* [PATCH v2 1/2] [gdb] Factor out is_dynamic_type_internal_1
2026-03-11 11:55 [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
@ 2026-03-11 11:55 ` Tom de Vries
2026-03-11 11:55 ` [PATCH v2 2/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
` (2 subsequent siblings)
3 siblings, 0 replies; 6+ messages in thread
From: Tom de Vries @ 2026-03-11 11:55 UTC (permalink / raw)
To: gdb-patches
Simplify is_dynamic_type_internal by factoring out is_dynamic_type_internal_1,
leaving only the handling of the top_level parameter in
is_dynamic_type_internal.
---
gdb/gdbtypes.c | 31 +++++++++++++++++++++----------
1 file changed, 21 insertions(+), 10 deletions(-)
diff --git a/gdb/gdbtypes.c b/gdb/gdbtypes.c
index f3422fe5609..00ef8122978 100644
--- a/gdb/gdbtypes.c
+++ b/gdb/gdbtypes.c
@@ -1913,15 +1913,10 @@ array_type_has_dynamic_stride (struct type *type)
/* Worker for is_dynamic_type. */
static bool
-is_dynamic_type_internal (struct type *type, bool top_level)
+is_dynamic_type_internal_1 (struct type *type)
{
type = check_typedef (type);
- /* We only want to recognize references and pointers at the outermost
- level. */
- if (top_level && type->is_pointer_or_reference ())
- type = check_typedef (type->target_type ());
-
/* Types that have a dynamic TYPE_DATA_LOCATION are considered
dynamic, even if the type itself is statically defined.
From a user's point of view, this may appear counter-intuitive;
@@ -1957,7 +1952,7 @@ is_dynamic_type_internal (struct type *type, bool top_level)
of the range type are static. It allows us to assume that
the subtype of a static range type is also static. */
return (!has_static_range (type->bounds ())
- || is_dynamic_type_internal (type->target_type (), false));
+ || is_dynamic_type_internal_1 (type->target_type ()));
}
case TYPE_CODE_STRING:
@@ -1968,10 +1963,10 @@ is_dynamic_type_internal (struct type *type, bool top_level)
gdb_assert (type->num_fields () == 1);
/* The array is dynamic if either the bounds are dynamic... */
- if (is_dynamic_type_internal (type->index_type (), false))
+ if (is_dynamic_type_internal_1 (type->index_type ()))
return true;
/* ... or the elements it contains have a dynamic contents... */
- if (is_dynamic_type_internal (type->target_type (), false))
+ if (is_dynamic_type_internal_1 (type->target_type ()))
return true;
/* ... or if it has a dynamic stride... */
if (array_type_has_dynamic_stride (type))
@@ -1992,7 +1987,7 @@ is_dynamic_type_internal (struct type *type, bool top_level)
if (f.is_static ())
continue;
/* If the field has dynamic type, then so does TYPE. */
- if (is_dynamic_type_internal (f.type (), false))
+ if (is_dynamic_type_internal_1 (f.type ()))
return true;
/* If the field is at a fixed offset, then it is not
dynamic. */
@@ -2012,6 +2007,22 @@ is_dynamic_type_internal (struct type *type, bool top_level)
return false;
}
+/* Worker for is_dynamic_type. If TOP_LEVEL and TYPE is a pointer or a
+ reference to a dynamic type, it is also considered a dynamic type. */
+
+static bool
+is_dynamic_type_internal (struct type *type, bool top_level)
+{
+ type = check_typedef (type);
+
+ /* We only want to recognize references and pointers at the outermost
+ level. */
+ if (top_level && type->is_pointer_or_reference ())
+ type = check_typedef (type->target_type ());
+
+ return is_dynamic_type_internal_1 (type);
+}
+
/* See gdbtypes.h. */
bool
--
2.51.0
^ permalink raw reply [flat|nested] 6+ messages in thread* [PATCH v2 2/2] [gdb] Enable ptype /o for some dynamic types
2026-03-11 11:55 [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
2026-03-11 11:55 ` [PATCH v2 1/2] [gdb] Factor out is_dynamic_type_internal_1 Tom de Vries
@ 2026-03-11 11:55 ` Tom de Vries
2026-03-25 12:48 ` [PING][PATCH v2 0/2] " Tom de Vries
2026-04-02 15:38 ` [PATCH " Guinevere Larsen
3 siblings, 0 replies; 6+ messages in thread
From: Tom de Vries @ 2026-03-11 11:55 UTC (permalink / raw)
To: gdb-patches
Printing the offsets of a struct containing a flexible array member using
"ptype /o" currently fails:
...
$ cat test.c
struct s {
int a;
int b[];
};
struct s foo;
$ gcc -g test.c -c
$ gdb -q -batch test.o -ex "ptype /o struct s"
warning: ptype/o does not work with dynamic types; disabling '/o'
type = struct s {
int a;
int b[];
}
...
This has been the case since gdb 14, containing commit 0c1aa2a0953 ("Disable
ptype/o for dynamic types").
If we revert the commit, we get instead:
...
$ gdb -q -batch test.o -ex "ptype /o struct s"
/* offset | size */ type = struct s {
/* 0 | 4 */ int a;
/* 4 | 0 */ int b[];
/* total size (bytes): 4 */
}
...
which is similar to what pahole prints:
...
struct s {
int a; /* 0 4 */
int b[]; /* 4 0 */
/* size: 4, cachelines: 1, members: 2 */
/* last cacheline: 4 bytes */
};
...
The problem is that the commit uses is_dynamic_type:
...
if (flags.print_offsets && is_dynamic_type (type))
{
warning (_("ptype/o does not work with dynamic types; disabling '/o'"));
flags.print_offsets = 0;
}
...
which is too restrictive.
Fix this by using a new function cannot_print_offsets instead.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=33966
---
gdb/gdbtypes.c | 39 ++++++++++++--
gdb/gdbtypes.h | 4 ++
gdb/testsuite/gdb.ada/ptype-o.exp | 2 +-
gdb/testsuite/gdb.base/ptype-offsets-c.c | 37 +++++++++++++
gdb/testsuite/gdb.base/ptype-offsets-c.exp | 63 ++++++++++++++++++++++
gdb/typeprint.c | 4 +-
6 files changed, 143 insertions(+), 6 deletions(-)
create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.c
create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.exp
diff --git a/gdb/gdbtypes.c b/gdb/gdbtypes.c
index 00ef8122978..8e90ded1d8c 100644
--- a/gdb/gdbtypes.c
+++ b/gdb/gdbtypes.c
@@ -1910,10 +1910,11 @@ array_type_has_dynamic_stride (struct type *type)
return prop != nullptr && prop->is_constant ();
}
-/* Worker for is_dynamic_type. */
+/* Worker for is_dynamic_type/cannot_print_offsets. */
static bool
-is_dynamic_type_internal_1 (struct type *type)
+is_dynamic_type_internal_1 (struct type *type,
+ bool cannot_print_offsets_p = false)
{
type = check_typedef (type);
@@ -1988,7 +1989,24 @@ is_dynamic_type_internal_1 (struct type *type)
continue;
/* If the field has dynamic type, then so does TYPE. */
if (is_dynamic_type_internal_1 (f.type ()))
- return true;
+ {
+ bool last_struct_field_p
+ = (type->code () == TYPE_CODE_STRUCT
+ && i == type->num_fields () - 1);
+ if (cannot_print_offsets_p && last_struct_field_p)
+ {
+ if (f.type ()->code () == TYPE_CODE_STRUCT)
+ /* The last field is a dynamic type and a struct. Check
+ if we can print the offsets for the struct. */
+ return is_dynamic_type_internal_1 (f.type (), true);
+
+ /* The last field is a dynamic type, this is ok to print
+ offsets for. */
+ return false;
+ }
+
+ return true;
+ }
/* If the field is at a fixed offset, then it is not
dynamic. */
if (!f.loc_is_dwarf_block ())
@@ -2031,6 +2049,21 @@ is_dynamic_type (struct type *type)
return is_dynamic_type_internal (type, true);
}
+/* See gdbtypes.h. */
+
+bool
+cannot_print_offsets (struct type *type)
+{
+ type = check_typedef (type);
+
+ /* We only want to recognize references and pointers at the outermost
+ level. */
+ if (type->is_pointer_or_reference ())
+ type = check_typedef (type->target_type ());
+
+ return is_dynamic_type_internal_1 (type, true);
+}
+
static struct type *resolve_dynamic_type_internal
(struct type *type, const property_addr_info *addr_stack,
const frame_info_ptr &frame, bool top_level);
diff --git a/gdb/gdbtypes.h b/gdb/gdbtypes.h
index 68c272d5fe5..29b03221cb2 100644
--- a/gdb/gdbtypes.h
+++ b/gdb/gdbtypes.h
@@ -2644,6 +2644,10 @@ extern struct type *resolve_dynamic_type
"dynamic". */
extern bool is_dynamic_type (struct type *type);
+/* Return true if TYPE cannot be printed using ptype /o. */
+
+extern bool cannot_print_offsets (struct type *type);
+
/* Resolve any dynamic components of FIELD. FIELD is updated.
ADDR_STACK and FRAME are used where necessary to supply information
for the resolution process; see resolve_dynamic_type.
diff --git a/gdb/testsuite/gdb.ada/ptype-o.exp b/gdb/testsuite/gdb.ada/ptype-o.exp
index 3bac7930762..03de262ecc1 100644
--- a/gdb/testsuite/gdb.ada/ptype-o.exp
+++ b/gdb/testsuite/gdb.ada/ptype-o.exp
@@ -37,7 +37,7 @@ foreach_gnat_encoding scenario flags {all minimal} {
"Warning: the current language does not match this frame."
if {$scenario == "minimal"} {
- set exp "ptype/o does not work with dynamic types.*"
+ set exp "ptype/o does not work with this dynamic type.*"
} else {
# In "all" mode this prints nonsense, but at least does not
# crash.
diff --git a/gdb/testsuite/gdb.base/ptype-offsets-c.c b/gdb/testsuite/gdb.base/ptype-offsets-c.c
new file mode 100644
index 00000000000..74ac9ed7b37
--- /dev/null
+++ b/gdb/testsuite/gdb.base/ptype-offsets-c.c
@@ -0,0 +1,37 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2026 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/>. */
+
+struct flexible_array_member
+{
+ int an_int;
+ int fam[];
+};
+
+struct nested_flexible_array_member
+{
+ int another_int;
+ struct flexible_array_member sfam;
+};
+
+int
+main (void)
+{
+ struct flexible_array_member fam;
+ struct nested_flexible_array_member nfam;
+
+ return 0;
+}
diff --git a/gdb/testsuite/gdb.base/ptype-offsets-c.exp b/gdb/testsuite/gdb.base/ptype-offsets-c.exp
new file mode 100644
index 00000000000..cffcbbeb2f7
--- /dev/null
+++ b/gdb/testsuite/gdb.base/ptype-offsets-c.exp
@@ -0,0 +1,63 @@
+# This testcase is part of GDB, the GNU debugger.
+
+# Copyright 2026 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 testcase exercises the "ptype /o" feature, which can be used to
+# print the offsets and sizes of each field of a struct/union.
+#
+# This is similar to ptype-offsets.exp, which uses C++ instead of C.
+
+standard_testfile .c
+
+# Test only works on LP64 targets. That's how we guarantee that the
+# expected holes will be present in the struct.
+if { ![is_lp64_target] } {
+ untested "test work only on lp64 targets"
+ return 0
+}
+
+if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
+ return -1
+}
+
+# Tests handling flexible array member. Regression tests for PR gdb/33966.
+set l {
+ "ptype /o struct flexible_array_member"
+ "/* offset | size */ type = struct flexible_array_member {"
+ "/* 0 | 4 */ int an_int;"
+ "/* 4 | 0 */ int fam[];"
+ ""
+ " /* total size (bytes): 4 */"
+ " }"
+}
+gdb_test "ptype /o struct flexible_array_member" \
+ [string_to_regexp [multi_line {*}$l]]
+
+set l {
+ "/* offset | size */ type = struct nested_flexible_array_member {"
+ "/* 0 | 4 */ int another_int;"
+ "/* 4 | 4 */ struct flexible_array_member {"
+ "/* 4 | 4 */ int an_int;"
+ "/* 8 | 0 */ int fam[];"
+ ""
+ " /* total size (bytes): 4 */"
+ " } sfam;"
+ ""
+ " /* total size (bytes): 8 */"
+ " }"
+}
+gdb_test "ptype /o struct nested_flexible_array_member" \
+ [string_to_regexp [multi_line {*}$l]]
diff --git a/gdb/typeprint.c b/gdb/typeprint.c
index e53a2b21af5..31a9c1814a3 100644
--- a/gdb/typeprint.c
+++ b/gdb/typeprint.c
@@ -451,9 +451,9 @@ whatis_exp (const char *exp, int show)
type = val->type ();
}
- if (flags.print_offsets && is_dynamic_type (type))
+ if (flags.print_offsets && cannot_print_offsets (type))
{
- warning (_("ptype/o does not work with dynamic types; disabling '/o'"));
+ warning (_("ptype/o does not work with this dynamic type; disabling '/o'"));
flags.print_offsets = 0;
}
--
2.51.0
^ permalink raw reply [flat|nested] 6+ messages in thread* [PING][PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types
2026-03-11 11:55 [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
2026-03-11 11:55 ` [PATCH v2 1/2] [gdb] Factor out is_dynamic_type_internal_1 Tom de Vries
2026-03-11 11:55 ` [PATCH v2 2/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
@ 2026-03-25 12:48 ` Tom de Vries
2026-04-02 15:38 ` [PATCH " Guinevere Larsen
3 siblings, 0 replies; 6+ messages in thread
From: Tom de Vries @ 2026-03-25 12:48 UTC (permalink / raw)
To: gdb-patches
On 3/11/26 12:55 PM, Tom de Vries wrote:
> This patch series fixes PR33966.
>
> The first patch refactors is_dynamic_types_internal.
>
> The second patch contains the fix.
>
> Tested on x86_64-linux.
>
Ping.
Thanks,
- Tom
> Changes in v2:
> - committed the first patch of v1
> - updated the first patch to not expose the top_level parameter in
> is_dynamic_types
> - updated the second patch by:
> - hoisting a check
> - moving the tests to a new testcase gdb.base/ptype-offsets-c.exp
>
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=33966
>
> Tom de Vries (2):
> [gdb] Factor out is_dynamic_type_internal_1
> [gdb] Enable ptype /o for some dynamic types
>
> gdb/gdbtypes.c | 68 ++++++++++++++++++----
> gdb/gdbtypes.h | 4 ++
> gdb/testsuite/gdb.ada/ptype-o.exp | 2 +-
> gdb/testsuite/gdb.base/ptype-offsets-c.c | 37 ++++++++++++
> gdb/testsuite/gdb.base/ptype-offsets-c.exp | 63 ++++++++++++++++++++
> gdb/typeprint.c | 4 +-
> 6 files changed, 163 insertions(+), 15 deletions(-)
> create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.c
> create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.exp
>
>
> base-commit: fe97d7c31d7e81eec11243e269139a08632d085f
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types
2026-03-11 11:55 [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types Tom de Vries
` (2 preceding siblings ...)
2026-03-25 12:48 ` [PING][PATCH v2 0/2] " Tom de Vries
@ 2026-04-02 15:38 ` Guinevere Larsen
2026-04-03 13:20 ` Tom de Vries
3 siblings, 1 reply; 6+ messages in thread
From: Guinevere Larsen @ 2026-04-02 15:38 UTC (permalink / raw)
To: Tom de Vries, gdb-patches
On 3/11/26 8:55 AM, Tom de Vries wrote:
> This patch series fixes PR33966.
>
> The first patch refactors is_dynamic_types_internal.
>
> The second patch contains the fix.
>
> Tested on x86_64-linux.
>
> Changes in v2:
> - committed the first patch of v1
> - updated the first patch to not expose the top_level parameter in
> is_dynamic_types
> - updated the second patch by:
> - hoisting a check
> - moving the tests to a new testcase gdb.base/ptype-offsets-c.exp
>
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=33966
>
> Tom de Vries (2):
> [gdb] Factor out is_dynamic_type_internal_1
> [gdb] Enable ptype /o for some dynamic types
>
> gdb/gdbtypes.c | 68 ++++++++++++++++++----
> gdb/gdbtypes.h | 4 ++
> gdb/testsuite/gdb.ada/ptype-o.exp | 2 +-
> gdb/testsuite/gdb.base/ptype-offsets-c.c | 37 ++++++++++++
> gdb/testsuite/gdb.base/ptype-offsets-c.exp | 63 ++++++++++++++++++++
> gdb/typeprint.c | 4 +-
> 6 files changed, 163 insertions(+), 15 deletions(-)
> create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.c
> create mode 100644 gdb/testsuite/gdb.base/ptype-offsets-c.exp
>
>
> base-commit: fe97d7c31d7e81eec11243e269139a08632d085f
Hi Tom!
I took a look at this, and I don't see anything that raises an eyebrow,
and it fixes the test, so
Reviewed-By: Guinevere Larsen <guinevere@redhat.com>
--
Cheers,
Guinevere Larsen
It/she
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH v2 0/2] [gdb] Enable ptype /o for some dynamic types
2026-04-02 15:38 ` [PATCH " Guinevere Larsen
@ 2026-04-03 13:20 ` Tom de Vries
0 siblings, 0 replies; 6+ messages in thread
From: Tom de Vries @ 2026-04-03 13:20 UTC (permalink / raw)
To: Guinevere Larsen, gdb-patches
On 4/2/26 5:38 PM, Guinevere Larsen wrote:
> Hi Tom!
>
> I took a look at this, and I don't see anything that raises an eyebrow,
> and it fixes the test, so
Hi Gwen,
thanks for the review, pushed.
- Tom
> Reviewed-By: Guinevere Larsen <guinevere@redhat.com>
>
> --
^ permalink raw reply [flat|nested] 6+ messages in thread