| Message ID | 1668482980-16163-1-git-send-email-wangyufen@huawei.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | c453e64cbc9532c0c2edfa999c35d29dad16b8bb |
| Delegated to: | BPF |
| Headers | show |
| Series | [bpf,v2] selftests/bpf: fix memory leak of lsm_cgroup | expand |
On Mon, Nov 14, 2022 at 7:09 PM Wang Yufen <wangyufen@huawei.com> wrote: > > kmemleak reports this issue: > > unreferenced object 0xffff88810b7835c0 (size 32): > comm "test_progs", pid 270, jiffies 4294969007 (age 1621.315s) > hex dump (first 32 bytes): > 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 03 00 00 00 03 00 00 00 0f 00 00 00 00 00 00 00 ................ > backtrace: > [<00000000376cdeab>] kmalloc_trace+0x27/0x110 > [<000000003bcdb3b6>] selinux_sk_alloc_security+0x66/0x110 > [<000000003959008f>] security_sk_alloc+0x47/0x80 > [<00000000e7bc6668>] sk_prot_alloc+0xbd/0x1a0 > [<0000000002d6343a>] sk_alloc+0x3b/0x940 > [<000000009812a46d>] unix_create1+0x8f/0x3d0 > [<000000005ed0976b>] unix_create+0xa1/0x150 > [<0000000086a1d27f>] __sock_create+0x233/0x4a0 > [<00000000cffe3a73>] __sys_socket_create.part.0+0xaa/0x110 > [<0000000007c63f20>] __sys_socket+0x49/0xf0 > [<00000000b08753c8>] __x64_sys_socket+0x42/0x50 > [<00000000b56e26b3>] do_syscall_64+0x3b/0x90 > [<000000009b4871b8>] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > The issue occurs in the following scenarios: > > unix_create1() > sk_alloc() > sk_prot_alloc() > security_sk_alloc() > call_int_hook() > hlist_for_each_entry() > entry1->hook.sk_alloc_security > <-- selinux_sk_alloc_security() succeeded, > <-- sk->security alloced here. > entry2->hook.sk_alloc_security > <-- bpf_lsm_sk_alloc_security() failed > goto out_free; > ... <-- the sk->security not freed, memleak > > The core problem is that the LSM is not yet fully stacked (work is > actively going on in this space) which means that some LSM hooks do > not support multiple LSMs at the same time. To fix, skip the > "EPERM" test when it runs in the environments that already have > non-bpf lsms installed > > Fixes: dca85aac8895 ("selftests/bpf: lsm_cgroup functional test") > Signed-off-by: Wang Yufen <wangyufen@huawei.com> > Cc: Stanislav Fomichev <sdf@google.com> Acked-by: Stanislav Fomichev <sdf@google.com> Thank you! > --- > tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c | 17 +++++++++++++---- > tools/testing/selftests/bpf/progs/lsm_cgroup.c | 8 ++++++++ > 2 files changed, 21 insertions(+), 4 deletions(-) > > diff --git a/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c b/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c > index 1102e4f..f117bfe 100644 > --- a/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c > +++ b/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c > @@ -173,10 +173,12 @@ static void test_lsm_cgroup_functional(void) > ASSERT_EQ(query_prog_cnt(cgroup_fd, NULL), 4, "total prog count"); > ASSERT_EQ(query_prog_cnt(cgroup_fd2, NULL), 1, "total prog count"); > > - /* AF_UNIX is prohibited. */ > - > fd = socket(AF_UNIX, SOCK_STREAM, 0); > - ASSERT_LT(fd, 0, "socket(AF_UNIX)"); > + if (!(skel->kconfig->CONFIG_SECURITY_APPARMOR > + || skel->kconfig->CONFIG_SECURITY_SELINUX > + || skel->kconfig->CONFIG_SECURITY_SMACK)) > + /* AF_UNIX is prohibited. */ > + ASSERT_LT(fd, 0, "socket(AF_UNIX)"); > close(fd); > > /* AF_INET6 gets default policy (sk_priority). */ > @@ -233,11 +235,18 @@ static void test_lsm_cgroup_functional(void) > > /* AF_INET6+SOCK_STREAM > * AF_PACKET+SOCK_RAW > + * AF_UNIX+SOCK_RAW if already have non-bpf lsms installed > * listen_fd > * client_fd > * accepted_fd > */ > - ASSERT_EQ(skel->bss->called_socket_post_create2, 5, "called_create2"); > + if (skel->kconfig->CONFIG_SECURITY_APPARMOR > + || skel->kconfig->CONFIG_SECURITY_SELINUX > + || skel->kconfig->CONFIG_SECURITY_SMACK) > + /* AF_UNIX+SOCK_RAW if already have non-bpf lsms installed */ > + ASSERT_EQ(skel->bss->called_socket_post_create2, 6, "called_create2"); > + else > + ASSERT_EQ(skel->bss->called_socket_post_create2, 5, "called_create2"); > > /* start_server > * bind(ETH_P_ALL) > diff --git a/tools/testing/selftests/bpf/progs/lsm_cgroup.c b/tools/testing/selftests/bpf/progs/lsm_cgroup.c > index 4f2d60b..02c11d1 100644 > --- a/tools/testing/selftests/bpf/progs/lsm_cgroup.c > +++ b/tools/testing/selftests/bpf/progs/lsm_cgroup.c > @@ -7,6 +7,10 @@ > > char _license[] SEC("license") = "GPL"; > > +extern bool CONFIG_SECURITY_SELINUX __kconfig __weak; > +extern bool CONFIG_SECURITY_SMACK __kconfig __weak; > +extern bool CONFIG_SECURITY_APPARMOR __kconfig __weak; > + > #ifndef AF_PACKET > #define AF_PACKET 17 > #endif > @@ -140,6 +144,10 @@ int BPF_PROG(socket_bind2, struct socket *sock, struct sockaddr *address, > int BPF_PROG(socket_alloc, struct sock *sk, int family, gfp_t priority) > { > called_socket_alloc++; > + /* if already have non-bpf lsms installed, EPERM will cause memory leak of non-bpf lsms */ > + if (CONFIG_SECURITY_SELINUX || CONFIG_SECURITY_SMACK || CONFIG_SECURITY_APPARMOR) > + return 1; > + > if (family == AF_UNIX) > return 0; /* EPERM */ > > -- > 1.8.3.1 >
Hello: This patch was applied to bpf/bpf-next.git (master) by Martin KaFai Lau <martin.lau@kernel.org>: On Tue, 15 Nov 2022 11:29:40 +0800 you wrote: > kmemleak reports this issue: > > unreferenced object 0xffff88810b7835c0 (size 32): > comm "test_progs", pid 270, jiffies 4294969007 (age 1621.315s) > hex dump (first 32 bytes): > 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 03 00 00 00 03 00 00 00 0f 00 00 00 00 00 00 00 ................ > backtrace: > [<00000000376cdeab>] kmalloc_trace+0x27/0x110 > [<000000003bcdb3b6>] selinux_sk_alloc_security+0x66/0x110 > [<000000003959008f>] security_sk_alloc+0x47/0x80 > [<00000000e7bc6668>] sk_prot_alloc+0xbd/0x1a0 > [<0000000002d6343a>] sk_alloc+0x3b/0x940 > [<000000009812a46d>] unix_create1+0x8f/0x3d0 > [<000000005ed0976b>] unix_create+0xa1/0x150 > [<0000000086a1d27f>] __sock_create+0x233/0x4a0 > [<00000000cffe3a73>] __sys_socket_create.part.0+0xaa/0x110 > [<0000000007c63f20>] __sys_socket+0x49/0xf0 > [<00000000b08753c8>] __x64_sys_socket+0x42/0x50 > [<00000000b56e26b3>] do_syscall_64+0x3b/0x90 > [<000000009b4871b8>] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > [...] Here is the summary with links: - [bpf,v2] selftests/bpf: fix memory leak of lsm_cgroup https://git.kernel.org/bpf/bpf-next/c/c453e64cbc95 You are awesome, thank you!
diff --git a/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c b/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c index 1102e4f..f117bfe 100644 --- a/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c +++ b/tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c @@ -173,10 +173,12 @@ static void test_lsm_cgroup_functional(void) ASSERT_EQ(query_prog_cnt(cgroup_fd, NULL), 4, "total prog count"); ASSERT_EQ(query_prog_cnt(cgroup_fd2, NULL), 1, "total prog count"); - /* AF_UNIX is prohibited. */ - fd = socket(AF_UNIX, SOCK_STREAM, 0); - ASSERT_LT(fd, 0, "socket(AF_UNIX)"); + if (!(skel->kconfig->CONFIG_SECURITY_APPARMOR + || skel->kconfig->CONFIG_SECURITY_SELINUX + || skel->kconfig->CONFIG_SECURITY_SMACK)) + /* AF_UNIX is prohibited. */ + ASSERT_LT(fd, 0, "socket(AF_UNIX)"); close(fd); /* AF_INET6 gets default policy (sk_priority). */ @@ -233,11 +235,18 @@ static void test_lsm_cgroup_functional(void) /* AF_INET6+SOCK_STREAM * AF_PACKET+SOCK_RAW + * AF_UNIX+SOCK_RAW if already have non-bpf lsms installed * listen_fd * client_fd * accepted_fd */ - ASSERT_EQ(skel->bss->called_socket_post_create2, 5, "called_create2"); + if (skel->kconfig->CONFIG_SECURITY_APPARMOR + || skel->kconfig->CONFIG_SECURITY_SELINUX + || skel->kconfig->CONFIG_SECURITY_SMACK) + /* AF_UNIX+SOCK_RAW if already have non-bpf lsms installed */ + ASSERT_EQ(skel->bss->called_socket_post_create2, 6, "called_create2"); + else + ASSERT_EQ(skel->bss->called_socket_post_create2, 5, "called_create2"); /* start_server * bind(ETH_P_ALL) diff --git a/tools/testing/selftests/bpf/progs/lsm_cgroup.c b/tools/testing/selftests/bpf/progs/lsm_cgroup.c index 4f2d60b..02c11d1 100644 --- a/tools/testing/selftests/bpf/progs/lsm_cgroup.c +++ b/tools/testing/selftests/bpf/progs/lsm_cgroup.c @@ -7,6 +7,10 @@ char _license[] SEC("license") = "GPL"; +extern bool CONFIG_SECURITY_SELINUX __kconfig __weak; +extern bool CONFIG_SECURITY_SMACK __kconfig __weak; +extern bool CONFIG_SECURITY_APPARMOR __kconfig __weak; + #ifndef AF_PACKET #define AF_PACKET 17 #endif @@ -140,6 +144,10 @@ int BPF_PROG(socket_bind2, struct socket *sock, struct sockaddr *address, int BPF_PROG(socket_alloc, struct sock *sk, int family, gfp_t priority) { called_socket_alloc++; + /* if already have non-bpf lsms installed, EPERM will cause memory leak of non-bpf lsms */ + if (CONFIG_SECURITY_SELINUX || CONFIG_SECURITY_SMACK || CONFIG_SECURITY_APPARMOR) + return 1; + if (family == AF_UNIX) return 0; /* EPERM */
kmemleak reports this issue: unreferenced object 0xffff88810b7835c0 (size 32): comm "test_progs", pid 270, jiffies 4294969007 (age 1621.315s) hex dump (first 32 bytes): 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 03 00 00 00 03 00 00 00 0f 00 00 00 00 00 00 00 ................ backtrace: [<00000000376cdeab>] kmalloc_trace+0x27/0x110 [<000000003bcdb3b6>] selinux_sk_alloc_security+0x66/0x110 [<000000003959008f>] security_sk_alloc+0x47/0x80 [<00000000e7bc6668>] sk_prot_alloc+0xbd/0x1a0 [<0000000002d6343a>] sk_alloc+0x3b/0x940 [<000000009812a46d>] unix_create1+0x8f/0x3d0 [<000000005ed0976b>] unix_create+0xa1/0x150 [<0000000086a1d27f>] __sock_create+0x233/0x4a0 [<00000000cffe3a73>] __sys_socket_create.part.0+0xaa/0x110 [<0000000007c63f20>] __sys_socket+0x49/0xf0 [<00000000b08753c8>] __x64_sys_socket+0x42/0x50 [<00000000b56e26b3>] do_syscall_64+0x3b/0x90 [<000000009b4871b8>] entry_SYSCALL_64_after_hwframe+0x63/0xcd The issue occurs in the following scenarios: unix_create1() sk_alloc() sk_prot_alloc() security_sk_alloc() call_int_hook() hlist_for_each_entry() entry1->hook.sk_alloc_security <-- selinux_sk_alloc_security() succeeded, <-- sk->security alloced here. entry2->hook.sk_alloc_security <-- bpf_lsm_sk_alloc_security() failed goto out_free; ... <-- the sk->security not freed, memleak The core problem is that the LSM is not yet fully stacked (work is actively going on in this space) which means that some LSM hooks do not support multiple LSMs at the same time. To fix, skip the "EPERM" test when it runs in the environments that already have non-bpf lsms installed Fixes: dca85aac8895 ("selftests/bpf: lsm_cgroup functional test") Signed-off-by: Wang Yufen <wangyufen@huawei.com> Cc: Stanislav Fomichev <sdf@google.com> --- tools/testing/selftests/bpf/prog_tests/lsm_cgroup.c | 17 +++++++++++++---- tools/testing/selftests/bpf/progs/lsm_cgroup.c | 8 ++++++++ 2 files changed, 21 insertions(+), 4 deletions(-)