MINOR: threads: add more consistency between certain variables in no-thread case
When threads are disabled, some variables such as tid and tid_bit are
still checked everywhere, the MAX_THREADS_MASK macro is ~0UL while
MAX_THREADS is 1, and the all_threads_mask variable is replaced with a
macro forced to zero. The compiler cannot optimize away all this code
involving checks on tid and tid_bit, and we end up in special cases
where all_threads_mask has to be specifically tested for being zero or
not. It is not even certain the code paths are always equivalent when
testing without threads and with nbthread 1.
Let's change this to make sure we always present a single thread when
threads are disabled, and have the relevant values declared as constants
so that the compiler can optimize all the tests away. Now we have
MAX_THREADS_MASK set to 1, all_threads_mask set to 1, tid set to zero
and tid_bit set to 1. Doing just this has removed 4 kB of code in the
no-thread case.
A few checks for all_threads_mask==0 have been removed since it never
happens anymore.
diff --git a/include/common/hathreads.h b/include/common/hathreads.h
index 068b2a1..4cf3db9 100644
--- a/include/common/hathreads.h
+++ b/include/common/hathreads.h
@@ -24,10 +24,6 @@
#include <common/config.h>
-#define MAX_THREADS_MASK ((unsigned long)-1)
-extern THREAD_LOCAL unsigned int tid; /* The thread id */
-extern THREAD_LOCAL unsigned long tid_bit; /* The bit corresponding to the thread id */
-
/* Note about all_threads_mask :
* - with threads support disabled, this symbol is defined as zero (0UL).
* - with threads enabled, this variable is never zero, it contains the mask
@@ -37,7 +33,14 @@
#ifndef USE_THREAD
#define MAX_THREADS 1
-#define all_threads_mask 0UL
+#define MAX_THREADS_MASK 1
+
+/* Only way found to replace variables with constants that are optimized away
+ * at build time.
+ */
+enum { all_threads_mask = 1UL };
+enum { tid_bit = 1UL };
+enum { tid = 0 };
#define __decl_hathreads(decl)
@@ -116,6 +119,9 @@
#define ha_sigmask(how, set, oldset) sigprocmask(how, set, oldset)
+static inline void ha_set_tid(unsigned int tid)
+{
+}
static inline void __ha_barrier_load(void)
{
@@ -138,6 +144,7 @@
#include <import/plock.h>
#define MAX_THREADS LONGBITS
+#define MAX_THREADS_MASK ((unsigned long)-1)
#define __decl_hathreads(decl) decl
@@ -266,10 +273,19 @@
int thread_no_sync(void);
int thread_need_sync(void);
+extern THREAD_LOCAL unsigned int tid; /* The thread id */
+extern THREAD_LOCAL unsigned long tid_bit; /* The bit corresponding to the thread id */
extern volatile unsigned long all_threads_mask;
#define ha_sigmask(how, set, oldset) pthread_sigmask(how, set, oldset)
+/* sets the thread ID and the TID bit for the current thread */
+static inline void ha_set_tid(unsigned int data)
+{
+ tid = data;
+ tid_bit = (1UL << tid);
+}
+
#if defined(DEBUG_THREAD) || defined(DEBUG_FULL)
diff --git a/src/cfgparse.c b/src/cfgparse.c
index 5abe9fa..1451243 100644
--- a/src/cfgparse.c
+++ b/src/cfgparse.c
@@ -7587,11 +7587,11 @@
nbproc = my_ffsl(bind_conf->bind_proc);
mask = bind_conf->bind_thread[nbproc - 1];
- if (mask && !(mask & (all_threads_mask ? all_threads_mask : 1UL))) {
+ if (mask && !(mask & all_threads_mask)) {
unsigned long new_mask = 0;
while (mask) {
- new_mask |= mask & (all_threads_mask ? all_threads_mask : 1UL);
+ new_mask |= mask & all_threads_mask;
mask >>= global.nbthread;
}
diff --git a/src/haproxy.c b/src/haproxy.c
index 810d598..42f1f3e 100644
--- a/src/haproxy.c
+++ b/src/haproxy.c
@@ -2459,8 +2459,7 @@
struct per_thread_deinit_fct *ptdf;
__decl_hathreads(static HA_SPINLOCK_T start_lock);
- tid = *((unsigned int *)data);
- tid_bit = (1UL << tid);
+ ha_set_tid(*((unsigned int *)data));
tv_update_date(-1,-1);
list_for_each_entry(ptif, &per_thread_init_list, list) {
diff --git a/src/hathreads.c b/src/hathreads.c
index bfd055a..595a717 100644
--- a/src/hathreads.c
+++ b/src/hathreads.c
@@ -19,8 +19,6 @@
#include <common/standard.h>
#include <proto/fd.h>
-THREAD_LOCAL unsigned int tid = 0;
-THREAD_LOCAL unsigned long tid_bit = (1UL << 0);
/* Dummy I/O handler used by the sync pipe.*/
void thread_sync_io_handler(int fd)
@@ -33,6 +31,9 @@
static int threads_sync_pipe[2];
static unsigned long threads_want_sync = 0;
volatile unsigned long all_threads_mask = 1; // nbthread 1 assumed by default
+THREAD_LOCAL unsigned int tid = 0;
+THREAD_LOCAL unsigned long tid_bit = (1UL << 0);
+
#if defined(DEBUG_THREAD) || defined(DEBUG_FULL)
struct lock_stat lock_stats[LOCK_LABELS];
@@ -127,7 +128,7 @@
{
static volatile unsigned long barrier = 0;
- if (!all_threads_mask)
+ if (!(all_threads_mask & (all_threads_mask - 1)))
return;
thread_sync_barrier(&barrier);
@@ -143,7 +144,7 @@
{
static volatile unsigned long barrier = 0;
- if (!all_threads_mask)
+ if (!(all_threads_mask & (all_threads_mask - 1)))
return;
if (threads_want_sync & tid_bit)