From: Xin Ouyang <Xin.Ouyang@windriver.com>
Date: Mon, 16 Jul 2012 13:29:34 +0800
Subject: [PATCH] tremor: add IT instructions for arm thumb2 tune flags.

Upstream-Status: Pending

In Thumb-2, most instructions do not have a built in condition code (except for 
conditional branches). Instead, short sequences of instructions which are to be 
executed conditionally can be preceded by a special "IT instruction" which 
describes the condition and which of the following instructions should be 
executed if the condition is false respectively. 

For the ARM/Thumb IT(If-Then) instruction:
http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0489c/Cjabicci.html

Signed-off-by: Xin Ouyang <Xin.Ouyang@windriver.com>
---
 asm_arm.h |   14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/asm_arm.h b/asm_arm.h
index c3bda00..823c54f 100755
--- a/asm_arm.h
+++ b/asm_arm.h
@@ -108,9 +108,11 @@ static inline void XNPROD31(ogg_int32_t  a, ogg_int32_t  b,
 static inline ogg_int32_t CLIP_TO_15(ogg_int32_t x) {
   int tmp;
   asm volatile("subs	%1, %0, #32768\n\t"
+	       "itt     pl\n\t"
 	       "movpl	%0, #0x7f00\n\t"
 	       "orrpl	%0, %0, #0xff\n"
 	       "adds	%1, %0, #32768\n\t"
+	       "it      mi\n\t"
 	       "movmi	%0, #0x8000"
 	       : "+r"(x),"=r"(tmp)
 	       :
@@ -139,10 +141,12 @@ static inline void lsp_loop_asm(ogg_uint32_t *qip,ogg_uint32_t *pip,
       
       "ldmdb   r0!,{r1,r3};"
       "subs    r1,r1,%4;"          //ilsp[j]-wi
+      "it      mi;"
       "rsbmi   r1,r1,#0;"          //labs(ilsp[j]-wi)
       "umull   %0,r2,r1,%0;"       //qi*=labs(ilsp[j]-wi)
       
       "subs    r1,r3,%4;"          //ilsp[j+1]-wi
+      "it      mi;"
       "rsbmi   r1,r1,#0;"          //labs(ilsp[j+1]-wi)
       "umull   %1,r3,r1,%1;"       //pi*=labs(ilsp[j+1]-wi)
       
@@ -167,6 +171,7 @@ static inline void lsp_loop_asm(ogg_uint32_t *qip,ogg_uint32_t *pip,
       "mov     r0,#0x4000;\n"
       
       "subs    r1,r1,%4;\n"          //ilsp[j]-wi
+      "it      mi;\n"
       "rsbmi   r1,r1,#0;\n"          //labs(ilsp[j]-wi)
       "umull   %0,r2,r1,%0;\n"       //qi*=labs(ilsp[j]-wi)
       "umull   %1,r3,r0,%1;\n"       //pi*=labs(ilsp[j+1]-wi)
@@ -190,18 +195,23 @@ static inline void lsp_loop_asm(ogg_uint32_t *qip,ogg_uint32_t *pip,
       "mov     r2,#0;"
       "orr     r1,%0,%1;"
       "tst     r1,#0xff000000;"
+      "itt     ne;"
       "addne   r2,r2,#8;"
       "movne   r1,r1,lsr #8;"
       "tst     r1,#0x00f00000;"
+      "itt     ne;"
       "addne   r2,r2,#4;"
       "movne   r1,r1,lsr #4;"
       "tst     r1,#0x000c0000;"
+      "itt     ne;"
       "addne   r2,r2,#2;"
       "movne   r1,r1,lsr #2;"
       "tst     r1,#0x00020000;"
+      "itt     ne;"
       "addne   r2,r2,#1;"
       "movne   r1,r1,lsr #1;"
       "tst     r1,#0x00010000;"
+      "it      ne;"
       "addne   r2,r2,#1;"
       "mov     %0,%0,lsr r2;"
       "mov     %1,%1,lsr r2;"
@@ -222,15 +232,19 @@ static inline void lsp_norm_asm(ogg_uint32_t *qip,ogg_int32_t *qexpp){
   ogg_int32_t qexp=*qexpp;
 
   asm("tst     %0,#0x0000ff00;"
+      "itt     eq;"
       "moveq   %0,%0,lsl #8;"
       "subeq   %1,%1,#8;"
       "tst     %0,#0x0000f000;"
+      "itt     eq;"
       "moveq   %0,%0,lsl #4;"
       "subeq   %1,%1,#4;"
       "tst     %0,#0x0000c000;"
+      "itt     eq;"
       "moveq   %0,%0,lsl #2;"
       "subeq   %1,%1,#2;"
       "tst     %0,#0x00008000;"
+      "itt     eq;"
       "moveq   %0,%0,lsl #1;"
       "subeq   %1,%1,#1;"
       : "+r"(qi),"+r"(qexp)
-- 
1.7.9.5