This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[PATCH] Fix the RTL of some sparc VIS patterns.
- From: David Miller <davem at davemloft dot net>
- To: gcc-patches at gcc dot gnu dot org
- Date: Thu, 13 Oct 2011 17:17:17 -0400 (EDT)
- Subject: [PATCH] Fix the RTL of some sparc VIS patterns.
Based upon a review of the sparc VIS support by Richard Henderson.
Committed to trunk.
gcc/
* config/sparc/sparc.md (UNSPEC_FPMERGE): Delete.
(UNSPEC_MUL16AU, UNSPEC_MUL8, UNSPEC_MUL8SU, UNSPEC_MULDSU): New
unspecs.
(fpmerge_vis): Remove inaccurate comment, represent using vec_select
of a vec_concat.
(vec_interleave_lowv8qi, vec_interleave_highv8qi): New insns.
(fmul8x16_vis, fmul8x16au_vis, fmul8sux16_vis, fmuld8sux16_vis):
Reimplement as unspecs and remove inaccurate comments.
(vis3_shift_patname): New code attr.
(<vis3_shift_insn><vbits>_vis): Rename to "v<vis3_shift_patname><mode>3".
(vis3_addsub_ss_patname): New code attr.
(<vis3_addsub_ss_insn><vbits>_vis): Rename to
"<vis3_addsub_ss_patname><mode>3".
* config/sparc/sparc.c (sparc_vis_init_builtins): Update to
accommodate pattern name changes.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@179943 138bc75d-0d04-0410-961f-82ee72b054a4
---
gcc/ChangeLog | 16 +++++++++
gcc/config/sparc/sparc.c | 32 +++++++++---------
gcc/config/sparc/sparc.md | 79 ++++++++++++++++++++++++++++++++------------
3 files changed, 89 insertions(+), 38 deletions(-)
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 18c6e88..6a514e8 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -64,6 +64,22 @@
2011-10-12 David S. Miller <davem@davemloft.net>
+ * config/sparc/sparc.md (UNSPEC_FPMERGE): Delete.
+ (UNSPEC_MUL16AU, UNSPEC_MUL8, UNSPEC_MUL8SU, UNSPEC_MULDSU): New
+ unspecs.
+ (fpmerge_vis): Remove inaccurate comment, represent using vec_select
+ of a vec_concat.
+ (vec_interleave_lowv8qi, vec_interleave_highv8qi): New insns.
+ (fmul8x16_vis, fmul8x16au_vis, fmul8sux16_vis, fmuld8sux16_vis):
+ Reimplement as unspecs and remove inaccurate comments.
+ (vis3_shift_patname): New code attr.
+ (<vis3_shift_insn><vbits>_vis): Rename to "v<vis3_shift_patname><mode>3".
+ (vis3_addsub_ss_patname): New code attr.
+ (<vis3_addsub_ss_insn><vbits>_vis): Rename to
+ "<vis3_addsub_ss_patname><mode>3".
+ * config/sparc/sparc.c (sparc_vis_init_builtins): Update to
+ accommodate pattern name changes.
+
* config/sparc/sparc.h: Do not force TARGET_VIS3 and TARGET_FMAF
to zero when assembler lacks support for such instructions.
* config/sparc/sparc.c (sparc_option_override): Clear MASK_VIS3
diff --git a/gcc/config/sparc/sparc.c b/gcc/config/sparc/sparc.c
index 5ecfe95..fc448cc 100644
--- a/gcc/config/sparc/sparc.c
+++ b/gcc/config/sparc/sparc.c
@@ -9496,21 +9496,21 @@ sparc_vis_init_builtins (void)
def_builtin_const ("__builtin_vis_fchksm16", CODE_FOR_fchksm16_vis,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fsll16", CODE_FOR_fsll16_vis,
+ def_builtin_const ("__builtin_vis_fsll16", CODE_FOR_vashlv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fslas16", CODE_FOR_fslas16_vis,
+ def_builtin_const ("__builtin_vis_fslas16", CODE_FOR_vssashlv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fsrl16", CODE_FOR_fsrl16_vis,
+ def_builtin_const ("__builtin_vis_fsrl16", CODE_FOR_vlshrv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fsra16", CODE_FOR_fsra16_vis,
+ def_builtin_const ("__builtin_vis_fsra16", CODE_FOR_vashrv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fsll32", CODE_FOR_fsll32_vis,
+ def_builtin_const ("__builtin_vis_fsll32", CODE_FOR_vashlv2si3,
v2si_ftype_v2si_v2si);
- def_builtin_const ("__builtin_vis_fslas32", CODE_FOR_fslas32_vis,
+ def_builtin_const ("__builtin_vis_fslas32", CODE_FOR_vssashlv2si3,
v2si_ftype_v2si_v2si);
- def_builtin_const ("__builtin_vis_fsrl32", CODE_FOR_fsrl32_vis,
+ def_builtin_const ("__builtin_vis_fsrl32", CODE_FOR_vlshrv2si3,
v2si_ftype_v2si_v2si);
- def_builtin_const ("__builtin_vis_fsra32", CODE_FOR_fsra32_vis,
+ def_builtin_const ("__builtin_vis_fsra32", CODE_FOR_vashrv2si3,
v2si_ftype_v2si_v2si);
if (TARGET_ARCH64)
@@ -9527,21 +9527,21 @@ sparc_vis_init_builtins (void)
def_builtin_const ("__builtin_vis_fpsub64", CODE_FOR_fpsub64_vis,
di_ftype_di_di);
- def_builtin_const ("__builtin_vis_fpadds16", CODE_FOR_fpadds16_vis,
+ def_builtin_const ("__builtin_vis_fpadds16", CODE_FOR_ssaddv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fpadds16s", CODE_FOR_fpadds16s_vis,
+ def_builtin_const ("__builtin_vis_fpadds16s", CODE_FOR_ssaddv2hi3,
v2hi_ftype_v2hi_v2hi);
- def_builtin_const ("__builtin_vis_fpsubs16", CODE_FOR_fpsubs16_vis,
+ def_builtin_const ("__builtin_vis_fpsubs16", CODE_FOR_sssubv4hi3,
v4hi_ftype_v4hi_v4hi);
- def_builtin_const ("__builtin_vis_fpsubs16s", CODE_FOR_fpsubs16s_vis,
+ def_builtin_const ("__builtin_vis_fpsubs16s", CODE_FOR_sssubv2hi3,
v2hi_ftype_v2hi_v2hi);
- def_builtin_const ("__builtin_vis_fpadds32", CODE_FOR_fpadds32_vis,
+ def_builtin_const ("__builtin_vis_fpadds32", CODE_FOR_ssaddv2si3,
v2si_ftype_v2si_v2si);
- def_builtin_const ("__builtin_vis_fpadds32s", CODE_FOR_fpadds32s_vis,
+ def_builtin_const ("__builtin_vis_fpadds32s", CODE_FOR_ssaddsi3,
v1si_ftype_v1si_v1si);
- def_builtin_const ("__builtin_vis_fpsubs32", CODE_FOR_fpsubs32_vis,
+ def_builtin_const ("__builtin_vis_fpsubs32", CODE_FOR_sssubv2si3,
v2si_ftype_v2si_v2si);
- def_builtin_const ("__builtin_vis_fpsubs32s", CODE_FOR_fpsubs32s_vis,
+ def_builtin_const ("__builtin_vis_fpsubs32s", CODE_FOR_sssubsi3,
v1si_ftype_v1si_v1si);
if (TARGET_ARCH64)
diff --git a/gcc/config/sparc/sparc.md b/gcc/config/sparc/sparc.md
index 24993fb..c41e259 100644
--- a/gcc/config/sparc/sparc.md
+++ b/gcc/config/sparc/sparc.md
@@ -53,7 +53,7 @@
(UNSPEC_FPACK32 41)
(UNSPEC_FPACKFIX 42)
(UNSPEC_FEXPAND 43)
- (UNSPEC_FPMERGE 44)
+ (UNSPEC_MUL16AU 44)
(UNSPEC_MUL16AL 45)
(UNSPEC_MUL8UL 46)
(UNSPEC_MULDUL 47)
@@ -89,6 +89,9 @@
(UNSPEC_FHADD 83)
(UNSPEC_FHSUB 84)
(UNSPEC_XMUL 85)
+ (UNSPEC_MUL8 86)
+ (UNSPEC_MUL8SU 87)
+ (UNSPEC_MULDSU 88)
])
(define_constants
@@ -8004,36 +8007,64 @@
[(set_attr "type" "fga")
(set_attr "fptype" "double")])
-;; It may be possible to describe this operation as (1 indexed):
-;; (vec_select (vec_duplicate (vec_duplicate (vec_concat 1 2)))
-;; 1,5,10,14,19,23,28,32)
-;; Note that (vec_merge:V8QI [(V4QI) (V4QI)] (10101010 = 170) doesn't work
-;; because vec_merge expects all the operands to be of the same type.
(define_insn "fpmerge_vis"
[(set (match_operand:V8QI 0 "register_operand" "=e")
- (unspec:V8QI [(match_operand:V4QI 1 "register_operand" "f")
- (match_operand:V4QI 2 "register_operand" "f")]
- UNSPEC_FPMERGE))]
+ (vec_select:V8QI
+ (vec_concat:V8QI (match_operand:V4QI 1 "register_operand" "f")
+ (match_operand:V4QI 2 "register_operand" "f"))
+ (parallel [(const_int 0) (const_int 4)
+ (const_int 1) (const_int 5)
+ (const_int 2) (const_int 6)
+ (const_int 3) (const_int 7)])))]
"TARGET_VIS"
"fpmerge\t%1, %2, %0"
[(set_attr "type" "fga")
(set_attr "fptype" "double")])
+(define_insn "vec_interleave_lowv8qi"
+ [(set (match_operand:V8QI 0 "register_operand" "=e")
+ (vec_select:V8QI
+ (vec_concat:V16QI (match_operand:V8QI 1 "register_operand" "f")
+ (match_operand:V8QI 2 "register_operand" "f"))
+ (parallel [(const_int 0) (const_int 8)
+ (const_int 1) (const_int 9)
+ (const_int 2) (const_int 10)
+ (const_int 3) (const_int 11)])))]
+ "TARGET_VIS"
+ "fpmerge\t%L1, %L2, %0"
+ [(set_attr "type" "fga")
+ (set_attr "fptype" "double")])
+
+(define_insn "vec_interleave_highv8qi"
+ [(set (match_operand:V8QI 0 "register_operand" "=e")
+ (vec_select:V8QI
+ (vec_concat:V16QI (match_operand:V8QI 1 "register_operand" "f")
+ (match_operand:V8QI 2 "register_operand" "f"))
+ (parallel [(const_int 4) (const_int 12)
+ (const_int 5) (const_int 13)
+ (const_int 6) (const_int 14)
+ (const_int 7) (const_int 15)])))]
+ "TARGET_VIS"
+ "fpmerge\t%H1, %H2, %0"
+ [(set_attr "type" "fga")
+ (set_attr "fptype" "double")])
+
;; Partitioned multiply instructions
(define_insn "fmul8x16_vis"
[(set (match_operand:V4HI 0 "register_operand" "=e")
- (mult:V4HI (match_operand:V4QI 1 "register_operand" "f")
- (match_operand:V4HI 2 "register_operand" "e")))]
+ (unspec:V4HI [(match_operand:V4QI 1 "register_operand" "f")
+ (match_operand:V4HI 2 "register_operand" "e")]
+ UNSPEC_MUL8))]
"TARGET_VIS"
"fmul8x16\t%1, %2, %0"
[(set_attr "type" "fpmul")
(set_attr "fptype" "double")])
-;; Only one of the following two insns can be a multiply.
(define_insn "fmul8x16au_vis"
[(set (match_operand:V4HI 0 "register_operand" "=e")
- (mult:V4HI (match_operand:V4QI 1 "register_operand" "f")
- (match_operand:V2HI 2 "register_operand" "f")))]
+ (unspec:V4HI [(match_operand:V4QI 1 "register_operand" "f")
+ (match_operand:V2HI 2 "register_operand" "f")]
+ UNSPEC_MUL16AU))]
"TARGET_VIS"
"fmul8x16au\t%1, %2, %0"
[(set_attr "type" "fpmul")
@@ -8049,11 +8080,11 @@
[(set_attr "type" "fpmul")
(set_attr "fptype" "double")])
-;; Only one of the following two insns can be a multiply.
(define_insn "fmul8sux16_vis"
[(set (match_operand:V4HI 0 "register_operand" "=e")
- (mult:V4HI (match_operand:V8QI 1 "register_operand" "e")
- (match_operand:V4HI 2 "register_operand" "e")))]
+ (unspec:V4HI [(match_operand:V8QI 1 "register_operand" "e")
+ (match_operand:V4HI 2 "register_operand" "e")]
+ UNSPEC_MUL8SU))]
"TARGET_VIS"
"fmul8sux16\t%1, %2, %0"
[(set_attr "type" "fpmul")
@@ -8069,11 +8100,11 @@
[(set_attr "type" "fpmul")
(set_attr "fptype" "double")])
-;; Only one of the following two insns can be a multiply.
(define_insn "fmuld8sux16_vis"
[(set (match_operand:V2SI 0 "register_operand" "=e")
- (mult:V2SI (match_operand:V4QI 1 "register_operand" "f")
- (match_operand:V2HI 2 "register_operand" "f")))]
+ (unspec:V2SI [(match_operand:V4QI 1 "register_operand" "f")
+ (match_operand:V2HI 2 "register_operand" "f")]
+ UNSPEC_MULDSU))]
"TARGET_VIS"
"fmuld8sux16\t%1, %2, %0"
[(set_attr "type" "fpmul")
@@ -8440,8 +8471,10 @@
(define_code_iterator vis3_shift [ashift ss_ashift lshiftrt ashiftrt])
(define_code_attr vis3_shift_insn
[(ashift "fsll") (ss_ashift "fslas") (lshiftrt "fsrl") (ashiftrt "fsra")])
+(define_code_attr vis3_shift_patname
+ [(ashift "ashl") (ss_ashift "ssashl") (lshiftrt "lshr") (ashiftrt "ashr")])
-(define_insn "<vis3_shift_insn><vbits>_vis"
+(define_insn "v<vis3_shift_patname><mode>3"
[(set (match_operand:V64N8 0 "register_operand" "=<vconstr>")
(vis3_shift:V64N8 (match_operand:V64N8 1 "register_operand" "<vconstr>")
(match_operand:V64N8 2 "register_operand" "<vconstr>")))]
@@ -8490,8 +8523,10 @@
(define_code_iterator vis3_addsub_ss [ss_plus ss_minus])
(define_code_attr vis3_addsub_ss_insn
[(ss_plus "fpadds") (ss_minus "fpsubs")])
+(define_code_attr vis3_addsub_ss_patname
+ [(ss_plus "ssadd") (ss_minus "sssub")])
-(define_insn "<vis3_addsub_ss_insn><vbits>_vis"
+(define_insn "<vis3_addsub_ss_patname><mode>3"
[(set (match_operand:VASS 0 "register_operand" "=<vconstr>")
(vis3_addsub_ss:VASS (match_operand:VASS 1 "register_operand" "<vconstr>")
(match_operand:VASS 2 "register_operand" "<vconstr>")))]
--
1.7.6.401.g6a319