From: will schmidt <will_schmidt@vnet.ibm.com>
To: Carl Love <cel@us.ibm.com>,
segher@gcc.gnu.org, dje.gcc@gmail.com, gcc-patches@gcc.gnu.org
Subject: Re: [PATCH 5/5] Conversions between 128-bit integer and floating point values.
Date: Thu, 24 Sep 2020 13:22:02 -0500 [thread overview]
Message-ID: <e73b7f6b52c0f148f181fddc84845b04dc4cadc4.camel@vnet.ibm.com> (raw)
In-Reply-To: <467724e7f9f9d7a34626cb630622500aaf3391ae.camel@us.ibm.com>
On Mon, 2020-09-21 at 16:57 -0700, Carl Love wrote:
> Segher, Will:
>
> Patch 5 adds the 128-bit integer to/from 128-floating point
> conversions. This patch has to invoke the routines to use the 128-
> bit
> hardware instructions if on Power 10 or use software routines if
> running on a pre Power 10 system via the resolve function.
>
> Add ifunc resolves for __fixkfti, __floatuntikf_sw, __fixkfti_swn,
> __fixunskfti_sw.
>
> The following changes were made to the previous version of the
> patch:
>
> Fixed typos in ChangeLog noted by Will.
>
> Turned off debug in test case.
>
> Removed extra blank lines, fixed spacing of #else in the test case.
>
> Added comment to fixunskfti-sw.c about changes made from the original
> file fixunskfti.c.
>
> The patch has been tested on
>
> powerpc64le-unknown-linux-gnu (Power 9 LE)
>
> with no regression errors.
>
> The P10 tests were run by hand on Mambo.
>
> Carl Love
> ---------------------------------------------------------
>
> gcc/ChangeLog
>
> 2020-09-21 Carl Love <cel@us.ibm.com>
> config/rs6000/rs6000.md (floatti<mode>2, floatunsti<mode>2,
> fix_trunc<mode>ti2, fixuns_trunc<mode>ti2): Add
> define_insn for mode IEEE 128.
ok
> libgcc/config/rs6000/fixkfi-sw.c: New file.
> libgcc/config/rs6000/fixkfi.c: Remove file.
Should that be fixkfti-sw.c (missing t).
Adjust to indicate this is a rename
libgcc/config/rs6000/fixkfti.c: Rename to
libgcc/config/rs6000/fixkfti-sw.c
> libgcc/config/rs6000/fixunskfti-sw.c: New file.
> libgcc/config/rs6000/fixunskfti.c: Remove file.
> libgcc/config/rs6000/float128-hw.c (__floattikf_hw,
> __floatuntikf_hw, __fixkfti_hw, __fixunskfti_hw):
> New functions.
> libgcc/config/rs6000/float128-ifunc.c (SW_OR_HW_ISA3_1):
> New macro.
> (__floattikf_resolve, __floatuntikf_resolve, __fixkfti_resolve,
> __fixunskfti_resolve): Add resolve functions.
> (__floattikf, __floatuntikf, __fixkfti, __fixunskfti): New
> functions.
> libgcc/config/rs6000/float128-sed (floattitf, __floatuntitf,
> __fixtfti, __fixunstfti): Add editor commands to change
> names.
> libgcc/config/rs6000/float128-sed-hw (__floattitf,
> __floatuntitf, __fixtfti, __fixunstfti): Add editor commands
> to change names.
> libgcc/config/rs6000/floattikf-sw.c: New file.
> libgcc/config/rs6000/floattikf.c: Remove file.
> libgcc/config/rs6000/floatuntikf-sw.c: New file.
> libgcc/config/rs6000/floatuntikf.c: Remove file.
> libgcc/config/rs6000/floatuntikf-sw.c: New file.
> libgcc/config/rs6000/quaad-float128.h (__floattikf_sw,
> __floatuntikf_sw, __fixkfti_sw, __fixunskfti_sw,
> __floattikf_hw,
> __floatuntikf_hw, __fixkfti_hw, __fixunskfti_hw, __floattikf,
> __floatuntikf, __fixkfti, __fixunskfti): New extern
> declarations.
> libgcc/config/rs6000/t-float128 (floattikf, floatuntikf,
> fixkfti, fixunskfti): Remove file names from fp128_ppc_funcs.
> (floattikf-sw, floatuntikf-sw, fixkfti-sw, fixunskfti-sw): Add
> file names to fp128_ppc_funcs.
>
> gcc/testsuite/ChangeLog
>
> 2020-09-21 Carl Love <cel@us.ibm.com>
> gcc.target/powerpc/fl128_conversions.c: New file.
> ---
> gcc/config/rs6000/rs6000.md | 36 +++
> .../gcc.target/powerpc/fp128_conversions.c | 286
> ++++++++++++++++++
> .../config/rs6000/{fixkfti.c => fixkfti-sw.c} | 4 +-
> .../rs6000/{fixunskfti.c => fixunskfti-sw.c} | 7 +-
> libgcc/config/rs6000/float128-hw.c | 24 ++
> libgcc/config/rs6000/float128-ifunc.c | 44 ++-
> libgcc/config/rs6000/float128-sed | 4 +
> libgcc/config/rs6000/float128-sed-hw | 4 +
> .../rs6000/{floattikf.c => floattikf-sw.c} | 4 +-
> .../{floatuntikf.c => floatuntikf-sw.c} | 4 +-
> libgcc/config/rs6000/quad-float128.h | 17 +-
> libgcc/config/rs6000/t-float128 | 3 +-
> 12 files changed, 417 insertions(+), 20 deletions(-)
> create mode 100644
> gcc/testsuite/gcc.target/powerpc/fp128_conversions.c
> rename libgcc/config/rs6000/{fixkfti.c => fixkfti-sw.c} (96%)
> rename libgcc/config/rs6000/{fixunskfti.c => fixunskfti-sw.c} (90%)
> rename libgcc/config/rs6000/{floattikf.c => floattikf-sw.c} (96%)
> rename libgcc/config/rs6000/{floatuntikf.c => floatuntikf-sw.c}
> (96%)
>
> diff --git a/gcc/config/rs6000/rs6000.md
> b/gcc/config/rs6000/rs6000.md
> index 694ff70635e..5db5d0b4505 100644
> --- a/gcc/config/rs6000/rs6000.md
> +++ b/gcc/config/rs6000/rs6000.md
> @@ -6390,6 +6390,42 @@
> xscvsxddp %x0,%x1"
> [(set_attr "type" "fp")])
>
> +(define_insn "floatti<mode>2"
> + [(set (match_operand:IEEE128 0 "vsx_register_operand" "=v")
> + (float:IEEE128 (match_operand:TI 1 "vsx_register_operand"
> "v")))]
> + "TARGET_POWER10"
> +{
> + return "xscvsqqp %0,%1";
> +}
> + [(set_attr "type" "fp")])
> +
> +(define_insn "floatunsti<mode>2"
> + [(set (match_operand:IEEE128 0 "vsx_register_operand" "=v")
> + (unsigned_float:IEEE128 (match_operand:TI 1
> "vsx_register_operand" "v")))]
> + "TARGET_POWER10"
> +{
> + return "xscvuqqp %0,%1";
> +}
> + [(set_attr "type" "fp")])
> +
> +(define_insn "fix_trunc<mode>ti2"
> + [(set (match_operand:TI 0 "vsx_register_operand" "=v")
> + (fix:TI (match_operand:IEEE128 1 "vsx_register_operand"
> "v")))]
> + "TARGET_POWER10"
> +{
> + return "xscvqpsqz %0,%1";
> +}
> + [(set_attr "type" "fp")])
> +
> +(define_insn "fixuns_trunc<mode>ti2"
> + [(set (match_operand:TI 0 "vsx_register_operand" "=v")
> + (unsigned_fix:TI (match_operand:IEEE128 1
> "vsx_register_operand" "v")))]
> + "TARGET_POWER10"
> +{
> + return "xscvqpuqz %0,%1";
> +}
> + [(set_attr "type" "fp")])
> +
> ; Allow the combiner to merge source memory operands to the
> conversion so that
> ; the optimizer/register allocator doesn't try to load the value too
> early in a
> ; GPR and then use store/load to move it to a FPR and suffer from a
> store-load
> diff --git a/gcc/testsuite/gcc.target/powerpc/fp128_conversions.c
> b/gcc/testsuite/gcc.target/powerpc/fp128_conversions.c
> new file mode 100644
> index 00000000000..dff24e1e2b4
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/powerpc/fp128_conversions.c
> @@ -0,0 +1,286 @@
> +/* { dg-do run } */
> +/* { dg-require-effective-target power10_hw } */
> +/* { dg-options "-mdejagnu-cpu=power10" } */
> +
> +/* Check that the expected 128-bit instructions are generated if the
> processor
> + supports the 128-bit integer instructions. */
> +/* { dg-final { scan-assembler-times {\mxscvsqqp\M} 1 { target {
> ppc_native_128bit } } } } */
> +/* { dg-final { scan-assembler-times {\mxscvuqqp\M} 1 { target {
> ppc_native_128bit } } } } */
> +/* { dg-final { scan-assembler-times {\mxscvqpsqz\M} 1 { target {
> ppc_native_128bit } } } } */
> +/* { dg-final { scan-assembler-times {\mxscvqpuqz\M} 1 { target {
> ppc_native_128bit } } } } */
> +
> +#include <stdio.h>
> +#include <math.h>
> +#include <fenv.h>
> +#include <stdlib.h>
> +#include <wchar.h>
> +
> +#define DEBUG 0
> +
> +void abort (void);
> +
> +float conv_i_2_fp( long long int a)
> +{
> + return (float) a;
> +}
> +
> +double conv_i_2_fpd( long long int a)
> +{
> + return (double) a;
> +}
> +
> +double conv_ui_2_fpd( unsigned long long int a)
> +{
> + return (double) a;
> +}
> +
> +__float128 conv_i128_2_fp128 (__int128_t a)
> +{
> + // default, gen inst KF mode
> + // -mabi=ibmlongdouble, gen inst floattiieee KF mode
> + // -mabi=ieeelongdouble gen inst floattiieee TF mode
> + return (__float128) a;
> +}
> +
> +__float128 conv_ui128_2_fp128 (__uint128_t a)
> +{
> + // default, gen inst KF mode
> + // -mabi=ibmlongdouble, gen inst floattiieee KF mode
> + // -mabi=ieeelongdouble gen inst floattiieee TF mode
> + return (__float128) a;
> +}
> +
> +__int128_t conv_fp128_2_i128 (__float128 a)
> +{
> + // default, gen inst KF mode
> + // -mabi=ibmlongdouble, gen inst floattiieee KF mode
> + // -mabi=ieeelongdouble gen inst floattiieee TF mode
> + return (__int128_t) a;
> +}
> +
> +__uint128_t conv_fp128_2_ui128 (__float128 a)
> +{
> + // default, gen inst KF mode
> + // -mabi=ibmlongdouble, gen inst floattiieee KF mode
> + // -mabi=ieeelongdouble gen inst floattiieee TF mode
> + return (__uint128_t) a;
> +}
> +
> +long double conv_i128_2_ld (__int128_t a)
> +{
> + // default, gen call __floattitf
> + // -mabi=ibmlongdouble, gen call __floattitf
> + // -mabi=ieeelongdouble gen inst floattiieee TF mode
> + return (long double) a;
> +}
> +
> +__ibm128 conv_i128_2_ibm128 (__int128_t a)
> +{
> + // default, gen call __floattitf
> + // -mabi=ibmlongdouble, gen call __floattitf
> + // -mabi=ieeelongdouble, message uses IBM long double, no binary
> output
> + return (__ibm128) a;
> +}
> +
> +int main()
> +{
> + float a, expected_result_float;
> + double b, expected_result_double;
> + long long int c, expected_result_llint;
> + unsigned long long int u;
> + __int128_t d;
> + __uint128_t u128;
> + unsigned long long expected_result_uint128[2] ;
> + __float128 e;
> + long double ld; // another 128-bit float version
> +
> + union conv_t {
> + float a;
> + double b;
> + long long int c;
> + long long int128[2] ;
> + unsigned long long uint128[2] ;
> + unsigned long long int u;
> + __int128_t d;
> + __uint128_t u128;
> + __float128 e;
> + long double ld; // another 128-bit float version
> + } conv, conv_result;
> +
> + c = 20;
> + expected_result_llint = 20.00000;
> + a = conv_i_2_fp (c);
> +
> + if (a != expected_result_llint) {
> +#if DEBUG
> + printf("ERROR: conv_i_2_fp(%lld) = %10.5f\n", c, a);
> + printf("\n does not match expected_result =
> %10.5f\n\n",
> + expected_result_llint);
> +#else
> + abort();
> +#endif
> + }
> +
> + c = 20;
> + expected_result_double = 20.00000;
> + b = conv_i_2_fpd (c);
> +
> + if (b != expected_result_double) {
> +#if DEBUG
> + printf("ERROR: conv_i_2_fpd(%lld) = %10.5f\n", d, b);
> + printf("\n does not match expected_result =
> %10.5f\n\n",
> + expected_result_double);
> + #else
> + abort();
> +#endif
> + }
> +
> + u = 20;
> + expected_result_double = 20.00000;
> + b = conv_ui_2_fpd (u);
> +
> + if (b != expected_result_double) {
> +#if DEBUG
> + printf("ERROR: conv_ui_2_fpd(%llu) = %10.5f\n", u, b);
> + printf("\n does not match expected_result =
> %10.5f\n\n",
> + expected_result_double);
> + #else
> + abort();
> +#endif
> + }
> +
> + /* Currently printing 128-bit float does not work correctly */
> + d = -3210;
> + d = (d * 10000000000) + 9876543210;
> + conv_result.e = conv_i128_2_fp128 (d);
> + expected_result_uint128[1] = 0xc02bd2f9068d1160;
> + expected_result_uint128[0] = 0x0;
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_i128_2_fp128(-32109876543210) = (result
> in hex) 0x%llx %llx\n",
> + conv.uint128[1], conv.uint128[0]);
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + d = 123;
> + d = (d * 10000000000) + 1234567890;
> + conv_result.ld = conv_i128_2_fp128 (d);
> + expected_result_uint128[1] = 0x0;
> + expected_result_uint128[0] = 0x4271eab4c8ed2000;
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_i128_2_fp128(1231234567890) = (result in
> hex) 0x%llx %llx\n",
> + conv.uint128[1], conv.uint128[0]);
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + /* Currently printing 128-bit float does not work correctly */
> + u128 = 8760;
> + u128 = (u128 * 10000000000) + 1234567890;
> + conv_result.e = conv_ui128_2_fp128 (u128);
> + expected_result_uint128[1] = 0x402d3eb101df8b48;
> + expected_result_uint128[0] = 0x0;
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_ui128_2_fp128(87601234567890) = (result
> in hex) 0x%llx %llx\n",
> + conv.uint128[1], conv.uint128[0]);
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + /* Currently printing 128-bit float does not work correctly */
> + u128 = 3210;
> + u128 = (u128 * 10000000000) + 9876543210;
> + expected_result_uint128[1] = 0x402bd3429c8feea0;
> + expected_result_uint128[0] = 0x0;
> + conv_result.e = conv_ui128_2_fp128 (u128);
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_ui128_2_fp128(32109876543210) = (result
> in hex) 0x%llx %llx\n",
> + conv.uint128[1], conv.uint128[0]);
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + conv.e = 12345.6789;
> + expected_result_uint128[1] = 0x1407374883526960;
> + expected_result_uint128[0] = 0x3039;
> +
> + conv_result.d = conv_fp128_2_i128 (conv.e);
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_fp128_2_i128(0x%llx %llx) = ",
> + conv.uint128[1], conv.uint128[0]);
> + printf("0x%llx %llx\n", conv_result.uint128[1],
> conv_result.uint128[0]);
> +
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + conv.e = -6789.12345;
> + expected_result_uint128[1] = 0x0;
> + expected_result_uint128[0] = 0xffffffffffffe57b;
> + conv_result.d = conv_fp128_2_i128 (conv.e);
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_fp128_2_i128(0x%llx %llx) = ",
> + conv.uint128[1], conv.uint128[0]);
> + printf("0x%llx %llx\n", conv_result.uint128[1],
> conv_result.uint128[0]);
> +
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + conv.e = 6789.12345;
> + expected_result_uint128[1] = 0x0;
> + expected_result_uint128[0] = 0x1a85;
> + conv_result.d = conv_fp128_2_ui128 (conv.e);
> +
> + if ((conv_result.uint128[1] != expected_result_uint128[1])
> + && (conv_result.uint128[0] !=
> expected_result_uint128[0])) {
> +#if DEBUG
> + printf("ERROR: conv_fp128_2_ui128(0x%llx %llx) = ",
> + conv.uint128[1], conv.uint128[0]);
> + printf("0x%llx %llx\n", conv_result.uint128[1],
> conv_result.uint128[0]);
> +
> + printf("\n does not match expected_result = (result in hex)
> 0x%llx %llx\n\n",
> + expected_result_uint128[1],
> expected_result_uint128[0]);
> + #else
> + abort();
> +#endif
> + }
> +
> + return 0;
> +}
> diff --git a/libgcc/config/rs6000/fixkfti.c
> b/libgcc/config/rs6000/fixkfti-sw.c
> similarity index 96%
> rename from libgcc/config/rs6000/fixkfti.c
> rename to libgcc/config/rs6000/fixkfti-sw.c
> index a22286228aa..d6bbbf889b7 100644
> --- a/libgcc/config/rs6000/fixkfti.c
> +++ b/libgcc/config/rs6000/fixkfti-sw.c
> @@ -5,7 +5,7 @@
> This file is part of the GNU C Library.
> Contributed by Steven Munroe (munroesj@linux.vnet.ibm.com)
> Code is based on the main soft-fp library written by:
> - Uros Bizjak (ubizjak@gmail.com).
> + Uros Bizjak (ubizjak@gmail.com).
>
> The GNU C Library is free software; you can redistribute it
> and/or
> modify it under the terms of the GNU Lesser General Public
> @@ -35,7 +35,7 @@
> #include "quad-float128.h"
>
> TItype
> -__fixkfti (TFtype a)
> +__fixkfti_sw (TFtype a)
> {
> FP_DECL_EX;
> FP_DECL_Q (A);
> diff --git a/libgcc/config/rs6000/fixunskfti.c
> b/libgcc/config/rs6000/fixunskfti-sw.c
> similarity index 90%
> rename from libgcc/config/rs6000/fixunskfti.c
> rename to libgcc/config/rs6000/fixunskfti-sw.c
> index ab232d92d24..5c8b94b8862 100644
> --- a/libgcc/config/rs6000/fixunskfti.c
> +++ b/libgcc/config/rs6000/fixunskfti-sw.c
> @@ -5,7 +5,10 @@
> This file is part of the GNU C Library.
> Contributed by Steven Munroe (munroesj@linux.vnet.ibm.com)
> Code is based on the main soft-fp library written by:
> - Uros Bizjak (ubizjak@gmail.com).
> + Uros Bizjak (ubizjak@gmail.com).
> +
> + The fixunskfti.c file renamed fixunskfti-sw.c. Updated the
> + source file names and made whitespace fixes
>
^ That should be in the patch description or embedded into the
changelog.
Nothing else jumps out at me below,
Thanks
-Will
> The GNU C Library is free software; you can redistribute it
> and/or
> modify it under the terms of the GNU Lesser General Public
> @@ -35,7 +38,7 @@
> #include "quad-float128.h"
>
> UTItype
> -__fixunskfti (TFtype a)
> +__fixunskfti_sw (TFtype a)
> {
> FP_DECL_EX;
> FP_DECL_Q (A);
> diff --git a/libgcc/config/rs6000/float128-hw.c
> b/libgcc/config/rs6000/float128-hw.c
> index 8705b53e22a..be8bd07e853 100644
> --- a/libgcc/config/rs6000/float128-hw.c
> +++ b/libgcc/config/rs6000/float128-hw.c
> @@ -86,6 +86,30 @@ __floatdikf_hw (DItype_ppc a)
> return (TFtype) a;
> }
>
> +TFtype
> +__floattikf_hw (TItype_ppc a)
> +{
> + return (TFtype) a;
> +}
> +
> +TFtype
> +__floatuntikf_hw (UTItype_ppc a)
> +{
> + return (TFtype) a;
> +}
> +
> +TItype_ppc
> +__fixkfti_hw (TFtype a)
> +{
> + return (TItype_ppc) a;
> +}
> +
> +UTItype_ppc
> +__fixunskfti_hw (TFtype a)
> +{
> + return (UTItype_ppc) a;
> +}
> +
> TFtype
> __floatundikf_hw (UDItype_ppc a)
> {
> diff --git a/libgcc/config/rs6000/float128-ifunc.c
> b/libgcc/config/rs6000/float128-ifunc.c
> index c2f65912a74..c221be2c864 100644
> --- a/libgcc/config/rs6000/float128-ifunc.c
> +++ b/libgcc/config/rs6000/float128-ifunc.c
> @@ -46,14 +46,9 @@
> #endif
>
> #define SW_OR_HW(SW, HW) (__builtin_cpu_supports ("ieee128") ? HW :
> SW)
> +#define SW_OR_HW_ISA3_1(SW, HW) (__builtin_cpu_supports ("arch_3_1")
> ? HW : SW)
>
> /* Resolvers. */
> -
> -/* We do not provide ifunc resolvers for __fixkfti, __fixunskfti,
> __floattikf,
> - and __floatuntikf. There is no ISA 3.0 instruction that converts
> between
> - 128-bit integer types and 128-bit IEEE floating point, or vice
> versa. So
> - use the emulator functions for these conversions. */
This would be good to turn into part of the changelog.
"Provide ifunc resolvers for 128-bit floating point conversions..."
> -
> static __typeof__ (__addkf3_sw) *
> __addkf3_resolve (void)
> {
> @@ -102,6 +97,18 @@ __floatdikf_resolve (void)
> return SW_OR_HW (__floatdikf_sw, __floatdikf_hw);
> }
>
> +static __typeof__ (__floattikf_sw) *
> +__floattikf_resolve (void)
> +{
> + return SW_OR_HW_ISA3_1 (__floattikf_sw, __floattikf_hw);
> +}
> +
> +static __typeof__ (__floatuntikf_sw) *
> +__floatuntikf_resolve (void)
> +{
> + return SW_OR_HW_ISA3_1 (__floatuntikf_sw, __floatuntikf_hw);
> +}
> +
> static __typeof__ (__floatunsikf_sw) *
> __floatunsikf_resolve (void)
> {
> @@ -114,6 +121,19 @@ __floatundikf_resolve (void)
> return SW_OR_HW (__floatundikf_sw, __floatundikf_hw);
> }
>
> +
> +static __typeof__ (__fixkfti_sw) *
> +__fixkfti_resolve (void)
> +{
> + return SW_OR_HW_ISA3_1 (__fixkfti_sw, __fixkfti_hw);
> +}
> +
> +static __typeof__ (__fixunskfti_sw) *
> +__fixunskfti_resolve (void)
> +{
> + return SW_OR_HW_ISA3_1 (__fixunskfti_sw, __fixunskfti_hw);
> +}
> +
> static __typeof__ (__fixkfsi_sw) *
> __fixkfsi_resolve (void)
> {
> @@ -303,6 +323,18 @@ TFtype __floatsikf (SItype_ppc)
> TFtype __floatdikf (DItype_ppc)
> __attribute__ ((__ifunc__ ("__floatdikf_resolve")));
>
> +TFtype __floattikf (TItype_ppc)
> + __attribute__ ((__ifunc__ ("__floattikf_resolve")));
> +
> +TFtype __floatuntikf (UTItype_ppc)
> + __attribute__ ((__ifunc__ ("__floatuntikf_resolve")));
> +
> +TItype_ppc __fixkfti (TFtype)
> + __attribute__ ((__ifunc__ ("__fixkfti_resolve")));
> +
> +UTItype_ppc __fixunskfti (TFtype)
> + __attribute__ ((__ifunc__ ("__fixunskfti_resolve")));
> +
> TFtype __floatunsikf (USItype_ppc)
> __attribute__ ((__ifunc__ ("__floatunsikf_resolve")));
>
> diff --git a/libgcc/config/rs6000/float128-sed
> b/libgcc/config/rs6000/float128-sed
> index d9a089ff9ba..c0fcddb1959 100644
> --- a/libgcc/config/rs6000/float128-sed
> +++ b/libgcc/config/rs6000/float128-sed
> @@ -8,6 +8,10 @@ s/__fixtfsi/__fixkfsi/g
> s/__fixunstfdi/__fixunskfdi/g
> s/__fixunstfsi/__fixunskfsi/g
> s/__floatditf/__floatdikf/g
> +s/__floattitf/__floattikf/g
> +s/__floatuntitf/__floatuntikf/g
> +s/__fixtfti/__fixkfti/g
> +s/__fixunstfti/__fixunskfti/g
> s/__floatsitf/__floatsikf/g
> s/__floatunditf/__floatundikf/g
> s/__floatunsitf/__floatunsikf/g
> diff --git a/libgcc/config/rs6000/float128-sed-hw
> b/libgcc/config/rs6000/float128-sed-hw
> index acf36b0c17d..3d2bf556da1 100644
> --- a/libgcc/config/rs6000/float128-sed-hw
> +++ b/libgcc/config/rs6000/float128-sed-hw
> @@ -8,6 +8,10 @@ s/__fixtfsi/__fixkfsi_sw/g
> s/__fixunstfdi/__fixunskfdi_sw/g
> s/__fixunstfsi/__fixunskfsi_sw/g
> s/__floatditf/__floatdikf_sw/g
> +s/__floattitf/__floattikf_sw/g
> +s/__floatuntitf/__floatuntikf_sw/g
> +s/__fixtfti/__fixkfti_sw/g
> +s/__fixunstfti/__fixunskfti_sw/g
> s/__floatsitf/__floatsikf_sw/g
> s/__floatunditf/__floatundikf_sw/g
> s/__floatunsitf/__floatunsikf_sw/g
> diff --git a/libgcc/config/rs6000/floattikf.c
> b/libgcc/config/rs6000/floattikf-sw.c
> similarity index 96%
> rename from libgcc/config/rs6000/floattikf.c
> rename to libgcc/config/rs6000/floattikf-sw.c
> index 4e8c40cfbe4..110706352bb 100644
> --- a/libgcc/config/rs6000/floattikf.c
> +++ b/libgcc/config/rs6000/floattikf-sw.c
> @@ -5,7 +5,7 @@
> This file is part of the GNU C Library.
> Contributed by Steven Munroe (munroesj@linux.vnet.ibm.com)
> Code is based on the main soft-fp library written by:
> - Uros Bizjak (ubizjak@gmail.com).
> + Uros Bizjak (ubizjak@gmail.com).
>
> The GNU C Library is free software; you can redistribute it
> and/or
> modify it under the terms of the GNU Lesser General Public
> @@ -35,7 +35,7 @@
> #include "quad-float128.h"
>
> TFtype
> -__floattikf (TItype i)
> +__floattikf_sw (TItype i)
> {
> FP_DECL_EX;
> FP_DECL_Q (A);
> diff --git a/libgcc/config/rs6000/floatuntikf.c
> b/libgcc/config/rs6000/floatuntikf-sw.c
> similarity index 96%
> rename from libgcc/config/rs6000/floatuntikf.c
> rename to libgcc/config/rs6000/floatuntikf-sw.c
> index 8bfba4267d4..5e712a67e26 100644
> --- a/libgcc/config/rs6000/floatuntikf.c
> +++ b/libgcc/config/rs6000/floatuntikf-sw.c
> @@ -5,7 +5,7 @@
> This file is part of the GNU C Library.
> Contributed by Steven Munroe (munroesj@linux.vnet.ibm.com)
> Code is based on the main soft-fp library written by:
> - Uros Bizjak (ubizjak@gmail.com).
> + Uros Bizjak (ubizjak@gmail.com).
>
> The GNU C Library is free software; you can redistribute it
> and/or
> modify it under the terms of the GNU Lesser General Public
> @@ -35,7 +35,7 @@
> #include "quad-float128.h"
>
> TFtype
> -__floatuntikf (UTItype i)
> +__floatuntikf_sw (UTItype i)
> {
> FP_DECL_EX;
> FP_DECL_Q (A);
> diff --git a/libgcc/config/rs6000/quad-float128.h
> b/libgcc/config/rs6000/quad-float128.h
> index 32ef328a8ea..24712b9277f 100644
> --- a/libgcc/config/rs6000/quad-float128.h
> +++ b/libgcc/config/rs6000/quad-float128.h
> @@ -87,19 +87,18 @@ extern USItype_ppc __fixunskfsi_sw (TFtype);
> extern UDItype_ppc __fixunskfdi_sw (TFtype);
> extern TFtype __floatsikf_sw (SItype_ppc);
> extern TFtype __floatdikf_sw (DItype_ppc);
> +extern TFtype __floattikf_sw (TItype_ppc);
> extern TFtype __floatunsikf_sw (USItype_ppc);
> extern TFtype __floatundikf_sw (UDItype_ppc);
> +extern TFtype __floatuntikf_sw (UTItype_ppc);
> +extern TItype_ppc __fixkfti_sw (TFtype);
> +extern UTItype_ppc __fixunskfti_sw (TFtype);
> extern IBM128_TYPE __extendkftf2_sw (TFtype);
> extern TFtype __trunctfkf2_sw (IBM128_TYPE);
> extern TCtype __mulkc3_sw (TFtype, TFtype, TFtype, TFtype);
> extern TCtype __divkc3_sw (TFtype, TFtype, TFtype, TFtype);
>
> #ifdef _ARCH_PPC64
> -/* We do not provide ifunc resolvers for __fixkfti, __fixunskfti,
> __floattikf,
> - and __floatuntikf. There is no ISA 3.0 instruction that converts
> between
> - 128-bit integer types and 128-bit IEEE floating point, or vice
> versa. So
> - use the emulator functions for these conversions. */
> -
> extern TItype_ppc __fixkfti (TFtype);
> extern UTItype_ppc __fixunskfti (TFtype);
> extern TFtype __floattikf (TItype_ppc);
> @@ -130,8 +129,12 @@ extern USItype_ppc __fixunskfsi_hw (TFtype);
> extern UDItype_ppc __fixunskfdi_hw (TFtype);
> extern TFtype __floatsikf_hw (SItype_ppc);
> extern TFtype __floatdikf_hw (DItype_ppc);
> +extern TFtype __floattikf_hw (TItype_ppc);
> extern TFtype __floatunsikf_hw (USItype_ppc);
> extern TFtype __floatundikf_hw (UDItype_ppc);
> +extern TFtype __floatuntikf_hw (UTItype_ppc);
> +extern TItype_ppc __fixkfti_hw (TFtype);
> +extern UTItype_ppc __fixunskfti_hw (TFtype);
> extern IBM128_TYPE __extendkftf2_hw (TFtype);
> extern TFtype __trunctfkf2_hw (IBM128_TYPE);
> extern TCtype __mulkc3_hw (TFtype, TFtype, TFtype, TFtype);
> @@ -162,8 +165,12 @@ extern USItype_ppc __fixunskfsi (TFtype);
> extern UDItype_ppc __fixunskfdi (TFtype);
> extern TFtype __floatsikf (SItype_ppc);
> extern TFtype __floatdikf (DItype_ppc);
> +extern TFtype __floattikf (TItype_ppc);
> extern TFtype __floatunsikf (USItype_ppc);
> extern TFtype __floatundikf (UDItype_ppc);
> +extern TFtype __floatuntikf (UTItype_ppc);
> +extern TItype_ppc __fixkfti (TFtype);
> +extern UTItype_ppc __fixunskfti (TFtype);
> extern IBM128_TYPE __extendkftf2 (TFtype);
> extern TFtype __trunctfkf2 (IBM128_TYPE);
>
> diff --git a/libgcc/config/rs6000/t-float128
> b/libgcc/config/rs6000/t-float128
> index d5413445189..325b22fd49e 100644
> --- a/libgcc/config/rs6000/t-float128
> +++ b/libgcc/config/rs6000/t-float128
> @@ -23,7 +23,8 @@ fp128_softfp_shared_obj = $(addsuffix
> -sw_s$(objext),$(fp128_softfp_funcs))
> fp128_softfp_obj = $(fp128_softfp_static_obj)
> $(fp128_softfp_shared_obj)
>
> # New functions for software emulation
> -fp128_ppc_funcs = floattikf floatuntikf fixkfti
> fixunskfti \
> +fp128_ppc_funcs = floattikf-sw floatuntikf-sw \
> + fixkfti-sw fixunskfti-sw \
> extendkftf2-sw trunctfkf2-sw \
> sfp-exceptions _mulkc3 _divkc3 _powikf2
>
next prev parent reply other threads:[~2020-09-24 18:22 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-21 21:17 [Patch 0/5] rs6000, 128-bit Binary Integer Operations Carl Love
2020-09-21 23:56 ` [PATCH 1/5] RS6000 Add 128-bit Binary Integer sign extend operations Carl Love
2020-09-24 18:20 ` will schmidt
2020-10-05 18:51 ` Carl Love
2020-10-07 21:08 ` will schmidt
2020-10-12 20:15 ` Carl Love
2020-10-12 20:51 ` Segher Boessenkool
2020-10-12 20:43 ` Segher Boessenkool
2020-10-12 21:06 ` Carl Love
2020-09-21 23:56 ` [PATCH 2/5] RS6000 add 128-bit Integer Operations Carl Love
2020-09-24 18:21 ` will schmidt
2020-10-05 18:52 ` [PATCH 2a/5] rs6000, vec_rlnm builtin fix arguments Carl Love
2020-10-07 21:08 ` will schmidt
2020-10-12 20:15 ` Carl Love
2020-10-12 23:37 ` Segher Boessenkool
2020-10-05 18:52 ` [PATCH 2b/5] RS6000 add 128-bit Integer Operations Carl Love
2020-10-07 21:53 ` will schmidt
2020-10-12 20:16 ` Carl Love
2020-10-13 0:23 ` Segher Boessenkool
2021-01-19 22:33 ` [PATCH 0/6 ver3] " Carl Love
2021-01-19 22:33 ` [PATCH 1/6 ver 3] rs6000, Fix arguments in altivec_vrlwmi and altivec_rlwdi builtins Carl Love
2021-01-19 22:33 ` [PATCH 2/6 ver 3] RS6000 Add 128-bit Binary Integer sign extend operations Carl Love
2021-01-19 22:33 ` [PATCH 3/6 ver 3] RS6000 add 128-bit Integer Operations part 1 Carl Love
2021-01-19 22:33 ` [PATCH 4/6 ver 3] Add TI to TD (128-bit DFP) and TD to TI support Carl Love
2021-01-19 22:33 ` [PATCH 5/6 ver 3] rs6000, Add test 128-bit shifts for just the int128 type Carl Love
2021-01-19 22:33 ` [PATCH 6/6 ver 3] Conversions between 128-bit integer and floating point values Carl Love
2020-09-21 23:56 ` [PATCH 3/5] Add TI to TD (128-bit DFP) and TD to TI support Carl Love
2020-09-24 18:21 ` will schmidt
2020-10-05 18:52 ` Carl Love
2020-10-08 14:22 ` will schmidt
2020-10-12 20:15 ` Carl Love
2020-09-21 23:56 ` [PATCH 4/5] Test 128-bit shifts for just the int128 type Carl Love
2020-09-24 18:21 ` will schmidt
2020-10-05 18:52 ` Carl Love
2020-10-08 14:50 ` will schmidt
2020-10-12 20:15 ` Carl Love
2020-09-21 23:57 ` [PATCH 5/5] Conversions between 128-bit integer and floating point values Carl Love
2020-09-24 18:22 ` will schmidt [this message]
2020-10-05 18:52 ` Carl Love
2020-10-08 15:35 ` will schmidt
2020-10-12 20:16 ` Carl Love
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e73b7f6b52c0f148f181fddc84845b04dc4cadc4.camel@vnet.ibm.com \
--to=will_schmidt@vnet.ibm.com \
--cc=cel@us.ibm.com \
--cc=dje.gcc@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=segher@gcc.gnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).