diff mbox series

Fix casting in-memory values of primitive types to const reference

Message ID 20240320160844.106-1-ssbssa@yahoo.de
State New
Headers
Series Fix casting in-memory values of primitive types to const reference |

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, 4:08 p.m. UTC 
  It's currently not possible to cast an in-memory value of a primitive
type to const reference:
```
(gdb) p Q.id
$1 = 42
(gdb) p (int&)Q.id
$2 = (int &) @0x22fd0c: 42
(gdb) p (const int&)Q.id
Attempt to take address of value not located in memory.
```

And if in a function call an argument needs the same kind of casting,
it also doesn't work:
```
(gdb) l f3
39      int f3(const int &i)
40      {
41        return i;
42      }
(gdb) p f3(Q.id)
Attempt to take address of value not located in memory.
```

It's because when the constness of the type changes in a call to
value_cast, a new not_lval value is allocated, which doesn't exist
in the target memory.

Fixed by ignoring const/volatile/restrict qualifications in
value_cast when comparing cast type to original type, so the new
value will point to the same location as the original value:
```
(gdb) p (int&)i
$2 = (int &) @0x39f72c: 1
(gdb) p (const int&)i
$3 = (const int &) @0x39f72c: 1
(gdb) p f3(Q.id)
$4 = 42
```

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=19423
---
 gdb/testsuite/gdb.cp/casts.exp      | 3 +++
 gdb/testsuite/gdb.cp/ref-params.cc  | 6 ++++++
 gdb/testsuite/gdb.cp/ref-params.exp | 1 +
 gdb/valops.c                        | 3 ++-
 4 files changed, 12 insertions(+), 1 deletion(-)

Comments

Tom Tromey March 20, 2024, 5:05 p.m. UTC | #1 
>>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:

Hannes> It's currently not possible to cast an in-memory value of a primitive
Hannes> type to const reference:

Thank you for the patch.

I think it is ok.  However I suspect there may be other bugs here.

Approved-By: Tom Tromey <tom@tromey.com>

Hannes> Fixed by ignoring const/volatile/restrict qualifications in
Hannes> value_cast when comparing cast type to original type, so the new
Hannes> value will point to the same location as the original value:

Hannes> -  if (types_deeply_equal (arg2->type (), type))
Hannes> +  if (types_deeply_equal (make_unqualified_type (arg2->type ()),
Hannes> +			  make_unqualified_type (type)))

This code seems slightly weird in that it isn't calling check_typedef.
So, I wonder what happens if typedefs are involved, for example if you
did:

typedef const int ci;

(gdb) print (ci&) Q.id

thanks,
Tom
Hannes Domani March 20, 2024, 5:25 p.m. UTC | #2 
Am Mittwoch, 20. März 2024 um 18:05:48 MEZ hat Tom Tromey <tom@tromey.com> Folgendes geschrieben:

> >>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:
>
> Hannes> It's currently not possible to cast an in-memory value of a primitive
> Hannes> type to const reference:
>
> Thank you for the patch.
>
> I think it is ok.  However I suspect there may be other bugs here.
>
> Approved-By: Tom Tromey <tom@tromey.com>

Pushed, thanks.


> Hannes> Fixed by ignoring const/volatile/restrict qualifications in
> Hannes> value_cast when comparing cast type to original type, so the new
> Hannes> value will point to the same location as the original value:
>
> Hannes> -  if (types_deeply_equal (arg2->type (), type))
> Hannes> +  if (types_deeply_equal (make_unqualified_type (arg2->type ()),
> Hannes> +              make_unqualified_type (type)))
>
> This code seems slightly weird in that it isn't calling check_typedef.
> So, I wonder what happens if typedefs are involved, for example if you
> did:
>
> typedef const int ci;
>
> (gdb) print (ci&) Q.id

This happens:

(gdb) p (ci&) Q.id
$1 = (const int &) @0x3cfb8c: 42

Because check_typedef is called when the reference is removed:
```
  /* Check if we are casting struct reference to struct reference.  */
  if (TYPE_IS_REFERENCE (check_typedef (type)))
    {
      /* We dereference type; then we recurse and finally
     we generate value of the given reference.  Nothing wrong with
     that.  */
      struct type *t1 = check_typedef (type);
      struct type *dereftype = check_typedef (t1->target_type ());
      struct value *val = value_cast (dereftype, arg2);

      return value_ref (val, t1->code ());
    }
```


Hannes
Tom Tromey March 20, 2024, 5:33 p.m. UTC | #3 
>>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:

Hannes> This happens:

Hannes> (gdb) p (ci&) Q.id
Hannes> $1 = (const int &) @0x3cfb8c: 42

Hannes> Because check_typedef is called when the reference is removed:

Thanks for looking.

Tom
diff mbox series

Patch

diff --git a/gdb/testsuite/gdb.cp/casts.exp b/gdb/testsuite/gdb.cp/casts.exp
index ca82ab084b9..9f7638c8aee 100644
--- a/gdb/testsuite/gdb.cp/casts.exp
+++ b/gdb/testsuite/gdb.cp/casts.exp
@@ -180,6 +180,9 @@  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 (const int &) gd.left" \
+    " = \\(const int \\&\\) @$nonzero_hex: 23"
+
 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"
diff --git a/gdb/testsuite/gdb.cp/ref-params.cc b/gdb/testsuite/gdb.cp/ref-params.cc
index f038d71fe10..3ef28688607 100644
--- a/gdb/testsuite/gdb.cp/ref-params.cc
+++ b/gdb/testsuite/gdb.cp/ref-params.cc
@@ -36,6 +36,11 @@  int f2(Child& C)
   return f1(C);			/* Set breakpoint marker2 here.  */
 }
 
+int f3(const int &i)
+{
+  return i;
+}
+
 struct OtherParent {
   OtherParent (int other_id0) : other_id(other_id0) { }
   int other_id;
@@ -64,6 +69,7 @@  int main(void)
 
   f2(Q);
   f2(QR);
+  f3(Q.id);
 
   MultiChild MQ(53);
   MultiChild& MQR = MQ;
diff --git a/gdb/testsuite/gdb.cp/ref-params.exp b/gdb/testsuite/gdb.cp/ref-params.exp
index 03bb8e62496..e5c28e6e2ad 100644
--- a/gdb/testsuite/gdb.cp/ref-params.exp
+++ b/gdb/testsuite/gdb.cp/ref-params.exp
@@ -62,3 +62,4 @@  gdb_test "print mf2(MQ)" ".* = 106"
 gdb_test "print f1(MQR)" ".* = 53"
 gdb_test "print mf1(MQR)" ".* = 106"
 gdb_test "print mf2(MQR)" ".* = 106"
+gdb_test "print f3(Q.id)" ".* = 42"
diff --git a/gdb/valops.c b/gdb/valops.c
index 0a4e57672ce..14c12142112 100644
--- a/gdb/valops.c
+++ b/gdb/valops.c
@@ -411,7 +411,8 @@  value_cast (struct type *type, struct value *arg2)
      In this case we want to preserve the LVAL of ARG2 as this allows the
      resulting value to be used in more places.  We do this by calling
      VALUE_COPY if appropriate.  */
-  if (types_deeply_equal (arg2->type (), type))
+  if (types_deeply_equal (make_unqualified_type (arg2->type ()),
+			  make_unqualified_type (type)))
     {
       /* If the types are exactly equal then we can avoid creating a new
 	 value completely.  */