* [PATCH v5] Allow calling of user-defined function call operators
[not found] <20240611181851.1784-1-ssbssa.ref@yahoo.de>
@ 2024-06-11 18:18 ` Hannes Domani
2024-06-11 19:58 ` Tom Tromey
0 siblings, 1 reply; 3+ messages in thread
From: Hannes Domani @ 2024-06-11 18:18 UTC (permalink / raw)
To: gdb-patches
Currently it's not possible to call user-defined function call
operators, at least not without specifying operator() directly:
```
(gdb) l 1
1 struct S {
2 int operator() (int x) { return x + 5; }
3 };
4
5 int main () {
6 S s;
7
8 return s(23);
9 }
(gdb) p s(10)
Invalid data type for function to be called.
(gdb) p s.operator()(10)
$1 = 15
```
This now looks if an user-defined call operator is available when
trying to 'call' a struct value, and calls it instead, making this
possible:
```
(gdb) p s(10)
$1 = 15
```
The change in operation::evaluate_funcall is to make sure the type
fields are only used for function types, only they use them as the
argument types.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=12213
---
v2:
- Move the logic into evaluate_subexp_do_call, to avoid duplication in
every evaluate_funcall of each operation subclass.
This makes it now work for some cases it didn't in v1, like if it's
called on a class member (`print c.m(5)` in the new test).
- Added tests for other (struct member) operations.
v3:
- Additional comments.
v4:
- Fix resetting ftype after callee changes, plus some tests for this
where noside == EVAL_AVOID_SIDE_EFFECTS.
v5:
- Add test for missing operator().
---
gdb/eval.c | 34 +++++++++++++++++++++++++++++---
gdb/testsuite/gdb.cp/userdef.cc | 22 +++++++++++++++++++++
gdb/testsuite/gdb.cp/userdef.exp | 10 ++++++++++
3 files changed, 63 insertions(+), 3 deletions(-)
diff --git a/gdb/eval.c b/gdb/eval.c
index 6b752e70635..40640ae97be 100644
--- a/gdb/eval.c
+++ b/gdb/eval.c
@@ -588,14 +588,38 @@ evaluate_subexp_do_call (expression *exp, enum noside noside,
{
if (callee == NULL)
error (_("Cannot evaluate function -- may be inlined"));
+
+ type *ftype = callee->type ();
+
+ /* If the callee is a struct, there might be a user-defined function call
+ operator that should be used instead. */
+ std::vector<value *> vals;
+ if (overload_resolution
+ && exp->language_defn->la_language == language_cplus
+ && check_typedef (ftype)->code () == TYPE_CODE_STRUCT)
+ {
+ /* Include space for the `this' pointer at the start. */
+ vals.resize (argvec.size () + 1);
+
+ vals[0] = value_addr (callee);
+ for (int i = 0; i < argvec.size (); ++i)
+ vals[i + 1] = argvec[i];
+
+ int static_memfuncp;
+ find_overload_match (vals, "operator()", METHOD, &vals[0], nullptr,
+ &callee, nullptr, &static_memfuncp, 0, noside);
+ if (!static_memfuncp)
+ argvec = vals;
+
+ ftype = callee->type ();
+ }
+
if (noside == EVAL_AVOID_SIDE_EFFECTS)
{
/* If the return type doesn't look like a function type,
call an error. This can happen if somebody tries to turn
a variable into a function call. */
- type *ftype = callee->type ();
-
if (ftype->code () == TYPE_CODE_INTERNAL_FUNCTION)
{
/* We don't know anything about what the internal
@@ -666,9 +690,13 @@ operation::evaluate_funcall (struct type *expect_type,
struct type *type = callee->type ();
if (type->code () == TYPE_CODE_PTR)
type = type->target_type ();
+ /* If type is a struct, num_fields would refer to the number of
+ members in the type, not the number of arguments. */
+ bool type_has_arguments
+ = type->code () == TYPE_CODE_FUNC || type->code () == TYPE_CODE_METHOD;
for (int i = 0; i < args.size (); ++i)
{
- if (i < type->num_fields ())
+ if (type_has_arguments && i < type->num_fields ())
vals[i] = args[i]->evaluate (type->field (i).type (), exp, noside);
else
vals[i] = args[i]->evaluate_with_coercion (exp, noside);
diff --git a/gdb/testsuite/gdb.cp/userdef.cc b/gdb/testsuite/gdb.cp/userdef.cc
index 774191726f3..7e045e46b3b 100644
--- a/gdb/testsuite/gdb.cp/userdef.cc
+++ b/gdb/testsuite/gdb.cp/userdef.cc
@@ -307,8 +307,21 @@ class Member
{
public:
int z;
+
+ int operator() ();
+ int operator() (int);
};
+int Member::operator() ()
+{
+ return z;
+}
+
+int Member::operator() (int value)
+{
+ return value * z;
+}
+
bool operator== (const Member &m1, const Member &m2)
{
return m1.z == m2.z;
@@ -335,9 +348,11 @@ int main (void)
Container c;
Member mem1, mem2;
int val;
+ Member Container::* mptr = &Container::m;
mem1.z = 5;
mem2.z = 7;
+ c.m.z = 8;
marker1(); // marker1-returns-here
cout << one; // marker1-returns-here
@@ -404,6 +419,13 @@ int main (void)
++three;
cout << "preinc " << three;
+ val = mem1 ();
+ cout << "funcall " << val << endl;
+ val = mem1 (10);
+ cout << "funcall 2 " << val << endl;
+ val = (c.*mptr) (11);
+ cout << "funcall 3 " << val << endl;
+
(*c).z = 1;
return 0;
diff --git a/gdb/testsuite/gdb.cp/userdef.exp b/gdb/testsuite/gdb.cp/userdef.exp
index e96636bef0c..869f4bccdce 100644
--- a/gdb/testsuite/gdb.cp/userdef.exp
+++ b/gdb/testsuite/gdb.cp/userdef.exp
@@ -132,4 +132,14 @@ gdb_test "ptype &*c" "type = (struct|class) Member {(\[\r\n \]+public:)?\[\r\n \
gdb_test "print operator== (mem1, mem2)" " = false"
gdb_test "print operator== (mem1, mem1)" " = true"
+gdb_test "print mem1()" " = 5"
+gdb_test "print mem1(10)" " = 50"
+gdb_test "print (*&mem1)(2)" " = 10"
+gdb_test "print (c.*mptr)(3)" " = 24"
+gdb_test "print (&c)->m(4)" " = 32"
+gdb_test "print c.m(5)" " = 40"
+gdb_test "print sizeof(mem1(0))/sizeof(int)" " = 1"
+gdb_test "ptype mem1(0)" "type = int"
+gdb_test "print one()" "Couldn't find method A1::operator\\(\\)"
+
gdb_exit
--
2.35.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v5] Allow calling of user-defined function call operators
2024-06-11 18:18 ` [PATCH v5] Allow calling of user-defined function call operators Hannes Domani
@ 2024-06-11 19:58 ` Tom Tromey
2024-06-12 13:26 ` Hannes Domani
0 siblings, 1 reply; 3+ messages in thread
From: Tom Tromey @ 2024-06-11 19:58 UTC (permalink / raw)
To: Hannes Domani; +Cc: gdb-patches
>>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:
Hannes> The change in operation::evaluate_funcall is to make sure the type
Hannes> fields are only used for function types, only they use them as the
Hannes> argument types.
Hannes> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=12213
Thank you. This is ok.
Approved-By: Tom Tromey <tom@tromey.com>
Tom
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v5] Allow calling of user-defined function call operators
2024-06-11 19:58 ` Tom Tromey
@ 2024-06-12 13:26 ` Hannes Domani
0 siblings, 0 replies; 3+ messages in thread
From: Hannes Domani @ 2024-06-12 13:26 UTC (permalink / raw)
To: Tom Tromey; +Cc: gdb-patches
Am Dienstag, 11. Juni 2024 um 21:58:10 MESZ hat Tom Tromey <tom@tromey.com> Folgendes geschrieben:
> >>>>> "Hannes" == Hannes Domani <ssbssa@yahoo.de> writes:
>
> Hannes> The change in operation::evaluate_funcall is to make sure the type
> Hannes> fields are only used for function types, only they use them as the
> Hannes> argument types.
>
> Hannes> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=12213
>
> Thank you. This is ok.
> Approved-By: Tom Tromey <tom@tromey.com>
Pushed, thanks.
Hannes
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-06-12 13:26 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <20240611181851.1784-1-ssbssa.ref@yahoo.de>
2024-06-11 18:18 ` [PATCH v5] Allow calling of user-defined function call operators Hannes Domani
2024-06-11 19:58 ` Tom Tromey
2024-06-12 13:26 ` Hannes Domani
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).