This is the mail archive of the
glibc-cvs@sourceware.org
mailing list for the glibc project.
GNU C Library master sources branch master updated. glibc-2.18-768-g38f3458
- From: azanella at sourceware dot org
- To: glibc-cvs at sourceware dot org
- Date: 8 Jan 2014 14:17:30 -0000
- Subject: GNU C Library master sources branch master updated. glibc-2.18-768-g38f3458
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "GNU C Library master sources".
The branch, master has been updated
via 38f3458175ecf7c3588bd5b6e465f4d9205fbe1c (commit)
from 3ff6304ee7294951adef3255457ca28ebaf0d7d1 (commit)
Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.
- Log -----------------------------------------------------------------
http://sourceware.org/git/gitweb.cgi?p=glibc.git;a=commitdiff;h=38f3458175ecf7c3588bd5b6e465f4d9205fbe1c
commit 38f3458175ecf7c3588bd5b6e465f4d9205fbe1c
Author: Adhemerval Zanella <azanella@linux.vnet.ibm.com>
Date: Wed Jan 8 05:10:41 2014 -0600
PowerPC: remove wrong truncl implementation for PowerPC64
The truncl assembly implementation (sysdeps/powerpc/powerpc64/fpu/s_truncl.S)
returns wrong results for some inputs where first double is a exact integer
and the precision is determined by second long double.
Checking on implementation comments and history, I am very confident the
assembly implementation was based on a version before commit
5c68d401698a58cf7da150d9cce769fa6679ba5f that fixes BZ#2423 (Errors in
long double (ldbl-128ibm) rounding functions in glibc-2.4).
By just removing the implementation and make the build select
sysdeps/ieee754/ldbl-128ibm/s_truncl.c instead it fixes tgammal
issues regarding wrong result sign.
diff --git a/ChangeLog b/ChangeLog
index 62ebe4b..1b19405 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,4 +1,11 @@
-2013-01-08 Adhemerval Zanella <azanella@linux.vnet.ibm.com>
+2014-01-08 Adhemerval Zanella <azanella@linux.vnet.ibm.com>
+
+ [BZ #16414]
+ * sysdeps/powerpc/powerpc64/fpu/s_truncl.S: Remove wrong
+ implementation.
+ * sysdeps/ieee754/ldbl-128ibm/s_truncl.c: Remove bogus comment.
+
+2014-01-08 Adhemerval Zanella <azanella@linux.vnet.ibm.com>
* sysdeps/powerpc/fpu/libm-test-ulps: Update.
diff --git a/NEWS b/NEWS
index 486e892..4fe2314 100644
--- a/NEWS
+++ b/NEWS
@@ -24,7 +24,7 @@ Version 2.19
16103, 16112, 16143, 16144, 16146, 16150, 16151, 16153, 16167, 16172,
16195, 16214, 16245, 16271, 16274, 16283, 16289, 16293, 16314, 16316,
16330, 16337, 16338, 16356, 16365, 16366, 16369, 16372, 16375, 16379,
- 16384, 16385, 16386, 16387, 16390, 16400, 16407, 16408.
+ 16384, 16385, 16386, 16387, 16390, 16400, 16407, 16408, 16414.
* Slovenian translations for glibc messages have been contributed by the
Translation Project's Slovenian team of translators.
diff --git a/sysdeps/ieee754/ldbl-128ibm/s_truncl.c b/sysdeps/ieee754/ldbl-128ibm/s_truncl.c
index 490ad26..aca3f3e 100644
--- a/sysdeps/ieee754/ldbl-128ibm/s_truncl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/s_truncl.c
@@ -17,9 +17,6 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-/* This has been coded in assembler because GCC makes such a mess of it
- when it's coded in C. */
-
#include <math.h>
#include <math_ldbl_opt.h>
#include <float.h>
diff --git a/sysdeps/powerpc/powerpc64/fpu/s_truncl.S b/sysdeps/powerpc/powerpc64/fpu/s_truncl.S
deleted file mode 100644
index e5a3008..0000000
--- a/sysdeps/powerpc/powerpc64/fpu/s_truncl.S
+++ /dev/null
@@ -1,120 +0,0 @@
-/* long double trunc function.
- IBM extended format long double version.
- Copyright (C) 2004-2014 Free Software Foundation, Inc.
- This file is part of the GNU C Library.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <http://www.gnu.org/licenses/>. */
-
-#include <sysdep.h>
-#include <math_ldbl_opt.h>
-
- .section ".toc","aw"
-.LC0: /* 2**52 */
- .tc FD_43300000_0[TC],0x4330000000000000
-.LC1: /* 0.5 */
- .tc FD_3fe00000_0[TC],0x3fe0000000000000
- .section ".text"
-
-/* long double [fp1,fp2] truncl (long double x [fp1,fp2]) */
-
-ENTRY (__truncl)
- mffs fp11 /* Save current FPU rounding mode. */
- lfd fp13,.LC0@toc(2)
- fabs fp0,fp1
- fabs fp9,fp2
- fsub fp12,fp13,fp13 /* generate 0.0 */
- fcmpu cr7,fp0,fp13 /* if (fabs(x) > TWO52) */
- fcmpu cr6,fp1,fp12 /* if (x > 0.0) */
- bnl- cr7,.L2
- mtfsfi 7,1 /* Set rounding mode toward 0. */
- ble- cr6,.L1
- fneg fp2,fp12
- fadd fp1,fp1,fp13 /* x+= TWO52; */
- fsub fp1,fp1,fp13 /* x-= TWO52; */
- fabs fp1,fp1 /* if (x == 0.0) x = 0.0; */
-.L0:
- mtfsf 0x01,fp11 /* restore previous rounding mode. */
- blr
-.L1:
- fneg fp2,fp12
- bge- cr6,.L0 /* if (x < 0.0) */
- fsub fp1,fp1,fp13 /* x-= TWO52; */
- fadd fp1,fp1,fp13 /* x+= TWO52; */
- fnabs fp1,fp1 /* if (x == 0.0) x = -0.0; */
- mtfsf 0x01,fp11 /* restore previous rounding mode. */
- blr
-
-/* The high double is > TWO52 so we need to round the low double and
- perhaps the high double. In this case we have to round the low
- double and handle any adjustment to the high double that may be
- caused by rounding (up). This is complicated by the fact that the
- high double may already be rounded and the low double may have the
- opposite sign to compensate.This gets a bit tricky so we use the
- following algorithm:
-
- tau = floor(x_high/TWO52);
- x0 = x_high - tau;
- x1 = x_low + tau;
- r1 = rint(x1);
- y_high = x0 + r1;
- y_low = x0 - y_high + r1;
- return y; */
-.L2:
- fcmpu cr7,fp9,fp13 /* if (|x_low| > TWO52) */
- fcmpu cr0,fp9,fp12 /* || (|x_low| == 0.0) */
- fcmpu cr5,fp2,fp12 /* if (x_low > 0.0) */
- bgelr- cr7 /* return x; */
- beqlr- cr0
- mtfsfi 7,1 /* Set rounding mode toward 0. */
- fdiv fp8,fp1,fp13 /* x_high/TWO52 */
-
- bng- cr6,.L6 /* if (x > 0.0) */
- fctidz fp0,fp8
- fcfid fp8,fp0 /* tau = floor(x_high/TWO52); */
- fadd fp8,fp8,fp8 /* tau++; Make tau even */
- bng cr5,.L4 /* if (x_low > 0.0) */
- fmr fp3,fp1
- fmr fp4,fp2
- b .L5
-.L4: /* if (x_low < 0.0) */
- fsub fp3,fp1,fp8 /* x0 = x_high - tau; */
- fadd fp4,fp2,fp8 /* x1 = x_low + tau; */
-.L5:
- fadd fp5,fp4,fp13 /* r1 = r1 + TWO52; */
- fsub fp5,fp5,fp13 /* r1 = r1 - TWO52; */
- b .L9
-.L6: /* if (x < 0.0) */
- fctidz fp0,fp8
- fcfid fp8,fp0 /* tau = floor(x_high/TWO52); */
- fadd fp8,fp8,fp8 /* tau++; Make tau even */
- bnl cr5,.L7 /* if (x_low < 0.0) */
- fmr fp3,fp1
- fmr fp4,fp2
- b .L8
-.L7: /* if (x_low > 0.0) */
- fsub fp3,fp1,fp8 /* x0 = x_high - tau; */
- fadd fp4,fp2,fp8 /* x1 = x_low + tau; */
-.L8:
- fsub fp5,fp4,fp13 /* r1-= TWO52; */
- fadd fp5,fp5,fp13 /* r1+= TWO52; */
-.L9:
- mtfsf 0x01,fp11 /* restore previous rounding mode. */
- fadd fp1,fp3,fp5 /* y_high = x0 + r1; */
- fsub fp2,fp3,fp1 /* y_low = x0 - y_high + r1; */
- fadd fp2,fp2,fp5
- blr
-END (__truncl)
-
-long_double_symbol (libm, __truncl, truncl)
-----------------------------------------------------------------------
Summary of changes:
ChangeLog | 9 ++-
NEWS | 2 +-
sysdeps/ieee754/ldbl-128ibm/s_truncl.c | 3 -
sysdeps/powerpc/powerpc64/fpu/s_truncl.S | 120 ------------------------------
4 files changed, 9 insertions(+), 125 deletions(-)
delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_truncl.S
hooks/post-receive
--
GNU C Library master sources