[PowerPC] Detect different long double floating-point formats

Current versions of GCC support switching the format used for "long double"
to either IBM double double or IEEE-128.  The resulting binary is marked
via different setting of the Tag_GNU_Power_ABI_FP GNU attribute.

This patch checks this attribute to detect the format of the default
"long double" type and sets GDB's notion of the format accordingly.

The patch also adds support for the "__ibm128" type, which always uses
IBM double double format independent of the format used for "long double".

A new test case verifies that all three types, "long double", "__float128",
and "__ibm128" are correctly detected in all three compiler settings,
the default setting, -mabi=ieeelongdouble, and -mabi=ibmlongdouble.

gdb/ChangeLog:
2017-11-21  Ulrich Weigand  <uweigand@de.ibm.com>

	* ppc-tdep.h (enum powerpc_long_double_abi): New data type.
	(struct gdbarch_tdep): New member long_double_abi.
	* rs6000-tdep.c (rs6000_gdbarch_init): Initialize long_double_abi
	member of tdep struct based on Tag_GNU_Power_ABI_FP attribute.
	* ppc-linux-tdep.c (ppc_linux_init_abi): Install long double data
	format depending on long_double_abi tdep member.
	(ppc_floatformat_for_type): Handle __ibm128 type.

gdb/testsuite/ChangeLog:
2017-11-21  Ulrich Weigand  <uweigand@de.ibm.com>

	* gdb.arch/ppc-longdouble.exp: New file.
	* gdb.arch/ppc-longdouble.c: Likewise.

7 files changed, 163 insertions, 12 deletions

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 88c7603..0ea3228 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,13 @@
+2017-11-21  Ulrich Weigand  <uweigand@de.ibm.com>
+
+	* ppc-tdep.h (enum powerpc_long_double_abi): New data type.
+	(struct gdbarch_tdep): New member long_double_abi.
+	* rs6000-tdep.c (rs6000_gdbarch_init): Initialize long_double_abi
+	member of tdep struct based on Tag_GNU_Power_ABI_FP attribute.
+	* ppc-linux-tdep.c (ppc_linux_init_abi): Install long double data
+	format depending on long_double_abi tdep member.
+	(ppc_floatformat_for_type): Handle __ibm128 type.
+
 2017-11-20  Simon Marchi  <simon.marchi@polymtl.ca>
 
 	* darwin-nat.c (set_enable_mach_exceptions): Constify parameter.
diff --git a/gdb/ppc-linux-tdep.c b/gdb/ppc-linux-tdep.c
index ee80a71..0e43a64 100644
--- a/gdb/ppc-linux-tdep.c
+++ b/gdb/ppc-linux-tdep.c
@@ -1623,12 +1623,17 @@ ppc_floatformat_for_type (struct gdbarch *gdbarch,
                           const char *name, int len)
 {
   if (len == 128 && name)
-    if (strcmp (name, "__float128") == 0
-        || strcmp (name, "_Float128") == 0
-        || strcmp (name, "_Float64x") == 0
-        || strcmp (name, "complex _Float128") == 0
-        || strcmp (name, "complex _Float64x") == 0)
-      return floatformats_ia64_quad;
+    {
+      if (strcmp (name, "__float128") == 0
+	  || strcmp (name, "_Float128") == 0
+	  || strcmp (name, "_Float64x") == 0
+	  || strcmp (name, "complex _Float128") == 0
+	  || strcmp (name, "complex _Float64x") == 0)
+	return floatformats_ia64_quad;
+
+      if (strcmp (name, "__ibm128") == 0)
+	return floatformats_ibm_long_double;
+    }
 
   return default_floatformat_for_type (gdbarch, name, len);
 }
@@ -1648,12 +1653,15 @@ ppc_linux_init_abi (struct gdbarch_info info,
   linux_init_abi (info, gdbarch);
 
   /* PPC GNU/Linux uses either 64-bit or 128-bit long doubles; where
-     128-bit, they are IBM long double, not IEEE quad long double as
-     in the System V ABI PowerPC Processor Supplement.  We can safely
-     let them default to 128-bit, since the debug info will give the
-     size of type actually used in each case.  */
+     128-bit, they can be either IBM long double or IEEE quad long double.
+     The 64-bit long double case will be detected automatically using
+     the size specified in debug info.  We use a .gnu.attribute flag
+     to distinguish between the IBM long double and IEEE quad cases.  */
   set_gdbarch_long_double_bit (gdbarch, 16 * TARGET_CHAR_BIT);
-  set_gdbarch_long_double_format (gdbarch, floatformats_ibm_long_double);
+  if (tdep->long_double_abi == POWERPC_LONG_DOUBLE_IEEE128)
+    set_gdbarch_long_double_format (gdbarch, floatformats_ia64_quad);
+  else
+    set_gdbarch_long_double_format (gdbarch, floatformats_ibm_long_double);
 
   /* Support for floating-point data type variants.  */
   set_gdbarch_floatformat_for_type (gdbarch, ppc_floatformat_for_type);
diff --git a/gdb/ppc-tdep.h b/gdb/ppc-tdep.h
index 4ae5c7f..3d1e327 100644
--- a/gdb/ppc-tdep.h
+++ b/gdb/ppc-tdep.h
@@ -202,6 +202,15 @@ enum powerpc_vector_abi
   POWERPC_VEC_LAST
 };
 
+/* long double ABI version used by the inferior.  */
+enum powerpc_long_double_abi
+{
+  POWERPC_LONG_DOUBLE_AUTO,
+  POWERPC_LONG_DOUBLE_IBM128,
+  POWERPC_LONG_DOUBLE_IEEE128,
+  POWERPC_LONG_DOUBLE_LAST
+};
+
 struct gdbarch_tdep
   {
     int wordsize;		/* Size in bytes of fixed-point word.  */
@@ -209,6 +218,9 @@ struct gdbarch_tdep
 
     enum powerpc_elf_abi elf_abi;	/* ELF ABI version.  */
 
+    /* Format to use for the "long double" data type.  */
+    enum powerpc_long_double_abi long_double_abi;
+
     /* How to pass vector arguments.  Never set to AUTO or LAST.  */
     enum powerpc_vector_abi vector_abi;
 
diff --git a/gdb/rs6000-tdep.c b/gdb/rs6000-tdep.c
index 6c44995..456dbcc 100644
--- a/gdb/rs6000-tdep.c
+++ b/gdb/rs6000-tdep.c
@@ -5949,6 +5949,7 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
   bfd abfd;
   enum auto_boolean soft_float_flag = powerpc_soft_float_global;
   int soft_float;
+  enum powerpc_long_double_abi long_double_abi = POWERPC_LONG_DOUBLE_AUTO;
   enum powerpc_vector_abi vector_abi = powerpc_vector_abi_global;
   enum powerpc_elf_abi elf_abi = POWERPC_ELF_AUTO;
   int have_fpu = 1, have_spe = 0, have_mq = 0, have_altivec = 0, have_dfp = 0,
@@ -6271,7 +6272,7 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
   if (soft_float_flag == AUTO_BOOLEAN_AUTO && from_elf_exec)
     {
       switch (bfd_elf_get_obj_attr_int (info.abfd, OBJ_ATTR_GNU,
-					Tag_GNU_Power_ABI_FP))
+					Tag_GNU_Power_ABI_FP) & 3)
 	{
 	case 1:
 	  soft_float_flag = AUTO_BOOLEAN_FALSE;
@@ -6284,6 +6285,22 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
 	}
     }
 
+  if (long_double_abi == POWERPC_LONG_DOUBLE_AUTO && from_elf_exec)
+    {
+      switch (bfd_elf_get_obj_attr_int (info.abfd, OBJ_ATTR_GNU,
+					Tag_GNU_Power_ABI_FP) >> 2)
+	{
+	case 1:
+	  long_double_abi = POWERPC_LONG_DOUBLE_IBM128;
+	  break;
+	case 3:
+	  long_double_abi = POWERPC_LONG_DOUBLE_IEEE128;
+	  break;
+	default:
+	  break;
+	}
+    }
+
   if (vector_abi == POWERPC_VEC_AUTO && from_elf_exec)
     {
       switch (bfd_elf_get_obj_attr_int (info.abfd, OBJ_ATTR_GNU,
@@ -6365,6 +6382,8 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
 	continue;
       if (tdep && tdep->soft_float != soft_float)
 	continue;
+      if (tdep && tdep->long_double_abi != long_double_abi)
+	continue;
       if (tdep && tdep->vector_abi != vector_abi)
 	continue;
       if (tdep && tdep->wordsize == wordsize)
@@ -6387,6 +6406,7 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
   tdep->wordsize = wordsize;
   tdep->elf_abi = elf_abi;
   tdep->soft_float = soft_float;
+  tdep->long_double_abi = long_double_abi;
   tdep->vector_abi = vector_abi;
 
   gdbarch = gdbarch_alloc (&info, tdep);
diff --git a/gdb/testsuite/ChangeLog b/gdb/testsuite/ChangeLog
index 16d2801..7434105 100644
--- a/gdb/testsuite/ChangeLog
+++ b/gdb/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2017-11-21  Ulrich Weigand  <uweigand@de.ibm.com>
+
+	* gdb.arch/ppc-longdouble.exp: New file.
+	* gdb.arch/ppc-longdouble.c: Likewise.
+
 2017-11-21  Pedro Alves  <palves@redhat.com>
 
 	* gdb.ada/minsyms.exp: Accept any address for 'some_minsym'.
diff --git a/gdb/testsuite/gdb.arch/ppc-longdouble.c b/gdb/testsuite/gdb.arch/ppc-longdouble.c
new file mode 100644
index 0000000..17296bf
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/ppc-longdouble.c
@@ -0,0 +1,38 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2017 Free Software Foundation, Inc.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program 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 General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+long double ld;
+__float128 float128;
+__ibm128 ibm128;
+
+/* FIXME: GCC will only emit a .gnu.attribute section specifying the
+   long double ABI if long double is used in a function prototype.  */
+void test (long double x)
+{
+}
+
+int main()
+{
+  ld = 1.375l;
+  float128 = 2.375l;
+  ibm128 = 3.375l;
+
+  return 0;
+}
diff --git a/gdb/testsuite/gdb.arch/ppc-longdouble.exp b/gdb/testsuite/gdb.arch/ppc-longdouble.exp
new file mode 100644
index 0000000..66c8f54
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/ppc-longdouble.exp
@@ -0,0 +1,58 @@
+# Copyright 2016-2017 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program 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 General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+if ![istarget "powerpc*"] then {
+    verbose "Skipping powerpc long-double floating point tests."
+    return
+}
+
+standard_testfile
+
+proc do_test { name {opts {}} } {
+    global srcdir subdir srcfile binfile
+    set ccopts {debug quiet}
+    foreach opt $opts {lappend ccopts "additional_flags=$opt"}
+    set lines [gdb_compile "${srcdir}/${subdir}/${srcfile}" "$binfile.$name" executable $ccopts]
+    # We get this warning even with the nowarnings option ...
+    regsub -all "(^|\n)\[^\n\]*using \[^\n\]* extended precision long double" $lines "" lines
+    if { $lines != "" } then {
+        return
+    }
+
+    clean_restart ${binfile}.${name}
+
+    if ![runto_main] then {
+        return
+    }
+
+    # Run to the breakpoint at return.
+    gdb_breakpoint [gdb_get_line_number "return"]
+    gdb_continue_to_breakpoint "return"
+
+    # Print the value of ld
+    gdb_test "print ld" ".* = 1\\.375.*" "the value of ld is 1.375 ($name)"
+    # Print the value of float128
+    gdb_test "print float128" ".* = 2\\.375.*" "the value of float128 is 2.375 ($name)"
+    # Print the value of ibm128
+    gdb_test "print ibm128" ".* = 3\\.375.*" "the value of ibm128 is 3.375 ($name)"
+}
+
+# Verify that we correctly detect the floating-point format used for
+# long double.  Re-run the test with -mabi=ieeelongdouble and mabi=ibmlongdouble
+
+do_test default { -mfloat128 }
+do_test ieee128 { -mfloat128 -mabi=ieeelongdouble }
+do_test ibm128 { -mfloat128 -mabi=ibmlongdouble }
+