[PATCH v2 10/14] arm: Implement hard-tp for GET_TLS

[Richard Henderson <rth at twiddle dot net>]

---
	* sysdeps/arm/sysdep.h (ARCH_HAS_HARD_TP): New macro.
	(GET_TLS): Use hard-tp if ARCH_HAS_HARD_TP.
	* sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S: Likewise.
	* sysdeps/unix/sysv/linux/arm/sysdep.h (GET_TLS): Don't override
	the default definition if ARCH_HAS_HARD_TP.
---
 ports/sysdeps/arm/sysdep.h                        | 14 +++++++++++---
 ports/sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S |  5 ++++-
 ports/sysdeps/unix/sysv/linux/arm/sysdep.h        | 16 +++++++++-------
 3 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/ports/sysdeps/arm/sysdep.h b/ports/sysdeps/arm/sysdep.h
index c09e680..03739a4 100644
--- a/ports/sysdeps/arm/sysdep.h
+++ b/ports/sysdeps/arm/sysdep.h
@@ -47,6 +47,9 @@
 #if __ARM_ARCH > 4
 # define ARCH_HAS_BLX
 #endif
+#if __ARM_ARCH > 6 || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6ZK__)
+# define ARCH_HAS_HARD_TP
+#endif
 #if __ARM_ARCH > 6 || defined(__ARM_ARCH_6T2__)
 # define ARCH_HAS_T2
 #endif
@@ -184,10 +187,14 @@
 /* Helper to get the TLS base pointer.  The interface is that TMP is a
    register that may be used to hold the LR, if necessary.  TMP may be
    LR itself to indicate that LR need not be saved.  The base pointer
-   is returned in R0.  Only R0 and TMP are modified.
+   is returned in R0.  Only R0 and TMP are modified.  */
 
-   At this generic level we have no tricks to pull.  Call the ABI routine.  */
-# define GET_TLS(TMP)					\
+# ifdef ARCH_HAS_HARD_TP
+/* If the cpu has cp15 available, use it.  */
+#  define GET_TLS(TMP)		mrc p15, 0, r0, c13, c0, 3
+# else
+/* At this generic level we have no tricks to pull.  Call the ABI routine.  */
+#  define GET_TLS(TMP)					\
 	push	{ r1, r2, r3, lr };			\
 	cfi_remember_state;				\
 	cfi_adjust_cfa_offset (16);			\
@@ -198,6 +205,7 @@
 	bl	__aeabi_read_tp;			\
 	pop	{ r1, r2, r3, lr };			\
 	cfi_restore_state
+# endif /* ARCH_HAS_HARD_TP */
 
 #endif	/* __ASSEMBLER__ */
 
diff --git a/ports/sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S b/ports/sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S
index ecdc322..21e3229 100644
--- a/ports/sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S
+++ b/ports/sysdeps/unix/sysv/linux/arm/aeabi_read_tp.S
@@ -41,7 +41,10 @@
 
 	.hidden __aeabi_read_tp
 ENTRY (__aeabi_read_tp)
-#ifdef __thumb2__
+#ifdef ARCH_HAS_HARD_TP
+	mrc	p15, 0, r0, c13, c0, 3
+	bx	lr
+#elif defined(__thumb2__)
 	movw	r0, #0x0fe0
 	movt	r0, #0xffff
 	bx	r0
diff --git a/ports/sysdeps/unix/sysv/linux/arm/sysdep.h b/ports/sysdeps/unix/sysv/linux/arm/sysdep.h
index 39872b8..89fea7a 100644
--- a/ports/sysdeps/unix/sysv/linux/arm/sysdep.h
+++ b/ports/sysdeps/unix/sysv/linux/arm/sysdep.h
@@ -45,26 +45,27 @@
 
 #ifdef __ASSEMBLER__
 
+#ifndef ARCH_HAS_HARD_TP
 /* Internal macro calling the linux kernel kuser_get_tls helper. 
    Note that in thumb mode, a constant pool break is often out of range, so
    we always expand the constant inline.  */
-#ifdef __thumb2__
-# define GET_TLS_BODY			\
+# ifdef __thumb2__
+#  define GET_TLS_BODY			\
 	movw	r0, #0x0fe0;		\
 	movt	r0, #0xffff;		\
 	blx	r0
-#else
-# define GET_TLS_BODY \
+# else
+#  define GET_TLS_BODY \
 	mov	r0, #0xffff0fff;	/* Point to the high page.  */	\
 	mov	lr, pc;			/* Save our return address.  */	\
 	sub	pc, r0, #31		/* Jump to the TLS entry.  */
-#endif
+# endif
 
 /* Helper to get the TLS base pointer.  Save LR in TMP, return in R0,
    and no other registers clobbered.  TMP may be LR itself to indicate
    that no save is necessary.  */
-#undef GET_TLS
-#define GET_TLS(TMP)			\
+# undef GET_TLS
+# define GET_TLS(TMP)			\
   .ifnc TMP, lr;			\
 	mov	TMP, lr;		\
 	cfi_register (lr, TMP);		\
@@ -74,6 +75,7 @@
   .else;				\
 	GET_TLS_BODY;			\
   .endif
+#endif /* ARCH_HAS_HARD_TP */
 
 /* Linux uses a negative return value to indicate syscall errors,
    unlike most Unices, which use the condition codes' carry flag.
-- 
1.8.1.2