This is the mail archive of the gdb@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: question about ARM watchpoints


On 09/05/2014 09:17 AM, Pedro Alves wrote:

> Hmm.  Sounds like we need a test to catch this.  :-)
> 
...

> Hmm...  I had an idea.  How about:
> 
> volatile int global;
> 
> set_global (int val)
> {
> 	global = val;
> }
> 
> main ()
> {
>    set_global (1);
>    set_global (2);
> }
> 
> #01 - run to main
> #02 - set a _software_ watchpoint on i1 (w/ hw watchpoints force-disabled)
> #03 - continue
> #04 - (GDB single-steps the inferior, and checks the watchpoint's
>      value at each step.)
> #05 - watchpoint triggers
> #06 - the PC now points to the instruction right after the instruction that
>      actually caused the memory write.  So this is the address a hardware
>      watchpoint should present the stop to the user too.  Store the PC address.
> #07 - delete the software watchpoint
> #08 - set a _hardware_ watchpoint on i1.
> #09 - continue
> #10 - hardware watchpoint triggers.  PC should point to the address
>       stored in #6.

Like this?  This "hopefully" is failing on ARMv7-M currently.

Passes on x86-64 native and gdbserver.

Also uploaded to:

  git@github.com:palves/gdb.git palves/hw_watch_test

----------->8-----------
>From e4035b618fdca66038bae8fd51b14cc2913f6298 Mon Sep 17 00:00:00 2001
From: Pedro Alves <palves@redhat.com>
Date: Fri, 5 Sep 2014 10:29:08 +0100
Subject: [PATCH] Add test to make sure GDB knows which "kind" of watchpoint
 the target has

This adds a test that makes sure GDB knows whether the target has
continuable, or non-continuable watchpoints.

That is, the test confirms that GDB presents a watchpoint value change
at the first instruction right after the instruction that changes
memory.

gdb/testsuite/ChangeLog:

2014-09-05  Pedro Alves  <palves@redhat.com>

	* gdb.base/watchpoint-stops-at-right-insn.c: New file.
	* gdb.base/watchpoint-stops-at-right-insn.exp: New file.
---
 .../gdb.base/watchpoint-stops-at-right-insn.c      |  33 ++++
 .../gdb.base/watchpoint-stops-at-right-insn.exp    | 177 +++++++++++++++++++++
 2 files changed, 210 insertions(+)
 create mode 100644 gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.c
 create mode 100644 gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.exp

diff --git a/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.c b/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.c
new file mode 100644
index 0000000..005d63f
--- /dev/null
+++ b/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.c
@@ -0,0 +1,33 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2014 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/>.  */
+
+volatile int global;
+
+void
+set_global (int val)
+{
+  global = val;
+}
+
+int
+main (void)
+{
+  set_global (1);
+  set_global (2);
+
+  return 0;
+}
diff --git a/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.exp b/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.exp
new file mode 100644
index 0000000..9ad6080
--- /dev/null
+++ b/gdb/testsuite/gdb.base/watchpoint-stops-at-right-insn.exp
@@ -0,0 +1,177 @@
+# This testcase is part of GDB, the GNU debugger.
+
+# Copyright 2014 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/>.
+
+# This file is part of the gdb testsuite.
+
+# Test that GDB presents a hardware watchpoint stop at the first
+# instruction right after the instruction that changes memory.
+#
+# Some targets trigger a hardware watchpoint after the instruction
+# that wrote memory executes, thus with the memory already changed and
+# the PC pointing to the instruction after the instruction that wrote
+# to memory.  These targets are said to have "continuable"
+# watchpoints, referring to the fact that to make progress after the
+# watchpoint triggers, GDB just needs to continue the target.
+#
+# Other targets trigger a hardware watchpoint at the instruction which
+# has attempted to write to the piece of memory under control of the
+# watchpoint, with the instruction actually not executed yet.  To be
+# able to check whether the watched value changed, GDB needs to
+# complete the memory write, single-stepping the target once.  These
+# targets are said to have "non-continuable" watchpoints.
+#
+# This test makes sure that GDB knows which kind of watchpoint the
+# target has, using this sequence of steps:
+#
+# 1 - run to main
+#
+# 2 - set a software watchpoint
+#
+# 3 - continue until watchpoint triggers
+#
+# 4 - the PC now points to the instruction right after the instruction
+#     that actually caused the memory write.  So this is the address a
+#     hardware watchpoint should present the stop to the user too.
+#     Store the PC address.
+#
+# 5 - replace the software watchpoint by a hardware watchpoint
+#
+# 6 - continue until hardware watchpoint triggers
+#
+# 7 - the PC must point to the same address the software watchpoint
+#     triggered at.
+#
+# If the target has continuable watchpoints, but GDB thinks it has
+# non-continuable watchpoints, GDB will stop the inferior two
+# instructions after the watched value change, rather than at the next
+# instruction.
+#
+# If the target has non-continuable watchpoints, while GDB thinks it
+# has continuable watchpoints, GDB will see a watchpoint trigger,
+# notice no value changed, and immediatly continue the target.  Now,
+# either the target manages to step-over the watchpoint transparently,
+# and GDB thus fails to present to value change to the user, or, the
+# watchpoint will keep re-triggering, with the program never making
+# any progress.
+
+standard_testfile
+
+# No use testing this if we can't use hardware watchpoints.
+if {[target_info exists gdb,no_hardware_watchpoints]} {
+    return -1
+}
+
+if { [prepare_for_testing ${testfile}.exp ${testfile} ${srcfile}] } {
+    untested ${testfile}.exp
+    return -1
+}
+
+if { ![runto main] } then {
+    fail "run to main"
+    return
+}
+
+# Get the current PC.  TEST is used as test prefix.
+
+proc get_pc {test} {
+    global hex gdb_prompt
+
+    set addr ""
+    gdb_test_multiple "p /x \$pc" "$test" {
+	-re " = ($hex).*$gdb_prompt $" {
+	    set addr $expect_out(1,string)
+	    pass "$test"
+	}
+    }
+
+    return $addr
+}
+
+# So we get an immediate warning/error if the target doesn't support a
+# given watchpoint type.
+gdb_test_no_output "set breakpoint always-inserted on"
+
+set hw_watchpoints_supported 0
+
+set test "set probe hw watchpoint"
+gdb_test_multiple "watch global" $test {
+    -re "You may have requested too many.*$gdb_prompt $" {
+	pass $test
+    }
+    -re "Target does not support.*$gdb_prompt $" {
+	pass $test
+    }
+    -re "$gdb_prompt $" {
+	pass $test
+	set hw_watchpoints_supported 1
+    }
+}
+
+if {!$hw_watchpoints_supported} {
+    unsupported "no hw watchpoints support"
+    return
+}
+
+delete_breakpoints
+
+proc test {always_inserted} {
+    global srcfile binfile
+
+    with_test_prefix "always-inserted $always_inserted" {
+
+	clean_restart $binfile
+
+	if { ![runto main] } then {
+	    fail "run to main"
+	    return
+	}
+
+	# Force use of software watchpoints.
+	gdb_test_no_output "set can-use-hw-watchpoints 0"
+
+	gdb_test "watch global" \
+	    "Watchpoint .*: global" \
+	    "set software watchpoint on global variable"
+
+	gdb_test "continue" \
+	    "Watchpoint .*: global.*Old value = 0.*New value = 1.*set_global \\(val=1\\).*$srcfile.*" \
+	    "software watchpoint triggers"
+
+	set sw_watch_pc [get_pc "get sw watchpoint PC"]
+
+	delete_breakpoints
+
+	# Allow hardware watchpoints again.
+	gdb_test_no_output "set can-use-hw-watchpoints 1"
+
+	gdb_test "watch global" \
+	    "Hardware watchpoint .*: global" \
+	    "set hardware watchpoint on global variable"
+
+	gdb_test "continue" \
+	    "Hardware watchpoint .*: global.*Old value = 1.*New value = 2.*set_global \\(val=2\\).*$srcfile.*" \
+	    "hardware watchpoint triggers"
+
+	set hw_watch_pc [get_pc "get hw watchpoint PC"]
+
+	gdb_assert {$sw_watch_pc == $sw_watch_pc} "hw watchpoint stops at right instruction"
+    }
+}
+
+foreach always_inserted {"off" "on" } {
+    test $always_inserted
+}
-- 
1.9.3


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]