* [PATCH 0/3] rs6000: Disparage lfiwzx and similar
@ 2022-04-21 21:28 Segher Boessenkool
2022-04-21 21:28 ` [PATCH 1/3] rs6000: Add effective target has_arch_ppc64 Segher Boessenkool
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Segher Boessenkool @ 2022-04-21 21:28 UTC (permalink / raw)
To: gcc-patches; +Cc: dje.gcc, Segher Boessenkool
This series adds "?" on the "Z" for lfiwzx and similar, so that we
preferably choose some D-form storage insn, instead of the X-form insn.
The D-form insns work on GPRs only, but it is still much preferred.
Committing to trunk.
Segher
Segher Boessenkool (3):
rs6000: Add effective target has_arch_ppc64
rs6000: Disparage lfiwzx and similar
rs6000/testsuite: xfail bswap-brw.c
gcc/config/rs6000/rs6000.md | 22 +++++++++++-----------
gcc/testsuite/gcc.target/powerpc/bswap-brw.c | 2 +-
gcc/testsuite/lib/target-supports.exp | 10 ++++++++++
3 files changed, 22 insertions(+), 12 deletions(-)
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 1/3] rs6000: Add effective target has_arch_ppc64
2022-04-21 21:28 [PATCH 0/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
@ 2022-04-21 21:28 ` Segher Boessenkool
2022-04-21 21:28 ` [PATCH 2/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
2022-04-21 21:28 ` [PATCH 3/3] rs6000/testsuite: xfail bswap-brw.c Segher Boessenkool
2 siblings, 0 replies; 4+ messages in thread
From: Segher Boessenkool @ 2022-04-21 21:28 UTC (permalink / raw)
To: gcc-patches; +Cc: dje.gcc, Segher Boessenkool
This is true if we have -mpowerpc64.
2022-04-21 Segher Boessenkool <segher@kernel.crashing.org>
gcc/testsuite/
* lib/target-supports.exp (check_effective_target_has_arch_ppc64): New.
---
gcc/testsuite/lib/target-supports.exp | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-supports.exp
index 90b900956822..8bdb4a845f54 100644
--- a/gcc/testsuite/lib/target-supports.exp
+++ b/gcc/testsuite/lib/target-supports.exp
@@ -6378,6 +6378,16 @@ proc check_effective_target_has_arch_pwr10 { } {
}]
}
+proc check_effective_target_has_arch_ppc64 { } {
+ return [check_no_compiler_messages arch_ppc64 assembly {
+ #ifndef _ARCH_PPC64
+ #error does not have ppc64 support.
+ #else
+ /* "has ppc64 support" */
+ #endif
+ }]
+}
+
# Return 1 if this is a PowerPC target supporting -mcpu=power10.
# Limit this to 64-bit linux systems for now until other targets support
# power10.
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 2/3] rs6000: Disparage lfiwzx and similar
2022-04-21 21:28 [PATCH 0/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
2022-04-21 21:28 ` [PATCH 1/3] rs6000: Add effective target has_arch_ppc64 Segher Boessenkool
@ 2022-04-21 21:28 ` Segher Boessenkool
2022-04-21 21:28 ` [PATCH 3/3] rs6000/testsuite: xfail bswap-brw.c Segher Boessenkool
2 siblings, 0 replies; 4+ messages in thread
From: Segher Boessenkool @ 2022-04-21 21:28 UTC (permalink / raw)
To: gcc-patches; +Cc: dje.gcc, Segher Boessenkool
RA now chooses GEN_OR_VSX_REGS in most cases. This is great in most
cases, but we often (or always?) use {l,st}{f,xs}iwzx now, which is
problematic because the integer load and store insns can use cheaper
addressing modes. We can fix that by putting a small penalty on the
instruction alternatives for those.
2022-04-21 Segher Boessenkool <segher@kernel.crashing.org>
PR target/103197
PR target/102146
* config/rs6000/rs6000.md (zero_extendqi<mode>2 for EXTQI): Disparage
the "Z" alternatives in {l,st}{f,xs}iwzx.
(zero_extendhi<mode>2 for EXTHI): Ditto.
(zero_extendsi<mode>2 for EXTSI): Ditto.
(*movsi_internal1): Ditto.
(*mov<mode>_internal1 for QHI): Ditto.
(movsd_hardfloat): Ditto.
---
gcc/config/rs6000/rs6000.md | 22 +++++++++++-----------
1 file changed, 11 insertions(+), 11 deletions(-)
diff --git a/gcc/config/rs6000/rs6000.md b/gcc/config/rs6000/rs6000.md
index f05b8358ba0a..a39b95f7dffa 100644
--- a/gcc/config/rs6000/rs6000.md
+++ b/gcc/config/rs6000/rs6000.md
@@ -835,8 +835,8 @@ (define_mode_iterator ALTIVEC_DFORM [DF
;; complex forms. Basic data transfer is done later.
(define_insn "zero_extendqi<mode>2"
- [(set (match_operand:EXTQI 0 "gpc_reg_operand" "=r,r,^wa,^v")
- (zero_extend:EXTQI (match_operand:QI 1 "reg_or_mem_operand" "m,r,Z,v")))]
+ [(set (match_operand:EXTQI 0 "gpc_reg_operand" "=r,r,wa,^v")
+ (zero_extend:EXTQI (match_operand:QI 1 "reg_or_mem_operand" "m,r,?Z,v")))]
""
"@
lbz%U1%X1 %0,%1
@@ -889,8 +889,8 @@ (define_insn_and_split "*zero_extendqi<mode>2_dot2"
(define_insn "zero_extendhi<mode>2"
- [(set (match_operand:EXTHI 0 "gpc_reg_operand" "=r,r,^wa,^v")
- (zero_extend:EXTHI (match_operand:HI 1 "reg_or_mem_operand" "m,r,Z,v")))]
+ [(set (match_operand:EXTHI 0 "gpc_reg_operand" "=r,r,wa,^v")
+ (zero_extend:EXTHI (match_operand:HI 1 "reg_or_mem_operand" "m,r,?Z,v")))]
""
"@
lhz%U1%X1 %0,%1
@@ -944,7 +944,7 @@ (define_insn_and_split "*zero_extendhi<mode>2_dot2"
(define_insn "zero_extendsi<mode>2"
[(set (match_operand:EXTSI 0 "gpc_reg_operand" "=r,r,d,wa,wa,r,wa")
- (zero_extend:EXTSI (match_operand:SI 1 "reg_or_mem_operand" "m,r,Z,Z,r,wa,wa")))]
+ (zero_extend:EXTSI (match_operand:SI 1 "reg_or_mem_operand" "m,r,?Z,?Z,r,wa,wa")))]
""
"@
lwz%U1%X1 %0,%1
@@ -7496,7 +7496,7 @@ (define_insn "*movsi_internal1"
[(set (match_operand:SI 0 "nonimmediate_operand"
"=r, r,
r, d, v,
- m, Z, Z,
+ m, ?Z, ?Z,
r, r, r, r,
wa, wa, wa, v,
wa, v, v,
@@ -7504,7 +7504,7 @@ (define_insn "*movsi_internal1"
r, *h, *h")
(match_operand:SI 1 "input_operand"
"r, U,
- m, Z, Z,
+ m, ?Z, ?Z,
r, d, v,
I, L, eI, n,
wa, O, wM, wB,
@@ -7785,11 +7785,11 @@ (define_expand "mov<mode>"
;; MTVSRWZ MF%1 MT%1 NOP
(define_insn "*mov<mode>_internal"
[(set (match_operand:QHI 0 "nonimmediate_operand"
- "=r, r, wa, m, Z, r,
+ "=r, r, wa, m, ?Z, r,
wa, wa, wa, v, ?v, r,
wa, r, *c*l, *h")
(match_operand:QHI 1 "input_operand"
- "r, m, Z, r, wa, i,
+ "r, m, ?Z, r, wa, i,
wa, O, wM, wB, wS, wa,
r, *h, r, 0"))]
"gpc_reg_operand (operands[0], <MODE>mode)
@@ -7973,10 +7973,10 @@ (define_insn "movsf_hardfloat"
;; FMR MR MT%0 MF%1 NOP
(define_insn "movsd_hardfloat"
[(set (match_operand:SD 0 "nonimmediate_operand"
- "=!r, d, m, Z, ?d, ?r,
+ "=!r, d, m, ?Z, ?d, ?r,
f, !r, *c*l, !r, *h")
(match_operand:SD 1 "input_operand"
- "m, Z, r, wx, r, d,
+ "m, ?Z, r, wx, r, d,
f, r, r, *h, 0"))]
"(register_operand (operands[0], SDmode)
|| register_operand (operands[1], SDmode))
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 3/3] rs6000/testsuite: xfail bswap-brw.c
2022-04-21 21:28 [PATCH 0/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
2022-04-21 21:28 ` [PATCH 1/3] rs6000: Add effective target has_arch_ppc64 Segher Boessenkool
2022-04-21 21:28 ` [PATCH 2/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
@ 2022-04-21 21:28 ` Segher Boessenkool
2 siblings, 0 replies; 4+ messages in thread
From: Segher Boessenkool @ 2022-04-21 21:28 UTC (permalink / raw)
To: gcc-patches; +Cc: dje.gcc, Segher Boessenkool
This testcase does not generate anywhere near optimal code for 32-bit
code. For p10 it actually now fails this testcase, after the previous
patch. Let's xfail it.
2022-04-21 Segher Boessenkool <segher@kernel.crashing.org>
gcc/testsuite/
PR target/103197
PR target/102146
* gcc.target/powerpc/bswap-brw.c: Add xfail on scan-assembler for -m32.
---
gcc/testsuite/gcc.target/powerpc/bswap-brw.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/gcc/testsuite/gcc.target/powerpc/bswap-brw.c b/gcc/testsuite/gcc.target/powerpc/bswap-brw.c
index b3f923eea848..0ed2a7bd1d25 100644
--- a/gcc/testsuite/gcc.target/powerpc/bswap-brw.c
+++ b/gcc/testsuite/gcc.target/powerpc/bswap-brw.c
@@ -17,6 +17,6 @@ bswap_int_dbl (unsigned int a)
/* Force the value to be loaded into a vector register. */
__asm__ (" # %x0" : "+wa" (b));
- /* { dg-final { scan-assembler {\mxxbrw\M} } } */
+ /* { dg-final { scan-assembler {\mxxbrw\M} {xfail {has_arch_pwr10 && {! has_arch_ppc64}}} } } */
return (double) __builtin_bswap32 (b);
}
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2022-04-21 21:28 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-21 21:28 [PATCH 0/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
2022-04-21 21:28 ` [PATCH 1/3] rs6000: Add effective target has_arch_ppc64 Segher Boessenkool
2022-04-21 21:28 ` [PATCH 2/3] rs6000: Disparage lfiwzx and similar Segher Boessenkool
2022-04-21 21:28 ` [PATCH 3/3] rs6000/testsuite: xfail bswap-brw.c Segher Boessenkool
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).