x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

When using eBPF to probe CS CALL instructions within a function,
a crash can be triggered.

The eBPF tool probes offset 257 of the __hrtimer_run_queues()
function:

<__hrtimer_run_queues+249>:  nopl   0x0(%rax,%rax,1)
<__hrtimer_run_queues+254>:  mov    %r14,%rdi
<__hrtimer_run_queues+257>:  cs call <__x86_indirect_thunk_r12>
<__hrtimer_run_queues+263>:  mov    %eax,%r12d
<__hrtimer_run_queues+266>:  xchg   %ax,%ax
<__hrtimer_run_queues+268>:  mov    %r13,%rdi

Which triggers this crash:

  BUG: unable to handle page fault for address: 00000000000f41c9
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x0002) - not-present page
  PGD 0 P4D 0
  Oops: 0002 [#1] SMP NOPTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P
  RIP: 0010:__hrtimer_run_queues+0x106/0x230

Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is
at the 6th byte of the above CS CALL instruction. Since the CS CALL
instruction occupies 6 bytes, the exception occurred in the middle of that
call instruction.

The root cause is that when using eBPF tools to probe in the middle of a
function, a kprobe with INT3 is used as the underlying implementation.

During single-step emulation of the original CALL instruction,
int3_emulate_call() assumes that the probed CALL instruction is 5 bytes
long. However, the actual CS-prefixed CALL instruction occupies 6 bytes,
so it constructs an incorrect exception return address. When the CPU
returns from the kprobe handler, the next instruction to be executed is at
the address of the last byte of that CS CALL instruction. Coincidentally,
starting from that address, the CPU fetches and decodes a completely
different instruction, which ultimately triggers a kernel crash.

Fix the issue by using the actual instruction length obtained from
the instruction decoder when constructing the exception return
address, rather than relying on the hardcoded CALL_INSN_SIZE macro.

[ mingo: Refined the changelog ]

Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step")
Suggested-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 514 insertions, 0 deletions

diff --git a/fs/smb/server/ndr.c b/fs/smb/server/ndr.c
new file mode 100644
index 000000000..58d71560f
--- /dev/null
+++ b/fs/smb/server/ndr.c
@@ -0,0 +1,514 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *   Copyright (C) 2021 Samsung Electronics Co., Ltd.
+ *   Author(s): Namjae Jeon <linkinjeon@kernel.org>
+ */
+
+#include <linux/fs.h>
+
+#include "glob.h"
+#include "ndr.h"
+
+static inline char *ndr_get_field(struct ndr *n)
+{
+	return n->data + n->offset;
+}
+
+static int try_to_realloc_ndr_blob(struct ndr *n, size_t sz)
+{
+	char *data;
+
+	data = krealloc(n->data, n->offset + sz + 1024, KSMBD_DEFAULT_GFP);
+	if (!data)
+		return -ENOMEM;
+
+	n->data = data;
+	n->length += 1024;
+	memset(n->data + n->offset, 0, 1024);
+	return 0;
+}
+
+static int ndr_write_int16(struct ndr *n, __u16 value)
+{
+	if (n->length <= n->offset + sizeof(value)) {
+		int ret;
+
+		ret = try_to_realloc_ndr_blob(n, sizeof(value));
+		if (ret)
+			return ret;
+	}
+
+	*(__le16 *)ndr_get_field(n) = cpu_to_le16(value);
+	n->offset += sizeof(value);
+	return 0;
+}
+
+static int ndr_write_int32(struct ndr *n, __u32 value)
+{
+	if (n->length <= n->offset + sizeof(value)) {
+		int ret;
+
+		ret = try_to_realloc_ndr_blob(n, sizeof(value));
+		if (ret)
+			return ret;
+	}
+
+	*(__le32 *)ndr_get_field(n) = cpu_to_le32(value);
+	n->offset += sizeof(value);
+	return 0;
+}
+
+static int ndr_write_int64(struct ndr *n, __u64 value)
+{
+	if (n->length <= n->offset + sizeof(value)) {
+		int ret;
+
+		ret = try_to_realloc_ndr_blob(n, sizeof(value));
+		if (ret)
+			return ret;
+	}
+
+	*(__le64 *)ndr_get_field(n) = cpu_to_le64(value);
+	n->offset += sizeof(value);
+	return 0;
+}
+
+static int ndr_write_bytes(struct ndr *n, void *value, size_t sz)
+{
+	if (n->length <= n->offset + sz) {
+		int ret;
+
+		ret = try_to_realloc_ndr_blob(n, sz);
+		if (ret)
+			return ret;
+	}
+
+	memcpy(ndr_get_field(n), value, sz);
+	n->offset += sz;
+	return 0;
+}
+
+static int ndr_write_string(struct ndr *n, char *value)
+{
+	size_t sz;
+
+	sz = strlen(value) + 1;
+	if (n->length <= n->offset + sz) {
+		int ret;
+
+		ret = try_to_realloc_ndr_blob(n, sz);
+		if (ret)
+			return ret;
+	}
+
+	memcpy(ndr_get_field(n), value, sz);
+	n->offset += sz;
+	n->offset = ALIGN(n->offset, 2);
+	return 0;
+}
+
+static int ndr_read_string(struct ndr *n, void *value, size_t sz)
+{
+	int len;
+
+	if (n->offset + sz > n->length)
+		return -EINVAL;
+
+	len = strnlen(ndr_get_field(n), sz);
+	if (value)
+		memcpy(value, ndr_get_field(n), len);
+	len++;
+	n->offset += len;
+	n->offset = ALIGN(n->offset, 2);
+	return 0;
+}
+
+static int ndr_read_bytes(struct ndr *n, void *value, size_t sz)
+{
+	if (n->offset + sz > n->length)
+		return -EINVAL;
+
+	if (value)
+		memcpy(value, ndr_get_field(n), sz);
+	n->offset += sz;
+	return 0;
+}
+
+static int ndr_read_int16(struct ndr *n, __u16 *value)
+{
+	if (n->offset + sizeof(__u16) > n->length)
+		return -EINVAL;
+
+	if (value)
+		*value = le16_to_cpu(*(__le16 *)ndr_get_field(n));
+	n->offset += sizeof(__u16);
+	return 0;
+}
+
+static int ndr_read_int32(struct ndr *n, __u32 *value)
+{
+	if (n->offset + sizeof(__u32) > n->length)
+		return -EINVAL;
+
+	if (value)
+		*value = le32_to_cpu(*(__le32 *)ndr_get_field(n));
+	n->offset += sizeof(__u32);
+	return 0;
+}
+
+static int ndr_read_int64(struct ndr *n, __u64 *value)
+{
+	if (n->offset + sizeof(__u64) > n->length)
+		return -EINVAL;
+
+	if (value)
+		*value = le64_to_cpu(*(__le64 *)ndr_get_field(n));
+	n->offset += sizeof(__u64);
+	return 0;
+}
+
+int ndr_encode_dos_attr(struct ndr *n, struct xattr_dos_attrib *da)
+{
+	char hex_attr[12] = {0};
+	int ret;
+
+	n->offset = 0;
+	n->length = 1024;
+	n->data = kzalloc(n->length, KSMBD_DEFAULT_GFP);
+	if (!n->data)
+		return -ENOMEM;
+
+	if (da->version == 3) {
+		snprintf(hex_attr, 10, "0x%x", da->attr);
+		ret = ndr_write_string(n, hex_attr);
+	} else {
+		ret = ndr_write_string(n, "");
+	}
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int16(n, da->version);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, da->version);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, da->flags);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, da->attr);
+	if (ret)
+		return ret;
+
+	if (da->version == 3) {
+		ret = ndr_write_int32(n, da->ea_size);
+		if (ret)
+			return ret;
+		ret = ndr_write_int64(n, da->size);
+		if (ret)
+			return ret;
+		ret = ndr_write_int64(n, da->alloc_size);
+	} else {
+		ret = ndr_write_int64(n, da->itime);
+	}
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int64(n, da->create_time);
+	if (ret)
+		return ret;
+
+	if (da->version == 3)
+		ret = ndr_write_int64(n, da->change_time);
+	return ret;
+}
+
+int ndr_decode_dos_attr(struct ndr *n, struct xattr_dos_attrib *da)
+{
+	char hex_attr[12];
+	unsigned int version2;
+	int ret;
+
+	n->offset = 0;
+	ret = ndr_read_string(n, hex_attr, sizeof(hex_attr));
+	if (ret)
+		return ret;
+
+	ret = ndr_read_int16(n, &da->version);
+	if (ret)
+		return ret;
+
+	if (da->version != 3 && da->version != 4) {
+		ksmbd_debug(VFS, "v%d version is not supported\n", da->version);
+		return -EINVAL;
+	}
+
+	ret = ndr_read_int32(n, &version2);
+	if (ret)
+		return ret;
+
+	if (da->version != version2) {
+		ksmbd_debug(VFS, "ndr version mismatched(version: %d, version2: %d)\n",
+		       da->version, version2);
+		return -EINVAL;
+	}
+
+	ret = ndr_read_int32(n, NULL);
+	if (ret)
+		return ret;
+
+	ret = ndr_read_int32(n, &da->attr);
+	if (ret)
+		return ret;
+
+	if (da->version == 4) {
+		ret = ndr_read_int64(n, &da->itime);
+		if (ret)
+			return ret;
+
+		ret = ndr_read_int64(n, &da->create_time);
+	} else {
+		ret = ndr_read_int32(n, NULL);
+		if (ret)
+			return ret;
+
+		ret = ndr_read_int64(n, NULL);
+		if (ret)
+			return ret;
+
+		ret = ndr_read_int64(n, NULL);
+		if (ret)
+			return ret;
+
+		ret = ndr_read_int64(n, &da->create_time);
+		if (ret)
+			return ret;
+
+		ret = ndr_read_int64(n, NULL);
+	}
+
+	return ret;
+}
+
+static int ndr_encode_posix_acl_entry(struct ndr *n, struct xattr_smb_acl *acl)
+{
+	int i, ret;
+
+	ret = ndr_write_int32(n, acl->count);
+	if (ret)
+		return ret;
+
+	n->offset = ALIGN(n->offset, 8);
+	ret = ndr_write_int32(n, acl->count);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, 0);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < acl->count; i++) {
+		n->offset = ALIGN(n->offset, 8);
+		ret = ndr_write_int16(n, acl->entries[i].type);
+		if (ret)
+			return ret;
+
+		ret = ndr_write_int16(n, acl->entries[i].type);
+		if (ret)
+			return ret;
+
+		if (acl->entries[i].type == SMB_ACL_USER) {
+			n->offset = ALIGN(n->offset, 8);
+			ret = ndr_write_int64(n, acl->entries[i].uid);
+		} else if (acl->entries[i].type == SMB_ACL_GROUP) {
+			n->offset = ALIGN(n->offset, 8);
+			ret = ndr_write_int64(n, acl->entries[i].gid);
+		}
+		if (ret)
+			return ret;
+
+		/* push permission */
+		ret = ndr_write_int32(n, acl->entries[i].perm);
+	}
+
+	return ret;
+}
+
+int ndr_encode_posix_acl(struct ndr *n,
+			 struct mnt_idmap *idmap,
+			 struct inode *inode,
+			 struct xattr_smb_acl *acl,
+			 struct xattr_smb_acl *def_acl)
+{
+	unsigned int ref_id = 0x00020000;
+	int ret;
+	vfsuid_t vfsuid;
+	vfsgid_t vfsgid;
+
+	n->offset = 0;
+	n->length = 1024;
+	n->data = kzalloc(n->length, KSMBD_DEFAULT_GFP);
+	if (!n->data)
+		return -ENOMEM;
+
+	if (acl) {
+		/* ACL ACCESS */
+		ret = ndr_write_int32(n, ref_id);
+		ref_id += 4;
+	} else {
+		ret = ndr_write_int32(n, 0);
+	}
+	if (ret)
+		return ret;
+
+	if (def_acl) {
+		/* DEFAULT ACL ACCESS */
+		ret = ndr_write_int32(n, ref_id);
+		ref_id += 4;
+	} else {
+		ret = ndr_write_int32(n, 0);
+	}
+	if (ret)
+		return ret;
+
+	vfsuid = i_uid_into_vfsuid(idmap, inode);
+	ret = ndr_write_int64(n, from_kuid(&init_user_ns, vfsuid_into_kuid(vfsuid)));
+	if (ret)
+		return ret;
+	vfsgid = i_gid_into_vfsgid(idmap, inode);
+	ret = ndr_write_int64(n, from_kgid(&init_user_ns, vfsgid_into_kgid(vfsgid)));
+	if (ret)
+		return ret;
+	ret = ndr_write_int32(n, inode->i_mode);
+	if (ret)
+		return ret;
+
+	if (acl) {
+		ret = ndr_encode_posix_acl_entry(n, acl);
+		if (def_acl && !ret)
+			ret = ndr_encode_posix_acl_entry(n, def_acl);
+	}
+	return ret;
+}
+
+int ndr_encode_v4_ntacl(struct ndr *n, struct xattr_ntacl *acl)
+{
+	unsigned int ref_id = 0x00020004;
+	int ret;
+
+	n->offset = 0;
+	n->length = 2048;
+	n->data = kzalloc(n->length, KSMBD_DEFAULT_GFP);
+	if (!n->data)
+		return -ENOMEM;
+
+	ret = ndr_write_int16(n, acl->version);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, acl->version);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int16(n, 2);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int32(n, ref_id);
+	if (ret)
+		return ret;
+
+	/* push hash type and hash 64bytes */
+	ret = ndr_write_int16(n, acl->hash_type);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_bytes(n, acl->hash, XATTR_SD_HASH_SIZE);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_bytes(n, acl->desc, acl->desc_len);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_int64(n, acl->current_time);
+	if (ret)
+		return ret;
+
+	ret = ndr_write_bytes(n, acl->posix_acl_hash, XATTR_SD_HASH_SIZE);
+	if (ret)
+		return ret;
+
+	/* push ndr for security descriptor */
+	ret = ndr_write_bytes(n, acl->sd_buf, acl->sd_size);
+	return ret;
+}
+
+int ndr_decode_v4_ntacl(struct ndr *n, struct xattr_ntacl *acl)
+{
+	unsigned int version2;
+	int ret;
+
+	n->offset = 0;
+	ret = ndr_read_int16(n, &acl->version);
+	if (ret)
+		return ret;
+	if (acl->version != 4) {
+		ksmbd_debug(VFS, "v%d version is not supported\n", acl->version);
+		return -EINVAL;
+	}
+
+	ret = ndr_read_int32(n, &version2);
+	if (ret)
+		return ret;
+	if (acl->version != version2) {
+		ksmbd_debug(VFS, "ndr version mismatched(version: %d, version2: %d)\n",
+		       acl->version, version2);
+		return -EINVAL;
+	}
+
+	/* Read Level */
+	ret = ndr_read_int16(n, NULL);
+	if (ret)
+		return ret;
+
+	/* Read Ref Id */
+	ret = ndr_read_int32(n, NULL);
+	if (ret)
+		return ret;
+
+	ret = ndr_read_int16(n, &acl->hash_type);
+	if (ret)
+		return ret;
+
+	ret = ndr_read_bytes(n, acl->hash, XATTR_SD_HASH_SIZE);
+	if (ret)
+		return ret;
+
+	ndr_read_bytes(n, acl->desc, 10);
+	if (strncmp(acl->desc, "posix_acl", 9)) {
+		pr_err("Invalid acl description : %s\n", acl->desc);
+		return -EINVAL;
+	}
+
+	/* Read Time */
+	ret = ndr_read_int64(n, NULL);
+	if (ret)
+		return ret;
+
+	/* Read Posix ACL hash */
+	ret = ndr_read_bytes(n, acl->posix_acl_hash, XATTR_SD_HASH_SIZE);
+	if (ret)
+		return ret;
+
+	acl->sd_size = n->length - n->offset;
+	acl->sd_buf = kzalloc(acl->sd_size, KSMBD_DEFAULT_GFP);
+	if (!acl->sd_buf)
+		return -ENOMEM;
+
+	ret = ndr_read_bytes(n, acl->sd_buf, acl->sd_size);
+	return ret;
+}