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, 406 insertions, 0 deletions

diff --git a/fs/afs/dynroot.c b/fs/afs/dynroot.c
new file mode 100644
index 000000000..6e3c8c691
--- /dev/null
+++ b/fs/afs/dynroot.c
@@ -0,0 +1,406 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* AFS dynamic root handling
+ *
+ * Copyright (C) 2018 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/fs.h>
+#include <linux/namei.h>
+#include <linux/dns_resolver.h>
+#include "internal.h"
+
+#define AFS_MIN_DYNROOT_CELL_INO 4 /* Allow for ., .., @cell, .@cell */
+#define AFS_MAX_DYNROOT_CELL_INO ((unsigned int)INT_MAX)
+
+static struct dentry *afs_lookup_atcell(struct inode *dir, struct dentry *dentry, ino_t ino);
+
+/*
+ * iget5() comparator for inode created by autocell operations
+ */
+static int afs_iget5_pseudo_test(struct inode *inode, void *opaque)
+{
+	struct afs_fid *fid = opaque;
+
+	return inode->i_ino == fid->vnode;
+}
+
+/*
+ * iget5() inode initialiser
+ */
+static int afs_iget5_pseudo_set(struct inode *inode, void *opaque)
+{
+	struct afs_super_info *as = AFS_FS_S(inode->i_sb);
+	struct afs_vnode *vnode = AFS_FS_I(inode);
+	struct afs_fid *fid = opaque;
+
+	vnode->volume		= as->volume;
+	vnode->fid		= *fid;
+	inode->i_ino		= fid->vnode;
+	inode->i_generation	= fid->unique;
+	return 0;
+}
+
+/*
+ * Create an inode for an autocell dynamic automount dir.
+ */
+static struct inode *afs_iget_pseudo_dir(struct super_block *sb, ino_t ino)
+{
+	struct afs_vnode *vnode;
+	struct inode *inode;
+	struct afs_fid fid = { .vnode = ino, .unique = 1, };
+
+	_enter("");
+
+	inode = iget5_locked(sb, fid.vnode,
+			     afs_iget5_pseudo_test, afs_iget5_pseudo_set, &fid);
+	if (!inode) {
+		_leave(" = -ENOMEM");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	_debug("GOT INODE %p { ino=%llu, vl=%llx, vn=%llx, u=%x }",
+	       inode, inode->i_ino, fid.vid, fid.vnode, fid.unique);
+
+	vnode = AFS_FS_I(inode);
+
+	if (inode_state_read_once(inode) & I_NEW) {
+		netfs_inode_init(&vnode->netfs, NULL, false);
+		simple_inode_init_ts(inode);
+		set_nlink(inode, 2);
+		inode->i_size		= 0;
+		inode->i_mode		= S_IFDIR | 0555;
+		inode->i_op		= &afs_autocell_inode_operations;
+		inode->i_uid		= GLOBAL_ROOT_UID;
+		inode->i_gid		= GLOBAL_ROOT_GID;
+		inode->i_blocks		= 0;
+		inode->i_generation	= 0;
+		inode->i_flags		|= S_AUTOMOUNT | S_NOATIME;
+
+		set_bit(AFS_VNODE_PSEUDODIR, &vnode->flags);
+		set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
+
+		unlock_new_inode(inode);
+	}
+	_leave(" = %p", inode);
+	return inode;
+}
+
+/*
+ * Try to automount the mountpoint with pseudo directory, if the autocell
+ * option is set.
+ */
+static struct dentry *afs_dynroot_lookup_cell(struct inode *dir, struct dentry *dentry,
+					      unsigned int flags)
+{
+	struct afs_cell *cell = NULL;
+	struct afs_net *net = afs_d2net(dentry);
+	struct inode *inode = NULL;
+	const char *name = dentry->d_name.name;
+	size_t len = dentry->d_name.len;
+	bool dotted = false;
+	int ret = -ENOENT;
+
+	/* Names prefixed with a dot are R/W mounts. */
+	if (name[0] == '.') {
+		name++;
+		len--;
+		dotted = true;
+	}
+
+	cell = afs_lookup_cell(net, name, len, NULL,
+			       AFS_LOOKUP_CELL_DYNROOT,
+			       afs_cell_trace_use_lookup_dynroot);
+	if (IS_ERR(cell)) {
+		ret = PTR_ERR(cell);
+		goto out_no_cell;
+	}
+
+	inode = afs_iget_pseudo_dir(dir->i_sb, cell->dynroot_ino * 2 + dotted);
+	if (IS_ERR(inode)) {
+		ret = PTR_ERR(inode);
+		goto out;
+	}
+
+	dentry->d_fsdata = cell;
+	return d_splice_alias(inode, dentry);
+
+out:
+	afs_unuse_cell(cell, afs_cell_trace_unuse_lookup_dynroot);
+out_no_cell:
+	if (!inode)
+		return d_splice_alias(inode, dentry);
+	return ret == -ENOENT ? NULL : ERR_PTR(ret);
+}
+
+/*
+ * Look up an entry in a dynroot directory.
+ */
+static struct dentry *afs_dynroot_lookup(struct inode *dir, struct dentry *dentry,
+					 unsigned int flags)
+{
+	_enter("%pd", dentry);
+
+	if (flags & LOOKUP_CREATE)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	if (dentry->d_name.len >= AFSNAMEMAX) {
+		_leave(" = -ENAMETOOLONG");
+		return ERR_PTR(-ENAMETOOLONG);
+	}
+
+	if (dentry->d_name.len == 5 &&
+	    memcmp(dentry->d_name.name, "@cell", 5) == 0)
+		return afs_lookup_atcell(dir, dentry, 2);
+
+	if (dentry->d_name.len == 6 &&
+	    memcmp(dentry->d_name.name, ".@cell", 6) == 0)
+		return afs_lookup_atcell(dir, dentry, 3);
+
+	return afs_dynroot_lookup_cell(dir, dentry, flags);
+}
+
+const struct inode_operations afs_dynroot_inode_operations = {
+	.lookup		= afs_dynroot_lookup,
+};
+
+static void afs_dynroot_d_release(struct dentry *dentry)
+{
+	struct afs_cell *cell = dentry->d_fsdata;
+
+	afs_unuse_cell(cell, afs_cell_trace_unuse_dynroot_mntpt);
+}
+
+/*
+ * Keep @cell symlink dentries around, but only keep cell autodirs when they're
+ * being used.
+ */
+static int afs_dynroot_delete_dentry(const struct dentry *dentry)
+{
+	const struct qstr *name = &dentry->d_name;
+
+	if (name->len == 5 && memcmp(name->name, "@cell", 5) == 0)
+		return 0;
+	if (name->len == 6 && memcmp(name->name, ".@cell", 6) == 0)
+		return 0;
+	return 1;
+}
+
+const struct dentry_operations afs_dynroot_dentry_operations = {
+	.d_delete	= afs_dynroot_delete_dentry,
+	.d_release	= afs_dynroot_d_release,
+	.d_automount	= afs_d_automount,
+};
+
+static void afs_atcell_delayed_put_cell(void *arg)
+{
+	struct afs_cell *cell = arg;
+
+	afs_put_cell(cell, afs_cell_trace_put_atcell);
+}
+
+/*
+ * Read @cell or .@cell symlinks.
+ */
+static const char *afs_atcell_get_link(struct dentry *dentry, struct inode *inode,
+				       struct delayed_call *done)
+{
+	struct afs_vnode *vnode = AFS_FS_I(inode);
+	struct afs_cell *cell;
+	struct afs_net *net = afs_i2net(inode);
+	const char *name;
+	bool dotted = vnode->fid.vnode == 3;
+
+	if (!rcu_access_pointer(net->ws_cell))
+		return ERR_PTR(-ENOENT);
+
+	if (!dentry) {
+		/* We're in RCU-pathwalk. */
+		cell = rcu_dereference(net->ws_cell);
+		if (dotted)
+			name = cell->name - 1;
+		else
+			name = cell->name;
+		/* Shouldn't need to set a delayed call. */
+		return name;
+	}
+
+	down_read(&net->cells_lock);
+
+	cell = rcu_dereference_protected(net->ws_cell, lockdep_is_held(&net->cells_lock));
+	if (dotted)
+		name = cell->name - 1;
+	else
+		name = cell->name;
+	afs_get_cell(cell, afs_cell_trace_get_atcell);
+	set_delayed_call(done, afs_atcell_delayed_put_cell, cell);
+
+	up_read(&net->cells_lock);
+	return name;
+}
+
+static const struct inode_operations afs_atcell_inode_operations = {
+	.get_link	= afs_atcell_get_link,
+};
+
+/*
+ * Create an inode for the @cell or .@cell symlinks.
+ */
+static struct dentry *afs_lookup_atcell(struct inode *dir, struct dentry *dentry, ino_t ino)
+{
+	struct afs_vnode *vnode;
+	struct inode *inode;
+	struct afs_fid fid = { .vnode = ino, .unique = 1, };
+
+	inode = iget5_locked(dir->i_sb, fid.vnode,
+			     afs_iget5_pseudo_test, afs_iget5_pseudo_set, &fid);
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
+
+	vnode = AFS_FS_I(inode);
+
+	if (inode_state_read_once(inode) & I_NEW) {
+		netfs_inode_init(&vnode->netfs, NULL, false);
+		simple_inode_init_ts(inode);
+		set_nlink(inode, 1);
+		inode->i_size		= 0;
+		inode->i_mode		= S_IFLNK | 0555;
+		inode->i_op		= &afs_atcell_inode_operations;
+		inode->i_uid		= GLOBAL_ROOT_UID;
+		inode->i_gid		= GLOBAL_ROOT_GID;
+		inode->i_blocks		= 0;
+		inode->i_generation	= 0;
+		inode->i_flags		|= S_NOATIME;
+
+		unlock_new_inode(inode);
+	}
+	return d_splice_alias(inode, dentry);
+}
+
+/*
+ * Transcribe the cell database into readdir content under net->cells_lock.
+ * Each cell produces two entries, one prefixed with a dot and one not.
+ */
+static int afs_dynroot_readdir_cells(struct afs_net *net, struct dir_context *ctx)
+{
+	const struct afs_cell *cell;
+	loff_t newpos;
+
+	_enter("%llu", ctx->pos);
+
+	for (;;) {
+		unsigned int ix = ctx->pos >> 1;
+
+		cell = idr_get_next(&net->cells_dyn_ino, &ix);
+		if (!cell)
+			return 0;
+		if (READ_ONCE(cell->state) == AFS_CELL_REMOVING ||
+		    READ_ONCE(cell->state) == AFS_CELL_DEAD) {
+			ctx->pos += 2;
+			ctx->pos &= ~1;
+			continue;
+		}
+
+		newpos = ix << 1;
+		if (newpos > ctx->pos)
+			ctx->pos = newpos;
+
+		_debug("pos %llu -> cell %u", ctx->pos, cell->dynroot_ino);
+
+		if ((ctx->pos & 1) == 0) {
+			if (!dir_emit(ctx, cell->name, cell->name_len,
+				      cell->dynroot_ino, DT_DIR))
+				return 0;
+			ctx->pos++;
+		}
+		if ((ctx->pos & 1) == 1) {
+			if (!dir_emit(ctx, cell->name - 1, cell->name_len + 1,
+				      cell->dynroot_ino + 1, DT_DIR))
+				return 0;
+			ctx->pos++;
+		}
+	}
+	return 0;
+}
+
+/*
+ * Read the AFS dynamic root directory.  This produces a list of cellnames,
+ * dotted and undotted, along with @cell and .@cell links if configured.
+ */
+static int afs_dynroot_readdir(struct file *file, struct dir_context *ctx)
+{
+	struct afs_net *net = afs_d2net(file->f_path.dentry);
+	int ret = 0;
+
+	if (!dir_emit_dots(file, ctx))
+		return 0;
+
+	if (ctx->pos == 2) {
+		if (rcu_access_pointer(net->ws_cell) &&
+		    !dir_emit(ctx, "@cell", 5, 2, DT_LNK))
+			return 0;
+		ctx->pos = 3;
+	}
+	if (ctx->pos == 3) {
+		if (rcu_access_pointer(net->ws_cell) &&
+		    !dir_emit(ctx, ".@cell", 6, 3, DT_LNK))
+			return 0;
+		ctx->pos = 4;
+	}
+
+	if ((unsigned long long)ctx->pos <= AFS_MAX_DYNROOT_CELL_INO) {
+		down_read(&net->cells_lock);
+		ret = afs_dynroot_readdir_cells(net, ctx);
+		up_read(&net->cells_lock);
+	}
+	return ret;
+}
+
+static const struct file_operations afs_dynroot_file_operations = {
+	.llseek		= generic_file_llseek,
+	.read		= generic_read_dir,
+	.iterate_shared	= afs_dynroot_readdir,
+	.fsync		= noop_fsync,
+};
+
+/*
+ * Create an inode for a dynamic root directory.
+ */
+struct inode *afs_dynroot_iget_root(struct super_block *sb)
+{
+	struct afs_super_info *as = AFS_FS_S(sb);
+	struct afs_vnode *vnode;
+	struct inode *inode;
+	struct afs_fid fid = { .vid = 0, .vnode = 1, .unique = 1,};
+
+	if (as->volume)
+		fid.vid = as->volume->vid;
+
+	inode = iget5_locked(sb, fid.vnode,
+			     afs_iget5_pseudo_test, afs_iget5_pseudo_set, &fid);
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
+
+	vnode = AFS_FS_I(inode);
+
+	/* there shouldn't be an existing inode */
+	if (inode_state_read_once(inode) & I_NEW) {
+		netfs_inode_init(&vnode->netfs, NULL, false);
+		simple_inode_init_ts(inode);
+		set_nlink(inode, 2);
+		inode->i_size		= 0;
+		inode->i_mode		= S_IFDIR | 0555;
+		inode->i_op		= &afs_dynroot_inode_operations;
+		inode->i_fop		= &afs_dynroot_file_operations;
+		inode->i_uid		= GLOBAL_ROOT_UID;
+		inode->i_gid		= GLOBAL_ROOT_GID;
+		inode->i_blocks		= 0;
+		inode->i_generation	= 0;
+		inode->i_flags		|= S_NOATIME;
+
+		set_bit(AFS_VNODE_PSEUDODIR, &vnode->flags);
+		unlock_new_inode(inode);
+	}
+	_leave(" = %p", inode);
+	return inode;
+}