tracing: comment and function graph latency format
Steven Rostedt
maintainerIngo,
I rebased the branch to remove the module ref count change. That will be a separate patch push to make sure everyone agrees with it.
Please pull the latest tip/tracing/core2 tree, which can be found at:
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
tip/tracing/core2
Steven Rostedt (2):
x86/tracing: comment need for atomic nop
tracing: add latency format to function_graph tracer
----
arch/x86/include/asm/nops.h | 2 + kernel/trace/trace_functions_graph.c | 74 +++++++++++++++++++++++++++++++--- 2 files changed, 70 insertions(+), 6 deletions(-)
Steven Rostedt
maintainerPATCH 1/2 x86/tracing: comment need for atomic nop
From: Steven Rostedt <[email protected]>
The dynamic function tracer relys on the macro P6_NOP5 always being an atomic NOP. If for some reason it is changed to be two operations (like a nop2 nop3) it can faults within the kernel when the function tracer modifies the code.
This patch adds a comment to note that the P6_NOPs are expected to be atomic. This will hopefully prevent anyone from changing that.
--- arch/x86/include/asm/nops.h | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-)
arch/x86/include/asm/nops.h +2 -0
@@ -65,6 +65,8 @@6: osp nopl 0x00(%eax,%eax,1) 7: nopl 0x00000000(%eax) 8: nopl 0x00000000(%eax,%eax,1) +Note: All the above are assumed to be a single instruction. +There is kernel code that depends on this. */ #define P6_NOP1 GENERIC_NOP1#define P6_NOP2 ".byte 0x66,0x90\n"
Signature · 3 lines
1.6.3.3 --
Steven Rostedt
maintainerPATCH 2/2 tracing: add latency format to function_graph tracer
In reply to Steven RostedtFrom: Steven Rostedt <[email protected]>
While debugging something with the function_graph tracer, I found the need to see the preempt count of the traces. Unfortunately, since the function graph tracer has its own output formatting, it does not honor the latency-format option.
This patch makes the function_graph tracer honor the latency-format option, but still keeps control of the output. But now we have the same details that the latency-format supplies.
# tracer: function_graph
#
# _-----=> irqs-off
# / _----=> need-resched
# | / _---=> hardirq/softirq
# || / _--=> preempt-depth
# ||| /
# ||||
# CPU|||| DURATION FUNCTION CALLS
# | |||| | | | | | |
3) d..1 1.333 us | idle_cpu();
3) d.h1 | tick_check_idle() {
3) d.h1 0.550 us | tick_check_oneshot_broadcast();
3) d.h1 | tick_nohz_stop_idle() {
3) d.h1 | ktime_get() {
3) d.h1 | ktime_get_ts() {--- kernel/trace/trace_functions_graph.c | 74 +++++++++++++++++++++++++++++++--- 1 files changed, 68 insertions(+), 6 deletions(-)
kernel/trace/trace_functions_graph.c +68 -6
@@ -364,6 +364,29 @@ print_graph_proc(struct trace_seq *s, pid_t pid) } +static enum print_line_t +print_graph_lat_fmt (struct trace_seq *s,struct trace_entry *entry) +{ +int hardirq, softirq; + + hardirq = entry->flags &TRACE_FLAG_HARDIRQ ; + softirq = entry->flags &TRACE_FLAG_SOFTIRQ ; + +if (!trace_seq_printf (s," %c%c%c" , + (entry->flags &TRACE_FLAG_IRQS_OFF ) ?'d' : + (entry->flags &TRACE_FLAG_IRQS_NOSUPPORT ) ? +'X' :'.' , + (entry->flags &TRACE_FLAG_NEED_RESCHED ) ? +'N' :'.' , + (hardirq && softirq) ?'H' : + hardirq ?'h' : softirq ?'s' :'.' )) +return 0 ; + +if (entry->preempt_count) +return trace_seq_printf (s,"%x" , entry->preempt_count); +return trace_seq_puts (s,"." ); +} +/* If the pid changed since the last trace, output this event */ static enum print_line_t verif_pid (struct trace_seq *s,pid_t pid,int cpu,struct fgraph_data *data)
@@ -521,6 +544,7 @@ print_graph_irq(struct trace_iterator *iter, unsigned long addr,if (ret ==TRACE_TYPE_PARTIAL_LINE )return TRACE_TYPE_PARTIAL_LINE ; } +/* Proc */ if (tracer_flags.val &TRACE_GRAPH_PRINT_PROC ) { ret =print_graph_proc (s, pid);
@@ -758,6 +782,13 @@ print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,return TRACE_TYPE_PARTIAL_LINE ; } +/* Latency format */ +if (trace_flags &TRACE_ITER_LATENCY_FMT ) { + ret =print_graph_lat_fmt (s, ent); +if (ret ==TRACE_TYPE_PARTIAL_LINE ) +return TRACE_TYPE_PARTIAL_LINE ; + } +return 0 ; }
@@ -952,28 +983,59 @@ print_graph_function(struct trace_iterator *iter)return TRACE_TYPE_HANDLED ; } +static void print_lat_header (struct seq_file *s) +{ +static const char spaces[] =" " /* 16 spaces */ +" " /* 4 spaces */ +" " ;/* 17 spaces */ +int size =0 ; + +if (tracer_flags.val &TRACE_GRAPH_PRINT_ABS_TIME ) + size +=16 ; +if (tracer_flags.val &TRACE_GRAPH_PRINT_CPU ) + size +=4 ; +if (tracer_flags.val &TRACE_GRAPH_PRINT_PROC ) + size +=17 ; + +seq_printf (s,"#%.*s _-----=> irqs-off \n" , size, spaces); +seq_printf (s,"#%.*s / _----=> need-resched \n" , size, spaces); +seq_printf (s,"#%.*s| / _---=> hardirq/softirq \n" , size, spaces); +seq_printf (s,"#%.*s|| / _--=> preempt-depth \n" , size, spaces); +seq_printf (s,"#%.*s||| / \n" , size, spaces); +seq_printf (s,"#%.*s|||| \n" , size, spaces); +} +static void print_graph_headers (struct seq_file *s) { +int lat = trace_flags &TRACE_ITER_LATENCY_FMT ; + +if (lat) +print_lat_header (s); +/* 1st line */ -seq_printf (s,"# " ); +seq_printf (s,"#" );if (tracer_flags.val &TRACE_GRAPH_PRINT_ABS_TIME )seq_printf (s," TIME " );if (tracer_flags.val &TRACE_GRAPH_PRINT_CPU ) -seq_printf (s,"CPU" ); +seq_printf (s," CPU" );if (tracer_flags.val &TRACE_GRAPH_PRINT_PROC ) -seq_printf (s," TASK/PID " ); +seq_printf (s," TASK/PID " ); +if (lat) +seq_printf (s,"||||" );if (tracer_flags.val &TRACE_GRAPH_PRINT_DURATION )seq_printf (s," DURATION " );seq_printf (s," FUNCTION CALLS\n" );/* 2nd line */ -seq_printf (s,"# " ); +seq_printf (s,"#" );if (tracer_flags.val &TRACE_GRAPH_PRINT_ABS_TIME )seq_printf (s," | " );if (tracer_flags.val &TRACE_GRAPH_PRINT_CPU ) -seq_printf (s,"| " ); +seq_printf (s," | " );if (tracer_flags.val &TRACE_GRAPH_PRINT_PROC ) -seq_printf (s," | | " ); +seq_printf (s," | | " ); +if (lat) +seq_printf (s,"||||" );if (tracer_flags.val &TRACE_GRAPH_PRINT_DURATION )seq_printf (s," | | " );seq_printf (s," | | | |\n" );
Signature · 3 lines
1.6.3.3 --