[rtems commit] libmisc: Add top to cpuusage.
Gedare Bloom
gedare at rtems.org
Thu Oct 30 14:53:34 UTC 2014
I missed reviewing this patch earlier, but there are some style
problems with it, please fix them on the head.
On Tue, Oct 28, 2014 at 11:17 AM, Jennifer Averett <jennifer at rtems.org> wrote:
> Module: rtems
> Branch: master
> Commit: 6031da438d219c6ec5d9d48f1df2aef91710cce3
> Changeset: http://git.rtems.org/rtems/commit/?id=6031da438d219c6ec5d9d48f1df2aef91710cce3
>
> Author: Jennifer Averett <jennifer.averett at oarcorp.com>
> Date: Mon Sep 29 10:20:27 2014 -0500
>
> libmisc: Add top to cpuusage.
>
> ---
>
> cpukit/libmisc/Makefile.am | 2 +-
> cpukit/libmisc/cpuuse/cpuusagetop.c | 337 +++++++++++++++++++++++++++++++++++
> cpukit/libmisc/cpuuse/cpuuse.h | 19 ++
> 3 files changed, 357 insertions(+), 1 deletions(-)
>
> diff --git a/cpukit/libmisc/Makefile.am b/cpukit/libmisc/Makefile.am
> index 2f41ffa..d26c484 100644
> --- a/cpukit/libmisc/Makefile.am
> +++ b/cpukit/libmisc/Makefile.am
> @@ -27,7 +27,7 @@ EXTRA_DIST += cpuuse/README
>
> noinst_LIBRARIES += libcpuuse.a
> libcpuuse_a_SOURCES = cpuuse/cpuusagereport.c cpuuse/cpuusagereset.c \
> - cpuuse/cpuuse.h cpuuse/cpuusagedata.c
> + cpuuse/cpuuse.h cpuuse/cpuusagedata.c cpuuse/cpuusagetop.c
>
> ## devnull
> noinst_LIBRARIES += libdevnull.a
> diff --git a/cpukit/libmisc/cpuuse/cpuusagetop.c b/cpukit/libmisc/cpuuse/cpuusagetop.c
> new file mode 100644
> index 0000000..7e7348a
> --- /dev/null
> +++ b/cpukit/libmisc/cpuuse/cpuusagetop.c
> @@ -0,0 +1,337 @@
> +/**
> + * @file
> + *
> + * @brief CPU Usage Top
> + * @ingroup libmisc_cpuuse CPU Usage
> + */
> +
> +/*
> + * COPYRIGHT (c) 2014.
> + * On-Line Applications Research Corporation (OAR).
> + *
> + * The license and distribution terms for this file may be
> + * found in the file LICENSE in this distribution or at
> + * http://www.rtems.org/license/LICENSE.
> + */
> +
> +#ifdef HAVE_CONFIG_H
> +#include "config.h"
> +#endif
> +
> +#include <string.h>
> +#include <stdlib.h>
> +#include <stdio.h>
> +#include <ctype.h>
> +#include <inttypes.h>
> +
> +#include <rtems/cpuuse.h>
> +#include <rtems/score/objectimpl.h>
> +#include <rtems/score/threadimpl.h>
> +#include <rtems/score/todimpl.h>
> +#include <rtems/score/watchdogimpl.h>
> +
> +
> +/*
> + * Common variable to sync the load monitor task.
> + */
> +static volatile int rtems_cpuusage_top_thread_active;
> +
> +
remove multiple blank lines
> +typedef struct {
> + void *context;
> + rtems_printk_plugin_t print;
> +}rtems_cpu_usage_plugin_t;
> +
> +#define RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS (20)
> +
> +/*
> + * rtems_cpuusage_top_thread
> + *
> + * DESCRIPTION:
> + *
> + * This function displays the load of the tasks on an ANSI terminal.
> + *
Get rid of "DESCRIPTION", prefer to use minimalist comment style on
static functions.
> + */
> +
> +static void
> +rtems_cpuusage_top_thread (rtems_task_argument arg)
> +{
> + uint32_t api_index;
> + Thread_Control* the_thread;
> + int i;
> + int j;
> + int k;
> + Objects_Information* information;
> + char name[13];
> + int task_count = 0;
> + uint32_t seconds, nanoseconds;
> + rtems_cpu_usage_plugin_t* plugin = (rtems_cpu_usage_plugin_t*)arg;
> + Thread_Control* load_tasks[RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS + 1];
> + unsigned long long load[RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS + 1];
> +
> + while (true)
> + {
Opening brace should go on whille() line.
> + #ifndef __RTEMS_USE_TICKS_FOR_STATISTICS__
> + Timestamp_Control uptime, total, ran, uptime_at_last_reset;
> + #else
> + uint32_t total_units = 0;
> + #endif
> +
> + rtems_cpuusage_top_thread_active = 1;
> +
> + memset (load_tasks, 0, sizeof (load_tasks));
> + memset (load, 0, sizeof (load));
> +
> + /*
> + * Iterate over the tasks and sort the highest load tasks
> + * into our local arrays. We only handle a limited number of
> + * tasks.
> + */
> + for ( api_index = 1 ; api_index <= OBJECTS_APIS_LAST ; api_index++ ) {
> + #if !defined(RTEMS_POSIX_API) || defined(RTEMS_DEBUG)
> + if ( !_Objects_Information_table[ api_index ] )
> + continue;
> + #endif
> +
> + information = _Objects_Information_table[ api_index ][ 1 ];
> + if ( information ) {
> + for ( k=1 ; k <= information->maximum ; k++ ) {
I prefer that nested iterators should be used in increasing order, e.g. i first.
> + the_thread = (Thread_Control *)information->local_table[ k ];
> + if ( the_thread ) {
> +
No blank space after the opening brace.
> + Thread_CPU_usage_t l = the_thread->cpu_time_used;
Avoid using l and I as variable names.They're hard to differentiate
from each other and from the digit 1. This should be a more useful
variable name anyway. A one-letter variable should really only be used
for iterators. We ought to have a rule about this if we don't.
> +
> + /*
> + * When not using nanosecond CPU usage resolution, we have to count
> + * the number of "ticks" we gave credit for to give the user a rough
> + * guideline as to what each number means proportionally.
> + */
> + #ifdef __RTEMS_USE_TICKS_FOR_STATISTICS__
> + total_units += l;
> + #endif
> +
> + /* Count the number of tasks and sort this load value */
> + task_count++;
> + for (i = 0; i < RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS; i++) {
> + if (load_tasks[i]) {
> + if ((l == 0) || (l < load[i]))
> + continue;
> + for (j = (RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS - 1); j >= i; j--){
> + load_tasks[j + 1] = load_tasks[j];
> + load[j + 1] = load[j];
> + }
> + }
> + load_tasks[i] = the_thread;
> + load[i] = l;
> + break;
> + }
> + }
> + }
> + }
> + }
> +
> + #ifndef __RTEMS_USE_TICKS_FOR_STATISTICS__
> + _Timestamp_Set_to_zero( &total );
> + uptime_at_last_reset = CPU_usage_Uptime_at_last_reset;
> + #endif
> +
> + _TOD_Get_uptime( &uptime );
> + seconds = _Timestamp_Get_seconds( &uptime );
> + nanoseconds = _Timestamp_Get_nanoseconds( &uptime ) /
> + TOD_NANOSECONDS_PER_MICROSECOND;
> + (*plugin->print)(plugin->context, "\x1b[H\x1b[J Press ENTER to exit.\n\n");
> + (*plugin->print)(plugin->context, "uptime: ");
> + (*plugin->print)(plugin->context,
> + "%7" PRIu32 ".%06" PRIu32 "\n", seconds, nanoseconds
> + );
> +
> + (*plugin->print)(
> + plugin->context,
> + "-------------------------------------------------------------------------------\n"
> + " CPU USAGE BY THREAD\n"
> + "------------+---------------------+---------------+---------------+------------\n"
> + #ifndef __RTEMS_USE_TICKS_FOR_STATISTICS__
> + " ID | NAME | RPRI | CPRI | SECONDS | PERCENT\n"
> + #else
> + " ID | NAME | RPRI | CPRI | TICKS | PERCENT\n"
> + #endif
> + "------------+---------------------+---------------+---------------+------------\n"
> + );
> +
> + for (i = 0; i < RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS; i++)
> + {
> +
Fix style here.
> + if (!load_tasks[i])
> + break;
> +
> + /*
> + * If this is the currently executing thread, account for time
> + * since the last context switch.
> + */
> + the_thread = load_tasks[i];
> +
> + rtems_object_get_name( the_thread->Object.id, sizeof(name), name );
> + (*plugin->print)(
> + plugin->context,
> + " 0x%08" PRIx32 " | %-19s | %3" PRId32 " | %3" PRId32 " |",
> + the_thread->Object.id,
> + name,
> + the_thread->real_priority,
> + the_thread->current_priority
> + );
> +
> + #ifndef __RTEMS_USE_TICKS_FOR_STATISTICS__
> + {
I don't know how to feel about artifically adding scoped blocks.
Anyone else weigh in?
> + Timestamp_Control last;
> + uint32_t ival, fval;
> +
> + /*
> + * If this is the currently executing thread, account for time
> + * since the last context switch.
> + */
> + ran = load[i];
> + if ( _Thread_Get_time_of_last_context_switch( the_thread, &last ) ) {
> + Timestamp_Control used;
> + _TOD_Get_uptime( &uptime );
> + _Timestamp_Subtract( &last, &uptime, &used );
> + _Timestamp_Add_to( &ran, &used );
> + } else {
> + _TOD_Get_uptime( &uptime );
> + }
> + _Timestamp_Subtract( &uptime_at_last_reset, &uptime, &total );
> + _Timestamp_Divide( &ran, &total, &ival, &fval );
> +
> + /*
> + * Print the information
> + */
> +
> + seconds = _Timestamp_Get_seconds( &ran );
> + nanoseconds = _Timestamp_Get_nanoseconds( &ran ) /
> + TOD_NANOSECONDS_PER_MICROSECOND;
> + (*plugin->print)( plugin->context,
> + "%7" PRIu32 ".%06" PRIu32 " |%4" PRIu32 ".%03" PRIu32 "\n",
> + seconds, nanoseconds,
> + ival, fval
> + );
> + }
> + #else
> + if (total_units) {
> + uint64_t ival_64;
> +
> + ival_64 = load[i];
> + ival_64 *= 100000;
> + ival = ival_64 / total_units;
> + } else {
> + ival = 0;
> + }
> +
> + fval = ival % 1000;
> + ival /= 1000;
> + (*plugin->print)( plugin->context,
> + "%14" PRIu32 " |%4" PRIu32 ".%03" PRIu32 "\n",
> + load[i],
> + ival,
> + fval
> + );
> + #endif
> + }
> +
> + if (task_count < RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS)
> + {
> + j = RTEMS_CPUUSAGE_TOP_MAX_LOAD_TASKS - task_count;
> + while (j > 0)
> + {
> + (*plugin->print)( plugin->context, "\x1b[K\n");
> + j--;
> + }
> + }
> +
> + rtems_cpuusage_top_thread_active = 0;
> +
> + rtems_task_wake_after (RTEMS_MICROSECONDS_TO_TICKS (5000000));
> + }
> +}
> +
> +void rtems_cpu_usage_top_with_plugin(
> + void *context,
> + rtems_printk_plugin_t print
> +)
> +{
> + rtems_status_code sc;
> + rtems_task_priority priority;
> + rtems_name name;
> + rtems_id id;
> + rtems_cpu_usage_plugin_t plugin;
> +
> + if ( !print )
> + return;
> +
> + plugin.context = context;
> + plugin.print = print;
> +
> + sc = rtems_task_set_priority (RTEMS_SELF, RTEMS_CURRENT_PRIORITY, &priority);
> +
> + if (sc != RTEMS_SUCCESSFUL)
> + {
> + (*print)(
> + context,
> + "error: cannot obtain the current priority: %s\n",
> + rtems_status_text (sc)
> + );
> + return;
> + }
> +
> + name = rtems_build_name('C', 'P', 'l', 't');
> +
> + sc = rtems_task_create (name, priority, 4 * 1024,
> + RTEMS_NO_FLOATING_POINT | RTEMS_LOCAL,
> + RTEMS_PREEMPT | RTEMS_TIMESLICE | RTEMS_NO_ASR,
> + &id);
> +
> + if (sc != RTEMS_SUCCESSFUL)
> + {
> + (*print)(
> + context,
> + "error: cannot create helper thread: %s\n",
> + rtems_status_text (sc)
> + );
> + return;
> + }
> +
> + sc = rtems_task_start (
> + id, rtems_cpuusage_top_thread, (rtems_task_argument)&plugin
> + );
> + if (sc != RTEMS_SUCCESSFUL)
> + {
> + (*print)(
> + context,
> + "error: cannot start helper thread: %s\n",
> + rtems_status_text (sc)
> + );
> + rtems_task_delete (id);
> + return;
> + }
> +
> + for (;;)
> + {
> + int c = getchar ();
> +
> + if ((c == '\r') || (c == '\n'))
> + {
> + int loops = 20;
> +
> + while (loops && rtems_cpuusage_top_thread_active)
> + rtems_task_wake_after (RTEMS_MICROSECONDS_TO_TICKS (100000));
> +
> + rtems_task_delete (id);
> +
> + (*print)(context, "load monitoring stopped.\n");
> + return;
> + }
> + }
> +}
> +
> +void rtems_cpu_usage_top( void )
> +{
> + rtems_cpu_usage_top_with_plugin( NULL, printk_plugin );
> +}
> diff --git a/cpukit/libmisc/cpuuse/cpuuse.h b/cpukit/libmisc/cpuuse/cpuuse.h
> index 1aee275..662d905 100644
> --- a/cpukit/libmisc/cpuuse/cpuuse.h
> +++ b/cpukit/libmisc/cpuuse/cpuuse.h
> @@ -63,6 +63,25 @@ void rtems_cpu_usage_report_with_plugin(
> void rtems_cpu_usage_report( void );
>
> /**
> + * @brief CPU usage Top plugin
> + *
> + * Report CPU Usage in top format to
> + * to a print plugin.
> + */
> +void rtems_cpu_usage_top_with_plugin(
> + void *context,
> + rtems_printk_plugin_t print
> +);
> +
> +/**
> + * @brief CPU usage top.
> + *
> + * CPU Usage top
> + */
> +
> +void rtems_cpu_usage_top( void );
> +
> +/**
> * @brief Reset CPU usage.
> *
> * CPU Usage Reporter
>
> _______________________________________________
> vc mailing list
> vc at rtems.org
> http://lists.rtems.org/mailman/listinfo/vc
More information about the devel
mailing list