the
hardware.
Avoid this by only doing the first pass over the existing events.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Make sure n_added is properly accounted so that we can rely on the value
to reflect the number of added counters. This is needed if its going to
be used for more than a boolean check.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Calling ioctl(PERF_EVENT_IOC_DISABLE) on a thottled counter would result
in a double disable, cure this by using x86_pmu_{start,stop} for
throttle/unthrottle and teach x86_pmu_stop() to check ->active_mask.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There is no concurrency on these variables, so don't use LOCK'ed ops.
As to the intel_pmu_handle_irq() status bit clean, nobody uses that so
remove it all together.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: paulus@samba.org
Cc: eranian@google.com
Cc: robert.richter@amd.com
Cc: fweisbec@gmail.com
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
LKML-Reference: <20100304140100.240023029@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Stephane Eranian
Peter Zijlstra
Lin Ming
Robert Richter
Zhang, Yanmin
Ingo Molnar
Tejun Heo
Frederic Weisbecker
Peter Zijlstra
Torok Edwin
Cyrill Gorcunov
Cyrill Gorcunov
Xiao Guangrong