public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* RE: PING:  [PATCH, ARM, iWMMXt][3/5]: built in define and expand
@ 2011-09-26  5:08 Xinyu Qi
  0 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2011-09-26  5:08 UTC (permalink / raw)
  To: Ramana Radhakrishnan; +Cc: gcc-patches

Ping. 

http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html

	* config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
 	(builtin_description bdesc_2arg): Revise built in declaration.
 	(builtin_description bdesc_1arg): Likewise.
 	(arm_init_iwmmxt_builtins): Revise built in initialization.
 	(arm_expand_builtin): Revise built in expansion.

At 2011-07-29 11:03:17,"Xinyu Qi" <xyqi@marvell.com> wrote:
> Ping.
> 
> http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> 
> At 2011-07-14 15:38:47,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > > Hi,
> > >
> > > It is the third part of iWMMXt maintenance.
> > >
> >
> > *config/arm/arm.c (enum arm_builtins): Built-in fcode.
> >  (builtin_description bdesc_2arg): Built in declare.
> >  (builtin_description bdesc_1arg): Ditto.
> >  (arm_init_iwmmxt_builtins): Built in initialize.
> >  (arm_expand_builtin): Built in expand.
> >
> >
> > Thanks,
> > Xinyu

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-11-18 23:47   ` Ramana Radhakrishnan
                       ` (3 preceding siblings ...)
  2012-02-03  2:08     ` Xinyu Qi
@ 2012-03-13  8:56     ` Xinyu Qi
  4 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2012-03-13  8:56 UTC (permalink / raw)
  To: Richard Earnshaw; +Cc: gcc-patches, Ramana Radhakrishnan

PING

At 2012-02-03 10:07:56,"Xinyu Qi" <xyqi@marvell.com> wrote:
> PING
> 
> http://gcc.gnu.org/ml/gcc-patches/2011-12/msg01789.html
> 
> At 2011-12-29 14:25:23,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > > At 2011-11-24 09:27:04,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > > > At 2011-11-19 07:08:22,"Ramana Radhakrishnan"
> > > > <ramana.radhakrishnan@linaro.org> wrote:
> > > > > On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> > > > > > Ping
> > > > > >
> > > > > > http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> > > > > >
> > > > > >        * config/arm/arm.c (enum arm_builtins): Revise built-in
> fcode.
> > > > > >        (builtin_description bdesc_2arg): Revise built in declaration.
> > > > > >        (builtin_description bdesc_1arg): Likewise.
> > > > > >        (arm_init_iwmmxt_builtins): Revise built in initialization.
> > > > > >        (arm_expand_builtin): Revise built in expansion.
> > > > > >
> > > > >
> > > > > This currently doesn't apply - can you take a look ?
> > > >
> > > > Hi Ramana,
> > > >
> > > > I resolve the patch conflict with the newest trunk gcc. The resolved
> > > > diff is attached.
> > > >
> > > > Thanks,
> > > > Xinyu
> > >
> > > Update the built in expand. Remove some redundant code.
> > > New diff is attached.
> > >
> > > Thanks,
> > > Xinyu
> >
> > 	* config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> > 	(IWMMXT2_BUILTIN): New define.
> > 	(IWMMXT2_BUILTIN2): Likewise.
> > 	(iwmmx2_mbuiltin): Likewise.
> > 	(builtin_description bdesc_2arg): Revise built in declaration.
> > 	(builtin_description bdesc_1arg): Likewise.
> > 	(arm_init_iwmmxt_builtins): Revise built in initialization.
> > 	(arm_expand_builtin): Revise built in expansion.
> >
> > Thanks,
> > Xinyu

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-11-18 23:47   ` Ramana Radhakrishnan
                       ` (2 preceding siblings ...)
  2011-12-29  6:39     ` Xinyu Qi
@ 2012-02-03  2:08     ` Xinyu Qi
  2012-03-13  8:56     ` Xinyu Qi
  4 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2012-02-03  2:08 UTC (permalink / raw)
  To: Richard Earnshaw; +Cc: gcc-patches, Ramana Radhakrishnan

PING

http://gcc.gnu.org/ml/gcc-patches/2011-12/msg01789.html

At 2011-12-29 14:25:23,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > At 2011-11-24 09:27:04,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > > At 2011-11-19 07:08:22,"Ramana Radhakrishnan"
> > > <ramana.radhakrishnan@linaro.org> wrote:
> > > > On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> > > > > Ping
> > > > >
> > > > > http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> > > > >
> > > > >        * config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> > > > >        (builtin_description bdesc_2arg): Revise built in declaration.
> > > > >        (builtin_description bdesc_1arg): Likewise.
> > > > >        (arm_init_iwmmxt_builtins): Revise built in initialization.
> > > > >        (arm_expand_builtin): Revise built in expansion.
> > > > >
> > > >
> > > > This currently doesn't apply - can you take a look ?
> > >
> > > Hi Ramana,
> > >
> > > I resolve the patch conflict with the newest trunk gcc. The resolved
> > > diff is attached.
> > >
> > > Thanks,
> > > Xinyu
> >
> > Update the built in expand. Remove some redundant code.
> > New diff is attached.
> >
> > Thanks,
> > Xinyu
> 
> 	* config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> 	(IWMMXT2_BUILTIN): New define.
> 	(IWMMXT2_BUILTIN2): Likewise.
> 	(iwmmx2_mbuiltin): Likewise.
> 	(builtin_description bdesc_2arg): Revise built in declaration.
> 	(builtin_description bdesc_1arg): Likewise.
> 	(arm_init_iwmmxt_builtins): Revise built in initialization.
> 	(arm_expand_builtin): Revise built in expansion.
> 
> Thanks,
> Xinyu

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-11-18 23:47   ` Ramana Radhakrishnan
  2011-11-24  7:09     ` Xinyu Qi
  2011-12-22  6:55     ` Xinyu Qi
@ 2011-12-29  6:39     ` Xinyu Qi
  2012-02-03  2:08     ` Xinyu Qi
  2012-03-13  8:56     ` Xinyu Qi
  4 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2011-12-29  6:39 UTC (permalink / raw)
  To: Richard Earnshaw; +Cc: gcc-patches, Ramana Radhakrishnan

[-- Attachment #1: Type: text/plain, Size: 1429 bytes --]

> At 2011-11-24 09:27:04,"Xinyu Qi" <xyqi@marvell.com> wrote:
> > At 2011-11-19 07:08:22,"Ramana Radhakrishnan"
> > <ramana.radhakrishnan@linaro.org> wrote:
> > > On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> > > > Ping
> > > >
> > > > http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> > > >
> > > >        * config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> > > >        (builtin_description bdesc_2arg): Revise built in declaration.
> > > >        (builtin_description bdesc_1arg): Likewise.
> > > >        (arm_init_iwmmxt_builtins): Revise built in initialization.
> > > >        (arm_expand_builtin): Revise built in expansion.
> > > >
> > >
> > > This currently doesn't apply - can you take a look ?
> >
> > Hi Ramana,
> >
> > I resolve the patch conflict with the newest trunk gcc. The resolved
> > diff is attached.
> >
> > Thanks,
> > Xinyu
> 
> Update the built in expand. Remove some redundant code.
> New diff is attached.
> 
> Thanks,
> Xinyu

	* config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
	(IWMMXT2_BUILTIN): New define.
	(IWMMXT2_BUILTIN2): Likewise.
	(iwmmx2_mbuiltin): Likewise.
	(builtin_description bdesc_2arg): Revise built in declaration.
	(builtin_description bdesc_1arg): Likewise.
	(arm_init_iwmmxt_builtins): Revise built in initialization.
	(arm_expand_builtin): Revise built in expansion.

Thanks,
Xinyu

[-- Attachment #2: 3_arm_c.diff --]
[-- Type: application/octet-stream, Size: 34444 bytes --]

Index: gcc/config/arm/arm.c
===================================================================
--- gcc/config/arm/arm.c	(revision 182703)
+++ gcc/config/arm/arm.c	(working copy)
@@ -19239,8 +19239,15 @@ static neon_builtin_datum neon_builtin_d
    FIXME?  */
 enum arm_builtins
 {
-  ARM_BUILTIN_GETWCX,
-  ARM_BUILTIN_SETWCX,
+  ARM_BUILTIN_GETWCGR0,
+  ARM_BUILTIN_GETWCGR1,
+  ARM_BUILTIN_GETWCGR2,
+  ARM_BUILTIN_GETWCGR3,
+
+  ARM_BUILTIN_SETWCGR0,
+  ARM_BUILTIN_SETWCGR1,
+  ARM_BUILTIN_SETWCGR2,
+  ARM_BUILTIN_SETWCGR3,
 
   ARM_BUILTIN_WZERO,
 
@@ -19263,7 +19270,11 @@ enum arm_builtins
   ARM_BUILTIN_WSADH,
   ARM_BUILTIN_WSADHZ,
 
-  ARM_BUILTIN_WALIGN,
+  ARM_BUILTIN_WALIGNI,
+  ARM_BUILTIN_WALIGNR0,
+  ARM_BUILTIN_WALIGNR1,
+  ARM_BUILTIN_WALIGNR2,
+  ARM_BUILTIN_WALIGNR3,
 
   ARM_BUILTIN_TMIA,
   ARM_BUILTIN_TMIAPH,
@@ -19399,6 +19410,81 @@ enum arm_builtins
   ARM_BUILTIN_WUNPCKELUH,
   ARM_BUILTIN_WUNPCKELUW,
 
+  ARM_BUILTIN_WABSB,
+  ARM_BUILTIN_WABSH,
+  ARM_BUILTIN_WABSW,
+
+  ARM_BUILTIN_WADDSUBHX,
+  ARM_BUILTIN_WSUBADDHX,
+
+  ARM_BUILTIN_WABSDIFFB,
+  ARM_BUILTIN_WABSDIFFH,
+  ARM_BUILTIN_WABSDIFFW,
+
+  ARM_BUILTIN_WADDCH,
+  ARM_BUILTIN_WADDCW,
+
+  ARM_BUILTIN_WAVG4,
+  ARM_BUILTIN_WAVG4R,
+
+  ARM_BUILTIN_WMADDSX,
+  ARM_BUILTIN_WMADDUX,
+
+  ARM_BUILTIN_WMADDSN,
+  ARM_BUILTIN_WMADDUN,
+
+  ARM_BUILTIN_WMULWSM,
+  ARM_BUILTIN_WMULWUM,
+
+  ARM_BUILTIN_WMULWSMR,
+  ARM_BUILTIN_WMULWUMR,
+
+  ARM_BUILTIN_WMULWL,
+
+  ARM_BUILTIN_WMULSMR,
+  ARM_BUILTIN_WMULUMR,
+
+  ARM_BUILTIN_WQMULM,
+  ARM_BUILTIN_WQMULMR,
+
+  ARM_BUILTIN_WQMULWM,
+  ARM_BUILTIN_WQMULWMR,
+
+  ARM_BUILTIN_WADDBHUSM,
+  ARM_BUILTIN_WADDBHUSL,
+
+  ARM_BUILTIN_WQMIABB,
+  ARM_BUILTIN_WQMIABT,
+  ARM_BUILTIN_WQMIATB,
+  ARM_BUILTIN_WQMIATT,
+
+  ARM_BUILTIN_WQMIABBN,
+  ARM_BUILTIN_WQMIABTN,
+  ARM_BUILTIN_WQMIATBN,
+  ARM_BUILTIN_WQMIATTN,
+
+  ARM_BUILTIN_WMIABB,
+  ARM_BUILTIN_WMIABT,
+  ARM_BUILTIN_WMIATB,
+  ARM_BUILTIN_WMIATT,
+
+  ARM_BUILTIN_WMIABBN,
+  ARM_BUILTIN_WMIABTN,
+  ARM_BUILTIN_WMIATBN,
+  ARM_BUILTIN_WMIATTN,
+
+  ARM_BUILTIN_WMIAWBB,
+  ARM_BUILTIN_WMIAWBT,
+  ARM_BUILTIN_WMIAWTB,
+  ARM_BUILTIN_WMIAWTT,
+
+  ARM_BUILTIN_WMIAWBBN,
+  ARM_BUILTIN_WMIAWBTN,
+  ARM_BUILTIN_WMIAWTBN,
+  ARM_BUILTIN_WMIAWTTN,
+
+  ARM_BUILTIN_WMERGE,
+
   ARM_BUILTIN_THREAD_POINTER,
 
   ARM_BUILTIN_NEON_BASE,
@@ -19931,6 +20017,10 @@ static const struct builtin_description 
   { FL_IWMMXT, CODE_FOR_##code, "__builtin_arm_" string, \
     ARM_BUILTIN_##builtin, UNKNOWN, 0 },
 
+#define IWMMXT2_BUILTIN(code, string, builtin) \
+  { FL_IWMMXT2, CODE_FOR_##code, "__builtin_arm_" string, \
+    ARM_BUILTIN_##builtin, UNKNOWN, 0 },
+
   IWMMXT_BUILTIN (addv8qi3, "waddb", WADDB)
   IWMMXT_BUILTIN (addv4hi3, "waddh", WADDH)
   IWMMXT_BUILTIN (addv2si3, "waddw", WADDW)
@@ -19987,44 +20077,45 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckihb, "wunpckihb", WUNPCKIHB)
   IWMMXT_BUILTIN (iwmmxt_wunpckihh, "wunpckihh", WUNPCKIHH)
   IWMMXT_BUILTIN (iwmmxt_wunpckihw, "wunpckihw", WUNPCKIHW)
-  IWMMXT_BUILTIN (iwmmxt_wmadds, "wmadds", WMADDS)
-  IWMMXT_BUILTIN (iwmmxt_wmaddu, "wmaddu", WMADDU)
+  IWMMXT2_BUILTIN (iwmmxt_waddsubhx, "waddsubhx", WADDSUBHX)
+  IWMMXT2_BUILTIN (iwmmxt_wsubaddhx, "wsubaddhx", WSUBADDHX)
+  IWMMXT2_BUILTIN (iwmmxt_wabsdiffb, "wabsdiffb", WABSDIFFB)
+  IWMMXT2_BUILTIN (iwmmxt_wabsdiffh, "wabsdiffh", WABSDIFFH)
+  IWMMXT2_BUILTIN (iwmmxt_wabsdiffw, "wabsdiffw", WABSDIFFW)
+  IWMMXT2_BUILTIN (iwmmxt_avg4, "wavg4", WAVG4)
+  IWMMXT2_BUILTIN (iwmmxt_avg4r, "wavg4r", WAVG4R)
+  IWMMXT2_BUILTIN (iwmmxt_wmulwsm, "wmulwsm", WMULWSM)
+  IWMMXT2_BUILTIN (iwmmxt_wmulwum, "wmulwum", WMULWUM)
+  IWMMXT2_BUILTIN (iwmmxt_wmulwsmr, "wmulwsmr", WMULWSMR)
+  IWMMXT2_BUILTIN (iwmmxt_wmulwumr, "wmulwumr", WMULWUMR)
+  IWMMXT2_BUILTIN (iwmmxt_wmulwl, "wmulwl", WMULWL)
+  IWMMXT2_BUILTIN (iwmmxt_wmulsmr, "wmulsmr", WMULSMR)
+  IWMMXT2_BUILTIN (iwmmxt_wmulumr, "wmulumr", WMULUMR)
+  IWMMXT2_BUILTIN (iwmmxt_wqmulm, "wqmulm", WQMULM)
+  IWMMXT2_BUILTIN (iwmmxt_wqmulmr, "wqmulmr", WQMULMR)
+  IWMMXT2_BUILTIN (iwmmxt_wqmulwm, "wqmulwm", WQMULWM)
+  IWMMXT2_BUILTIN (iwmmxt_wqmulwmr, "wqmulwmr", WQMULWMR)
+  IWMMXT_BUILTIN (iwmmxt_walignr0, "walignr0", WALIGNR0)
+  IWMMXT_BUILTIN (iwmmxt_walignr1, "walignr1", WALIGNR1)
+  IWMMXT_BUILTIN (iwmmxt_walignr2, "walignr2", WALIGNR2)
+  IWMMXT_BUILTIN (iwmmxt_walignr3, "walignr3", WALIGNR3)
 
 #define IWMMXT_BUILTIN2(code, builtin) \
   { FL_IWMMXT, CODE_FOR_##code, NULL, ARM_BUILTIN_##builtin, UNKNOWN, 0 },
 
+#define IWMMXT2_BUILTIN2(code, builtin) \
+  { FL_IWMMXT2, CODE_FOR_##code, NULL, ARM_BUILTIN_##builtin, UNKNOWN, 0 },
+
+  IWMMXT2_BUILTIN2 (iwmmxt_waddbhusm, WADDBHUSM)
+  IWMMXT2_BUILTIN2 (iwmmxt_waddbhusl, WADDBHUSL)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhss, WPACKHSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwss, WPACKWSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdss, WPACKDSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhus, WPACKHUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwus, WPACKWUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdus, WPACKDUS)
-  IWMMXT_BUILTIN2 (ashlv4hi3_di,    WSLLH)
-  IWMMXT_BUILTIN2 (ashlv4hi3_iwmmxt, WSLLHI)
-  IWMMXT_BUILTIN2 (ashlv2si3_di,    WSLLW)
-  IWMMXT_BUILTIN2 (ashlv2si3_iwmmxt, WSLLWI)
-  IWMMXT_BUILTIN2 (ashldi3_di,      WSLLD)
-  IWMMXT_BUILTIN2 (ashldi3_iwmmxt,  WSLLDI)
-  IWMMXT_BUILTIN2 (lshrv4hi3_di,    WSRLH)
-  IWMMXT_BUILTIN2 (lshrv4hi3_iwmmxt, WSRLHI)
-  IWMMXT_BUILTIN2 (lshrv2si3_di,    WSRLW)
-  IWMMXT_BUILTIN2 (lshrv2si3_iwmmxt, WSRLWI)
-  IWMMXT_BUILTIN2 (lshrdi3_di,      WSRLD)
-  IWMMXT_BUILTIN2 (lshrdi3_iwmmxt,  WSRLDI)
-  IWMMXT_BUILTIN2 (ashrv4hi3_di,    WSRAH)
-  IWMMXT_BUILTIN2 (ashrv4hi3_iwmmxt, WSRAHI)
-  IWMMXT_BUILTIN2 (ashrv2si3_di,    WSRAW)
-  IWMMXT_BUILTIN2 (ashrv2si3_iwmmxt, WSRAWI)
-  IWMMXT_BUILTIN2 (ashrdi3_di,      WSRAD)
-  IWMMXT_BUILTIN2 (ashrdi3_iwmmxt,  WSRADI)
-  IWMMXT_BUILTIN2 (rorv4hi3_di,     WRORH)
-  IWMMXT_BUILTIN2 (rorv4hi3,        WRORHI)
-  IWMMXT_BUILTIN2 (rorv2si3_di,     WRORW)
-  IWMMXT_BUILTIN2 (rorv2si3,        WRORWI)
-  IWMMXT_BUILTIN2 (rordi3_di,       WRORD)
-  IWMMXT_BUILTIN2 (rordi3,          WRORDI)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacuz,   WMACUZ)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacsz,   WMACSZ)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacuz, WMACUZ)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacsz, WMACSZ)
 };
 
 static const struct builtin_description bdesc_1arg[] =
@@ -20047,6 +20138,12 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckelsb, "wunpckelsb", WUNPCKELSB)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsh, "wunpckelsh", WUNPCKELSH)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsw, "wunpckelsw", WUNPCKELSW)
+  IWMMXT2_BUILTIN (iwmmxt_wabsv8qi3, "wabsb", WABSB)
+  IWMMXT2_BUILTIN (iwmmxt_wabsv4hi3, "wabsh", WABSH)
+  IWMMXT2_BUILTIN (iwmmxt_wabsv2si3, "wabsw", WABSW)
+  IWMMXT_BUILTIN (tbcstv8qi, "tbcstb", TBCSTB)
+  IWMMXT_BUILTIN (tbcstv4hi, "tbcsth", TBCSTH)
+  IWMMXT_BUILTIN (tbcstv2si, "tbcstw", TBCSTW)
 };
 
 /* Set up all the iWMMXt builtins.  This is not called if
@@ -20062,9 +20159,6 @@ arm_init_iwmmxt_builtins (void)
   tree V4HI_type_node = build_vector_type_for_mode (intHI_type_node, V4HImode);
   tree V8QI_type_node = build_vector_type_for_mode (intQI_type_node, V8QImode);
 
-  tree int_ftype_int
-    = build_function_type_list (integer_type_node,
-				integer_type_node, NULL_TREE);
   tree v8qi_ftype_v8qi_v8qi_int
     = build_function_type_list (V8QI_type_node,
 				V8QI_type_node, V8QI_type_node,
@@ -20126,6 +20220,9 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v2si_v2si
     = build_function_type_list (V4HI_type_node,
 				V2SI_type_node, V2SI_type_node, NULL_TREE);
+  tree v8qi_ftype_v4hi_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V4HI_type_node, V8QI_type_node, NULL_TREE);
   tree v2si_ftype_v4hi_v4hi
     = build_function_type_list (V2SI_type_node,
 				V4HI_type_node, V4HI_type_node, NULL_TREE);
@@ -20140,12 +20237,10 @@ arm_init_iwmmxt_builtins (void)
     = build_function_type_list (V2SI_type_node,
 				V2SI_type_node, long_long_integer_type_node,
 				NULL_TREE);
-  tree void_ftype_int_int
-    = build_function_type_list (void_type_node,
-				integer_type_node, integer_type_node,
-				NULL_TREE);
   tree di_ftype_void
     = build_function_type_list (long_long_unsigned_type_node, NULL_TREE);
+  tree int_ftype_void
+    = build_function_type_list (integer_type_node, NULL_TREE);
   tree di_ftype_v8qi
     = build_function_type_list (long_long_integer_type_node,
 				V8QI_type_node, NULL_TREE);
@@ -20161,6 +20256,15 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v8qi
     = build_function_type_list (V4HI_type_node,
 				V8QI_type_node, NULL_TREE);
+  tree v8qi_ftype_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V8QI_type_node, NULL_TREE);
+  tree v4hi_ftype_v4hi
+    = build_function_type_list (V4HI_type_node,
+	                        V4HI_type_node, NULL_TREE);
+  tree v2si_ftype_v2si
+    = build_function_type_list (V2SI_type_node,
+	                        V2SI_type_node, NULL_TREE);
 
   tree di_ftype_di_v4hi_v4hi
     = build_function_type_list (long_long_unsigned_type_node,
@@ -20173,6 +20277,48 @@ arm_init_iwmmxt_builtins (void)
 				V4HI_type_node,V4HI_type_node,
 				NULL_TREE);
 
+  tree v2si_ftype_v2si_v4hi_v4hi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V4HI_type_node,
+                                V4HI_type_node, NULL_TREE);
+
+  tree v2si_ftype_v2si_v8qi_v8qi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V8QI_type_node,
+                                V8QI_type_node, NULL_TREE);
+
+  tree di_ftype_di_v2si_v2si
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 V2SI_type_node, V2SI_type_node,
+                                 NULL_TREE);
+
+   tree di_ftype_di_di_int
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree void_ftype_void
+     = build_function_type_list (void_type_node,
+                                 NULL_TREE);
+
+   tree void_ftype_int
+     = build_function_type_list (void_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree v8qi_ftype_char
+     = build_function_type_list (V8QI_type_node,
+                                 signed_char_type_node, NULL_TREE);
+
+   tree v4hi_ftype_short
+     = build_function_type_list (V4HI_type_node,
+                                 short_integer_type_node, NULL_TREE);
+
+   tree v2si_ftype_int
+     = build_function_type_list (V2SI_type_node,
+                                 integer_type_node, NULL_TREE);
+
   /* Normal vector binops.  */
   tree v8qi_ftype_v8qi_v8qi
     = build_function_type_list (V8QI_type_node,
@@ -20230,9 +20376,19 @@ arm_init_iwmmxt_builtins (void)
   def_mbuiltin (FL_IWMMXT, "__builtin_arm_" NAME, (TYPE),	\
 		ARM_BUILTIN_ ## CODE)
 
+#define iwmmx2_mbuiltin(NAME, TYPE, CODE)                      \
+  def_mbuiltin (FL_IWMMXT2, "__builtin_arm_" NAME, (TYPE),     \
+               ARM_BUILTIN_ ## CODE)
+
   iwmmx_mbuiltin ("wzero", di_ftype_void, WZERO);
-  iwmmx_mbuiltin ("setwcx", void_ftype_int_int, SETWCX);
-  iwmmx_mbuiltin ("getwcx", int_ftype_int, GETWCX);
+  iwmmx_mbuiltin ("setwcgr0", void_ftype_int, SETWCGR0);
+  iwmmx_mbuiltin ("setwcgr1", void_ftype_int, SETWCGR1);
+  iwmmx_mbuiltin ("setwcgr2", void_ftype_int, SETWCGR2);
+  iwmmx_mbuiltin ("setwcgr3", void_ftype_int, SETWCGR3);
+  iwmmx_mbuiltin ("getwcgr0", int_ftype_void, GETWCGR0);
+  iwmmx_mbuiltin ("getwcgr1", int_ftype_void, GETWCGR1);
+  iwmmx_mbuiltin ("getwcgr2", int_ftype_void, GETWCGR2);
+  iwmmx_mbuiltin ("getwcgr3", int_ftype_void, GETWCGR3);
 
   iwmmx_mbuiltin ("wsllh", v4hi_ftype_v4hi_di, WSLLH);
   iwmmx_mbuiltin ("wsllw", v2si_ftype_v2si_di, WSLLW);
@@ -20264,8 +20420,14 @@ arm_init_iwmmxt_builtins (void)
 
   iwmmx_mbuiltin ("wshufh", v4hi_ftype_v4hi_int, WSHUFH);
 
-  iwmmx_mbuiltin ("wsadb", v2si_ftype_v8qi_v8qi, WSADB);
-  iwmmx_mbuiltin ("wsadh", v2si_ftype_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wsadb", v2si_ftype_v2si_v8qi_v8qi, WSADB);
+  iwmmx_mbuiltin ("wsadh", v2si_ftype_v2si_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wmadds", v2si_ftype_v4hi_v4hi, WMADDS);
+  iwmmx2_mbuiltin ("wmaddsx", v2si_ftype_v4hi_v4hi, WMADDSX);
+  iwmmx2_mbuiltin ("wmaddsn", v2si_ftype_v4hi_v4hi, WMADDSN);
+  iwmmx_mbuiltin ("wmaddu", v2si_ftype_v4hi_v4hi, WMADDU);
+  iwmmx2_mbuiltin ("wmaddux", v2si_ftype_v4hi_v4hi, WMADDUX);
+  iwmmx2_mbuiltin ("wmaddun", v2si_ftype_v4hi_v4hi, WMADDUN);
   iwmmx_mbuiltin ("wsadbz", v2si_ftype_v8qi_v8qi, WSADBZ);
   iwmmx_mbuiltin ("wsadhz", v2si_ftype_v4hi_v4hi, WSADHZ);
 
@@ -20287,6 +20449,9 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmovmskh", int_ftype_v4hi, TMOVMSKH);
   iwmmx_mbuiltin ("tmovmskw", int_ftype_v2si, TMOVMSKW);
 
+  iwmmx2_mbuiltin ("waddbhusm", v8qi_ftype_v4hi_v8qi, WADDBHUSM);
+  iwmmx2_mbuiltin ("waddbhusl", v8qi_ftype_v4hi_v8qi, WADDBHUSL);
+
   iwmmx_mbuiltin ("wpackhss", v8qi_ftype_v4hi_v4hi, WPACKHSS);
   iwmmx_mbuiltin ("wpackhus", v8qi_ftype_v4hi_v4hi, WPACKHUS);
   iwmmx_mbuiltin ("wpackwus", v4hi_ftype_v2si_v2si, WPACKWUS);
@@ -20312,7 +20477,7 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("wmacu", di_ftype_di_v4hi_v4hi, WMACU);
   iwmmx_mbuiltin ("wmacuz", di_ftype_v4hi_v4hi, WMACUZ);
 
-  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGN);
+  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGNI);
   iwmmx_mbuiltin ("tmia", di_ftype_di_int_int, TMIA);
   iwmmx_mbuiltin ("tmiaph", di_ftype_di_int_int, TMIAPH);
   iwmmx_mbuiltin ("tmiabb", di_ftype_di_int_int, TMIABB);
@@ -20320,7 +20485,48 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmiatb", di_ftype_di_int_int, TMIATB);
   iwmmx_mbuiltin ("tmiatt", di_ftype_di_int_int, TMIATT);
 
+  iwmmx2_mbuiltin ("wabsb", v8qi_ftype_v8qi, WABSB);
+  iwmmx2_mbuiltin ("wabsh", v4hi_ftype_v4hi, WABSH);
+  iwmmx2_mbuiltin ("wabsw", v2si_ftype_v2si, WABSW);
+
+  iwmmx2_mbuiltin ("wqmiabb", v2si_ftype_v2si_v4hi_v4hi, WQMIABB);
+  iwmmx2_mbuiltin ("wqmiabt", v2si_ftype_v2si_v4hi_v4hi, WQMIABT);
+  iwmmx2_mbuiltin ("wqmiatb", v2si_ftype_v2si_v4hi_v4hi, WQMIATB);
+  iwmmx2_mbuiltin ("wqmiatt", v2si_ftype_v2si_v4hi_v4hi, WQMIATT);
+
+  iwmmx2_mbuiltin ("wqmiabbn", v2si_ftype_v2si_v4hi_v4hi, WQMIABBN);
+  iwmmx2_mbuiltin ("wqmiabtn", v2si_ftype_v2si_v4hi_v4hi, WQMIABTN);
+  iwmmx2_mbuiltin ("wqmiatbn", v2si_ftype_v2si_v4hi_v4hi, WQMIATBN);
+  iwmmx2_mbuiltin ("wqmiattn", v2si_ftype_v2si_v4hi_v4hi, WQMIATTN);
+
+  iwmmx2_mbuiltin ("wmiabb", di_ftype_di_v4hi_v4hi, WMIABB);
+  iwmmx2_mbuiltin ("wmiabt", di_ftype_di_v4hi_v4hi, WMIABT);
+  iwmmx2_mbuiltin ("wmiatb", di_ftype_di_v4hi_v4hi, WMIATB);
+  iwmmx2_mbuiltin ("wmiatt", di_ftype_di_v4hi_v4hi, WMIATT);
+
+  iwmmx2_mbuiltin ("wmiabbn", di_ftype_di_v4hi_v4hi, WMIABBN);
+  iwmmx2_mbuiltin ("wmiabtn", di_ftype_di_v4hi_v4hi, WMIABTN);
+  iwmmx2_mbuiltin ("wmiatbn", di_ftype_di_v4hi_v4hi, WMIATBN);
+  iwmmx2_mbuiltin ("wmiattn", di_ftype_di_v4hi_v4hi, WMIATTN);
+
+  iwmmx2_mbuiltin ("wmiawbb", di_ftype_di_v2si_v2si, WMIAWBB);
+  iwmmx2_mbuiltin ("wmiawbt", di_ftype_di_v2si_v2si, WMIAWBT);
+  iwmmx2_mbuiltin ("wmiawtb", di_ftype_di_v2si_v2si, WMIAWTB);
+  iwmmx2_mbuiltin ("wmiawtt", di_ftype_di_v2si_v2si, WMIAWTT);
+
+  iwmmx2_mbuiltin ("wmiawbbn", di_ftype_di_v2si_v2si, WMIAWBBN);
+  iwmmx2_mbuiltin ("wmiawbtn", di_ftype_di_v2si_v2si, WMIAWBTN);
+  iwmmx2_mbuiltin ("wmiawtbn", di_ftype_di_v2si_v2si, WMIAWTBN);
+  iwmmx2_mbuiltin ("wmiawttn", di_ftype_di_v2si_v2si, WMIAWTTN);
+
+  iwmmx2_mbuiltin ("wmerge", di_ftype_di_di_int, WMERGE);
+
+  iwmmx_mbuiltin ("tbcstb", v8qi_ftype_char, TBCSTB);
+  iwmmx_mbuiltin ("tbcsth", v4hi_ftype_short, TBCSTH);
+  iwmmx_mbuiltin ("tbcstw", v2si_ftype_int, TBCSTW);
+
 #undef iwmmx_mbuiltin
+#undef iwmmx2_mbuiltin
 }
 
 static void
@@ -20926,6 +21132,10 @@ arm_expand_builtin (tree exp,
   enum machine_mode mode0;
   enum machine_mode mode1;
   enum machine_mode mode2;
+  int opint;
+  int selector;
+  int mask;
+  int imm;
 
   if (fcode >= ARM_BUILTIN_NEON_BASE)
     return arm_expand_neon_builtin (fcode, exp, target);
@@ -20960,6 +21170,24 @@ arm_expand_builtin (tree exp,
 	  error ("selector must be an immediate");
 	  return gen_reg_rtx (tmode);
 	}
+
+      opint = INTVAL (op1);
+      if (fcode == ARM_BUILTIN_TEXTRMSB || fcode == ARM_BUILTIN_TEXTRMUB)
+	{
+	  if (opint > 7 || opint < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else if (fcode == ARM_BUILTIN_TEXTRMSH || fcode == ARM_BUILTIN_TEXTRMUH)
+	{
+	  if (opint > 3 || opint < 0)
+	    error ("the range of selector should be in 0 to 3");
+	}
+      else /* ARM_BUILTIN_TEXTRMSW || ARM_BUILTIN_TEXTRMUW.  */
+	{
+	  if (opint > 1 || opint < 0)
+	    error ("the range of selector should be in 0 to 1");
+	}
+
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20970,11 +21198,61 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
+    case ARM_BUILTIN_WALIGNI:
+      /* If op2 is immediate, call walighi, else call walighr.  */
+      arg0 = CALL_EXPR_ARG (exp, 0);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      arg2 = CALL_EXPR_ARG (exp, 2);
+      op0 = expand_normal (arg0);
+      op1 = expand_normal (arg1);
+      op2 = expand_normal (arg2);
+      if (GET_CODE (op2) == CONST_INT)
+        {
+	  icode = CODE_FOR_iwmmxt_waligni;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          gcc_assert ((*insn_data[icode].operand[3].predicate) (op2, mode2));
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else
+        {
+	  icode = CODE_FOR_iwmmxt_walignr;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          if (!(*insn_data[icode].operand[3].predicate) (op2, mode2))
+	    op2 = copy_to_mode_reg (mode2, op2);
+	}
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+	target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target, op0, op1, op2);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
+      return target;
+
     case ARM_BUILTIN_TINSRB:
     case ARM_BUILTIN_TINSRH:
     case ARM_BUILTIN_TINSRW:
+    case ARM_BUILTIN_WMERGE:
       icode = (fcode == ARM_BUILTIN_TINSRB ? CODE_FOR_iwmmxt_tinsrb
 	       : fcode == ARM_BUILTIN_TINSRH ? CODE_FOR_iwmmxt_tinsrh
+	       : fcode == ARM_BUILTIN_WMERGE ? CODE_FOR_iwmmxt_wmerge
 	       : CODE_FOR_iwmmxt_tinsrw);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
@@ -20993,10 +21271,30 @@ arm_expand_builtin (tree exp,
 	op1 = copy_to_mode_reg (mode1, op1);
       if (! (*insn_data[icode].operand[3].predicate) (op2, mode2))
 	{
-	  /* @@@ better error message */
 	  error ("selector must be an immediate");
 	  return const0_rtx;
 	}
+      if (icode == CODE_FOR_iwmmxt_wmerge)
+	{
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      if ((icode == CODE_FOR_iwmmxt_tinsrb)
+	  || (icode == CODE_FOR_iwmmxt_tinsrh)
+	  || (icode == CODE_FOR_iwmmxt_tinsrw))
+        {
+	  mask = 0x01;
+	  selector= INTVAL (op2);
+	  if (icode == CODE_FOR_iwmmxt_tinsrb && (selector < 0 || selector > 7))
+	    error ("the range of selector should be in 0 to 7");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrh && (selector < 0 ||selector > 3))
+	    error ("the range of selector should be in 0 to 3");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrw && (selector < 0 ||selector > 1))
+	    error ("the range of selector should be in 0 to 1");
+	  mask <<= selector;
+	  op2 = gen_rtx_CONST_INT (SImode, mask);
+	}
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -21007,19 +21305,42 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_SETWCX:
+    case ARM_BUILTIN_SETWCGR0:
+    case ARM_BUILTIN_SETWCGR1:
+    case ARM_BUILTIN_SETWCGR2:
+    case ARM_BUILTIN_SETWCGR3:
+      icode = (fcode == ARM_BUILTIN_SETWCGR0 ? CODE_FOR_iwmmxt_setwcgr0
+	       : fcode == ARM_BUILTIN_SETWCGR1 ? CODE_FOR_iwmmxt_setwcgr1
+	       : fcode == ARM_BUILTIN_SETWCGR2 ? CODE_FOR_iwmmxt_setwcgr2
+	       : CODE_FOR_iwmmxt_setwcgr3);
       arg0 = CALL_EXPR_ARG (exp, 0);
-      arg1 = CALL_EXPR_ARG (exp, 1);
-      op0 = force_reg (SImode, expand_normal (arg0));
-      op1 = expand_normal (arg1);
-      emit_insn (gen_iwmmxt_tmcr (op1, op0));
+      op0 = expand_normal (arg0);
+      mode0 = insn_data[icode].operand[0].mode;
+      if (!(*insn_data[icode].operand[0].predicate) (op0, mode0))
+        op0 = copy_to_mode_reg (mode0, op0);
+      pat = GEN_FCN (icode) (op0);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
       return 0;
 
-    case ARM_BUILTIN_GETWCX:
-      arg0 = CALL_EXPR_ARG (exp, 0);
-      op0 = expand_normal (arg0);
-      target = gen_reg_rtx (SImode);
-      emit_insn (gen_iwmmxt_tmrc (target, op0));
+    case ARM_BUILTIN_GETWCGR0:
+    case ARM_BUILTIN_GETWCGR1:
+    case ARM_BUILTIN_GETWCGR2:
+    case ARM_BUILTIN_GETWCGR3:
+      icode = (fcode == ARM_BUILTIN_GETWCGR0 ? CODE_FOR_iwmmxt_getwcgr0
+	       : fcode == ARM_BUILTIN_GETWCGR1 ? CODE_FOR_iwmmxt_getwcgr1
+	       : fcode == ARM_BUILTIN_GETWCGR2 ? CODE_FOR_iwmmxt_getwcgr2
+	       : CODE_FOR_iwmmxt_getwcgr3);
+      tmode = insn_data[icode].operand[0].mode;
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+        target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target);
+      if (!pat)
+        return 0;
+      emit_insn (pat);
       return target;
 
     case ARM_BUILTIN_WSHUFH:
@@ -21036,10 +21357,12 @@ arm_expand_builtin (tree exp,
 	op0 = copy_to_mode_reg (mode1, op0);
       if (! (*insn_data[icode].operand[2].predicate) (op1, mode2))
 	{
-	  /* @@@ better error message */
 	  error ("mask must be an immediate");
 	  return const0_rtx;
 	}
+      selector = INTVAL (op1);
+      if (selector < 0 || selector > 255)
+	error ("the range of mask should be in 0 to 255");
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -21050,10 +21373,18 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_WSADB:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadb, exp, target);
-    case ARM_BUILTIN_WSADH:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadh, exp, target);
+    case ARM_BUILTIN_WMADDS:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmadds, exp, target);
+    case ARM_BUILTIN_WMADDSX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsx, exp, target);
+    case ARM_BUILTIN_WMADDSN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsn, exp, target);
+    case ARM_BUILTIN_WMADDU:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddu, exp, target);
+    case ARM_BUILTIN_WMADDUX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddux, exp, target);
+    case ARM_BUILTIN_WMADDUN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddun, exp, target);
     case ARM_BUILTIN_WSADBZ:
       return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadbz, exp, target);
     case ARM_BUILTIN_WSADHZ:
@@ -21062,13 +21393,38 @@ arm_expand_builtin (tree exp,
       /* Several three-argument builtins.  */
     case ARM_BUILTIN_WMACS:
     case ARM_BUILTIN_WMACU:
-    case ARM_BUILTIN_WALIGN:
     case ARM_BUILTIN_TMIA:
     case ARM_BUILTIN_TMIAPH:
     case ARM_BUILTIN_TMIATT:
     case ARM_BUILTIN_TMIATB:
     case ARM_BUILTIN_TMIABT:
     case ARM_BUILTIN_TMIABB:
+    case ARM_BUILTIN_WQMIABB:
+    case ARM_BUILTIN_WQMIABT:
+    case ARM_BUILTIN_WQMIATB:
+    case ARM_BUILTIN_WQMIATT:
+    case ARM_BUILTIN_WQMIABBN:
+    case ARM_BUILTIN_WQMIABTN:
+    case ARM_BUILTIN_WQMIATBN:
+    case ARM_BUILTIN_WQMIATTN:
+    case ARM_BUILTIN_WMIABB:
+    case ARM_BUILTIN_WMIABT:
+    case ARM_BUILTIN_WMIATB:
+    case ARM_BUILTIN_WMIATT:
+    case ARM_BUILTIN_WMIABBN:
+    case ARM_BUILTIN_WMIABTN:
+    case ARM_BUILTIN_WMIATBN:
+    case ARM_BUILTIN_WMIATTN:
+    case ARM_BUILTIN_WMIAWBB:
+    case ARM_BUILTIN_WMIAWBT:
+    case ARM_BUILTIN_WMIAWTB:
+    case ARM_BUILTIN_WMIAWTT:
+    case ARM_BUILTIN_WMIAWBBN:
+    case ARM_BUILTIN_WMIAWBTN:
+    case ARM_BUILTIN_WMIAWTBN:
+    case ARM_BUILTIN_WMIAWTTN:
+    case ARM_BUILTIN_WSADB:
+    case ARM_BUILTIN_WSADH:
       icode = (fcode == ARM_BUILTIN_WMACS ? CODE_FOR_iwmmxt_wmacs
 	       : fcode == ARM_BUILTIN_WMACU ? CODE_FOR_iwmmxt_wmacu
 	       : fcode == ARM_BUILTIN_TMIA ? CODE_FOR_iwmmxt_tmia
@@ -21077,7 +21433,32 @@ arm_expand_builtin (tree exp,
 	       : fcode == ARM_BUILTIN_TMIABT ? CODE_FOR_iwmmxt_tmiabt
 	       : fcode == ARM_BUILTIN_TMIATB ? CODE_FOR_iwmmxt_tmiatb
 	       : fcode == ARM_BUILTIN_TMIATT ? CODE_FOR_iwmmxt_tmiatt
-	       : CODE_FOR_iwmmxt_walign);
+	       : fcode == ARM_BUILTIN_WQMIABB ? CODE_FOR_iwmmxt_wqmiabb
+	       : fcode == ARM_BUILTIN_WQMIABT ? CODE_FOR_iwmmxt_wqmiabt
+	       : fcode == ARM_BUILTIN_WQMIATB ? CODE_FOR_iwmmxt_wqmiatb
+	       : fcode == ARM_BUILTIN_WQMIATT ? CODE_FOR_iwmmxt_wqmiatt
+	       : fcode == ARM_BUILTIN_WQMIABBN ? CODE_FOR_iwmmxt_wqmiabbn
+	       : fcode == ARM_BUILTIN_WQMIABTN ? CODE_FOR_iwmmxt_wqmiabtn
+	       : fcode == ARM_BUILTIN_WQMIATBN ? CODE_FOR_iwmmxt_wqmiatbn
+	       : fcode == ARM_BUILTIN_WQMIATTN ? CODE_FOR_iwmmxt_wqmiattn
+	       : fcode == ARM_BUILTIN_WMIABB ? CODE_FOR_iwmmxt_wmiabb
+	       : fcode == ARM_BUILTIN_WMIABT ? CODE_FOR_iwmmxt_wmiabt
+	       : fcode == ARM_BUILTIN_WMIATB ? CODE_FOR_iwmmxt_wmiatb
+	       : fcode == ARM_BUILTIN_WMIATT ? CODE_FOR_iwmmxt_wmiatt
+	       : fcode == ARM_BUILTIN_WMIABBN ? CODE_FOR_iwmmxt_wmiabbn
+	       : fcode == ARM_BUILTIN_WMIABTN ? CODE_FOR_iwmmxt_wmiabtn
+	       : fcode == ARM_BUILTIN_WMIATBN ? CODE_FOR_iwmmxt_wmiatbn
+	       : fcode == ARM_BUILTIN_WMIATTN ? CODE_FOR_iwmmxt_wmiattn
+	       : fcode == ARM_BUILTIN_WMIAWBB ? CODE_FOR_iwmmxt_wmiawbb
+	       : fcode == ARM_BUILTIN_WMIAWBT ? CODE_FOR_iwmmxt_wmiawbt
+	       : fcode == ARM_BUILTIN_WMIAWTB ? CODE_FOR_iwmmxt_wmiawtb
+	       : fcode == ARM_BUILTIN_WMIAWTT ? CODE_FOR_iwmmxt_wmiawtt
+	       : fcode == ARM_BUILTIN_WMIAWBBN ? CODE_FOR_iwmmxt_wmiawbbn
+	       : fcode == ARM_BUILTIN_WMIAWBTN ? CODE_FOR_iwmmxt_wmiawbtn
+	       : fcode == ARM_BUILTIN_WMIAWTBN ? CODE_FOR_iwmmxt_wmiawtbn
+	       : fcode == ARM_BUILTIN_WMIAWTTN ? CODE_FOR_iwmmxt_wmiawttn
+	       : fcode == ARM_BUILTIN_WSADB ? CODE_FOR_iwmmxt_wsadb
+	       : CODE_FOR_iwmmxt_wsadh);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
       arg2 = CALL_EXPR_ARG (exp, 2);
@@ -21110,6 +21491,123 @@ arm_expand_builtin (tree exp,
       emit_insn (gen_iwmmxt_clrdi (target));
       return target;
 
+    case ARM_BUILTIN_WSRLHI:
+    case ARM_BUILTIN_WSRLWI:
+    case ARM_BUILTIN_WSRLDI:
+    case ARM_BUILTIN_WSLLHI:
+    case ARM_BUILTIN_WSLLWI:
+    case ARM_BUILTIN_WSLLDI:
+    case ARM_BUILTIN_WSRAHI:
+    case ARM_BUILTIN_WSRAWI:
+    case ARM_BUILTIN_WSRADI:
+    case ARM_BUILTIN_WRORHI:
+    case ARM_BUILTIN_WRORWI:
+    case ARM_BUILTIN_WRORDI:
+    case ARM_BUILTIN_WSRLH:
+    case ARM_BUILTIN_WSRLW:
+    case ARM_BUILTIN_WSRLD:
+    case ARM_BUILTIN_WSLLH:
+    case ARM_BUILTIN_WSLLW:
+    case ARM_BUILTIN_WSLLD:
+    case ARM_BUILTIN_WSRAH:
+    case ARM_BUILTIN_WSRAW:
+    case ARM_BUILTIN_WSRAD:
+    case ARM_BUILTIN_WRORH:
+    case ARM_BUILTIN_WRORW:
+    case ARM_BUILTIN_WRORD:
+      icode = (fcode == ARM_BUILTIN_WSRLHI ? CODE_FOR_lshrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLWI ? CODE_FOR_lshrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLDI ? CODE_FOR_lshrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSLLHI ? CODE_FOR_ashlv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLWI ? CODE_FOR_ashlv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLDI ? CODE_FOR_ashldi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSRAHI ? CODE_FOR_ashrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRAWI ? CODE_FOR_ashrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRADI ? CODE_FOR_ashrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WRORHI ? CODE_FOR_rorv4hi3
+ 	       : fcode == ARM_BUILTIN_WRORWI ? CODE_FOR_rorv2si3
+ 	       : fcode == ARM_BUILTIN_WRORDI ? CODE_FOR_rordi3
+	       : fcode == ARM_BUILTIN_WSRLH  ? CODE_FOR_lshrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRLW  ? CODE_FOR_lshrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRLD  ? CODE_FOR_lshrdi3_di
+	       : fcode == ARM_BUILTIN_WSLLH  ? CODE_FOR_ashlv4hi3_di
+	       : fcode == ARM_BUILTIN_WSLLW  ? CODE_FOR_ashlv2si3_di
+	       : fcode == ARM_BUILTIN_WSLLD  ? CODE_FOR_ashldi3_di
+	       : fcode == ARM_BUILTIN_WSRAH  ? CODE_FOR_ashrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRAW  ? CODE_FOR_ashrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRAD  ? CODE_FOR_ashrdi3_di
+	       : fcode == ARM_BUILTIN_WRORH  ? CODE_FOR_rorv4hi3_di
+	       : fcode == ARM_BUILTIN_WRORW  ? CODE_FOR_rorv2si3_di
+	       : fcode == ARM_BUILTIN_WRORD  ? CODE_FOR_rordi3_di
+	       : CODE_FOR_nothing);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      op1 = expand_normal (arg1);
+      if (GET_MODE (op1) == VOIDmode)
+	{
+	  imm = INTVAL (op1);
+	  if ((fcode == ARM_BUILTIN_WRORHI || fcode == ARM_BUILTIN_WRORWI
+	       || fcode == ARM_BUILTIN_WRORH || fcode == ARM_BUILTIN_WRORW)
+	      && (imm < 0 || imm > 32))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORHI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORWI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORH)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi16 in code.");
+	      else
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi32 in code.");
+	    }
+	  else if ((fcode == ARM_BUILTIN_WRORDI || fcode == ARM_BUILTIN_WRORD)
+		   && (imm < 0 || imm > 64))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORDI)
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_rori_si64 in code.");
+	      else
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_ror_si64 in code.");
+	    }
+	  else if (imm < 0)
+	    {
+	      if (fcode == ARM_BUILTIN_WSRLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRADI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi32 in code.");
+	      else
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_si64 in code.");
+	    }
+	}
+      return arm_expand_binop_builtin (icode, exp, target);
+    
     case ARM_BUILTIN_THREAD_POINTER:
       return arm_load_tp (target);
 

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-11-18 23:47   ` Ramana Radhakrishnan
  2011-11-24  7:09     ` Xinyu Qi
@ 2011-12-22  6:55     ` Xinyu Qi
  2011-12-29  6:39     ` Xinyu Qi
                       ` (2 subsequent siblings)
  4 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2011-12-22  6:55 UTC (permalink / raw)
  To: Richard Earnshaw; +Cc: gcc-patches, Ramana Radhakrishnan

[-- Attachment #1: Type: text/plain, Size: 969 bytes --]

At 2011-11-24 09:27:04,"Xinyu Qi" <xyqi@marvell.com> wrote:  
> At 2011-11-19 07:08:22,"Ramana Radhakrishnan"
> <ramana.radhakrishnan@linaro.org> wrote:
> > On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> > > Ping
> > >
> > > http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> > >
> > >        * config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> > >        (builtin_description bdesc_2arg): Revise built in declaration.
> > >        (builtin_description bdesc_1arg): Likewise.
> > >        (arm_init_iwmmxt_builtins): Revise built in initialization.
> > >        (arm_expand_builtin): Revise built in expansion.
> > >
> >
> > This currently doesn't apply - can you take a look ?
> 
> Hi Ramana,
> 
> I resolve the patch conflict with the newest trunk gcc. The resolved diff is
> attached.
> 
> Thanks,
> Xinyu

Update the built in expand. Remove some redundant code.
New diff is attached.

Thanks,
Xinyu

[-- Attachment #2: 3_arm_c.diff --]
[-- Type: application/octet-stream, Size: 34016 bytes --]

Index: gcc/config/arm/arm.c
===================================================================
--- gcc/config/arm/arm.c	(revision 182518)
+++ gcc/config/arm/arm.c	(working copy)
@@ -19230,8 +19230,15 @@ static neon_builtin_datum neon_builtin_d
    FIXME?  */
 enum arm_builtins
 {
-  ARM_BUILTIN_GETWCX,
-  ARM_BUILTIN_SETWCX,
+  ARM_BUILTIN_GETWCGR0,
+  ARM_BUILTIN_GETWCGR1,
+  ARM_BUILTIN_GETWCGR2,
+  ARM_BUILTIN_GETWCGR3,
+
+  ARM_BUILTIN_SETWCGR0,
+  ARM_BUILTIN_SETWCGR1,
+  ARM_BUILTIN_SETWCGR2,
+  ARM_BUILTIN_SETWCGR3,
 
   ARM_BUILTIN_WZERO,
 
@@ -19254,7 +19261,11 @@ enum arm_builtins
   ARM_BUILTIN_WSADH,
   ARM_BUILTIN_WSADHZ,
 
-  ARM_BUILTIN_WALIGN,
+  ARM_BUILTIN_WALIGNI,
+  ARM_BUILTIN_WALIGNR0,
+  ARM_BUILTIN_WALIGNR1,
+  ARM_BUILTIN_WALIGNR2,
+  ARM_BUILTIN_WALIGNR3,
 
   ARM_BUILTIN_TMIA,
   ARM_BUILTIN_TMIAPH,
@@ -19390,6 +19401,81 @@ enum arm_builtins
   ARM_BUILTIN_WUNPCKELUH,
   ARM_BUILTIN_WUNPCKELUW,
 
+  ARM_BUILTIN_WABSB,
+  ARM_BUILTIN_WABSH,
+  ARM_BUILTIN_WABSW,
+
+  ARM_BUILTIN_WADDSUBHX,
+  ARM_BUILTIN_WSUBADDHX,
+
+  ARM_BUILTIN_WABSDIFFB,
+  ARM_BUILTIN_WABSDIFFH,
+  ARM_BUILTIN_WABSDIFFW,
+
+  ARM_BUILTIN_WADDCH,
+  ARM_BUILTIN_WADDCW,
+
+  ARM_BUILTIN_WAVG4,
+  ARM_BUILTIN_WAVG4R,
+
+  ARM_BUILTIN_WMADDSX,
+  ARM_BUILTIN_WMADDUX,
+
+  ARM_BUILTIN_WMADDSN,
+  ARM_BUILTIN_WMADDUN,
+
+  ARM_BUILTIN_WMULWSM,
+  ARM_BUILTIN_WMULWUM,
+
+  ARM_BUILTIN_WMULWSMR,
+  ARM_BUILTIN_WMULWUMR,
+
+  ARM_BUILTIN_WMULWL,
+
+  ARM_BUILTIN_WMULSMR,
+  ARM_BUILTIN_WMULUMR,
+
+  ARM_BUILTIN_WQMULM,
+  ARM_BUILTIN_WQMULMR,
+
+  ARM_BUILTIN_WQMULWM,
+  ARM_BUILTIN_WQMULWMR,
+
+  ARM_BUILTIN_WADDBHUSM,
+  ARM_BUILTIN_WADDBHUSL,
+
+  ARM_BUILTIN_WQMIABB,
+  ARM_BUILTIN_WQMIABT,
+  ARM_BUILTIN_WQMIATB,
+  ARM_BUILTIN_WQMIATT,
+
+  ARM_BUILTIN_WQMIABBN,
+  ARM_BUILTIN_WQMIABTN,
+  ARM_BUILTIN_WQMIATBN,
+  ARM_BUILTIN_WQMIATTN,
+
+  ARM_BUILTIN_WMIABB,
+  ARM_BUILTIN_WMIABT,
+  ARM_BUILTIN_WMIATB,
+  ARM_BUILTIN_WMIATT,
+
+  ARM_BUILTIN_WMIABBN,
+  ARM_BUILTIN_WMIABTN,
+  ARM_BUILTIN_WMIATBN,
+  ARM_BUILTIN_WMIATTN,
+
+  ARM_BUILTIN_WMIAWBB,
+  ARM_BUILTIN_WMIAWBT,
+  ARM_BUILTIN_WMIAWTB,
+  ARM_BUILTIN_WMIAWTT,
+
+  ARM_BUILTIN_WMIAWBBN,
+  ARM_BUILTIN_WMIAWBTN,
+  ARM_BUILTIN_WMIAWTBN,
+  ARM_BUILTIN_WMIAWTTN,
+
+  ARM_BUILTIN_WMERGE,
+
   ARM_BUILTIN_THREAD_POINTER,
 
   ARM_BUILTIN_NEON_BASE,
@@ -19978,44 +20064,54 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckihb, "wunpckihb", WUNPCKIHB)
   IWMMXT_BUILTIN (iwmmxt_wunpckihh, "wunpckihh", WUNPCKIHH)
   IWMMXT_BUILTIN (iwmmxt_wunpckihw, "wunpckihw", WUNPCKIHW)
-  IWMMXT_BUILTIN (iwmmxt_wmadds, "wmadds", WMADDS)
-  IWMMXT_BUILTIN (iwmmxt_wmaddu, "wmaddu", WMADDU)
+  IWMMXT_BUILTIN (iwmmxt_waddsubhx, "waddsubhx", WADDSUBHX)
+  IWMMXT_BUILTIN (iwmmxt_wsubaddhx, "wsubaddhx", WSUBADDHX)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffb, "wabsdiffb", WABSDIFFB)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffh, "wabsdiffh", WABSDIFFH)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffw, "wabsdiffw", WABSDIFFW)
+  IWMMXT_BUILTIN (iwmmxt_avg4, "wavg4", WAVG4)
+  IWMMXT_BUILTIN (iwmmxt_avg4r, "wavg4r", WAVG4R)
+  IWMMXT_BUILTIN (iwmmxt_wmulwsm, "wmulwsm", WMULWSM)
+  IWMMXT_BUILTIN (iwmmxt_wmulwum, "wmulwum", WMULWUM)
+  IWMMXT_BUILTIN (iwmmxt_wmulwsmr, "wmulwsmr", WMULWSMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulwumr, "wmulwumr", WMULWUMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulwl, "wmulwl", WMULWL)
+  IWMMXT_BUILTIN (iwmmxt_wmulsmr, "wmulsmr", WMULSMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulumr, "wmulumr", WMULUMR)
+  IWMMXT_BUILTIN (iwmmxt_wqmulm, "wqmulm", WQMULM)
+  IWMMXT_BUILTIN (iwmmxt_wqmulmr, "wqmulmr", WQMULMR)
+  IWMMXT_BUILTIN (iwmmxt_wqmulwm, "wqmulwm", WQMULWM)
+  IWMMXT_BUILTIN (iwmmxt_wqmulwmr, "wqmulwmr", WQMULWMR)
+  IWMMXT_BUILTIN (iwmmxt_walignr0, "walignr0", WALIGNR0)
+  IWMMXT_BUILTIN (iwmmxt_walignr1, "walignr1", WALIGNR1)
+  IWMMXT_BUILTIN (iwmmxt_walignr2, "walignr2", WALIGNR2)
+  IWMMXT_BUILTIN (iwmmxt_walignr3, "walignr3", WALIGNR3)
 
 #define IWMMXT_BUILTIN2(code, builtin) \
   { FL_IWMMXT, CODE_FOR_##code, NULL, ARM_BUILTIN_##builtin, UNKNOWN, 0 },
 
+  IWMMXT_BUILTIN2 (iwmmxt_waddbhusm, WADDBHUSM)
+  IWMMXT_BUILTIN2 (iwmmxt_waddbhusl, WADDBHUSL)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhss, WPACKHSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwss, WPACKWSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdss, WPACKDSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhus, WPACKHUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwus, WPACKWUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdus, WPACKDUS)
-  IWMMXT_BUILTIN2 (ashlv4hi3_di,    WSLLH)
-  IWMMXT_BUILTIN2 (ashlv4hi3_iwmmxt, WSLLHI)
-  IWMMXT_BUILTIN2 (ashlv2si3_di,    WSLLW)
-  IWMMXT_BUILTIN2 (ashlv2si3_iwmmxt, WSLLWI)
-  IWMMXT_BUILTIN2 (ashldi3_di,      WSLLD)
-  IWMMXT_BUILTIN2 (ashldi3_iwmmxt,  WSLLDI)
-  IWMMXT_BUILTIN2 (lshrv4hi3_di,    WSRLH)
-  IWMMXT_BUILTIN2 (lshrv4hi3_iwmmxt, WSRLHI)
-  IWMMXT_BUILTIN2 (lshrv2si3_di,    WSRLW)
-  IWMMXT_BUILTIN2 (lshrv2si3_iwmmxt, WSRLWI)
-  IWMMXT_BUILTIN2 (lshrdi3_di,      WSRLD)
-  IWMMXT_BUILTIN2 (lshrdi3_iwmmxt,  WSRLDI)
-  IWMMXT_BUILTIN2 (ashrv4hi3_di,    WSRAH)
-  IWMMXT_BUILTIN2 (ashrv4hi3_iwmmxt, WSRAHI)
-  IWMMXT_BUILTIN2 (ashrv2si3_di,    WSRAW)
-  IWMMXT_BUILTIN2 (ashrv2si3_iwmmxt, WSRAWI)
-  IWMMXT_BUILTIN2 (ashrdi3_di,      WSRAD)
-  IWMMXT_BUILTIN2 (ashrdi3_iwmmxt,  WSRADI)
-  IWMMXT_BUILTIN2 (rorv4hi3_di,     WRORH)
-  IWMMXT_BUILTIN2 (rorv4hi3,        WRORHI)
-  IWMMXT_BUILTIN2 (rorv2si3_di,     WRORW)
-  IWMMXT_BUILTIN2 (rorv2si3,        WRORWI)
-  IWMMXT_BUILTIN2 (rordi3_di,       WRORD)
-  IWMMXT_BUILTIN2 (rordi3,          WRORDI)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacuz,   WMACUZ)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacsz,   WMACSZ)
+  IWMMXT_BUILTIN2 (ashlv4hi3_di, WSLLH)
+  IWMMXT_BUILTIN2 (ashlv2si3_di, WSLLW)
+  IWMMXT_BUILTIN2 (ashldi3_di, WSLLD)
+  IWMMXT_BUILTIN2 (lshrv4hi3_di, WSRLH)
+  IWMMXT_BUILTIN2 (lshrv2si3_di, WSRLW)
+  IWMMXT_BUILTIN2 (lshrdi3_di, WSRLD)
+  IWMMXT_BUILTIN2 (ashrv4hi3_di, WSRAH)
+  IWMMXT_BUILTIN2 (ashrv2si3_di, WSRAW)
+  IWMMXT_BUILTIN2 (ashrdi3_di, WSRAD)
+  IWMMXT_BUILTIN2 (rorv4hi3_di, WRORH)
+  IWMMXT_BUILTIN2 (rorv2si3_di, WRORW)
+  IWMMXT_BUILTIN2 (rordi3_di, WRORD)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacuz, WMACUZ)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacsz, WMACSZ)
 };
 
 static const struct builtin_description bdesc_1arg[] =
@@ -20038,6 +20134,12 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckelsb, "wunpckelsb", WUNPCKELSB)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsh, "wunpckelsh", WUNPCKELSH)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsw, "wunpckelsw", WUNPCKELSW)
+  IWMMXT_BUILTIN (iwmmxt_wabsv8qi3, "wabsb", WABSB)
+  IWMMXT_BUILTIN (iwmmxt_wabsv4hi3, "wabsh", WABSH)
+  IWMMXT_BUILTIN (iwmmxt_wabsv2si3, "wabsw", WABSW)
+  IWMMXT_BUILTIN (tbcstv8qi, "tbcstb", TBCSTB)
+  IWMMXT_BUILTIN (tbcstv4hi, "tbcsth", TBCSTH)
+  IWMMXT_BUILTIN (tbcstv2si, "tbcstw", TBCSTW)
 };
 
 /* Set up all the iWMMXt builtins.  This is not called if
@@ -20053,9 +20155,6 @@ arm_init_iwmmxt_builtins (void)
   tree V4HI_type_node = build_vector_type_for_mode (intHI_type_node, V4HImode);
   tree V8QI_type_node = build_vector_type_for_mode (intQI_type_node, V8QImode);
 
-  tree int_ftype_int
-    = build_function_type_list (integer_type_node,
-				integer_type_node, NULL_TREE);
   tree v8qi_ftype_v8qi_v8qi_int
     = build_function_type_list (V8QI_type_node,
 				V8QI_type_node, V8QI_type_node,
@@ -20117,6 +20216,9 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v2si_v2si
     = build_function_type_list (V4HI_type_node,
 				V2SI_type_node, V2SI_type_node, NULL_TREE);
+  tree v8qi_ftype_v4hi_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V4HI_type_node, V8QI_type_node, NULL_TREE);
   tree v2si_ftype_v4hi_v4hi
     = build_function_type_list (V2SI_type_node,
 				V4HI_type_node, V4HI_type_node, NULL_TREE);
@@ -20131,12 +20233,10 @@ arm_init_iwmmxt_builtins (void)
     = build_function_type_list (V2SI_type_node,
 				V2SI_type_node, long_long_integer_type_node,
 				NULL_TREE);
-  tree void_ftype_int_int
-    = build_function_type_list (void_type_node,
-				integer_type_node, integer_type_node,
-				NULL_TREE);
   tree di_ftype_void
     = build_function_type_list (long_long_unsigned_type_node, NULL_TREE);
+  tree int_ftype_void
+    = build_function_type_list (integer_type_node, NULL_TREE);
   tree di_ftype_v8qi
     = build_function_type_list (long_long_integer_type_node,
 				V8QI_type_node, NULL_TREE);
@@ -20152,6 +20252,15 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v8qi
     = build_function_type_list (V4HI_type_node,
 				V8QI_type_node, NULL_TREE);
+  tree v8qi_ftype_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V8QI_type_node, NULL_TREE);
+  tree v4hi_ftype_v4hi
+    = build_function_type_list (V4HI_type_node,
+	                        V4HI_type_node, NULL_TREE);
+  tree v2si_ftype_v2si
+    = build_function_type_list (V2SI_type_node,
+	                        V2SI_type_node, NULL_TREE);
 
   tree di_ftype_di_v4hi_v4hi
     = build_function_type_list (long_long_unsigned_type_node,
@@ -20164,6 +20273,48 @@ arm_init_iwmmxt_builtins (void)
 				V4HI_type_node,V4HI_type_node,
 				NULL_TREE);
 
+  tree v2si_ftype_v2si_v4hi_v4hi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V4HI_type_node,
+                                V4HI_type_node, NULL_TREE);
+
+  tree v2si_ftype_v2si_v8qi_v8qi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V8QI_type_node,
+                                V8QI_type_node, NULL_TREE);
+
+  tree di_ftype_di_v2si_v2si
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 V2SI_type_node, V2SI_type_node,
+                                 NULL_TREE);
+
+   tree di_ftype_di_di_int
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree void_ftype_void
+     = build_function_type_list (void_type_node,
+                                 NULL_TREE);
+
+   tree void_ftype_int
+     = build_function_type_list (void_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree v8qi_ftype_char
+     = build_function_type_list (V8QI_type_node,
+                                 signed_char_type_node, NULL_TREE);
+
+   tree v4hi_ftype_short
+     = build_function_type_list (V4HI_type_node,
+                                 short_integer_type_node, NULL_TREE);
+
+   tree v2si_ftype_int
+     = build_function_type_list (V2SI_type_node,
+                                 integer_type_node, NULL_TREE);
+
   /* Normal vector binops.  */
   tree v8qi_ftype_v8qi_v8qi
     = build_function_type_list (V8QI_type_node,
@@ -20222,8 +20373,14 @@ arm_init_iwmmxt_builtins (void)
 		ARM_BUILTIN_ ## CODE)
 
   iwmmx_mbuiltin ("wzero", di_ftype_void, WZERO);
-  iwmmx_mbuiltin ("setwcx", void_ftype_int_int, SETWCX);
-  iwmmx_mbuiltin ("getwcx", int_ftype_int, GETWCX);
+  iwmmx_mbuiltin ("setwcgr0", void_ftype_int, SETWCGR0);
+  iwmmx_mbuiltin ("setwcgr1", void_ftype_int, SETWCGR1);
+  iwmmx_mbuiltin ("setwcgr2", void_ftype_int, SETWCGR2);
+  iwmmx_mbuiltin ("setwcgr3", void_ftype_int, SETWCGR3);
+  iwmmx_mbuiltin ("getwcgr0", int_ftype_void, GETWCGR0);
+  iwmmx_mbuiltin ("getwcgr1", int_ftype_void, GETWCGR1);
+  iwmmx_mbuiltin ("getwcgr2", int_ftype_void, GETWCGR2);
+  iwmmx_mbuiltin ("getwcgr3", int_ftype_void, GETWCGR3);
 
   iwmmx_mbuiltin ("wsllh", v4hi_ftype_v4hi_di, WSLLH);
   iwmmx_mbuiltin ("wsllw", v2si_ftype_v2si_di, WSLLW);
@@ -20255,8 +20412,14 @@ arm_init_iwmmxt_builtins (void)
 
   iwmmx_mbuiltin ("wshufh", v4hi_ftype_v4hi_int, WSHUFH);
 
-  iwmmx_mbuiltin ("wsadb", v2si_ftype_v8qi_v8qi, WSADB);
-  iwmmx_mbuiltin ("wsadh", v2si_ftype_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wsadb", v2si_ftype_v2si_v8qi_v8qi, WSADB);
+  iwmmx_mbuiltin ("wsadh", v2si_ftype_v2si_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wmadds", v2si_ftype_v4hi_v4hi, WMADDS);
+  iwmmx_mbuiltin ("wmaddsx", v2si_ftype_v4hi_v4hi, WMADDSX);
+  iwmmx_mbuiltin ("wmaddsn", v2si_ftype_v4hi_v4hi, WMADDSN);
+  iwmmx_mbuiltin ("wmaddu", v2si_ftype_v4hi_v4hi, WMADDU);
+  iwmmx_mbuiltin ("wmaddux", v2si_ftype_v4hi_v4hi, WMADDUX);
+  iwmmx_mbuiltin ("wmaddun", v2si_ftype_v4hi_v4hi, WMADDUN);
   iwmmx_mbuiltin ("wsadbz", v2si_ftype_v8qi_v8qi, WSADBZ);
   iwmmx_mbuiltin ("wsadhz", v2si_ftype_v4hi_v4hi, WSADHZ);
 
@@ -20278,6 +20441,9 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmovmskh", int_ftype_v4hi, TMOVMSKH);
   iwmmx_mbuiltin ("tmovmskw", int_ftype_v2si, TMOVMSKW);
 
+  iwmmx_mbuiltin ("waddbhusm", v8qi_ftype_v4hi_v8qi, WADDBHUSM);
+  iwmmx_mbuiltin ("waddbhusl", v8qi_ftype_v4hi_v8qi, WADDBHUSL);
+
   iwmmx_mbuiltin ("wpackhss", v8qi_ftype_v4hi_v4hi, WPACKHSS);
   iwmmx_mbuiltin ("wpackhus", v8qi_ftype_v4hi_v4hi, WPACKHUS);
   iwmmx_mbuiltin ("wpackwus", v4hi_ftype_v2si_v2si, WPACKWUS);
@@ -20303,7 +20469,7 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("wmacu", di_ftype_di_v4hi_v4hi, WMACU);
   iwmmx_mbuiltin ("wmacuz", di_ftype_v4hi_v4hi, WMACUZ);
 
-  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGN);
+  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGNI);
   iwmmx_mbuiltin ("tmia", di_ftype_di_int_int, TMIA);
   iwmmx_mbuiltin ("tmiaph", di_ftype_di_int_int, TMIAPH);
   iwmmx_mbuiltin ("tmiabb", di_ftype_di_int_int, TMIABB);
@@ -20311,6 +20477,46 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmiatb", di_ftype_di_int_int, TMIATB);
   iwmmx_mbuiltin ("tmiatt", di_ftype_di_int_int, TMIATT);
 
+  iwmmx_mbuiltin ("wabsb", v8qi_ftype_v8qi, WABSB);
+  iwmmx_mbuiltin ("wabsh", v4hi_ftype_v4hi, WABSH);
+  iwmmx_mbuiltin ("wabsw", v2si_ftype_v2si, WABSW);
+
+  iwmmx_mbuiltin ("wqmiabb", v2si_ftype_v2si_v4hi_v4hi, WQMIABB);
+  iwmmx_mbuiltin ("wqmiabt", v2si_ftype_v2si_v4hi_v4hi, WQMIABT);
+  iwmmx_mbuiltin ("wqmiatb", v2si_ftype_v2si_v4hi_v4hi, WQMIATB);
+  iwmmx_mbuiltin ("wqmiatt", v2si_ftype_v2si_v4hi_v4hi, WQMIATT);
+
+  iwmmx_mbuiltin ("wqmiabbn", v2si_ftype_v2si_v4hi_v4hi, WQMIABBN);
+  iwmmx_mbuiltin ("wqmiabtn", v2si_ftype_v2si_v4hi_v4hi, WQMIABTN);
+  iwmmx_mbuiltin ("wqmiatbn", v2si_ftype_v2si_v4hi_v4hi, WQMIATBN);
+  iwmmx_mbuiltin ("wqmiattn", v2si_ftype_v2si_v4hi_v4hi, WQMIATTN);
+
+  iwmmx_mbuiltin ("wmiabb", di_ftype_di_v4hi_v4hi, WMIABB);
+  iwmmx_mbuiltin ("wmiabt", di_ftype_di_v4hi_v4hi, WMIABT);
+  iwmmx_mbuiltin ("wmiatb", di_ftype_di_v4hi_v4hi, WMIATB);
+  iwmmx_mbuiltin ("wmiatt", di_ftype_di_v4hi_v4hi, WMIATT);
+
+  iwmmx_mbuiltin ("wmiabbn", di_ftype_di_v4hi_v4hi, WMIABBN);
+  iwmmx_mbuiltin ("wmiabtn", di_ftype_di_v4hi_v4hi, WMIABTN);
+  iwmmx_mbuiltin ("wmiatbn", di_ftype_di_v4hi_v4hi, WMIATBN);
+  iwmmx_mbuiltin ("wmiattn", di_ftype_di_v4hi_v4hi, WMIATTN);
+
+  iwmmx_mbuiltin ("wmiawbb", di_ftype_di_v2si_v2si, WMIAWBB);
+  iwmmx_mbuiltin ("wmiawbt", di_ftype_di_v2si_v2si, WMIAWBT);
+  iwmmx_mbuiltin ("wmiawtb", di_ftype_di_v2si_v2si, WMIAWTB);
+  iwmmx_mbuiltin ("wmiawtt", di_ftype_di_v2si_v2si, WMIAWTT);
+
+  iwmmx_mbuiltin ("wmiawbbn", di_ftype_di_v2si_v2si, WMIAWBBN);
+  iwmmx_mbuiltin ("wmiawbtn", di_ftype_di_v2si_v2si, WMIAWBTN);
+  iwmmx_mbuiltin ("wmiawtbn", di_ftype_di_v2si_v2si, WMIAWTBN);
+  iwmmx_mbuiltin ("wmiawttn", di_ftype_di_v2si_v2si, WMIAWTTN);
+
+  iwmmx_mbuiltin ("wmerge", di_ftype_di_di_int, WMERGE);
+
+  iwmmx_mbuiltin ("tbcstb", v8qi_ftype_char, TBCSTB);
+  iwmmx_mbuiltin ("tbcsth", v4hi_ftype_short, TBCSTH);
+  iwmmx_mbuiltin ("tbcstw", v2si_ftype_int, TBCSTW);
+
 #undef iwmmx_mbuiltin
 }
 
@@ -20917,6 +21123,10 @@ arm_expand_builtin (tree exp,
   enum machine_mode mode0;
   enum machine_mode mode1;
   enum machine_mode mode2;
+  int opint;
+  int selector;
+  int mask;
+  int imm;
 
   if (fcode >= ARM_BUILTIN_NEON_BASE)
     return arm_expand_neon_builtin (fcode, exp, target);
@@ -20951,6 +21161,24 @@ arm_expand_builtin (tree exp,
 	  error ("selector must be an immediate");
 	  return gen_reg_rtx (tmode);
 	}
+
+      opint = INTVAL (op1);
+      if (fcode == ARM_BUILTIN_TEXTRMSB || fcode == ARM_BUILTIN_TEXTRMUB)
+	{
+	  if (opint > 7 || opint < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else if (fcode == ARM_BUILTIN_TEXTRMSH || fcode == ARM_BUILTIN_TEXTRMUH)
+	{
+	  if (opint > 3 || opint < 0)
+	    error ("the range of selector should be in 0 to 3");
+	}
+      else /* ARM_BUILTIN_TEXTRMSW || ARM_BUILTIN_TEXTRMUW.  */
+	{
+	  if (opint > 1 || opint < 0)
+	    error ("the range of selector should be in 0 to 1");
+	}
+
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20961,11 +21189,61 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
+    case ARM_BUILTIN_WALIGNI:
+      /* If op2 is immediate, call walighi, else call walighr.  */
+      arg0 = CALL_EXPR_ARG (exp, 0);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      arg2 = CALL_EXPR_ARG (exp, 2);
+      op0 = expand_normal (arg0);
+      op1 = expand_normal (arg1);
+      op2 = expand_normal (arg2);
+      if (GET_CODE (op2) == CONST_INT)
+        {
+	  icode = CODE_FOR_iwmmxt_waligni;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          gcc_assert ((*insn_data[icode].operand[3].predicate) (op2, mode2));
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else
+        {
+	  icode = CODE_FOR_iwmmxt_walignr;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          if (!(*insn_data[icode].operand[3].predicate) (op2, mode2))
+	    op2 = copy_to_mode_reg (mode2, op2);
+	}
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+	target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target, op0, op1, op2);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
+      return target;
+
     case ARM_BUILTIN_TINSRB:
     case ARM_BUILTIN_TINSRH:
     case ARM_BUILTIN_TINSRW:
+    case ARM_BUILTIN_WMERGE:
       icode = (fcode == ARM_BUILTIN_TINSRB ? CODE_FOR_iwmmxt_tinsrb
 	       : fcode == ARM_BUILTIN_TINSRH ? CODE_FOR_iwmmxt_tinsrh
+	       : fcode == ARM_BUILTIN_WMERGE ? CODE_FOR_iwmmxt_wmerge
 	       : CODE_FOR_iwmmxt_tinsrw);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
@@ -20984,10 +21262,30 @@ arm_expand_builtin (tree exp,
 	op1 = copy_to_mode_reg (mode1, op1);
       if (! (*insn_data[icode].operand[3].predicate) (op2, mode2))
 	{
-	  /* @@@ better error message */
 	  error ("selector must be an immediate");
 	  return const0_rtx;
 	}
+      if (icode == CODE_FOR_iwmmxt_wmerge)
+	{
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      if ((icode == CODE_FOR_iwmmxt_tinsrb)
+	  || (icode == CODE_FOR_iwmmxt_tinsrh)
+	  || (icode == CODE_FOR_iwmmxt_tinsrw))
+        {
+	  mask = 0x01;
+	  selector= INTVAL (op2);
+	  if (icode == CODE_FOR_iwmmxt_tinsrb && (selector < 0 || selector > 7))
+	    error ("the range of selector should be in 0 to 7");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrh && (selector < 0 ||selector > 3))
+	    error ("the range of selector should be in 0 to 3");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrw && (selector < 0 ||selector > 1))
+	    error ("the range of selector should be in 0 to 1");
+	  mask <<= selector;
+	  op2 = gen_rtx_CONST_INT (SImode, mask);
+	}
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20998,19 +21296,42 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_SETWCX:
+    case ARM_BUILTIN_SETWCGR0:
+    case ARM_BUILTIN_SETWCGR1:
+    case ARM_BUILTIN_SETWCGR2:
+    case ARM_BUILTIN_SETWCGR3:
+      icode = (fcode == ARM_BUILTIN_SETWCGR0 ? CODE_FOR_iwmmxt_setwcgr0
+	       : fcode == ARM_BUILTIN_SETWCGR1 ? CODE_FOR_iwmmxt_setwcgr1
+	       : fcode == ARM_BUILTIN_SETWCGR2 ? CODE_FOR_iwmmxt_setwcgr2
+	       : CODE_FOR_iwmmxt_setwcgr3);
       arg0 = CALL_EXPR_ARG (exp, 0);
-      arg1 = CALL_EXPR_ARG (exp, 1);
-      op0 = force_reg (SImode, expand_normal (arg0));
-      op1 = expand_normal (arg1);
-      emit_insn (gen_iwmmxt_tmcr (op1, op0));
+      op0 = expand_normal (arg0);
+      mode0 = insn_data[icode].operand[0].mode;
+      if (!(*insn_data[icode].operand[0].predicate) (op0, mode0))
+        op0 = copy_to_mode_reg (mode0, op0);
+      pat = GEN_FCN (icode) (op0);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
       return 0;
 
-    case ARM_BUILTIN_GETWCX:
-      arg0 = CALL_EXPR_ARG (exp, 0);
-      op0 = expand_normal (arg0);
-      target = gen_reg_rtx (SImode);
-      emit_insn (gen_iwmmxt_tmrc (target, op0));
+    case ARM_BUILTIN_GETWCGR0:
+    case ARM_BUILTIN_GETWCGR1:
+    case ARM_BUILTIN_GETWCGR2:
+    case ARM_BUILTIN_GETWCGR3:
+      icode = (fcode == ARM_BUILTIN_GETWCGR0 ? CODE_FOR_iwmmxt_getwcgr0
+	       : fcode == ARM_BUILTIN_GETWCGR1 ? CODE_FOR_iwmmxt_getwcgr1
+	       : fcode == ARM_BUILTIN_GETWCGR2 ? CODE_FOR_iwmmxt_getwcgr2
+	       : CODE_FOR_iwmmxt_getwcgr3);
+      tmode = insn_data[icode].operand[0].mode;
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+        target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target);
+      if (!pat)
+        return 0;
+      emit_insn (pat);
       return target;
 
     case ARM_BUILTIN_WSHUFH:
@@ -21027,10 +21348,12 @@ arm_expand_builtin (tree exp,
 	op0 = copy_to_mode_reg (mode1, op0);
       if (! (*insn_data[icode].operand[2].predicate) (op1, mode2))
 	{
-	  /* @@@ better error message */
 	  error ("mask must be an immediate");
 	  return const0_rtx;
 	}
+      selector = INTVAL (op1);
+      if (selector < 0 || selector > 255)
+	error ("the range of mask should be in 0 to 255");
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -21041,10 +21364,18 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_WSADB:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadb, exp, target);
-    case ARM_BUILTIN_WSADH:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadh, exp, target);
+    case ARM_BUILTIN_WMADDS:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmadds, exp, target);
+    case ARM_BUILTIN_WMADDSX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsx, exp, target);
+    case ARM_BUILTIN_WMADDSN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsn, exp, target);
+    case ARM_BUILTIN_WMADDU:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddu, exp, target);
+    case ARM_BUILTIN_WMADDUX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddux, exp, target);
+    case ARM_BUILTIN_WMADDUN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddun, exp, target);
     case ARM_BUILTIN_WSADBZ:
       return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadbz, exp, target);
     case ARM_BUILTIN_WSADHZ:
@@ -21053,13 +21384,38 @@ arm_expand_builtin (tree exp,
       /* Several three-argument builtins.  */
     case ARM_BUILTIN_WMACS:
     case ARM_BUILTIN_WMACU:
-    case ARM_BUILTIN_WALIGN:
     case ARM_BUILTIN_TMIA:
     case ARM_BUILTIN_TMIAPH:
     case ARM_BUILTIN_TMIATT:
     case ARM_BUILTIN_TMIATB:
     case ARM_BUILTIN_TMIABT:
     case ARM_BUILTIN_TMIABB:
+    case ARM_BUILTIN_WQMIABB:
+    case ARM_BUILTIN_WQMIABT:
+    case ARM_BUILTIN_WQMIATB:
+    case ARM_BUILTIN_WQMIATT:
+    case ARM_BUILTIN_WQMIABBN:
+    case ARM_BUILTIN_WQMIABTN:
+    case ARM_BUILTIN_WQMIATBN:
+    case ARM_BUILTIN_WQMIATTN:
+    case ARM_BUILTIN_WMIABB:
+    case ARM_BUILTIN_WMIABT:
+    case ARM_BUILTIN_WMIATB:
+    case ARM_BUILTIN_WMIATT:
+    case ARM_BUILTIN_WMIABBN:
+    case ARM_BUILTIN_WMIABTN:
+    case ARM_BUILTIN_WMIATBN:
+    case ARM_BUILTIN_WMIATTN:
+    case ARM_BUILTIN_WMIAWBB:
+    case ARM_BUILTIN_WMIAWBT:
+    case ARM_BUILTIN_WMIAWTB:
+    case ARM_BUILTIN_WMIAWTT:
+    case ARM_BUILTIN_WMIAWBBN:
+    case ARM_BUILTIN_WMIAWBTN:
+    case ARM_BUILTIN_WMIAWTBN:
+    case ARM_BUILTIN_WMIAWTTN:
+    case ARM_BUILTIN_WSADB:
+    case ARM_BUILTIN_WSADH:
       icode = (fcode == ARM_BUILTIN_WMACS ? CODE_FOR_iwmmxt_wmacs
 	       : fcode == ARM_BUILTIN_WMACU ? CODE_FOR_iwmmxt_wmacu
 	       : fcode == ARM_BUILTIN_TMIA ? CODE_FOR_iwmmxt_tmia
@@ -21068,7 +21424,32 @@ arm_expand_builtin (tree exp,
 	       : fcode == ARM_BUILTIN_TMIABT ? CODE_FOR_iwmmxt_tmiabt
 	       : fcode == ARM_BUILTIN_TMIATB ? CODE_FOR_iwmmxt_tmiatb
 	       : fcode == ARM_BUILTIN_TMIATT ? CODE_FOR_iwmmxt_tmiatt
-	       : CODE_FOR_iwmmxt_walign);
+	       : fcode == ARM_BUILTIN_WQMIABB ? CODE_FOR_iwmmxt_wqmiabb
+	       : fcode == ARM_BUILTIN_WQMIABT ? CODE_FOR_iwmmxt_wqmiabt
+	       : fcode == ARM_BUILTIN_WQMIATB ? CODE_FOR_iwmmxt_wqmiatb
+	       : fcode == ARM_BUILTIN_WQMIATT ? CODE_FOR_iwmmxt_wqmiatt
+	       : fcode == ARM_BUILTIN_WQMIABBN ? CODE_FOR_iwmmxt_wqmiabbn
+	       : fcode == ARM_BUILTIN_WQMIABTN ? CODE_FOR_iwmmxt_wqmiabtn
+	       : fcode == ARM_BUILTIN_WQMIATBN ? CODE_FOR_iwmmxt_wqmiatbn
+	       : fcode == ARM_BUILTIN_WQMIATTN ? CODE_FOR_iwmmxt_wqmiattn
+	       : fcode == ARM_BUILTIN_WMIABB ? CODE_FOR_iwmmxt_wmiabb
+	       : fcode == ARM_BUILTIN_WMIABT ? CODE_FOR_iwmmxt_wmiabt
+	       : fcode == ARM_BUILTIN_WMIATB ? CODE_FOR_iwmmxt_wmiatb
+	       : fcode == ARM_BUILTIN_WMIATT ? CODE_FOR_iwmmxt_wmiatt
+	       : fcode == ARM_BUILTIN_WMIABBN ? CODE_FOR_iwmmxt_wmiabbn
+	       : fcode == ARM_BUILTIN_WMIABTN ? CODE_FOR_iwmmxt_wmiabtn
+	       : fcode == ARM_BUILTIN_WMIATBN ? CODE_FOR_iwmmxt_wmiatbn
+	       : fcode == ARM_BUILTIN_WMIATTN ? CODE_FOR_iwmmxt_wmiattn
+	       : fcode == ARM_BUILTIN_WMIAWBB ? CODE_FOR_iwmmxt_wmiawbb
+	       : fcode == ARM_BUILTIN_WMIAWBT ? CODE_FOR_iwmmxt_wmiawbt
+	       : fcode == ARM_BUILTIN_WMIAWTB ? CODE_FOR_iwmmxt_wmiawtb
+	       : fcode == ARM_BUILTIN_WMIAWTT ? CODE_FOR_iwmmxt_wmiawtt
+	       : fcode == ARM_BUILTIN_WMIAWBBN ? CODE_FOR_iwmmxt_wmiawbbn
+	       : fcode == ARM_BUILTIN_WMIAWBTN ? CODE_FOR_iwmmxt_wmiawbtn
+	       : fcode == ARM_BUILTIN_WMIAWTBN ? CODE_FOR_iwmmxt_wmiawtbn
+	       : fcode == ARM_BUILTIN_WMIAWTTN ? CODE_FOR_iwmmxt_wmiawttn
+	       : fcode == ARM_BUILTIN_WSADB ? CODE_FOR_iwmmxt_wsadb
+	       : CODE_FOR_iwmmxt_wsadh);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
       arg2 = CALL_EXPR_ARG (exp, 2);
@@ -21101,6 +21482,123 @@ arm_expand_builtin (tree exp,
       emit_insn (gen_iwmmxt_clrdi (target));
       return target;
 
+    case ARM_BUILTIN_WSRLHI:
+    case ARM_BUILTIN_WSRLWI:
+    case ARM_BUILTIN_WSRLDI:
+    case ARM_BUILTIN_WSLLHI:
+    case ARM_BUILTIN_WSLLWI:
+    case ARM_BUILTIN_WSLLDI:
+    case ARM_BUILTIN_WSRAHI:
+    case ARM_BUILTIN_WSRAWI:
+    case ARM_BUILTIN_WSRADI:
+    case ARM_BUILTIN_WRORHI:
+    case ARM_BUILTIN_WRORWI:
+    case ARM_BUILTIN_WRORDI:
+    case ARM_BUILTIN_WSRLH:
+    case ARM_BUILTIN_WSRLW:
+    case ARM_BUILTIN_WSRLD:
+    case ARM_BUILTIN_WSLLH:
+    case ARM_BUILTIN_WSLLW:
+    case ARM_BUILTIN_WSLLD:
+    case ARM_BUILTIN_WSRAH:
+    case ARM_BUILTIN_WSRAW:
+    case ARM_BUILTIN_WSRAD:
+    case ARM_BUILTIN_WRORH:
+    case ARM_BUILTIN_WRORW:
+    case ARM_BUILTIN_WRORD:
+      icode = (fcode == ARM_BUILTIN_WSRLHI ? CODE_FOR_lshrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLWI ? CODE_FOR_lshrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLDI ? CODE_FOR_lshrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSLLHI ? CODE_FOR_ashlv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLWI ? CODE_FOR_ashlv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLDI ? CODE_FOR_ashldi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSRAHI ? CODE_FOR_ashrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRAWI ? CODE_FOR_ashrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRADI ? CODE_FOR_ashrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WRORHI ? CODE_FOR_rorv4hi3
+ 	       : fcode == ARM_BUILTIN_WRORWI ? CODE_FOR_rorv2si3
+ 	       : fcode == ARM_BUILTIN_WRORDI ? CODE_FOR_rordi3
+	       : fcode == ARM_BUILTIN_WSRLH  ? CODE_FOR_lshrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRLW  ? CODE_FOR_lshrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRLD  ? CODE_FOR_lshrdi3_di
+	       : fcode == ARM_BUILTIN_WSLLH  ? CODE_FOR_ashlv4hi3_di
+	       : fcode == ARM_BUILTIN_WSLLW  ? CODE_FOR_ashlv2si3_di
+	       : fcode == ARM_BUILTIN_WSLLD  ? CODE_FOR_ashldi3_di
+	       : fcode == ARM_BUILTIN_WSRAH  ? CODE_FOR_ashrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRAW  ? CODE_FOR_ashrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRAD  ? CODE_FOR_ashrdi3_di
+	       : fcode == ARM_BUILTIN_WRORH  ? CODE_FOR_rorv4hi3_di
+	       : fcode == ARM_BUILTIN_WRORW  ? CODE_FOR_rorv2si3_di
+	       : fcode == ARM_BUILTIN_WRORD  ? CODE_FOR_rordi3_di
+	       : CODE_FOR_nothing);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      op1 = expand_normal (arg1);
+      if (GET_MODE (op1) == VOIDmode)
+	{
+	  imm = INTVAL (op1);
+	  if ((fcode == ARM_BUILTIN_WRORHI || fcode == ARM_BUILTIN_WRORWI
+	       || fcode == ARM_BUILTIN_WRORH || fcode == ARM_BUILTIN_WRORW)
+	      && (imm < 0 || imm > 32))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORHI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORWI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORH)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi16 in code.");
+	      else
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi32 in code.");
+	    }
+	  else if ((fcode == ARM_BUILTIN_WRORDI || fcode == ARM_BUILTIN_WRORD)
+		   && (imm < 0 || imm > 64))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORDI)
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_rori_si64 in code.");
+	      else
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_ror_si64 in code.");
+	    }
+	  else if (imm < 0)
+	    {
+	      if (fcode == ARM_BUILTIN_WSRLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRADI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi32 in code.");
+	      else
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_si64 in code.");
+	    }
+	}
+      return arm_expand_binop_builtin (icode, exp, target);
+    
     case ARM_BUILTIN_THREAD_POINTER:
       return arm_load_tp (target);
 

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-11-18 23:47   ` Ramana Radhakrishnan
@ 2011-11-24  7:09     ` Xinyu Qi
  2011-12-22  6:55     ` Xinyu Qi
                       ` (3 subsequent siblings)
  4 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2011-11-24  7:09 UTC (permalink / raw)
  To: Ramana Radhakrishnan; +Cc: gcc-patches

[-- Attachment #1: Type: text/plain, Size: 760 bytes --]

At 2011-11-19 07:08:22,"Ramana Radhakrishnan" <ramana.radhakrishnan@linaro.org> wrote: 
> On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> > Ping
> >
> > http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
> >
> >        * config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
> >        (builtin_description bdesc_2arg): Revise built in declaration.
> >        (builtin_description bdesc_1arg): Likewise.
> >        (arm_init_iwmmxt_builtins): Revise built in initialization.
> >        (arm_expand_builtin): Revise built in expansion.
> >
> 
> This currently doesn't apply - can you take a look ?

Hi Ramana,

I resolve the patch conflict with the newest trunk gcc. The resolved diff is attached.

Thanks,
Xinyu

[-- Attachment #2: 3_arm_c.diff --]
[-- Type: application/octet-stream, Size: 38151 bytes --]

Index: gcc/config/arm/arm.c
===================================================================
--- gcc/config/arm/arm.c	(revision 181648)
+++ gcc/config/arm/arm.c	(working copy)
@@ -19100,8 +19100,15 @@ static neon_builtin_datum neon_builtin_d
    FIXME?  */
 enum arm_builtins
 {
-  ARM_BUILTIN_GETWCX,
-  ARM_BUILTIN_SETWCX,
+  ARM_BUILTIN_GETWCGR0,
+  ARM_BUILTIN_GETWCGR1,
+  ARM_BUILTIN_GETWCGR2,
+  ARM_BUILTIN_GETWCGR3,
+
+  ARM_BUILTIN_SETWCGR0,
+  ARM_BUILTIN_SETWCGR1,
+  ARM_BUILTIN_SETWCGR2,
+  ARM_BUILTIN_SETWCGR3,
 
   ARM_BUILTIN_WZERO,
 
@@ -19124,7 +19131,11 @@ enum arm_builtins
   ARM_BUILTIN_WSADH,
   ARM_BUILTIN_WSADHZ,
 
-  ARM_BUILTIN_WALIGN,
+  ARM_BUILTIN_WALIGNI,
+  ARM_BUILTIN_WALIGNR0,
+  ARM_BUILTIN_WALIGNR1,
+  ARM_BUILTIN_WALIGNR2,
+  ARM_BUILTIN_WALIGNR3,
 
   ARM_BUILTIN_TMIA,
   ARM_BUILTIN_TMIAPH,
@@ -19260,6 +19271,97 @@ enum arm_builtins
   ARM_BUILTIN_WUNPCKELUH,
   ARM_BUILTIN_WUNPCKELUW,
 
+  ARM_BUILTIN_WABSB,
+  ARM_BUILTIN_WABSH,
+  ARM_BUILTIN_WABSW,
+
+  ARM_BUILTIN_WADDSUBHX,
+  ARM_BUILTIN_WSUBADDHX,
+
+  ARM_BUILTIN_WABSDIFFB,
+  ARM_BUILTIN_WABSDIFFH,
+  ARM_BUILTIN_WABSDIFFW,
+
+  ARM_BUILTIN_WADDCH,
+  ARM_BUILTIN_WADDCW,
+
+  ARM_BUILTIN_WAVG4,
+  ARM_BUILTIN_WAVG4R,
+
+  ARM_BUILTIN_WMADDSX,
+  ARM_BUILTIN_WMADDUX,
+
+  ARM_BUILTIN_WMADDSN,
+  ARM_BUILTIN_WMADDUN,
+
+  ARM_BUILTIN_WMULWSM,
+  ARM_BUILTIN_WMULWUM,
+
+  ARM_BUILTIN_WMULWSMR,
+  ARM_BUILTIN_WMULWUMR,
+
+  ARM_BUILTIN_WMULWL,
+
+  ARM_BUILTIN_WMULSMR,
+  ARM_BUILTIN_WMULUMR,
+
+  ARM_BUILTIN_WQMULM,
+  ARM_BUILTIN_WQMULMR,
+
+  ARM_BUILTIN_WQMULWM,
+  ARM_BUILTIN_WQMULWMR,
+
+  ARM_BUILTIN_WADDBHUSM,
+  ARM_BUILTIN_WADDBHUSL,
+
+  ARM_BUILTIN_WQMIABB,
+  ARM_BUILTIN_WQMIABT,
+  ARM_BUILTIN_WQMIATB,
+  ARM_BUILTIN_WQMIATT,
+
+  ARM_BUILTIN_WQMIABBN,
+  ARM_BUILTIN_WQMIABTN,
+  ARM_BUILTIN_WQMIATBN,
+  ARM_BUILTIN_WQMIATTN,
+
+  ARM_BUILTIN_WMIABB,
+  ARM_BUILTIN_WMIABT,
+  ARM_BUILTIN_WMIATB,
+  ARM_BUILTIN_WMIATT,
+
+  ARM_BUILTIN_WMIABBN,
+  ARM_BUILTIN_WMIABTN,
+  ARM_BUILTIN_WMIATBN,
+  ARM_BUILTIN_WMIATTN,
+
+  ARM_BUILTIN_WMIAWBB,
+  ARM_BUILTIN_WMIAWBT,
+  ARM_BUILTIN_WMIAWTB,
+  ARM_BUILTIN_WMIAWTT,
+
+  ARM_BUILTIN_WMIAWBBN,
+  ARM_BUILTIN_WMIAWBTN,
+  ARM_BUILTIN_WMIAWTBN,
+  ARM_BUILTIN_WMIAWTTN,
+
+  ARM_BUILTIN_WMERGE,
+
+  ARM_BUILTIN_TANDCB,
+  ARM_BUILTIN_TANDCH,
+  ARM_BUILTIN_TANDCW,
+
+  ARM_BUILTIN_TORCB,
+  ARM_BUILTIN_TORCH,
+  ARM_BUILTIN_TORCW,
+
+  ARM_BUILTIN_TORVSCB,
+  ARM_BUILTIN_TORVSCH,
+  ARM_BUILTIN_TORVSCW,
+
+  ARM_BUILTIN_TEXTRCB,
+  ARM_BUILTIN_TEXTRCH,
+  ARM_BUILTIN_TEXTRCW,
+
   ARM_BUILTIN_THREAD_POINTER,
 
   ARM_BUILTIN_NEON_BASE,
@@ -19848,44 +19950,56 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckihb, "wunpckihb", WUNPCKIHB)
   IWMMXT_BUILTIN (iwmmxt_wunpckihh, "wunpckihh", WUNPCKIHH)
   IWMMXT_BUILTIN (iwmmxt_wunpckihw, "wunpckihw", WUNPCKIHW)
-  IWMMXT_BUILTIN (iwmmxt_wmadds, "wmadds", WMADDS)
-  IWMMXT_BUILTIN (iwmmxt_wmaddu, "wmaddu", WMADDU)
+  IWMMXT_BUILTIN (iwmmxt_waddsubhx, "waddsubhx", WADDSUBHX)
+  IWMMXT_BUILTIN (iwmmxt_wsubaddhx, "wsubaddhx", WSUBADDHX)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffb, "wabsdiffb", WABSDIFFB)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffh, "wabsdiffh", WABSDIFFH)
+  IWMMXT_BUILTIN (iwmmxt_wabsdiffw, "wabsdiffw", WABSDIFFW)
+  IWMMXT_BUILTIN (addcv4hi3, "waddhc", WADDCH)
+  IWMMXT_BUILTIN (addcv2si3, "waddwc", WADDCW)
+  IWMMXT_BUILTIN (iwmmxt_avg4, "wavg4", WAVG4)
+  IWMMXT_BUILTIN (iwmmxt_avg4r, "wavg4r", WAVG4R)
+  IWMMXT_BUILTIN (iwmmxt_wmulwsm, "wmulwsm", WMULWSM)
+  IWMMXT_BUILTIN (iwmmxt_wmulwum, "wmulwum", WMULWUM)
+  IWMMXT_BUILTIN (iwmmxt_wmulwsmr, "wmulwsmr", WMULWSMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulwumr, "wmulwumr", WMULWUMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulwl, "wmulwl", WMULWL)
+  IWMMXT_BUILTIN (iwmmxt_wmulsmr, "wmulsmr", WMULSMR)
+  IWMMXT_BUILTIN (iwmmxt_wmulumr, "wmulumr", WMULUMR)
+  IWMMXT_BUILTIN (iwmmxt_wqmulm, "wqmulm", WQMULM)
+  IWMMXT_BUILTIN (iwmmxt_wqmulmr, "wqmulmr", WQMULMR)
+  IWMMXT_BUILTIN (iwmmxt_wqmulwm, "wqmulwm", WQMULWM)
+  IWMMXT_BUILTIN (iwmmxt_wqmulwmr, "wqmulwmr", WQMULWMR)
+  IWMMXT_BUILTIN (iwmmxt_walignr0, "walignr0", WALIGNR0)
+  IWMMXT_BUILTIN (iwmmxt_walignr1, "walignr1", WALIGNR1)
+  IWMMXT_BUILTIN (iwmmxt_walignr2, "walignr2", WALIGNR2)
+  IWMMXT_BUILTIN (iwmmxt_walignr3, "walignr3", WALIGNR3)
 
 #define IWMMXT_BUILTIN2(code, builtin) \
   { FL_IWMMXT, CODE_FOR_##code, NULL, ARM_BUILTIN_##builtin, UNKNOWN, 0 },
 
+  IWMMXT_BUILTIN2 (iwmmxt_waddbhusm, WADDBHUSM)
+  IWMMXT_BUILTIN2 (iwmmxt_waddbhusl, WADDBHUSL)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhss, WPACKHSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwss, WPACKWSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdss, WPACKDSS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackhus, WPACKHUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackwus, WPACKWUS)
   IWMMXT_BUILTIN2 (iwmmxt_wpackdus, WPACKDUS)
-  IWMMXT_BUILTIN2 (ashlv4hi3_di,    WSLLH)
-  IWMMXT_BUILTIN2 (ashlv4hi3_iwmmxt, WSLLHI)
-  IWMMXT_BUILTIN2 (ashlv2si3_di,    WSLLW)
-  IWMMXT_BUILTIN2 (ashlv2si3_iwmmxt, WSLLWI)
-  IWMMXT_BUILTIN2 (ashldi3_di,      WSLLD)
-  IWMMXT_BUILTIN2 (ashldi3_iwmmxt,  WSLLDI)
-  IWMMXT_BUILTIN2 (lshrv4hi3_di,    WSRLH)
-  IWMMXT_BUILTIN2 (lshrv4hi3_iwmmxt, WSRLHI)
-  IWMMXT_BUILTIN2 (lshrv2si3_di,    WSRLW)
-  IWMMXT_BUILTIN2 (lshrv2si3_iwmmxt, WSRLWI)
-  IWMMXT_BUILTIN2 (lshrdi3_di,      WSRLD)
-  IWMMXT_BUILTIN2 (lshrdi3_iwmmxt,  WSRLDI)
-  IWMMXT_BUILTIN2 (ashrv4hi3_di,    WSRAH)
-  IWMMXT_BUILTIN2 (ashrv4hi3_iwmmxt, WSRAHI)
-  IWMMXT_BUILTIN2 (ashrv2si3_di,    WSRAW)
-  IWMMXT_BUILTIN2 (ashrv2si3_iwmmxt, WSRAWI)
-  IWMMXT_BUILTIN2 (ashrdi3_di,      WSRAD)
-  IWMMXT_BUILTIN2 (ashrdi3_iwmmxt,  WSRADI)
-  IWMMXT_BUILTIN2 (rorv4hi3_di,     WRORH)
-  IWMMXT_BUILTIN2 (rorv4hi3,        WRORHI)
-  IWMMXT_BUILTIN2 (rorv2si3_di,     WRORW)
-  IWMMXT_BUILTIN2 (rorv2si3,        WRORWI)
-  IWMMXT_BUILTIN2 (rordi3_di,       WRORD)
-  IWMMXT_BUILTIN2 (rordi3,          WRORDI)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacuz,   WMACUZ)
-  IWMMXT_BUILTIN2 (iwmmxt_wmacsz,   WMACSZ)
+  IWMMXT_BUILTIN2 (ashlv4hi3_di, WSLLH)
+  IWMMXT_BUILTIN2 (ashlv2si3_di, WSLLW)
+  IWMMXT_BUILTIN2 (ashldi3_di, WSLLD)
+  IWMMXT_BUILTIN2 (lshrv4hi3_di, WSRLH)
+  IWMMXT_BUILTIN2 (lshrv2si3_di, WSRLW)
+  IWMMXT_BUILTIN2 (lshrdi3_di, WSRLD)
+  IWMMXT_BUILTIN2 (ashrv4hi3_di, WSRAH)
+  IWMMXT_BUILTIN2 (ashrv2si3_di, WSRAW)
+  IWMMXT_BUILTIN2 (ashrdi3_di, WSRAD)
+  IWMMXT_BUILTIN2 (rorv4hi3_di, WRORH)
+  IWMMXT_BUILTIN2 (rorv2si3_di, WRORW)
+  IWMMXT_BUILTIN2 (rordi3_di, WRORD)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacuz, WMACUZ)
+  IWMMXT_BUILTIN2 (iwmmxt_wmacsz, WMACSZ)
 };
 
 static const struct builtin_description bdesc_1arg[] =
@@ -19908,6 +20022,12 @@ static const struct builtin_description 
   IWMMXT_BUILTIN (iwmmxt_wunpckelsb, "wunpckelsb", WUNPCKELSB)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsh, "wunpckelsh", WUNPCKELSH)
   IWMMXT_BUILTIN (iwmmxt_wunpckelsw, "wunpckelsw", WUNPCKELSW)
+  IWMMXT_BUILTIN (iwmmxt_wabsv8qi3, "wabsb", WABSB)
+  IWMMXT_BUILTIN (iwmmxt_wabsv4hi3, "wabsh", WABSH)
+  IWMMXT_BUILTIN (iwmmxt_wabsv2si3, "wabsw", WABSW)
+  IWMMXT_BUILTIN (tbcstv8qi, "tbcstb", TBCSTB)
+  IWMMXT_BUILTIN (tbcstv4hi, "tbcsth", TBCSTH)
+  IWMMXT_BUILTIN (tbcstv2si, "tbcstw", TBCSTW)
 };
 
 /* Set up all the iWMMXt builtins.  This is not called if
@@ -19923,9 +20043,6 @@ arm_init_iwmmxt_builtins (void)
   tree V4HI_type_node = build_vector_type_for_mode (intHI_type_node, V4HImode);
   tree V8QI_type_node = build_vector_type_for_mode (intQI_type_node, V8QImode);
 
-  tree int_ftype_int
-    = build_function_type_list (integer_type_node,
-				integer_type_node, NULL_TREE);
   tree v8qi_ftype_v8qi_v8qi_int
     = build_function_type_list (V8QI_type_node,
 				V8QI_type_node, V8QI_type_node,
@@ -19987,6 +20104,9 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v2si_v2si
     = build_function_type_list (V4HI_type_node,
 				V2SI_type_node, V2SI_type_node, NULL_TREE);
+  tree v8qi_ftype_v4hi_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V4HI_type_node, V8QI_type_node, NULL_TREE);
   tree v2si_ftype_v4hi_v4hi
     = build_function_type_list (V2SI_type_node,
 				V4HI_type_node, V4HI_type_node, NULL_TREE);
@@ -20001,12 +20121,10 @@ arm_init_iwmmxt_builtins (void)
     = build_function_type_list (V2SI_type_node,
 				V2SI_type_node, long_long_integer_type_node,
 				NULL_TREE);
-  tree void_ftype_int_int
-    = build_function_type_list (void_type_node,
-				integer_type_node, integer_type_node,
-				NULL_TREE);
   tree di_ftype_void
     = build_function_type_list (long_long_unsigned_type_node, NULL_TREE);
+  tree int_ftype_void
+    = build_function_type_list (integer_type_node, NULL_TREE);
   tree di_ftype_v8qi
     = build_function_type_list (long_long_integer_type_node,
 				V8QI_type_node, NULL_TREE);
@@ -20022,6 +20140,15 @@ arm_init_iwmmxt_builtins (void)
   tree v4hi_ftype_v8qi
     = build_function_type_list (V4HI_type_node,
 				V8QI_type_node, NULL_TREE);
+  tree v8qi_ftype_v8qi
+    = build_function_type_list (V8QI_type_node,
+	                        V8QI_type_node, NULL_TREE);
+  tree v4hi_ftype_v4hi
+    = build_function_type_list (V4HI_type_node,
+	                        V4HI_type_node, NULL_TREE);
+  tree v2si_ftype_v2si
+    = build_function_type_list (V2SI_type_node,
+	                        V2SI_type_node, NULL_TREE);
 
   tree di_ftype_di_v4hi_v4hi
     = build_function_type_list (long_long_unsigned_type_node,
@@ -20034,6 +20161,48 @@ arm_init_iwmmxt_builtins (void)
 				V4HI_type_node,V4HI_type_node,
 				NULL_TREE);
 
+  tree v2si_ftype_v2si_v4hi_v4hi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V4HI_type_node,
+                                V4HI_type_node, NULL_TREE);
+
+  tree v2si_ftype_v2si_v8qi_v8qi
+    = build_function_type_list (V2SI_type_node,
+                                V2SI_type_node, V8QI_type_node,
+                                V8QI_type_node, NULL_TREE);
+
+  tree di_ftype_di_v2si_v2si
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 V2SI_type_node, V2SI_type_node,
+                                 NULL_TREE);
+
+   tree di_ftype_di_di_int
+     = build_function_type_list (long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 long_long_unsigned_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree void_ftype_void
+     = build_function_type_list (void_type_node,
+                                 NULL_TREE);
+
+   tree void_ftype_int
+     = build_function_type_list (void_type_node,
+                                 integer_type_node, NULL_TREE);
+
+   tree v8qi_ftype_char
+     = build_function_type_list (V8QI_type_node,
+                                 signed_char_type_node, NULL_TREE);
+
+   tree v4hi_ftype_short
+     = build_function_type_list (V4HI_type_node,
+                                 short_integer_type_node, NULL_TREE);
+
+   tree v2si_ftype_int
+     = build_function_type_list (V2SI_type_node,
+                                 integer_type_node, NULL_TREE);
+
   /* Normal vector binops.  */
   tree v8qi_ftype_v8qi_v8qi
     = build_function_type_list (V8QI_type_node,
@@ -20092,8 +20261,14 @@ arm_init_iwmmxt_builtins (void)
 		ARM_BUILTIN_ ## CODE)
 
   iwmmx_mbuiltin ("wzero", di_ftype_void, WZERO);
-  iwmmx_mbuiltin ("setwcx", void_ftype_int_int, SETWCX);
-  iwmmx_mbuiltin ("getwcx", int_ftype_int, GETWCX);
+  iwmmx_mbuiltin ("setwcgr0", void_ftype_int, SETWCGR0);
+  iwmmx_mbuiltin ("setwcgr1", void_ftype_int, SETWCGR1);
+  iwmmx_mbuiltin ("setwcgr2", void_ftype_int, SETWCGR2);
+  iwmmx_mbuiltin ("setwcgr3", void_ftype_int, SETWCGR3);
+  iwmmx_mbuiltin ("getwcgr0", int_ftype_void, GETWCGR0);
+  iwmmx_mbuiltin ("getwcgr1", int_ftype_void, GETWCGR1);
+  iwmmx_mbuiltin ("getwcgr2", int_ftype_void, GETWCGR2);
+  iwmmx_mbuiltin ("getwcgr3", int_ftype_void, GETWCGR3);
 
   iwmmx_mbuiltin ("wsllh", v4hi_ftype_v4hi_di, WSLLH);
   iwmmx_mbuiltin ("wsllw", v2si_ftype_v2si_di, WSLLW);
@@ -20125,8 +20300,14 @@ arm_init_iwmmxt_builtins (void)
 
   iwmmx_mbuiltin ("wshufh", v4hi_ftype_v4hi_int, WSHUFH);
 
-  iwmmx_mbuiltin ("wsadb", v2si_ftype_v8qi_v8qi, WSADB);
-  iwmmx_mbuiltin ("wsadh", v2si_ftype_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wsadb", v2si_ftype_v2si_v8qi_v8qi, WSADB);
+  iwmmx_mbuiltin ("wsadh", v2si_ftype_v2si_v4hi_v4hi, WSADH);
+  iwmmx_mbuiltin ("wmadds", v2si_ftype_v4hi_v4hi, WMADDS);
+  iwmmx_mbuiltin ("wmaddsx", v2si_ftype_v4hi_v4hi, WMADDSX);
+  iwmmx_mbuiltin ("wmaddsn", v2si_ftype_v4hi_v4hi, WMADDSN);
+  iwmmx_mbuiltin ("wmaddu", v2si_ftype_v4hi_v4hi, WMADDU);
+  iwmmx_mbuiltin ("wmaddux", v2si_ftype_v4hi_v4hi, WMADDUX);
+  iwmmx_mbuiltin ("wmaddun", v2si_ftype_v4hi_v4hi, WMADDUN);
   iwmmx_mbuiltin ("wsadbz", v2si_ftype_v8qi_v8qi, WSADBZ);
   iwmmx_mbuiltin ("wsadhz", v2si_ftype_v4hi_v4hi, WSADHZ);
 
@@ -20148,6 +20329,9 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmovmskh", int_ftype_v4hi, TMOVMSKH);
   iwmmx_mbuiltin ("tmovmskw", int_ftype_v2si, TMOVMSKW);
 
+  iwmmx_mbuiltin ("waddbhusm", v8qi_ftype_v4hi_v8qi, WADDBHUSM);
+  iwmmx_mbuiltin ("waddbhusl", v8qi_ftype_v4hi_v8qi, WADDBHUSL);
+
   iwmmx_mbuiltin ("wpackhss", v8qi_ftype_v4hi_v4hi, WPACKHSS);
   iwmmx_mbuiltin ("wpackhus", v8qi_ftype_v4hi_v4hi, WPACKHUS);
   iwmmx_mbuiltin ("wpackwus", v4hi_ftype_v2si_v2si, WPACKWUS);
@@ -20173,7 +20357,7 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("wmacu", di_ftype_di_v4hi_v4hi, WMACU);
   iwmmx_mbuiltin ("wmacuz", di_ftype_v4hi_v4hi, WMACUZ);
 
-  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGN);
+  iwmmx_mbuiltin ("walign", v8qi_ftype_v8qi_v8qi_int, WALIGNI);
   iwmmx_mbuiltin ("tmia", di_ftype_di_int_int, TMIA);
   iwmmx_mbuiltin ("tmiaph", di_ftype_di_int_int, TMIAPH);
   iwmmx_mbuiltin ("tmiabb", di_ftype_di_int_int, TMIABB);
@@ -20181,6 +20365,62 @@ arm_init_iwmmxt_builtins (void)
   iwmmx_mbuiltin ("tmiatb", di_ftype_di_int_int, TMIATB);
   iwmmx_mbuiltin ("tmiatt", di_ftype_di_int_int, TMIATT);
 
+  iwmmx_mbuiltin ("wabsb", v8qi_ftype_v8qi, WABSB);
+  iwmmx_mbuiltin ("wabsh", v4hi_ftype_v4hi, WABSH);
+  iwmmx_mbuiltin ("wabsw", v2si_ftype_v2si, WABSW);
+
+  iwmmx_mbuiltin ("wqmiabb", v2si_ftype_v2si_v4hi_v4hi, WQMIABB);
+  iwmmx_mbuiltin ("wqmiabt", v2si_ftype_v2si_v4hi_v4hi, WQMIABT);
+  iwmmx_mbuiltin ("wqmiatb", v2si_ftype_v2si_v4hi_v4hi, WQMIATB);
+  iwmmx_mbuiltin ("wqmiatt", v2si_ftype_v2si_v4hi_v4hi, WQMIATT);
+
+  iwmmx_mbuiltin ("wqmiabbn", v2si_ftype_v2si_v4hi_v4hi, WQMIABBN);
+  iwmmx_mbuiltin ("wqmiabtn", v2si_ftype_v2si_v4hi_v4hi, WQMIABTN);
+  iwmmx_mbuiltin ("wqmiatbn", v2si_ftype_v2si_v4hi_v4hi, WQMIATBN);
+  iwmmx_mbuiltin ("wqmiattn", v2si_ftype_v2si_v4hi_v4hi, WQMIATTN);
+
+  iwmmx_mbuiltin ("wmiabb", di_ftype_di_v4hi_v4hi, WMIABB);
+  iwmmx_mbuiltin ("wmiabt", di_ftype_di_v4hi_v4hi, WMIABT);
+  iwmmx_mbuiltin ("wmiatb", di_ftype_di_v4hi_v4hi, WMIATB);
+  iwmmx_mbuiltin ("wmiatt", di_ftype_di_v4hi_v4hi, WMIATT);
+
+  iwmmx_mbuiltin ("wmiabbn", di_ftype_di_v4hi_v4hi, WMIABBN);
+  iwmmx_mbuiltin ("wmiabtn", di_ftype_di_v4hi_v4hi, WMIABTN);
+  iwmmx_mbuiltin ("wmiatbn", di_ftype_di_v4hi_v4hi, WMIATBN);
+  iwmmx_mbuiltin ("wmiattn", di_ftype_di_v4hi_v4hi, WMIATTN);
+
+  iwmmx_mbuiltin ("wmiawbb", di_ftype_di_v2si_v2si, WMIAWBB);
+  iwmmx_mbuiltin ("wmiawbt", di_ftype_di_v2si_v2si, WMIAWBT);
+  iwmmx_mbuiltin ("wmiawtb", di_ftype_di_v2si_v2si, WMIAWTB);
+  iwmmx_mbuiltin ("wmiawtt", di_ftype_di_v2si_v2si, WMIAWTT);
+
+  iwmmx_mbuiltin ("wmiawbbn", di_ftype_di_v2si_v2si, WMIAWBBN);
+  iwmmx_mbuiltin ("wmiawbtn", di_ftype_di_v2si_v2si, WMIAWBTN);
+  iwmmx_mbuiltin ("wmiawtbn", di_ftype_di_v2si_v2si, WMIAWTBN);
+  iwmmx_mbuiltin ("wmiawttn", di_ftype_di_v2si_v2si, WMIAWTTN);
+
+  iwmmx_mbuiltin ("wmerge", di_ftype_di_di_int, WMERGE);
+
+  iwmmx_mbuiltin ("tandcb", void_ftype_void, TANDCB);
+  iwmmx_mbuiltin ("tandch", void_ftype_void, TANDCH);
+  iwmmx_mbuiltin ("tandcw", void_ftype_void, TANDCW);
+
+  iwmmx_mbuiltin ("torcb", void_ftype_void, TORCB);
+  iwmmx_mbuiltin ("torch", void_ftype_void, TORCH);
+  iwmmx_mbuiltin ("torcw", void_ftype_void, TORCW);
+
+  iwmmx_mbuiltin ("torvscb", void_ftype_void, TORVSCB);
+  iwmmx_mbuiltin ("torvsch", void_ftype_void, TORVSCH);
+  iwmmx_mbuiltin ("torvscw", void_ftype_void, TORVSCW);
+
+  iwmmx_mbuiltin ("textrcb", void_ftype_int, TEXTRCB);
+  iwmmx_mbuiltin ("textrch", void_ftype_int, TEXTRCH);
+  iwmmx_mbuiltin ("textrcw", void_ftype_int, TEXTRCW);
+
+  iwmmx_mbuiltin ("tbcstb", v8qi_ftype_char, TBCSTB);
+  iwmmx_mbuiltin ("tbcsth", v4hi_ftype_short, TBCSTH);
+  iwmmx_mbuiltin ("tbcstw", v2si_ftype_int, TBCSTW);
+
 #undef iwmmx_mbuiltin
 }
 
@@ -20787,6 +21027,10 @@ arm_expand_builtin (tree exp,
   enum machine_mode mode0;
   enum machine_mode mode1;
   enum machine_mode mode2;
+  int opint;
+  int selector;
+  int mask;
+  int imm;
 
   if (fcode >= ARM_BUILTIN_NEON_BASE)
     return arm_expand_neon_builtin (fcode, exp, target);
@@ -20821,6 +21065,24 @@ arm_expand_builtin (tree exp,
 	  error ("selector must be an immediate");
 	  return gen_reg_rtx (tmode);
 	}
+
+      opint = INTVAL (op1);
+      if (fcode == ARM_BUILTIN_TEXTRMSB || fcode == ARM_BUILTIN_TEXTRMUB)
+	{
+	  if (opint > 7 || opint < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else if (fcode == ARM_BUILTIN_TEXTRMSH || fcode == ARM_BUILTIN_TEXTRMUH)
+	{
+	  if (opint > 3 || opint < 0)
+	    error ("the range of selector should be in 0 to 3");
+	}
+      else /* ARM_BUILTIN_TEXTRMSW || ARM_BUILTIN_TEXTRMUW.  */
+	{
+	  if (opint > 1 || opint < 0)
+	    error ("the range of selector should be in 0 to 1");
+	}
+
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20831,11 +21093,61 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
+    case ARM_BUILTIN_WALIGNI:
+      /* If op2 is immediate, call walighi, else call walighr.  */
+      arg0 = CALL_EXPR_ARG (exp, 0);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      arg2 = CALL_EXPR_ARG (exp, 2);
+      op0 = expand_normal (arg0);
+      op1 = expand_normal (arg1);
+      op2 = expand_normal (arg2);
+      if (GET_CODE (op2) == CONST_INT)
+        {
+	  icode = CODE_FOR_iwmmxt_waligni;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          gcc_assert ((*insn_data[icode].operand[3].predicate) (op2, mode2));
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      else
+        {
+	  icode = CODE_FOR_iwmmxt_walignr;
+          tmode = insn_data[icode].operand[0].mode;
+	  mode0 = insn_data[icode].operand[1].mode;
+	  mode1 = insn_data[icode].operand[2].mode;
+	  mode2 = insn_data[icode].operand[3].mode;
+          if (!(*insn_data[icode].operand[1].predicate) (op0, mode0))
+	    op0 = copy_to_mode_reg (mode0, op0);
+          if (!(*insn_data[icode].operand[2].predicate) (op1, mode1))
+	    op1 = copy_to_mode_reg (mode1, op1);
+          if (!(*insn_data[icode].operand[3].predicate) (op2, mode2))
+	    op2 = copy_to_mode_reg (mode2, op2);
+	}
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+	target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target, op0, op1, op2);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
+      return target;
+
     case ARM_BUILTIN_TINSRB:
     case ARM_BUILTIN_TINSRH:
     case ARM_BUILTIN_TINSRW:
+    case ARM_BUILTIN_WMERGE:
       icode = (fcode == ARM_BUILTIN_TINSRB ? CODE_FOR_iwmmxt_tinsrb
 	       : fcode == ARM_BUILTIN_TINSRH ? CODE_FOR_iwmmxt_tinsrh
+	       : fcode == ARM_BUILTIN_WMERGE ? CODE_FOR_iwmmxt_wmerge
 	       : CODE_FOR_iwmmxt_tinsrw);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
@@ -20858,6 +21170,27 @@ arm_expand_builtin (tree exp,
 	  error ("selector must be an immediate");
 	  return const0_rtx;
 	}
+      if (icode == CODE_FOR_iwmmxt_wmerge)
+	{
+	  selector = INTVAL (op2);
+	  if (selector > 7 || selector < 0)
+	    error ("the range of selector should be in 0 to 7");
+	}
+      if ((icode == CODE_FOR_iwmmxt_tinsrb)
+	  || (icode == CODE_FOR_iwmmxt_tinsrh)
+	  || (icode == CODE_FOR_iwmmxt_tinsrw))
+        {
+	  mask=0x01;
+	  selector= INTVAL (op2);
+	  if (icode == CODE_FOR_iwmmxt_tinsrb && (selector < 0 || selector > 7))
+	    error ("the range of selector should be in 0 to 7");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrh && (selector < 0 ||selector > 3))
+	    error ("the range of selector should be in 0 to 3");
+	  else if (icode == CODE_FOR_iwmmxt_tinsrw && (selector < 0 ||selector > 1))
+	    error ("the range of selector should be in 0 to 1");
+	  mask <<= selector;
+	  op2 = gen_rtx_CONST_INT (SImode, mask);
+	}
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20868,19 +21201,42 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_SETWCX:
+    case ARM_BUILTIN_SETWCGR0:
+    case ARM_BUILTIN_SETWCGR1:
+    case ARM_BUILTIN_SETWCGR2:
+    case ARM_BUILTIN_SETWCGR3:
+      icode = (fcode == ARM_BUILTIN_SETWCGR0 ? CODE_FOR_iwmmxt_setwcgr0
+	       : fcode == ARM_BUILTIN_SETWCGR1 ? CODE_FOR_iwmmxt_setwcgr1
+	       : fcode == ARM_BUILTIN_SETWCGR2 ? CODE_FOR_iwmmxt_setwcgr2
+	       : CODE_FOR_iwmmxt_setwcgr3);
       arg0 = CALL_EXPR_ARG (exp, 0);
-      arg1 = CALL_EXPR_ARG (exp, 1);
-      op0 = force_reg (SImode, expand_normal (arg0));
-      op1 = expand_normal (arg1);
-      emit_insn (gen_iwmmxt_tmcr (op1, op0));
+      op0 = expand_normal (arg0);
+      mode0 = insn_data[icode].operand[0].mode;
+      if (!(*insn_data[icode].operand[0].predicate) (op0, mode0))
+        op0 = copy_to_mode_reg (mode0, op0);
+      pat = GEN_FCN (icode) (op0);
+      if (!pat)
+	return 0;
+      emit_insn (pat);
       return 0;
 
-    case ARM_BUILTIN_GETWCX:
-      arg0 = CALL_EXPR_ARG (exp, 0);
-      op0 = expand_normal (arg0);
-      target = gen_reg_rtx (SImode);
-      emit_insn (gen_iwmmxt_tmrc (target, op0));
+    case ARM_BUILTIN_GETWCGR0:
+    case ARM_BUILTIN_GETWCGR1:
+    case ARM_BUILTIN_GETWCGR2:
+    case ARM_BUILTIN_GETWCGR3:
+      icode = (fcode == ARM_BUILTIN_GETWCGR0 ? CODE_FOR_iwmmxt_getwcgr0
+	       : fcode == ARM_BUILTIN_GETWCGR1 ? CODE_FOR_iwmmxt_getwcgr1
+	       : fcode == ARM_BUILTIN_GETWCGR2 ? CODE_FOR_iwmmxt_getwcgr2
+	       : CODE_FOR_iwmmxt_getwcgr3);
+      tmode = insn_data[icode].operand[0].mode;
+      if (target == 0
+	  || GET_MODE (target) != tmode
+	  || !(*insn_data[icode].operand[0].predicate) (target, tmode))
+        target = gen_reg_rtx (tmode);
+      pat = GEN_FCN (icode) (target);
+      if (!pat)
+        return 0;
+      emit_insn (pat);
       return target;
 
     case ARM_BUILTIN_WSHUFH:
@@ -20901,6 +21257,9 @@ arm_expand_builtin (tree exp,
 	  error ("mask must be an immediate");
 	  return const0_rtx;
 	}
+      selector = INTVAL (op1);
+      if (selector < 0 || selector > 255)
+	error ("the range of mask should be in 0 to 255");
       if (target == 0
 	  || GET_MODE (target) != tmode
 	  || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
@@ -20911,10 +21270,18 @@ arm_expand_builtin (tree exp,
       emit_insn (pat);
       return target;
 
-    case ARM_BUILTIN_WSADB:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadb, exp, target);
-    case ARM_BUILTIN_WSADH:
-      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadh, exp, target);
+    case ARM_BUILTIN_WMADDS:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmadds, exp, target);
+    case ARM_BUILTIN_WMADDSX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsx, exp, target);
+    case ARM_BUILTIN_WMADDSN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddsn, exp, target);
+    case ARM_BUILTIN_WMADDU:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddu, exp, target);
+    case ARM_BUILTIN_WMADDUX:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddux, exp, target);
+    case ARM_BUILTIN_WMADDUN:
+      return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wmaddun, exp, target);
     case ARM_BUILTIN_WSADBZ:
       return arm_expand_binop_builtin (CODE_FOR_iwmmxt_wsadbz, exp, target);
     case ARM_BUILTIN_WSADHZ:
@@ -20923,13 +21290,38 @@ arm_expand_builtin (tree exp,
       /* Several three-argument builtins.  */
     case ARM_BUILTIN_WMACS:
     case ARM_BUILTIN_WMACU:
-    case ARM_BUILTIN_WALIGN:
     case ARM_BUILTIN_TMIA:
     case ARM_BUILTIN_TMIAPH:
     case ARM_BUILTIN_TMIATT:
     case ARM_BUILTIN_TMIATB:
     case ARM_BUILTIN_TMIABT:
     case ARM_BUILTIN_TMIABB:
+    case ARM_BUILTIN_WQMIABB:
+    case ARM_BUILTIN_WQMIABT:
+    case ARM_BUILTIN_WQMIATB:
+    case ARM_BUILTIN_WQMIATT:
+    case ARM_BUILTIN_WQMIABBN:
+    case ARM_BUILTIN_WQMIABTN:
+    case ARM_BUILTIN_WQMIATBN:
+    case ARM_BUILTIN_WQMIATTN:
+    case ARM_BUILTIN_WMIABB:
+    case ARM_BUILTIN_WMIABT:
+    case ARM_BUILTIN_WMIATB:
+    case ARM_BUILTIN_WMIATT:
+    case ARM_BUILTIN_WMIABBN:
+    case ARM_BUILTIN_WMIABTN:
+    case ARM_BUILTIN_WMIATBN:
+    case ARM_BUILTIN_WMIATTN:
+    case ARM_BUILTIN_WMIAWBB:
+    case ARM_BUILTIN_WMIAWBT:
+    case ARM_BUILTIN_WMIAWTB:
+    case ARM_BUILTIN_WMIAWTT:
+    case ARM_BUILTIN_WMIAWBBN:
+    case ARM_BUILTIN_WMIAWBTN:
+    case ARM_BUILTIN_WMIAWTBN:
+    case ARM_BUILTIN_WMIAWTTN:
+    case ARM_BUILTIN_WSADB:
+    case ARM_BUILTIN_WSADH:
       icode = (fcode == ARM_BUILTIN_WMACS ? CODE_FOR_iwmmxt_wmacs
 	       : fcode == ARM_BUILTIN_WMACU ? CODE_FOR_iwmmxt_wmacu
 	       : fcode == ARM_BUILTIN_TMIA ? CODE_FOR_iwmmxt_tmia
@@ -20938,7 +21330,32 @@ arm_expand_builtin (tree exp,
 	       : fcode == ARM_BUILTIN_TMIABT ? CODE_FOR_iwmmxt_tmiabt
 	       : fcode == ARM_BUILTIN_TMIATB ? CODE_FOR_iwmmxt_tmiatb
 	       : fcode == ARM_BUILTIN_TMIATT ? CODE_FOR_iwmmxt_tmiatt
-	       : CODE_FOR_iwmmxt_walign);
+	       : fcode == ARM_BUILTIN_WQMIABB ? CODE_FOR_iwmmxt_wqmiabb
+	       : fcode == ARM_BUILTIN_WQMIABT ? CODE_FOR_iwmmxt_wqmiabt
+	       : fcode == ARM_BUILTIN_WQMIATB ? CODE_FOR_iwmmxt_wqmiatb
+	       : fcode == ARM_BUILTIN_WQMIATT ? CODE_FOR_iwmmxt_wqmiatt
+	       : fcode == ARM_BUILTIN_WQMIABBN ? CODE_FOR_iwmmxt_wqmiabbn
+	       : fcode == ARM_BUILTIN_WQMIABTN ? CODE_FOR_iwmmxt_wqmiabtn
+	       : fcode == ARM_BUILTIN_WQMIATBN ? CODE_FOR_iwmmxt_wqmiatbn
+	       : fcode == ARM_BUILTIN_WQMIATTN ? CODE_FOR_iwmmxt_wqmiattn
+	       : fcode == ARM_BUILTIN_WMIABB ? CODE_FOR_iwmmxt_wmiabb
+	       : fcode == ARM_BUILTIN_WMIABT ? CODE_FOR_iwmmxt_wmiabt
+	       : fcode == ARM_BUILTIN_WMIATB ? CODE_FOR_iwmmxt_wmiatb
+	       : fcode == ARM_BUILTIN_WMIATT ? CODE_FOR_iwmmxt_wmiatt
+	       : fcode == ARM_BUILTIN_WMIABBN ? CODE_FOR_iwmmxt_wmiabbn
+	       : fcode == ARM_BUILTIN_WMIABTN ? CODE_FOR_iwmmxt_wmiabtn
+	       : fcode == ARM_BUILTIN_WMIATBN ? CODE_FOR_iwmmxt_wmiatbn
+	       : fcode == ARM_BUILTIN_WMIATTN ? CODE_FOR_iwmmxt_wmiattn
+	       : fcode == ARM_BUILTIN_WMIAWBB ? CODE_FOR_iwmmxt_wmiawbb
+	       : fcode == ARM_BUILTIN_WMIAWBT ? CODE_FOR_iwmmxt_wmiawbt
+	       : fcode == ARM_BUILTIN_WMIAWTB ? CODE_FOR_iwmmxt_wmiawtb
+	       : fcode == ARM_BUILTIN_WMIAWTT ? CODE_FOR_iwmmxt_wmiawtt
+	       : fcode == ARM_BUILTIN_WMIAWBBN ? CODE_FOR_iwmmxt_wmiawbbn
+	       : fcode == ARM_BUILTIN_WMIAWBTN ? CODE_FOR_iwmmxt_wmiawbtn
+	       : fcode == ARM_BUILTIN_WMIAWTBN ? CODE_FOR_iwmmxt_wmiawtbn
+	       : fcode == ARM_BUILTIN_WMIAWTTN ? CODE_FOR_iwmmxt_wmiawttn
+	       : fcode == ARM_BUILTIN_WSADB ? CODE_FOR_iwmmxt_wsadb
+	       : CODE_FOR_iwmmxt_wsadh);
       arg0 = CALL_EXPR_ARG (exp, 0);
       arg1 = CALL_EXPR_ARG (exp, 1);
       arg2 = CALL_EXPR_ARG (exp, 2);
@@ -20971,9 +21388,196 @@ arm_expand_builtin (tree exp,
       emit_insn (gen_iwmmxt_clrdi (target));
       return target;
 
+    case ARM_BUILTIN_TANDCB:
+    case ARM_BUILTIN_TANDCH:
+    case ARM_BUILTIN_TANDCW:
+    case ARM_BUILTIN_TORCB:
+    case ARM_BUILTIN_TORCH:
+    case ARM_BUILTIN_TORCW:
+    case ARM_BUILTIN_TORVSCB:
+    case ARM_BUILTIN_TORVSCH:
+    case ARM_BUILTIN_TORVSCW:
+      icode = (fcode == ARM_BUILTIN_TANDCB ? CODE_FOR_iwmmxt_tandcv8qi3
+	       : fcode == ARM_BUILTIN_TANDCH ? CODE_FOR_iwmmxt_tandcv4hi3
+	       : fcode == ARM_BUILTIN_TANDCW ? CODE_FOR_iwmmxt_tandcv2si3
+	       : fcode == ARM_BUILTIN_TORCB ? CODE_FOR_iwmmxt_torcv8qi3
+	       : fcode == ARM_BUILTIN_TORCH ? CODE_FOR_iwmmxt_torcv4hi3
+	       : fcode == ARM_BUILTIN_TORCW ? CODE_FOR_iwmmxt_torcv2si3
+	       : fcode == ARM_BUILTIN_TORVSCB ? CODE_FOR_iwmmxt_torvscv8qi3
+	       : fcode == ARM_BUILTIN_TORVSCH ? CODE_FOR_iwmmxt_torvscv4hi3
+	       : fcode == ARM_BUILTIN_TORVSCW ? CODE_FOR_iwmmxt_torvscv2si3
+	       : CODE_FOR_iwmmxt_tandcv2si3);
+      pat = GEN_FCN (icode) (0);
+      if (!pat)
+        return 0;
+      emit_insn (pat);
+      return target;
+
+    case ARM_BUILTIN_TEXTRCB:
+    case ARM_BUILTIN_TEXTRCH:
+    case ARM_BUILTIN_TEXTRCW:
+      icode = (fcode == ARM_BUILTIN_TEXTRCB ? CODE_FOR_iwmmxt_textrcv8qi3
+ 	       : fcode == ARM_BUILTIN_TEXTRCH ? CODE_FOR_iwmmxt_textrcv4hi3
+	       : fcode == ARM_BUILTIN_TEXTRCW ? CODE_FOR_iwmmxt_textrcv2si3
+	       : CODE_FOR_iwmmxt_textrcv2si3);
+      arg0 = CALL_EXPR_ARG (exp, 0);
+      op0 = expand_normal (arg0);
+      mode0 = insn_data[icode].operand[0].mode;
+      if (!(*insn_data[icode].operand[0].predicate) (op0, mode0))
+	{
+	  /* Need a better error message here.  */
+	  error ("selector should be an immediate");
+	  return const0_rtx;
+	}
+      pat = GEN_FCN (icode) (op0);
+      if (!pat)
+      return 0;
+      emit_insn (pat);
+      return 0;
     case ARM_BUILTIN_THREAD_POINTER:
       return arm_load_tp (target);
 
+    case ARM_BUILTIN_WSRLHI:
+    case ARM_BUILTIN_WSRLWI:
+    case ARM_BUILTIN_WSRLDI:
+    case ARM_BUILTIN_WSLLHI:
+    case ARM_BUILTIN_WSLLWI:
+    case ARM_BUILTIN_WSLLDI:
+    case ARM_BUILTIN_WSRAHI:
+    case ARM_BUILTIN_WSRAWI:
+    case ARM_BUILTIN_WSRADI:
+    case ARM_BUILTIN_WRORHI:
+    case ARM_BUILTIN_WRORWI:
+    case ARM_BUILTIN_WRORDI:
+    case ARM_BUILTIN_WSRLH:
+    case ARM_BUILTIN_WSRLW:
+    case ARM_BUILTIN_WSRLD:
+    case ARM_BUILTIN_WSLLH:
+    case ARM_BUILTIN_WSLLW:
+    case ARM_BUILTIN_WSLLD:
+    case ARM_BUILTIN_WSRAH:
+    case ARM_BUILTIN_WSRAW:
+    case ARM_BUILTIN_WSRAD:
+    case ARM_BUILTIN_WRORH:
+    case ARM_BUILTIN_WRORW:
+    case ARM_BUILTIN_WRORD:
+      icode = (fcode == ARM_BUILTIN_WSRLHI ? CODE_FOR_lshrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLWI ? CODE_FOR_lshrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRLDI ? CODE_FOR_lshrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSLLHI ? CODE_FOR_ashlv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLWI ? CODE_FOR_ashlv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSLLDI ? CODE_FOR_ashldi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WSRAHI ? CODE_FOR_ashrv4hi3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRAWI ? CODE_FOR_ashrv2si3_iwmmxt
+ 	       : fcode == ARM_BUILTIN_WSRADI ? CODE_FOR_ashrdi3_iwmmxt
+	       : fcode == ARM_BUILTIN_WRORHI ? CODE_FOR_rorv4hi3
+ 	       : fcode == ARM_BUILTIN_WRORWI ? CODE_FOR_rorv2si3
+ 	       : fcode == ARM_BUILTIN_WRORDI ? CODE_FOR_rordi3
+	       : fcode == ARM_BUILTIN_WSRLH  ? CODE_FOR_lshrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRLW  ? CODE_FOR_lshrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRLD  ? CODE_FOR_lshrdi3_di
+	       : fcode == ARM_BUILTIN_WSLLH  ? CODE_FOR_ashlv4hi3_di
+	       : fcode == ARM_BUILTIN_WSLLW  ? CODE_FOR_ashlv2si3_di
+	       : fcode == ARM_BUILTIN_WSLLD  ? CODE_FOR_ashldi3_di
+	       : fcode == ARM_BUILTIN_WSRAH  ? CODE_FOR_ashrv4hi3_di
+	       : fcode == ARM_BUILTIN_WSRAW  ? CODE_FOR_ashrv2si3_di
+	       : fcode == ARM_BUILTIN_WSRAD  ? CODE_FOR_ashrdi3_di
+	       : fcode == ARM_BUILTIN_WRORH  ? CODE_FOR_rorv4hi3_di
+	       : fcode == ARM_BUILTIN_WRORW  ? CODE_FOR_rorv2si3_di
+	       : fcode == ARM_BUILTIN_WRORD  ? CODE_FOR_rordi3_di
+	       : CODE_FOR_nothing);
+      arg1 = CALL_EXPR_ARG (exp, 1);
+      op1 = expand_normal (arg1);
+      if (GET_MODE (op1) == VOIDmode)
+	{
+	  imm = INTVAL (op1);
+	  if ((fcode == ARM_BUILTIN_WRORHI || fcode == ARM_BUILTIN_WRORWI
+	       || fcode == ARM_BUILTIN_WRORH || fcode == ARM_BUILTIN_WRORW)
+	      && (imm < 0 || imm > 32))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORHI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORWI)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_rori_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WRORH)
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi16 in code.");
+	      else
+		error ("the range of count should be in 0 to 32.  please check the intrinsic _mm_ror_pi32 in code.");
+	    }
+	  else if ((fcode == ARM_BUILTIN_WRORDI || fcode == ARM_BUILTIN_WRORD)
+		   && (imm < 0 || imm > 64))
+	    {
+	      if (fcode == ARM_BUILTIN_WRORDI)
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_rori_si64 in code.");
+	      else
+		error ("the range of count should be in 0 to 64.  please check the intrinsic _mm_ror_si64 in code.");
+	    }
+	  else if (imm < 0)
+	    {
+	      if (fcode == ARM_BUILTIN_WSRLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLDI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_slli_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAHI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAWI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRADI)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srai_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_srl_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_pi32 in code.");
+	      else if (fcode == ARM_BUILTIN_WSLLD)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sll_si64 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAH)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi16 in code.");
+	      else if (fcode == ARM_BUILTIN_WSRAW)
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_pi32 in code.");
+	      else
+		error ("the count should be no less than 0.  please check the intrinsic _mm_sra_si64 in code.");
+	    }
+	  icode = (fcode == ARM_BUILTIN_WSRLHI ? CODE_FOR_lshriv4hi3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRLWI ? CODE_FOR_lshriv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRLDI ? CODE_FOR_lshridi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WSLLHI ? CODE_FOR_ashliv4hi3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSLLWI ? CODE_FOR_ashliv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSLLDI ? CODE_FOR_ashlidi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WSRAHI ? CODE_FOR_ashriv4hi3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRAWI ? CODE_FOR_ashriv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRADI ? CODE_FOR_ashridi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WRORHI ? CODE_FOR_roriv4hi3
+ 		   : fcode == ARM_BUILTIN_WRORWI ? CODE_FOR_roriv2si3
+ 		   : fcode == ARM_BUILTIN_WRORDI ? CODE_FOR_roridi3
+		   : fcode == ARM_BUILTIN_WSRLH ? CODE_FOR_lshriv4hi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WSRLW ? CODE_FOR_lshriv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRLD ? CODE_FOR_lshridi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WSLLH ? CODE_FOR_ashliv4hi3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSLLW ? CODE_FOR_ashliv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSLLD ? CODE_FOR_ashlidi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WSRAH ? CODE_FOR_ashriv4hi3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRAW ? CODE_FOR_ashriv2si3_iwmmxt
+ 		   : fcode == ARM_BUILTIN_WSRAD ? CODE_FOR_ashridi3_iwmmxt
+		   : fcode == ARM_BUILTIN_WRORH ? CODE_FOR_roriv4hi3
+ 		   : fcode == ARM_BUILTIN_WRORW ? CODE_FOR_roriv2si3
+ 		   : fcode == ARM_BUILTIN_WRORD ? CODE_FOR_roridi3
+		   : CODE_FOR_nothing);
+	}
+      return arm_expand_binop_builtin (icode, exp, target);
     default:
       break;
     }

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand
  2011-10-20  8:02 ` Xinyu Qi
@ 2011-11-18 23:47   ` Ramana Radhakrishnan
  2011-11-24  7:09     ` Xinyu Qi
                       ` (4 more replies)
  0 siblings, 5 replies; 9+ messages in thread
From: Ramana Radhakrishnan @ 2011-11-18 23:47 UTC (permalink / raw)
  To: Xinyu Qi; +Cc: gcc-patches

On 20 October 2011 08:39, Xinyu Qi <xyqi@marvell.com> wrote:
> Ping
>
> http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html
>
>        * config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
>        (builtin_description bdesc_2arg): Revise built in declaration.
>        (builtin_description bdesc_1arg): Likewise.
>        (arm_init_iwmmxt_builtins): Revise built in initialization.
>        (arm_expand_builtin): Revise built in expansion.
>

This currently doesn't apply - can you take a look ?

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING:  [PATCH, ARM, iWMMXt][3/5]: built in define and expand
@ 2011-10-20  8:02 ` Xinyu Qi
  2011-11-18 23:47   ` Ramana Radhakrishnan
  0 siblings, 1 reply; 9+ messages in thread
From: Xinyu Qi @ 2011-10-20  8:02 UTC (permalink / raw)
  To: Ramana Radhakrishnan, gcc-patches

Ping

http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html

	* config/arm/arm.c (enum arm_builtins): Revise built-in fcode.
 	(builtin_description bdesc_2arg): Revise built in declaration.
 	(builtin_description bdesc_1arg): Likewise.
 	(arm_init_iwmmxt_builtins): Revise built in initialization.
 	(arm_expand_builtin): Revise built in expansion.

^ permalink raw reply	[flat|nested] 9+ messages in thread

* RE: PING:  [PATCH, ARM, iWMMXt][3/5]: built in define and expand
@ 2011-07-29  4:25 Xinyu Qi
  0 siblings, 0 replies; 9+ messages in thread
From: Xinyu Qi @ 2011-07-29  4:25 UTC (permalink / raw)
  To: gcc-patches

Ping.

http://gcc.gnu.org/ml/gcc-patches/2011-07/msg01103.html

At 2011-07-14 15:38:47,"Xinyu Qi" <xyqi@marvell.com> wrote: 
> > Hi,
> >
> > It is the third part of iWMMXt maintenance.
> >
> 
> *config/arm/arm.c (enum arm_builtins): Built-in fcode.
>  (builtin_description bdesc_2arg): Built in declare.
>  (builtin_description bdesc_1arg): Ditto.
>  (arm_init_iwmmxt_builtins): Built in initialize.
>  (arm_expand_builtin): Built in expand.
> 
> 
> Thanks,
> Xinyu

^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, other threads:[~2012-03-13  8:56 UTC | newest]

Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-09-26  5:08 PING: [PATCH, ARM, iWMMXt][3/5]: built in define and expand Xinyu Qi
     [not found] <Acw7xRAvD+dx7PPOTJq7pOnokjiE5gGK4C2gAurXkPALlovaUATBQ7Sg>
2011-10-20  8:02 ` Xinyu Qi
2011-11-18 23:47   ` Ramana Radhakrishnan
2011-11-24  7:09     ` Xinyu Qi
2011-12-22  6:55     ` Xinyu Qi
2011-12-29  6:39     ` Xinyu Qi
2012-02-03  2:08     ` Xinyu Qi
2012-03-13  8:56     ` Xinyu Qi
  -- strict thread matches above, loose matches on Subject: below --
2011-07-29  4:25 Xinyu Qi

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).