[v2] ldbl-128ibm-compat: Add printf_size
Commit Message
Changes since v1 (https://sourceware.org/ml/libc-alpha/2018-06/msg00929.html):
- Do _not_ change the behavior of __printf_fp (i.e.: is_binary128
keeps its precedence over is_long_double).
- In ___ieee128_printf_size, set is_binary128 to is_long_double,
instead of hardcoding it to one (which would break the printing of
values with double type).
-- 8< --
Since the addition of the _Float128 API, strfromf128 and printf_size use
__printf_fp to print _Float128 values. This is achieved by setting the
'is_binary128' member of the 'printf_info' structure to one. Now that
the format of long double on powerpc64le is getting a third option, this
mechanism is reused for long double values that have binary128 format
(i.e.: when -mabi=ieeelongdouble).
This patch adds __printf_sizeieee128 as an exported symbol, but doesn't
provide redirections from printf_size, yet. All redirections will be
installed in a future commit, once all other functions that print or
read long double values with binary128 format are ready. In
__printf_fp, when 'is_binary128' is one, the floating-point argument is
treated as if it was of _Float128 type, regardless of the value of
'is_long_double', thus __printf_sizeieee128 sets 'is_binary128' to the
same value of 'is_long_double'. Otherwise, double values would not be
printed correctly.
Tested for powerpc64le.
* stdio-common/Makefile (test-srcs): Add tst-printfsz-islongdouble.
(tests-special) Add $(objpfx)tst-printfsz-islongdouble.out.
($(objpfx)tst-printfsz-islongdouble.out): New build and run rule.
* stdio-common/tst-printfsz-islongdouble.c: New file.
* stdio-common/tst-printfsz-islongdouble.sh: Likewise.
* sysdeps/ieee754/ldbl-128ibm-compat/Makefile:
[subdir == stdio-common] (routines): Add ieee128-printf_size.
[subdir == stdio-common] (tests-internal): Add
test-printf-size-ieee128, and test-printf-size-ibm128.
[subdir == stdio-common] (CFLAGS-test-printf-size-ieee128.c)
(CFLAGS-test-printf-size-ibm128.c): New variables.
[subdir == stdio-common] (tests-special): Add
$(objpfx)test-printf-size-ieee128.out and
$(objpfx)test-printf-size-ibm128.out.
[subdir == stdio-common] ($(objpfx)test-printf-size-ieee128.out)
($(objpfx)test-printf-size-ibm128.out): New build and run rules.
* sysdeps/ieee754/ldbl-128ibm-compat/Versions (libc): Add
__printf_sizeieee128.
* sysdeps/ieee754/ldbl-128ibm-compat/ieee128-printf_size.c:
New file.
* sysdeps/ieee754/ldbl-128ibm-compat/test-printf-size-ibm128.c:
Likewise.
* sysdeps/ieee754/ldbl-128ibm-compat/test-printf-size-ieee128.c:
Likewise.
---
stdio-common/Makefile | 10 ++++-
stdio-common/tst-printfsz-islongdouble.c | 51 ++++++++++++++++++++++
stdio-common/tst-printfsz-islongdouble.sh | 38 ++++++++++++++++
sysdeps/ieee754/ldbl-128ibm-compat/Makefile | 22 ++++++++++
sysdeps/ieee754/ldbl-128ibm-compat/Versions | 2 +
.../ldbl-128ibm-compat/ieee128-printf_size.c | 32 ++++++++++++++
.../ldbl-128ibm-compat/test-printf-size-ibm128.c | 1 +
.../ldbl-128ibm-compat/test-printf-size-ieee128.c | 1 +
8 files changed, 155 insertions(+), 2 deletions(-)
create mode 100644 stdio-common/tst-printfsz-islongdouble.c
create mode 100644 stdio-common/tst-printfsz-islongdouble.sh
create mode 100644 sysdeps/ieee754/ldbl-128ibm-compat/Makefile
create mode 100644 sysdeps/ieee754/ldbl-128ibm-compat/ieee128-printf_size.c
create mode 100644 sysdeps/ieee754/ldbl-128ibm-compat/test-printf-size-ibm128.c
create mode 100644 sysdeps/ieee754/ldbl-128ibm-compat/test-printf-size-ieee128.c
Comments
On Fri, 29 Jun 2018, Gabriel F. T. Gomes wrote:
> This patch adds __printf_sizeieee128 as an exported symbol, but doesn't
> provide redirections from printf_size, yet. All redirections will be
> installed in a future commit, once all other functions that print or
> read long double values with binary128 format are ready. In
> __printf_fp, when 'is_binary128' is one, the floating-point argument is
> treated as if it was of _Float128 type, regardless of the value of
> 'is_long_double', thus __printf_sizeieee128 sets 'is_binary128' to the
> same value of 'is_long_double'. Otherwise, double values would not be
> printed correctly.
This patch is OK.
On Sat, 30 Jun 2018, Joseph Myers wrote:
>On Fri, 29 Jun 2018, Gabriel F. T. Gomes wrote:
>
>> This patch adds __printf_sizeieee128 as an exported symbol, but doesn't
>> provide redirections from printf_size, yet. All redirections will be
>> installed in a future commit, once all other functions that print or
>> read long double values with binary128 format are ready. In
>> __printf_fp, when 'is_binary128' is one, the floating-point argument is
>> treated as if it was of _Float128 type, regardless of the value of
>> 'is_long_double', thus __printf_sizeieee128 sets 'is_binary128' to the
>> same value of 'is_long_double'. Otherwise, double values would not be
>> printed correctly.
>
>This patch is OK.
Thanks. Now pushed.
@@ -63,13 +63,14 @@ tests := tstscanf test_rdwr test-popen tstgetln test-fseek \
tst-vfprintf-mbs-prec \
tst-scanf-round \
-test-srcs = tst-unbputc tst-printf
+test-srcs = tst-unbputc tst-printf tst-printfsz-islongdouble
ifeq ($(run-built-tests),yes)
tests-special += $(objpfx)tst-unbputc.out $(objpfx)tst-printf.out \
$(objpfx)tst-printf-bz18872-mem.out \
$(objpfx)tst-setvbuf1-cmp.out \
- $(objpfx)tst-vfprintf-width-prec-mem.out
+ $(objpfx)tst-vfprintf-width-prec-mem.out \
+ $(objpfx)tst-printfsz-islongdouble.out
generated += tst-printf-bz18872.c tst-printf-bz18872.mtrace \
tst-printf-bz18872-mem.out \
tst-vfprintf-width-prec.mtrace tst-vfprintf-width-prec-mem.out
@@ -103,6 +104,11 @@ $(objpfx)tst-printf.out: tst-printf.sh $(objpfx)tst-printf
$(SHELL) $< $(common-objpfx) '$(test-program-prefix)' > $@; \
$(evaluate-test)
+$(objpfx)tst-printfsz-islongdouble.out: \
+ tst-printfsz-islongdouble.sh $(objpfx)tst-printfsz-islongdouble
+ $(SHELL) $^ '$(test-program-prefix)' $@; \
+ $(evaluate-test)
+
# We generate this source because it requires a printf invocation with
# 10K arguments.
$(objpfx)tst-printf-bz18872.c: tst-printf-bz18872.sh
new file mode 100644
@@ -0,0 +1,51 @@
+/* Test for the behaviour of 'is_long_double' in printf_size.
+ Copyright (C) 2018 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 <printf.h>
+#include <stdio.h>
+#include <string.h>
+
+#include <support/check.h>
+
+static int
+do_test (void)
+{
+ double d = 2000;
+ double *dptr = &d;
+ long double ld = 4000;
+ long double *ldptr = & ld;
+ struct printf_info info;
+
+ memset (&info, 0, sizeof (info));
+ info.spec = L'f';
+
+ /* Print a value with double type. */
+ printf_size (stdout, &info, (void *) &dptr);
+
+ /* Printf a value with long double type. */
+ info.is_long_double = 1;
+ printf_size (stdout, &info, (void *) &ldptr);
+
+ /* Setting both 'is_long_double' and 'is_binary128' to one is out of
+ the scope of this test, because such configuration is only valid
+ when _Float128 and long double are ABI-distinct (which is not
+ always true in this arch-independent test). */
+ return 0;
+}
+
+#include <support/test-driver.c>
new file mode 100644
@@ -0,0 +1,38 @@
+#!/bin/sh
+# Test for the behaviour of 'is_binary128' in printf_size.
+# Copyright (C) 2018 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/>.
+
+set -e
+
+test_program=$1; shift
+test_program_prefix=$1; shift
+test_program_output=$1; shift
+
+status=0
+
+${test_program_prefix} \
+ ${test_program} \
+ > ${test_program_output} || status=1
+
+echo -n "2k4k" | cmp - ${test_program_output} > /dev/null 2>&1 ||
+{
+ status=1
+ echo "*** output comparison failed"
+}
+
+exit $status
new file mode 100644
@@ -0,0 +1,22 @@
+ifeq ($(subdir),stdio-common)
+routines += ieee128-printf_size
+
+tests-internal += test-printf-size-ieee128 test-printf-size-ibm128
+CFLAGS-test-printf-size-ieee128.c += -mfloat128 -mabi=ieeelongdouble -Wno-psabi
+CFLAGS-test-printf-size-ibm128.c += -mabi=ibmlongdouble -Wno-psabi
+
+ifeq ($(run-built-tests),yes)
+tests-special += $(objpfx)test-printf-size-ieee128.out
+tests-special += $(objpfx)test-printf-size-ibm128.out
+endif
+
+$(objpfx)test-printf-size-ieee128.out: \
+ tst-printfsz-islongdouble.sh $(objpfx)test-printf-size-ieee128
+ $(SHELL) $^ '$(test-program-prefix)' $@; \
+ $(evaluate-test)
+
+$(objpfx)test-printf-size-ibm128.out: \
+ tst-printfsz-islongdouble.sh $(objpfx)test-printf-size-ibm128
+ $(SHELL) $^ '$(test-program-prefix)' $@; \
+ $(evaluate-test)
+endif
@@ -118,4 +118,6 @@ libc {
__strtoieee128_l;
__wcstoieee128;
__wcstoieee128_l;
+
+ __printf_sizeieee128;
}
new file mode 100644
@@ -0,0 +1,32 @@
+/* Wrapper for printf_size. IEEE128 version.
+ Copyright (C) 2018 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 <printf.h>
+
+extern __typeof (printf_size) __printf_size;
+
+int
+___ieee128_printf_size (FILE *fp, const struct printf_info *info,
+ const void *const *args)
+{
+ struct printf_info info_ieee128 = *info;
+
+ info_ieee128.is_binary128 = info->is_long_double;
+ return __printf_size (fp, &info_ieee128, args);
+}
+strong_alias (___ieee128_printf_size, __printf_sizeieee128)
new file mode 100644
@@ -0,0 +1 @@
+#include <tst-printfsz-islongdouble.c>
new file mode 100644
@@ -0,0 +1 @@
+#include <tst-printfsz-islongdouble.c>