This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [rfc] add ppc testcase to test fpscr
- From: Thiago Jung Bauermann <bauerman at br dot ibm dot com>
- To: Daniel Jacobowitz <drow at false dot org>
- Cc: Joel Brobecker <brobecker at adacore dot com>, gdb-patches ml <gdb-patches at sourceware dot org>
- Date: Wed, 03 Sep 2008 17:12:25 -0300
- Subject: Re: [rfc] add ppc testcase to test fpscr
- References: <1219360611.8989.6.camel@localhost.localdomain> <20080821233115.GA1239@caradoc.them.org> <1219362081.29526.2.camel@localhost.localdomain> <20080822024659.GA12951@caradoc.them.org> <1219428669.8167.2.camel@localhost.localdomain> <20080902215458.GB25623@caradoc.them.org>
El mar, 02-09-2008 a las 17:54 -0400, Daniel Jacobowitz escribiÃ:
> On Fri, Aug 22, 2008 at 03:11:09PM -0300, Thiago Jung Bauermann wrote:
> > +# Run the program, when the prompt comes back it will be for the invalid
> > +# operation breakpoint.
> > +gdb_test "run" "" ""
>
> This is similar to what Joel said about testing bare metal targets -
> using "run" will break gdbserver testing. Pick almost any test and you
> can see how this is usually handled, e.g. by gdb_run_cmd or "runto
> main".
I chose gdb_test because it waits for $gdb_prompt to show up, which
seems to be a good way to avoid creating a racy testcase.
But I just changed to use gdb_run_cmd instead. (I'm glad I remembered my
original reasoning, at least one of my testcase gaffes can be
explained!)
> > +# First, verify if FPSCR is all zeroes.
> > +gdb_test "print \$fpscr" " = 0\r" "FPSCR is all zeroes"
>
> Oh, also there's PowerPC targets without hardware floating
> point... and GDB is sometimes smart enough nowadays to not display
> dummy FP registers. So having fpscr not exist is legitimate.
> There's also the E500, which does not have a classic FPU.
Good point. What about checking for "= void" when printing the FPSCR to
determine if the register exists? That's what I do below.
This version also incorporates Joel's comments. WDYT?
--
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center
2008-09-03 Thiago Jung Bauermann <bauerman@br.ibm.com>
* gdb.arch/ppc-fp.exp: New file.
* gdb.arch/ppc-fp.c: New file.
diff --git a/gdb/testsuite/gdb.arch/ppc-fp.c b/gdb/testsuite/gdb.arch/ppc-fp.c
new file mode 100644
index 0000000..841b112
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/ppc-fp.c
@@ -0,0 +1,36 @@
+/* Copyright 2008 Free Software Foundation, Inc.
+
+ This file is part of GDB.
+
+ 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/>. */
+
+int
+main (int argc, char *argv[])
+{
+ double result;
+
+ asm ("fdiv %0, %1, %1\n" /* Invalid operation. */
+ : "=f" (result)
+ : "f" (0.0));
+
+ asm ("mtfsf 0xff, %0\n" /* Reset FPSCR. */
+ :
+ : "f" (0.0));
+
+ asm ("fdiv %0, %1, %2\n" /* Division by zero. */
+ : "=f" (result)
+ : "f" (1.25), "f" (0.0));
+
+ return 0;
+}
diff --git a/gdb/testsuite/gdb.arch/ppc-fp.exp b/gdb/testsuite/gdb.arch/ppc-fp.exp
new file mode 100644
index 0000000..33824f6
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/ppc-fp.exp
@@ -0,0 +1,101 @@
+# Copyright (C) 2008 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/>.
+#
+# Please email any bugs, comments, and/or additions to this file to:
+# bug-gdb@prep.ai.mit.edu
+#
+
+# Tests for Powerpc floating point register setting and fetching
+
+if $tracelevel then {
+ strace $tracelevel
+}
+
+if ![istarget "powerpc*"] then {
+ verbose "Skipping powerpc floating point register tests."
+ verbose -log "Skipping powerpc floating point register tests."
+ return
+}
+
+set testfile "ppc-fp"
+set binfile ${objdir}/${subdir}/${testfile}
+set srcfile ${testfile}.c
+
+if [get_compiler_info $binfile] {
+ warning "get_compiler failed"
+ return -1
+}
+
+if ![test_compiler_info gcc*] {
+ # We use GCC's extended asm syntax
+ warning "unknown compiler"
+ return -1
+}
+
+if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {quiet debug additional_flags=-O0}] != "" } {
+ verbose "Skipping FPSCR tests."
+ return -1
+}
+
+# Start with a fresh gdb.
+
+gdb_exit
+gdb_start
+gdb_reinitialize_dir $srcdir/$subdir
+gdb_load ${binfile}
+
+gdb_breakpoint [gdb_get_line_number "Invalid operation."]
+gdb_breakpoint [gdb_get_line_number "Division by zero."]
+
+# Run the program, when the prompt comes back it will be for the invalid
+# operation breakpoint.
+gdb_run_cmd
+
+# First, verify if FPSCR exists and is all zeroes.
+gdb_test_multiple "print \$fpscr" "FPSCR exists" {
+ -re " = 0" { pass "FPSCR is all zeroes" }
+
+ -re " = void" {
+ # There's no FPSCR. Skip this testcase.
+ verbose "Skipping powerpc floating point register tests."
+ verbose -log "Skipping powerpc floating point register tests."
+ return
+ }
+}
+
+# Step over invalid operation.
+gdb_test "next" "" ""
+
+# Verify that the following bits are set (See Power ISA for details):
+#
+# 32 - Floating-Point Exception Summary (FX)
+# 34 - Floating-Point Invalid Operation Summary (VX)
+# 42 - Floating-Point Invalid Operation Exception (VXZDZ)
+# 47 - Floating-Point Result Class Descriptor (C)
+# 51 - Floating-Point Unordered or NaN (FU or ?)
+gdb_test "print/t \$fpscr" " = 10100000001000010001000000000000" "FPSCR for invalid operation"
+
+gdb_continue_to_breakpoint "go to division by zero"
+
+# Step over division by zero.
+gdb_test "next" "" ""
+
+# Verify that the following bits are set (See Power ISA for details):
+#
+# 32 - Floating-Point Exception Summary (FX)
+# 37 - Floating-Point Zero Divide Exception (ZX)
+# 49 - Floating-Point Greater Than or Positive (FG or >)
+# 51 - Floating-Point Unordered or NaN (FU or ?)
+gdb_test "print/t \$fpscr" " = 10000100000000000101000000000000" "FPSCR for division by zero"