Skip to contentSkip to conversation

bpf: File verification with LSM and fsverity

PATCH Superseded Song Liu 5 participants 22 messages Nov 23, 2023 – Nov 28, 2023

Subsystems (10)
Documentation VFS Kselftest BPF +6 more subsystems

v13 of 15 Superseded Latest is v15 →

lore
Share & tools
lore

Song Liu

author maintainer
#0 lore raw

v13 cover bpf: File verification with LSM and fsverity

Changes v12 => v13:
1. Only keep 4/9 through 9/9 of v12, as the first 3 patches already applied;
2. Use new macro __bpf_kfunc_[start|end]_defs().

Changes v11 => v12:
1. Fix typo (data_ptr => sig_ptr) in bpf_get_file_xattr().

Changes v10 => v11:
1. Let __bpf_dynptr_data() return const void *. (Andrii)
2. Optimize code to reuse output from __bpf_dynptr_size(). (Andrii)
3. Add __diag_ignore_all("-Wmissing-declarations") for kfunc definition.
4. Fix an off indentation. (Andrii)

Changes v9 => v10:
1. Remove WARN_ON_ONCE() from check_reg_const_str. (Alexei)

Changes v8 => v9:
1. Fix test_progs kfunc_dynptr_param/dynptr_data_null.

Changes v7 => v8:
1. Do not use bpf_dynptr_slice* in the kernel. Add __bpf_dynptr_data* and use them in ther kernel. (Andrii)

Changes v6 => v7:
1. Change "__const_str" annotation to "__str". (Alexei, Andrii)
2. Add KF_TRUSTED_ARGS flag for both new kfuncs. (KP)
3. Only allow bpf_get_file_xattr() to read xattr with "user." prefix.
4. Add Acked-by from Eric Biggers.

Changes v5 => v6:
1. Let fsverity_init_bpf() return void. (Eric Biggers)
2. Sort things in alphabetic orders. (Eric Biggers)

Changes v4 => v5:
1. Revise commit logs. (Alexei)

Changes v3 => v4:
1. Fix error reported by CI.
2. Update comments of bpf_dynptr_slice* that they may return error pointer.

Changes v2 => v3:
1. Rebase and resolve conflicts.

Changes v1 => v2:
1. Let bpf_get_file_xattr() use const string for arg "name". (Alexei)
2. Add recursion prevention with allowlist. (Alexei)
3. Let bpf_get_file_xattr() use __vfs_getxattr() to avoid recursion, as vfs_getxattr() calls into other LSM hooks.
4. Do not use dynptr->data directly, use helper insteadd. (Andrii)
5. Fixes with bpf_get_fsverity_digest. (Eric Biggers)
6. Add documentation. (Eric Biggers)
7. Fix some compile warnings. (kernel test robot)

This set enables file verification with BPF LSM and fsverity.

In this solution, fsverity is used to provide reliable and efficient hash of files; and BPF LSM is used to implement signature verification (against asymmetric keys), and to enforce access control.

This solution can be used to implement access control in complicated cases. For example: only signed python binary and signed python script and access special files/devices/ports.

Thanks,
Song

Song Liu (6):
bpf: Add kfunc bpf_get_file_xattr
bpf, fsverity: Add kfunc bpf_get_fsverity_digest Documentation/bpf: Add documentation for filesystem kfuncs selftests/bpf: Sort config in alphabetic order selftests/bpf: Add tests for filesystem kfuncs selftests/bpf: Add test that uses fsverity and xattr to sign a file

 Documentation/bpf/fs_kfuncs.rst               |  21 +++
 Documentation/bpf/index.rst                   |   1 +
 fs/verity/fsverity_private.h                  |  10 ++
 fs/verity/init.c                              |   1 +
 fs/verity/measure.c                           |  84 +++++++++
 kernel/trace/bpf_trace.c                      |  63 +++++++
 tools/testing/selftests/bpf/bpf_kfuncs.h      |  10 ++
 tools/testing/selftests/bpf/config            |   3 +-
 .../selftests/bpf/prog_tests/fs_kfuncs.c      | 132 ++++++++++++++
 .../bpf/prog_tests/verify_pkcs7_sig.c         | 163 +++++++++++++++++-
 .../selftests/bpf/progs/test_fsverity.c       |  46 +++++
 .../selftests/bpf/progs/test_get_xattr.c      |  37 ++++
 .../selftests/bpf/progs/test_sig_in_xattr.c   |  82 +++++++++
 .../bpf/progs/test_verify_pkcs7_sig.c         |   8 +-
 .../testing/selftests/bpf/verify_sig_setup.sh |  25 +++
 15 files changed, 677 insertions(+), 9 deletions(-)
 create mode 100644 Documentation/bpf/fs_kfuncs.rst
 create mode 100644 tools/testing/selftests/bpf/prog_tests/fs_kfuncs.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_fsverity.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_get_xattr.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_sig_in_xattr.c

--
2.34.1

author maintainer
#1 lore raw

PATCH v13 1/6 bpf: Add kfunc bpf_get_file_xattr

It is common practice for security solutions to store tags/labels in xattrs. To implement similar functionalities in BPF LSM, add new kfunc bpf_get_file_xattr().

The first use case of bpf_get_file_xattr() is to implement file verifications with asymmetric keys. Specificially, security applications could use fsverity for file hashes and use xattr to store file signatures. (kfunc for fsverity hash will be added in a separate commit.)

Currently, only xattrs with "user." prefix can be read with kfunc bpf_get_file_xattr(). As use cases evolve, we may add a dedicated prefix for bpf_get_file_xattr().

To avoid recursion, bpf_get_file_xattr can be only called from LSM hooks.

---
 kernel/trace/bpf_trace.c | 63 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
kernel/trace/bpf_trace.c +63 -0
index f0b8b7c29126..55758a6fbe90 100644
@@ -24,6 +24,7 @@
 #include <linux/key.h>
 #include <linux/verification.h>
 #include <linux/namei.h>
+#include <linux/fileattr.h>
 
 #include <net/bpf_sk_storage.h>
 
@@ -1431,6 +1432,68 @@ static int __init bpf_key_sig_kfuncs_init(void)
 late_initcall(bpf_key_sig_kfuncs_init);
 #endif /* CONFIG_KEYS */
 
+/* filesystem kfuncs */
+__bpf_kfunc_start_defs();
+
+/**
+ * bpf_get_file_xattr - get xattr of a file
+ * @file: file to get xattr from
+ * @name__str: name of the xattr
+ * @value_ptr: output buffer of the xattr value
+ *
+ * Get xattr *name__str* of *file* and store the output in *value_ptr*.
+ *
+ * For security reasons, only *name__str* with prefix "user." is allowed.
+ *
+ * Return: 0 on success, a negative value on error.
+ */
+__bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,
+				   struct bpf_dynptr_kern *value_ptr)
+{
+	struct dentry *dentry;
+	u32 value_len;
+	void *value;
+
+	if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
+		return -EPERM;
+
+	value_len = __bpf_dynptr_size(value_ptr);
+	value = __bpf_dynptr_data_rw(value_ptr, value_len);
+	if (!value)
+		return -EINVAL;
+
+	dentry = file_dentry(file);
+	return __vfs_getxattr(dentry, dentry->d_inode, name__str, value, value_len);
+}
+
+__bpf_kfunc_end_defs();
+
+BTF_SET8_START(fs_kfunc_set_ids)
+BTF_ID_FLAGS(func, bpf_get_file_xattr, KF_SLEEPABLE | KF_TRUSTED_ARGS)
+BTF_SET8_END(fs_kfunc_set_ids)
+
+static int bpf_get_file_xattr_filter(const struct bpf_prog *prog, u32 kfunc_id)
+{
+	if (!btf_id_set8_contains(&fs_kfunc_set_ids, kfunc_id))
+		return 0;
+
+	/* Only allow to attach from LSM hooks, to avoid recursion */
+	return prog->type != BPF_PROG_TYPE_LSM ? -EACCES : 0;
+}
+
+const struct btf_kfunc_id_set bpf_fs_kfunc_set = {
+	.owner = THIS_MODULE,
+	.set = &fs_kfunc_set_ids,
+	.filter = bpf_get_file_xattr_filter,
+};
+
+static int __init bpf_fs_kfuncs_init(void)
+{
+	return register_btf_kfunc_id_set(BPF_PROG_TYPE_LSM, &bpf_fs_kfunc_set);
+}
+
+late_initcall(bpf_fs_kfuncs_init);
+
 static const struct bpf_func_proto *
 bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 {
Signature · 1 line
2.34.1
author maintainer
#2 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Quoted 77 lines · Song Liu

It is common practice for security solutions to store tags/labels in xattrs. To implement similar functionalities in BPF LSM, add new kfunc bpf_get_file_xattr().

The first use case of bpf_get_file_xattr() is to implement file verifications with asymmetric keys. Specificially, security applications could use fsverity for file hashes and use xattr to store file signatures. (kfunc for fsverity hash will be added in a separate commit.)

Currently, only xattrs with "user." prefix can be read with kfunc bpf_get_file_xattr(). As use cases evolve, we may add a dedicated prefix for bpf_get_file_xattr().

To avoid recursion, bpf_get_file_xattr can be only called from LSM hooks.

---
 kernel/trace/bpf_trace.c | 63 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
kernel/trace/bpf_trace.c +41 -0
index f0b8b7c29126..55758a6fbe90 100644
@@ -24,6 +24,7 @@
 #include <linux/key.h>
 #include <linux/verification.h>
 #include <linux/namei.h>
+#include <linux/fileattr.h>
 
 #include <net/bpf_sk_storage.h>
 
@@ -1431,6 +1432,68 @@ static int __init bpf_key_sig_kfuncs_init(void)
 late_initcall(bpf_key_sig_kfuncs_init);
 #endif /* CONFIG_KEYS */
 
+/* filesystem kfuncs */
+__bpf_kfunc_start_defs();
+
+/**
+ * bpf_get_file_xattr - get xattr of a file
+ * @file: file to get xattr from
+ * @name__str: name of the xattr
+ * @value_ptr: output buffer of the xattr value
+ *
+ * Get xattr *name__str* of *file* and store the output in *value_ptr*.
+ *
+ * For security reasons, only *name__str* with prefix "user." is allowed.
+ *
+ * Return: 0 on success, a negative value on error.
+ */
+__bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,
+                                  struct bpf_dynptr_kern *value_ptr)
+{
+       struct dentry *dentry;
+       u32 value_len;
+       void *value;
+
+       if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
+               return -EPERM;
+
+       value_len = __bpf_dynptr_size(value_ptr);
+       value = __bpf_dynptr_data_rw(value_ptr, value_len);
+       if (!value)
+               return -EINVAL;
+
+       dentry = file_dentry(file);
+       return __vfs_getxattr(dentry, dentry->d_inode, name__str, value, value_len);
+}
+
+__bpf_kfunc_end_defs();
+
+BTF_SET8_START(fs_kfunc_set_ids)
+BTF_ID_FLAGS(func, bpf_get_file_xattr, KF_SLEEPABLE | KF_TRUSTED_ARGS)
+BTF_SET8_END(fs_kfunc_set_ids)
+

kernel/trace/bpf_trace.c static int __init bpf_key_sig_kfuncs_init(void) · line 1484

Quoted 10 lines
kernel/trace/bpf_trace.c +10 -0
+static int bpf_get_file_xattr_filter(const struct bpf_prog *prog, u32 kfunc_id)
+{
+       if (!btf_id_set8_contains(&fs_kfunc_set_ids, kfunc_id))
+               return 0;
+
+       /* Only allow to attach from LSM hooks, to avoid recursion */
+       return prog->type != BPF_PROG_TYPE_LSM ? -EACCES : 0;
+}
+
+const struct btf_kfunc_id_set bpf_fs_kfunc_set = {

Missed static here. If there will be v14, I will fix it here.

Thanks,
Song

Quoted 18 lines
kernel/trace/bpf_trace.c +12 -1
+       .owner = THIS_MODULE,
+       .set = &fs_kfunc_set_ids,
+       .filter = bpf_get_file_xattr_filter,
+};
+
+static int __init bpf_fs_kfuncs_init(void)
+{
+       return register_btf_kfunc_id_set(BPF_PROG_TYPE_LSM, &bpf_fs_kfunc_set);
+}
+
+late_initcall(bpf_fs_kfuncs_init);
+
 static const struct bpf_func_proto *
 bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 {
--

2.34.1

#3 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

In reply to Song Liu
Quoted 9 lines · Song Liu

It is common practice for security solutions to store tags/labels in xattrs. To implement similar functionalities in BPF LSM, add new kfunc bpf_get_file_xattr().

The first use case of bpf_get_file_xattr() is to implement file verifications with asymmetric keys. Specificially, security applications could use fsverity for file hashes and use xattr to store file signatures. (kfunc for fsverity hash will be added in a separate commit.)

Quoted 8 lines

Currently, only xattrs with "user." prefix can be read with kfunc bpf_get_file_xattr(). As use cases evolve, we may add a dedicated prefix for bpf_get_file_xattr().

To avoid recursion, bpf_get_file_xattr can be only called from LSM hooks.

---

Looks ok to me. But see below for a question.

If you ever allow the retrieval of additional extended attributes through bfs_get_file_xattr() or other bpf interfaces we would like to be Cced, please. The xattr stuff is (/me looks for suitable words)...

Over the last months we've moved POSIX_ACL retrieval out of these low-level functions. They now have a dedicated api. The same is going to happen for fscaps as well.

But even with these out of the way we would want the bpf helpers to always maintain an allowlist of retrievable attributes.

Quoted 41 lines
 kernel/trace/bpf_trace.c | 63 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
kernel/trace/bpf_trace.c +23 -0
index f0b8b7c29126..55758a6fbe90 100644
@@ -24,6 +24,7 @@
 #include <linux/key.h>
 #include <linux/verification.h>
 #include <linux/namei.h>
+#include <linux/fileattr.h>
 
 #include <net/bpf_sk_storage.h>
 
@@ -1431,6 +1432,68 @@ static int __init bpf_key_sig_kfuncs_init(void)
 late_initcall(bpf_key_sig_kfuncs_init);
 #endif /* CONFIG_KEYS */
 
+/* filesystem kfuncs */
+__bpf_kfunc_start_defs();
+
+/**
+ * bpf_get_file_xattr - get xattr of a file
+ * @file: file to get xattr from
+ * @name__str: name of the xattr
+ * @value_ptr: output buffer of the xattr value
+ *
+ * Get xattr *name__str* of *file* and store the output in *value_ptr*.
+ *
+ * For security reasons, only *name__str* with prefix "user." is allowed.
+ *
+ * Return: 0 on success, a negative value on error.
+ */
+__bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,
+				   struct bpf_dynptr_kern *value_ptr)
+{
+	struct dentry *dentry;
+	u32 value_len;
+	void *value;
+

kernel/trace/bpf_trace.c static int __init bpf_key_sig_kfuncs_init(void) · line 1466

Quoted 10 lines
kernel/trace/bpf_trace.c +10 -0
+	if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
+		return -EPERM;
+
+	value_len = __bpf_dynptr_size(value_ptr);
+	value = __bpf_dynptr_data_rw(value_ptr, value_len);
+	if (!value)
+		return -EINVAL;
+
+	dentry = file_dentry(file);
+	return __vfs_getxattr(dentry, dentry->d_inode, name__str, value, value_len);

By calling __vfs_getxattr() from bpf_get_file_xattr() you're skipping at least inode_permission() from xattr_permission(). I'm probably just missing or forgot the context. But why is that ok?

Quoted 36 lines
kernel/trace/bpf_trace.c +30 -0
+}
+
+__bpf_kfunc_end_defs();
+
+BTF_SET8_START(fs_kfunc_set_ids)
+BTF_ID_FLAGS(func, bpf_get_file_xattr, KF_SLEEPABLE | KF_TRUSTED_ARGS)
+BTF_SET8_END(fs_kfunc_set_ids)
+
+static int bpf_get_file_xattr_filter(const struct bpf_prog *prog, u32 kfunc_id)
+{
+	if (!btf_id_set8_contains(&fs_kfunc_set_ids, kfunc_id))
+		return 0;
+
+	/* Only allow to attach from LSM hooks, to avoid recursion */
+	return prog->type != BPF_PROG_TYPE_LSM ? -EACCES : 0;
+}
+
+const struct btf_kfunc_id_set bpf_fs_kfunc_set = {
+	.owner = THIS_MODULE,
+	.set = &fs_kfunc_set_ids,
+	.filter = bpf_get_file_xattr_filter,
+};
+
+static int __init bpf_fs_kfuncs_init(void)
+{
+	return register_btf_kfunc_id_set(BPF_PROG_TYPE_LSM, &bpf_fs_kfunc_set);
+}
+
+late_initcall(bpf_fs_kfuncs_init);
+
 static const struct bpf_func_proto *
 bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 {
Signature · 1 line
2.34.1
author maintainer
#4 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Quoted 26 lines · Christian Brauner
Quoted 9 lines · Song Liu

It is common practice for security solutions to store tags/labels in xattrs. To implement similar functionalities in BPF LSM, add new kfunc bpf_get_file_xattr().

The first use case of bpf_get_file_xattr() is to implement file verifications with asymmetric keys. Specificially, security applications could use fsverity for file hashes and use xattr to store file signatures. (kfunc for fsverity hash will be added in a separate commit.)

Quoted 8 lines

Currently, only xattrs with "user." prefix can be read with kfunc bpf_get_file_xattr(). As use cases evolve, we may add a dedicated prefix for bpf_get_file_xattr().

To avoid recursion, bpf_get_file_xattr can be only called from LSM hooks.

---

Looks ok to me. But see below for a question.

If you ever allow the retrieval of additional extended attributes through bfs_get_file_xattr() or other bpf interfaces we would like to be Cced, please. The xattr stuff is (/me looks for suitable words)...

Quoted 6 lines

Over the last months we've moved POSIX_ACL retrieval out of these low-level functions. They now have a dedicated api. The same is going to happen for fscaps as well.

But even with these out of the way we would want the bpf helpers to always maintain an allowlist of retrievable attributes.

Agreed. We will be very specific which attributes are available to bpf helpers/kfuncs.

Quoted 48 lines
Quoted 48 lines
 kernel/trace/bpf_trace.c | 63 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
kernel/trace/bpf_trace.c +29 -0
index f0b8b7c29126..55758a6fbe90 100644
@@ -24,6 +24,7 @@
 #include <linux/key.h>
 #include <linux/verification.h>
 #include <linux/namei.h>
+#include <linux/fileattr.h>
 
 #include <net/bpf_sk_storage.h>
 
@@ -1431,6 +1432,68 @@ static int __init bpf_key_sig_kfuncs_init(void)
 late_initcall(bpf_key_sig_kfuncs_init);
 #endif /* CONFIG_KEYS */
 
+/* filesystem kfuncs */
+__bpf_kfunc_start_defs();
+
+/**
+ * bpf_get_file_xattr - get xattr of a file
+ * @file: file to get xattr from
+ * @name__str: name of the xattr
+ * @value_ptr: output buffer of the xattr value
+ *
+ * Get xattr *name__str* of *file* and store the output in *value_ptr*.
+ *
+ * For security reasons, only *name__str* with prefix "user." is allowed.
+ *
+ * Return: 0 on success, a negative value on error.
+ */
+__bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,
+                                struct bpf_dynptr_kern *value_ptr)
+{
+     struct dentry *dentry;
+     u32 value_len;
+     void *value;
+
+     if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
+             return -EPERM;
+
+     value_len = __bpf_dynptr_size(value_ptr);
+     value = __bpf_dynptr_data_rw(value_ptr, value_len);
+     if (!value)
Quoted 8 lines

kernel/trace/bpf_trace.c static int __init bpf_key_sig_kfuncs_init(void) · line 1466

Quoted 4 lines
kernel/trace/bpf_trace.c +4 -0
+             return -EINVAL;
+
+     dentry = file_dentry(file);
+     return __vfs_getxattr(dentry, dentry->d_inode, name__str, value, value_len);

By calling __vfs_getxattr() from bpf_get_file_xattr() you're skipping at least inode_permission() from xattr_permission(). I'm probably just missing or forgot the context. But why is that ok?

AFAICT, the XATTR_USER_PREFIX above is equivalent to the prefix check in xattr_permission().

For inode_permission(), I think it is not required because we already have the "struct file" of the target file. Did I misunderstand something here?

Thanks,
Song

Quoted 2 lines
Quoted 2 lines
kernel/trace/bpf_trace.c +2 -0
+}
+

[...]

#5 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Quoted 94 lines · Song Liu
Quoted 26 lines · Christian Brauner
Quoted 9 lines · Song Liu

It is common practice for security solutions to store tags/labels in xattrs. To implement similar functionalities in BPF LSM, add new kfunc bpf_get_file_xattr().

The first use case of bpf_get_file_xattr() is to implement file verifications with asymmetric keys. Specificially, security applications could use fsverity for file hashes and use xattr to store file signatures. (kfunc for fsverity hash will be added in a separate commit.)

Quoted 8 lines

Currently, only xattrs with "user." prefix can be read with kfunc bpf_get_file_xattr(). As use cases evolve, we may add a dedicated prefix for bpf_get_file_xattr().

To avoid recursion, bpf_get_file_xattr can be only called from LSM hooks.

---

Looks ok to me. But see below for a question.

If you ever allow the retrieval of additional extended attributes through bfs_get_file_xattr() or other bpf interfaces we would like to be Cced, please. The xattr stuff is (/me looks for suitable words)...

Quoted 6 lines

Over the last months we've moved POSIX_ACL retrieval out of these low-level functions. They now have a dedicated api. The same is going to happen for fscaps as well.

But even with these out of the way we would want the bpf helpers to always maintain an allowlist of retrievable attributes.

Agreed. We will be very specific which attributes are available to bpf helpers/kfuncs.

Quoted 48 lines
Quoted 48 lines
 kernel/trace/bpf_trace.c | 63 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
kernel/trace/bpf_trace.c +29 -0
index f0b8b7c29126..55758a6fbe90 100644
@@ -24,6 +24,7 @@
 #include <linux/key.h>
 #include <linux/verification.h>
 #include <linux/namei.h>
+#include <linux/fileattr.h>
 
 #include <net/bpf_sk_storage.h>
 
@@ -1431,6 +1432,68 @@ static int __init bpf_key_sig_kfuncs_init(void)
 late_initcall(bpf_key_sig_kfuncs_init);
 #endif /* CONFIG_KEYS */
 
+/* filesystem kfuncs */
+__bpf_kfunc_start_defs();
+
+/**
+ * bpf_get_file_xattr - get xattr of a file
+ * @file: file to get xattr from
+ * @name__str: name of the xattr
+ * @value_ptr: output buffer of the xattr value
+ *
+ * Get xattr *name__str* of *file* and store the output in *value_ptr*.
+ *
+ * For security reasons, only *name__str* with prefix "user." is allowed.
+ *
+ * Return: 0 on success, a negative value on error.
+ */
+__bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,
+                                struct bpf_dynptr_kern *value_ptr)
+{
+     struct dentry *dentry;
+     u32 value_len;
+     void *value;
+
+     if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
+             return -EPERM;
+
+     value_len = __bpf_dynptr_size(value_ptr);
+     value = __bpf_dynptr_data_rw(value_ptr, value_len);
+     if (!value)
Quoted 8 lines

kernel/trace/bpf_trace.c static int __init bpf_key_sig_kfuncs_init(void) · line 1466

Quoted 4 lines
kernel/trace/bpf_trace.c +4 -0
+             return -EINVAL;
+
+     dentry = file_dentry(file);
+     return __vfs_getxattr(dentry, dentry->d_inode, name__str, value, value_len);

By calling __vfs_getxattr() from bpf_get_file_xattr() you're skipping at least inode_permission() from xattr_permission(). I'm probably just missing or forgot the context. But why is that ok?

Quoted 6 lines

AFAICT, the XATTR_USER_PREFIX above is equivalent to the prefix check in xattr_permission().

For inode_permission(), I think it is not required because we already have the "struct file" of the target file. Did I misunderstand something here?

I had overlooked that you don't allow writing xattrs. But there's still some issues:

So if you look at the system call interface:

fgetxattr(fd)
-> getxattr()
-> do_getxattr()

      -> vfs_getxattr()
         -> xattr_permission()
         -> __vfs_getxattr()

and io_uring:

do_getxattr()
-> vfs_getxattr()
-> xattr_permission()
-> __vfs_getxattr()

you can see that xattr_permission() is a _read/write-time check_, not an open check. That's because the read/write permissions may depend on what xattr is read/written. Since you don't know what xattr will be read/written at open-time.

So there needs to be a good reason for bpf_get_file_xattr() to deviate from the system call and io_uring interface. And I'd like to hear it, please. :)

I think I might see the argument because you document the helper as "may only be called from BPF LSM function" in which case you're trying to say that bpf_get_file_xattr() is equivalent to a call to __vfs_getxattr() from an LSM to get at it's own security xattr.

But if that's the case you really should have a way to verify that these helpers are only callable from a specific BPF context. Because you otherwise omit read/write-time permission checking when retrieving xattrs which is a potentialy security issue and may be abused by a BPF program to skip permission checks that are otherwise enforced.

Is there a way for BPF to enforce/verify that a function is only called from a specific BPF program? It should be able to recognize that, no? And then refuse to load that BPF program if a helper is called outside it's intended context.

author maintainer
#6 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Hi Christian,

Thanks again for your comments.

Quoted 1 line · Christian Brauner

[...]

Quoted 41 lines
Quoted 7 lines

AFAICT, the XATTR_USER_PREFIX above is equivalent to the prefix check in xattr_permission().

For inode_permission(), I think it is not required because we already have the "struct file" of the target file. Did I misunderstand something here?

I had overlooked that you don't allow writing xattrs. But there's still some issues:

So if you look at the system call interface:

fgetxattr(fd)
-> getxattr()
-> do_getxattr()

      -> vfs_getxattr()
         -> xattr_permission()
         -> __vfs_getxattr()

and io_uring:

do_getxattr()
-> vfs_getxattr()
-> xattr_permission()
-> __vfs_getxattr()

you can see that xattr_permission() is a _read/write-time check_, not an open check. That's because the read/write permissions may depend on what xattr is read/written. Since you don't know what xattr will be read/written at open-time.

So there needs to be a good reason for bpf_get_file_xattr() to deviate from the system call and io_uring interface. And I'd like to hear it, please. :)

I think I might see the argument because you document the helper as "may only be called from BPF LSM function" in which case you're trying to say that bpf_get_file_xattr() is equivalent to a call to __vfs_getxattr() from an LSM to get at it's own security xattr.

Quoted 5 lines

But if that's the case you really should have a way to verify that these helpers are only callable from a specific BPF context. Because you otherwise omit read/write-time permission checking when retrieving xattrs which is a potentialy security issue and may be abused by a BPF program to skip permission checks that are otherwise enforced.

What do you mean by "a specific BPF context"? Current implementation makes sure the helper only works on LSM hooks with "struct file *" in the argument list. Specifically, we can only use them from the following hooks:

security_binder_transfer_file
security_bprm_creds_from_file
security_file_permission
security_file_alloc_security
security_file_free_security
security_file_ioctl
security_mmap_file
security_file_lock
security_file_fcntl
security_file_set_fowner
security_file_receive
security_file_open
security_file_truncate
security_kernel_read_file
security_kernel_post_read_file

Note that, we disallow pointer-walking with the kfunc, so the kfunc is not allowed from hooks with indirect access to "struct file". For example, we cannot use it with security_bprm_creds_for_exec(struct linux_binprm *bprm) as this hook only has bprm, and calling bpf_get_file_xattr(bprm->file) is not allowed.

Quoted 4 lines

Is there a way for BPF to enforce/verify that a function is only called from a specific BPF program? It should be able to recognize that, no? And then refuse to load that BPF program if a helper is called outside it's intended context.

Similarly, I am not quite sure what you mean by "a specific BPF program". My answer to this is probably the same as above.

Going back to xattr_permission itself. AFAICT, it does 3 checks:

1. MAY_WRITE check;
2. prefix check;
3. inode_permission().

We don't need MAY_WRITE check as bpf_get_file_xattr is read only. We have the prefix check embedded in bpf_get_file_xattr():

if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
return -EPERM;

inode_permission() is a little trickier here, which checks against idmap. However, I don't think the check makes sense in the context of LSM. In this case, we have two processes: one security daemon, which owns the BPF LSM program, and a process being monitored. idmap here, from file_mnt_idmap(file), is the idmap from the being monitored process. However, whether the BPF LSM program have the permission to read the xattr should be determined by the security daemon.

Overall, we can technically add xattr_permission() check here. But I don't think that's the right check for the LSM use case.

Does this make sense? Did I miss or misunderstand something?

Thanks,
Song

author maintainer
#7 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Quoted 7 lines · Song Liu

Hi Christian,

Thanks again for your comments.

Quoted 1 line · Christian Brauner

[...]

Quoted 12 lines

Going back to xattr_permission itself. AFAICT, it does 3 checks:

1. MAY_WRITE check;
2. prefix check;
3. inode_permission().

We don't need MAY_WRITE check as bpf_get_file_xattr is read only. We have the prefix check embedded in bpf_get_file_xattr():

if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
return -EPERM;

Quoted 8 lines

inode_permission() is a little trickier here, which checks against idmap. However, I don't think the check makes sense in the context of LSM. In this case, we have two processes: one security daemon, which owns the BPF LSM program, and a process being monitored. idmap here, from file_mnt_idmap(file), is the idmap from the being monitored process. However, whether the BPF LSM program have the permission to read the xattr should be determined by the security daemon.

Maybe we should check against nop_mnt_idmap? Would something like the following make more sense?

Thanks,
Song

kernel/trace/bpf_trace.c +0 -0
index 0019e990408e..62fc51bc57af 100644
@@ -1453,6 +1453,7 @@ __bpf_kfunc int bpf_get_file_xattr(struct file

*file, const char *name__str,

        struct dentry *dentry;
        u32 value_len;
        void *value;

+ int ret;

if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
return -EPERM;
@@ -1463,6 +1464,9 @@ __bpf_kfunc int bpf_get_file_xattr(struct file *file, const char *name__str,

                return -EINVAL;

        dentry = file_dentry(file);
+       ret = inode_permission(&nop_mnt_idmap, dentry->d_inode, MAY_READ);
+       if (ret)
+               return ret;
        return __vfs_getxattr(dentry, dentry->d_inode, name__str,

value, value_len);
}

#8 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

In reply to Song Liu
Quoted 58 lines · Song Liu

Hi Christian,

Thanks again for your comments.

Quoted 1 line · Christian Brauner

[...]

Quoted 41 lines
Quoted 7 lines

AFAICT, the XATTR_USER_PREFIX above is equivalent to the prefix check in xattr_permission().

For inode_permission(), I think it is not required because we already have the "struct file" of the target file. Did I misunderstand something here?

I had overlooked that you don't allow writing xattrs. But there's still some issues:

So if you look at the system call interface:

fgetxattr(fd)
-> getxattr()
-> do_getxattr()

      -> vfs_getxattr()
         -> xattr_permission()
         -> __vfs_getxattr()

and io_uring:

do_getxattr()
-> vfs_getxattr()
-> xattr_permission()
-> __vfs_getxattr()

you can see that xattr_permission() is a _read/write-time check_, not an open check. That's because the read/write permissions may depend on what xattr is read/written. Since you don't know what xattr will be read/written at open-time.

So there needs to be a good reason for bpf_get_file_xattr() to deviate from the system call and io_uring interface. And I'd like to hear it, please. :)

I think I might see the argument because you document the helper as "may only be called from BPF LSM function" in which case you're trying to say that bpf_get_file_xattr() is equivalent to a call to __vfs_getxattr() from an LSM to get at it's own security xattr.

Quoted 5 lines

But if that's the case you really should have a way to verify that these helpers are only callable from a specific BPF context. Because you otherwise omit read/write-time permission checking when retrieving xattrs which is a potentialy security issue and may be abused by a BPF program to skip permission checks that are otherwise enforced.

What do you mean by "a specific BPF context"? Current implementation makes sure the helper only works on LSM hooks with "struct file *" in the argument list. Specifically, we can only use them from the following hooks:

Quoted 15 lines

security_binder_transfer_file
security_bprm_creds_from_file
security_file_permission
security_file_alloc_security
security_file_free_security
security_file_ioctl
security_mmap_file
security_file_lock
security_file_fcntl
security_file_set_fowner
security_file_receive
security_file_open
security_file_truncate
security_kernel_read_file
security_kernel_post_read_file

Ok, good!

Quoted 5 lines

Note that, we disallow pointer-walking with the kfunc, so the kfunc is not allowed from hooks with indirect access to "struct file". For example, we cannot use it with security_bprm_creds_for_exec(struct linux_binprm *bprm) as this hook only has bprm, and calling bpf_get_file_xattr(bprm->file) is not allowed.

Great.

Quoted 8 lines
Quoted 5 lines

Is there a way for BPF to enforce/verify that a function is only called from a specific BPF program? It should be able to recognize that, no? And then refuse to load that BPF program if a helper is called outside it's intended context.

Similarly, I am not quite sure what you mean by "a specific BPF program". My answer to this is probably the same as above.

Yes, this is exactly what I meant.

Quoted 22 lines

Going back to xattr_permission itself. AFAICT, it does 3 checks:

1. MAY_WRITE check;
2. prefix check;
3. inode_permission().

We don't need MAY_WRITE check as bpf_get_file_xattr is read only. We have the prefix check embedded in bpf_get_file_xattr():

if (strncmp(name__str, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN))
return -EPERM;

inode_permission() is a little trickier here, which checks against idmap. However, I don't think the check makes sense in the context of LSM. In this case, we have two processes: one security daemon, which owns the BPF LSM program, and a process being monitored. idmap here, from file_mnt_idmap(file), is the idmap from the being monitored process. However, whether the BPF LSM program have the permission to read the xattr should be determined by the security daemon.

Quoted 4 lines

Overall, we can technically add xattr_permission() check here. But I don't think that's the right check for the LSM use case.

Does this make sense? Did I miss or misunderstand something?

If the helper is only callable from an LSM context then this should be fine.

author maintainer
#9 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Hi Christian,

Quoted 2 lines · Christian Brauner

On Mon, Nov 27, 2023 at 10:05:23AM -0800, Song Liu wrote:

[...]

Quoted 8 lines
Quoted 5 lines

Overall, we can technically add xattr_permission() check here. But I don't think that's the right check for the LSM use case.

Does this make sense? Did I miss or misunderstand something?

If the helper is only callable from an LSM context then this should be fine.

If everything looks good, would you please give an official Acked-by or Reviewed-by?

Thanks,
Song

maintainer Acked-by
#10 lore raw

Re: [PATCH v13 bpf-next 1/6] bpf: Add kfunc bpf_get_file_xattr

Quoted 10 lines · Song Liu

Hi Christian,

Quoted 2 lines · Christian Brauner

On Mon, Nov 27, 2023 at 10:05:23AM -0800, Song Liu wrote:

[...]

Quoted 4 lines
Quoted 4 lines

Overall, we can technically add xattr_permission() check here. But I don't think that's the right check for the LSM use case.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

Does this make sense? Did I miss or misunderstand something?

If the helper is only callable from an LSM context then this should be fine.

If everything looks good, would you please give an official Acked-by or Reviewed-by?

Yeah looks ok to me,

author maintainer
#11 lore raw

PATCH v13 2/6 bpf, fsverity: Add kfunc bpf_get_fsverity_digest

In reply to Song Liu

fsverity provides fast and reliable hash of files, namely fsverity_digest. The digest can be used by security solutions to verify file contents.

Add new kfunc bpf_get_fsverity_digest() so that we can access fsverity from BPF LSM programs. This kfunc is added to fs/verity/measure.c because some data structure used in the function is private to fsverity (fs/verity/fsverity_private.h).

To avoid recursion, bpf_get_fsverity_digest is only allowed in BPF LSM programs.

---
 fs/verity/fsverity_private.h | 10 +++++
 fs/verity/init.c             |  1 +
 fs/verity/measure.c          | 84 ++++++++++++++++++++++++++++++++++++
 3 files changed, 95 insertions(+)
fs/verity/fsverity_private.h +10 -0
index d071a6e32581..a6a6b2749241 100644
@@ -100,6 +100,16 @@ fsverity_msg(const struct inode *inode, const char *level,
 #define fsverity_err(inode, fmt, ...)		\
 	fsverity_msg((inode), KERN_ERR, fmt, ##__VA_ARGS__)
 
+/* measure.c */
+
+#ifdef CONFIG_BPF_SYSCALL
+void __init fsverity_init_bpf(void);
+#else
+static inline void fsverity_init_bpf(void)
+{
+}
+#endif
+
 /* open.c */
 
 int fsverity_init_merkle_tree_params(struct merkle_tree_params *params,
fs/verity/init.c +1 -0
index a29f062f6047..1e207c0f71de 100644
@@ -69,6 +69,7 @@ static int __init fsverity_init(void)
 	fsverity_init_workqueue();
 	fsverity_init_sysctl();
 	fsverity_init_signature();
+	fsverity_init_bpf();
 	return 0;
 }
 late_initcall(fsverity_init)
fs/verity/measure.c +84 -0

93 lines not on this page · raw message

Signature · 1 line
2.34.1
author maintainer
#12 lore raw

PATCH v13 3/6 Documentation/bpf: Add documentation for filesystem kfuncs

In reply to Song Liu

Add a brief introduction for file system kfuncs:

bpf_get_file_xattr()
bpf_get_fsverity_digest()

The documentation highlights the strategy to avoid recursions of these kfuncs.

---
 Documentation/bpf/fs_kfuncs.rst | 21 +++++++++++++++++++++
 Documentation/bpf/index.rst     |  1 +
 2 files changed, 22 insertions(+)
 create mode 100644 Documentation/bpf/fs_kfuncs.rst
Documentation/bpf/fs_kfuncs.rst new file +21 -0
new file mode 100644
index 000000000000..8762c3233a3d
@@ -0,0 +1,21 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+.. _fs_kfuncs-header-label:
+
+=====================
+BPF filesystem kfuncs
+=====================
+
+BPF LSM programs need to access filesystem data from LSM hooks. The following
+BPF kfuncs can be used to get these data.
+
+ * ``bpf_get_file_xattr()``
+
+ * ``bpf_get_fsverity_digest()``
+
+To avoid recursions, these kfuncs follow the following rules:
+
+1. These kfuncs are only permitted from BPF LSM function.
+2. These kfuncs should not call into other LSM hooks, i.e. security_*(). For
+   example, ``bpf_get_file_xattr()`` does not use ``vfs_getxattr()``, because
+   the latter calls LSM hook ``security_inode_getxattr``.
Documentation/bpf/index.rst +1 -0
index aeaeb35e6d4a..0bb5cb8157f1 100644
@@ -21,6 +21,7 @@ that goes into great technical depth about the BPF Architecture.
    helpers
    kfuncs
    cpumasks
+   fs_kfuncs
    programs
    maps
    bpf_prog_run
Signature · 1 line
2.34.1
author maintainer
#13 lore raw

PATCH v13 4/6 selftests/bpf: Sort config in alphabetic order

In reply to Song Liu

Move CONFIG_VSOCKETS up, so the CONFIGs are in alphabetic order.

---
 tools/testing/selftests/bpf/config | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/testing/selftests/bpf/config +1 -1
index 3ec5927ec3e5..782876452acf 100644
@@ -82,7 +82,7 @@ CONFIG_SECURITY=y
 CONFIG_SECURITYFS=y
 CONFIG_TEST_BPF=m
 CONFIG_USERFAULTFD=y
+CONFIG_VSOCKETS=y
 CONFIG_VXLAN=y
 CONFIG_XDP_SOCKETS=y
 CONFIG_XFRM_INTERFACE=y
-CONFIG_VSOCKETS=y
Signature · 1 line
2.34.1
author maintainer
#14 lore raw

PATCH v13 5/6 selftests/bpf: Add tests for filesystem kfuncs

In reply to Song Liu

Add selftests for two new filesystem kfuncs:
1. bpf_get_file_xattr
2. bpf_get_fsverity_digest

These tests simply make sure the two kfuncs work. Another selftest will be added to demonstrate how to use these kfuncs to verify file signature.

CONFIG_FS_VERITY is added to selftests config. However, this is not sufficient to guarantee bpf_get_fsverity_digest works. This is because fsverity need to be enabled at file system level (for example, with tune2fs on ext4). If local file system doesn't have this feature enabled, just skip the test.

---
 tools/testing/selftests/bpf/bpf_kfuncs.h      |   3 +
 tools/testing/selftests/bpf/config            |   1 +
 .../selftests/bpf/prog_tests/fs_kfuncs.c      | 132 ++++++++++++++++++
 .../selftests/bpf/progs/test_fsverity.c       |  46 ++++++
 .../selftests/bpf/progs/test_get_xattr.c      |  37 +++++
 5 files changed, 219 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/fs_kfuncs.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_fsverity.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_get_xattr.c
tools/testing/selftests/bpf/bpf_kfuncs.h +3 -0
index 5ca68ff0b59f..c2c084a44eae 100644
@@ -55,4 +55,7 @@ void *bpf_cast_to_kern_ctx(void *) __ksym;
 
 void *bpf_rdonly_cast(void *obj, __u32 btf_id) __ksym;
 
+extern int bpf_get_file_xattr(struct file *file, const char *name,
+			      struct bpf_dynptr *value_ptr) __ksym;
+extern int bpf_get_fsverity_digest(struct file *file, struct bpf_dynptr *digest_ptr) __ksym;
 #endif
tools/testing/selftests/bpf/config +1 -0
index 782876452acf..c125c441abc7 100644
@@ -23,6 +23,7 @@ CONFIG_FPROBE=y
 CONFIG_FTRACE_SYSCALLS=y
 CONFIG_FUNCTION_ERROR_INJECTION=y
 CONFIG_FUNCTION_TRACER=y
+CONFIG_FS_VERITY=y
 CONFIG_GENEVE=y
 CONFIG_IKCONFIG=y
 CONFIG_IKCONFIG_PROC=y
tools/testing/selftests/bpf/prog_tests/fs_kfuncs.c new file +132 -0

132 lines not on this page · raw message

tools/testing/selftests/bpf/progs/test_fsverity.c new file +46 -0

46 lines not on this page · raw message

tools/testing/selftests/bpf/progs/test_get_xattr.c new file +37 -0
new file mode 100644
index 000000000000..7eb2a4e5a3e5
@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2023 Meta Platforms, Inc. and affiliates. */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+#include "bpf_kfuncs.h"
+
+char _license[] SEC("license") = "GPL";
+
+__u32 monitored_pid;
+__u32 found_xattr;
+
+static const char expected_value[] = "hello";
+char value[32];
+
+SEC("lsm.s/file_open")
+int BPF_PROG(test_file_open, struct file *f)
+{
+	struct bpf_dynptr value_ptr;
+	__u32 pid;
+	int ret;
+
+	pid = bpf_get_current_pid_tgid() >> 32;
+	if (pid != monitored_pid)
+		return 0;
+
+	bpf_dynptr_from_mem(value, sizeof(value), 0, &value_ptr);
+
+	ret = bpf_get_file_xattr(f, "user.kfuncs", &value_ptr);
+	if (ret != sizeof(expected_value))
+		return 0;
+	if (bpf_strncmp(value, ret, expected_value))
+		return 0;
+	found_xattr = 1;
+	return 0;
+}
Signature · 1 line
2.34.1
#15 lore raw

Re: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs

kernel test robot noticed the following build errors:

Hi Song,

kernel test robot noticed the following build errors:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/Song-Liu/bpf-Add-kfunc-bpf_get_file_xattr/20231124-074239
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/20231123233936.3079687-6-song%40kernel.org
patch subject: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20231124/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

>> progs/test_fsverity.c:16:22: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                      ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^
   progs/test_fsverity.c:17:13: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'
      17 | char digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |             ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^

2 errors generated.

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
author maintainer
#16 lore raw

Re: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs

Quoted 33 lines · kernel test robot

Hi Song,

kernel test robot noticed the following build errors:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/Song-Liu/bpf-Add-kfunc-bpf_get_file_xattr/20231124-074239
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/20231123233936.3079687-6-song%40kernel.org
patch subject: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20231124/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

Quoted 1 line
Quoted 1 line

progs/test_fsverity.c:16:22: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'

      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                      ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^
   progs/test_fsverity.c:17:13: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'
      17 | char digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |             ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^
Quoted 1 line

2 errors generated.

In the patch, I added CONFIG_FS_VERITY to the config (tools/testing/selftests/bpf/config). I guess the test robot didn't update the config accordingly?

Thanks,
Song

#17 lore raw

Re: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs

Quoted 36 lines · Song Liu
Quoted 33 lines · kernel test robot

Hi Song,

kernel test robot noticed the following build errors:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/Song-Liu/bpf-Add-kfunc-bpf_get_file_xattr/20231124-074239
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/20231123233936.3079687-6-song%40kernel.org
patch subject: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20231124/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

Quoted 1 line
Quoted 1 line

progs/test_fsverity.c:16:22: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'

      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                      ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^
   progs/test_fsverity.c:17:13: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'
      17 | char digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |             ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_fsverity.c:16:36: note: forward declaration of 'struct fsverity_digest'
      16 | char expected_digest[sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                    ^
Quoted 1 line

2 errors generated.

Quoted 3 lines

In the patch, I added CONFIG_FS_VERITY to the config (tools/testing/selftests/bpf/config). I guess the test robot didn't update the config accordingly?

Thanks for this information. The bot does use a static config to build selftests. Dynamic config updating is not supported yet. We will fix this soon.

Best Regards,
Yujie

#18 lore raw

Re: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs

In reply to Song Liu
Quoted 11 lines · Song Liu
tools/testing/selftests/bpf/progs/test_get_xattr.c new file +10 -0
+static const char expected_value[] = "hello";
+char value[32];
+
+SEC("lsm.s/file_open")
+int BPF_PROG(test_file_open, struct file *f)
+{
+       struct bpf_dynptr value_ptr;
+       __u32 pid;
+       int ret;
+

tools/testing/selftests/bpf/progs/test_get_xattr.c line 33

Quoted 10 lines
tools/testing/selftests/bpf/progs/test_get_xattr.c new file +10 -0
+       pid = bpf_get_current_pid_tgid() >> 32;
+       if (pid != monitored_pid)
+               return 0;
+
+       bpf_dynptr_from_mem(value, sizeof(value), 0, &value_ptr);
+
+       ret = bpf_get_file_xattr(f, "user.kfuncs", &value_ptr);
+       if (ret != sizeof(expected_value))
+               return 0;
+       if (bpf_strncmp(value, ret, expected_value))

Hmm. It doesn't work like:
if (bpf_strncmp(value, ret, "hello"))

?

author maintainer
#19 lore raw

Re: [PATCH v13 bpf-next 5/6] selftests/bpf: Add tests for filesystem kfuncs

Quoted 18 lines · Alexei Starovoitov
Quoted 16 lines · Song Liu
tools/testing/selftests/bpf/progs/test_get_xattr.c new file +15 -0
+static const char expected_value[] = "hello";
+char value[32];
+
+SEC("lsm.s/file_open")
+int BPF_PROG(test_file_open, struct file *f)
+{
+       struct bpf_dynptr value_ptr;
+       __u32 pid;
+       int ret;
+
+       pid = bpf_get_current_pid_tgid() >> 32;
+       if (pid != monitored_pid)
+               return 0;
+
+       bpf_dynptr_from_mem(value, sizeof(value), 0, &value_ptr);
Quoted 8 lines

tools/testing/selftests/bpf/progs/test_get_xattr.c line 33

Quoted 5 lines
tools/testing/selftests/bpf/progs/test_get_xattr.c new file +5 -0
+
+       ret = bpf_get_file_xattr(f, "user.kfuncs", &value_ptr);
+       if (ret != sizeof(expected_value))
+               return 0;
+       if (bpf_strncmp(value, ret, expected_value))

Hmm. It doesn't work like:
if (bpf_strncmp(value, ret, "hello"))

This also works. I used expected_value because there is a size check above. We can also make do something like

if (ret != sizeof("hello"))

             return 0;
if (bpf_strncmp(value, ret, "hello"))
             return 0;

Both of the two work.

Thanks,
Song

author maintainer
#20 lore raw

PATCH v13 6/6 selftests/bpf: Add test that uses fsverity and xattr to sign a file

In reply to Song Liu

This selftests shows a proof of concept method to use BPF LSM to enforce file signature. This test is added to verify_pkcs7_sig, so that some existing logic can be reused.

This file signature method uses fsverity, which provides reliable and efficient hash (known as digest) of the file. The file digest is signed with asymmetic key, and the signature is stored in xattr. At the run time, BPF LSM reads file digest and the signature, and then checks them against the public key.

Note that this solution does NOT require FS_VERITY_BUILTIN_SIGNATURES. fsverity is only used to provide file digest. The signature verification and access control is all implemented in BPF LSM.

---
 tools/testing/selftests/bpf/bpf_kfuncs.h      |   7 +
 .../bpf/prog_tests/verify_pkcs7_sig.c         | 163 +++++++++++++++++-
 .../selftests/bpf/progs/test_sig_in_xattr.c   |  82 +++++++++
 .../bpf/progs/test_verify_pkcs7_sig.c         |   8 +-
 .../testing/selftests/bpf/verify_sig_setup.sh |  25 +++
 5 files changed, 277 insertions(+), 8 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/progs/test_sig_in_xattr.c
tools/testing/selftests/bpf/bpf_kfuncs.h +7 -0
index c2c084a44eae..b4e78c1eb37b 100644
@@ -58,4 +58,11 @@ void *bpf_rdonly_cast(void *obj, __u32 btf_id) __ksym;
 extern int bpf_get_file_xattr(struct file *file, const char *name,
 			      struct bpf_dynptr *value_ptr) __ksym;
 extern int bpf_get_fsverity_digest(struct file *file, struct bpf_dynptr *digest_ptr) __ksym;
+
+extern struct bpf_key *bpf_lookup_user_key(__u32 serial, __u64 flags) __ksym;
+extern struct bpf_key *bpf_lookup_system_key(__u64 id) __ksym;
+extern void bpf_key_put(struct bpf_key *key) __ksym;
+extern int bpf_verify_pkcs7_signature(struct bpf_dynptr *data_ptr,
+				      struct bpf_dynptr *sig_ptr,
+				      struct bpf_key *trusted_keyring) __ksym;
 #endif
tools/testing/selftests/bpf/prog_tests/verify_pkcs7_sig.c +162 -1

187 lines not on this page · raw message

tools/testing/selftests/bpf/progs/test_sig_in_xattr.c new file +82 -0

82 lines not on this page · raw message

tools/testing/selftests/bpf/progs/test_verify_pkcs7_sig.c +1 -7
index 7748cc23de8a..f42e9f3831a1 100644
@@ -10,17 +10,11 @@
 #include <errno.h>
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_tracing.h>
+#include "bpf_kfuncs.h"
 
 #define MAX_DATA_SIZE (1024 * 1024)
 #define MAX_SIG_SIZE 1024
 
-extern struct bpf_key *bpf_lookup_user_key(__u32 serial, __u64 flags) __ksym;
-extern struct bpf_key *bpf_lookup_system_key(__u64 id) __ksym;
-extern void bpf_key_put(struct bpf_key *key) __ksym;
-extern int bpf_verify_pkcs7_signature(struct bpf_dynptr *data_ptr,
-				      struct bpf_dynptr *sig_ptr,
-				      struct bpf_key *trusted_keyring) __ksym;
-
 __u32 monitored_pid;
 __u32 user_keyring_serial;
 __u64 system_keyring_id;
tools/testing/selftests/bpf/verify_sig_setup.sh +25 -0
index ba08922b4a27..7e6caa134e1a 100755
@@ -60,6 +60,27 @@ cleanup() {
 	rm -rf ${tmp_dir}
 }
 
+fsverity_create_sign_file() {
+	local tmp_dir="$1"
+
+	data_file=${tmp_dir}/data-file
+	sig_file=${tmp_dir}/sig-file
+	dd if=/dev/urandom of=$data_file bs=1 count=12345 2> /dev/null
+	fsverity sign --key ${tmp_dir}/signing_key.pem $data_file $sig_file
+
+	# We do not want to enable fsverity on $data_file yet. Try whether
+	# the file system support fsverity on a different file.
+	touch ${tmp_dir}/tmp-file
+	fsverity enable ${tmp_dir}/tmp-file
+}
+
+fsverity_enable_file() {
+    local tmp_dir="$1"
+
+	data_file=${tmp_dir}/data-file
+	fsverity enable $data_file
+}
+
 catch()
 {
 	local exit_code="$1"
@@ -86,6 +107,10 @@ main()
 		setup "${tmp_dir}"
 	elif [[ "${action}" == "cleanup" ]]; then
 		cleanup "${tmp_dir}"
+	elif [[ "${action}" == "fsverity-create-sign" ]]; then
+		fsverity_create_sign_file "${tmp_dir}"
+	elif [[ "${action}" == "fsverity-enable" ]]; then
+		fsverity_enable_file "${tmp_dir}"
 	else
 		echo "Unknown action: ${action}"
 		exit 1
Signature · 1 line
2.34.1
#21 lore raw

Re: [PATCH v13 bpf-next 6/6] selftests/bpf: Add test that uses fsverity and xattr to sign a file

kernel test robot noticed the following build errors:

Hi Song,

kernel test robot noticed the following build errors:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/Song-Liu/bpf-Add-kfunc-bpf_get_file_xattr/20231124-074239
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/20231123233936.3079687-7-song%40kernel.org
patch subject: [PATCH v13 bpf-next 6/6] selftests/bpf: Add test that uses fsverity and xattr to sign a file
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20231125/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

>> progs/test_sig_in_xattr.c:36:26: error: invalid application of 'sizeof' to an incomplete type 'struct fsverity_digest'
      36 | char digest[MAGIC_SIZE + sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                          ^     ~~~~~~~~~~~~~~~~~~~~~~~~
   progs/test_sig_in_xattr.c:36:40: note: forward declaration of 'struct fsverity_digest'
      36 | char digest[MAGIC_SIZE + sizeof(struct fsverity_digest) + SHA256_DIGEST_SIZE];
         |                                        ^

1 error generated.

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki