Skip to contentSkip to conversation

bpf: fix reuse of DEVMAP

PATCH Superseded Yureka Lilian 3 participants 7 messages Aug 13, 2025

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover bpf: fix reuse of DEVMAP

This patch series includes the previous fix to libbpf for re-using DEVMAP maps, but modifies it to preserve compatibility with older kernels, thanks to the feedback given by Martin KaFai Lau.

Additionally adds a basic selftest covering the re-use of DEVMAP maps, as requested by Eduard Zingerman.

Yureka Lilian (2):
bpf: fix reuse of DEVMAP
bpf: add test for DEVMAP reuse

 tools/lib/bpf/libbpf.c                        | 14 +++-
 .../bpf/prog_tests/pinning_devmap_reuse.c     | 68 +++++++++++++++++++
 .../selftests/bpf/progs/test_pinning_devmap.c | 20 ++++++
 3 files changed, 101 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_pinning_devmap.c
Signature · 1 line
2.50.1
#1 lore raw

PATCH v2 1/2 bpf: fix reuse of DEVMAP

Previously, re-using pinned DEVMAP maps would always fail, because get_map_info on a DEVMAP always returns flags with BPF_F_RDONLY_PROG set, but BPF_F_RDONLY_PROG being set on a map during creation is invalid.

Thus, ignore the BPF_F_RDONLY_PROG flag on both sides when checking for compatibility with an existing DEVMAP.

Ignoring it on both sides ensures that it continues to work on older kernels which don't set BPF_F_RDONLY_PROG on get_map_info.

The same problem is handled in a third-party ebpf library:
- https://github.com/cilium/ebpf/issues/925
- https://github.com/cilium/ebpf/pull/930

Fixes 0cdbb4b09a06 ("devmap: Allow map lookups from eBPF")
---
 tools/lib/bpf/libbpf.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
tools/lib/bpf/libbpf.c +13 -1
index d41ee26b9..049b0c400 100644
@@ -5076,6 +5076,7 @@ static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd)
 {
 	struct bpf_map_info map_info;
 	__u32 map_info_len = sizeof(map_info);
+	__u32 map_flags_for_check = map->def.map_flags;
 	int err;
 
 	memset(&map_info, 0, map_info_len);
@@ -5088,11 +5089,22 @@ static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd)
 		return false;
 	}
 
+	/* get_map_info on a DEVMAP will always return flags with
+	 * BPF_F_RDONLY_PROG set, but it will never be set on a map
+	 * being created.
+	 * Thus, ignore the BPF_F_RDONLY_PROG flag on both sides when
+	 * checking for compatibility with an existing DEVMAP.
+	 */
+	if (map->def.type == BPF_MAP_TYPE_DEVMAP || map->def.type == BPF_MAP_TYPE_DEVMAP_HASH) {
+		map_info.map_flags |= BPF_F_RDONLY_PROG;
+		map_flags_for_check |= BPF_F_RDONLY_PROG;
+	}
+
 	return (map_info.type == map->def.type &&
 		map_info.key_size == map->def.key_size &&
 		map_info.value_size == map->def.value_size &&
 		map_info.max_entries == map->def.max_entries &&
-		map_info.map_flags == map->def.map_flags &&
+		map_info.map_flags == map_flags_for_check &&
 		map_info.map_extra == map->map_extra);
 }
 
Signature · 1 line
2.50.1
maintainer Reviewed-by on v2 1/2
#2 lore raw
Quoted 10 lines · Yureka Lilian

Previously, re-using pinned DEVMAP maps would always fail, because get_map_info on a DEVMAP always returns flags with BPF_F_RDONLY_PROG set, but BPF_F_RDONLY_PROG being set on a map during creation is invalid.

Thus, ignore the BPF_F_RDONLY_PROG flag on both sides when checking for compatibility with an existing DEVMAP.

Ignoring it on both sides ensures that it continues to work on older kernels which don't set BPF_F_RDONLY_PROG on get_map_info.

Quoted 7 lines

The same problem is handled in a third-party ebpf library:
- https://github.com/cilium/ebpf/issues/925
- https://github.com/cilium/ebpf/pull/930

Fixes 0cdbb4b09a06 ("devmap: Allow map lookups from eBPF")

---

[...]

#3 lore raw
In reply to Yureka Lilian
Quoted 37 lines · Yureka Lilian

Previously, re-using pinned DEVMAP maps would always fail, because get_map_info on a DEVMAP always returns flags with BPF_F_RDONLY_PROG set, but BPF_F_RDONLY_PROG being set on a map during creation is invalid.

Thus, ignore the BPF_F_RDONLY_PROG flag on both sides when checking for compatibility with an existing DEVMAP.

Ignoring it on both sides ensures that it continues to work on older kernels which don't set BPF_F_RDONLY_PROG on get_map_info.

The same problem is handled in a third-party ebpf library:
- https://github.com/cilium/ebpf/issues/925
- https://github.com/cilium/ebpf/pull/930

Fixes 0cdbb4b09a06 ("devmap: Allow map lookups from eBPF")
---
 tools/lib/bpf/libbpf.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
tools/lib/bpf/libbpf.c +1 -0
index d41ee26b9..049b0c400 100644
@@ -5076,6 +5076,7 @@ static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd)
 {
        struct bpf_map_info map_info;
        __u32 map_info_len = sizeof(map_info);
+       __u32 map_flags_for_check = map->def.map_flags;
        int err;
 
        memset(&map_info, 0, map_info_len);
@@ -5088,11 +5089,22 @@ static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd)
                return false;
        }
 

tools/lib/bpf/libbpf.c static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd) · line 5101

Quoted 10 lines
tools/lib/bpf/libbpf.c +10 -0
+       /* get_map_info on a DEVMAP will always return flags with
+        * BPF_F_RDONLY_PROG set, but it will never be set on a map
+        * being created.
+        * Thus, ignore the BPF_F_RDONLY_PROG flag on both sides when
+        * checking for compatibility with an existing DEVMAP.
+        */
+       if (map->def.type == BPF_MAP_TYPE_DEVMAP || map->def.type == BPF_MAP_TYPE_DEVMAP_HASH) {
+               map_info.map_flags |= BPF_F_RDONLY_PROG;
+               map_flags_for_check |= BPF_F_RDONLY_PROG;
+       }

can we instead clear BPF_F_RDONLY_PROG in map_info.map_flags? and then keep using map->def.map_flags directly

pw-bot: cr

Quoted 13 lines
tools/lib/bpf/libbpf.c +2 -2
+
        return (map_info.type == map->def.type &&
                map_info.key_size == map->def.key_size &&
                map_info.value_size == map->def.value_size &&
                map_info.max_entries == map->def.max_entries &&
-               map_info.map_flags == map->def.map_flags &&
+               map_info.map_flags == map_flags_for_check &&
                map_info.map_extra == map->map_extra);
 }
 
--

2.50.1

#4 lore raw

PATCH v2 2/2 bpf: add test for DEVMAP reuse

In reply to Yureka Lilian

The test covers basic re-use of a pinned DEVMAP map, with both matching and mismatching parameters.

---
 .../bpf/prog_tests/pinning_devmap_reuse.c     | 68 +++++++++++++++++++
 .../selftests/bpf/progs/test_pinning_devmap.c | 20 ++++++
 2 files changed, 88 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_pinning_devmap.c
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +68 -0
new file mode 100644
index 000000000..06befb03b
@@ -0,0 +1,68 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <test_progs.h>
+
+void test_pinning_devmap_reuse(void)
+{
+	const char *pinpath1 = "/sys/fs/bpf/pinmap1";
+	const char *pinpath2 = "/sys/fs/bpf/pinmap2";
+	const char *file = "./test_pinning_devmap.bpf.o";
+	struct bpf_object *obj1 = NULL, *obj2 = NULL;
+	int err;
+	__u32 duration = 0;
+	DECLARE_LIBBPF_OPTS(bpf_object_open_opts, opts);
+
+	/* load the object a first time */
+	obj1 = bpf_object__open_file(file, NULL);
+	err = libbpf_get_error(obj1);
+	if (CHECK(err, "first open", "err %d\n", err)) {
+		obj1 = NULL;
+		goto out;
+	}
+	err = bpf_object__load(obj1);
+	if (CHECK(err, "first load", "err %d\n", err))
+		goto out;
+
+	/* load the object a second time, re-using the pinned map */
+	obj2 = bpf_object__open_file(file, NULL);
+	if (CHECK(err, "second open", "err %d\n", err)) {
+		obj2 = NULL;
+		goto out;
+	}
+	err = bpf_object__load(obj2);
+	if (CHECK(err, "second load", "err %d\n", err))
+		goto out;
+
+	/* we can close the reference safely without
+	 * the map's refcount falling to 0
+	 */
+	bpf_object__close(obj1);
+	obj1 = NULL;
+
+	/* now, swap the pins */
+	err = renameat2(0, pinpath1, 0, pinpath2, RENAME_EXCHANGE);
+	if (CHECK(err, "swap pins", "err %d\n", err))
+		goto out;
+
+	/* load the object again, this time the re-use should fail */
+	obj1 = bpf_object__open_file(file, NULL);
+	err = libbpf_get_error(obj1);
+	if (CHECK(err, "third open", "err %d\n", err)) {
+		obj1 = NULL;
+		goto out;
+	}
+	err = bpf_object__load(obj1);
+	if (CHECK(err != -EINVAL, "param mismatch load", "err %d\n", err))
+		goto out;
+
+out:
+	unlink(pinpath1);
+	unlink(pinpath2);
+	if (obj1)
+		bpf_object__close(obj1);
+	if (obj2)
+		bpf_object__close(obj2);
+}
tools/testing/selftests/bpf/progs/test_pinning_devmap.c new file +20 -0
new file mode 100644
index 000000000..c855f8f87
@@ -0,0 +1,20 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+
+struct {
+	__uint(type, BPF_MAP_TYPE_DEVMAP);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u32);
+	__uint(pinning, LIBBPF_PIN_BY_NAME);
+} pinmap1 SEC(".maps");
+
+struct {
+	__uint(type, BPF_MAP_TYPE_DEVMAP);
+	__uint(max_entries, 2);
+	__type(key, __u32);
+	__type(value, __u32);
+	__uint(pinning, LIBBPF_PIN_BY_NAME);
+} pinmap2 SEC(".maps");
Signature · 1 line
2.50.1
maintainer
#5 lore raw

Re: [PATCH v2 2/2] bpf: add test for DEVMAP reuse

Quoted 5 lines · Yureka Lilian

The test covers basic re-use of a pinned DEVMAP map, with both matching and mismatching parameters.

---

Thank you for adding the test case, please find a few comments below.

Quoted 21 lines
 .../bpf/prog_tests/pinning_devmap_reuse.c     | 68 +++++++++++++++++++
 .../selftests/bpf/progs/test_pinning_devmap.c | 20 ++++++
 2 files changed, 88 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_pinning_devmap.c
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +9 -0
new file mode 100644
index 000000000..06befb03b
@@ -0,0 +1,68 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <test_progs.h>
+
+void test_pinning_devmap_reuse(void)
+{

tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c line 19

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +10 -0
+	const char *pinpath1 = "/sys/fs/bpf/pinmap1";
+	const char *pinpath2 = "/sys/fs/bpf/pinmap2";
+	const char *file = "./test_pinning_devmap.bpf.o";
+	struct bpf_object *obj1 = NULL, *obj2 = NULL;
+	int err;
+	__u32 duration = 0;
+	DECLARE_LIBBPF_OPTS(bpf_object_open_opts, opts);
+
+	/* load the object a first time */
+	obj1 = bpf_object__open_file(file, NULL);

The test can be simplified by including test_pinning_devmap.skel.h and calling test_pinning_devmap__open_and_load(), thus avoiding separate calls to open_file and load.

tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c line 26

Quoted 7 lines
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +7 -0
+	err = libbpf_get_error(obj1);
+	if (CHECK(err, "first open", "err %d\n", err)) {
+		obj1 = NULL;
+		goto out;
+	}
+	err = bpf_object__load(obj1);
+	if (CHECK(err, "first load", "err %d\n", err))

Please don't use CHECK in new tests.

Quoted 29 lines
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +29 -0
+		goto out;
+
+	/* load the object a second time, re-using the pinned map */
+	obj2 = bpf_object__open_file(file, NULL);
+	if (CHECK(err, "second open", "err %d\n", err)) {
+		obj2 = NULL;
+		goto out;
+	}
+	err = bpf_object__load(obj2);
+	if (CHECK(err, "second load", "err %d\n", err))
+		goto out;
+
+	/* we can close the reference safely without
+	 * the map's refcount falling to 0
+	 */
+	bpf_object__close(obj1);
+	obj1 = NULL;
+
+	/* now, swap the pins */
+	err = renameat2(0, pinpath1, 0, pinpath2, RENAME_EXCHANGE);
+	if (CHECK(err, "swap pins", "err %d\n", err))
+		goto out;
+
+	/* load the object again, this time the re-use should fail */
+	obj1 = bpf_object__open_file(file, NULL);
+	err = libbpf_get_error(obj1);
+	if (CHECK(err, "third open", "err %d\n", err)) {
+		obj1 = NULL;
+		goto out;

tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c line 65

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +10 -0
+	}
+	err = bpf_object__load(obj1);
+	if (CHECK(err != -EINVAL, "param mismatch load", "err %d\n", err))
+		goto out;
+
+out:
+	unlink(pinpath1);
+	unlink(pinpath2);
+	if (obj1)
+		bpf_object__close(obj1);

Nit: bpf_object__close() can handle NULLs.

Quoted 3 lines
tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c new file +3 -0
+	if (obj2)
+		bpf_object__close(obj2);
+}

[...]

maintainer
#6 lore raw
In reply to Yureka Lilian
Quoted 6 lines · Yureka Lilian

This patch series includes the previous fix to libbpf for re-using DEVMAP maps, but modifies it to preserve compatibility with older kernels, thanks to the feedback given by Martin KaFai Lau.

Additionally adds a basic selftest covering the re-use of DEVMAP maps, as requested by Eduard Zingerman.

Nit: could you please use tag libbpf instead of bpf?

Quoted 11 lines

Yureka Lilian (2):
bpf: fix reuse of DEVMAP
bpf: add test for DEVMAP reuse

 tools/lib/bpf/libbpf.c                        | 14 +++-
 .../bpf/prog_tests/pinning_devmap_reuse.c     | 68 +++++++++++++++++++
 .../selftests/bpf/progs/test_pinning_devmap.c | 20 ++++++
 3 files changed, 101 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/pinning_devmap_reuse.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_pinning_devmap.c