diff mbox series

Fix reinterpret_cast for classes with multiple inheritance

Message ID 20240320160115.94-1-ssbssa@yahoo.de
State New
Headers
Series Fix reinterpret_cast for classes with multiple inheritance |

Checks

Context Check Description
linaro-tcwg-bot/tcwg_gdb_build--master-aarch64 fail Patch failed to apply
linaro-tcwg-bot/tcwg_gdb_build--master-arm fail Patch failed to apply

Commit Message

Hannes Domani March 20, 2024, 4:01 p.m. UTC 
  Currently a reinterpret_cast may change the pointer value if
multiple inheritance is involved:
```
(gdb) p r
$1 = (Right *) 0x22f75c
(gdb) p reinterpret_cast<LeftRight*>(r)
$2 = (LeftRight *) 0x22f758
```

It's because value_cast is called in this case, which automatically
does up- and downcasting.

Fixed by simply using the target pointer type in a copy of the
original value:
```
(gdb) p r
$1 = (Right *) 0x3bf87c
(gdb) p reinterpret_cast<LeftRight*>(r)
$2 = (LeftRight *) 0x3bf87c
```

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=18861
---
 gdb/testsuite/gdb.cp/casts.cc  |  8 ++++++++
 gdb/testsuite/gdb.cp/casts.exp | 10 ++++++++++
 gdb/valops.c                   | 11 +++++++++--
 3 files changed, 27 insertions(+), 2 deletions(-)

Comments

Tom Tromey March 20, 2024, 4:57 p.m. UTC | #1 
>>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:

Hannes> Currently a reinterpret_cast may change the pointer value if
Hannes> multiple inheritance is involved:

Thanks for the patch.

Hannes> +  else if (dest_code == TYPE_CODE_PTR && arg_code == TYPE_CODE_PTR)
Hannes> +    {
Hannes> +      /* Don't do any up- or downcasting.  */
Hannes> +      result = arg->copy ();
Hannes> +      result->deprecated_set_type (dest_type);
Hannes> +      result->set_enclosing_type (dest_type);
Hannes> +      result->set_pointed_to_offset (0);	/* pai: chk_val */

I don't think anybody knows what this comment means in the other spots
it appears.  I guess "pai" is some gdb maintainer from the distant past,
though I don't know their real name.

Anyway, please remove this comment.

Ok with this change.
Approved-By: Tom Tromey <tom@tromey.com>

Tom
Hannes Domani March 20, 2024, 5:03 p.m. UTC | #2 
Am Mittwoch, 20. März 2024 um 17:57:50 MEZ hat Tom Tromey <tom@tromey.com> Folgendes geschrieben:

> >>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:
>
> Hannes> Currently a reinterpret_cast may change the pointer value if
> Hannes> multiple inheritance is involved:
>
> Thanks for the patch.
>
> Hannes> +  else if (dest_code == TYPE_CODE_PTR && arg_code == TYPE_CODE_PTR)
> Hannes> +    {
> Hannes> +      /* Don't do any up- or downcasting.  */
> Hannes> +      result = arg->copy ();
> Hannes> +      result->deprecated_set_type (dest_type);
> Hannes> +      result->set_enclosing_type (dest_type);
> Hannes> +      result->set_pointed_to_offset (0);    /* pai: chk_val */
>
> I don't think anybody knows what this comment means in the other spots
> it appears.  I guess "pai" is some gdb maintainer from the distant past,
> though I don't know their real name.
>
> Anyway, please remove this comment.
>
> Ok with this change.
> Approved-By: Tom Tromey <tom@tromey.com>

Pushed with this change, thanks.


Hannes
diff mbox series

Patch

diff --git a/gdb/testsuite/gdb.cp/casts.cc b/gdb/testsuite/gdb.cp/casts.cc
index 5c7f9dc8a1c..eacd8bc0a44 100644
--- a/gdb/testsuite/gdb.cp/casts.cc
+++ b/gdb/testsuite/gdb.cp/casts.cc
@@ -88,6 +88,14 @@  main (int argc, char **argv)
   unsigned long long gd_value = (unsigned long long) (std::uintptr_t)&gd;
   unsigned long long r_value = (unsigned long long) (Right *) &gd;
 
+  LeftRight *lr = &gd;
+  Left *l = lr;
+  Right *r = lr;
+  LeftRight *lr_l = reinterpret_cast<LeftRight *>(l);
+  LeftRight *lr_r = reinterpret_cast<LeftRight *>(r);
+  Left *l_lr = reinterpret_cast<Left *>(lr);
+  Right *r_lr = reinterpret_cast<Right *>(lr);
+
   VirtualLeftRight *vlr = new VirtualLeftRight ();
   VirtualLeft *vl = vlr;
   VirtualRight *vr = vlr;
diff --git a/gdb/testsuite/gdb.cp/casts.exp b/gdb/testsuite/gdb.cp/casts.exp
index 7bfc93b1a02..ca82ab084b9 100644
--- a/gdb/testsuite/gdb.cp/casts.exp
+++ b/gdb/testsuite/gdb.cp/casts.exp
@@ -180,6 +180,16 @@  gdb_test "print (unsigned long long) (LeftRight *) (Right *) &gd == gd_value" \
 gdb_test "print (unsigned long long) (LeftRight *) (Right *) r_value == gd_value" \
     " = true"
 
+gdb_test "print reinterpret_cast<LeftRight *>(l) == lr_l" " = true"
+gdb_test "print reinterpret_cast<LeftRight *>(r) == lr_r" " = true"
+gdb_test "print reinterpret_cast<Left *>(lr) == l_lr" " = true"
+gdb_test "print reinterpret_cast<Right *>(lr) == r_lr" " = true"
+
+gdb_test "print &reinterpret_cast<LeftRight &>(*l) == lr_l" " = true"
+gdb_test "print &reinterpret_cast<LeftRight &>(*r) == lr_r" " = true"
+gdb_test "print &reinterpret_cast<Left &>(*lr) == l_lr" " = true"
+gdb_test "print &reinterpret_cast<Right &>(*lr) == r_lr" " = true"
+
 gdb_test "print dynamic_cast<VirtualLeftRight *> (vlr) == vlr" " = true"
 gdb_test "print dynamic_cast<VirtualLeftRight *> (vl) == vlr" " = true"
 gdb_test "print dynamic_cast<VirtualLeftRight *> (vr) == vlr" " = true"
diff --git a/gdb/valops.c b/gdb/valops.c
index be907440a59..0a4e57672ce 100644
--- a/gdb/valops.c
+++ b/gdb/valops.c
@@ -694,10 +694,17 @@  value_reinterpret_cast (struct type *type, struct value *arg)
       || (dest_code == TYPE_CODE_MEMBERPTR && arg_code == TYPE_CODE_INT)
       || (dest_code == TYPE_CODE_INT && arg_code == TYPE_CODE_MEMBERPTR)
       || (dest_code == arg_code
-	  && (dest_code == TYPE_CODE_PTR
-	      || dest_code == TYPE_CODE_METHODPTR
+	  && (dest_code == TYPE_CODE_METHODPTR
 	      || dest_code == TYPE_CODE_MEMBERPTR)))
     result = value_cast (dest_type, arg);
+  else if (dest_code == TYPE_CODE_PTR && arg_code == TYPE_CODE_PTR)
+    {
+      /* Don't do any up- or downcasting.  */
+      result = arg->copy ();
+      result->deprecated_set_type (dest_type);
+      result->set_enclosing_type (dest_type);
+      result->set_pointed_to_offset (0);	/* pai: chk_val */
+    }
   else
     error (_("Invalid reinterpret_cast"));