Skip to contentSkip to conversation

tools/bpf/bpftool: Reset vmlinux BTF after map dump

PATCH Superseded Yichong Chen 2 participants 4 messages Jun 22 1 message not in the archive

Subsystems (2)

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

PATCH 1/2 tools/bpf/bpftool: Reset vmlinux BTF after map dump

map dump frees the global btf_vmlinux object.

In batch mode, a later map dump can reuse the stale pointer.

Reset the global pointer after freeing it.

---
 tools/bpf/bpftool/map.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/bpf/bpftool/map.c +7 -1
index 7ebf7dbcfba4..8aab9ce743cb 100644
@@ -790,6 +790,12 @@ static int maps_have_btf(int *fds, int nb_fds)
 
 static struct btf *btf_vmlinux;
 
+static void free_btf_vmlinux(void)
+{
+	btf__free(btf_vmlinux);
+	btf_vmlinux = NULL;
+}
+
 static int get_map_kv_btf(const struct bpf_map_info *info, struct btf **btf)
 {
 	int err = 0;
@@ -958,7 +964,7 @@ static int do_dump(int argc, char **argv)
 		close(fds[i]);
 exit_free:
 	free(fds);
-	btf__free(btf_vmlinux);
+	free_btf_vmlinux();
 	return err;
 }
 
Signature · 1 line
2.51.0
#2 lore raw
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Pre-existing issues:
- [Critical] Incomplete Fix: print_key_value() frees the global btf_vmlinux object but leaves the pointer dangling, leading to a Use-After-Free in batch mode.
- [High] Severe memory leak in cgroup tree traversal and batch execution.
--

commit c154c8297093a9827de6d90793d321cdcca31123
Author: Yichong Chen <[email protected]>

tools/bpf/bpftool: Reset vmlinux BTF after map dump

map dump frees the global btf_vmlinux object.

In batch mode, a later map dump can reuse the stale pointer.

Reset the global pointer after freeing it.

Quoted 17 lines
tools/bpf/bpftool/map.c +6 -0
index 71a45d96617ed..dc22fe8750467 100644
@@ -790,6 +790,12 @@ static int maps_have_btf(int *fds, int nb_fds)
 
 static struct btf *btf_vmlinux;
 
+static void free_btf_vmlinux(void)
+{
+	btf__free(btf_vmlinux);
+	btf_vmlinux = NULL;
+}
+
 static int get_map_kv_btf(const struct bpf_map_info *info, struct btf **btf)
 {
 	int err = 0;

[ ... ]

tools/bpf/bpftool/map.c static int do_dump(int argc, char **argv) · line 969

Quoted 8 lines
tools/bpf/bpftool/map.c +1 -1
@@ -958,7 +964,7 @@ static int do_dump(int argc, char **argv)
 		close(fds[i]);
 exit_free:
 	free(fds);
-	btf__free(btf_vmlinux);
+	free_btf_vmlinux();
 	return err;
 }

[Severity: Critical]
This is a pre-existing issue, but does this commit leave an incomplete fix?

Looking at print_key_value() in map.c, it fetches the BTF object using get_map_kv_btf(), which aliases the local btf pointer to the global btf_vmlinux cache:

static int get_map_kv_btf(...) {
    ...
    *btf = btf_vmlinux;
    ...
}

print_key_value() then unconditionally frees it:

static void print_key_value(...) {
    ...
    btf__free(btf);
}

Does this destroy the global BTF object but leave the static btf_vmlinux pointer non-NULL? In batch mode, if a subsequent command executes, could get_map_kv_btf() return this dangling pointer and trigger a use-after-free when accessing map data?

[Severity: High]
This isn't a bug introduced by this patch, but while reviewing the use of btf_vmlinux, does this code leak the BTF object in cgroup.c?

In do_show_tree_fn(), which is invoked by nftw() for every directory in the cgroup tree:

static int do_show_tree_fn(...) {
    ...
    btf_vmlinux = libbpf_find_kernel_btf();
    for (i = 0; i < ARRAY_SIZE(cgroup_attach_types); i++)
        show_bpf_progs(cgroup_fd, cgroup_attach_types[i], ftw->level);
    ...
}

Could this unconditionally overwrite the static btf_vmlinux variable without freeing the previous allocation? On systems with many cgroups, wouldn't traversing the tree leak memory and potentially trigger the OOM killer?

Does do_show() in cgroup.c leak the BTF object across batch commands in the same way, since it also fails to free the allocation?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#3 lore raw

PATCH 2/2 tools/bpf/bpftool: Reset vmlinux BTF after struct_ops commands

In reply to a message not in the archive

struct_ops frees the global btf_vmlinux object.

In batch mode, a later struct_ops command can reuse stale state.

Reset the BTF pointer and cached map info state.

---
 tools/bpf/bpftool/struct_ops.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
tools/bpf/bpftool/struct_ops.c +10 -1
index aa43dead249c..4dba7c96e5f6 100644
@@ -21,6 +21,15 @@ static __u32 map_info_alloc_len;
 static struct btf *btf_vmlinux;
 static __s32 map_info_type_id;
 
+static void free_btf_vmlinux(void)
+{
+	btf__free(btf_vmlinux);
+	btf_vmlinux = NULL;
+	map_info_type = NULL;
+	map_info_alloc_len = 0;
+	map_info_type_id = 0;
+}
+
 struct res {
 	unsigned int nr_maps;
 	unsigned int nr_errs;
@@ -642,7 +651,7 @@ int do_struct_ops(int argc, char **argv)
 
 	err = cmd_select(cmds, argc, argv, do_help);
 
-	btf__free(btf_vmlinux);
+	free_btf_vmlinux();
 
 	return err;
 }
Signature · 1 line
2.51.0
#4 lore raw

Re: [PATCH 2/2] tools/bpf/bpftool: Reset vmlinux BTF after struct_ops commands

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Use-After-Free of btf_vmlinux in map.c during batch execution. --

commit 39dc823c4a9209628a9e626b6a80fe537d15b79e
Author: Yichong Chen <[email protected]>

tools/bpf/bpftool: Reset vmlinux BTF after struct_ops commands

This commit fixes a stale state reuse issue in batch mode by correctly resetting the BTF pointer and cached map info state after struct_ops commands.

Quoted 4 lines
tools/bpf/bpftool/struct_ops.c +0 -0
index aa43dead249cb..4dba7c96e5f60 100644

[ ... ]

tools/bpf/bpftool/struct_ops.c int do_struct_ops(int argc, char **argv) · line 654

Quoted 6 lines
tools/bpf/bpftool/struct_ops.c +1 -1
@@ -642,7 +651,7 @@ int do_struct_ops(int argc, char **argv)
 
 	err = cmd_select(cmds, argc, argv, do_help);
 
-	btf__free(btf_vmlinux);
+	free_btf_vmlinux();

[Severity: High]
This is a pre-existing issue, but does a similar use-after-free exist in tools/bpf/bpftool/map.c during batch execution?

Looking at map.c, get_map_kv_btf() can return the static btf_vmlinux pointer without taking a reference:

tools/bpf/bpftool/map.c:get_map_kv_btf() {
    ...
        *btf = btf_vmlinux;
}

Then print_key_value() frees it unconditionally:

tools/bpf/bpftool/map.c:print_key_value() {
    ...
    btf__free(btf);
}

Since it doesn't NULL the static btf_vmlinux pointer (unlike what this patch does for struct_ops.c), could a subsequent map command in batch mode bypass the (!btf_vmlinux) initialization check and reuse the freed pointer?

Quoted 3 lines
tools/bpf/bpftool/struct_ops.c +0 -0
 	return err;
 }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2