Menu
▾
▴
Re: [LTP] [PATCH 2/2] Remove libjvmsim
|
From: Subrata M. <su...@li...> - 2008-11-28 10:43:07
|
On Wed, 2008-11-26 at 10:40 +0530, Chirag Jog wrote:
> This patch removes the libjvmsim itself. Hence no
> testcase can use it anymore
>
> Signed-Off-By: Chirag <ch...@li...>
> Acked-By: Sripathi Kodi <sri...@in...>
Thanks for this too.
Regards--
Subrata
>
>
>
> ltp/testcases/realtime/config.mk | 7
> testcases/realtime/include/libjvmsim.h | 70 -----
> testcases/realtime/lib/libjvmsim.c | 410 ---------------------------------
> 3 files changed, 3 insertions(+), 484 deletions(-)
>
> Index: ltp/testcases/realtime/include/libjvmsim.h
> ===================================================================
> --- ltp.orig/testcases/realtime/include/libjvmsim.h
> +++ /dev/null
> @@ -1,70 +0,0 @@
> -/******************************************************************************
> - *
> - * Copyright © International Business Machines Corp., 2006, 2008
> - *
> - * This program is free software; you can redistribute it and/or modify
> - * it under the terms of the GNU General Public License as published by
> - * the Free Software Foundation; either version 2 of the License, or
> - * (at your option) any later version.
> - *
> - * This program is distributed in the hope that it will be useful,
> - * but WITHOUT ANY WARRANTY; without even the implied warranty of
> - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
> - * the GNU General Public License for more details.
> - *
> - * You should have received a copy of the GNU General Public License
> - * along with this program; if not, write to the Free Software
> - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
> - *
> - * NAME
> - * libjvmsim.h
> - *
> - * DESCRIPTION
> - * Provide threads to mimic real-time jvm activity
> - *
> - * Note: this "library" is written as a single header file to simplify the
> - * creation of new test cases and keep the build system simple. At some
> - * point it may make sense to break it up into .h and .c files and link to
> - * it as an object file.
> - *
> - * USAGE:
> - * To be included in testcases.
> - *
> - * AUTHOR
> - * Darren Hart <dv...@us...>
> - *
> - * HISTORY
> - * 2006-May-9: Initial version by Darren Hart
> - *
> - *****************************************************************************/
> -
> -#ifndef LIBJVMSIM_H
> -#define LIBJVMSIM_H
> -
> -#define SIT_AND_WAIT \
> -struct thread *t = (struct thread *)arg; \
> -int i = 0; \
> -while (!thread_quit(t)) { \
> - pthread_mutex_lock(&block_mutex); \
> - pthread_cond_wait(&block_cond, &block_mutex); \
> - busy_work_us(50); /* this should only happen on exit atm */ \
> - pthread_mutex_unlock(&block_mutex); \
> - i++; \
> -} \
> -debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i); \
> -return NULL;
> -
> -/* Function Prototypes */
> -void *jvm_gc_alarm(void *arg);
> -void *jvm_gc_collect(void *arg);
> -void *jvm_gc_trace(void *arg);
> -void *jvm_gc_finalize(void *arg);
> -void *jvm_posix_signal_dispatch(void *arg);
> -void *jvm_sigquit(void *arg);
> -void *jvm_async_event_server(void *arg);
> -void *jvm_timer_dispatch(void *arg);
> -void *jvm_ras_trace(void *arg);
> -void *jvmsim_monitor(void *arg);
> -void jvmsim_init();
> -
> -#endif /* LIBJVMSIM_H */
> Index: ltp/testcases/realtime/lib/libjvmsim.c
> ===================================================================
> --- ltp.orig/testcases/realtime/lib/libjvmsim.c
> +++ /dev/null
> @@ -1,410 +0,0 @@
> -/******************************************************************************
> - *
> - * Copyright © International Business Machines Corp., 2006, 2008
> - *
> - * This program is free software; you can redistribute it and/or modify
> - * it under the terms of the GNU General Public License as published by
> - * the Free Software Foundation; either version 2 of the License, or
> - * (at your option) any later version.
> - *
> - * This program is distributed in the hope that it will be useful,
> - * but WITHOUT ANY WARRANTY; without even the implied warranty of
> - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
> - * the GNU General Public License for more details.
> - *
> - * You should have received a copy of the GNU General Public License
> - * along with this program; if not, write to the Free Software
> - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
> - *
> - * NAME
> - * libjvmsim.c
> - *
> - * DESCRIPTION
> - * Provide threads to mimic real-time jvm activity
> - *
> - * Note: this "library" is written as a single header file to simplify the
> - * creation of new test cases and keep the build system simple. At some
> - * point it may make sense to break it up into .h and .c files and link to
> - * it as an object file.
> -
> - * USAGE:
> - * To be included in testcases.
> - *
> - * AUTHOR
> - * Darren Hart <dv...@us...>
> - *
> - * HISTORY
> - * 2006-May-9: Initial version by Darren Hart
> - *
> - *****************************************************************************/
> -
> -#include <stdio.h>
> -#include <sys/resource.h>
> -#include <sys/time.h>
> -#include <semaphore.h>
> -#include <librttest.h>
> -
> -/* this mutex and cond pair are used by the gc alarm and collect threads */
> -static pthread_mutex_t gc_mutex = PTHREAD_MUTEX_INITIALIZER;
> -static pthread_cond_t gc_cond;
> -
> -/* this semaphore is used by posix signal dispatch, and any other simulated
> - * by a sit and wait routine on a POSIX semaphore */
> -sem_t block_sem;
> -
> -/* this mutex and cond pair are used by all threads simulated by a
> - * sit and wait routine on a pthread mutex and cond pair */
> -static pthread_mutex_t block_mutex = PTHREAD_MUTEX_INITIALIZER;
> -static pthread_cond_t block_cond;
> -#define SIT_AND_WAIT \
> -struct thread *t = (struct thread *)arg; \
> -int i = 0; \
> -int ret; \
> -nsec_t timeout; \
> -struct timespec ts_timeout; \
> -while (!thread_quit(t)) { \
> - timeout = rt_gettime() + 50 * NS_PER_MS; \
> - nsec_to_ts(timeout, &ts_timeout); \
> - pthread_mutex_lock(&block_mutex); \
> - ret = pthread_cond_timedwait(&block_cond, &block_mutex, &ts_timeout); \
> - if (ret && ret != ETIMEDOUT) \
> - printf("%s: %s\n", __FUNCTION__, strerror(ret)); \
> - busy_work_us(50); /* this should only happen on exit atm */ \
> - pthread_mutex_unlock(&block_mutex); \
> - i++; \
> -} \
> -debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i); \
> -return NULL;
> -
> -/* High level description of JVM threads
> - * There may be other threads used for profiling, debug and trace
> - * purposes depending on options and application.
> - */
> -
> -/* Main/Primordial  thread  (SCHED_OTHER prio)
> - * o Creates other internal threads and runs user's main().
> - * o Typically would create the initial RT thread but otherwise
> - * is effectively in a pthread_cond wait state doing no work
> - * o Covered by testcase main() really (Not implemented here)
> - */
> -
> -/* GC Alarm  (RT FIFO 89)
> - * o Does clock_nanosleep for 451us
> - * o Upon awakening, does a pthread_mutex_lock(m) and then a
> - * pthread_cond_signal(s) to a GC collect thread waiting on signal(s)
> - * o Repeat this process
> - */
> -void *jvm_gc_alarm(void *arg)
> -{
> - struct thread *t = (struct thread *)arg;
> - int i = 0;
> - while (!thread_quit(t)) {
> - rt_nanosleep(451*NS_PER_US);
> - pthread_mutex_lock(&gc_mutex);
> - pthread_cond_signal(&gc_cond);
> - pthread_mutex_unlock(&gc_mutex);
> - i++;
> - }
> - /* do it once more to make sure the collect thread can exit */
> - pthread_mutex_lock(&gc_mutex);
> - pthread_cond_signal(&gc_cond);
> - pthread_mutex_unlock(&gc_mutex);
> - debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i);
> - return NULL;
> -}
> -
> -/* GC Collect (RT FIFO appPrio+1)
> - * o The default should be to have one such thread. There is a JVM option that
> - * can control the number of such worker threads created but the common case
> - * we should only have one running.
> - * o Does a pthread_cond_wait(m,s)
> - * o On pthread_cond_signal() from the alarm thread sees if there is GC work to
> - * be performed (small/trivial calculation - i.e. noise)
> - * o If no work, then the server thread repeats this whole process
> - * Performs gc work (which I don't think you would want to emulate)
> - * o Repeats this process
> - */
> -void *jvm_gc_collect(void *arg)
> -{
> - struct thread *t = (struct thread *)arg;
> - int i = 0;
> - int ret;
> - nsec_t timeout;
> - struct timespec ts_timeout;
> -
> - while (!thread_quit(t)) {
> - timeout = rt_gettime() + 50 * NS_PER_MS;
> - nsec_to_ts(timeout, &ts_timeout);
> - pthread_mutex_lock(&gc_mutex);
> - ret = pthread_cond_timedwait(&gc_cond, &gc_mutex, &ts_timeout);
> - if (ret && ret != ETIMEDOUT)
> - printf("%s: %s\n", __FUNCTION__, strerror(ret));
> - busy_work_us(50);
> - pthread_mutex_unlock(&gc_mutex);
> - i++;
> - }
> - debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i);
> - return NULL;
> -}
> -
> -/* GC Trace Thread (SCHED_OTHER)
> - * o A thread that sleeps using pthread_cond_timedwait for 50 millis.
> - * o It wakes up, releases the  mutex does a getrusage(RUSAGE_SELF, ×)
> - * call and some small calculations (~50us or so) then goes back to the
> - * timewait (grabbing mutex of course)
> - */
> -void *jvm_gc_trace(void *arg)
> -{
> - struct thread *t = (struct thread *)arg;
> - pthread_cond_t c;
> - pthread_condattr_t c_attr;
> - pthread_mutex_t m = PTHREAD_MUTEX_INITIALIZER;
> - nsec_t abs_timeout;
> - struct timespec ts_timeout;
> - struct rusage usage;
> - int i, ret;
> -
> - /* use the same clock as the timeout value */
> - pthread_condattr_init(&c_attr);
> - if (pthread_condattr_setclock(&c_attr, CLOCK_MONOTONIC) != 0) {
> - perror("pthread_condattr_setclock");
> - exit(-1);
> - }
> - if (pthread_cond_init(&c, &c_attr) != 0) {
> - perror("pthread_cond_init");
> - exit(-1);
> - }
> -
> - i=0;
> - while (!thread_quit(t)) {
> - pthread_mutex_lock(&m);
> - abs_timeout = rt_gettime() + 50 * NS_PER_MS;
> - nsec_to_ts(abs_timeout, &ts_timeout);
> - do {
> - ret = pthread_cond_timedwait(&c, &m, &ts_timeout);
> - } while (ret == 0);
> - if (ret != ETIMEDOUT)
> - printf("%s: %s\n", __FUNCTION__, strerror(ret));
> - getrusage(RUSAGE_SELF, &usage);
> - busy_work_us(50);
> - pthread_mutex_unlock(&m);
> - i++;
> - }
> - pthread_condattr_destroy(&c_attr);
> - debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i);
> - return NULL;
> -}
> -
> -/* GC Finalize (SCHED_OTHER)
> - * o GC could dispatch work off to this thread depending on users
> - * application. But, for the common case I believe this thread should be
> - * in a wait state and can be ignored,
> - */
> -/* sit and wait */
> -void *jvm_gc_finalize(void *arg)
> -{
> -SIT_AND_WAIT
> -}
> -
> -/* GC Processor info thread (RT FIFO ??)
> - * o JVM startup creates one such thread per processor and binds one
> - * thread per the individual processors (including CPU 0). Each threads
> - * tries to calculate its CPU frequency, Â and when done, Â the thread dies.
> - * o For the common case, these threads have exited and so are not running.
> - * I do have comeback as to how long they run to perform the frequency
> - * calculation. (will append that later)
> - */
> -/* can ignore this one */
> -/*void *jvm_gc_cpu_info(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* POSIX Signal Dispatch (RT FIFO 87)
> - * o should be in a sem_wait() state. Async signals such as SIGABORT, SIGQUIT,
> - * SIGSTOP are delegated off to this thread from a signal handler using
> - * sem_post().
> - * o For the common case, this thread should be in a sem_wait() state and so
> - * doing nothing. I.E. can be ignored.
> - */
> -/* good to have, sit and do nothing */
> -void *jvm_posix_signal_dispatch(void *arg)
> -{
> - struct thread *t = (struct thread *)arg;
> - int i = 0;
> - nsec_t timeout;
> - struct timespec ts_timeout;
> -
> - while (!thread_quit(t)) {
> - clock_gettime(CLOCK_REALTIME, &ts_timeout);
> - ts_to_nsec(&ts_timeout, &timeout);
> - timeout += 50 * NS_PER_MS;
> - nsec_to_ts(timeout, &ts_timeout);
> - sem_timedwait(&block_sem, &ts_timeout);
> - i++;
> - }
> - debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i);
> - return NULL;
> -}
> -
> -/* POSIX Signal Dispatch NH
> - * o This thread is analogous to the previous thread. The reason we have 2 is
> - * that the former thread may touch the heap while this thread cannot.
> - * I.E. can be ignored.
> - */
> -/*void *jvm_posix_signal_dispatch_nh(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* JIT Sampler (RT FIFO 1)
> - * o Somewhat like the alarm thread and used by JIT.
> - * o For configuration running AOT this thread is not running and can be
> - * ignored
> - */
> -/* not needed with aot */
> -/*void *jvm_jit_sampler(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* JIT Compile (should be SCHED_OTHER - but think it is not)
> - * o Does a JIT compile.
> - * o For Configurations running AOT this thread is not running and can
> - * be ignored
> - */
> -/* not needed with aot */
> -/*void *jvm_jit_compile(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* SigQuit Thread (SCHED_OTHER but the priority will likely be raised)
> - * o SIGQUIT helper thread. The thread is in a pthread_cond_wait() so,
> - * for our purposes, should not be doing any work.
> - */
> -/* good to have - sit there and do nothing until exit*/
> -void *jvm_sigquit(void *arg)
> -{
> -SIT_AND_WAIT
> -}
> -
> -/* Async EventServerThreads (priority described below)
> - * o A pool of threads that are used to service hander fire requests and bind
> - * to handler.
> - * o 6 Â or so threads are created in a wait state at priority 83 (I believe).
> - * The priority of these theads is changed to the appropriate RT priority
> - * before they exit the wait state and they go back to prio 83 when they are
> - * done servicing the request.
> - * o For the common case, these threads should just be waiting.
> - */
> -void *jvm_async_event_server(void *arg)
> -{
> -SIT_AND_WAIT
> -}
> -
> -/* Async EventServerThreads NH
> - * o Same as previous (# threads may be different) but used for handlers that
> - * can't access the Java heap.
> - */
> -/*void *jvm_async_event_server_nh(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* Timer Dispatch (RT FIFO 83 currently)
> - * o A thread used as part of implementation for Timers that need to fire at
> - * certain times.
> - * o In a wait state if no timers or handlers are created (default)
> - */
> -void *jvm_timer_dispatch(void *arg)
> -{
> -SIT_AND_WAIT
> -}
> -
> -/* Timer Dispatch NH
> - * o same as previous except for threads that cannot access the Java heap.
> - */
> -/*void *jvm_timer_dispatch_nh(void *arg)
> -{
> - //struct thread *t = (struct thread *)arg;
> -}*/
> -
> -/* RAS trace thread (SCHED_OTHER)
> - * o in support of some JVM tracing activities
> - * o is in a wait state
> - */
> -void *jvm_ras_trace(void *arg)
> -{
> -SIT_AND_WAIT
> -}
> -
> -/* cleanup and wakeup blocked threads at join_threads() time */
> -void *jvmsim_monitor(void *arg) {
> - struct thread *t = (struct thread *)arg;
> - int i = 0;
> -
> - while (!thread_quit(t)) {
> - usleep(10*US_PER_MS);
> - i++;
> - }
> -
> - /* unblock all the sit and wait threads */
> - pthread_mutex_lock(&block_mutex);
> - pthread_cond_broadcast(&block_cond);
> - pthread_mutex_unlock(&block_mutex);
> -
> - sem_post(&block_sem); /* allow jvm_posix_signal_dispatch() to exit */
> -
> - debug(DBG_INFO, "JVMSIM: %s thread woke %d times\n", __FUNCTION__, i);
> - return NULL;
> -}
> -
> -/* FIXME: the priorities used below will need to be adjusted
> - * and librt.h should be made to provide access to the highest
> - * prio running thread - perhaps with a listener interface... (so the gc
> - * can be boosted to app+1 prio.
> - */
> -
> -/* convenience functions for starting groups of jvm simulation threads */
> -void jvmsim_init(void)
> -{
> - pthread_condattr_t c_attr;
> - sem_init(&block_sem, 0, 0);
> -
> - if (pthread_condattr_setclock(&c_attr, CLOCK_MONOTONIC) != 0) {
> - perror("pthread_condattr_setclock");
> - exit(-1);
> - }
> - if (pthread_cond_init(&gc_cond, &c_attr) != 0) {
> - perror("pthread_cond_init");
> - exit(-1);
> - }
> - if (pthread_cond_init(&block_cond, &c_attr) != 0) {
> - perror("pthread_cond_init");
> - exit(-1);
> - }
> -
> - create_other_thread(jvmsim_monitor, (void *)0);
> - create_fifo_thread(jvm_gc_alarm, (void *)0, 89);
> - /* The gc collect thread varies - this can be modelled better */
> - create_fifo_thread(jvm_gc_collect, (void *)0, 82);
> - create_other_thread(jvm_gc_trace, (void *)12);
> - create_other_thread(jvm_gc_finalize, (void *)0);
> - create_fifo_thread(jvm_posix_signal_dispatch, (void *)0, 87);
> - create_other_thread(jvm_sigquit, (void *)88);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 85);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 85);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_async_event_server, (void *)0, 83);
> - create_fifo_thread(jvm_timer_dispatch, (void *)0, 85);
> - create_fifo_thread(jvm_timer_dispatch, (void *)0, 83);
> - create_other_thread(jvm_ras_trace, (void *)0);
> -}
> -
> Index: ltp/testcases/realtime/config.mk
> ===================================================================
> --- ltp.orig/testcases/realtime/config.mk
> +++ ltp/testcases/realtime/config.mk
> @@ -18,8 +18,7 @@ endif
> #
> CPPFLAGS += -I$(srcdir)/include -D_GNU_SOURCE
> CFLAGS += -Wall
> -LDLIBS += $(srcdir)/lib/libjvmsim.o \
> - $(srcdir)/lib/librttest.o \
> - $(srcdir)/lib/libstats.o \
> - -lpthread -lrt -lm
> +LDLIBS += $(srcdir)/lib/librttest.o \
> + $(srcdir)/lib/libstats.o \
> + -lpthread -lrt -lm
>
|
