PowerPC: abort transaction in syscallsazanella/lockelision-ppc

Linux kernel powerpc documentation states issuing a syscall inside a
transaction is not recommended and may lead to undefined behavior. It
also states syscalls does not abort transactoin neither they run in
transactional state.

To avoid side-effects being visible outside transactions, GLIBC with
lock elision enable issues a transaction abort instruction just before all
syscalls if hardware supports hardware transactions.

8 files changed, 90 insertions, 2 deletions

diff --git a/ChangeLog b/ChangeLog
index 615c845967..490112d6bc 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,22 @@
 2014-11-10  Adhemerval Zanella  <azanella@linux.vnet.ibm.com>
 
+	* sysdeps/powerpc/nptl/tls.h (tcbhead_t): Add tm_capable field.
+	(TLS_INIT_TP): Add tm_capable initialization.
+	(TLS_DEFINE_INIT_TP): Likewise.
+	(THREAD_GET_TM_CAPABLE): New file: get tm_capable field value from
+	TCB.
+	(THREAD_SET_TM_CAPABLE): New file: set tm_capable field value in TCB.
+	* sysdeps/powerpc/nptl/tcb-offsets.sym (TM_CAPABLE): Add field offset
+	calculation.
+	* sysdeps/powerpc/powerpc32/sysdep.h (DO_CALL): Abort hardware
+	transactoion is lock elision is built and TCB tm_capable is set.
+	* sysdeps/powerpc/powerpc64/sysdep.h (DO_CALL): Likewise.
+	* sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h
+	(INTERNAL_SYSCALL_NCS): Likewise.
+	* sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h
+	(INTERNAL_SYSCALL_NCS): Likewise.
+	* sysdeps/powerpc/sysdep.h (ABORT_TRANSACTION): New define.
+
 	* sysdeps/powerpc/nptl/elide.h: New file: generic lock elision support
 	for powerpc.
 	* sysdeps/unix/sysv/linux/powerpc/bits/pthreadtypes.h
diff --git a/sysdeps/powerpc/nptl/tcb-offsets.sym b/sysdeps/powerpc/nptl/tcb-offsets.sym
index f996759027..d955142aff 100644
--- a/sysdeps/powerpc/nptl/tcb-offsets.sym
+++ b/sysdeps/powerpc/nptl/tcb-offsets.sym
@@ -19,6 +19,7 @@ POINTER_GUARD			(offsetof (tcbhead_t, pointer_guard) - TLS_TCB_OFFSET - sizeof (
 TAR_SAVE			(offsetof (tcbhead_t, tar_save) - TLS_TCB_OFFSET - sizeof (tcbhead_t))
 DSO_SLOT1			(offsetof (tcbhead_t, dso_slot1) - TLS_TCB_OFFSET - sizeof (tcbhead_t))
 DSO_SLOT2			(offsetof (tcbhead_t, dso_slot2) - TLS_TCB_OFFSET - sizeof (tcbhead_t))
+TM_CAPABLE			(offsetof (tcbhead_t, tm_capable) - TLS_TCB_OFFSET - sizeof (tcbhead_t))
 #ifndef __ASSUME_PRIVATE_FUTEX
 PRIVATE_FUTEX_OFFSET		thread_offsetof (header.private_futex)
 #endif
diff --git a/sysdeps/powerpc/nptl/tls.h b/sysdeps/powerpc/nptl/tls.h
index b80a5fbf54..37280d2fec 100644
--- a/sysdeps/powerpc/nptl/tls.h
+++ b/sysdeps/powerpc/nptl/tls.h
@@ -63,6 +63,8 @@ typedef union dtv
    are private.  */
 typedef struct
 {
+  /* Indicate if hwcap2 has PPC_FEATURE2_HAS_HTM capability.  */
+  int tm_capable;
   /* Reservation for Dynamic System Optimizer ABI.  */
   uintptr_t dso_slot2;
   uintptr_t dso_slot1;
@@ -130,11 +132,17 @@ register void *__thread_register __asm__ ("r13");
    special attention since 'errno' is not yet available and if the
    operation can cause a failure 'errno' must not be touched.  */
 # define TLS_INIT_TP(tcbp) \
-    (__thread_register = (void *) (tcbp) + TLS_TCB_OFFSET, NULL)
+  ({ 									      \
+    __thread_register = (void *) (tcbp) + TLS_TCB_OFFSET;		      \
+    THREAD_SET_TM_CAPABLE (GLRO (dl_hwcap2) & PPC_FEATURE2_HAS_HTM ? 1 : 0);  \
+    NULL;								      \
+  })
 
 /* Value passed to 'clone' for initialization of the thread register.  */
 # define TLS_DEFINE_INIT_TP(tp, pd) \
-  void *tp = (void *) (pd) + TLS_TCB_OFFSET + TLS_PRE_TCB_SIZE
+    void *tp = (void *) (pd) + TLS_TCB_OFFSET + TLS_PRE_TCB_SIZE;	      \
+    (((tcbhead_t *) ((char *) tp - TLS_TCB_OFFSET))[-1].tm_capable) =	      \
+      THREAD_GET_TM_CAPABLE ();
 
 /* Return the address of the dtv for the current thread.  */
 # define THREAD_DTV() \
@@ -188,6 +196,13 @@ register void *__thread_register __asm__ ("r13");
 		     + TLS_PRE_TCB_SIZE))[-1].pointer_guard		      \
      = THREAD_GET_POINTER_GUARD())
 
+/* tm_capable field in TCB head.  */
+# define THREAD_GET_TM_CAPABLE() \
+    (((tcbhead_t *) ((char *) __thread_register				      \
+		     - TLS_TCB_OFFSET))[-1].tm_capable)
+# define THREAD_SET_TM_CAPABLE(value) \
+    (THREAD_GET_TM_CAPABLE () = (value))
+
 /* l_tls_offset == 0 is perfectly valid on PPC, so we have to use some
    different value to mean unset l_tls_offset.  */
 # define NO_TLS_OFFSET		-1
diff --git a/sysdeps/powerpc/powerpc32/sysdep.h b/sysdeps/powerpc/powerpc32/sysdep.h
index c8a56aadbf..c4b3ca8696 100644
--- a/sysdeps/powerpc/powerpc32/sysdep.h
+++ b/sysdeps/powerpc/powerpc32/sysdep.h
@@ -88,7 +88,23 @@ GOT_LABEL:			;					      \
   cfi_endproc;								      \
   ASM_SIZE_DIRECTIVE(name)
 
+#if !defined IS_IN_rtld && defined (ENABLE_LOCK_ELISION)
+# define ABORT_TRANSACTION \
+    cmpwi    2,0;		\
+    beq      1f;		\
+    lwz      0,TM_CAPABLE(2);	\
+    cmpwi    0,0;		\
+    beq	     1f;		\
+    li	     0,_ABORT_SYSCALL;	\
+    tabort.  0;			\
+    .align 4;			\
+1:
+#else
+# define ABORT_TRANSACTION
+#endif
+
 #define DO_CALL(syscall)						      \
+    ABORT_TRANSACTION							      \
     li 0,syscall;							      \
     sc
 
diff --git a/sysdeps/powerpc/powerpc64/sysdep.h b/sysdeps/powerpc/powerpc64/sysdep.h
index b28fb9d8aa..78722c6873 100644
--- a/sysdeps/powerpc/powerpc64/sysdep.h
+++ b/sysdeps/powerpc/powerpc64/sysdep.h
@@ -283,7 +283,23 @@ LT_LABELSUFFIX(name,_name_end): ; \
   TRACEBACK_MASK(name,mask)	\
   END_2(name)
 
+#if !defined IS_IN_rtld && defined (ENABLE_LOCK_ELISION)
+# define ABORT_TRANSACTION \
+    cmpdi    13,0;		\
+    beq      1f;		\
+    lwz      0,TM_CAPABLE(13);	\
+    cmpwi    0,0;		\
+    beq	     1f;		\
+    li	     0,_ABORT_SYSCALL;	\
+    tabort.  0;			\
+    .align 4;                   \
+1:
+#else
+# define ABORT_TRANSACTION
+#endif
+
 #define DO_CALL(syscall) \
+    ABORT_TRANSACTION \
     li 0,syscall; \
     sc
 
diff --git a/sysdeps/powerpc/sysdep.h b/sysdeps/powerpc/sysdep.h
index e6627c071f..c683066673 100644
--- a/sysdeps/powerpc/sysdep.h
+++ b/sysdeps/powerpc/sysdep.h
@@ -21,6 +21,10 @@
  */
 #define _SYSDEPS_SYSDEP_H 1
 #include <bits/hwcap.h>
+#ifdef ENABLE_LOCK_ELISION
+#include <tls.h>
+#include <htm.h>
+#endif
 
 #define PPC_FEATURE_970 (PPC_FEATURE_POWER4 + PPC_FEATURE_HAS_ALTIVEC)
 
@@ -164,4 +168,21 @@
 #define ALIGNARG(log2) log2
 #define ASM_SIZE_DIRECTIVE(name) .size name,.-name
 
+#else
+
+/* Linux kernel powerpc documentation states issuing a syscall inside a
+   transaction is not recommended and may lead to undefined behavior. It
+   also states syscalls does not abort transactoin neither run in
+   transactional state.  To avoid such traps, we abort transaction just
+   before syscalls.  */
+#if !defined IS_IN_rtld && defined (ENABLE_LOCK_ELISION)
+# define ABORT_TRANSACTION \
+  ({ 						\
+    if (THREAD_GET_TM_CAPABLE ())		\
+      __builtin_tabort (_ABORT_SYSCALL);	\
+  })
+#else
+# define ABORT_TRANSACTION
+#endif
+
 #endif	/* __ASSEMBLER__ */
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h b/sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h
index 1a5e37a1d9..0947ca34a6 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h
@@ -194,6 +194,7 @@
     register long int r11 __asm__ ("r11");				\
     register long int r12 __asm__ ("r12");				\
     LOADARGS_##nr(name, args);						\
+    ABORT_TRANSACTION;							\
     __asm__ __volatile__						\
       ("sc   \n\t"							\
        "mfcr %0"							\
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h b/sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h
index 93e454e902..a3cc3025e0 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h
@@ -201,6 +201,7 @@
     register long int r7  __asm__ ("r7");				\
     register long int r8  __asm__ ("r8");				\
     LOADARGS_##nr (name, ##args);					\
+    ABORT_TRANSACTION;							\
     __asm__ __volatile__						\
       ("sc\n\t"								\
        "mfcr  %0\n\t"							\