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, 199 insertions, 0 deletions

diff --git a/fs/smb/client/dfs.h b/fs/smb/client/dfs.h
new file mode 100644
index 000000000..16ac2cdd5
--- /dev/null
+++ b/fs/smb/client/dfs.h
@@ -0,0 +1,199 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (c) 2022 Paulo Alcantara <palcantara@suse.de>
+ */
+
+#ifndef _CIFS_DFS_H
+#define _CIFS_DFS_H
+
+#include "cifsglob.h"
+#include "cifsproto.h"
+#include "fs_context.h"
+#include "dfs_cache.h"
+#include "cifs_unicode.h"
+#include <linux/namei.h>
+#include <linux/errno.h>
+
+#define DFS_INTERLINK(v) \
+	(((v) & DFSREF_REFERRAL_SERVER) && !((v) & DFSREF_STORAGE_SERVER))
+
+struct dfs_ref {
+	char *path;
+	char *full_path;
+	struct cifs_ses *ses;
+	struct dfs_cache_tgt_list tl;
+	struct dfs_cache_tgt_iterator *tit;
+};
+
+struct dfs_ref_walk {
+	struct cifs_mount_ctx	*mnt_ctx;
+	struct dfs_ref		*ref;
+	struct dfs_ref		refs[MAX_NESTED_LINKS];
+};
+
+#define ref_walk_start(w)	((w)->refs)
+#define ref_walk_end(w)	(&(w)->refs[ARRAY_SIZE((w)->refs) - 1])
+#define ref_walk_cur(w)	((w)->ref)
+#define ref_walk_descend(w)	(--ref_walk_cur(w) >= ref_walk_start(w))
+
+#define ref_walk_tit(w)	(ref_walk_cur(w)->tit)
+#define ref_walk_path(w)	(ref_walk_cur(w)->path)
+#define ref_walk_fpath(w)	(ref_walk_cur(w)->full_path)
+#define ref_walk_tl(w)		(&ref_walk_cur(w)->tl)
+#define ref_walk_ses(w)	(ref_walk_cur(w)->ses)
+
+static inline struct dfs_ref_walk *ref_walk_alloc(void)
+{
+	struct dfs_ref_walk *rw;
+
+	rw = kmalloc_obj(*rw);
+	if (!rw)
+		return ERR_PTR(-ENOMEM);
+	return rw;
+}
+
+static inline void ref_walk_init(struct dfs_ref_walk *rw,
+				 struct cifs_mount_ctx *mnt_ctx)
+{
+	memset(rw, 0, sizeof(*rw));
+	rw->mnt_ctx = mnt_ctx;
+	ref_walk_cur(rw) = ref_walk_start(rw);
+}
+
+static inline void __ref_walk_free(struct dfs_ref *ref)
+{
+	kfree(ref->path);
+	kfree(ref->full_path);
+	dfs_cache_free_tgts(&ref->tl);
+	if (ref->ses)
+		cifs_put_smb_ses(ref->ses);
+	memset(ref, 0, sizeof(*ref));
+}
+
+static inline void ref_walk_free(struct dfs_ref_walk *rw)
+{
+	struct dfs_ref *ref;
+
+	if (!rw)
+		return;
+
+	for (ref = ref_walk_start(rw); ref <= ref_walk_end(rw); ref++)
+		__ref_walk_free(ref);
+	kfree(rw);
+}
+
+static inline int ref_walk_advance(struct dfs_ref_walk *rw)
+{
+	struct dfs_ref *ref = ref_walk_cur(rw) + 1;
+
+	if (ref > ref_walk_end(rw))
+		return -ELOOP;
+	__ref_walk_free(ref);
+	ref_walk_cur(rw) = ref;
+	return 0;
+}
+
+static inline struct dfs_cache_tgt_iterator *
+ref_walk_next_tgt(struct dfs_ref_walk *rw)
+{
+	struct dfs_ref *ref = ref_walk_cur(rw);
+	struct dfs_cache_tgt_iterator *tit;
+
+	if (IS_ERR(ref->tit))
+		return NULL;
+
+	if (!ref->tit)
+		tit = dfs_cache_get_tgt_iterator(&ref->tl);
+	else
+		tit = dfs_cache_get_next_tgt(&ref->tl, ref->tit);
+
+	if (!tit) {
+		ref->tit = ERR_PTR(-ENOENT);
+		return NULL;
+	}
+	ref->tit = tit;
+	return ref->tit;
+}
+
+static inline int ref_walk_get_tgt(struct dfs_ref_walk *rw,
+				   struct dfs_info3_param *tgt)
+{
+	zfree_dfs_info_param(tgt);
+	return dfs_cache_get_tgt_referral(ref_walk_path(rw) + 1,
+					  ref_walk_tit(rw), tgt);
+}
+
+static inline void ref_walk_set_tgt_hint(struct dfs_ref_walk *rw)
+{
+	dfs_cache_noreq_update_tgthint(ref_walk_path(rw) + 1,
+				       ref_walk_tit(rw));
+}
+
+static inline void ref_walk_set_tcon(struct dfs_ref_walk *rw,
+				     struct cifs_tcon *tcon)
+{
+	struct dfs_ref *ref = ref_walk_start(rw);
+
+	for (; ref <= ref_walk_cur(rw); ref++) {
+		if (WARN_ON_ONCE(!ref->ses))
+			continue;
+		list_add(&ref->ses->dlist, &tcon->dfs_ses_list);
+		ref->ses = NULL;
+	}
+}
+
+static inline void ref_walk_mark_end(struct dfs_ref_walk *rw)
+{
+	struct dfs_ref *ref = ref_walk_cur(rw) - 1;
+
+	WARN_ON_ONCE(ref < ref_walk_start(rw));
+	dfs_cache_noreq_update_tgthint(ref->path + 1, ref->tit);
+	ref->tit = ERR_PTR(-ENOENT); /* end marker */
+}
+
+int dfs_parse_target_referral(const char *full_path,
+			      const struct dfs_info3_param *ref,
+			      struct smb3_fs_context *ctx);
+int dfs_mount_share(struct cifs_mount_ctx *mnt_ctx);
+
+static inline char *dfs_get_path(struct cifs_sb_info *cifs_sb, const char *path)
+{
+	return dfs_cache_canonical_path(path, cifs_sb->local_nls, cifs_remap(cifs_sb));
+}
+
+static inline int dfs_get_referral(struct cifs_mount_ctx *mnt_ctx,
+				   const char *path,
+				   struct dfs_cache_tgt_list *tl)
+{
+	struct smb3_fs_context *ctx = mnt_ctx->fs_ctx;
+	struct cifs_sb_info *cifs_sb = mnt_ctx->cifs_sb;
+	struct cifs_ses *rses = ctx->dfs_root_ses ?: mnt_ctx->ses;
+
+	return dfs_cache_find(mnt_ctx->xid, rses, cifs_sb->local_nls,
+			      cifs_remap(cifs_sb), path, NULL, tl);
+}
+
+/*
+ * cifs_get_smb_ses() already guarantees an active reference of
+ * @ses->dfs_root_ses when a new session is created, so we need to put extra
+ * references of all DFS root sessions that were used across the mount process
+ * in dfs_mount_share().
+ */
+static inline void dfs_put_root_smb_sessions(struct list_head *head)
+{
+	struct cifs_ses *ses, *n;
+
+	list_for_each_entry_safe(ses, n, head, dlist) {
+		list_del_init(&ses->dlist);
+		cifs_put_smb_ses(ses);
+	}
+}
+
+static inline const char *dfs_ses_refpath(struct cifs_ses *ses)
+{
+	const char *path = ses->server->leaf_fullpath;
+
+	return path ? path + 1 : ERR_PTR(-ENOENT);
+}
+
+#endif /* _CIFS_DFS_H */