* [RFC PATCH 1/4] CodeSamples: Add rule to generate Makefile.arch and api.h
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
@ 2017-05-29 22:14 ` Akira Yokosawa
2017-05-29 22:16 ` [RFC PATCH 2/4] CodeSamples: Remove generated files from repository Akira Yokosawa
` (4 subsequent siblings)
5 siblings, 0 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-29 22:14 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From b4845ac3b17aaade7258528fa114def5f6554242 Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Sun, 28 May 2017 09:29:44 +0900
Subject: [RFC PATCH 1/4] CodeSamples: Add rule to generate Makefile.arch and api.h
To generate suitable Makefile.arch and api.h for host arch,
describe the dependency in depends.mk and include it in Makefiles.
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/Makefile | 18 +++++++++++++
CodeSamples/SMPdesign/Makefile | 3 +++
CodeSamples/advsync/Makefile | 3 +++
CodeSamples/count/Makefile | 3 +++
CodeSamples/datastruct/Issaquah/Makefile | 3 +++
CodeSamples/datastruct/hash/Makefile | 3 +++
CodeSamples/datastruct/log/Makefile | 3 +++
CodeSamples/datastruct/skiplist/Makefile | 3 +++
CodeSamples/defer/Makefile | 3 +++
CodeSamples/depends.mk | 43 ++++++++++++++++++++++++++++++++
CodeSamples/intro/Makefile | 3 +++
CodeSamples/locking/Makefile | 3 +++
CodeSamples/toolsoftrade/Makefile | 3 +++
13 files changed, 94 insertions(+)
create mode 100644 CodeSamples/depends.mk
diff --git a/CodeSamples/Makefile b/CodeSamples/Makefile
index f5461e9..4168220 100644
--- a/CodeSamples/Makefile
+++ b/CodeSamples/Makefile
@@ -1,3 +1,7 @@
+.PHONY: all clean pthreads pthreads-x86 pthreads-ppc64 pthreads-arm
+
+top := .
+
all:
(cd SMPdesign; make)
(cd advsync; make)
@@ -6,6 +10,17 @@ all:
(cd intro; make)
(cd toolsoftrade; make)
+include depends.mk
+
+api.h Makefile.arch:
+ifeq ($(strip $(target)),)
+ make pthreads
+ $(warning "Could not figure out which target to use (arch:$(arch)).\
+ Used 'make pthreads' in CodeSamples/Makefile.")
+else
+ make pthreads-$(target)
+endif
+
pthreads:
echo "#ifndef __PERFBOOK_API_H__" > api.h
echo "#define __PERFBOOK_API_H__" >> api.h
@@ -125,3 +140,6 @@ pthreads-arm:
echo "#endif /* #ifndef __PERFBOOK_API_H__ */" >> api.h
echo "# MECHANICALLY GENERATED, DO NOT EDIT!!!" > Makefile.arch
cat arch-arm/Makefile.arch >> Makefile.arch
+
+clean:
+ rm -f Makefile.arch api.h
diff --git a/CodeSamples/SMPdesign/Makefile b/CodeSamples/SMPdesign/Makefile
index ae0cdd7..d03575c 100644
--- a/CodeSamples/SMPdesign/Makefile
+++ b/CodeSamples/SMPdesign/Makefile
@@ -25,6 +25,9 @@ include ../Makefile.arch
# http://www.ibm.com/developerworks/wikis/display/hpccentral/Tuning+options+to+consider+with+gcc#Tuningoptionstoconsiderwithgcc-mcpuandmtune
# GCC_ARGS=-g -O3 -m64 -mcpu=power5 -mtune=power5
+top := ..
+include $(top)/depends.mk
+
smpalloc: smpalloc.c ../api.h
cc $(GCC_ARGS) -g -o smpalloc -DTEST smpalloc.c -lpthread
diff --git a/CodeSamples/advsync/Makefile b/CodeSamples/advsync/Makefile
index b02ca37..bb4c8fc 100644
--- a/CodeSamples/advsync/Makefile
+++ b/CodeSamples/advsync/Makefile
@@ -20,6 +20,9 @@ PROGS = q singleq wfenqueue
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
q: q.c q.h queuetorture.h ../api.h
cc $(GCC_ARGS) -o q -DTEST q.c -lpthread
diff --git a/CodeSamples/count/Makefile b/CodeSamples/count/Makefile
index 05fcf04..d3fbb83 100644
--- a/CodeSamples/count/Makefile
+++ b/CodeSamples/count/Makefile
@@ -35,6 +35,9 @@ RCU_SRCS = ../defer/rcu_nest32.h ../defer/rcu_nest32.c
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
count_atomic: count_atomic.c ../api.h counttorture.h
cc $(GCC_ARGS) $(CFLAGS) -o count_atomic count_atomic.c -lpthread
diff --git a/CodeSamples/datastruct/Issaquah/Makefile b/CodeSamples/datastruct/Issaquah/Makefile
index 013738c..bb3d6a3 100644
--- a/CodeSamples/datastruct/Issaquah/Makefile
+++ b/CodeSamples/datastruct/Issaquah/Makefile
@@ -21,6 +21,9 @@ LIB = ../../lib
all: $(PROGS)
+top := ../..
+include $(top)/depends.mk
+
# NOTE: For decent scalability on update-side tests as of early 2015,
# use something like jemalloc() instead of glibc malloc().
# If you install jemalloc at /home/paulmck/jemalloc, you will
diff --git a/CodeSamples/datastruct/hash/Makefile b/CodeSamples/datastruct/hash/Makefile
index a87bdbb..85dfa5d 100644
--- a/CodeSamples/datastruct/hash/Makefile
+++ b/CodeSamples/datastruct/hash/Makefile
@@ -20,6 +20,9 @@ PROGS = hash_bkt hash_bkt_hazptr hash_bkt_rcu hash_global hash_resize
all: $(PROGS)
+top := ../..
+include $(top)/depends.mk
+
hash_bkt: hash_bkt.c ../../api.h hashtorture.h
cc $(GCC_ARGS) -DTEST_HASH -o hash_bkt hash_bkt.c -lpthread
diff --git a/CodeSamples/datastruct/log/Makefile b/CodeSamples/datastruct/log/Makefile
index f225c92..cd2c571 100644
--- a/CodeSamples/datastruct/log/Makefile
+++ b/CodeSamples/datastruct/log/Makefile
@@ -21,6 +21,9 @@ LIB = ../../lib
all: $(PROGS)
+top := ../..
+include $(top)/depends.mk
+
log_glock: log_glock.c ../../api.h logtorture.h log.h
cc $(GCC_ARGS) -DTEST_LOG -I $(LIB) -g -o log_glock log_glock.c $(LIB)/random.c -lpthread
diff --git a/CodeSamples/datastruct/skiplist/Makefile b/CodeSamples/datastruct/skiplist/Makefile
index 7289979..ab6f524 100644
--- a/CodeSamples/datastruct/skiplist/Makefile
+++ b/CodeSamples/datastruct/skiplist/Makefile
@@ -21,6 +21,9 @@ LIB = ../../lib
all: $(PROGS)
+top := ../..
+include $(top)/depends.mk
+
skiplist: skiplist.c ../../api.h skiplisttorture.h skiplist.h
cc $(GCC_ARGS) -DTEST_SKIPLIST -I $(LIB) -g -o skiplist skiplist.c $(LIB)/random.c -lpthread -lurcu -lurcu-signal
diff --git a/CodeSamples/defer/Makefile b/CodeSamples/defer/Makefile
index 35db164..5e0f1e7 100644
--- a/CodeSamples/defer/Makefile
+++ b/CodeSamples/defer/Makefile
@@ -46,6 +46,9 @@ PROGS = bug_rcu_dp \
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
# Note that bug_srcu_a is disabled until completed.
bug_srcu_a: bug_srcu_a.c srcu.c ../api.h
cc $(GCC_ARGS) -o bug_srcu_a -DTEST bug_srcu_a.c -lurcu -lpthread
diff --git a/CodeSamples/depends.mk b/CodeSamples/depends.mk
new file mode 100644
index 0000000..409eeee
--- /dev/null
+++ b/CodeSamples/depends.mk
@@ -0,0 +1,43 @@
+ifeq ($(strip $(arch)),)
+arch := $(shell uname -m)
+endif
+
+ifeq ($(arch),i686)
+target := x86
+else ifeq ($(arch),x86_64)
+target := x86
+else ifeq ($(arch),ppc64)
+target := ppc64
+else ifneq (,$(findstring arm,$(arch)))
+target := arm
+else
+target :=
+endif
+
+api_depend_common := $(top)/linux/common.h \
+ $(top)/api-pthreads/api-pthreads.h \
+ $(top)/linux/list.h
+ifeq ($(target),x86)
+api_depend := $(top)/arch-x86/arch-x86.h
+arch_depend := $(top)/arch-x86/Makefile.arch
+else ifeq ($(target),ppc64)
+api_depend := $(top)/arch-ppc64/arch-ppc64.h
+arch_depend := $(top)/arch-ppc64/Makefile.arch
+else ifeq ($(target),arm)
+api_depend := $(top)/arch-arm/arch-arm.h
+arch_depend := $(top)/arch-arm/Makefile.arch
+else
+api_depend :=
+arch_depend :=
+endif
+
+ifneq ($(top),.)
+$(top)/api.h: $(api_depend) $(api_depend_common)
+ make -C $(top) api.h
+
+$(top)/Makefile.arch: $(arch_depend)
+ make -C $(top) Makefile.arch
+else
+$(top)/api.h: $(api_depend) $(api_depend_common)
+$(top)/Makefile.arch: $(arch_depend)
+endif
diff --git a/CodeSamples/intro/Makefile b/CodeSamples/intro/Makefile
index e323834..ba2f257 100644
--- a/CodeSamples/intro/Makefile
+++ b/CodeSamples/intro/Makefile
@@ -2,6 +2,9 @@ PROGS = initrace initraceperf lockinc nakedinc perthreadinc threadcreate
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
initrace: initrace.c ../api.h
cc -g -Wall -o initrace initrace.c -lpthread
diff --git a/CodeSamples/locking/Makefile b/CodeSamples/locking/Makefile
index 15cf8f1..5862448 100644
--- a/CodeSamples/locking/Makefile
+++ b/CodeSamples/locking/Makefile
@@ -2,6 +2,9 @@ PROGS = locked_list xchglock
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
locked_list: locked_list.c ../api.h
cc -g -Wall -o locked_list locked_list.c
diff --git a/CodeSamples/toolsoftrade/Makefile b/CodeSamples/toolsoftrade/Makefile
index 4b7f219..b6216e8 100644
--- a/CodeSamples/toolsoftrade/Makefile
+++ b/CodeSamples/toolsoftrade/Makefile
@@ -26,6 +26,9 @@ PROGS = forkjoin \
all: $(PROGS)
+top := ..
+include $(top)/depends.mk
+
forkjoin: forkjoin.c ../api.h
cc $(GCC_ARGS) -o forkjoin forkjoin.c
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* [RFC PATCH 2/4] CodeSamples: Remove generated files from repository
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
2017-05-29 22:14 ` [RFC PATCH 1/4] CodeSamples: Add rule to generate Makefile.arch and api.h Akira Yokosawa
@ 2017-05-29 22:16 ` Akira Yokosawa
2017-05-29 22:17 ` [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
` (3 subsequent siblings)
5 siblings, 0 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-29 22:16 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From 10024f2b41c129e5e7e892c60946c8dd5dd58007 Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Sun, 28 May 2017 09:40:21 +0900
Subject: [RFC PATCH 2/4] CodeSamples: Remove generated files from repository
Now, as Makefile.arch and api.h are generated when necessary,
they are no longer necessary in the repository.
Also fix entry for gprof-helper.c in .gitignore and remove an
executable file sneaked into repository under datastruct/Issaquah/.
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/.gitignore | 4 +-
CodeSamples/Makefile.arch | 6 -
CodeSamples/api.h | 808 -------------------------
CodeSamples/datastruct/Issaquah/existence_test | Bin 26359 -> 0 bytes
4 files changed, 3 insertions(+), 815 deletions(-)
delete mode 100644 CodeSamples/Makefile.arch
delete mode 100644 CodeSamples/api.h
delete mode 100755 CodeSamples/datastruct/Issaquah/existence_test
diff --git a/CodeSamples/.gitignore b/CodeSamples/.gitignore
index 9e3642f..e116055 100644
--- a/CodeSamples/.gitignore
+++ b/CodeSamples/.gitignore
@@ -1 +1,3 @@
-./gprof-helper.c
+/gprof-helper.c
+/Makefile.arch
+/api.h
diff --git a/CodeSamples/Makefile.arch b/CodeSamples/Makefile.arch
deleted file mode 100644
index 07a4b33..0000000
--- a/CodeSamples/Makefile.arch
+++ /dev/null
@@ -1,6 +0,0 @@
-# MECHANICALLY GENERATED, DO NOT EDIT!!!
-
-# For use on 64-bit systems, add -m32. You might also need to install
-# g++-multilib (thanks to robert.berger@reliableembeddedsystems.com).
-
-GCC_ARGS=-g -O3 -Wall
diff --git a/CodeSamples/api.h b/CodeSamples/api.h
deleted file mode 100644
index 3b4bc16..0000000
--- a/CodeSamples/api.h
+++ /dev/null
@@ -1,808 +0,0 @@
-#ifndef __PERFBOOK_API_H__
-#define __PERFBOOK_API_H__
-/* MECHANICALLY GENERATED, DO NOT EDIT!!! */
-
-/*
- * common.h: Common Linux kernel-isms.
- *
- * 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; but version 2 of the License only due
- * to code included from the Linux kernel.
- *
- * 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, you can access it online at
- * http://www.gnu.org/licenses/gpl-2.0.html.
- *
- * Copyright (c) 2006 Paul E. McKenney, IBM.
- *
- * Much code taken from the Linux kernel. For such code, the option
- * to redistribute under later versions of GPL might not be available.
- */
-
-#ifndef __always_inline
-#define __always_inline inline
-#endif
-#ifndef __maybe_unused
-#define __maybe_unused __attribute__((unused))
-#endif
-
-#define BUILD_BUG_ON(condition) ((void)sizeof(char[1 - 2*!!(condition)]))
-#define BUILD_BUG_ON_ZERO(e) (sizeof(char[1 - 2 * !!(e)]) - 1)
-
-#ifdef __ASSEMBLY__
-# define stringify_in_c(...) __VA_ARGS__
-# define ASM_CONST(x) x
-#else
-/* This version of stringify will deal with commas... */
-# define __stringify_in_c(...) #__VA_ARGS__
-# define stringify_in_c(...) __stringify_in_c(__VA_ARGS__) " "
-# define __ASM_CONST(x) x##UL
-# define ASM_CONST(x) __ASM_CONST(x)
-#endif
-
-
-/*
- * arch-i386.h: Expose x86 atomic instructions. 80486 and better only.
- *
- * 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, but version 2 only due to inclusion
- * of Linux-kernel code.
- *
- * 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, you can access it online at
- * http://www.gnu.org/licenses/gpl-2.0.html.
- *
- * Copyright (c) 2006 Paul E. McKenney, IBM.
- *
- * Much code taken from the Linux kernel. For such code, the option
- * to redistribute under later versions of GPL might not be available.
- */
-
-/*
- * Machine parameters.
- */
-
-#define CONFIG_SMP
-
-#define CACHE_LINE_SIZE 64
-#define ____cacheline_internodealigned_in_smp \
- __attribute__((__aligned__(1 << 6)))
-
-#define LOCK_PREFIX "lock ; "
-
-/* These are x86-specific, used by some header files */
-#define atomic_clear_mask(mask, addr) \
-__asm__ __volatile__(LOCK_PREFIX "andl %0,%1" \
-: : "r" (~(mask)),"m" (*addr) : "memory")
-
-#define atomic_set_mask(mask, addr) \
-__asm__ __volatile__(LOCK_PREFIX "orl %0,%1" \
-: : "r" (mask),"m" (*(addr)) : "memory")
-
-/* Atomic operations are already serializing on x86 */
-#define smp_mb__before_atomic_dec() barrier()
-#define smp_mb__after_atomic_dec() barrier()
-#define smp_mb__before_atomic_inc() barrier()
-#define smp_mb__after_atomic_inc() barrier()
-
-#define smp_mb() \
-__asm__ __volatile__("mfence" : : : "memory")
-/* __asm__ __volatile__("lock; addl $0,0(%%esp)" : : : "memory") */
-
-
-/*
- * Generate 64-bit timestamp.
- */
-
-static __inline__ long long get_timestamp(void)
-{
- unsigned int __a,__d;
-
- __asm__ __volatile__("rdtsc" : "=a" (__a), "=d" (__d));
- return ((long long)__a) | (((long long)__d)<<32);
-}
-
-/*
- * api_pthreads.h: API mapping to pthreads environment.
- *
- * 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. However, please note that much
- * of the code in this file derives from the Linux kernel, and that such
- * code may not be available except under GPLv2.
- *
- * 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, you can access it online at
- * http://www.gnu.org/licenses/gpl-2.0.html.
- *
- * Copyright (c) 2006 Paul E. McKenney, IBM.
- */
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <unistd.h>
-#include <strings.h>
-#include <errno.h>
-#include <limits.h>
-#include <sys/types.h>
-#include <sys/wait.h>
-#include <poll.h>
-#include <sys/time.h>
-#include <string.h>
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#endif /* #ifndef _GNU_SOURCE */
-#ifndef __USE_GNU
-#define __USE_GNU
-#endif /* #ifndef __USE_GNU */
-#include <pthread.h>
-#include <sched.h>
-#include <sys/param.h>
-/* #include "atomic.h" */
-
-/*
- * Compiler magic.
- */
-#ifndef offsetof
-#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
-#endif /* #ifndef offsetof */
-#ifndef container_of
-#define container_of(ptr, type, member) ({ \
- const typeof( ((type *)0)->member ) *__mptr = (ptr); \
- (type *)( (char *)__mptr - offsetof(type,member) );})
-#endif /* #ifndef offsetof */
-#define barrier() __asm__ __volatile__("": : :"memory")
-
-/*
- * Default machine parameters.
- */
-
-#ifndef CACHE_LINE_SIZE
-#define CACHE_LINE_SIZE 128
-#endif /* #ifndef CACHE_LINE_SIZE */
-
-/*
- * Exclusive locking primitives.
- */
-
-typedef pthread_mutex_t spinlock_t;
-
-#define DEFINE_SPINLOCK(lock) spinlock_t lock = PTHREAD_MUTEX_INITIALIZER;
-#define __SPIN_LOCK_UNLOCKED(lockp) PTHREAD_MUTEX_INITIALIZER
-
-static __inline__ void spin_lock_init(spinlock_t *sp)
-{
- int ret;
-
-retry:
- ret = pthread_mutex_init(sp, NULL);
- if (ret) {
- if (ret == EINTR)
- goto retry;
- fprintf(stderr, "spin_lock_init:pthread_mutex_init %d\n", ret);
- abort();
- }
-}
-
-static __inline__ void spin_lock(spinlock_t *sp)
-{
- if (pthread_mutex_lock(sp) != 0) {
- perror("spin_lock:pthread_mutex_lock");
- abort();
- }
-}
-
-static __inline__ int spin_trylock(spinlock_t *sp)
-{
- int retval;
-
- if ((retval = pthread_mutex_trylock(sp)) == 0)
- return 1;
- if (retval == EBUSY)
- return 0;
- perror("spin_trylock:pthread_mutex_trylock");
- abort();
-}
-
-static __inline__ void spin_unlock(spinlock_t *sp)
-{
- if (pthread_mutex_unlock(sp) != 0) {
- perror("spin_unlock:pthread_mutex_unlock");
- abort();
- }
-}
-
-static __inline__ int spin_is_locked(spinlock_t *sp)
-{
- if (spin_trylock(sp)) {
- spin_unlock(sp);
- return 0;
- }
- return 1;
-}
-
-#define spin_lock_irqsave(l, f) do { f = 1; spin_lock(l); } while (0)
-#define spin_unlock_irqrestore(l, f) do { f = 0; spin_unlock(l); } while (0)
-
-#define ACCESS_ONCE(x) (*(volatile typeof(x) *)&(x))
-#define READ_ONCE(x) ({ typeof(x) ___x = ACCESS_ONCE(x); ___x; })
-#define WRITE_ONCE(x, val) ({ ACCESS_ONCE(x) = (val); })
-#ifndef unlikely
-#define unlikely(x) x
-#endif /* #ifndef unlikely */
-#ifndef likely
-#define likely(x) x
-#endif /* #ifndef likely */
-#define prefetch(x) x
-
-/*
- * Thread creation/destruction primitives.
- */
-
-typedef pthread_t thread_id_t;
-
-#define NR_THREADS 128
-
-#define __THREAD_ID_MAP_EMPTY 0
-#define __THREAD_ID_MAP_WAITING 1
-thread_id_t __thread_id_map[NR_THREADS];
-spinlock_t __thread_id_map_mutex;
-
-#define for_each_thread(t) \
- for (t = 0; t < NR_THREADS; t++)
-
-#define for_each_running_thread(t) \
- for (t = 0; t < NR_THREADS; t++) \
- if ((__thread_id_map[t] != __THREAD_ID_MAP_EMPTY) && \
- (__thread_id_map[t] != __THREAD_ID_MAP_WAITING))
-
-#define for_each_tid(t, tid) \
- for (t = 0; t < NR_THREADS; t++) \
- if ((((tid) = __thread_id_map[t]) != __THREAD_ID_MAP_EMPTY) && \
- ((tid) != __THREAD_ID_MAP_WAITING))
-
-pthread_key_t thread_id_key;
-
-static __inline__ int num_online_threads(void)
-{
- int t;
- int nonline = 0;
-
- for_each_running_thread(t)
- nonline++;
- return nonline;
-}
-
-static __inline__ int __smp_thread_id(void)
-{
- int i;
- thread_id_t tid = pthread_self();
-
- for (i = 0; i < NR_THREADS; i++) {
- if (__thread_id_map[i] == tid) {
- long v = i + 1; /* must be non-NULL. */
-
- if (pthread_setspecific(thread_id_key, (void *)v) != 0) {
- perror("pthread_setspecific");
- exit(-1);
- }
- return i;
- }
- }
- spin_lock(&__thread_id_map_mutex);
- for (i = 0; i < NR_THREADS; i++) {
- if (__thread_id_map[i] == tid) {
- spin_unlock(&__thread_id_map_mutex);
- return i;
- }
- }
- spin_unlock(&__thread_id_map_mutex);
- fprintf(stderr, "smp_thread_id: Rogue thread, id: %d(%#x)\n",
- (int)tid, (int)tid);
- exit(-1);
-}
-
-static __inline__ int smp_thread_id(void)
-{
- void *id;
-
- id = pthread_getspecific(thread_id_key);
- if (id == NULL)
- return __smp_thread_id();
- return (long)(id - 1);
-}
-
-static __inline__ thread_id_t create_thread(void *(*func)(void *), void *arg)
-{
- thread_id_t tid;
- int i;
-
- spin_lock(&__thread_id_map_mutex);
- for (i = 0; i < NR_THREADS; i++) {
- if (__thread_id_map[i] == __THREAD_ID_MAP_EMPTY)
- break;
- }
- if (i >= NR_THREADS) {
- spin_unlock(&__thread_id_map_mutex);
- fprintf(stderr, "Thread limit of %d exceeded!\n", NR_THREADS);
- exit(-1);
- }
- __thread_id_map[i] = __THREAD_ID_MAP_WAITING;
- if (pthread_create(&tid, NULL, func, arg) != 0) {
- perror("create_thread:pthread_create");
- exit(-1);
- }
- __thread_id_map[i] = tid;
- spin_unlock(&__thread_id_map_mutex);
- return tid;
-}
-
-static __inline__ void *wait_thread(thread_id_t tid)
-{
- int i;
- void *vp;
-
- for (i = 0; i < NR_THREADS; i++) {
- if (__thread_id_map[i] == tid)
- break;
- }
- if (i >= NR_THREADS){
- fprintf(stderr, "wait_thread: bad tid = %d(%#x)\n",
- (int)tid, (int)tid);
- exit(-1);
- }
- if (pthread_join(tid, &vp) != 0) {
- perror("wait_thread:pthread_join");
- exit(-1);
- }
- __thread_id_map[i] = __THREAD_ID_MAP_EMPTY;
- return vp;
-}
-
-static __inline__ void wait_all_threads(void)
-{
- int i;
- thread_id_t tid;
-
- for (i = 1; i < NR_THREADS; i++) {
- tid = __thread_id_map[i];
- if (tid != __THREAD_ID_MAP_EMPTY &&
- tid != __THREAD_ID_MAP_WAITING)
- (void)wait_thread(tid);
- }
-}
-
-/*
- * Wait on all child processes.
- */
-static __inline__ void waitall(void)
-{
- int pid;
- int status;
-
- for (;;) {
- pid = wait(&status);
- if (pid == -1) {
- if (errno == ECHILD)
- break;
- perror("wait");
- exit(-1);
- }
- poll(NULL, 0, 1);
- }
-}
-
-static __inline__ void run_on(int cpu)
-{
- cpu_set_t mask;
-
- CPU_ZERO(&mask);
- CPU_SET(cpu, &mask);
- sched_setaffinity(0, sizeof(mask), &mask);
-}
-
-/*
- * timekeeping -- very crude -- should use MONOTONIC...
- */
-
-static __inline__ long long get_microseconds(void)
-{
- struct timeval tv;
-
- if (gettimeofday(&tv, NULL) != 0)
- abort();
- return ((long long)tv.tv_sec) * 1000000LL + (long long)tv.tv_usec;
-}
-
-/*
- * Per-thread variables.
- */
-
-#define DEFINE_PER_THREAD(type, name) \
- struct { \
- __typeof__(type) v \
- __attribute__((__aligned__(CACHE_LINE_SIZE))); \
- } __per_thread_##name[NR_THREADS];
-#define DECLARE_PER_THREAD(type, name) extern DEFINE_PER_THREAD(type, name)
-
-#define per_thread(name, thread) __per_thread_##name[thread].v
-#define __get_thread_var(name) per_thread(name, smp_thread_id())
-
-#define init_per_thread(name, v) \
- do { \
- int __i_p_t_i; \
- for (__i_p_t_i = 0; __i_p_t_i < NR_THREADS; __i_p_t_i++) \
- per_thread(name, __i_p_t_i) = v; \
- } while (0)
-
-/*
- * CPU traversal primitives.
- */
-
-#ifndef NR_CPUS
-#define NR_CPUS 16
-#endif /* #ifndef NR_CPUS */
-
-#define for_each_possible_cpu(cpu) \
- for (cpu = 0; cpu < NR_CPUS; cpu++)
-#define for_each_online_cpu(cpu) \
- for (cpu = 0; cpu < NR_CPUS; cpu++)
-
-/*
- * Per-CPU variables.
- */
-
-#define DEFINE_PER_CPU(type, name) \
- struct { \
- __typeof__(type) v \
- __attribute__((__aligned__(CACHE_LINE_SIZE))); \
- } __per_cpu_##name[NR_CPUS]
-#define DECLARE_PER_CPU(type, name) extern DEFINE_PER_CPU(type, name)
-
-DEFINE_PER_THREAD(int, smp_processor_id);
-
-static __inline__ int smp_processor_id(void)
-{
- return __get_thread_var(smp_processor_id);
-}
-
-static __inline__ void set_smp_processor_id(int cpu)
-{
- __get_thread_var(smp_processor_id) = cpu;
-}
-
-#define per_cpu(name, thread) __per_cpu_##name[thread].v
-#define __get_cpu_var(name) per_cpu(name, smp_processor_id())
-
-#define init_per_cpu(name, v) \
- do { \
- int __i_p_c_i; \
- for (__i_p_c_i = 0; __i_p_c_i < NR_CPUS; __i_p_c_i++) \
- per_cpu(name, __i_p_c_i) = v; \
- } while (0)
-
-/*
- * CPU state checking (crowbarred).
- */
-
-#define idle_cpu(cpu) 0
-#define in_softirq() 1
-#define hardirq_count() 0
-#define PREEMPT_SHIFT 0
-#define SOFTIRQ_SHIFT (PREEMPT_SHIFT + PREEMPT_BITS)
-#define HARDIRQ_SHIFT (SOFTIRQ_SHIFT + SOFTIRQ_BITS)
-#define PREEMPT_BITS 8
-#define SOFTIRQ_BITS 8
-
-/*
- * CPU hotplug.
- */
-
-struct notifier_block {
- int (*notifier_call)(struct notifier_block *, unsigned long, void *);
- struct notifier_block *next;
- int priority;
-};
-
-#define CPU_ONLINE 0x0002 /* CPU (unsigned)v is up */
-#define CPU_UP_PREPARE 0x0003 /* CPU (unsigned)v coming up */
-#define CPU_UP_CANCELED 0x0004 /* CPU (unsigned)v NOT coming up */
-#define CPU_DOWN_PREPARE 0x0005 /* CPU (unsigned)v going down */
-#define CPU_DOWN_FAILED 0x0006 /* CPU (unsigned)v NOT going down */
-#define CPU_DEAD 0x0007 /* CPU (unsigned)v dead */
-#define CPU_DYING 0x0008 /* CPU (unsigned)v not running any task,
- * not handling interrupts, soon dead */
-#define CPU_POST_DEAD 0x0009 /* CPU (unsigned)v dead, cpu_hotplug
- * lock is dropped */
-
-/* Used for CPU hotplug events occuring while tasks are frozen due to a suspend
- * operation in progress
- */
-#define CPU_TASKS_FROZEN 0x0010
-
-#define CPU_ONLINE_FROZEN (CPU_ONLINE | CPU_TASKS_FROZEN)
-#define CPU_UP_PREPARE_FROZEN (CPU_UP_PREPARE | CPU_TASKS_FROZEN)
-#define CPU_UP_CANCELED_FROZEN (CPU_UP_CANCELED | CPU_TASKS_FROZEN)
-#define CPU_DOWN_PREPARE_FROZEN (CPU_DOWN_PREPARE | CPU_TASKS_FROZEN)
-#define CPU_DOWN_FAILED_FROZEN (CPU_DOWN_FAILED | CPU_TASKS_FROZEN)
-#define CPU_DEAD_FROZEN (CPU_DEAD | CPU_TASKS_FROZEN)
-#define CPU_DYING_FROZEN (CPU_DYING | CPU_TASKS_FROZEN)
-
-/* Hibernation and suspend events */
-#define PM_HIBERNATION_PREPARE 0x0001 /* Going to hibernate */
-#define PM_POST_HIBERNATION 0x0002 /* Hibernation finished */
-#define PM_SUSPEND_PREPARE 0x0003 /* Going to suspend the system */
-#define PM_POST_SUSPEND 0x0004 /* Suspend finished */
-#define PM_RESTORE_PREPARE 0x0005 /* Going to restore a saved image */
-#define PM_POST_RESTORE 0x0006 /* Restore failed */
-
-#define NOTIFY_DONE 0x0000 /* Don't care */
-#define NOTIFY_OK 0x0001 /* Suits me */
-#define NOTIFY_STOP_MASK 0x8000 /* Don't call further */
-#define NOTIFY_BAD (NOTIFY_STOP_MASK|0x0002)
- /* Bad/Veto action */
-/*
- * Clean way to return from the notifier and stop further calls.
- */
-#define NOTIFY_STOP (NOTIFY_OK|NOTIFY_STOP_MASK)
-
-/*
- * Bug checks.
- */
-
-#define BUG_ON(c) do { if (c) abort(); } while (0)
-
-/*
- * Initialization -- Must be called before calling any primitives.
- */
-
-static __inline__ void smp_init(void)
-{
- int i;
-
- spin_lock_init(&__thread_id_map_mutex);
- __thread_id_map[0] = pthread_self();
- for (i = 1; i < NR_THREADS; i++)
- __thread_id_map[i] = __THREAD_ID_MAP_EMPTY;
- init_per_thread(smp_processor_id, 0);
- if (pthread_key_create(&thread_id_key, NULL) != 0) {
- perror("pthread_key_create");
- exit(-1);
- }
-}
-
-/*
- * api-gcc.h: API mapping to pthreads gcc environment.
- * Uses C11-like gcc intrinsics.
- *
- * 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. However, please note that much
- * of the code in this file derives from the Linux kernel, and that such
- * code may not be available except under GPLv2.
- *
- * 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, you can access it online at
- * http://www.gnu.org/licenses/gpl-2.0.html.
- *
- * Copyright (c) 2016 Paul E. McKenney, IBM.
- */
-
-/*
- * Atomic data structure, initialization, and access.
- */
-
-typedef struct { volatile int counter; } atomic_t;
-
-#define ATOMIC_INIT(i) { (i) }
-
-#define atomic_read(v) \
- __atomic_load_n(&(v)->counter, __ATOMIC_RELAXED)
-#define atomic_set(v, i) \
- __atomic_store_n(&(v)->counter, (i), __ATOMIC_RELAXED)
-#define smp_load_acquire(p) \
- __atomic_load_n(p, __ATOMIC_ACQUIRE)
-#define smp_store_release(p, i) \
- __atomic_store_n(p, (i), __ATOMIC_RELEASE)
-
-/*
- * Atomic operations.
- */
-
-/**
- * atomic_add - add integer to atomic variable
- * @i: integer value to add
- * @v: pointer of type atomic_t
- *
- * Atomically adds @i to @v.
- */
-static __inline__ void atomic_add(int i, atomic_t *v)
-{
- __atomic_add_fetch(&v->counter, i, __ATOMIC_RELAXED);
-}
-
-/**
- * atomic_sub - subtract the atomic variable
- * @i: integer value to subtract
- * @v: pointer of type atomic_t
- *
- * Atomically subtracts @i from @v.
- */
-static __inline__ void atomic_sub(int i, atomic_t *v)
-{
- __atomic_sub_fetch(&v->counter, i, __ATOMIC_RELAXED);
-}
-
-/**
- * atomic_sub_and_test - subtract value from variable and test result
- * @i: integer value to subtract
- * @v: pointer of type atomic_t
- *
- * Atomically subtracts @i from @v and returns
- * true if the result is zero, or false for all
- * other cases.
- */
-static __inline__ int atomic_sub_and_test(int i, atomic_t *v)
-{
- return __atomic_sub_fetch(&v->counter, i, __ATOMIC_SEQ_CST) == 0;
-}
-
-/**
- * atomic_inc - increment atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1.
- */
-static __inline__ void atomic_inc(atomic_t *v)
-{
- atomic_add(1, v);
-}
-
-/**
- * atomic_dec - decrement atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically decrements @v by 1.
- */
-static __inline__ void atomic_dec(atomic_t *v)
-{
- atomic_sub(1, v);
-}
-
-/**
- * atomic_dec_and_test - decrement and test
- * @v: pointer of type atomic_t
- *
- * Atomically decrements @v by 1 and
- * returns true if the result is 0, or false for all other
- * cases.
- */
-static __inline__ int atomic_dec_and_test(atomic_t *v)
-{
- return atomic_sub_and_test(1, v);
-}
-
-/**
- * atomic_inc_and_test - increment and test
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-static __inline__ int atomic_inc_and_test(atomic_t *v)
-{
- return __atomic_add_fetch(&v->counter, 1, __ATOMIC_SEQ_CST) == 0;
-}
-
-/**
- * atomic_add_negative - add and test if negative
- * @v: pointer of type atomic_t
- * @i: integer value to add
- *
- * Atomically adds @i to @v and returns true
- * if the result is negative, or false when
- * result is greater than or equal to zero.
- */
-static __inline__ int atomic_add_negative(int i, atomic_t *v)
-{
- return __atomic_add_fetch(&v->counter, 1, __ATOMIC_SEQ_CST) < 0;
-}
-
-/**
- * atomic_add_return - add and return
- * @v: pointer of type atomic_t
- * @i: integer value to add
- *
- * Atomically adds @i to @v and returns @i + @v
- */
-static __inline__ int atomic_add_return(int i, atomic_t *v)
-{
- return __atomic_add_fetch(&v->counter, 1, __ATOMIC_SEQ_CST);
-}
-
-static __inline__ int atomic_sub_return(int i, atomic_t *v)
-{
- return atomic_add_return(-i, v);
-}
-
-struct __xchg_dummy {
- unsigned long a[100];
-};
-#define __xg(x) ((struct __xchg_dummy *)(x))
-
-#define cmpxchg(ptr, o, n) \
-({ \
- typeof(*ptr) _____actual = (o); \
- \
- (void)__atomic_compare_exchange_n(ptr, (void *)&_____actual, (n), 1, \
- __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST); \
- _____actual; \
-})
-
-static __inline__ int atomic_cmpxchg(atomic_t *v, int old, int new)
-{
- return cmpxchg(&v->counter, old, new);
-}
-
-#define xchg(ptr, v) __atomic_exchange_n((ptr), (v), __ATOMIC_SEQ_CST)
-#define atomic_xchg(ptr, v) \
- __atomic_exchange_n(&(ptr)->counter, (v), __ATOMIC_SEQ_CST)
-
-/**
- * atomic_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns non-zero if @v was not @u, and zero otherwise.
- */
-#define atomic_add_unless(v, a, u) \
-({ \
- int c, old; \
- c = atomic_read(v); \
- for (;;) { \
- if (unlikely(c == (u))) \
- break; \
- old = atomic_cmpxchg((v), c, c + (a)); \
- if (likely(old == c)) \
- break; \
- c = old; \
- } \
- c != (u); \
-})
-#define atomic_inc_not_zero(v) atomic_add_unless((v), 1, 0)
-
-#define atomic_inc_return(v) (atomic_add_return(1,v))
-#define atomic_dec_return(v) (atomic_sub_return(1,v))
-
-#define _LGPL_SOURCE
-#include <urcu/rculist.h>
-#include <urcu/rcuhlist.h>
-#include <urcu-pointer.h>
-#endif /* #ifndef __PERFBOOK_API_H__ */
diff --git a/CodeSamples/datastruct/Issaquah/existence_test b/CodeSamples/datastruct/Issaquah/existence_test
deleted file mode 100755
index a3ef38115581c15944e82e5d57ddc86e8da4afb9..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001
literal 26359
zcmeHwdwf*Yz3<vHJIT&u$OIArA>cp~<T*(oV0dYGKSWFc6`#W-nIuz^Oq!Viv|0lJ
zg$8Q0YV9qywYAn(ZMF4P(Vkcov{zfJ)>5rKz8ec|salS&n)Cgx$IMPnKK0(uIe*>j
z&S%Y9zxQuF_gZ`Hy>jKUHOno_66W#<uOPJZYTKzo{NM&fsS-iq6O+XxQ6jQ|BCV<u
zJP0)qV8{Z_M(`pe0VlWH4nrYe2GW3*0g(`pWpGXsf`Rg)K+{h_`p2Lntb~xiR7ks|
zdq64zy$CdhBY+v`7i}=?hRkl*W>^ZC!O(J$7B2k+I}u2)L8yH**>LA@A@01^b{HN5
zj)8QB&w-BUzg))tI?~<xb*3rHWO+-huDqpTMoX+MwJns0hh{3h$eN_DSbL7t&3sbl
zXavd@A+T&o_uf(8@xkW2YsUEIPx?;%uevwhzUm^<Mj#lu-6Cj-H*4~Uqb&6-UiRB#
z2VTGK^6GcGi|;)1_1+_2*jW6teSgnC`rP00K^}1Fs+TzGm_X<CMR<_`5EH&^5d7*O
z_^*RtSF@Ay`&{%O(tmKl)E^jRKQ@StDobBMq}WEyMfv*;fSB;RgW$43aNi(!66~+M
z8*LI-f%tp>T)_V4GOz=<6Li+qa4|+ajridn0|&&6LGo)sFTP62j}UocKj8h~M+E;+
z9cU-*$>xq|q=BuZLWEau39pMb#S+PA$C8#vA`wjp%@D2!TQZskKOc{^rGeH|GP*4s
zYl|fZp;B!v@%qgo9FBH$w8g^!BFR{Mu%PB?YK^yr6Uj(NG8`7+MyR8MsgC;8j6|%d
zEz&|KD@A>zr6mkx7)UbI6Fo<SSDg*vhFDuTm54To=r%e7v+)iQsf%|2h_|%Blg9Rr
zSX;6&T;IG|v_e(9UO)_tpi!Ilpf!U0HFiX!A{=dqBqN~KB@&7VR9k~sv1ZlcCE=N&
zS)r<Qx+0ybL>Y6e^T=PX>{t9{-(j#Z8gV4F^e<BuC<j4fKdnGWo*c_1=jo_d;tFE1
zVeFx+0rMh1lcez)F6mif!vKCyn*SKavTW2o+fWueEaDvS+Kl0p!x)|1&lruYfH4Nm
zBE}dfMl;6XR?HZBf{fu|DPx}ylNsj;F`Y3ytze91R>e3VL^WeLyMQq|)B3JsP9Gh=
zaQneuPZ6TK>;3-z{$0JvtiBt7?Ry5k^-Y7~uE(4USunWzXTOhJ^p|~%UkR5n7tKt*
z{R*V}%b1&H(%ybX!pv1Oi|FlM2{U)i9ID>FTf)p`bCB@2CCuD5PbS<WVdlEIf^esV
znfvBy!tD}fDKys*-Xvj`M)OL-wGw8jG}jWYkuXcAc_ZNp3A2=%HxUj>n5EUs482_-
zVU}8RJ7FPVmR|EV!pA@9?>}eHo9hqsq0+kEIKD4AN%UQait9RRrA}1#c3=GQlLvcx
zdMf{N$(~n1uidb&c~cLhk6&XupF6jC6Yjx%XF}<FVA2=$bse)i7w&`^kvij+-G~qP
zQQF&&I{1C`BM-FCJ(3u|PY!ooz1DZh(qFWTWXV4FNJrnHEGn?N9<wjJ_#hqI_wcFk
zSwMf+vC+2_fEN`#;z8n&$USVA64(EX^w@aQ)6>;bY8Smd2{~2v&V_eH(OY)HYlxlP
zy|gsg)z=XH?p`3e9xJ-=l7osb+C7`9w;$>RVc#V+M++3bZ!=VNRokg^4=*j{Kq~eW
zm-fBzPpV737Tq(On!65dj2_OSBpHz0ok<=s$vw0GfFy{oXw2JjB#@8Z9u(`X(1<iJ
z{e4G}?%8+mBr5AVRvhg<iAMG$ON+FZdoC(1*mHU5=v~hyPu+g#4HBzL>0j+`sQNv`
zyTQ>7fo^bhJyv|MXW!3d#=<p613j|dzX~qqE8@=ACm)-9EtL5>%zUw^biO_<iwg&5
z@e;_IEItb5CX4?-TKa+f0T+b+KGIP~<M!1S>?t^0qYAs$Ib7rCpmn&W;0+WMGoxx|
z-@nl*$g@tFi5g_StG8&6jrxHqayIq^esmmk8`8%h4d|dQSlzxz?5?%D1^(=wgFO;1
z=&rH5M|9geG%OD8IpO9X1iRZS<-2+v6u7BBKTrkT&aNj@zU~dBMSBXm9`hgUYk-HW
zhpuCaUVysv9cGR&dV#)F)0ID!y(D+9wvl);aDDH)Y5fD}>qvh#a0BW3LixyYzQxAG
z0Hy2SgW}0Stsp)B7NL+Mh<pbP9mp3{=^FlWl71&isglZG)t2D55j{rw1B(92zAd7!
z`y-VxU0&>eRed8R!*?B+3+%K|vs!&%*LxU>jI9FdVt?D!8`$|02^1jmMj+8ln1?+m
z;J(`c8x=e60jDPRJ*}m0cHw@naJ%*lHYGH3I|-^N0=w!nKvFvNB035kT1|p+=p=~3
zCWWL!gpG=wL2znf-vVvi@4~IrcHhT6G%a)#fXu%#x9cUqP|}N0v2W`K(uG|=0Iu?-
zqa#qxT?ei0A9bR&1a|%%-J{#S`4w<tOnPhI($dL?6)5}7(o#QL6?%wM+ig4P?v7&p
zwbnZqdb_2Vw6{k|_dSE#D{{VzYbOjQlc_?hl%11##MSpc_|hfQjfR`9nvptFHv&mb
z*(y`4$G(ZWT;G98Xp*>)&io&NGc8P3!Z}L8&c_uq%f0Vat@Iuj?k}3}X5dg#1wF2E
z@&Px$N|sZq52ZAnHHU$-?<SqkF;bCakR_&_Iv+Ysj&>~t&w%!N0PUU4W119}Eu!yK
zN|**Uf)WZ)4Qh{;_y9)I#a0F6;Vd>Un(lh63U?-yN;8&weY^i7GjKkPG{+6oG^PVa
ztv`QDC#oM6<nM$yD{43-S=X>G+tqU}C0mh}0=TCXAowY;caXq$Tms8gdJVn#@I=O&
zkiwS}|C$RwQsLjjklXi57yc8Z^?kJ-EEuECIq%%{eKGjB{oqHZsNuY&v|`@@KZdBj
zJhFE&Ipy%9{hnR@$tjg5lsuYi^`7aas6q$HDpqCx^*<$R*CA`}C#iqONOj>Q;foI*
z9cG3?8NYk5ySOwxI>Nr1zudxgJ*Jb%!UIe-_H-S!`=9Cd0+o8NOHN99dffBRv*B!C
zKm6-bb4_$_IloS{V@1IQid^2zGpixjibt)f!7eXWPg-UWyIkjNkn62rORP1P48|LS
zWevgTw)$waA=)s(C)6q{tQKWyalQ818i^$bljJ(g*i_4{w3u>Iuff3MmJilN8iL7K
zLvSHHm|Rw}ZHiA=W4-fx!8Z!w*7y4Rxjw%4AN~E$pqF2dqTyoWR>Vb!8&KUfh?k<T
zH6z}MxRZ3mHz9ru@tugDN8F3}4aCnQeh+aUVjDx%$B2s%7eIe0;!?y*5!WDYKztG6
zorv2J-->uA;?fWM`@fBNDaPFQ5Z{Zq0+n??;-!eWtl?!{woX`=JJwjgZSRJiAaKP9
zz3;*vV3F@E&-bqm<R$HG;*2r#r_L;$$VEa8!d~PVs}6Zh#xI{!>;_IwRu|<vS9uoY
z`@6hL@`D%n^8G*@T$FDgT9L=mvlF2hW%Lqa0giQ7dDmR+>9V7~eEaaBgNv~|l0M#!
z_<8WjX9(bH&$lJM7x<5XACH*w7XVMcjv&}5b3Rw#McgsQz+q`6FX9C9<{yM(eL29%
z=TcxH-(=vZ&y8Eaasf_0C)0ejz|kK$i~%<qp#iw}fn#~X$Ab09+ahF)G`0cq52n%P
zeyyK;yokFSIQnaFj2{DTG;k?0p$t4{!@%-9F&*G5pj{5yINq_`by4H3^15n+{sh6o
zw)Ka1`uis!P5xb;#ly0ecX`iy91Y_j)}%mJfoCPg>qwdh^?S8<QGUTy_M-fvuB^5B
z6`s}kML?jY{fD6DFlzj9`p>p`Ib;#Zi#WE^&jUvnKf69FfzL|dvl95M1U@T)&r0C4
z68NkH{{JcgEMWVYV-qutDQr4?N2f6r6@0G4{W@%&$bGthe9|9(3j52PkMHlo-VJ$n
z*bdhwmp}>E5}a?Igun~SAMqYmyPS)5B5*ClJxi|V_V2PCruQP|8iI1%Gv~Y>OQ-mB
z;NfGOvd0mBxKlm30Qi4>anIZcH6lM-5UxVF5#ct3`w<Qyynyf)!U=?I>_v=3n1nC~
zVL8IN2u9C}B}?W7C!bT7YD=cd*TmW)9r0jQXijL>jLMWm&Yq%RxyoZMu)dMMe2na{
zX!KQz*duaL*ow}FXg4Ft-4MA-(0p0o(0Z&tL*BX{&<P}{a3zGu=i#P?_xTF`+Bx4v
zWZPNTUj&qOoJc3@C?m_c5|Um|jtwQ?{H8=^Z<oldeT<x(uLJ4f!#TTp8IuUY-gi2a
zCxgTO$`&NE>}>m6h`m<!ZXi8d@ssuVX=G|eGQ0EwBy(Pdu$>+H9n#L-_{pwd(o=$z
z{V-GQXjIviOy=6zi?|Eu`xy!$dkHo9t|4{l%SeixYd{;FU0sLBX~j?WJZY>%IM34F
zeCFL#Vmn`l$?OG+<~Ya6d4_`g&RpuNQJe*NeNv)G<c`jsvlo26{~+^n=>#*Ly+UF`
zOs-`1g`YX*@^Kh-j1r#f>FX-`<XO#REAjbNaOnD+;Z_21U^a+3BP{J~-~`;u8D+^b
zF0li{Ngb_0CvYj0<eaQQf8a*=n=?j(1%X3Qo-<a1MS*`ObFl_T2O3Bnr@`U?cV~0P
zYcLp?L%Tr@mIfA6^8^h}4qQ!D6E!$J&;ZACN-aJC_morwg48_8dO(6zfm~{ytaDTy
z;4(93x&{{nx=0OK+acpAsR^7*aHh3hf=dHbnKR3ZNN{;z4mDR<k4Vic0&Qfjww{*Y
z%D~kG7g>8H^P0e|baJ)zH3@DA+(Gb64Q>qlgy0$tUKDtphR)L9rof}5uC>afY(ro#
zeXi9$HwU~djdj*@(r$|=m_p(EbDo5Kf6l}7*|~?FKB7uQIG@n7pQ{3~oiN#dp_?39
z_GN&x?VnJZ4Tjl~4Te~@35lhH5_>AJZvz-k1tVlzdfH+m8}SH!Jh>j4F0Apq>EXkM
zqAda`-I_#BH)xJCaXBc?Pna%c5*^H`AylRy8zv}x(gQ#WEB6T7?a2CMhsb@A{gnJ`
z5a+%`f7y$^j1tLxnKIszPHQBd(23kX(h+Y79%JR~0NM6U`!VR!tZzD#Py<8MjrIUa
z-rPSLhP-)1)6NvcdGn7;$QGllKO;4oKM=hSQMOH?oLRWbIM(D-V8~gF(y^v6Nd<h2
zW=&Nkz>K{PY7U=`1+CClvKNrO`~<X+{fnq-t3pAJQdLP2Pf5YZj%H!aq)|^vk<5oR
ziyAy7#n4-ZRL(6h)@jY>YX{C+YFeNeg!3TXSV#iP>k2}r%Q3RValQvp>x^NN=6B+x
z)%;c3tZ~X=&RV3jY;+zVZL!jFkrN_qiPEx3dbd=0*8qPgl5Jms-;-yPZ=+>RhlMfG
zrgi|Nim~5^i=JmK%exko;zT<ECdZY~Iv#vx+w^1ndS+#fWu5XO?QgXxZ3P>t_OQhp
z%W4Eq(*8MQCv;%j-#W}H@?0w>N^?(2XRgKIT>Pz)LNFN_>#5Qpz-&blqU;y6wBO<m
zGPd00&1w@K_9><NwbDtqkn5W)ux4bVWCIdcY6<I>Q<+VIKSHq9PKEg7c|c7uxT(tk
zX|dx*Y<~&G(%@PpVqh%-sc9CSvhJ|3A><rI*beY5*pOaq)JW?`(hGuDAy{kgPJ3Z+
zZZDoQVs}U{(xCRjz!@*5TeqNItovk59EWu2mu=@~@FWQDRGB{@E2(zI#mLsX;OM6x
z55WJWEWBr}3P`L)coR}9AvHnP)U%XAy@`+|yU|NlEqLk>P`&Xq+e?P6-0GzxZ^^N-
z9c))B)@LOv8p2ylL19G%&CBXzSK>{qQg+L?*)6kB`6c$0tq}Yo!lZ>%c#MptU}Pg^
zZ!EXIfV6c&8tMe!HOf#BZ}LotLiGu4XreUqKIN@-%FG8DGh(K-fNj`oS>*45lCLR6
zQ7D=fZHFeW#WZ}LLh_9WRnn&{DPq;qB?`~93fT|xl&Wuo@B2#C-Dy>MX;mcOMpf(S
z*s!#!nbs5RMny`=-Qc)SDS25cVP`FpwaiMHX+6(E7^j$f!T5w?_F-~8OIF~xH1lk0
zBQ=*<)-zx_hHx5kzF2ZHn`Oo~<eHEwKU>VUMzMv2EbBOA{FpLwL~2RPg^V2B@iH^j
zmaLQla85?xIFM$pHq3KqWS+|Qx+&xxiJ%JBpqaro+FqtKR)B+3R%k53szKGxGNbV{
zYFr^Tejg4_A@>}FYFSx|@DK<>%;gr2{|CJPTGrg_V8Q=G4fs)m{yP-l{}2pk!jQbr
zUMc2HVP6P%<^yGAdAYMa?=KwI3OpPYf#SCKyzx-(ImJUCfI0${KO0=vyA-XXqBlJ|
zl_K#|+nIYRyvTilcT>2;`)uy3Ygk>@+!vuc_x0N(#m|=eN2AFr=5x2!Dm$(e@lBNB
zze0Nc*XYk`4)#_y7690A5KqX3zy%=KY9Ww~fq)AEbKQ=_Xn5gKvwV-5<$Kh60MR!=
z_gHeW@A0U~zFH)B)at;arv5gk{#n(`3H(QL-<K$;77{jB09jvYCLSw*n@CYB0Q*#=
zv!7@F?5$wRd7N`Tn~MRbl(#gs7;r+|l9G!7r-DgO3A<k}QZ^R@P9>AMwzDV?Zpg)e
zvxJ)DV!&DYDw29J;8Zt)MJ@)Md1`{G76Z<FdSMm=&H_bKivi~h1*ye=Q=>Qw@{UOf
zE(V-AJ+#clfU{gWu?#e4g~W0(;H+eU2tPCbZEy{vj!{BRe4JJE$zF|;b6SbFt5E>C
zm5&0e)&;)d)()WLy1+L=PmG0J7x+e5*Fgl1%XNWov<B6>z<06+)w;kpMuTcy;2Wz!
zwJz`#Yf!BVeB(5z)&;)t8dU27Ur>W;UErIbLA5UMP1K-T7x+r$9m1>&e3Pt0$d0E(
ztqXjUb&k}!z&BlkYF*$9SqYgzwJz|@wA5NmtqXj!ES9vVM6C;aRn}8dvsxGUs;xgr
zP^}Ami>zIeS*;6vtF0ahs&#?yObx1afp3il)w;lUmIl?jz_->?(_XbM@YQOc)w;m9
z&U#JS)$0PEUKjWt=8aOW3w)0-iH0oK1-_rF0#WM%-!F9QLnG&#LfJN(f^0j?j%+)`
zvQ0?rI|j>j!TSJ)vz^Ixf%j>P?M<!=yt$tHfG@1^zUi5c)`oQfQpy%ONH=Is-o$g*
z4!JJKD`k={)(|RF5U0-+*6RZQk?#VK>jM9a?6TzNy1@UE?C5%3;D4Di>2-nsk90(?
z3r4+w9_;-)=AVy%V>BiR0GNcGe<CZRJ|_KT&|fz6mlgeF>ot(<O|t0mm;%6<GK_Eb
zCYd}Ti7cA2P<1~Fp*RGpRVpjessedjq@xK4wu137VNGIDh~i#QJT*$|=g7qgfv4s?
zB*iIQV_4bt#Q?HxQiG#_@vie&xy<0Lz)awB1(<u09!(2K)5Jush`bv-WSkfi-g9Kp
zixTM%nWZC8OC`Y{<Hx(%BZ=O>Qk4|noEC4D;uE)^z}lpKQT!+}>zS%Np9r^<=TqtV
zc-$yG)9H4ZoLs525U1t?Rf{|3;|NcqkdmCF%5@XwK2t%QO5+<rjUiZ5z6`ySxzC>M
z<$F@!nr2ATeV}CFULr9UgL(Q%!0_&ZOKNEa@=Me*<7rT4P|Gw57`#-m#9%s%igR5m
zb_4l!st6UpQK=$E0fUz+wj0boqvCp{BE;O@4rDJ?luk2AnbC6QI>g)({WEe*l?J#%
zlB~o@1OH|)cM-#EKC28Mrxn)Upu#%MdIq$=T!YCqNIP(Yu4DyTb6Ne?xl`Hx{sfL=
z2-fKjFf+@5IvZ7}3hO-(vi5>_#)l@SO(0G)c<Bfq%}LA$;FgYDM+Hzp{Zk=cP8Ewr
zVX7+ie=n^+3_?4JOD`aimGKC*8NAg0TZ3r;r_|3EwvGO6Kwd!=%eNZ+AEov00^vpy
zS3G3&7b5G^3|{J=W-wm{r__HsFh>6YAbY7|)$fe{&yg$}({T{8_JO!&3}(ryB~Y8e
zOa0po<{w5spDEy)S&#e<gE#?e?QyEu4%GKSo{>W001~r;AWq*0=s7^szm5dojHo(@
zAIjE%>^;wur0{Qm8wUna#R2gOP&bm=<hh5`dw`n_YKTNX8pUJ;>kHFNvYZUKW|K}@
zHG=nI&##Rzd#;$nA&``dNxj4~<0HUyww%vKaLIys9JXIXur|!X6kHYzvoOuzWx?EM
zFuw$+EchpAL96h6#dgLbSR1z(6)#c1;H3&Ln!UsXz$sOX1P0g4hrnA4<Te1-MUO*8
z`BOm6;wU)%2vSqPcp67<&$TTwQ!!-dE6h|%WoiW&(wSmY{u<Qnr$^z}=n=b0nAT*D
zZw28#5;v8hpUE6=1#z0eOLwm^m>+;sy89g}fC}n=63AasMLpkhlKSr>S&j!Ufbb_0
z8>DUyUegRt>VDH;UILfY{l3yIs;wVFyY+cFDDye}c7!~1=qF)j?g2@Yci?Nq+&@c*
z1kbhhJlv34NLd1A+!Q#=JB5#|todBY<o<v&2Fseyl}zprHNcfj?i~v7bD;5K0U+lK
zSBeE($#?^vkD*_$WEOBG<1O&;KC4$U3%HW;p5nO!ky*){1g>*jinx-	<jwVBtz;
zK36jS7kK}NOAjFOzsj#U0B|MafBlD&vQR!|3$iUkkhhP8@=@Cajp3`Qr^^-W#K)C#
zzJ;1?{~9g}Yq(XGUzq33ADchi8etWnfITCHH-BE<{Jd3#@TgSe<(&>XW&k>g&Ycwi
z7Ujo?Tsei7H$N|rLKZ-5SQ!n~Xji$TtPIVUp<((nOspy-i{G+V6;gEA1YpXcJ3p;5
zi}-*R%ToLW#z)M`CP4yoF^vHOCS!%lq7*HHr~}uu=t)v>mhp9@rXnA7WYkI4seT$9
zoh8+t974{>d#XPu^EL*bX;GWZ_gGEMGSp&Cm0pbVsbWCIj1TD&AtRbx23%E0L@;B9
z)llQ33llP?JaBDdRoXJal5tjLA(=PkoevGAS)#I#__9H&vZ-nkWObc-P{pUZWl5h*
zX^bFwXb*g!d8$vs(?Zr6ev^OcnAA4ijb0_4sF8#j;~<vtDdgsp;xb?5ZnPGcgbG&<
z;bo=EX=H3>S`5YQ7MqnRbQVCB7DF8nob49N<YA7srMXUXfiPX|(gAR;22tr}_!poP
z&CAfw@YBkCHwO^|d)7MZoNU>*GTq7aEOaN^v}U&5|A+h5;NF5hW_l)`rKjr&eT}`M
zO7=tcxAbT<b~FmMIMcw%#^#GDLZ(xWM^&VI=qZH|XN{R2CfkmZV*jFQ*;!5Rl0XSZ
z0W~0@U$Iuo`24}ui_nKcy8o%BFMALxeoB?>I#UOCv1z7nO(z3;;SAYjkqEhYaf?=+
zvBF&-nTE?=p|eMGGbzjF#bJPDF)I%MdVvcshr!t{BN8#kjlj%7x&@8ZZoBT@jZIrq
z6|~GgU>t$L0X={vzCd>gl&I<>3pGOaB5F82qd{?qBHZyjqL#Nw>!bW(_?6#m6Ap#S
zLm^R~N_5E2*_-O?%VTq@=ctd`XEe2?%CY@l9&4*_Ni{?jp9acfvj*_h#gd7C$&k2o
zAblV+KS2Jkm~womoQ&1u>%eXNT)M438Ui3hq&*hG<`zBzNH#P?8<9p9lb5zf6G-AN
zZj!jK>u(7R21I;+9E*bwcaRonMG6H?`$%B|v$gRHsc0&iLC@f~&@&VrGW`5mi*pgl
zts(DWtogrxrR5F1bK$ptX?wi)c$|<^c(ZMJKfc^qu(kR}cioi&y36~{0pOmy!db8|
z;gqj&&Ij^)9%sb6Y1Y&38p!+d70w86-2(4~ht4`bwJ^2CJN%u6PE%{)!n+n8wbm{?
zbD=jkWSw0;zkFd;VbiUJP2Vf}pfGjgf-{?%@J!QMc~d&&&Fv|S%~<oT@5x^ezG=38
z{eRRhKDEA4p5_0IhVwtuF2czus!}$NOj~Il+*V@onW$G+TR}42g#Kl_QH-YIKcxw=
zUGPDLx74yddoEamI=^&>?Kz97D{arsj5|Mvy5D-IBwn_Im9@?5d*{OQXP`~A*wzM6
z+jdB5U`LnjaUQm<^SYdcK*yW~5?=(BowjumP>qUn6gj<@IS<)ZSn)={yA$S%ci3L<
zSG_yD9|-50Z+dg@Ty>Xc@<{K#D_48p6-mTN??1#7VC?(iYUf8#a={Z)u5^debFQ{I
zS;AiLjkdGn&6}KUka+crcY?nGHofCEB6XJXrDDfU0Go)OK=dkwuG+El$#21zz23$H
z$8FpDx69uF^iQv+%mUPP`9pwiumL@Cz}x)L0q@6OdI(*H#pCU|!U=k^&~8z>T%LIU
zj1j<Ndrt7@ybC=!Ace#6#>Q|mGhT$Z#yX;5bDDv`fdo8qAQ8JP3a)TC(t<M&L^#om
zvkU4{jiS9Hx&>&`MH@aB3b)p|bTxIvQ|<B$1n`FIA_<;bU`8gDN`yP2Es<@}22tOT
zz)=T@WVjh8KZuriTT@W}q}q7SLbM^s7ns7~w&=E`;6V#KAR^i!&s*Rr3tKy4$!NG8
zFEL~+G)0s6K&n388f(L$3~7y2lwpc=bVM%Y$qp^iwzzbbn&n{%>Vv0ntRcKPdZ{!j
zeG50XM4A#V6|j@eB@bZ;*QYwd_3@TeD~b-EYRNCP=)C;aDjfbo3I>v*O`YTrZ)spH
zB&Y-#kY8ej)mKsjjIc0KGxg0;(U^?HT13V@Wxh4iu{qiy2i!K7ck(z1Ef@}Oj?uxe
zIz1x6I+T?e98k=e(y4EWMml8Hn_xVdDepvNOH@^9hFRt(1G)^WV>mLTSRzv~syJtg
zaBCd*Objpa6iXt$wGEkxHq~!!&wRT!NDl_EjLrcqH(X!mDoxaQv;${#gcEFGnMxkm
zT;LQ;H{lGCCfaXooKuBb>WC&g&_2+_Of@38eASv|;0bSW*-vbZCF`4sRabkZ2)D-*
z$d{~_%%LSo0e@!Agzj{qqnn}P*sEkS$3j_GSSEYwRFLqNWTdVonr<)|x=g_u8XjIp
z6wuJf15kiyk2K&gkhSLom*|%f%bVk^(ekEPvK(iTG}guAo6DD=?AJ$H+gqZEa<+{`
zvLjWWEMJvKM9^BBMPsV19u0O-4HMLknud#>f{bO{VzbUbQh-wzmZ$7zYE%Fu8x(tw
z(WJV7>=cm(lqK_lR>Ei&B6cih7vOA{1@`$&Ekdsv7-YB9pDD6Kw?^7k9~0qJ8_#B$
z$(F|hS2E60l^LDF17%^CQXOyrCh8*fo6$E_R>CM#!<=Ys54U&3>!XQ8yaQ(GuI{z6
zC8DG_>|qKOZe!!;VK2IUDz~Ld^nrGq3WIUj^{u49_Jh0iqw%5<wKcKv%1=G@7nb^D
z8M1E?4e@Zijb<8KlbKHC%Cepn4EW#^>}U<qNUKQHN7^z2xG2DMVK{<EByGkcR9e2t
z%eQ}l_cJb6=^*0@jb~h`(hlQlm9`l_q0&J_@;U{-GZAtGUL|oFin%=}uV?W490DC+
zJQ6cpq=Sq{tF+CSPrQf+i@fdz{8LS5{JN$yenaC8?{UEIY8>NXnAuPl<9wCItXW<2
z0iTXwiQ0dG>H)`BwIt))RJwq%d}9l_VSJ~?8&!7${;9&*j1Or#<HH)y_$iGya(u^>
zUjkZU-M>I_{<R)~Wy$zEDjj6ZO%vi7b2G_ZmfQ_$MsVeM2o$%0Zbrz)Y6H*x(lyA9
zpeqJS4}sd9k(L`%CL7!Xx@n;F5GZ>62*P07+~9d|p!5(ZZuR^If%#|rtV+xHw<_&t
zoXbbtBH1gL@F76J&p1z|Wz0PRDq_t2BPn2V!!4uNfP|j|a>-qxq43-nvblwlX>CKG
zxKT8eH1~pr!gDt$K2YlrC~gmF=>hf09iXA`+ylB{pw=N!+!Pv0n)^OO;kol;b2}tc
z0z;s<okMAls7SXvgF{)f;ivJJg11!iUfax#i=XH;<ym1Kajje=vCgJ}_cUpJJw?cF
ziS?#7T^9HnQQAT`Bv}2hEb+#QfsMRTIT+79BFX^I!xz$exdo*4%GTr3JD@djQ|R~7
z>+~2k1d6*sTAQq2m$m`*&HbS<+&;<Fz!0eQ;0Q~u^;Gn~5L_i?P{<Kahe+p$`$Fy<
znUdkw&ro=7{@g3Aq;oa|in~GXoEfd$^cf1zZJ)OWY8?W_^;HCcottsJO3S!Yrc(}M
zZV{2kSfyNx#0_kl`*m%aa_>Wt85y>$Poqx;&o3e4mgC0dP6W2Vn-G{=?nfHE+}7(x
zFg8CGkB-AW!8|ZuhF^pdm$qV-ia6P!;%1YQ6Y+KtYKteMp+$>V%}7R?G=zs|p}JHI
zCq!bN6e83dNi>U4!=-J|r(&$`BxMU`UOchS4TM3%Y>oUX(cY32A$j~?2=id1<>7sy
zj<{U=grd#i#*Rp9G~C<(Ndt*c5+~}3P<=9v$z7;HMW3$+3lv9M@o+8F6o)eWgro(f
zO$o(YTcenJhZ>@FsittG1M@K)R;a;PTVvd$BXxCLIco}z#EhCWO3Ea$-Y`k6DP6>$
zk{Am*<|>$|;MpF&)#5vRcq*H|NaIz4z&%Lv$#X+N2qJAR(@r@<Gkmv0-|gTlHB5*N
z-%_M;%!2%0i{FSn1-=^K@EA~D<YQQifbAk_m*0;0;Q(K#!-GZXH+c#W&PBjuCdtQd
z&GrM}H)iHC$xDDR--+=Xu@9kuGe%>V`D56O!0*bK=C@_j!LSKw+9i&5d0mEp?Rkmg
zw`qI9M}Ley^0B>lAs9Y>%eEg3oxqU~51*wz^6f>Sue8H&>57ryyMyL3cCQA)@YRB^
z7JLOJlc3QD@?PiSWB1#H%2_~1^hG{iHz62(JY#b|2xBBg{TRF7)O;+12f_Cs_}&|W
z?=~0TofxZmUhHE-@ZIa;YrhUJql|zrRIV>$_ve6FK3*bzj3ML{ReqXH)aV3lFdu9S
zhA%VTWa{mIyZ9XFbD(c1zGq#0Jm9YYe8wl5FzMeR%{rr9dG0Rwj%XYr$w&RKAQ=CS
zLEo_<^ik)VE<Sz}`5ySz0B0`pF#Hw4=&Ql(p(Z$VK3GSx{1A77ZzcHlP6VK@8Xe2m
zijl6+F!`qHG$Yz#@Zg8BxjKUQc)$RN4UMn(n>5J}1%>$SXr|O<XyXR)T{Bk*zN}#`
z!#8aZ-}o9Kj+#mUv*F_#?o69}<Sr9p;UImaFV=iW%4_+VLY&6|Okd>Tb=Dxho7V}^
zD=F&7@S&Njwx?+qW14rNAp&QTj9txK+>Y0#D`2+_cw}AHExMK*Gh|yvn%9vCw&OWy
z=*#Fcush~>PFS5-t;THyE6W_86|5|999FO@P>#0>Rt7n)Dp(by`Nq~OM|?F-D!OEo
z-|8w@4LBV46pYO@T;^<FuN>j=wh+cKMdyep<QJ_9<`>24n_>my^$uJde-x}v5aKwa
zV3r&<s4$)=n6-in{fLoQK7Vk->W;^8K+%gB3HEmd^Vx*z*9smfVfJ4IkCHI^se)Al
zXMa@i$&$`~r(pNUN`H74&Rkyk4X_&?mjRISit&SB3=Xc#D}pYF{1XPj6VvUG1RG~H
z%M#cATbKdRLsK3WzQmpk*p!EbFScg`zFSY_JYtwYw-NHhRKymf(VPX(Rb@*e+>E#Y
z8+wS)?y*zKguZA$iL~vqzn$<6ir>PsqiX=W?SBJsyRLugy9cnjsP7j5xL!*kCjB>t
zh?sPK8_W$xwnWm8Bd!JfpoY1BZrTf<R}}*`--TOZq$nFiuLf-D-@-{!YXS4DTK2RN
z(C?m$;1(yo1bUtJe}re4xB;+fFF|p#{62u~h2_VyviVV{xhVgOLGUYpSAw5BJbU|3
zfce!9&#|I?ewE-pJNuu5<j2C%UdXfM(q1Lt3X~sV%CBHX^hNvU4}$srEx%dfNJ0Ki
zz<yny!#zb}A7FkFWAgKjLGpJ2Hu*0={vQI|4t?y|)W=6vwSZ}xFh3ACR}hI~fcYWb
zOGuM$y9Q0l6aikU%V)UfWHAA-`|NR^!S;-+yiNzqkIRhxjez-;4M&#YA|RRpn~VB)
z0WNTr*Eaz3nBoX%^yd!E-wAjmF!up=pN;;Irh^Qs5HA4Usr#4AFJO*$s{8`t4f5mp
zm?_V{0Cv~^2Y@->n;`WK6Q2Nn2L6qM0QF@d=9g%^)e#;`MfwUN=GUft8bCDLLp2f=
z;OEFg{uO}vUY?<!$%MY>fBhi19q=nGf7BE8bpqyRYizL8_q9R%{3_dA)b|wN9+$r_
z4x)1oVD2Y8EB^1K{}lFu%3c5slb=Uc0Zn^2sW6#HrWzZuF%%9j*>Lu{@S0WYH}K^4
z@X`U$>Lo4`9*Z}%#Ots>B-eZ4NNStFt|2x<lF^3HoXXjA1qpb7zYWlwN*uw?g=0gi
zwG}%<%FeR2OQo6R>lU50ENmc}G3ksIxww>8GMf$h$@`#3>}mRdov!pQN}4=y+g5FU
zap{XSD0VmHW@dQl`D+)QwQ30*c2VTX_9DDu&Do0=tqGsKeEIri8^RkFEnc&XWhRfF
zuTP}Zx%X!4!hM2cO^rEm9tZV@`7TU|+a~Sx*aAtlZ4T9K6I>wcBkI)_i(7_Sal->`
z&`~#4pL(Ckw%VQ?$~|t@4G=YZG6Ob$GMd~wlInzhw=|FGcf%REf%{FWm8lxfY|nD{
zSe@$c)}c0G(}(vD+81|i%A_|I2mCLBj0*!cJ}Eb_%)$+cco<tl4K3LD;YQnl|3{!+
z6AH`S=Zs33%IwW&2zs}C@INJR=PScJa6>iSnB+5=%oe9=orC@vgIl2~qQ>@c^H%Jg
z>J5GO|2~krV|sThV}HN~-VpyqBE3(T5n+48rgM9G7m`Prx}7nvMyMAL1~w{Jozb+*
sQ`i4@$SC9gXfW+v?YL!Vl>>FD{}z!UqqE6U#D3zx*lo;YOKl$iCp%eIu>b%7
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
2017-05-29 22:14 ` [RFC PATCH 1/4] CodeSamples: Add rule to generate Makefile.arch and api.h Akira Yokosawa
2017-05-29 22:16 ` [RFC PATCH 2/4] CodeSamples: Remove generated files from repository Akira Yokosawa
@ 2017-05-29 22:17 ` Akira Yokosawa
2017-05-30 0:10 ` Paul E. McKenney
2017-05-29 22:18 ` [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c Akira Yokosawa
` (2 subsequent siblings)
5 siblings, 1 reply; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-29 22:17 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From e325a5ff132d8e4a79dfe465b18f573b97da75db Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Sun, 28 May 2017 12:38:27 +0900
Subject: [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
On x86_64, casting between "int" and "void *" in threadcreate.c
causes GCC to emit warnings such as:
warning: cast from pointer to integer of different size
warning: cast from integer to pointer of different size
Let's adopt C99 way of writing portable code.
Also do the same changes where appropriate. (Other than
threadcreate.c, no warnings were observed on x86_64, but we can
remove a few casts to "long".)
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/SMPdesign/matmul.c | 13 +++++++------
CodeSamples/SMPdesign/smpalloc.c | 11 ++++++-----
CodeSamples/defer/gettimestampmp.c | 2 +-
CodeSamples/intro/threadcreate.c | 7 ++++---
4 files changed, 18 insertions(+), 15 deletions(-)
diff --git a/CodeSamples/SMPdesign/matmul.c b/CodeSamples/SMPdesign/matmul.c
index 6a07730..9f0b250 100644
--- a/CodeSamples/SMPdesign/matmul.c
+++ b/CodeSamples/SMPdesign/matmul.c
@@ -18,12 +18,13 @@
* Copyright (c) 2009 Paul E. McKenney, IBM Corporation.
*/
+#include <inttypes.h>
#include "../api.h"
float *a;
float *b;
float *c;
-long dim = 1000;
+intptr_t dim = 1000;
int nthread = 1;
#define GOFLAG_INIT 0
@@ -38,8 +39,8 @@ atomic_t nstarted;
void *matmul_thread(void *me_in)
{
- long me = (long)me_in;
- int i, j, k;
+ intptr_t me = (intptr_t)me_in;
+ intptr_t i, j, k;
atomic_inc(&nstarted);
while (goflag == GOFLAG_INIT)
@@ -58,7 +59,7 @@ void *matmul_thread(void *me_in)
int main(int argc, char *argv[])
{
- int i, j;
+ intptr_t i, j;
long long startcreatetime;
long long starttime;
long long endtime;
@@ -87,7 +88,7 @@ int main(int argc, char *argv[])
goflag = GOFLAG_INIT;
startcreatetime = get_microseconds();
for (i = 0; i < nthread; i++)
- create_thread(matmul_thread, (void *)(long)i);
+ create_thread(matmul_thread, (void *)i);
while (atomic_read(&nstarted) != nthread)
barrier();
starttime = get_microseconds();
@@ -95,7 +96,7 @@ int main(int argc, char *argv[])
while (atomic_read(&ndone) != nthread)
poll(NULL, 0, 1);
endtime = get_microseconds();
- printf("dim = %ld, nthread = %d, duration = %lld : %lld us\n",
+ printf("dim = %" PRIdPTR ", nthread = %d, duration = %lld : %lld us\n",
dim, nthread, endtime - startcreatetime, endtime - starttime);
return 0;
}
diff --git a/CodeSamples/SMPdesign/smpalloc.c b/CodeSamples/SMPdesign/smpalloc.c
index 95b21a1..0882f63 100644
--- a/CodeSamples/SMPdesign/smpalloc.c
+++ b/CodeSamples/SMPdesign/smpalloc.c
@@ -19,6 +19,7 @@
* Copyright (c) 2006 Paul E. McKenney, IBM Corporation.
*/
+#include <inttypes.h>
#include "../api.h"
#define TARGET_POOL_SIZE 3
@@ -123,7 +124,7 @@ void *memblock_test(void *arg)
long cnt = 0;
long cntfail = 0;
int i;
- int runlength = (int)(long)arg;
+ intptr_t runlength = (intptr_t)arg;
struct memblock *p[MAX_RUN];
if (runlength > MAX_RUN)
@@ -159,7 +160,7 @@ int main(int argc, char *argv[])
long long nc;
long long nf;
int nkids = 1;
- int runlength = 1;
+ intptr_t runlength = 1;
int totbefore;
smp_init();
@@ -176,12 +177,12 @@ int main(int argc, char *argv[])
if (argc > 2) {
runlength = strtoul(argv[2], NULL, 0);
if (runlength > MAX_RUN) {
- fprintf(stderr, "nkids = %d too large, max = %d\n",
+ fprintf(stderr, "nkids = %" PRIdPTR " too large, max = %d\n",
runlength, MAX_RUN);
usage(argv[0]);
}
}
- printf("%d %d ", nkids, runlength);
+ printf("%d %" PRIdPTR, nkids, runlength);
init_per_thread(results, 0L);
init_per_thread(failures, 0L);
@@ -189,7 +190,7 @@ int main(int argc, char *argv[])
goflag = 1;
for (i = 0; i < nkids; i++)
- create_thread(memblock_test, (void *)(long)runlength);
+ create_thread(memblock_test, (void *)runlength);
sleep(1);
goflag = 0;
diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
index bc0b9ea..e794e82 100644
--- a/CodeSamples/defer/gettimestampmp.c
+++ b/CodeSamples/defer/gettimestampmp.c
@@ -29,7 +29,7 @@ long curtimestamp = 0;
void *collect_timestamps(void *mask_in)
{
- long mask = (long)mask_in;
+ intptr_t mask = (intptr_t)mask_in;
while (curtimestamp < MAX_TIMESTAMPS) {
while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
diff --git a/CodeSamples/intro/threadcreate.c b/CodeSamples/intro/threadcreate.c
index 26b7ba9..470bb11 100644
--- a/CodeSamples/intro/threadcreate.c
+++ b/CodeSamples/intro/threadcreate.c
@@ -18,13 +18,14 @@
* Copyright (c) 2006 Paul E. McKenney, IBM Corporation.
*/
+#include <inttypes.h>
#include "../api.h"
void *thread_test(void *arg)
{
- int myarg = (int)arg;
+ intptr_t myarg = (intptr_t)arg;
- printf("child thread %d: smp_thread_id() = %d\n",
+ printf("child thread %" PRIdPTR ": smp_thread_id() = %d\n",
myarg, smp_thread_id());
return NULL;
}
@@ -38,7 +39,7 @@ void usage(char *progname)
int main(int argc, char *argv[])
{
- int i;
+ intptr_t i;
int nkids = 1;
smp_init();
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
2017-05-29 22:17 ` [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
@ 2017-05-30 0:10 ` Paul E. McKenney
0 siblings, 0 replies; 18+ messages in thread
From: Paul E. McKenney @ 2017-05-30 0:10 UTC (permalink / raw)
To: Akira Yokosawa; +Cc: perfbook
On Tue, May 30, 2017 at 07:17:27AM +0900, Akira Yokosawa wrote:
> >From e325a5ff132d8e4a79dfe465b18f573b97da75db Mon Sep 17 00:00:00 2001
> From: Akira Yokosawa <akiyks@gmail.com>
> Date: Sun, 28 May 2017 12:38:27 +0900
> Subject: [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
>
> On x86_64, casting between "int" and "void *" in threadcreate.c
> causes GCC to emit warnings such as:
>
> warning: cast from pointer to integer of different size
> warning: cast from integer to pointer of different size
>
> Let's adopt C99 way of writing portable code.
>
> Also do the same changes where appropriate. (Other than
> threadcreate.c, no warnings were observed on x86_64, but we can
> remove a few casts to "long".)
>
> Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
> ---
> CodeSamples/SMPdesign/matmul.c | 13 +++++++------
> CodeSamples/SMPdesign/smpalloc.c | 11 ++++++-----
> CodeSamples/defer/gettimestampmp.c | 2 +-
> CodeSamples/intro/threadcreate.c | 7 ++++---
> 4 files changed, 18 insertions(+), 15 deletions(-)
>
> diff --git a/CodeSamples/SMPdesign/matmul.c b/CodeSamples/SMPdesign/matmul.c
> index 6a07730..9f0b250 100644
> --- a/CodeSamples/SMPdesign/matmul.c
> +++ b/CodeSamples/SMPdesign/matmul.c
> @@ -18,12 +18,13 @@
> * Copyright (c) 2009 Paul E. McKenney, IBM Corporation.
> */
>
> +#include <inttypes.h>
> #include "../api.h"
>
> float *a;
> float *b;
> float *c;
> -long dim = 1000;
> +intptr_t dim = 1000;
This really is an int, used only for array index calculations. I don't
see how it helps to make it be an intptr_t. What am I missing?
> int nthread = 1;
>
> #define GOFLAG_INIT 0
> @@ -38,8 +39,8 @@ atomic_t nstarted;
>
> void *matmul_thread(void *me_in)
> {
> - long me = (long)me_in;
> - int i, j, k;
> + intptr_t me = (intptr_t)me_in;
> + intptr_t i, j, k;
Hmmm... Now "me" is a long and used as such, but the conversion from
a pointer on 32-bit systems could get warnings on some systems. But
it is guaranteed to be a small integer. Would something like this work?
int me = (int)(long)me_in;
Or this?
int me = (intptr_t)me_in;
Or even this?
int me = ((intptr_t)me_in) & 0xffffffff;
I suppose that the painfully correct way to do this would be to pass in
a pointer to a structure containing a lone int. Any other approaches
that work?
The variables i, j, and k, like dim, are used only as int to compute
array indexes.
I have similar concerns about the remaining cases. I am not inalterably
opposed, but it seems better to keep intptr_t usage more confined.
Thanx, Paul
> atomic_inc(&nstarted);
> while (goflag == GOFLAG_INIT)
> @@ -58,7 +59,7 @@ void *matmul_thread(void *me_in)
>
> int main(int argc, char *argv[])
> {
> - int i, j;
> + intptr_t i, j;
> long long startcreatetime;
> long long starttime;
> long long endtime;
> @@ -87,7 +88,7 @@ int main(int argc, char *argv[])
> goflag = GOFLAG_INIT;
> startcreatetime = get_microseconds();
> for (i = 0; i < nthread; i++)
> - create_thread(matmul_thread, (void *)(long)i);
> + create_thread(matmul_thread, (void *)i);
> while (atomic_read(&nstarted) != nthread)
> barrier();
> starttime = get_microseconds();
> @@ -95,7 +96,7 @@ int main(int argc, char *argv[])
> while (atomic_read(&ndone) != nthread)
> poll(NULL, 0, 1);
> endtime = get_microseconds();
> - printf("dim = %ld, nthread = %d, duration = %lld : %lld us\n",
> + printf("dim = %" PRIdPTR ", nthread = %d, duration = %lld : %lld us\n",
> dim, nthread, endtime - startcreatetime, endtime - starttime);
> return 0;
> }
> diff --git a/CodeSamples/SMPdesign/smpalloc.c b/CodeSamples/SMPdesign/smpalloc.c
> index 95b21a1..0882f63 100644
> --- a/CodeSamples/SMPdesign/smpalloc.c
> +++ b/CodeSamples/SMPdesign/smpalloc.c
> @@ -19,6 +19,7 @@
> * Copyright (c) 2006 Paul E. McKenney, IBM Corporation.
> */
>
> +#include <inttypes.h>
> #include "../api.h"
>
> #define TARGET_POOL_SIZE 3
> @@ -123,7 +124,7 @@ void *memblock_test(void *arg)
> long cnt = 0;
> long cntfail = 0;
> int i;
> - int runlength = (int)(long)arg;
> + intptr_t runlength = (intptr_t)arg;
> struct memblock *p[MAX_RUN];
>
> if (runlength > MAX_RUN)
> @@ -159,7 +160,7 @@ int main(int argc, char *argv[])
> long long nc;
> long long nf;
> int nkids = 1;
> - int runlength = 1;
> + intptr_t runlength = 1;
> int totbefore;
>
> smp_init();
> @@ -176,12 +177,12 @@ int main(int argc, char *argv[])
> if (argc > 2) {
> runlength = strtoul(argv[2], NULL, 0);
> if (runlength > MAX_RUN) {
> - fprintf(stderr, "nkids = %d too large, max = %d\n",
> + fprintf(stderr, "nkids = %" PRIdPTR " too large, max = %d\n",
> runlength, MAX_RUN);
> usage(argv[0]);
> }
> }
> - printf("%d %d ", nkids, runlength);
> + printf("%d %" PRIdPTR, nkids, runlength);
>
> init_per_thread(results, 0L);
> init_per_thread(failures, 0L);
> @@ -189,7 +190,7 @@ int main(int argc, char *argv[])
>
> goflag = 1;
> for (i = 0; i < nkids; i++)
> - create_thread(memblock_test, (void *)(long)runlength);
> + create_thread(memblock_test, (void *)runlength);
>
> sleep(1);
> goflag = 0;
> diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> index bc0b9ea..e794e82 100644
> --- a/CodeSamples/defer/gettimestampmp.c
> +++ b/CodeSamples/defer/gettimestampmp.c
> @@ -29,7 +29,7 @@ long curtimestamp = 0;
>
> void *collect_timestamps(void *mask_in)
> {
> - long mask = (long)mask_in;
> + intptr_t mask = (intptr_t)mask_in;
>
> while (curtimestamp < MAX_TIMESTAMPS) {
> while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> diff --git a/CodeSamples/intro/threadcreate.c b/CodeSamples/intro/threadcreate.c
> index 26b7ba9..470bb11 100644
> --- a/CodeSamples/intro/threadcreate.c
> +++ b/CodeSamples/intro/threadcreate.c
> @@ -18,13 +18,14 @@
> * Copyright (c) 2006 Paul E. McKenney, IBM Corporation.
> */
>
> +#include <inttypes.h>
> #include "../api.h"
>
> void *thread_test(void *arg)
> {
> - int myarg = (int)arg;
> + intptr_t myarg = (intptr_t)arg;
>
> - printf("child thread %d: smp_thread_id() = %d\n",
> + printf("child thread %" PRIdPTR ": smp_thread_id() = %d\n",
> myarg, smp_thread_id());
> return NULL;
> }
> @@ -38,7 +39,7 @@ void usage(char *progname)
>
> int main(int argc, char *argv[])
> {
> - int i;
> + intptr_t i;
> int nkids = 1;
>
> smp_init();
> --
> 2.7.4
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe perfbook" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
` (2 preceding siblings ...)
2017-05-29 22:17 ` [RFC PATCH 3/4] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
@ 2017-05-29 22:18 ` Akira Yokosawa
2017-05-30 0:12 ` Paul E. McKenney
2017-05-30 0:02 ` [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Paul E. McKenney
2017-05-30 12:05 ` [RFC PATCH v2 0/2] " Akira Yokosawa
5 siblings, 1 reply; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-29 22:18 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From fd5fb9afb6b99b15370a9823b42fa2a92342b92a Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Sun, 28 May 2017 15:42:45 +0900
Subject: [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
They ensure curtimestamp is read and written once in every
iteration.
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/defer/gettimestampmp.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
index e794e82..7012caa 100644
--- a/CodeSamples/defer/gettimestampmp.c
+++ b/CodeSamples/defer/gettimestampmp.c
@@ -32,7 +32,7 @@ void *collect_timestamps(void *mask_in)
intptr_t mask = (intptr_t)mask_in;
while (curtimestamp < MAX_TIMESTAMPS) {
- while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
+ while ((READ_ONCE(curtimestamp) & CURTIMESTAMP_MASK) != mask)
continue;
if (curtimestamp >= MAX_TIMESTAMPS)
break;
@@ -40,6 +40,7 @@ void *collect_timestamps(void *mask_in)
/* Don't need memory barrier -- no other shared vars!!! */
ts[curtimestamp++] = get_timestamp();
+ barrier();
}
smp_mb();
return (NULL);
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
2017-05-29 22:18 ` [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c Akira Yokosawa
@ 2017-05-30 0:12 ` Paul E. McKenney
0 siblings, 0 replies; 18+ messages in thread
From: Paul E. McKenney @ 2017-05-30 0:12 UTC (permalink / raw)
To: Akira Yokosawa; +Cc: perfbook
On Tue, May 30, 2017 at 07:18:51AM +0900, Akira Yokosawa wrote:
> >From fd5fb9afb6b99b15370a9823b42fa2a92342b92a Mon Sep 17 00:00:00 2001
> From: Akira Yokosawa <akiyks@gmail.com>
> Date: Sun, 28 May 2017 15:42:45 +0900
> Subject: [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
>
> They ensure curtimestamp is read and written once in every
> iteration.
>
> Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
I was actually more or less OK with this one, but it does not apply
cleanly without 3/4.
Nevertheless, given that it needs change anyway, would it be better to
replace the ++ with a WRITE_ONCE and capture the value in the "while"
expression?
Thanx, Paul
> ---
> CodeSamples/defer/gettimestampmp.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> index e794e82..7012caa 100644
> --- a/CodeSamples/defer/gettimestampmp.c
> +++ b/CodeSamples/defer/gettimestampmp.c
> @@ -32,7 +32,7 @@ void *collect_timestamps(void *mask_in)
> intptr_t mask = (intptr_t)mask_in;
>
> while (curtimestamp < MAX_TIMESTAMPS) {
> - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> + while ((READ_ONCE(curtimestamp) & CURTIMESTAMP_MASK) != mask)
> continue;
> if (curtimestamp >= MAX_TIMESTAMPS)
> break;
> @@ -40,6 +40,7 @@ void *collect_timestamps(void *mask_in)
> /* Don't need memory barrier -- no other shared vars!!! */
>
> ts[curtimestamp++] = get_timestamp();
> + barrier();
> }
> smp_mb();
> return (NULL);
> --
> 2.7.4
>
>
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [RFC PATCH 0/4] CodeSamples: Cleanups and fixes
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
` (3 preceding siblings ...)
2017-05-29 22:18 ` [RFC PATCH 4/4] CodeSamples/defer: Add compiler barriers in gettimestampmp.c Akira Yokosawa
@ 2017-05-30 0:02 ` Paul E. McKenney
2017-05-30 1:44 ` Akira Yokosawa
2017-05-30 12:05 ` [RFC PATCH v2 0/2] " Akira Yokosawa
5 siblings, 1 reply; 18+ messages in thread
From: Paul E. McKenney @ 2017-05-30 0:02 UTC (permalink / raw)
To: Akira Yokosawa; +Cc: perfbook
On Tue, May 30, 2017 at 07:13:23AM +0900, Akira Yokosawa wrote:
> >From fd5fb9afb6b99b15370a9823b42fa2a92342b92a Mon Sep 17 00:00:00 2001
> From: Akira Yokosawa <akiyks@gmail.com>
> Date: Tue, 30 May 2017 06:52:40 +0900
> Subject: [RFC PATCH 0/4] CodeSamples: Cleanups and fixes
>
> Hi Paul,
>
> While I was looking around CodeSamples, I encountered several
> warnings in build and an infinite loop in one of sample codes.
> Also, I found that CodeSamples/api.h and CodeSamples/Makefile.arch
> can be removed from repository if their recipe are properly
> written in Makefiles.
>
> Patch 1 adds rules to generate Makefile.arch and api.h that
> suit the architecture of you host. As I don't have ppc64 or
> arm host environment, I tested it only on x86_32 and x86_64.
> I'd like to know if it works on ppc64 and arm.
>
> Patch 2 removes the redundant files.
I took the first two, and it is nice to just be able to type "make"
in the CodeSamples directory! ;-)
> Patch 3 and 4 are fixes of warnings and infinite loops.
I have some concerns here, and will reply to the patches.
Thanx, Paul
> Thoughts?
>
> Thanks, Akira
> --
> Akira Yokosawa (4):
> CodeSamples: Add rule to generate Makefile.arch and api.h
> CodeSamples: Remove generated files from repository
> CodeSamples: Use 'intptr_t' to be compatible with 'void *'
> CodeSamples/defer: Add compiler barriers in gettimestampmp.c
>
> CodeSamples/.gitignore | 4 +-
> CodeSamples/Makefile | 18 +
> CodeSamples/Makefile.arch | 6 -
> CodeSamples/SMPdesign/Makefile | 3 +
> CodeSamples/SMPdesign/matmul.c | 13 +-
> CodeSamples/SMPdesign/smpalloc.c | 11 +-
> CodeSamples/advsync/Makefile | 3 +
> CodeSamples/api.h | 808 -------------------------
> CodeSamples/count/Makefile | 3 +
> CodeSamples/datastruct/Issaquah/Makefile | 3 +
> CodeSamples/datastruct/Issaquah/existence_test | Bin 26359 -> 0 bytes
> CodeSamples/datastruct/hash/Makefile | 3 +
> CodeSamples/datastruct/log/Makefile | 3 +
> CodeSamples/datastruct/skiplist/Makefile | 3 +
> CodeSamples/defer/Makefile | 3 +
> CodeSamples/defer/gettimestampmp.c | 5 +-
> CodeSamples/depends.mk | 43 ++
> CodeSamples/intro/Makefile | 3 +
> CodeSamples/intro/threadcreate.c | 7 +-
> CodeSamples/locking/Makefile | 3 +
> CodeSamples/toolsoftrade/Makefile | 3 +
> 21 files changed, 117 insertions(+), 831 deletions(-)
> delete mode 100644 CodeSamples/Makefile.arch
> delete mode 100644 CodeSamples/api.h
> delete mode 100755 CodeSamples/datastruct/Issaquah/existence_test
> create mode 100644 CodeSamples/depends.mk
>
> --
> 2.7.4
>
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH 0/4] CodeSamples: Cleanups and fixes
2017-05-30 0:02 ` [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Paul E. McKenney
@ 2017-05-30 1:44 ` Akira Yokosawa
0 siblings, 0 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-30 1:44 UTC (permalink / raw)
To: paulmck; +Cc: perfbook
2017/05/30 9:02、Paul E. McKenney <paulmck@linux.vnet.ibm.com> のメッセージ:
> On Tue, May 30, 2017 at 07:13:23AM +0900, Akira Yokosawa wrote:
>>> From fd5fb9afb6b99b15370a9823b42fa2a92342b92a Mon Sep 17 00:00:00 2001
>> From: Akira Yokosawa <akiyks@gmail.com>
>> Date: Tue, 30 May 2017 06:52:40 +0900
>> Subject: [RFC PATCH 0/4] CodeSamples: Cleanups and fixes
>>
>> Hi Paul,
>>
>> While I was looking around CodeSamples, I encountered several
>> warnings in build and an infinite loop in one of sample codes.
>> Also, I found that CodeSamples/api.h and CodeSamples/Makefile.arch
>> can be removed from repository if their recipe are properly
>> written in Makefiles.
>>
>> Patch 1 adds rules to generate Makefile.arch and api.h that
>> suit the architecture of you host. As I don't have ppc64 or
>> arm host environment, I tested it only on x86_32 and x86_64.
>> I'd like to know if it works on ppc64 and arm.
>>
>> Patch 2 removes the redundant files.
>
> I took the first two, and it is nice to just be able to type "make"
> in the CodeSamples directory! ;-)
Does this work on ppc64 and arm (whatever arm
platform you have access to)?
>
>> Patch 3 and 4 are fixes of warnings and infinite loops.
>
> I have some concerns here, and will reply to the patches.
>
I see your points. I'll respin.
Thanks, Akira
(from mobile, might be QP encoded)
> Thanx, Paul
>
>> Thoughts?
>>
>> Thanks, Akira
>> --
>> Akira Yokosawa (4):
>> CodeSamples: Add rule to generate Makefile.arch and api.h
>> CodeSamples: Remove generated files from repository
>> CodeSamples: Use 'intptr_t' to be compatible with 'void *'
>> CodeSamples/defer: Add compiler barriers in gettimestampmp.c
>>
>> CodeSamples/.gitignore | 4 +-
>> CodeSamples/Makefile | 18 +
>> CodeSamples/Makefile.arch | 6 -
>> CodeSamples/SMPdesign/Makefile | 3 +
>> CodeSamples/SMPdesign/matmul.c | 13 +-
>> CodeSamples/SMPdesign/smpalloc.c | 11 +-
>> CodeSamples/advsync/Makefile | 3 +
>> CodeSamples/api.h | 808 -------------------------
>> CodeSamples/count/Makefile | 3 +
>> CodeSamples/datastruct/Issaquah/Makefile | 3 +
>> CodeSamples/datastruct/Issaquah/existence_test | Bin 26359 -> 0 bytes
>> CodeSamples/datastruct/hash/Makefile | 3 +
>> CodeSamples/datastruct/log/Makefile | 3 +
>> CodeSamples/datastruct/skiplist/Makefile | 3 +
>> CodeSamples/defer/Makefile | 3 +
>> CodeSamples/defer/gettimestampmp.c | 5 +-
>> CodeSamples/depends.mk | 43 ++
>> CodeSamples/intro/Makefile | 3 +
>> CodeSamples/intro/threadcreate.c | 7 +-
>> CodeSamples/locking/Makefile | 3 +
>> CodeSamples/toolsoftrade/Makefile | 3 +
>> 21 files changed, 117 insertions(+), 831 deletions(-)
>> delete mode 100644 CodeSamples/Makefile.arch
>> delete mode 100644 CodeSamples/api.h
>> delete mode 100755 CodeSamples/datastruct/Issaquah/existence_test
>> create mode 100644 CodeSamples/depends.mk
>>
>> --
>> 2.7.4
>>
>
^ permalink raw reply [flat|nested] 18+ messages in thread
* [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-05-29 22:13 [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Akira Yokosawa
` (4 preceding siblings ...)
2017-05-30 0:02 ` [RFC PATCH 0/4] CodeSamples: Cleanups and fixes Paul E. McKenney
@ 2017-05-30 12:05 ` Akira Yokosawa
2017-05-30 12:06 ` [RFC PATCH v2 1/2] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
` (2 more replies)
5 siblings, 3 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-30 12:05 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Tue, 30 May 2017 20:44:52 +0900
Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
Hi Paul,
This is the respin of the latter two patches of v1. I'm keeping RFC
because of some questions.
"long" -> "intptr_t" changes in Patch 1 have no effect on a platform
where "long" and "intptr_t" have the same width, but I think they
are good in portability POV.
WRITE_ONCE() in Patch 2 is placed under the assignment to the array
because I could not translate post increment in any other way.
Does the WRITE_ONCE() ensure the outer "while" capture the value?
Thanks, Akira
--
Akira Yokosawa (2):
CodeSamples: Use 'intptr_t' to be compatible with 'void *'
CodeSamples/defer: Add compiler barriers in gettimestampmp.c
CodeSamples/SMPdesign/matmul.c | 4 ++--
CodeSamples/SMPdesign/smpalloc.c | 4 ++--
CodeSamples/defer/gettimestampmp.c | 7 ++++---
CodeSamples/intro/threadcreate.c | 4 ++--
4 files changed, 10 insertions(+), 9 deletions(-)
--
2.7.4
^ permalink raw reply [flat|nested] 18+ messages in thread* [RFC PATCH v2 1/2] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
2017-05-30 12:05 ` [RFC PATCH v2 0/2] " Akira Yokosawa
@ 2017-05-30 12:06 ` Akira Yokosawa
2017-05-30 12:07 ` [RFC PATCH v2 2/2] CodeSamples/defer: Add compiler barriers in gettimestampmp.c Akira Yokosawa
2017-05-31 18:46 ` [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes Paul E. McKenney
2 siblings, 0 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-30 12:06 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From 177ee2891184f0a97b66d748763dbc7af75033ff Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Tue, 30 May 2017 20:21:41 +0900
Subject: [RFC PATCH v2 1/2] CodeSamples: Use 'intptr_t' to be compatible with 'void *'
On x86_64, casting between "int" and "void *" in threadcreate.c
causes GCC to emit warnings such as:
warning: cast from pointer to integer of different size
warning: cast from integer to pointer of different size
Let's adopt C99 way of writing portable code.
Also do the same changes where appropriate. (Other than
threadcreate.c, no warnings were observed on x86_64.)
[Revised according to Paul's suggestion to keep intptr_t usage
confined.]
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/SMPdesign/matmul.c | 4 ++--
CodeSamples/SMPdesign/smpalloc.c | 4 ++--
CodeSamples/defer/gettimestampmp.c | 2 +-
CodeSamples/intro/threadcreate.c | 4 ++--
4 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/CodeSamples/SMPdesign/matmul.c b/CodeSamples/SMPdesign/matmul.c
index 6a07730..3f70807 100644
--- a/CodeSamples/SMPdesign/matmul.c
+++ b/CodeSamples/SMPdesign/matmul.c
@@ -38,7 +38,7 @@ atomic_t nstarted;
void *matmul_thread(void *me_in)
{
- long me = (long)me_in;
+ long me = (intptr_t)me_in;
int i, j, k;
atomic_inc(&nstarted);
@@ -87,7 +87,7 @@ int main(int argc, char *argv[])
goflag = GOFLAG_INIT;
startcreatetime = get_microseconds();
for (i = 0; i < nthread; i++)
- create_thread(matmul_thread, (void *)(long)i);
+ create_thread(matmul_thread, (void *)(intptr_t)i);
while (atomic_read(&nstarted) != nthread)
barrier();
starttime = get_microseconds();
diff --git a/CodeSamples/SMPdesign/smpalloc.c b/CodeSamples/SMPdesign/smpalloc.c
index 95b21a1..454846b 100644
--- a/CodeSamples/SMPdesign/smpalloc.c
+++ b/CodeSamples/SMPdesign/smpalloc.c
@@ -123,7 +123,7 @@ void *memblock_test(void *arg)
long cnt = 0;
long cntfail = 0;
int i;
- int runlength = (int)(long)arg;
+ int runlength = (intptr_t)arg;
struct memblock *p[MAX_RUN];
if (runlength > MAX_RUN)
@@ -189,7 +189,7 @@ int main(int argc, char *argv[])
goflag = 1;
for (i = 0; i < nkids; i++)
- create_thread(memblock_test, (void *)(long)runlength);
+ create_thread(memblock_test, (void *)(intptr_t)runlength);
sleep(1);
goflag = 0;
diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
index bc0b9ea..2abade4 100644
--- a/CodeSamples/defer/gettimestampmp.c
+++ b/CodeSamples/defer/gettimestampmp.c
@@ -29,7 +29,7 @@ long curtimestamp = 0;
void *collect_timestamps(void *mask_in)
{
- long mask = (long)mask_in;
+ long mask = (intptr_t)mask_in;
while (curtimestamp < MAX_TIMESTAMPS) {
while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
diff --git a/CodeSamples/intro/threadcreate.c b/CodeSamples/intro/threadcreate.c
index 26b7ba9..94cde89 100644
--- a/CodeSamples/intro/threadcreate.c
+++ b/CodeSamples/intro/threadcreate.c
@@ -22,7 +22,7 @@
void *thread_test(void *arg)
{
- int myarg = (int)arg;
+ int myarg = (intptr_t)arg;
printf("child thread %d: smp_thread_id() = %d\n",
myarg, smp_thread_id());
@@ -54,7 +54,7 @@ int main(int argc, char *argv[])
printf("Parent thread spawning %d threads.\n", nkids);
for (i = 0; i < nkids; i++)
- create_thread(thread_test, (void *)i);
+ create_thread(thread_test, (void *)(intptr_t)i);
wait_all_threads();
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* [RFC PATCH v2 2/2] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
2017-05-30 12:05 ` [RFC PATCH v2 0/2] " Akira Yokosawa
2017-05-30 12:06 ` [RFC PATCH v2 1/2] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
@ 2017-05-30 12:07 ` Akira Yokosawa
2017-05-31 18:46 ` [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes Paul E. McKenney
2 siblings, 0 replies; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-30 12:07 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
From: Akira Yokosawa <akiyks@gmail.com>
Date: Tue, 30 May 2017 20:40:04 +0900
Subject: [RFC PATCH v2 2/2] CodeSamples/defer: Add compiler barriers in gettimestampmp.c
They ensure curtimestamp is read and written once in every
iteration.
[Revised according to Paul's suggestion to use WRITE_ONCE()]
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
---
CodeSamples/defer/gettimestampmp.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
index 2abade4..fc5718d 100644
--- a/CodeSamples/defer/gettimestampmp.c
+++ b/CodeSamples/defer/gettimestampmp.c
@@ -32,14 +32,15 @@ void *collect_timestamps(void *mask_in)
long mask = (intptr_t)mask_in;
while (curtimestamp < MAX_TIMESTAMPS) {
- while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
+ while ((READ_ONCE(curtimestamp) & CURTIMESTAMP_MASK) != mask)
continue;
if (curtimestamp >= MAX_TIMESTAMPS)
break;
/* Don't need memory barrier -- no other shared vars!!! */
- ts[curtimestamp++] = get_timestamp();
+ ts[curtimestamp] = get_timestamp();
+ WRITE_ONCE(curtimestamp, curtimestamp + 1);
}
smp_mb();
return (NULL);
--
2.7.4
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-05-30 12:05 ` [RFC PATCH v2 0/2] " Akira Yokosawa
2017-05-30 12:06 ` [RFC PATCH v2 1/2] CodeSamples: Use 'intptr_t' to be compatible with 'void *' Akira Yokosawa
2017-05-30 12:07 ` [RFC PATCH v2 2/2] CodeSamples/defer: Add compiler barriers in gettimestampmp.c Akira Yokosawa
@ 2017-05-31 18:46 ` Paul E. McKenney
2017-05-31 21:19 ` Akira Yokosawa
2017-06-01 1:45 ` Junchang Wang
2 siblings, 2 replies; 18+ messages in thread
From: Paul E. McKenney @ 2017-05-31 18:46 UTC (permalink / raw)
To: Akira Yokosawa; +Cc: perfbook
On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
> From: Akira Yokosawa <akiyks@gmail.com>
> Date: Tue, 30 May 2017 20:44:52 +0900
> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
>
> Hi Paul,
>
> This is the respin of the latter two patches of v1. I'm keeping RFC
> because of some questions.
>
> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
> where "long" and "intptr_t" have the same width, but I think they
> are good in portability POV.
>
> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
> because I could not translate post increment in any other way.
> Does the WRITE_ONCE() ensure the outer "while" capture the value?
Wow, that loop is old code!!! My current compiler creates an infinite
loop for it, so yes, there is more required. Plus there are confusing
and redundant comparisons, so that it is not entirely clear to me that
the loop is guaranteed to terminate properly.
So I took both patches, but rewrote the loop in the second patch as
shown below.
If you are OK with this rewrite, I will push them.
Thanx, Paul
------------------------------------------------------------------------
commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
Author: Akira Yokosawa <akiyks@gmail.com>
Date: Tue May 30 20:40:04 2017 +0900
CodeSamples/defer: Rework loop in gettimestampmp.c
Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
once in every iteration. The READ_ONCE() is not optional, as modern
compilers can (and do) emit an infinite loop for the earlier code.
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
[ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
index 2abade42e233..8780b71f33d7 100644
--- a/CodeSamples/defer/gettimestampmp.c
+++ b/CodeSamples/defer/gettimestampmp.c
@@ -30,16 +30,19 @@ long curtimestamp = 0;
void *collect_timestamps(void *mask_in)
{
long mask = (intptr_t)mask_in;
+ long cts;
- while (curtimestamp < MAX_TIMESTAMPS) {
- while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
- continue;
- if (curtimestamp >= MAX_TIMESTAMPS)
+ for (;;) {
+ cts = READ_ONCE(curtimestamp);
+ if (cts >= MAX_TIMESTAMPS)
break;
+ if ((cts & CURTIMESTAMP_MASK) != mask)
+ continue;
/* Don't need memory barrier -- no other shared vars!!! */
- ts[curtimestamp++] = get_timestamp();
+ ts[cts] = get_timestamp();
+ WRITE_ONCE(curtimestamp, cts + 1);
}
smp_mb();
return (NULL);
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-05-31 18:46 ` [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes Paul E. McKenney
@ 2017-05-31 21:19 ` Akira Yokosawa
2017-06-01 0:05 ` Paul E. McKenney
2017-06-01 1:45 ` Junchang Wang
1 sibling, 1 reply; 18+ messages in thread
From: Akira Yokosawa @ 2017-05-31 21:19 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: perfbook, Akira Yokosawa
On 2017/05/31 11:46:55 -0700, Paul E. McKenney wrote:
> On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
>> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
>> From: Akira Yokosawa <akiyks@gmail.com>
>> Date: Tue, 30 May 2017 20:44:52 +0900
>> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
>>
>> Hi Paul,
>>
>> This is the respin of the latter two patches of v1. I'm keeping RFC
>> because of some questions.
>>
>> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
>> where "long" and "intptr_t" have the same width, but I think they
>> are good in portability POV.
>>
>> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
>> because I could not translate post increment in any other way.
>> Does the WRITE_ONCE() ensure the outer "while" capture the value?
>
> Wow, that loop is old code!!! My current compiler creates an infinite
> loop for it, so yes, there is more required.
You mean on GCC for ppc64?
> Plus there are confusing
> and redundant comparisons, so that it is not entirely clear to me that
> the loop is guaranteed to terminate properly
>
> So I took both patches, but rewrote the loop in the second patch as
> shown below.
>
> If you are OK with this rewrite, I will push them.
I'm OK with this, but it is a whole rewrite of the code, so
Reported-by: Akira Yokosawa <akiyks@gmail.com>
looks appropriate in this case.
It's up to you which tag to use.
Thanks, Akira
>
> Thanx, Paul
>
> ------------------------------------------------------------------------
>
> commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
> Author: Akira Yokosawa <akiyks@gmail.com>
> Date: Tue May 30 20:40:04 2017 +0900
>
> CodeSamples/defer: Rework loop in gettimestampmp.c
>
> Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
> once in every iteration. The READ_ONCE() is not optional, as modern
> compilers can (and do) emit an infinite loop for the earlier code.
>
> Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
> [ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
>
> diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> index 2abade42e233..8780b71f33d7 100644
> --- a/CodeSamples/defer/gettimestampmp.c
> +++ b/CodeSamples/defer/gettimestampmp.c
> @@ -30,16 +30,19 @@ long curtimestamp = 0;
> void *collect_timestamps(void *mask_in)
> {
> long mask = (intptr_t)mask_in;
> + long cts;
>
> - while (curtimestamp < MAX_TIMESTAMPS) {
> - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> - continue;
> - if (curtimestamp >= MAX_TIMESTAMPS)
> + for (;;) {
> + cts = READ_ONCE(curtimestamp);
> + if (cts >= MAX_TIMESTAMPS)
> break;
> + if ((cts & CURTIMESTAMP_MASK) != mask)
> + continue;
>
> /* Don't need memory barrier -- no other shared vars!!! */
>
> - ts[curtimestamp++] = get_timestamp();
> + ts[cts] = get_timestamp();
> + WRITE_ONCE(curtimestamp, cts + 1);
> }
> smp_mb();
> return (NULL);
>
>
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-05-31 21:19 ` Akira Yokosawa
@ 2017-06-01 0:05 ` Paul E. McKenney
0 siblings, 0 replies; 18+ messages in thread
From: Paul E. McKenney @ 2017-06-01 0:05 UTC (permalink / raw)
To: Akira Yokosawa; +Cc: perfbook
On Thu, Jun 01, 2017 at 06:19:16AM +0900, Akira Yokosawa wrote:
> On 2017/05/31 11:46:55 -0700, Paul E. McKenney wrote:
> > On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
> >> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
> >> From: Akira Yokosawa <akiyks@gmail.com>
> >> Date: Tue, 30 May 2017 20:44:52 +0900
> >> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
> >>
> >> Hi Paul,
> >>
> >> This is the respin of the latter two patches of v1. I'm keeping RFC
> >> because of some questions.
> >>
> >> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
> >> where "long" and "intptr_t" have the same width, but I think they
> >> are good in portability POV.
> >>
> >> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
> >> because I could not translate post increment in any other way.
> >> Does the WRITE_ONCE() ensure the outer "while" capture the value?
> >
> > Wow, that loop is old code!!! My current compiler creates an infinite
> > loop for it, so yes, there is more required.
>
> You mean on GCC for ppc64?
No, x86 GCC on the code before you changed it. ;-)
Which reminds me... How about if I get you direct access to some ppc64
machines? If you are at all interested, please let me know and also
please sign up here:
http://osuosl.org/services/powerdev/request_hosting/
I believe that there is also similar access to ARM systems, so please
let me know if you are interested.
> > Plus there are confusing
> > and redundant comparisons, so that it is not entirely clear to me that
> > the loop is guaranteed to terminate properly
> >
> > So I took both patches, but rewrote the loop in the second patch as
> > shown below.
> >
> > If you are OK with this rewrite, I will push them.
>
> I'm OK with this, but it is a whole rewrite of the code, so
>
> Reported-by: Akira Yokosawa <akiyks@gmail.com>
>
> looks appropriate in this case.
> It's up to you which tag to use.
I will stick with what I have. Lazy and all that. ;-)
Thanx, Paul
> Thanks, Akira
>
> >
> > Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
> > Author: Akira Yokosawa <akiyks@gmail.com>
> > Date: Tue May 30 20:40:04 2017 +0900
> >
> > CodeSamples/defer: Rework loop in gettimestampmp.c
> >
> > Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
> > once in every iteration. The READ_ONCE() is not optional, as modern
> > compilers can (and do) emit an infinite loop for the earlier code.
> >
> > Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
> > [ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> >
> > diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> > index 2abade42e233..8780b71f33d7 100644
> > --- a/CodeSamples/defer/gettimestampmp.c
> > +++ b/CodeSamples/defer/gettimestampmp.c
> > @@ -30,16 +30,19 @@ long curtimestamp = 0;
> > void *collect_timestamps(void *mask_in)
> > {
> > long mask = (intptr_t)mask_in;
> > + long cts;
> >
> > - while (curtimestamp < MAX_TIMESTAMPS) {
> > - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> > - continue;
> > - if (curtimestamp >= MAX_TIMESTAMPS)
> > + for (;;) {
> > + cts = READ_ONCE(curtimestamp);
> > + if (cts >= MAX_TIMESTAMPS)
> > break;
> > + if ((cts & CURTIMESTAMP_MASK) != mask)
> > + continue;
> >
> > /* Don't need memory barrier -- no other shared vars!!! */
> >
> > - ts[curtimestamp++] = get_timestamp();
> > + ts[cts] = get_timestamp();
> > + WRITE_ONCE(curtimestamp, cts + 1);
> > }
> > smp_mb();
> > return (NULL);
> >
> >
>
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-05-31 18:46 ` [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes Paul E. McKenney
2017-05-31 21:19 ` Akira Yokosawa
@ 2017-06-01 1:45 ` Junchang Wang
2017-06-01 4:19 ` Paul E. McKenney
1 sibling, 1 reply; 18+ messages in thread
From: Junchang Wang @ 2017-06-01 1:45 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: Akira Yokosawa, perfbook
Hi Paul and Akira,
The code looks clear and easy to understand now :-).
While checking the patch, I have one question. Are there any technical
reason that we prefer for(;;) instead of while(1) in CodeSamples? Just
out of curiosity :-)
Thanks
On Thu, Jun 1, 2017 at 2:46 AM, Paul E. McKenney
<paulmck@linux.vnet.ibm.com> wrote:
> On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
>> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
>> From: Akira Yokosawa <akiyks@gmail.com>
>> Date: Tue, 30 May 2017 20:44:52 +0900
>> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
>>
>> Hi Paul,
>>
>> This is the respin of the latter two patches of v1. I'm keeping RFC
>> because of some questions.
>>
>> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
>> where "long" and "intptr_t" have the same width, but I think they
>> are good in portability POV.
>>
>> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
>> because I could not translate post increment in any other way.
>> Does the WRITE_ONCE() ensure the outer "while" capture the value?
>
> Wow, that loop is old code!!! My current compiler creates an infinite
> loop for it, so yes, there is more required. Plus there are confusing
> and redundant comparisons, so that it is not entirely clear to me that
> the loop is guaranteed to terminate properly.
>
> So I took both patches, but rewrote the loop in the second patch as
> shown below.
>
> If you are OK with this rewrite, I will push them.
>
> Thanx, Paul
>
> ------------------------------------------------------------------------
>
> commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
> Author: Akira Yokosawa <akiyks@gmail.com>
> Date: Tue May 30 20:40:04 2017 +0900
>
> CodeSamples/defer: Rework loop in gettimestampmp.c
>
> Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
> once in every iteration. The READ_ONCE() is not optional, as modern
> compilers can (and do) emit an infinite loop for the earlier code.
>
> Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
> [ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
>
> diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> index 2abade42e233..8780b71f33d7 100644
> --- a/CodeSamples/defer/gettimestampmp.c
> +++ b/CodeSamples/defer/gettimestampmp.c
> @@ -30,16 +30,19 @@ long curtimestamp = 0;
> void *collect_timestamps(void *mask_in)
> {
> long mask = (intptr_t)mask_in;
> + long cts;
>
> - while (curtimestamp < MAX_TIMESTAMPS) {
> - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> - continue;
> - if (curtimestamp >= MAX_TIMESTAMPS)
> + for (;;) {
> + cts = READ_ONCE(curtimestamp);
> + if (cts >= MAX_TIMESTAMPS)
> break;
> + if ((cts & CURTIMESTAMP_MASK) != mask)
> + continue;
>
> /* Don't need memory barrier -- no other shared vars!!! */
>
> - ts[curtimestamp++] = get_timestamp();
> + ts[cts] = get_timestamp();
> + WRITE_ONCE(curtimestamp, cts + 1);
> }
> smp_mb();
> return (NULL);
>
> --
> To unsubscribe from this list: send the line "unsubscribe perfbook" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-06-01 1:45 ` Junchang Wang
@ 2017-06-01 4:19 ` Paul E. McKenney
2017-06-01 4:34 ` Junchang Wang
0 siblings, 1 reply; 18+ messages in thread
From: Paul E. McKenney @ 2017-06-01 4:19 UTC (permalink / raw)
To: Junchang Wang; +Cc: Akira Yokosawa, perfbook
On Thu, Jun 01, 2017 at 09:45:02AM +0800, Junchang Wang wrote:
> Hi Paul and Akira,
>
> The code looks clear and easy to understand now :-).
Glad you like it! I am sure that other code in there could use
similar help.
> While checking the patch, I have one question. Are there any technical
> reason that we prefer for(;;) instead of while(1) in CodeSamples? Just
> out of curiosity :-)
No. Just thirty-five years of habit plus the preferences of the Linux
kernel community. And the latter are probably because "while (1)"
has one more character than does "for (;;)". ;-)
Thanx, Paul
> Thanks
>
> On Thu, Jun 1, 2017 at 2:46 AM, Paul E. McKenney
> <paulmck@linux.vnet.ibm.com> wrote:
> > On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
> >> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
> >> From: Akira Yokosawa <akiyks@gmail.com>
> >> Date: Tue, 30 May 2017 20:44:52 +0900
> >> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
> >>
> >> Hi Paul,
> >>
> >> This is the respin of the latter two patches of v1. I'm keeping RFC
> >> because of some questions.
> >>
> >> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
> >> where "long" and "intptr_t" have the same width, but I think they
> >> are good in portability POV.
> >>
> >> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
> >> because I could not translate post increment in any other way.
> >> Does the WRITE_ONCE() ensure the outer "while" capture the value?
> >
> > Wow, that loop is old code!!! My current compiler creates an infinite
> > loop for it, so yes, there is more required. Plus there are confusing
> > and redundant comparisons, so that it is not entirely clear to me that
> > the loop is guaranteed to terminate properly.
> >
> > So I took both patches, but rewrote the loop in the second patch as
> > shown below.
> >
> > If you are OK with this rewrite, I will push them.
> >
> > Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
> > Author: Akira Yokosawa <akiyks@gmail.com>
> > Date: Tue May 30 20:40:04 2017 +0900
> >
> > CodeSamples/defer: Rework loop in gettimestampmp.c
> >
> > Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
> > once in every iteration. The READ_ONCE() is not optional, as modern
> > compilers can (and do) emit an infinite loop for the earlier code.
> >
> > Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
> > [ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> >
> > diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
> > index 2abade42e233..8780b71f33d7 100644
> > --- a/CodeSamples/defer/gettimestampmp.c
> > +++ b/CodeSamples/defer/gettimestampmp.c
> > @@ -30,16 +30,19 @@ long curtimestamp = 0;
> > void *collect_timestamps(void *mask_in)
> > {
> > long mask = (intptr_t)mask_in;
> > + long cts;
> >
> > - while (curtimestamp < MAX_TIMESTAMPS) {
> > - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
> > - continue;
> > - if (curtimestamp >= MAX_TIMESTAMPS)
> > + for (;;) {
> > + cts = READ_ONCE(curtimestamp);
> > + if (cts >= MAX_TIMESTAMPS)
> > break;
> > + if ((cts & CURTIMESTAMP_MASK) != mask)
> > + continue;
> >
> > /* Don't need memory barrier -- no other shared vars!!! */
> >
> > - ts[curtimestamp++] = get_timestamp();
> > + ts[cts] = get_timestamp();
> > + WRITE_ONCE(curtimestamp, cts + 1);
> > }
> > smp_mb();
> > return (NULL);
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe perfbook" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
>
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
2017-06-01 4:19 ` Paul E. McKenney
@ 2017-06-01 4:34 ` Junchang Wang
0 siblings, 0 replies; 18+ messages in thread
From: Junchang Wang @ 2017-06-01 4:34 UTC (permalink / raw)
To: Paul E. McKenney; +Cc: Akira Yokosawa, perfbook
On Thu, Jun 1, 2017 at 12:19 PM, Paul E. McKenney
<paulmck@linux.vnet.ibm.com> wrote:
> On Thu, Jun 01, 2017 at 09:45:02AM +0800, Junchang Wang wrote:
>> Hi Paul and Akira,
>>
>> The code looks clear and easy to understand now :-).
>
> Glad you like it! I am sure that other code in there could use
> similar help.
I would be very happy to gradually go through these sections this summer. :-)
>
>> While checking the patch, I have one question. Are there any technical
>> reason that we prefer for(;;) instead of while(1) in CodeSamples? Just
>> out of curiosity :-)
>
> No. Just thirty-five years of habit plus the preferences of the Linux
> kernel community. And the latter are probably because "while (1)"
> has one more character than does "for (;;)". ;-)
>
Got you :-) Thanks.
--Jason
> Thanx, Paul
>
>> Thanks
>>
>> On Thu, Jun 1, 2017 at 2:46 AM, Paul E. McKenney
>> <paulmck@linux.vnet.ibm.com> wrote:
>> > On Tue, May 30, 2017 at 09:05:15PM +0900, Akira Yokosawa wrote:
>> >> >From 489b5e3bdeba2f9b733dbe3d85390368dd159174 Mon Sep 17 00:00:00 2001
>> >> From: Akira Yokosawa <akiyks@gmail.com>
>> >> Date: Tue, 30 May 2017 20:44:52 +0900
>> >> Subject: [RFC PATCH v2 0/2] CodeSamples: Cleanups and fixes
>> >>
>> >> Hi Paul,
>> >>
>> >> This is the respin of the latter two patches of v1. I'm keeping RFC
>> >> because of some questions.
>> >>
>> >> "long" -> "intptr_t" changes in Patch 1 have no effect on a platform
>> >> where "long" and "intptr_t" have the same width, but I think they
>> >> are good in portability POV.
>> >>
>> >> WRITE_ONCE() in Patch 2 is placed under the assignment to the array
>> >> because I could not translate post increment in any other way.
>> >> Does the WRITE_ONCE() ensure the outer "while" capture the value?
>> >
>> > Wow, that loop is old code!!! My current compiler creates an infinite
>> > loop for it, so yes, there is more required. Plus there are confusing
>> > and redundant comparisons, so that it is not entirely clear to me that
>> > the loop is guaranteed to terminate properly.
>> >
>> > So I took both patches, but rewrote the loop in the second patch as
>> > shown below.
>> >
>> > If you are OK with this rewrite, I will push them.
>> >
>> > Thanx, Paul
>> >
>> > ------------------------------------------------------------------------
>> >
>> > commit 8a54d9aeeeefa1909db062dc893705ff8fefd702
>> > Author: Akira Yokosawa <akiyks@gmail.com>
>> > Date: Tue May 30 20:40:04 2017 +0900
>> >
>> > CodeSamples/defer: Rework loop in gettimestampmp.c
>> >
>> > Add READ_ONCE() and WRITE_ONCE() ensure curtimestamp is read and written
>> > once in every iteration. The READ_ONCE() is not optional, as modern
>> > compilers can (and do) emit an infinite loop for the earlier code.
>> >
>> > Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
>> > [ paulmck: Rework loop to eliminate redundant fetches and comparisons. ]
>> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
>> >
>> > diff --git a/CodeSamples/defer/gettimestampmp.c b/CodeSamples/defer/gettimestampmp.c
>> > index 2abade42e233..8780b71f33d7 100644
>> > --- a/CodeSamples/defer/gettimestampmp.c
>> > +++ b/CodeSamples/defer/gettimestampmp.c
>> > @@ -30,16 +30,19 @@ long curtimestamp = 0;
>> > void *collect_timestamps(void *mask_in)
>> > {
>> > long mask = (intptr_t)mask_in;
>> > + long cts;
>> >
>> > - while (curtimestamp < MAX_TIMESTAMPS) {
>> > - while ((curtimestamp & CURTIMESTAMP_MASK) != mask)
>> > - continue;
>> > - if (curtimestamp >= MAX_TIMESTAMPS)
>> > + for (;;) {
>> > + cts = READ_ONCE(curtimestamp);
>> > + if (cts >= MAX_TIMESTAMPS)
>> > break;
>> > + if ((cts & CURTIMESTAMP_MASK) != mask)
>> > + continue;
>> >
>> > /* Don't need memory barrier -- no other shared vars!!! */
>> >
>> > - ts[curtimestamp++] = get_timestamp();
>> > + ts[cts] = get_timestamp();
>> > + WRITE_ONCE(curtimestamp, cts + 1);
>> > }
>> > smp_mb();
>> > return (NULL);
>> >
>> > --
>> > To unsubscribe from this list: send the line "unsubscribe perfbook" in
>> > the body of a message to majordomo@vger.kernel.org
>> > More majordomo info at http://vger.kernel.org/majordomo-info.html
>>
>
^ permalink raw reply [flat|nested] 18+ messages in thread