[PATCH] s390: spin lock retry

Split spin lock and r/w lock implementation into a single try which is done
inline and an out of line function that repeatedly tries to get the lock
before doing the cpu_relax().  Add a system control to set the number of
retries before a cpu is yielded.

The reason for the spin lock retry is that the diagnose 0x44 that is used to
give up the virtual cpu is quite expensive.  For spin locks that are held only
for a short period of time the costs of the diagnoses outweights the savings
for spin locks that are held for a longer timer.  The default retry count is
1000.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

4 files changed, 135 insertions, 11 deletions

diff --git a/arch/s390/kernel/head64.S b/arch/s390/kernel/head64.S
index f525c0c21250..28c50bdf7d40 100644
--- a/arch/s390/kernel/head64.S
+++ b/arch/s390/kernel/head64.S
@@ -658,10 +658,8 @@ startup:basr  %r13,0                     # get base
 #
 	la     %r1,0f-.LPG1(%r13)	# set program check address
 	stg    %r1,__LC_PGM_NEW_PSW+8
-	mvc    __LC_DIAG44_OPCODE(8),.Lnop-.LPG1(%r13)
 	diag   0,0,0x44			# test diag 0x44
 	oi     7(%r12),32		# set diag44 flag
-	mvc    __LC_DIAG44_OPCODE(8),.Ldiag44-.LPG1(%r13)
 0:	
 
 #
@@ -702,7 +700,6 @@ startup:basr  %r13,0                     # get base
 .L4malign:.quad 0xffffffffffc00000
 .Lscan2g:.quad 0x80000000 + 0x20000 - 8 # 2GB + 128K - 8
 .Lnop:	.long  0x07000700
-.Ldiag44:.long 0x83000044
 
 	.org PARMAREA-64
 .Lduct:	.long 0,0,0,0,0,0,0,0
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
index a12183989a79..5ba5a5485da9 100644
--- a/arch/s390/kernel/setup.c
+++ b/arch/s390/kernel/setup.c
@@ -431,12 +431,6 @@ setup_lowcore(void)
 		ctl_set_bit(14, 29);
 	}
 #endif
-#ifdef CONFIG_ARCH_S390X
-	if (MACHINE_HAS_DIAG44)
-		lc->diag44_opcode = 0x83000044;
-	else
-		lc->diag44_opcode = 0x07000700;
-#endif /* CONFIG_ARCH_S390X */
 	set_prefix((u32)(unsigned long) lc);
 }
 
diff --git a/arch/s390/lib/Makefile b/arch/s390/lib/Makefile
index a8758b1d20a9..b701efa1f00e 100644
--- a/arch/s390/lib/Makefile
+++ b/arch/s390/lib/Makefile
@@ -5,5 +5,5 @@
 EXTRA_AFLAGS := -traditional
 
 lib-y += delay.o string.o
-lib-$(CONFIG_ARCH_S390_31) += uaccess.o
-lib-$(CONFIG_ARCH_S390X) += uaccess64.o
+lib-$(CONFIG_ARCH_S390_31) += uaccess.o spinlock.o
+lib-$(CONFIG_ARCH_S390X) += uaccess64.o spinlock.o
diff --git a/arch/s390/lib/spinlock.c b/arch/s390/lib/spinlock.c
new file mode 100644
index 000000000000..888b5596c195
--- /dev/null
+++ b/arch/s390/lib/spinlock.c
@@ -0,0 +1,133 @@
+/*
+ *  arch/s390/lib/spinlock.c
+ *    Out of line spinlock code.
+ *
+ *  S390 version
+ *    Copyright (C) 2004 IBM Deutschland Entwicklung GmbH, IBM Corporation
+ *    Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
+ */
+
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/spinlock.h>
+#include <linux/init.h>
+#include <asm/io.h>
+
+atomic_t spin_retry_counter;
+int spin_retry = 1000;
+
+/**
+ * spin_retry= parameter
+ */
+static int __init spin_retry_setup(char *str)
+{
+	spin_retry = simple_strtoul(str, &str, 0);
+	return 1;
+}
+__setup("spin_retry=", spin_retry_setup);
+
+static inline void
+_diag44(void)
+{
+#ifdef __s390x__
+	if (MACHINE_HAS_DIAG44)
+#endif
+		asm volatile("diag 0,0,0x44");
+}
+
+void
+_raw_spin_lock_wait(spinlock_t *lp, unsigned int pc)
+{
+	int count = spin_retry;
+
+	while (1) {
+		if (count-- <= 0) {
+			_diag44();
+			count = spin_retry;
+		}
+		atomic_inc(&spin_retry_counter);
+		if (_raw_compare_and_swap(&lp->lock, 0, pc) == 0)
+			return;
+	}
+}
+EXPORT_SYMBOL(_raw_spin_lock_wait);
+
+int
+_raw_spin_trylock_retry(spinlock_t *lp, unsigned int pc)
+{
+	int count = spin_retry;
+
+	while (count-- > 0) {
+		atomic_inc(&spin_retry_counter);
+		if (_raw_compare_and_swap(&lp->lock, 0, pc) == 0)
+			return 1;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(_raw_spin_trylock_retry);
+
+void
+_raw_read_lock_wait(rwlock_t *rw)
+{
+	unsigned int old;
+	int count = spin_retry;
+
+	while (1) {
+		if (count-- <= 0) {
+			_diag44();
+			count = spin_retry;
+		}
+		atomic_inc(&spin_retry_counter);
+		old = rw->lock & 0x7fffffffU;
+		if (_raw_compare_and_swap(&rw->lock, old, old + 1) == old)
+			return;
+	}
+}
+EXPORT_SYMBOL(_raw_read_lock_wait);
+
+int
+_raw_read_trylock_retry(rwlock_t *rw)
+{
+	unsigned int old;
+	int count = spin_retry;
+
+	while (count-- > 0) {
+		atomic_inc(&spin_retry_counter);
+		old = rw->lock & 0x7fffffffU;
+		if (_raw_compare_and_swap(&rw->lock, old, old + 1) == old)
+			return 1;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(_raw_read_trylock_retry);
+
+void
+_raw_write_lock_wait(rwlock_t *rw)
+{
+	int count = spin_retry;
+
+	while (1) {
+		if (count-- <= 0) {
+			_diag44();
+			count = spin_retry;
+		}
+		atomic_inc(&spin_retry_counter);
+		if (_raw_compare_and_swap(&rw->lock, 0, 0x80000000) == 0)
+			return;
+	}
+}
+EXPORT_SYMBOL(_raw_write_lock_wait);
+
+int
+_raw_write_trylock_retry(rwlock_t *rw)
+{
+	int count = spin_retry;
+
+	while (count-- > 0) {
+		atomic_inc(&spin_retry_counter);
+		if (_raw_compare_and_swap(&rw->lock, 0, 0x80000000) == 0)
+			return 1;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(_raw_write_trylock_retry);