lk/linux-kernel-test
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

Browse Source 
* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  tracing: Fix sign fields in ftrace_define_fields_##call()
  tracing/syscalls: Fix typo in SYSCALL_DEFINE0
  tracing/kprobe: Show sign of fields in trace_kprobe format files
  ksym_tracer: Remove trace_stat
  ksym_tracer: Fix race when incrementing count
  ksym_tracer: Fix to allow writing newline to ksym_trace_filter
  ksym_tracer: Fix to make the tracer work
  tracing: Kconfig spelling fixes and cleanups
  tracing: Fix setting tracer specific options
  Documentation: Update ftrace-design.txt
  Documentation: Update tracepoint-analysis.txt
  Documentation: Update mmiotrace.txt
This commit is contained in:
Linus Torvalds
2009-12-31 11:52:01 -08:00
parent 4e58fb7305 fb7ae981cb
commit b21c070403
11 changed files with 185 additions and 203 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
Documentation/trace/ftrace-design.txt
View File

@@ -53,14 +53,14 @@ size of the mcount call that is embedded in the function).
For example, if the function foo() calls bar(), when the bar() function calls For example, if the function foo() calls bar(), when the bar() function calls
mcount(), the arguments mcount() will pass to the tracer are: mcount(), the arguments mcount() will pass to the tracer are:
"frompc" - the address bar() will use to return to foo() "frompc" - the address bar() will use to return to foo()
"selfpc" - the address bar() (with _mcount() size adjustment) "selfpc" - the address bar() (with mcount() size adjustment)
Also keep in mind that this mcount function will be called *a lot*, so Also keep in mind that this mcount function will be called *a lot*, so
optimizing for the default case of no tracer will help the smooth running of optimizing for the default case of no tracer will help the smooth running of
your system when tracing is disabled. So the start of the mcount function is your system when tracing is disabled. So the start of the mcount function is
typically the bare min with checking things before returning. That also means typically the bare minimum with checking things before returning. That also
the code flow should usually kept linear (i.e. no branching in the nop case). means the code flow should usually be kept linear (i.e. no branching in the nop
This is of course an optimization and not a hard requirement. case). This is of course an optimization and not a hard requirement.
Here is some pseudo code that should help (these functions should actually be Here is some pseudo code that should help (these functions should actually be
implemented in assembly): implemented in assembly):
@@ -131,10 +131,10 @@ some functions to save (hijack) and restore the return address.
The mcount function should check the function pointers ftrace_graph_return The mcount function should check the function pointers ftrace_graph_return
(compare to ftrace_stub) and ftrace_graph_entry (compare to (compare to ftrace_stub) and ftrace_graph_entry (compare to
ftrace_graph_entry_stub). If either of those are not set to the relevant stub ftrace_graph_entry_stub). If either of those is not set to the relevant stub
function, call the arch-specific function ftrace_graph_caller which in turn function, call the arch-specific function ftrace_graph_caller which in turn
calls the arch-specific function prepare_ftrace_return. Neither of these calls the arch-specific function prepare_ftrace_return. Neither of these
function names are strictly required, but you should use them anyways to stay function names is strictly required, but you should use them anyway to stay
consistent across the architecture ports -- easier to compare & contrast consistent across the architecture ports -- easier to compare & contrast
things. things.
@@ -144,7 +144,7 @@ but the first argument should be a pointer to the "frompc". Typically this is
located on the stack. This allows the function to hijack the return address located on the stack. This allows the function to hijack the return address
temporarily to have it point to the arch-specific function return_to_handler. temporarily to have it point to the arch-specific function return_to_handler.
That function will simply call the common ftrace_return_to_handler function and That function will simply call the common ftrace_return_to_handler function and
that will return the original return address with which, you can return to the that will return the original return address with which you can return to the
original call site. original call site.
Here is the updated mcount pseudo code: Here is the updated mcount pseudo code:

15
Documentation/trace/mmiotrace.txt
View File

@@ -44,7 +44,8 @@ Check for lost events.
Usage Usage
----- -----
Make sure debugfs is mounted to /sys/kernel/debug. If not, (requires root privileges) Make sure debugfs is mounted to /sys/kernel/debug.
If not (requires root privileges):
$ mount -t debugfs debugfs /sys/kernel/debug $ mount -t debugfs debugfs /sys/kernel/debug
Check that the driver you are about to trace is not loaded. Check that the driver you are about to trace is not loaded.
@@ -91,7 +92,7 @@ $ dmesg > dmesg.txt
$ tar zcf pciid-nick-mmiotrace.tar.gz mydump.txt lspci.txt dmesg.txt $ tar zcf pciid-nick-mmiotrace.tar.gz mydump.txt lspci.txt dmesg.txt
and then send the .tar.gz file. The trace compresses considerably. Replace and then send the .tar.gz file. The trace compresses considerably. Replace
"pciid" and "nick" with the PCI ID or model name of your piece of hardware "pciid" and "nick" with the PCI ID or model name of your piece of hardware
under investigation and your nick name. under investigation and your nickname.
How Mmiotrace Works How Mmiotrace Works
@@ -100,7 +101,7 @@ How Mmiotrace Works
Access to hardware IO-memory is gained by mapping addresses from PCI bus by Access to hardware IO-memory is gained by mapping addresses from PCI bus by
calling one of the ioremap_*() functions. Mmiotrace is hooked into the calling one of the ioremap_*() functions. Mmiotrace is hooked into the
__ioremap() function and gets called whenever a mapping is created. Mapping is __ioremap() function and gets called whenever a mapping is created. Mapping is
an event that is recorded into the trace log. Note, that ISA range mappings an event that is recorded into the trace log. Note that ISA range mappings
are not caught, since the mapping always exists and is returned directly. are not caught, since the mapping always exists and is returned directly.
MMIO accesses are recorded via page faults. Just before __ioremap() returns, MMIO accesses are recorded via page faults. Just before __ioremap() returns,
@@ -122,11 +123,11 @@ Trace Log Format
---------------- ----------------
The raw log is text and easily filtered with e.g. grep and awk. One record is The raw log is text and easily filtered with e.g. grep and awk. One record is
one line in the log. A record starts with a keyword, followed by keyword one line in the log. A record starts with a keyword, followed by keyword-
dependant arguments. Arguments are separated by a space, or continue until the dependent arguments. Arguments are separated by a space, or continue until the
end of line. The format for version 20070824 is as follows: end of line. The format for version 20070824 is as follows:
Explanation Keyword Space separated arguments Explanation Keyword Space-separated arguments
--------------------------------------------------------------------------- ---------------------------------------------------------------------------
read event R width, timestamp, map id, physical, value, PC, PID read event R width, timestamp, map id, physical, value, PC, PID
@@ -136,7 +137,7 @@ iounmap event UNMAP timestamp, map id, PC, PID
marker MARK timestamp, text marker MARK timestamp, text
version VERSION the string "20070824" version VERSION the string "20070824"
info for reader LSPCI one line from lspci -v info for reader LSPCI one line from lspci -v
PCI address map PCIDEV space separated /proc/bus/pci/devices data PCI address map PCIDEV space-separated /proc/bus/pci/devices data
unk. opcode UNKNOWN timestamp, map id, physical, data, PC, PID unk. opcode UNKNOWN timestamp, map id, physical, data, PC, PID
Timestamp is in seconds with decimals. Physical is a PCI bus address, virtual Timestamp is in seconds with decimals. Physical is a PCI bus address, virtual

60
Documentation/trace/tracepoint-analysis.txt
View File

@@ -10,8 +10,8 @@ Tracepoints (see Documentation/trace/tracepoints.txt) can be used without
creating custom kernel modules to register probe functions using the event creating custom kernel modules to register probe functions using the event
tracing infrastructure. tracing infrastructure.
Simplistically, tracepoints will represent an important event that when can Simplistically, tracepoints represent important events that can be
be taken in conjunction with other tracepoints to build a "Big Picture" of taken in conjunction with other tracepoints to build a "Big Picture" of
what is going on within the system. There are a large number of methods for what is going on within the system. There are a large number of methods for
gathering and interpreting these events. Lacking any current Best Practises, gathering and interpreting these events. Lacking any current Best Practises,
this document describes some of the methods that can be used. this document describes some of the methods that can be used.
@@ -33,12 +33,12 @@ calling
will give a fair indication of the number of events available. will give a fair indication of the number of events available.
2.2 PCL 2.2 PCL (Performance Counters for Linux)
------- -------
Discovery and enumeration of all counters and events, including tracepoints Discovery and enumeration of all counters and events, including tracepoints,
are available with the perf tool. Getting a list of available events is a are available with the perf tool. Getting a list of available events is a
simple case of simple case of:
$ perf list 2>&1 | grep Tracepoint $ perf list 2>&1 | grep Tracepoint
ext4:ext4_free_inode [Tracepoint event] ext4:ext4_free_inode [Tracepoint event]
@@ -49,19 +49,19 @@ simple case of
[ .... remaining output snipped .... ] [ .... remaining output snipped .... ]
2. Enabling Events 3. Enabling Events
================== ==================
2.1 System-Wide Event Enabling 3.1 System-Wide Event Enabling
------------------------------ ------------------------------
See Documentation/trace/events.txt for a proper description on how events See Documentation/trace/events.txt for a proper description on how events
can be enabled system-wide. A short example of enabling all events related can be enabled system-wide. A short example of enabling all events related
to page allocation would look something like to page allocation would look something like:
$ for i in `find /sys/kernel/debug/tracing/events -name "enable" | grep mm_`; do echo 1 > $i; done $ for i in `find /sys/kernel/debug/tracing/events -name "enable" | grep mm_`; do echo 1 > $i; done
2.2 System-Wide Event Enabling with SystemTap 3.2 System-Wide Event Enabling with SystemTap
--------------------------------------------- ---------------------------------------------
In SystemTap, tracepoints are accessible using the kernel.trace() function In SystemTap, tracepoints are accessible using the kernel.trace() function
@@ -86,7 +86,7 @@ were allocating the pages.
print_count() print_count()
} }
2.3 System-Wide Event Enabling with PCL 3.3 System-Wide Event Enabling with PCL
--------------------------------------- ---------------------------------------
By specifying the -a switch and analysing sleep, the system-wide events By specifying the -a switch and analysing sleep, the system-wide events
@@ -107,16 +107,16 @@ for a duration of time can be examined.
Similarly, one could execute a shell and exit it as desired to get a report Similarly, one could execute a shell and exit it as desired to get a report
at that point. at that point.
2.4 Local Event Enabling 3.4 Local Event Enabling
------------------------ ------------------------
Documentation/trace/ftrace.txt describes how to enable events on a per-thread Documentation/trace/ftrace.txt describes how to enable events on a per-thread
basis using set_ftrace_pid. basis using set_ftrace_pid.
2.5 Local Event Enablement with PCL 3.5 Local Event Enablement with PCL
----------------------------------- -----------------------------------
Events can be activate and tracked for the duration of a process on a local Events can be activated and tracked for the duration of a process on a local
basis using PCL such as follows. basis using PCL such as follows.
$ perf stat -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \ $ perf stat -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \
@@ -131,18 +131,18 @@ basis using PCL such as follows.
0.973913387 seconds time elapsed 0.973913387 seconds time elapsed
3. Event Filtering 4. Event Filtering
================== ==================
Documentation/trace/ftrace.txt covers in-depth how to filter events in Documentation/trace/ftrace.txt covers in-depth how to filter events in
ftrace. Obviously using grep and awk of trace_pipe is an option as well ftrace. Obviously using grep and awk of trace_pipe is an option as well
as any script reading trace_pipe. as any script reading trace_pipe.
4. Analysing Event Variances with PCL 5. Analysing Event Variances with PCL
===================================== =====================================
Any workload can exhibit variances between runs and it can be important Any workload can exhibit variances between runs and it can be important
to know what the standard deviation in. By and large, this is left to the to know what the standard deviation is. By and large, this is left to the
performance analyst to do it by hand. In the event that the discrete event performance analyst to do it by hand. In the event that the discrete event
occurrences are useful to the performance analyst, then perf can be used. occurrences are useful to the performance analyst, then perf can be used.
@@ -166,7 +166,7 @@ In the event that some higher-level event is required that depends on some
aggregation of discrete events, then a script would need to be developed. aggregation of discrete events, then a script would need to be developed.
Using --repeat, it is also possible to view how events are fluctuating over Using --repeat, it is also possible to view how events are fluctuating over
time on a system wide basis using -a and sleep. time on a system-wide basis using -a and sleep.
$ perf stat -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \ $ perf stat -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \
-e kmem:mm_pagevec_free \ -e kmem:mm_pagevec_free \
@@ -180,7 +180,7 @@ time on a system wide basis using -a and sleep.
1.002251757 seconds time elapsed ( +- 0.005% ) 1.002251757 seconds time elapsed ( +- 0.005% )
5. Higher-Level Analysis with Helper Scripts 6. Higher-Level Analysis with Helper Scripts
============================================ ============================================
When events are enabled the events that are triggering can be read from When events are enabled the events that are triggering can be read from
@@ -190,11 +190,11 @@ be gathered on-line as appropriate. Examples of post-processing might include
o Reading information from /proc for the PID that triggered the event o Reading information from /proc for the PID that triggered the event
o Deriving a higher-level event from a series of lower-level events. o Deriving a higher-level event from a series of lower-level events.
o Calculate latencies between two events o Calculating latencies between two events
Documentation/trace/postprocess/trace-pagealloc-postprocess.pl is an example Documentation/trace/postprocess/trace-pagealloc-postprocess.pl is an example
script that can read trace_pipe from STDIN or a copy of a trace. When used script that can read trace_pipe from STDIN or a copy of a trace. When used
on-line, it can be interrupted once to generate a report without existing on-line, it can be interrupted once to generate a report without exiting
and twice to exit. and twice to exit.
Simplistically, the script just reads STDIN and counts up events but it Simplistically, the script just reads STDIN and counts up events but it
@@ -212,12 +212,12 @@ also can do more such as
processes, the parent process responsible for creating all the helpers processes, the parent process responsible for creating all the helpers
can be identified can be identified
6. Lower-Level Analysis with PCL 7. Lower-Level Analysis with PCL
================================ ================================
There may also be a requirement to identify what functions with a program There may also be a requirement to identify what functions within a program
were generating events within the kernel. To begin this sort of analysis, the were generating events within the kernel. To begin this sort of analysis, the
data must be recorded. At the time of writing, this required root data must be recorded. At the time of writing, this required root:
$ perf record -c 1 \ $ perf record -c 1 \
-e kmem:mm_page_alloc -e kmem:mm_page_free_direct \ -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \
@@ -253,11 +253,11 @@ perf report.
# (For more details, try: perf report --sort comm,dso,symbol) # (For more details, try: perf report --sort comm,dso,symbol)
# #
According to this, the vast majority of events occured triggered on events According to this, the vast majority of events triggered on events
within the VDSO. With simple binaries, this will often be the case so lets within the VDSO. With simple binaries, this will often be the case so let's
take a slightly different example. In the course of writing this, it was take a slightly different example. In the course of writing this, it was
noticed that X was generating an insane amount of page allocations so lets look noticed that X was generating an insane amount of page allocations so let's look
at it at it:
$ perf record -c 1 -f \ $ perf record -c 1 -f \
-e kmem:mm_page_alloc -e kmem:mm_page_free_direct \ -e kmem:mm_page_alloc -e kmem:mm_page_free_direct \
@@ -280,8 +280,8 @@ This was interrupted after a few seconds and
# (For more details, try: perf report --sort comm,dso,symbol) # (For more details, try: perf report --sort comm,dso,symbol)
# #
So, almost half of the events are occuring in a library. To get an idea which So, almost half of the events are occurring in a library. To get an idea which
symbol. symbol:
$ perf report --sort comm,dso,symbol $ perf report --sort comm,dso,symbol
# Samples: 27666 # Samples: 27666
@@ -297,7 +297,7 @@ symbol.
0.01% Xorg /opt/gfx-test/lib/libpixman-1.so.0.13.1 [.] get_fast_path 0.01% Xorg /opt/gfx-test/lib/libpixman-1.so.0.13.1 [.] get_fast_path
0.00% Xorg [kernel] [k] ftrace_trace_userstack 0.00% Xorg [kernel] [k] ftrace_trace_userstack
To see where within the function pixmanFillsse2 things are going wrong To see where within the function pixmanFillsse2 things are going wrong:
$ perf annotate pixmanFillsse2 $ perf annotate pixmanFillsse2
[ ... ] [ ... ]

2
include/linux/syscalls.h
View File

@@ -195,7 +195,7 @@ struct perf_event_attr;
static const struct syscall_metadata __used \ static const struct syscall_metadata __used \
__attribute__((__aligned__(4))) \ __attribute__((__aligned__(4))) \
__attribute__((section("__syscalls_metadata"))) \ __attribute__((section("__syscalls_metadata"))) \
__syscall_meta_##sname = { \ __syscall_meta__##sname = { \
.name = "sys_"#sname, \ .name = "sys_"#sname, \
.nb_args = 0, \ .nb_args = 0, \
.enter_event = &event_enter__##sname, \ .enter_event = &event_enter__##sname, \

7
include/trace/ftrace.h
View File

@@ -414,7 +414,8 @@ ftrace_raw_output_##call(struct trace_iterator *iter, int flags) \
BUILD_BUG_ON(len > MAX_FILTER_STR_VAL); \ BUILD_BUG_ON(len > MAX_FILTER_STR_VAL); \
ret = trace_define_field(event_call, #type "[" #len "]", #item, \ ret = trace_define_field(event_call, #type "[" #len "]", #item, \
offsetof(typeof(field), item), \ offsetof(typeof(field), item), \
sizeof(field.item), 0, FILTER_OTHER); \ sizeof(field.item), \
is_signed_type(type), FILTER_OTHER); \
if (ret) \ if (ret) \
return ret; return ret;
@@ -422,8 +423,8 @@ ftrace_raw_output_##call(struct trace_iterator *iter, int flags) \
#define __dynamic_array(type, item, len) \ #define __dynamic_array(type, item, len) \
ret = trace_define_field(event_call, "__data_loc " #type "[]", #item, \ ret = trace_define_field(event_call, "__data_loc " #type "[]", #item, \
offsetof(typeof(field), __data_loc_##item), \ offsetof(typeof(field), __data_loc_##item), \
sizeof(field.__data_loc_##item), 0, \ sizeof(field.__data_loc_##item), \
FILTER_OTHER); is_signed_type(type), FILTER_OTHER);
#undef __string #undef __string
#define __string(item, src) __dynamic_array(char, item, -1) #define __string(item, src) __dynamic_array(char, item, -1)

10
kernel/hw_breakpoint.c
View File

@@ -40,6 +40,7 @@
#include <linux/percpu.h> #include <linux/percpu.h>
#include <linux/sched.h> #include <linux/sched.h>
#include <linux/init.h> #include <linux/init.h>
#include <linux/cpu.h>
#include <linux/smp.h> #include <linux/smp.h>
#include <linux/hw_breakpoint.h> #include <linux/hw_breakpoint.h>
@@ -388,7 +389,8 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr,
if (!cpu_events) if (!cpu_events)
return ERR_PTR(-ENOMEM); return ERR_PTR(-ENOMEM);
for_each_possible_cpu(cpu) { get_online_cpus();
for_each_online_cpu(cpu) {
pevent = per_cpu_ptr(cpu_events, cpu); pevent = per_cpu_ptr(cpu_events, cpu);
bp = perf_event_create_kernel_counter(attr, cpu, -1, triggered); bp = perf_event_create_kernel_counter(attr, cpu, -1, triggered);
@@ -399,18 +401,20 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr,
goto fail; goto fail;
} }
} }
put_online_cpus();
return cpu_events; return cpu_events;
fail: fail:
for_each_possible_cpu(cpu) { for_each_online_cpu(cpu) {
pevent = per_cpu_ptr(cpu_events, cpu); pevent = per_cpu_ptr(cpu_events, cpu);
if (IS_ERR(*pevent)) if (IS_ERR(*pevent))
break; break;
unregister_hw_breakpoint(*pevent); unregister_hw_breakpoint(*pevent);
} }
put_online_cpus();
free_percpu(cpu_events); free_percpu(cpu_events);
/* return the error if any */
return ERR_PTR(err); return ERR_PTR(err);
} }
EXPORT_SYMBOL_GPL(register_wide_hw_breakpoint); EXPORT_SYMBOL_GPL(register_wide_hw_breakpoint);

112
kernel/trace/Kconfig
View File

@@ -12,17 +12,17 @@ config NOP_TRACER
config HAVE_FTRACE_NMI_ENTER config HAVE_FTRACE_NMI_ENTER
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_FUNCTION_TRACER config HAVE_FUNCTION_TRACER
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_FUNCTION_GRAPH_TRACER config HAVE_FUNCTION_GRAPH_TRACER
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_FUNCTION_GRAPH_FP_TEST config HAVE_FUNCTION_GRAPH_FP_TEST
bool bool
@@ -34,17 +34,17 @@ config HAVE_FUNCTION_GRAPH_FP_TEST
config HAVE_FUNCTION_TRACE_MCOUNT_TEST config HAVE_FUNCTION_TRACE_MCOUNT_TEST
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_DYNAMIC_FTRACE config HAVE_DYNAMIC_FTRACE
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_FTRACE_MCOUNT_RECORD config HAVE_FTRACE_MCOUNT_RECORD
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config HAVE_HW_BRANCH_TRACER config HAVE_HW_BRANCH_TRACER
bool bool
@@ -52,7 +52,7 @@ config HAVE_HW_BRANCH_TRACER
config HAVE_SYSCALL_TRACEPOINTS config HAVE_SYSCALL_TRACEPOINTS
bool bool
help help
See Documentation/trace/ftrace-implementation.txt See Documentation/trace/ftrace-design.txt
config TRACER_MAX_TRACE config TRACER_MAX_TRACE
bool bool
@@ -83,7 +83,7 @@ config RING_BUFFER_ALLOW_SWAP
# This allows those options to appear when no other tracer is selected. But the # This allows those options to appear when no other tracer is selected. But the
# options do not appear when something else selects it. We need the two options # options do not appear when something else selects it. We need the two options
# GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
# hidding of the automatic options. # hiding of the automatic options.
config TRACING config TRACING
bool bool
@@ -119,7 +119,7 @@ menuconfig FTRACE
bool "Tracers" bool "Tracers"
default y if DEBUG_KERNEL default y if DEBUG_KERNEL
help help
Enable the kernel tracing infrastructure. Enable the kernel tracing infrastructure.
if FTRACE if FTRACE
@@ -133,7 +133,7 @@ config FUNCTION_TRACER
help help
Enable the kernel to trace every kernel function. This is done Enable the kernel to trace every kernel function. This is done
by using a compiler feature to insert a small, 5-byte No-Operation by using a compiler feature to insert a small, 5-byte No-Operation
instruction to the beginning of every kernel function, which NOP instruction at the beginning of every kernel function, which NOP
sequence is then dynamically patched into a tracer call when sequence is then dynamically patched into a tracer call when
tracing is enabled by the administrator. If it's runtime disabled tracing is enabled by the administrator. If it's runtime disabled
(the bootup default), then the overhead of the instructions is very (the bootup default), then the overhead of the instructions is very
@@ -150,7 +150,7 @@ config FUNCTION_GRAPH_TRACER
and its entry. and its entry.
Its first purpose is to trace the duration of functions and Its first purpose is to trace the duration of functions and
draw a call graph for each thread with some information like draw a call graph for each thread with some information like
the return value. This is done by setting the current return the return value. This is done by setting the current return
address on the current task structure into a stack of calls. address on the current task structure into a stack of calls.
@@ -173,7 +173,7 @@ config IRQSOFF_TRACER
echo 0 > /sys/kernel/debug/tracing/tracing_max_latency echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
(Note that kernel size and overhead increases with this option (Note that kernel size and overhead increase with this option
enabled. This option and the preempt-off timing option can be enabled. This option and the preempt-off timing option can be
used together or separately.) used together or separately.)
@@ -186,7 +186,7 @@ config PREEMPT_TRACER
select TRACER_MAX_TRACE select TRACER_MAX_TRACE
select RING_BUFFER_ALLOW_SWAP select RING_BUFFER_ALLOW_SWAP
help help
This option measures the time spent in preemption off critical This option measures the time spent in preemption-off critical
sections, with microsecond accuracy. sections, with microsecond accuracy.
The default measurement method is a maximum search, which is The default measurement method is a maximum search, which is
@@ -195,7 +195,7 @@ config PREEMPT_TRACER
echo 0 > /sys/kernel/debug/tracing/tracing_max_latency echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
(Note that kernel size and overhead increases with this option (Note that kernel size and overhead increase with this option
enabled. This option and the irqs-off timing option can be enabled. This option and the irqs-off timing option can be
used together or separately.) used together or separately.)
@@ -222,7 +222,7 @@ config ENABLE_DEFAULT_TRACERS
depends on !GENERIC_TRACER depends on !GENERIC_TRACER
select TRACING select TRACING
help help
This tracer hooks to various trace points in the kernel This tracer hooks to various trace points in the kernel,
allowing the user to pick and choose which trace point they allowing the user to pick and choose which trace point they
want to trace. It also includes the sched_switch tracer plugin. want to trace. It also includes the sched_switch tracer plugin.
@@ -265,19 +265,19 @@ choice
The likely/unlikely profiler only looks at the conditions that The likely/unlikely profiler only looks at the conditions that
are annotated with a likely or unlikely macro. are annotated with a likely or unlikely macro.
The "all branch" profiler will profile every if statement in the The "all branch" profiler will profile every if-statement in the
kernel. This profiler will also enable the likely/unlikely kernel. This profiler will also enable the likely/unlikely
profiler as well. profiler.
Either of the above profilers add a bit of overhead to the system. Either of the above profilers adds a bit of overhead to the system.
If unsure choose "No branch profiling". If unsure, choose "No branch profiling".
config BRANCH_PROFILE_NONE config BRANCH_PROFILE_NONE
bool "No branch profiling" bool "No branch profiling"
help help
No branch profiling. Branch profiling adds a bit of overhead. No branch profiling. Branch profiling adds a bit of overhead.
Only enable it if you want to analyse the branching behavior. Only enable it if you want to analyse the branching behavior.
Otherwise keep it disabled. Otherwise keep it disabled.
config PROFILE_ANNOTATED_BRANCHES config PROFILE_ANNOTATED_BRANCHES
bool "Trace likely/unlikely profiler" bool "Trace likely/unlikely profiler"
@@ -288,7 +288,7 @@ config PROFILE_ANNOTATED_BRANCHES
/sys/kernel/debug/tracing/profile_annotated_branch /sys/kernel/debug/tracing/profile_annotated_branch
Note: this will add a significant overhead, only turn this Note: this will add a significant overhead; only turn this
on if you need to profile the system's use of these macros. on if you need to profile the system's use of these macros.
config PROFILE_ALL_BRANCHES config PROFILE_ALL_BRANCHES
@@ -305,7 +305,7 @@ config PROFILE_ALL_BRANCHES
This configuration, when enabled, will impose a great overhead This configuration, when enabled, will impose a great overhead
on the system. This should only be enabled when the system on the system. This should only be enabled when the system
is to be analyzed is to be analyzed in much detail.
endchoice endchoice
config TRACING_BRANCHES config TRACING_BRANCHES
@@ -335,7 +335,7 @@ config POWER_TRACER
depends on X86 depends on X86
select GENERIC_TRACER select GENERIC_TRACER
help help
This tracer helps developers to analyze and optimize the kernels This tracer helps developers to analyze and optimize the kernel's
power management decisions, specifically the C-state and P-state power management decisions, specifically the C-state and P-state
behavior. behavior.
@@ -391,14 +391,14 @@ config HW_BRANCH_TRACER
select GENERIC_TRACER select GENERIC_TRACER
help help
This tracer records all branches on the system in a circular This tracer records all branches on the system in a circular
buffer giving access to the last N branches for each cpu. buffer, giving access to the last N branches for each cpu.
config KMEMTRACE config KMEMTRACE
bool "Trace SLAB allocations" bool "Trace SLAB allocations"
select GENERIC_TRACER select GENERIC_TRACER
help help
kmemtrace provides tracing for slab allocator functions, such as kmemtrace provides tracing for slab allocator functions, such as
kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected kmalloc, kfree, kmem_cache_alloc, kmem_cache_free, etc. Collected
data is then fed to the userspace application in order to analyse data is then fed to the userspace application in order to analyse
allocation hotspots, internal fragmentation and so on, making it allocation hotspots, internal fragmentation and so on, making it
possible to see how well an allocator performs, as well as debug possible to see how well an allocator performs, as well as debug
@@ -417,15 +417,15 @@ config WORKQUEUE_TRACER
bool "Trace workqueues" bool "Trace workqueues"
select GENERIC_TRACER select GENERIC_TRACER
help help
The workqueue tracer provides some statistical informations The workqueue tracer provides some statistical information
about each cpu workqueue thread such as the number of the about each cpu workqueue thread such as the number of the
works inserted and executed since their creation. It can help works inserted and executed since their creation. It can help
to evaluate the amount of work each of them have to perform. to evaluate the amount of work each of them has to perform.
For example it can help a developer to decide whether he should For example it can help a developer to decide whether he should
choose a per cpu workqueue instead of a singlethreaded one. choose a per-cpu workqueue instead of a singlethreaded one.
config BLK_DEV_IO_TRACE config BLK_DEV_IO_TRACE
bool "Support for tracing block io actions" bool "Support for tracing block IO actions"
depends on SYSFS depends on SYSFS
depends on BLOCK depends on BLOCK
select RELAY select RELAY
@@ -456,15 +456,15 @@ config KPROBE_EVENT
select TRACING select TRACING
default y default y
help help
This allows the user to add tracing events (similar to tracepoints) on the fly This allows the user to add tracing events (similar to tracepoints)
via the ftrace interface. See Documentation/trace/kprobetrace.txt on the fly via the ftrace interface. See
for more details. Documentation/trace/kprobetrace.txt for more details.
Those events can be inserted wherever kprobes can probe, and record Those events can be inserted wherever kprobes can probe, and record
various register and memory values. various register and memory values.
This option is also required by perf-probe subcommand of perf tools. If This option is also required by perf-probe subcommand of perf tools.
you want to use perf tools, this option is strongly recommended. If you want to use perf tools, this option is strongly recommended.
config DYNAMIC_FTRACE config DYNAMIC_FTRACE
bool "enable/disable ftrace tracepoints dynamically" bool "enable/disable ftrace tracepoints dynamically"
@@ -472,32 +472,32 @@ config DYNAMIC_FTRACE
depends on HAVE_DYNAMIC_FTRACE depends on HAVE_DYNAMIC_FTRACE
default y default y
help help
This option will modify all the calls to ftrace dynamically This option will modify all the calls to ftrace dynamically
(will patch them out of the binary image and replaces them (will patch them out of the binary image and replace them
with a No-Op instruction) as they are called. A table is with a No-Op instruction) as they are called. A table is
created to dynamically enable them again. created to dynamically enable them again.
This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
has native performance as long as no tracing is active. otherwise has native performance as long as no tracing is active.
The changes to the code are done by a kernel thread that The changes to the code are done by a kernel thread that
wakes up once a second and checks to see if any ftrace calls wakes up once a second and checks to see if any ftrace calls
were made. If so, it runs stop_machine (stops all CPUS) were made. If so, it runs stop_machine (stops all CPUS)
and modifies the code to jump over the call to ftrace. and modifies the code to jump over the call to ftrace.
config FUNCTION_PROFILER config FUNCTION_PROFILER
bool "Kernel function profiler" bool "Kernel function profiler"
depends on FUNCTION_TRACER depends on FUNCTION_TRACER
default n default n
help help
This option enables the kernel function profiler. A file is created This option enables the kernel function profiler. A file is created
in debugfs called function_profile_enabled which defaults to zero. in debugfs called function_profile_enabled which defaults to zero.
When a 1 is echoed into this file profiling begins, and when a When a 1 is echoed into this file profiling begins, and when a
zero is entered, profiling stops. A file in the trace_stats zero is entered, profiling stops. A "functions" file is created in
directory called functions, that show the list of functions that the trace_stats directory; this file shows the list of functions that
have been hit and their counters. have been hit and their counters.
If in doubt, say N If in doubt, say N.
config FTRACE_MCOUNT_RECORD config FTRACE_MCOUNT_RECORD
def_bool y def_bool y
@@ -556,8 +556,8 @@ config RING_BUFFER_BENCHMARK
tristate "Ring buffer benchmark stress tester" tristate "Ring buffer benchmark stress tester"
depends on RING_BUFFER depends on RING_BUFFER
help help
This option creates a test to stress the ring buffer and bench mark it. This option creates a test to stress the ring buffer and benchmark it.
It creates its own ring buffer such that it will not interfer with It creates its own ring buffer such that it will not interfere with
any other users of the ring buffer (such as ftrace). It then creates any other users of the ring buffer (such as ftrace). It then creates
a producer and consumer that will run for 10 seconds and sleep for a producer and consumer that will run for 10 seconds and sleep for
10 seconds. Each interval it will print out the number of events 10 seconds. Each interval it will print out the number of events
@@ -566,7 +566,7 @@ config RING_BUFFER_BENCHMARK
It does not disable interrupts or raise its priority, so it may be It does not disable interrupts or raise its priority, so it may be
affected by processes that are running. affected by processes that are running.
If unsure, say N If unsure, say N.
endif # FTRACE endif # FTRACE

2
kernel/trace/trace.c
View File

@@ -3949,7 +3949,7 @@ trace_options_write(struct file *filp, const char __user *ubuf, size_t cnt,
if (!!(topt->flags->val & topt->opt->bit) != val) { if (!!(topt->flags->val & topt->opt->bit) != val) {
mutex_lock(&trace_types_lock); mutex_lock(&trace_types_lock);
ret = __set_tracer_option(current_trace, topt->flags, ret = __set_tracer_option(current_trace, topt->flags,
topt->opt, val); topt->opt, !val);
mutex_unlock(&trace_types_lock); mutex_unlock(&trace_types_lock);
if (ret) if (ret)
return ret; return ret;

7
kernel/trace/trace_export.c
View File

@@ -158,7 +158,8 @@ ftrace_format_##name(struct ftrace_event_call *unused, \
BUILD_BUG_ON(len > MAX_FILTER_STR_VAL); \ BUILD_BUG_ON(len > MAX_FILTER_STR_VAL); \
ret = trace_define_field(event_call, #type "[" #len "]", #item, \ ret = trace_define_field(event_call, #type "[" #len "]", #item, \
offsetof(typeof(field), item), \ offsetof(typeof(field), item), \
sizeof(field.item), 0, FILTER_OTHER); \ sizeof(field.item), \
is_signed_type(type), FILTER_OTHER); \
if (ret) \ if (ret) \
return ret; return ret;
@@ -168,8 +169,8 @@ ftrace_format_##name(struct ftrace_event_call *unused, \
ret = trace_define_field(event_call, #type "[" #len "]", #item, \ ret = trace_define_field(event_call, #type "[" #len "]", #item, \
offsetof(typeof(field), \ offsetof(typeof(field), \
container.item), \ container.item), \
sizeof(field.container.item), 0, \ sizeof(field.container.item), \
FILTER_OTHER); \ is_signed_type(type), FILTER_OTHER); \
if (ret) \ if (ret) \
return ret; return ret;

7
kernel/trace/trace_kprobe.c
View File

@@ -1201,10 +1201,11 @@ static int __probe_event_show_format(struct trace_seq *s,
#undef SHOW_FIELD #undef SHOW_FIELD
#define SHOW_FIELD(type, item, name) \ #define SHOW_FIELD(type, item, name) \
do { \ do { \
ret = trace_seq_printf(s, "\tfield: " #type " %s;\t" \ ret = trace_seq_printf(s, "\tfield:" #type " %s;\t" \
"offset:%u;\tsize:%u;\n", name, \ "offset:%u;\tsize:%u;\tsigned:%d;\n", name,\
(unsigned int)offsetof(typeof(field), item),\ (unsigned int)offsetof(typeof(field), item),\
(unsigned int)sizeof(type)); \ (unsigned int)sizeof(type), \
is_signed_type(type)); \
if (!ret) \ if (!ret) \
return 0; \ return 0; \
} while (0) } while (0)

152
kernel/trace/trace_ksym.c
View File

@@ -26,12 +26,13 @@
#include <linux/fs.h> #include <linux/fs.h>
#include "trace_output.h" #include "trace_output.h"
#include "trace_stat.h"
#include "trace.h" #include "trace.h"
#include <linux/hw_breakpoint.h> #include <linux/hw_breakpoint.h>
#include <asm/hw_breakpoint.h> #include <asm/hw_breakpoint.h>
#include <asm/atomic.h>
/* /*
* For now, let us restrict the no. of symbols traced simultaneously to number * For now, let us restrict the no. of symbols traced simultaneously to number
* of available hardware breakpoint registers. * of available hardware breakpoint registers.
@@ -44,7 +45,7 @@ struct trace_ksym {
struct perf_event **ksym_hbp; struct perf_event **ksym_hbp;
struct perf_event_attr attr; struct perf_event_attr attr;
#ifdef CONFIG_PROFILE_KSYM_TRACER #ifdef CONFIG_PROFILE_KSYM_TRACER
unsigned long counter; atomic64_t counter;
#endif #endif
struct hlist_node ksym_hlist; struct hlist_node ksym_hlist;
}; };
@@ -69,9 +70,8 @@ void ksym_collect_stats(unsigned long hbp_hit_addr)
rcu_read_lock(); rcu_read_lock();
hlist_for_each_entry_rcu(entry, node, &ksym_filter_head, ksym_hlist) { hlist_for_each_entry_rcu(entry, node, &ksym_filter_head, ksym_hlist) {
if ((entry->attr.bp_addr == hbp_hit_addr) && if (entry->attr.bp_addr == hbp_hit_addr) {
(entry->counter <= MAX_UL_INT)) { atomic64_inc(&entry->counter);
entry->counter++;
break; break;
} }
} }
@@ -197,7 +197,6 @@ int process_new_ksym_entry(char *ksymname, int op, unsigned long addr)
entry->attr.bp_addr = addr; entry->attr.bp_addr = addr;
entry->attr.bp_len = HW_BREAKPOINT_LEN_4; entry->attr.bp_len = HW_BREAKPOINT_LEN_4;
ret = -EAGAIN;
entry->ksym_hbp = register_wide_hw_breakpoint(&entry->attr, entry->ksym_hbp = register_wide_hw_breakpoint(&entry->attr,
ksym_hbp_handler); ksym_hbp_handler);
@@ -300,8 +299,8 @@ static ssize_t ksym_trace_filter_write(struct file *file,
* 2: echo 0 > ksym_trace_filter * 2: echo 0 > ksym_trace_filter
* 3: echo "*:---" > ksym_trace_filter * 3: echo "*:---" > ksym_trace_filter
*/ */
if (!buf[0] || !strcmp(buf, "0") || if (!input_string[0] || !strcmp(input_string, "0") ||
!strcmp(buf, "*:---")) { !strcmp(input_string, "*:---")) {
__ksym_trace_reset(); __ksym_trace_reset();
ret = 0; ret = 0;
goto out; goto out;
@@ -444,102 +443,77 @@ struct tracer ksym_tracer __read_mostly =
.print_line = ksym_trace_output .print_line = ksym_trace_output
}; };
__init static int init_ksym_trace(void)
{
struct dentry *d_tracer;
struct dentry *entry;
d_tracer = tracing_init_dentry();
ksym_filter_entry_count = 0;
entry = debugfs_create_file("ksym_trace_filter", 0644, d_tracer,
NULL, &ksym_tracing_fops);
if (!entry)
pr_warning("Could not create debugfs "
"'ksym_trace_filter' file\n");
return register_tracer(&ksym_tracer);
}
device_initcall(init_ksym_trace);
#ifdef CONFIG_PROFILE_KSYM_TRACER #ifdef CONFIG_PROFILE_KSYM_TRACER
static int ksym_tracer_stat_headers(struct seq_file *m) static int ksym_profile_show(struct seq_file *m, void *v)
{ {
seq_puts(m, " Access Type "); struct hlist_node *node;
seq_puts(m, " Symbol Counter\n");
seq_puts(m, " ----------- ");
seq_puts(m, " ------ -------\n");
return 0;
}
static int ksym_tracer_stat_show(struct seq_file *m, void *v)
{
struct hlist_node *stat = v;
struct trace_ksym *entry; struct trace_ksym *entry;
int access_type = 0; int access_type = 0;
char fn_name[KSYM_NAME_LEN]; char fn_name[KSYM_NAME_LEN];
entry = hlist_entry(stat, struct trace_ksym, ksym_hlist); seq_puts(m, " Access Type ");
seq_puts(m, " Symbol Counter\n");
seq_puts(m, " ----------- ");
seq_puts(m, " ------ -------\n");
access_type = entry->attr.bp_type; rcu_read_lock();
hlist_for_each_entry_rcu(entry, node, &ksym_filter_head, ksym_hlist) {
switch (access_type) { access_type = entry->attr.bp_type;
case HW_BREAKPOINT_R:
seq_puts(m, " R "); switch (access_type) {
break; case HW_BREAKPOINT_R:
case HW_BREAKPOINT_W: seq_puts(m, " R ");
seq_puts(m, " W "); break;
break; case HW_BREAKPOINT_W:
case HW_BREAKPOINT_R | HW_BREAKPOINT_W: seq_puts(m, " W ");
seq_puts(m, " RW "); break;
break; case HW_BREAKPOINT_R | HW_BREAKPOINT_W:
default: seq_puts(m, " RW ");
seq_puts(m, " NA "); break;
default:
seq_puts(m, " NA ");
}
if (lookup_symbol_name(entry->attr.bp_addr, fn_name) >= 0)
seq_printf(m, " %-36s", fn_name);
else
seq_printf(m, " %-36s", "<NA>");
seq_printf(m, " %15llu\n",
(unsigned long long)atomic64_read(&entry->counter));
} }
rcu_read_unlock();
if (lookup_symbol_name(entry->attr.bp_addr, fn_name) >= 0)
seq_printf(m, " %-36s", fn_name);
else
seq_printf(m, " %-36s", "<NA>");
seq_printf(m, " %15lu\n", entry->counter);
return 0; return 0;
} }
static void *ksym_tracer_stat_start(struct tracer_stat *trace) static int ksym_profile_open(struct inode *node, struct file *file)
{ {
return ksym_filter_head.first; return single_open(file, ksym_profile_show, NULL);
} }
static void * static const struct file_operations ksym_profile_fops = {
ksym_tracer_stat_next(void *v, int idx) .open = ksym_profile_open,
{ .read = seq_read,
struct hlist_node *stat = v; .llseek = seq_lseek,
.release = single_release,
return stat->next;
}
static struct tracer_stat ksym_tracer_stats = {
.name = "ksym_tracer",
.stat_start = ksym_tracer_stat_start,
.stat_next = ksym_tracer_stat_next,
.stat_headers = ksym_tracer_stat_headers,
.stat_show = ksym_tracer_stat_show
}; };
__init static int ksym_tracer_stat_init(void)
{
int ret;
ret = register_stat_tracer(&ksym_tracer_stats);
if (ret) {
printk(KERN_WARNING "Warning: could not register "
"ksym tracer stats\n");
return 1;
}
return 0;
}
fs_initcall(ksym_tracer_stat_init);
#endif /* CONFIG_PROFILE_KSYM_TRACER */ #endif /* CONFIG_PROFILE_KSYM_TRACER */
__init static int init_ksym_trace(void)
{
struct dentry *d_tracer;
d_tracer = tracing_init_dentry();
trace_create_file("ksym_trace_filter", 0644, d_tracer,
NULL, &ksym_tracing_fops);
#ifdef CONFIG_PROFILE_KSYM_TRACER
trace_create_file("ksym_profile", 0444, d_tracer,
NULL, &ksym_profile_fops);
#endif
return register_tracer(&ksym_tracer);
}
device_initcall(init_ksym_trace);