Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
* [PATCH] should_validate_memtags: Do not dereference references
@ 2025-08-01 16:39 Keith Seitz
  2025-08-02 22:50 ` Kevin Buettner
  0 siblings, 1 reply; 3+ messages in thread
From: Keith Seitz @ 2025-08-01 16:39 UTC (permalink / raw)
  To: gdb-patches

should_validate_memtags uses value_as_address to evalute
whether an address for a value is tagged. The comments for
that function simply say, "Extract a value as a C pointer."

While that sounds innoncuous, that function calls coerce_array,
which will dereference any references.  This is not what is
desired here.

This can be demonstrated on an MTE-enabled host, such as aarch64-
based Ampere (example taken from tests introduced in this patch):

(gdb) p b.get_foo ()
Could not validate memory tag: Value can't be converted to integer.
$2 = (const foo &) @0xffffffffed88: {m_a = 42}

While the command completes, gdb didn't actually attempt to
evaluate any memory tags.

Fix this by using unpack_pointer instead.

Tested on x86_64 Fedora 40 and aarch64 RHEL 9.6.
---
 gdb/printcmd.c                             |  4 +-
 gdb/testsuite/gdb.cp/method-ref-return.cc  | 42 +++++++++++++
 gdb/testsuite/gdb.cp/method-ref-return.exp | 70 ++++++++++++++++++++++
 gdb/value.c                                |  0
 4 files changed, 115 insertions(+), 1 deletion(-)
 create mode 100644 gdb/testsuite/gdb.cp/method-ref-return.cc
 create mode 100644 gdb/testsuite/gdb.cp/method-ref-return.exp
 mode change 100644 => 100755 gdb/value.c

diff --git a/gdb/printcmd.c b/gdb/printcmd.c
index 19fbc20074e..d1cf72d371d 100644
--- a/gdb/printcmd.c
+++ b/gdb/printcmd.c
@@ -1290,7 +1290,9 @@ should_validate_memtags (gdbarch *gdbarch, struct value *value)
     return false;
 
   /* We do.  Check whether it includes any tags.  */
-  return target_is_address_tagged (gdbarch, value_as_address (value));
+  struct type *val_type = value->type ();
+  const gdb_byte *data = value->contents ().data ();
+  return target_is_address_tagged (gdbarch, unpack_pointer (val_type, data));
 }
 
 /* Helper for parsing arguments for print_command_1.  */
diff --git a/gdb/testsuite/gdb.cp/method-ref-return.cc b/gdb/testsuite/gdb.cp/method-ref-return.cc
new file mode 100644
index 00000000000..4169bfe6908
--- /dev/null
+++ b/gdb/testsuite/gdb.cp/method-ref-return.cc
@@ -0,0 +1,42 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2025 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/>.  */
+
+/* Test that we can access class method/data member via reference.  */
+
+struct foo
+{
+  foo () : m_a (42) {}
+  int get_a () const { return m_a; }
+  int m_a;
+};
+
+struct bar
+{
+  bar () : m_foo () {}
+  const foo &get_foo () const { return m_foo; }
+  foo m_foo;
+};
+
+int
+main (int argc, char *argv[])
+{
+  bar b;
+  const foo &ref = b.get_foo ();
+  int ret = ref.m_a;  // breakpoint here
+  ret += ref.get_a ();
+  return ret;
+}
\ No newline at end of file
diff --git a/gdb/testsuite/gdb.cp/method-ref-return.exp b/gdb/testsuite/gdb.cp/method-ref-return.exp
new file mode 100644
index 00000000000..1ac5ac930f8
--- /dev/null
+++ b/gdb/testsuite/gdb.cp/method-ref-return.exp
@@ -0,0 +1,70 @@
+# Copyright 2025 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/>.
+
+# Test calling methods and accessing members via reference.
+
+require allow_cplus_tests
+
+standard_testfile .cc
+
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug c++}]} {
+    return -1
+}
+
+if {![runto_main]} {
+    return -1
+}
+
+# Set a breakpoint after the bar object is created and the reference is obtained.
+gdb_breakpoint [gdb_get_line_number "breakpoint here"]
+gdb_continue_to_breakpoint "after reference assignment"
+
+# Test that we can call the method through reference and get the expected result.
+gdb_test_multiple "print b.get_foo()" "print method call returning reference" {
+    -re "\\$\[0-9\]+ = \\(const foo &\\) @$hex: \\{m_a = 42\\}\r\n$gdb_prompt $" {
+	pass $gdb_test_name
+    }
+    -re "Could not validate memory tag: Value can't be converted to integer\\." {
+	fail "$gdb_test_name"
+    }
+}
+
+# Test accessing the member through the reference.
+gdb_test "print b.get_foo ().m_a" \
+    "\\$\[0-9\]+ = 42" \
+    "print member access through reference"
+
+# Test calling method on the referenced object.
+gdb_test "print b.get_foo ().get_a()" \
+    "\\$\[0-9\]+ = 42" \
+    "print method call on referenced object"
+
+# Test that the stored reference works correctly.
+gdb_test_multiple "print ref" "print stored reference" {
+    -re "\\$\[0-9\]+ = \\(const foo &\\) @$hex: \\{m_a = 42\\}\r\n$gdb_prompt $" {
+	pass $gdb_test_name
+    }
+    -re "Could not validate memory tag: Value can't be converted to integer\\." {
+	fail "$gdb_test_name"
+    }
+}
+
+gdb_test "print ref.m_a" \
+    "\\$\[0-9\]+ = 42" \
+    "print member through stored reference"
+
+gdb_test "print ref.get_a()" \
+    "\\$\[0-9\]+ = 42" \
+    "print method call through stored reference"
\ No newline at end of file
diff --git a/gdb/value.c b/gdb/value.c
old mode 100644
new mode 100755

base-commit: 8a6f3a0462063fec4c16b5ecf98bc54a702ac3e2
-- 
2.49.0


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] should_validate_memtags: Do not dereference references
  2025-08-01 16:39 [PATCH] should_validate_memtags: Do not dereference references Keith Seitz
@ 2025-08-02 22:50 ` Kevin Buettner
  2025-08-08 18:04   ` Keith Seitz
  0 siblings, 1 reply; 3+ messages in thread
From: Kevin Buettner @ 2025-08-02 22:50 UTC (permalink / raw)
  To: Keith Seitz; +Cc: gdb-patches

On Fri,  1 Aug 2025 09:39:43 -0700
Keith Seitz <keiths@redhat.com> wrote:

> should_validate_memtags uses value_as_address to evalute
> whether an address for a value is tagged. The comments for
> that function simply say, "Extract a value as a C pointer."
> 
> While that sounds innoncuous, that function calls coerce_array,
> which will dereference any references.  This is not what is
> desired here.
> 
> This can be demonstrated on an MTE-enabled host, such as aarch64-
> based Ampere (example taken from tests introduced in this patch):
> 
> (gdb) p b.get_foo ()
> Could not validate memory tag: Value can't be converted to integer.
> $2 = (const foo &) @0xffffffffed88: {m_a = 42}
> 
> While the command completes, gdb didn't actually attempt to
> evaluate any memory tags.
> 
> Fix this by using unpack_pointer instead.
> 
> Tested on x86_64 Fedora 40 and aarch64 RHEL 9.6.
> ---
>  gdb/printcmd.c                             |  4 +-
>  gdb/testsuite/gdb.cp/method-ref-return.cc  | 42 +++++++++++++
>  gdb/testsuite/gdb.cp/method-ref-return.exp | 70 ++++++++++++++++++++++
>  gdb/value.c                                |  0
>  4 files changed, 115 insertions(+), 1 deletion(-)
>  create mode 100644 gdb/testsuite/gdb.cp/method-ref-return.cc
>  create mode 100644 gdb/testsuite/gdb.cp/method-ref-return.exp
>  mode change 100644 => 100755 gdb/value.c

Aside from the mode change of gdb/value.c, this LGTM.  Approved, with
that change undone...

Approved-by: Kevin Buettner <kevinb@redhat.com>


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] should_validate_memtags: Do not dereference references
  2025-08-02 22:50 ` Kevin Buettner
@ 2025-08-08 18:04   ` Keith Seitz
  0 siblings, 0 replies; 3+ messages in thread
From: Keith Seitz @ 2025-08-08 18:04 UTC (permalink / raw)
  To: gdb-patches

On 8/2/25 3:50 PM, Kevin Buettner wrote:
> 
> Aside from the mode change of gdb/value.c, this LGTM.  Approved, with
> that change undone...

Weird but fixed.

> Approved-by: Kevin Buettner <kevinb@redhat.com>
Thank you for the review, I've pushed this.Keith



^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2025-08-08 18:05 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-08-01 16:39 [PATCH] should_validate_memtags: Do not dereference references Keith Seitz
2025-08-02 22:50 ` Kevin Buettner
2025-08-08 18:04   ` Keith Seitz

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox