[PATCH 09/10] x86 rwsem: down_read_critical implementation

From: Michel Lespinasse
Date: Mon May 17 2010 - 18:27:46 EST


The basic properties that have been true so far still hold:
- RWSEM_ACTIVE_READ_BIAS & RWSEM_ACTIVE_MASK == 1
- RWSEM_ACTIVE_WRITE_BIAS & RWSEM_ACTIVE_MASK == 1
- RWSEM_WAITING_BIAS & RWSEM_ACTIVE_MASK == 0
- RWSEM_ACTIVE_WRITE_BIAS < 0 and RWSEM_WAITING_BIAS < 0

A new property is introduced: RWSEM_ACTIVE_WRITE_BIAS is set to be
'more negative' than RWSEM_WAITING_BIAS. This way, even the first writer
will set the rwsem count to be < RWSEM_WAITING_BIAS and down_read_critical()
can make use of this to determine if there are any active writers.

Signed-off-by: Michel Lespinasse <walken@xxxxxxxxxx>
---
arch/x86/include/asm/rwsem.h | 53 ++++++++++++++++++++++++++++++++---------
arch/x86/lib/rwsem_64.S | 14 +++++++++-
arch/x86/lib/semaphore_32.S | 21 +++++++++++++++-
3 files changed, 72 insertions(+), 16 deletions(-)

diff --git a/arch/x86/include/asm/rwsem.h b/arch/x86/include/asm/rwsem.h
index 86e0473..fbb068c 100644
--- a/arch/x86/include/asm/rwsem.h
+++ b/arch/x86/include/asm/rwsem.h
@@ -16,11 +16,10 @@
* if there are writers (and maybe) readers waiting (in which case it goes to
* sleep).
*
- * The value of WAITING_BIAS supports up to 32766 waiting processes. This can
- * be extended to 65534 by manually checking the whole MSW rather than relying
- * on the S flag.
+ * The WRITE_BIAS value supports up to 32767 processes simultaneously
+ * trying to acquire a write lock.
*
- * The value of ACTIVE_BIAS supports up to 65535 active processes.
+ * The value of ACTIVE_BIAS supports up to 32767 active processes.
*
* This should be totally fair - if anything is waiting, a process that wants a
* lock will go to the back of the queue. When the currently active lock is
@@ -48,6 +47,8 @@ struct rwsem_waiter;
extern asmregparm struct rw_semaphore *
rwsem_down_read_failed(struct rw_semaphore *sem);
extern asmregparm struct rw_semaphore *
+ rwsem_down_read_unfair_failed(struct rw_semaphore *sem);
+extern asmregparm struct rw_semaphore *
rwsem_down_write_failed(struct rw_semaphore *sem);
extern asmregparm struct rw_semaphore *
rwsem_wake(struct rw_semaphore *);
@@ -62,17 +63,22 @@ extern asmregparm struct rw_semaphore *
* for 64 bits.
*/

+
#ifdef CONFIG_X86_64
-# define RWSEM_ACTIVE_MASK 0xffffffffL
+# define RWSEM_UNLOCKED_VALUE 0x0000000000000000L
+# define RWSEM_ACTIVE_MASK 0x000000007fffffffL
+# define RWSEM_ACTIVE_READ_BIAS 0x0000000000000001L
+# define RWSEM_ACTIVE_WRITE_BIAS 0xffffffff00000001L
+# define RWSEM_WAITING_BIAS 0xffffffff80000000L
#else
-# define RWSEM_ACTIVE_MASK 0x0000ffffL
+# define RWSEM_UNLOCKED_VALUE 0x00000000L
+# define RWSEM_ACTIVE_MASK 0x00007fffL
+# define RWSEM_ACTIVE_READ_BIAS 0x00000001L
+# define RWSEM_ACTIVE_WRITE_BIAS 0xffff0001L
+# define RWSEM_WAITING_BIAS 0xffff8000L
#endif

-#define RWSEM_UNLOCKED_VALUE 0x00000000L
-#define RWSEM_ACTIVE_BIAS 0x00000001L
-#define RWSEM_WAITING_BIAS (-RWSEM_ACTIVE_MASK-1)
-#define RWSEM_ACTIVE_READ_BIAS RWSEM_ACTIVE_BIAS
-#define RWSEM_ACTIVE_WRITE_BIAS (RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
+#define __HAVE_DOWN_READ_UNFAIR

typedef signed long rwsem_count_t;

@@ -129,6 +135,28 @@ static inline void __down_read(struct rw_semaphore *sem)
}

/*
+ * lock for reading - skip waiting threads
+ */
+static inline void __down_read_unfair(struct rw_semaphore *sem)
+{
+ rwsem_count_t tmp;
+
+ tmp = RWSEM_ACTIVE_READ_BIAS;
+ asm volatile("# beginning down_read_unfair\n\t"
+ LOCK_PREFIX " xadd %1,(%2)\n\t"
+ /* adds 0x00000001, returns the old value */
+ " cmp %4,%1\n\t"
+ /* was the count >= RWSEM_WAITING_BIAS before? */
+ " jge 1f\n"
+ " call call_rwsem_down_read_unfair_failed\n"
+ "1:\n"
+ "# ending down_read_unfair"
+ : "+m" (sem->count), "=r" (tmp)
+ : "a" (sem), "1" (tmp), "re" (RWSEM_WAITING_BIAS)
+ : "memory", "cc");
+}
+
+/*
* trylock for reading -- returns 1 if successful, 0 if contention
*/
static inline int __down_read_trylock(struct rw_semaphore *sem)
@@ -242,7 +270,8 @@ static inline void __downgrade_write(struct rw_semaphore *sem)
"1:\n\t"
"# ending __downgrade_write\n"
: "+m" (sem->count)
- : "a" (sem), "er" (-RWSEM_WAITING_BIAS)
+ : "a" (sem),
+ "er" (RWSEM_ACTIVE_READ_BIAS - RWSEM_ACTIVE_WRITE_BIAS)
: "memory", "cc");
}

diff --git a/arch/x86/lib/rwsem_64.S b/arch/x86/lib/rwsem_64.S
index 41fcf00..e99d187 100644
--- a/arch/x86/lib/rwsem_64.S
+++ b/arch/x86/lib/rwsem_64.S
@@ -51,6 +51,16 @@ ENTRY(call_rwsem_down_read_failed)
ret
ENDPROC(call_rwsem_down_read_failed)

+ENTRY(call_rwsem_down_read_unfair_failed)
+ save_common_regs
+ pushq %rdx
+ movq %rax,%rdi
+ call rwsem_down_read_unfair_failed
+ popq %rdx
+ restore_common_regs
+ ret
+ ENDPROC(call_rwsem_down_read_failed)
+
ENTRY(call_rwsem_down_write_failed)
save_common_regs
movq %rax,%rdi
@@ -60,8 +70,8 @@ ENTRY(call_rwsem_down_write_failed)
ENDPROC(call_rwsem_down_write_failed)

ENTRY(call_rwsem_wake)
- decl %edx /* do nothing if still outstanding active readers */
- jnz 1f
+ cmpl $0x80000001, %edx
+ jne 1f /* do nothing unless there are waiters and no active threads */
save_common_regs
movq %rax,%rdi
call rwsem_wake
diff --git a/arch/x86/lib/semaphore_32.S b/arch/x86/lib/semaphore_32.S
index 648fe47..fea9a53 100644
--- a/arch/x86/lib/semaphore_32.S
+++ b/arch/x86/lib/semaphore_32.S
@@ -89,6 +89,23 @@ ENTRY(call_rwsem_down_read_failed)
CFI_ENDPROC
ENDPROC(call_rwsem_down_read_failed)

+ENTRY(call_rwsem_down_read_unfair_failed)
+ CFI_STARTPROC
+ push %ecx
+ CFI_ADJUST_CFA_OFFSET 4
+ CFI_REL_OFFSET ecx,0
+ push %edx
+ CFI_ADJUST_CFA_OFFSET 4
+ CFI_REL_OFFSET edx,0
+ call rwsem_down_read_unfair_failed
+ pop %edx
+ CFI_ADJUST_CFA_OFFSET -4
+ pop %ecx
+ CFI_ADJUST_CFA_OFFSET -4
+ ret
+ CFI_ENDPROC
+ ENDPROC(call_rwsem_down_read_failed)
+
ENTRY(call_rwsem_down_write_failed)
CFI_STARTPROC
push %ecx
@@ -103,8 +120,8 @@ ENTRY(call_rwsem_down_write_failed)

ENTRY(call_rwsem_wake)
CFI_STARTPROC
- decw %dx /* do nothing if still outstanding active readers */
- jnz 1f
+ cmpw $0x8001, %dx
+ jne 1f /* do nothing unless there are waiters and no active threads */
push %ecx
CFI_ADJUST_CFA_OFFSET 4
CFI_REL_OFFSET ecx,0
--
1.7.0.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/