From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-294583-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 30306 invoked by alias); 16 Jun 2011 15:54:15 -0000
Received: (qmail 30295 invoked by uid 22791); 16 Jun 2011 15:54:14 -0000
X-SWARE-Spam-Status: No, hits=-1.2 required=5.0	tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,RCVD_IN_DNSWL_NONE,TW_TQ
X-Spam-Check-By: sourceware.org
Received: from mo-p00-ob.rzone.de (HELO mo-p00-ob.rzone.de) (81.169.146.162)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Thu, 16 Jun 2011 15:53:43 +0000
X-RZG-AUTH: :LXoWVUeid/7A29J/hMvvT2k715jHQaJercGObUOFkj18odoYNahU4Q==
X-RZG-CLASS-ID: mo00
Received: from [192.168.0.22]	(business-188-111-022-002.static.arcor-ip.net [188.111.22.2])	by post.strato.de (fruni mo63) (RZmta 25.18)	with ESMTPA id 206ef8n5GEkGrn ; Thu, 16 Jun 2011 17:53:34 +0200 (MEST)
Message-ID: <4DFA26FE.1000400@gjlay.de>
Date: Thu, 16 Jun 2011 16:08:00 -0000
From: Georg-Johann Lay <avr@gjlay.de>
User-Agent: Thunderbird 2.0.0.24 (X11/20100302)
MIME-Version: 1.0
To: gcc-patches@gcc.gnu.org
CC: Denis Chertykov <chertykov@gmail.com>,  Anatoly Sokolov <aesok@post.ru>, "Eric B. Weddington" <eric.weddington@atmel.com>
Subject: [Patch, AVR]: QI builtins for parity, popcount, 1<< n
Content-Type: multipart/mixed; boundary="------------070800060208080504040207"
X-IsSubscribed: yes
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
X-SW-Source: 2011-06/txt/msg01268.txt.bz2

This is a multi-part message in MIME format.
--------------070800060208080504040207
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Content-length: 1196

The recent implementation of some asm function in libgcc added
__popcountqi2 and __parityqi2. This patch makes these functions
available as __builtin_avr_popcount8 resp. __builtin_avr_parity8.

Moreover, just out of a mood, I wrote a builtin for 1<<n.
1<<n is sometimes used to set a variable SFR bit. The builtin supplies
a fast, loop-free implementation.

The nice thing is that all these can represented explicitly so that
gcc sees what's going on and can fold if it sees operation on a
compile time constant.

Johann

	* config/avr/avr.c (avr_init_builtins, avr_builtin_id,
	bdesc_1arg): Add support for __builtin_avr_parity8,
	__builtin_avr_popcount8, __builtin_avr_pow2.
	(adjust_insn_length): Filter out "ashlqi2_1" insn in length
	computation.
	* config/avr/avr.c (avr_cpu_cpp_builtins): New builtin defines for
	__BUILTIN_AVR_PARITY8, __BUILTIN_AVR_POPCOUNT8,	__BUILTIN_AVR_POW2.
	* config/avr/avr.md (parityqi2): New expander.
	(popcountqi2): New expander.
	(*parityqihi2.libgcc): New insn.
	(*popcountqi2.libgcc): New insn.
	(ashlqi2_1): New insn.
	* doc/extend.texi (AVR Built-in Functions): Document new builtins
	__builtin_avr_parity8, __builtin_avr_popcount8,
	__builtin_avr_pow2.

--------------070800060208080504040207
Content-Type: text/x-patch;
 name="builtin-pop.diff"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="builtin-pop.diff"
Content-length: 5232

Index: doc/extend.texi
===================================================================
--- doc/extend.texi	(Revision 175104)
+++ doc/extend.texi	(Arbeitskopie)
@@ -8203,6 +8203,21 @@ int __builtin_avr_fmuls (char, char)
 int __builtin_avr_fmulsu (char, unsigned char)
 @end smallexample
 
+The following built-in functions are a supplements to gcc's
+@code{__builtin_parity*} resp. @code{__builtin_popcount*} built-ins
+for 8-bit values and are implemented as library calls:
+@smallexample
+unsigned char __builtin_avr_parity8 (unsigned char)
+unsigned char __builtin_avr_popcount8 (unsigned char)
+@end smallexample
+
+@smallexample
+unsigned char __builtin_avr_pow2 (unsigned char)
+@end smallexample
+This built-in supplies a fast, loop-free implementation for the @var{N}-th
+power of two as of @code{1 << (N & 7)} that takes about seven ticks resp.
+seven instructions.
+
 In order to delay execution for a specific number of cycles, GCC
 implements
 @smallexample
Index: config/avr/avr.md
===================================================================
--- config/avr/avr.md	(Revision 175104)
+++ config/avr/avr.md	(Arbeitskopie)
@@ -3321,6 +3321,62 @@ (define_insn "delay_cycles_4"
   [(set_attr "length" "9")
    (set_attr "cc" "clobber")])
 
+;; __builtin_avr_parity8
+(define_expand "parityqi2"
+  [(set (reg:QI 24)
+        (match_operand:QI 1 "register_operand" ""))
+   (set (reg:HI 24)
+        (zero_extend:HI (parity:QI (reg:QI 24))))
+   (set (match_operand:QI 0 "register_operand" "")
+        (reg:QI 24))]
+  ""
+  "")
+
+(define_insn "*parityqihi2.libgcc"
+  [(set (reg:HI 24)
+        (zero_extend:HI (parity:QI (reg:QI 24))))]
+  ""
+  "%~call __parityqi2"
+  [(set_attr "type" "xcall")
+   (set_attr "cc" "clobber")])
+
+;; __builtin_avr_popcount8
+(define_expand "popcountqi2"
+  [(set (reg:QI 24)
+        (match_operand:QI 1 "register_operand" ""))
+   (set (reg:QI 24)
+        (popcount:QI (reg:QI 24)))
+   (set (match_operand:QI 0 "register_operand" "")
+        (reg:QI 24))]
+  ""
+  "")
+
+(define_insn "*popcountqi2.libgcc"
+  [(set (reg:QI 24)
+        (popcount:QI (reg:QI 24)))]
+  ""
+  "%~call __popcountqi2"
+  [(set_attr "type" "xcall")
+   (set_attr "cc" "clobber")])
+
+;; __builtin_avr_pow2
+;; $0 = (1 << ($1 & 7))
+(define_insn "ashlqi2_1"
+  [(set (match_operand:QI 0 "register_operand" "=&d")
+        (ashift:QI (const_int 1)
+                   (and:QI (match_operand:QI 1 "register_operand" "r")
+                           (const_int 7))))]
+  ""
+  "ldi %0, 1
+	sbrc %1, 1
+	ldi %0, 4
+	sbrc %1, 0
+	lsl %0
+	sbrc %1, 2
+	swap %0"
+  [(set_attr "length" "7")
+   (set_attr "cc" "clobber")])
+
 ;; CPU instructions
 
 ;; NOP taking 1 or 2 Ticks 
Index: config/avr/avr-c.c
===================================================================
--- config/avr/avr-c.c	(Revision 175104)
+++ config/avr/avr-c.c	(Arbeitskopie)
@@ -93,6 +93,9 @@ avr_cpu_cpp_builtins (struct cpp_reader
   cpp_define (pfile, "__BUILTIN_AVR_SLEEP");
   cpp_define (pfile, "__BUILTIN_AVR_SWAP");
   cpp_define (pfile, "__BUILTIN_AVR_DELAY_CYCLES");
+  cpp_define (pfile, "__BUILTIN_AVR_PARITY8");
+  cpp_define (pfile, "__BUILTIN_AVR_POPCOUNT8");
+  cpp_define (pfile, "__BUILTIN_AVR_POW2");
 
   if (AVR_HAVE_MUL)
     {
Index: config/avr/avr.c
===================================================================
--- config/avr/avr.c	(Revision 175104)
+++ config/avr/avr.c	(Arbeitskopie)
@@ -4740,9 +4740,11 @@ adjust_insn_length (rtx insn, int len)
 	      break;
 	    }
 	}
-      else if (GET_CODE (op[1]) == ASHIFT
-	  || GET_CODE (op[1]) == ASHIFTRT
-	  || GET_CODE (op[1]) == LSHIFTRT)
+      else if ((GET_CODE (op[1]) == ASHIFT
+                || GET_CODE (op[1]) == ASHIFTRT
+                || GET_CODE (op[1]) == LSHIFTRT)
+               /* Filter out "ashlqi2_1" */
+               && AND != GET_CODE (XEXP (op[1], 1)))
 	{
 	  rtx ops[10];
 	  ops[0] = op[0];
@@ -6614,7 +6616,10 @@ enum avr_builtin_id
     AVR_BUILTIN_FMUL,
     AVR_BUILTIN_FMULS,
     AVR_BUILTIN_FMULSU,
-    AVR_BUILTIN_DELAY_CYCLES
+    AVR_BUILTIN_DELAY_CYCLES,
+    AVR_BUILTIN_PARITY8,
+    AVR_BUILTIN_POPCOUNT8,
+    AVR_BUILTIN_POW2
   };
 
 #define DEF_BUILTIN(NAME, TYPE, CODE)                                   \
@@ -6665,6 +6670,12 @@ avr_init_builtins (void)
   DEF_BUILTIN ("__builtin_avr_swap", uchar_ftype_uchar, AVR_BUILTIN_SWAP);
   DEF_BUILTIN ("__builtin_avr_delay_cycles", void_ftype_ulong, 
                AVR_BUILTIN_DELAY_CYCLES);
+  DEF_BUILTIN ("__builtin_avr_parity8", uchar_ftype_uchar,
+               AVR_BUILTIN_PARITY8);
+  DEF_BUILTIN ("__builtin_avr_popcount8", uchar_ftype_uchar,
+               AVR_BUILTIN_POPCOUNT8);
+  DEF_BUILTIN ("__builtin_avr_pow2", uchar_ftype_uchar,
+               AVR_BUILTIN_POW2);
 
   if (AVR_HAVE_MUL)
     {
@@ -6694,6 +6705,9 @@ static const struct avr_builtin_descript
 bdesc_1arg[] =
   {
     { CODE_FOR_rotlqi3_4, "__builtin_avr_swap", AVR_BUILTIN_SWAP }
+    , { CODE_FOR_parityqi2, "__builtin_avr_parity8", AVR_BUILTIN_PARITY8 }
+    , { CODE_FOR_popcountqi2, "__builtin_avr_popcount8", AVR_BUILTIN_POPCOUNT8 }
+    , { CODE_FOR_ashlqi2_1, "__builtin_avr_pow2", AVR_BUILTIN_POW2 }
   };
 
 static const struct avr_builtin_description

--------------070800060208080504040207--