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, 343 insertions, 0 deletions

diff --git a/fs/hfsplus/bfind.c b/fs/hfsplus/bfind.c
new file mode 100644
index 000000000..ca9813f58
--- /dev/null
+++ b/fs/hfsplus/bfind.c
@@ -0,0 +1,343 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/fs/hfsplus/bfind.c
+ *
+ * Copyright (C) 2001
+ * Brad Boyer (flar@allandria.com)
+ * (C) 2003 Ardis Technologies <roman@ardistech.com>
+ *
+ * Search routines for btrees
+ */
+
+#include <linux/slab.h>
+#include "hfsplus_fs.h"
+
+int hfs_find_init(struct hfs_btree *tree, struct hfs_find_data *fd)
+{
+	void *ptr;
+
+	fd->tree = tree;
+	fd->bnode = NULL;
+	hfs_find_result_init(fd);
+	ptr = kzalloc(tree->max_key_len * 2 + 4, GFP_KERNEL);
+	if (!ptr)
+		return -ENOMEM;
+	fd->search_key = ptr;
+	fd->key = ptr + tree->max_key_len + 2;
+	hfs_dbg("cnid %d, caller %ps\n",
+		tree->cnid, __builtin_return_address(0));
+	mutex_lock_nested(&tree->tree_lock,
+			hfsplus_btree_lock_class(tree));
+	return 0;
+}
+
+void hfs_find_exit(struct hfs_find_data *fd)
+{
+	hfs_bnode_put(fd->bnode);
+	kfree(fd->search_key);
+	hfs_dbg("cnid %d, caller %ps\n",
+		fd->tree->cnid, __builtin_return_address(0));
+	mutex_unlock(&fd->tree->tree_lock);
+	fd->tree = NULL;
+}
+
+int hfs_find_1st_rec_by_cnid(struct hfs_bnode *bnode,
+				struct hfs_find_data *fd,
+				int *begin,
+				int *end,
+				int *cur_rec)
+{
+	__be32 cur_cnid;
+	__be32 search_cnid;
+
+	if (bnode->tree->cnid == HFSPLUS_EXT_CNID) {
+		cur_cnid = fd->key->ext.cnid;
+		search_cnid = fd->search_key->ext.cnid;
+	} else if (bnode->tree->cnid == HFSPLUS_CAT_CNID) {
+		cur_cnid = fd->key->cat.parent;
+		search_cnid = fd->search_key->cat.parent;
+	} else if (bnode->tree->cnid == HFSPLUS_ATTR_CNID) {
+		cur_cnid = fd->key->attr.cnid;
+		search_cnid = fd->search_key->attr.cnid;
+	} else {
+		cur_cnid = 0;	/* used-uninitialized warning */
+		search_cnid = 0;
+		BUG();
+	}
+
+	if (cur_cnid == search_cnid) {
+		(*end) = (*cur_rec);
+		if ((*begin) == (*end))
+			return 1;
+	} else {
+		if (be32_to_cpu(cur_cnid) < be32_to_cpu(search_cnid))
+			(*begin) = (*cur_rec) + 1;
+		else
+			(*end) = (*cur_rec) - 1;
+	}
+
+	return 0;
+}
+
+int hfs_find_rec_by_key(struct hfs_bnode *bnode,
+				struct hfs_find_data *fd,
+				int *begin,
+				int *end,
+				int *cur_rec)
+{
+	int cmpval;
+
+	cmpval = bnode->tree->keycmp(fd->key, fd->search_key);
+	if (!cmpval) {
+		(*end) = (*cur_rec);
+		return 1;
+	}
+	if (cmpval < 0)
+		(*begin) = (*cur_rec) + 1;
+	else
+		*(end) = (*cur_rec) - 1;
+
+	return 0;
+}
+
+/* Find the record in bnode that best matches key (not greater than...)*/
+int __hfs_brec_find(struct hfs_bnode *bnode, struct hfs_find_data *fd,
+					search_strategy_t rec_found)
+{
+	u16 off, len, keylen;
+	int rec;
+	int b, e;
+	int res = -ENOENT;
+
+	BUG_ON(!rec_found);
+	hfs_find_result_init(fd);
+	if (hfs_bnode_num_recs_invalid(bnode))
+		goto fail;
+
+	b = 0;
+	e = bnode->num_recs - 1;
+	do {
+		rec = (e + b) / 2;
+		len = hfs_brec_lenoff(bnode, rec, &off);
+		keylen = hfs_brec_keylen(bnode, rec);
+		if (hfs_brec_len_invalid(bnode, len) ||
+		    hfs_brec_len_invalid(bnode, keylen)) {
+			res = -EINVAL;
+			goto fail;
+		}
+		hfs_bnode_read(bnode, fd->key, off, keylen);
+		if (rec_found(bnode, fd, &b, &e, &rec)) {
+			res = 0;
+			goto done;
+		}
+	} while (b <= e);
+
+	if (rec != e && e >= 0) {
+		len = hfs_brec_lenoff(bnode, e, &off);
+		keylen = hfs_brec_keylen(bnode, e);
+		if (hfs_brec_len_invalid(bnode, keylen) ||
+		    hfs_brec_len_invalid(bnode, len)) {
+			res = -EINVAL;
+			goto fail;
+		}
+		hfs_bnode_read(bnode, fd->key, off, keylen);
+	}
+
+done:
+	fd->record = e;
+	fd->keyoffset = off;
+	fd->keylength = keylen;
+	fd->entryoffset = off + keylen;
+	fd->entrylength = len - keylen;
+
+fail:
+	return res;
+}
+
+/* Traverse a B*Tree from the root to a leaf finding best fit to key */
+/* Return allocated copy of node found, set recnum to best record */
+int hfs_brec_find(struct hfs_find_data *fd, search_strategy_t do_key_compare)
+{
+	struct hfs_btree *tree;
+	struct hfs_bnode *bnode;
+	u32 nidx, parent;
+	__be32 data;
+	int height, res;
+
+	hfs_find_result_init(fd);
+
+	tree = fd->tree;
+	if (fd->bnode)
+		hfs_bnode_put(fd->bnode);
+	fd->bnode = NULL;
+	nidx = tree->root;
+	if (!nidx)
+		return -ENOENT;
+	height = tree->depth;
+	res = 0;
+	parent = 0;
+	for (;;) {
+		bnode = hfs_bnode_find(tree, nidx);
+		if (IS_ERR(bnode)) {
+			res = PTR_ERR(bnode);
+			bnode = NULL;
+			break;
+		}
+		if (bnode->height != height)
+			goto invalid;
+		if (bnode->type != (--height ? HFS_NODE_INDEX : HFS_NODE_LEAF))
+			goto invalid;
+		bnode->parent = parent;
+
+		res = __hfs_brec_find(bnode, fd, do_key_compare);
+		if (!height)
+			break;
+		if (fd->record < 0)
+			goto release;
+
+		parent = nidx;
+		hfs_bnode_read(bnode, &data, fd->entryoffset, 4);
+		nidx = be32_to_cpu(data);
+		hfs_bnode_put(bnode);
+	}
+	fd->bnode = bnode;
+	return res;
+
+invalid:
+	pr_err("inconsistency in B*Tree (%d,%d,%d,%u,%u)\n",
+		height, bnode->height, bnode->type, nidx, parent);
+	res = -EIO;
+release:
+	hfs_bnode_put(bnode);
+	return res;
+}
+
+int hfs_brec_read(struct hfs_find_data *fd, void *rec, u32 rec_len)
+{
+	int res;
+
+	res = hfs_brec_find(fd, hfs_find_rec_by_key);
+	if (res)
+		return res;
+	if (fd->entrylength > rec_len)
+		return -EINVAL;
+	hfs_bnode_read(fd->bnode, rec, fd->entryoffset, fd->entrylength);
+	return 0;
+}
+
+int hfs_brec_goto(struct hfs_find_data *fd, int cnt)
+{
+	struct hfs_btree *tree;
+	struct hfs_bnode *bnode;
+	int idx, res = 0;
+	u16 off, len, keylen;
+
+	bnode = fd->bnode;
+	tree = bnode->tree;
+
+	if (cnt < 0) {
+		cnt = -cnt;
+		while (cnt > fd->record) {
+			cnt -= fd->record + 1;
+			fd->record = bnode->num_recs - 1;
+			idx = bnode->prev;
+			if (!idx) {
+				res = -ENOENT;
+				goto out;
+			}
+			hfs_bnode_put(bnode);
+			bnode = hfs_bnode_find(tree, idx);
+			if (IS_ERR(bnode)) {
+				res = PTR_ERR(bnode);
+				bnode = NULL;
+				goto out;
+			}
+		}
+		fd->record -= cnt;
+	} else {
+		while (cnt >= bnode->num_recs - fd->record) {
+			cnt -= bnode->num_recs - fd->record;
+			fd->record = 0;
+			idx = bnode->next;
+			if (!idx) {
+				res = -ENOENT;
+				goto out;
+			}
+			hfs_bnode_put(bnode);
+			bnode = hfs_bnode_find(tree, idx);
+			if (IS_ERR(bnode)) {
+				res = PTR_ERR(bnode);
+				bnode = NULL;
+				goto out;
+			}
+		}
+		fd->record += cnt;
+	}
+
+	len = hfs_brec_lenoff(bnode, fd->record, &off);
+	keylen = hfs_brec_keylen(bnode, fd->record);
+	if (hfs_brec_len_invalid(bnode, len) ||
+	    hfs_brec_len_invalid(bnode, keylen)) {
+		res = -EINVAL;
+		goto out;
+	}
+	fd->keyoffset = off;
+	fd->keylength = keylen;
+	fd->entryoffset = off + keylen;
+	fd->entrylength = len - keylen;
+	hfs_bnode_read(bnode, fd->key, off, keylen);
+out:
+	fd->bnode = bnode;
+	return res;
+}
+
+/**
+ * hfsplus_brec_read_cat - read and validate a catalog record
+ * @fd: find data structure
+ * @entry: pointer to catalog entry to read into
+ *
+ * Reads a catalog record and validates its size matches the expected
+ * size based on the record type.
+ *
+ * Returns 0 on success, or negative error code on failure.
+ */
+int hfsplus_brec_read_cat(struct hfs_find_data *fd, hfsplus_cat_entry *entry)
+{
+	int res;
+	u32 expected_size;
+
+	res = hfs_brec_read(fd, entry, sizeof(hfsplus_cat_entry));
+	if (res)
+		return res;
+
+	/* Validate catalog record size based on type */
+	switch (be16_to_cpu(entry->type)) {
+	case HFSPLUS_FOLDER:
+		expected_size = sizeof(struct hfsplus_cat_folder);
+		break;
+	case HFSPLUS_FILE:
+		expected_size = sizeof(struct hfsplus_cat_file);
+		break;
+	case HFSPLUS_FOLDER_THREAD:
+	case HFSPLUS_FILE_THREAD:
+		/* Ensure we have at least the fixed fields before reading nodeName.length */
+		if (fd->entrylength < HFSPLUS_MIN_THREAD_SZ) {
+			pr_err("thread record too short (got %u)\n", fd->entrylength);
+			return -EIO;
+		}
+		expected_size = hfsplus_cat_thread_size(&entry->thread);
+		break;
+	default:
+		pr_err("unknown catalog record type %d\n",
+		       be16_to_cpu(entry->type));
+		return -EIO;
+	}
+
+	if (fd->entrylength != expected_size) {
+		pr_err("catalog record size mismatch (type %d, got %u, expected %u)\n",
+		       be16_to_cpu(entry->type), fd->entrylength, expected_size);
+		return -EIO;
+	}
+
+	return 0;
+}