tracing: function graph output for preempt/irqs-off tracers
Subsystems (2)
cover tracing: function graph output for preempt/irqs-off tracers
hi,
I'm sending reworked version of the graph output support for preemptirqsoff/preemptoff/irqsoff tracers.
I made the graph output as an output event, so it could be shared within tracers - patch 1/4.
I also added raw trace output for graph tracer. I have this one around for long time and it was quite handy for investigating graph tracer issues. (patch 4/4)
attached patches:
- 1/4 adding ftrace events for graph tracer
- 2/4 graph output support for irqsoff tracer
- 3/4 graph output support for preemptirqsoff/preemptoff tracers
- 4/4 raw output for graph tracer
wbr,
jirka
--- kernel/trace/trace.c | 2 +- kernel/trace/trace.h | 11 ++- kernel/trace/trace_functions_graph.c | 117 +++++++++++++++++++---- kernel/trace/trace_irqsoff.c | 174 ++++++++++++++++++++++++++++++++-- 4 files changed, 277 insertions(+), 27 deletions(-)
PATCH 1/4 tracing: adding ftrace events for graph tracer
hi,
this patch adds ftrace events for graph tracer, so the graph output could be shared within other tracers.
wbr,
jirka
--- kernel/trace/trace.h | 3 ++- kernel/trace/trace_functions_graph.c | 25 ++++++++++++++++++++++--- 2 files changed, 24 insertions(+), 4 deletions(-)
kernel/trace/trace.h +2 -1
@@ -490,7 +490,8 @@ extern int trace_clock_id;/* Standard output formatting function used for function return traces */ #ifdef CONFIG_FUNCTION_GRAPH_TRACER -extern enum print_line_t print_graph_function (struct trace_iterator *iter); +extern enum print_line_t print_graph_function (struct trace_iterator *iter, +int flags);extern enum print_line_t trace_print_graph_duration (unsigned long long duration,struct trace_seq *s);
kernel/trace/trace_functions_graph.c +22 -3
@@ -38,7 +38,7 @@ struct fgraph_data {#define TRACE_GRAPH_PRINT_OVERHEAD 0x4#define TRACE_GRAPH_PRINT_PROC 0x8#define TRACE_GRAPH_PRINT_DURATION 0x10 -#define TRACE_GRAPH_PRINT_ABS_TIME 0X20 +#define TRACE_GRAPH_PRINT_ABS_TIME 0x20static struct tracer_opt trace_opts[] = {/* Display overruns? (for self-debug purpose) */
@@ -985,7 +985,7 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent,enum print_line_t -print_graph_function (struct trace_iterator *iter) +print_graph_function (struct trace_iterator *iter,int flags) {struct ftrace_graph_ent_entry *field;struct fgraph_data *data = iter->private;
@@ -1143,6 +1143,16 @@ static void graph_trace_close(struct trace_iterator *iter) } } +static struct trace_event graph_trace_entry_event = { + .type =TRACE_GRAPH_ENT , + .trace = print_graph_function, +}; + +static struct trace_event graph_trace_ret_event = { + .type =TRACE_GRAPH_RET , + .trace = print_graph_function, +}; +static struct tracer graph_trace__read_mostly = { .name ="function_graph" , .open = graph_trace_open,
@@ -1152,7 +1162,6 @@ static struct tracer graph_trace __read_mostly = { .wait_pipe = poll_wait_pipe, .init = graph_trace_init, .reset = graph_trace_reset, - .print_line = print_graph_function, .print_header = print_graph_headers, .flags = &tracer_flags,#ifdef CONFIG_FTRACE_SELFTEST
@@ -1164,6 +1173,16 @@ static __init int init_graph_trace(void) { max_bytes_for_cpu =snprintf (NULL ,0 ,"%d" , nr_cpu_ids -1 ); +if (!register_ftrace_event (&graph_trace_entry_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } + +if (!register_ftrace_event (&graph_trace_ret_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } +return register_tracer (&graph_trace); }
Signature · 1 line
1.6.6
Steven Rostedt
maintainerRe: [PATCH 1/4] tracing: adding ftrace events for graph tracer
Quoted 5 lines · Jiri Olsa
hi,
this patch adds ftrace events for graph tracer, so the graph output could be shared within other tracers.
Sorry but NAK to this,
What use to be:
2) + 25.633 us | } 3) ! 252.132 us | } /* do_softirq */ 2) ! 898.211 us | } /* rcu_process_callbacks */ 2) | /* softirq_exit: vec=9 [action=RCU] */ 3) 5.534 us | rcu_irq_exit(); 2) 6.405 us | preempt_schedule(); 3) 5.221 us | idle_cpu(); 2) 6.014 us | rcu_bh_qs(); 3) ! 289.729 us | } /* irq_exit */ 2) | /* softirq_entry: vec=1 [action=TIMER] */ 3) ! 860.416 us | } /* smp_apic_timer_interrupt */ 3) <========== | 2) 5.785 us | preempt_schedule();
Is now:
gdmgreeter-3382 [002] 119.609346: 2) + 25.633 us | }
pcscd-2849 [003] 119.609352: 3) ! 252.132 us | } /* do_softirq */
gdmgreeter-3382 [002] 119.609353: 2) ! 898.211 us | } /* rcu_process_callbacks */
gdmgreeter-3382 [002] 119.609357: softirq_exit: vec=9 [action=RCU]
pcscd-2849 [003] 119.609360: 3) 5.534 us | rcu_irq_exit();
gdmgreeter-3382 [002] 119.609363: 2) 6.405 us | preempt_schedule();
pcscd-2849 [003] 119.609370: 3) 5.221 us | idle_cpu();
gdmgreeter-3382 [002] 119.609376: 2) 6.014 us | rcu_bh_qs();
pcscd-2849 [003] 119.609383: 3) ! 289.729 us | } /* irq_exit */
gdmgreeter-3382 [002] 119.609387: softirq_entry: vec=1 [action=TIMER]
pcscd-2849 [003] 119.609389: 3) ! 860.416 us | } /* smp_apic_timer_interrupt */
3) <========== |
gdmgreeter-3382 [002] 119.609393: 2) 5.785 us | preempt_schedule();Keep the function graph output as is. You can still add the event callbacks but do not remove the print_line of the function graph tracer. That simply breaks what we worked hard to format nicely.
-- Steve
Quoted 85 lines
Signed-off-by Jiri Olsa <[email protected]>--- kernel/trace/trace.h | 3 ++- kernel/trace/trace_functions_graph.c | 25 ++++++++++++++++++++++--- 2 files changed, 24 insertions(+), 4 deletions(-)kernel/trace/trace.h +2
-1@@ -490,7 +490,8 @@ extern int trace_clock_id;/* Standard output formatting function used for function return traces */ #ifdef CONFIG_FUNCTION_GRAPH_TRACER -extern enum print_line_t print_graph_function (struct trace_iterator *iter); +extern enum print_line_t print_graph_function (struct trace_iterator *iter, +int flags);extern enum print_line_t trace_print_graph_duration (unsigned long long duration,struct trace_seq *s);kernel/trace/trace_functions_graph.c +22
-3@@ -38,7 +38,7 @@ struct fgraph_data {#define TRACE_GRAPH_PRINT_OVERHEAD 0x4#define TRACE_GRAPH_PRINT_PROC 0x8#define TRACE_GRAPH_PRINT_DURATION 0x10 -#define TRACE_GRAPH_PRINT_ABS_TIME 0X20 +#define TRACE_GRAPH_PRINT_ABS_TIME 0x20static struct tracer_opt trace_opts[] = {/* Display overruns? (for self-debug purpose) */ @@ -985,7 +985,7 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent,enum print_line_t -print_graph_function (struct trace_iterator *iter) +print_graph_function (struct trace_iterator *iter,int flags) {struct ftrace_graph_ent_entry *field;struct fgraph_data *data = iter->private;@@ -1143,6 +1143,16 @@ static void graph_trace_close(struct trace_iterator *iter) } } +static struct trace_event graph_trace_entry_event = { + .type =TRACE_GRAPH_ENT , + .trace = print_graph_function, +}; + +static struct trace_event graph_trace_ret_event = { + .type =TRACE_GRAPH_RET , + .trace = print_graph_function, +}; +static struct tracer graph_trace__read_mostly = { .name ="function_graph" , .open = graph_trace_open,@@ -1152,7 +1162,6 @@ static struct tracer graph_trace __read_mostly = { .wait_pipe = poll_wait_pipe, .init = graph_trace_init, .reset = graph_trace_reset, - .print_line = print_graph_function, .print_header = print_graph_headers, .flags = &tracer_flags,#ifdef CONFIG_FTRACE_SELFTEST @@ -1164,6 +1173,16 @@ static __init int init_graph_trace(void) { max_bytes_for_cpu =snprintf (NULL ,0 ,"%d" , nr_cpu_ids -1 ); +if (!register_ftrace_event (&graph_trace_entry_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } + +if (!register_ftrace_event (&graph_trace_ret_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } +return register_tracer (&graph_trace); }
Re: [PATCH 1/4] tracing: adding ftrace events for graph tracer
Quoted 8 lines · Steven Rostedt
Quoted 5 lines · Jiri Olsa
hi,
this patch adds ftrace events for graph tracer, so the graph output could be shared within other tracers.
Sorry but NAK to this,
right, I haven't realized this to be a problem,
I'll send new version shortly
wbr,
jirka
Quoted 128 lines
What use to be:
2) + 25.633 us | } 3) ! 252.132 us | } /* do_softirq */ 2) ! 898.211 us | } /* rcu_process_callbacks */ 2) | /* softirq_exit: vec=9 [action=RCU] */ 3) 5.534 us | rcu_irq_exit(); 2) 6.405 us | preempt_schedule(); 3) 5.221 us | idle_cpu(); 2) 6.014 us | rcu_bh_qs(); 3) ! 289.729 us | } /* irq_exit */ 2) | /* softirq_entry: vec=1 [action=TIMER] */ 3) ! 860.416 us | } /* smp_apic_timer_interrupt */ 3) <========== | 2) 5.785 us | preempt_schedule();Is now:
gdmgreeter-3382 [002] 119.609346: 2) + 25.633 us | } pcscd-2849 [003] 119.609352: 3) ! 252.132 us | } /* do_softirq */ gdmgreeter-3382 [002] 119.609353: 2) ! 898.211 us | } /* rcu_process_callbacks */ gdmgreeter-3382 [002] 119.609357: softirq_exit: vec=9 [action=RCU] pcscd-2849 [003] 119.609360: 3) 5.534 us | rcu_irq_exit(); gdmgreeter-3382 [002] 119.609363: 2) 6.405 us | preempt_schedule(); pcscd-2849 [003] 119.609370: 3) 5.221 us | idle_cpu(); gdmgreeter-3382 [002] 119.609376: 2) 6.014 us | rcu_bh_qs(); pcscd-2849 [003] 119.609383: 3) ! 289.729 us | } /* irq_exit */ gdmgreeter-3382 [002] 119.609387: softirq_entry: vec=1 [action=TIMER] pcscd-2849 [003] 119.609389: 3) ! 860.416 us | } /* smp_apic_timer_interrupt */ 3) <========== | gdmgreeter-3382 [002] 119.609393: 2) 5.785 us | preempt_schedule();Keep the function graph output as is. You can still add the event callbacks but do not remove the
print_lineof the function graph tracer. That simply breaks what we worked hard to format nicely.-- Steve
Quoted 85 lines
Signed-off-by Jiri Olsa <[email protected]>--- kernel/trace/trace.h | 3 ++- kernel/trace/trace_functions_graph.c | 25 ++++++++++++++++++++++--- 2 files changed, 24 insertions(+), 4 deletions(-)kernel/trace/trace.h +2
-1@@ -490,7 +490,8 @@ extern int trace_clock_id;/* Standard output formatting function used for function return traces */ #ifdef CONFIG_FUNCTION_GRAPH_TRACER -extern enum print_line_t print_graph_function (struct trace_iterator *iter); +extern enum print_line_t print_graph_function (struct trace_iterator *iter, +int flags);extern enum print_line_t trace_print_graph_duration (unsigned long long duration,struct trace_seq *s);kernel/trace/trace_functions_graph.c +22
-3@@ -38,7 +38,7 @@ struct fgraph_data {#define TRACE_GRAPH_PRINT_OVERHEAD 0x4#define TRACE_GRAPH_PRINT_PROC 0x8#define TRACE_GRAPH_PRINT_DURATION 0x10 -#define TRACE_GRAPH_PRINT_ABS_TIME 0X20 +#define TRACE_GRAPH_PRINT_ABS_TIME 0x20static struct tracer_opt trace_opts[] = {/* Display overruns? (for self-debug purpose) */ @@ -985,7 +985,7 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent,enum print_line_t -print_graph_function (struct trace_iterator *iter) +print_graph_function (struct trace_iterator *iter,int flags) {struct ftrace_graph_ent_entry *field;struct fgraph_data *data = iter->private;@@ -1143,6 +1143,16 @@ static void graph_trace_close(struct trace_iterator *iter) } } +static struct trace_event graph_trace_entry_event = { + .type =TRACE_GRAPH_ENT , + .trace = print_graph_function, +}; + +static struct trace_event graph_trace_ret_event = { + .type =TRACE_GRAPH_RET , + .trace = print_graph_function, +}; +static struct tracer graph_trace__read_mostly = { .name ="function_graph" , .open = graph_trace_open,@@ -1152,7 +1162,6 @@ static struct tracer graph_trace __read_mostly = { .wait_pipe = poll_wait_pipe, .init = graph_trace_init, .reset = graph_trace_reset, - .print_line = print_graph_function, .print_header = print_graph_headers, .flags = &tracer_flags,#ifdef CONFIG_FTRACE_SELFTEST @@ -1164,6 +1173,16 @@ static __init int init_graph_trace(void) { max_bytes_for_cpu =snprintf (NULL ,0 ,"%d" , nr_cpu_ids -1 ); +if (!register_ftrace_event (&graph_trace_entry_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } + +if (!register_ftrace_event (&graph_trace_ret_event)) { +pr_warning ("Warning: could not register graph trace events\n" ); +return 1 ; + } +return register_tracer (&graph_trace); }
PATCH 2/4 tracing: graph output support for irqsoff tracer
In reply to Jiri Olsahi,
when I was using irqsoff tracer, I thought it could be any use to see the disabled IRQs flow output same way as function graph output.
The graph output is enabled by setting new 'display-graph' trace option of irqsoff tracer.
It looks like this:
[root@dell-pe1950-05 tracing]# echo irqsoff-graph > ./trace_options [root@dell-pe1950-05 tracing]# cat ./trace
# tracer: irqsoff
#
# _-----=> irqs-off
# / _----=> need-resched
# | / _---=> hardirq/softirq
# || / _--=> preempt-depth
# ||| / _-=> lock-depth
# |||| /
# CPU||||| DURATION FUNCTION CALLS
# | ||||| | | | | | |
hald-add-1048 3d...0 0us : _raw_spin_lock_irqsave <-scsi_dispatch_cmd
3) d...0 | ata_scsi_queuecmd() {
3) d...0 0.318 us | _raw_spin_lock();
3) d...0 | ata_scsi_find_dev() {
3) d...0 | __ata_scsi_find_dev() {
3) d...0 0.273 us | ata_find_dev();
3) d...0 0.798 us | }
3) d...0 1.419 us | }
3) d...0 | __ata_scsi_queuecmd() {
3) d...0 | ata_qc_new_init() {
3) d...0 | ata_qc_reinit() {
3) d...0 0.258 us | ata_tf_init();
3) d...0 0.784 us | }
3) d...0 1.441 us | }
3) d...0 0.348 us | atapi_xlat();
3) d...0 | ata_qc_issue() {...
SNIP
...
3) d...0+ 31.244 us | }
3) d...0+ 32.393 us | }
3) d...0+ 99.146 us | }
3) d...0! 100.455 us | }
3) d...0! 103.669 us | }
3) d...0 0.267 us | _raw_spin_lock();
3) d...0! 107.264 us | }
3) d...0 | spin_unlock_irqrestore() {
3) d...0 | _raw_spin_unlock_irqrestore() {hald-add-1048 3d...0 109us : _raw_spin_unlock_irqrestore <-return_to_handler
hald-add-1048 3d...0 109us : trace_hardirqs_on <-return_to_handler
hald-add-1048 3d...0 109us : <stack trace>
=> trace_hardirqs_on
=> _raw_spin_unlock_irqrestore
=> return_to_handler
=> scsi_dispatch_cmd
=> return_to_handler
=> __generic_unplug_device
=> scsi_request_fn
=> return_to_handler
[root@dell-pe1950-05 tracing]#
wbr,
jirka
--- kernel/trace/trace.h | 8 ++ kernel/trace/trace_functions_graph.c | 25 ++---- kernel/trace/trace_irqsoff.c | 162 ++++++++++++++++++++++++++++++++-- 3 files changed, 172 insertions(+), 23 deletions(-)
kernel/trace/trace.h +8 -0
@@ -494,6 +494,14 @@ extern enum print_line_t print_graph_function(struct trace_iterator *iter, int flags);extern enum print_line_t trace_print_graph_duration (unsigned long long duration,struct trace_seq *s); +void graph_trace_open (struct trace_iterator *iter); +void graph_trace_close (struct trace_iterator *iter); +int __trace_graph_entry (struct trace_array *tr,struct ftrace_graph_ent *trace, +unsigned long flags,int pc); +void __trace_graph_return (struct trace_array *tr, +struct ftrace_graph_ret *trace, +unsigned long flags,int pc); +#ifdef CONFIG_DYNAMIC_FTRACE /* TODO: make this variable */
kernel/trace/trace_functions_graph.c +9 -16
@@ -177,7 +177,7 @@ unsigned long ftrace_return_to_handler(unsigned long frame_pointer)return ret; } -static int __trace_graph_entry (struct trace_array *tr, +int __trace_graph_entry (struct trace_array *tr,struct ftrace_graph_ent *trace,unsigned long flags,int pc)
@@ -236,7 +236,7 @@ int trace_graph_entry(struct ftrace_graph_ent *trace)return ret; } -static void __trace_graph_return (struct trace_array *tr, +void __trace_graph_return (struct trace_array *tr,struct ftrace_graph_ret *trace,unsigned long flags,int pc)
@@ -911,9 +911,7 @@ static enum print_line_t print_graph_comment(struct trace_seq *s, struct trace_entry *ent, struct trace_iterator *iter) { - unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); struct fgraph_data *data = iter->private; -struct trace_event *event;int depth =0 ;int ret;int i;
@@ -960,14 +958,6 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent,if (ret !=TRACE_TYPE_HANDLED )return ret;break ; -default : - event =ftrace_find_event (ent->type); -if (!event) -return TRACE_TYPE_UNHANDLED ; - - ret = event->trace (iter, sym_flags); -if (ret !=TRACE_TYPE_HANDLED ) -return ret; }/* Strip ending newline */
@@ -1033,8 +1023,11 @@ print_graph_function(struct trace_iterator *iter, int flags)trace_assign_type (field, entry);return print_graph_return (&field->ret, s, entry, iter); } -default : +case TRACE_BPRINT : +case TRACE_PRINT :return print_graph_comment (s, entry, iter); +default : +return TRACE_TYPE_UNHANDLED ; }return TRACE_TYPE_HANDLED ;
@@ -1062,7 +1055,7 @@ static void print_lat_header(struct seq_file *s)seq_printf (s,"#%.*s|||| / \n" , size, spaces); } -static void print_graph_headers (struct seq_file *s) +void print_graph_headers (struct seq_file *s) {int lat = trace_flags &TRACE_ITER_LATENCY_FMT ;
@@ -1098,7 +1091,7 @@ static void print_graph_headers(struct seq_file *s)seq_printf (s," | | | |\n" ); } -static void graph_trace_open (struct trace_iterator *iter) +void graph_trace_open (struct trace_iterator *iter) {/* pid and depth on the last trace processed */ struct fgraph_data *data;
@@ -1133,7 +1126,7 @@ static void graph_trace_open(struct trace_iterator *iter)pr_warning ("function graph tracer: not enough memory\n" ); } -static void graph_trace_close (struct trace_iterator *iter) +void graph_trace_close (struct trace_iterator *iter) {struct fgraph_data *data = iter->private;
kernel/trace/trace_irqsoff.c +155 -7
214 lines not on this page · raw message
Signature · 1 line
1.6.6
PATCH 3/4 tracing: graph output support for preemptirqsoff/preemptoff tracers
In reply to Jiri Olsahi,
adding support for graph output for preemptirqsoff/preemptoff tracers.
wbr,
jirka
--- kernel/trace/trace_irqsoff.c | 22 +++++++++++++++++----- 1 files changed, 17 insertions(+), 5 deletions(-)
kernel/trace/trace_irqsoff.c +17 -5
@@ -58,12 +58,12 @@ irq_trace(void)# define irq_trace () (0)#endif -#define TRACE_IRQSOFF_GRAPH 1 +#define TRACE_DISPLAY_GRAPH 1static struct tracer_opt trace_opts[] = { #ifdef CONFIG_FUNCTION_GRAPH_TRACER/* display latency trace as call graph */ - {TRACER_OPT (irqsoff-graph,TRACE_IRQSOFF_GRAPH ) }, + {TRACER_OPT (display-graph,TRACE_DISPLAY_GRAPH ) }, #endif { }/* Empty entry */ };
@@ -73,7 +73,7 @@ static struct tracer_flags tracer_flags = { .opts = trace_opts, }; -#define is_graph () (tracer_flags.val & TRACE_IRQSOFF_GRAPH) +#define is_graph () (tracer_flags.val & TRACE_DISPLAY_GRAPH)/* * Sequence count - we record it when starting a measurement and
@@ -133,7 +133,7 @@ static int irqsoff_set_flag(u32 old_flags, u32 bit, int set) {int cpu; -if (!(bit &TRACE_IRQSOFF_GRAPH )) +if (!(bit &TRACE_DISPLAY_GRAPH ))return -EINVAL ;if (!(is_graph () ^ set))
@@ -550,8 +550,8 @@ static struct tracer irqsoff_tracer __read_mostly = .reset = irqsoff_tracer_reset, .start = irqsoff_tracer_start, .stop = irqsoff_tracer_stop, - .flags = &tracer_flags, .print_max = 1, + .flags = &tracer_flags, .set_flag = irqsoff_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_irqsoff,
@@ -583,9 +583,15 @@ static struct tracer preemptoff_tracer __read_mostly = .start = irqsoff_tracer_start, .stop = irqsoff_tracer_stop, .print_max = 1, + .flags = &tracer_flags, + .set_flag = irqsoff_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_preemptoff, #endif +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + .open = irqsoff_trace_open, + .close = irqsoff_trace_close, +#endif };# define register_preemptoff (trace) register_tracer(&trace)#else
@@ -611,9 +617,15 @@ static struct tracer preemptirqsoff_tracer __read_mostly = .start = irqsoff_tracer_start, .stop = irqsoff_tracer_stop, .print_max = 1, + .flags = &tracer_flags, + .set_flag = irqsoff_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_preemptirqsoff, #endif +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + .open = irqsoff_trace_open, + .close = irqsoff_trace_close, +#endif };# define register_preemptirqsoff (trace) register_tracer(&trace)
Signature · 1 line
1.6.6
PATCH 4/4 tracing: raw output for graph tracer
In reply to Jiri Olsahi,
this patch adds raw trace output for graph tracer.
I have this one around for long time and it was quite handy for investigating graph tracer issues.
wbr,
jirka
--- kernel/trace/trace.c | 2 +- kernel/trace/trace_functions_graph.c | 69 ++++++++++++++++++++++++++++++++++ 2 files changed, 70 insertions(+), 1 deletions(-)
kernel/trace/trace.c +1 -1
@@ -1900,7 +1900,7 @@ static enum print_line_t print_raw_fmt(struct trace_iterator *iter) entry = iter->ent;if (trace_flags &TRACE_ITER_CONTEXT_INFO ) { -if (!trace_seq_printf (s,"%d %d %llu " , +if (!trace_seq_printf (s," %6d %3d %20llu " , entry->pid, iter->cpu, iter->ts)) goto partial; }
kernel/trace/trace_functions_graph.c +69 -0
101 lines not on this page · raw message
Signature · 1 line
1.6.6
hi,
any feedback?
thanks,
jirka
Quoted 32 lines · Jiri Olsa
hi,
I'm sending reworked version of the graph output support for preemptirqsoff/preemptoff/irqsoff tracers.
I made the graph output as an output event, so it could be shared within tracers - patch 1/4.
I also added raw trace output for graph tracer. I have this one around for long time and it was quite handy for investigating graph tracer issues. (patch 4/4)
attached patches:
- 1/4 adding ftrace events for graph tracer
- 2/4 graph output support for irqsoff tracer
- 3/4 graph output support for preemptirqsoff/preemptoff tracers
- 4/4 raw output for graph tracerwbr,
jirka--- kernel/trace/trace.c | 2 +- kernel/trace/trace.h | 11 ++- kernel/trace/trace_functions_graph.c | 117 +++++++++++++++++++---- kernel/trace/trace_irqsoff.c | 174 ++++++++++++++++++++++++++++++++-- 4 files changed, 277 insertions(+), 27 deletions(-)--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Steven Rostedt
maintainerQuoted 2 lines · Jiri Olsa
hi,
any feedback?
Ah, I forgot about this patch set. I'll take a look at it today.
Thanks,
-- Steve
Quoted 37 lines
thanks,
jirkaQuoted 32 lines · Jiri Olsa
hi,
I'm sending reworked version of the graph output support for preemptirqsoff/preemptoff/irqsoff tracers.
I made the graph output as an output event, so it could be shared within tracers - patch 1/4.
I also added raw trace output for graph tracer. I have this one around for long time and it was quite handy for investigating graph tracer issues. (patch 4/4)
attached patches:
- 1/4 adding ftrace events for graph tracer
- 2/4 graph output support for irqsoff tracer
- 3/4 graph output support for preemptirqsoff/preemptoff tracers
- 4/4 raw output for graph tracerwbr,
jirka--- kernel/trace/trace.c | 2 +- kernel/trace/trace.h | 11 ++- kernel/trace/trace_functions_graph.c | 117 +++++++++++++++++++---- kernel/trace/trace_irqsoff.c | 174 ++++++++++++++++++++++++++++++++-- 4 files changed, 277 insertions(+), 27 deletions(-)--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/