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, 484 insertions, 0 deletions

diff --git a/fs/vboxsf/super.c b/fs/vboxsf/super.c
new file mode 100644
index 000000000..a61fbab51
--- /dev/null
+++ b/fs/vboxsf/super.c
@@ -0,0 +1,484 @@
+// SPDX-License-Identifier: MIT
+/*
+ * VirtualBox Guest Shared Folders support: Virtual File System.
+ *
+ * Module initialization/finalization
+ * File system registration/deregistration
+ * Superblock reading
+ * Few utility functions
+ *
+ * Copyright (C) 2006-2018 Oracle Corporation
+ */
+
+#include <linux/idr.h>
+#include <linux/fs_parser.h>
+#include <linux/magic.h>
+#include <linux/module.h>
+#include <linux/nls.h>
+#include <linux/statfs.h>
+#include <linux/vbox_utils.h>
+#include "vfsmod.h"
+
+#define VBOXSF_SUPER_MAGIC 0x786f4256 /* 'VBox' little endian */
+
+static const unsigned char VBSF_MOUNT_SIGNATURE[4] __nonstring = "\000\377\376\375";
+
+static int follow_symlinks;
+module_param(follow_symlinks, int, 0444);
+MODULE_PARM_DESC(follow_symlinks,
+		 "Let host resolve symlinks rather than showing them");
+
+static DEFINE_IDA(vboxsf_bdi_ida);
+static DEFINE_MUTEX(vboxsf_setup_mutex);
+static bool vboxsf_setup_done;
+static struct super_operations vboxsf_super_ops; /* forward declaration */
+static struct kmem_cache *vboxsf_inode_cachep;
+
+static char * const vboxsf_default_nls = CONFIG_NLS_DEFAULT;
+
+enum  { opt_nls, opt_uid, opt_gid, opt_ttl, opt_dmode, opt_fmode,
+	opt_dmask, opt_fmask };
+
+static const struct fs_parameter_spec vboxsf_fs_parameters[] = {
+	fsparam_string	("nls",		opt_nls),
+	fsparam_uid	("uid",		opt_uid),
+	fsparam_gid	("gid",		opt_gid),
+	fsparam_u32	("ttl",		opt_ttl),
+	fsparam_u32oct	("dmode",	opt_dmode),
+	fsparam_u32oct	("fmode",	opt_fmode),
+	fsparam_u32oct	("dmask",	opt_dmask),
+	fsparam_u32oct	("fmask",	opt_fmask),
+	{}
+};
+
+static int vboxsf_parse_param(struct fs_context *fc, struct fs_parameter *param)
+{
+	struct vboxsf_fs_context *ctx = fc->fs_private;
+	struct fs_parse_result result;
+	int opt;
+
+	opt = fs_parse(fc, vboxsf_fs_parameters, param, &result);
+	if (opt < 0)
+		return opt;
+
+	switch (opt) {
+	case opt_nls:
+		if (ctx->nls_name || fc->purpose != FS_CONTEXT_FOR_MOUNT) {
+			vbg_err("vboxsf: Cannot reconfigure nls option\n");
+			return -EINVAL;
+		}
+		ctx->nls_name = param->string;
+		param->string = NULL;
+		break;
+	case opt_uid:
+		ctx->o.uid = result.uid;
+		break;
+	case opt_gid:
+		ctx->o.gid = result.gid;
+		break;
+	case opt_ttl:
+		ctx->o.ttl = msecs_to_jiffies(result.uint_32);
+		break;
+	case opt_dmode:
+		if (result.uint_32 & ~0777)
+			return -EINVAL;
+		ctx->o.dmode = result.uint_32;
+		ctx->o.dmode_set = true;
+		break;
+	case opt_fmode:
+		if (result.uint_32 & ~0777)
+			return -EINVAL;
+		ctx->o.fmode = result.uint_32;
+		ctx->o.fmode_set = true;
+		break;
+	case opt_dmask:
+		if (result.uint_32 & ~07777)
+			return -EINVAL;
+		ctx->o.dmask = result.uint_32;
+		break;
+	case opt_fmask:
+		if (result.uint_32 & ~07777)
+			return -EINVAL;
+		ctx->o.fmask = result.uint_32;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int vboxsf_fill_super(struct super_block *sb, struct fs_context *fc)
+{
+	struct vboxsf_fs_context *ctx = fc->fs_private;
+	struct shfl_string *folder_name, root_path;
+	struct vboxsf_sbi *sbi;
+	struct dentry *droot;
+	struct inode *iroot;
+	char *nls_name;
+	size_t size;
+	int err;
+
+	if (!fc->source)
+		return -EINVAL;
+
+	sbi = kzalloc_obj(*sbi);
+	if (!sbi)
+		return -ENOMEM;
+
+	sbi->o = ctx->o;
+	idr_init(&sbi->ino_idr);
+	spin_lock_init(&sbi->ino_idr_lock);
+	sbi->next_generation = 1;
+	sbi->bdi_id = -1;
+
+	/* Load nls if not utf8 */
+	nls_name = ctx->nls_name ? ctx->nls_name : vboxsf_default_nls;
+	if (strcmp(nls_name, "utf8") != 0) {
+		if (nls_name == vboxsf_default_nls)
+			sbi->nls = load_nls_default();
+		else
+			sbi->nls = load_nls(nls_name);
+
+		if (!sbi->nls) {
+			vbg_err("vboxsf: Count not load '%s' nls\n", nls_name);
+			err = -EINVAL;
+			goto fail_destroy_idr;
+		}
+	}
+
+	sbi->bdi_id = ida_alloc(&vboxsf_bdi_ida, GFP_KERNEL);
+	if (sbi->bdi_id < 0) {
+		err = sbi->bdi_id;
+		goto fail_free;
+	}
+
+	err = super_setup_bdi_name(sb, "vboxsf-%d", sbi->bdi_id);
+	if (err)
+		goto fail_free;
+	sb->s_bdi->ra_pages = 0;
+	sb->s_bdi->io_pages = 0;
+
+	/* Turn source into a shfl_string and map the folder */
+	size = strlen(fc->source) + 1;
+	folder_name = kmalloc(SHFLSTRING_HEADER_SIZE + size, GFP_KERNEL);
+	if (!folder_name) {
+		err = -ENOMEM;
+		goto fail_free;
+	}
+	folder_name->size = size;
+	folder_name->length = size - 1;
+	strscpy(folder_name->string.utf8, fc->source, size);
+	err = vboxsf_map_folder(folder_name, &sbi->root);
+	kfree(folder_name);
+	if (err) {
+		vbg_err("vboxsf: Host rejected mount of '%s' with error %d\n",
+			fc->source, err);
+		goto fail_free;
+	}
+
+	root_path.length = 1;
+	root_path.size = 2;
+	root_path.string.utf8[0] = '/';
+	root_path.string.utf8[1] = 0;
+	err = vboxsf_stat(sbi, &root_path, &sbi->root_info);
+	if (err)
+		goto fail_unmap;
+
+	/*
+	 * A failed query leaves sbi->case_insensitive false, so the
+	 * mount defaults to reporting case-sensitive behavior. Do not
+	 * fail the mount over an advisory attribute.
+	 */
+	vboxsf_query_case_sensitive(sbi);
+
+	sb->s_magic = VBOXSF_SUPER_MAGIC;
+	sb->s_blocksize = 1024;
+	sb->s_maxbytes = MAX_LFS_FILESIZE;
+	sb->s_op = &vboxsf_super_ops;
+	set_default_d_op(sb, &vboxsf_dentry_ops);
+
+	iroot = iget_locked(sb, 0);
+	if (!iroot) {
+		err = -ENOMEM;
+		goto fail_unmap;
+	}
+	vboxsf_init_inode(sbi, iroot, &sbi->root_info, false);
+	unlock_new_inode(iroot);
+
+	droot = d_make_root(iroot);
+	if (!droot) {
+		err = -ENOMEM;
+		goto fail_unmap;
+	}
+
+	sb->s_root = droot;
+	sb->s_fs_info = sbi;
+	return 0;
+
+fail_unmap:
+	vboxsf_unmap_folder(sbi->root);
+fail_free:
+	if (sbi->bdi_id >= 0)
+		ida_free(&vboxsf_bdi_ida, sbi->bdi_id);
+	if (sbi->nls)
+		unload_nls(sbi->nls);
+fail_destroy_idr:
+	idr_destroy(&sbi->ino_idr);
+	kfree(sbi);
+	return err;
+}
+
+static void vboxsf_inode_init_once(void *data)
+{
+	struct vboxsf_inode *sf_i = data;
+
+	mutex_init(&sf_i->handle_list_mutex);
+	inode_init_once(&sf_i->vfs_inode);
+}
+
+static struct inode *vboxsf_alloc_inode(struct super_block *sb)
+{
+	struct vboxsf_inode *sf_i;
+
+	sf_i = alloc_inode_sb(sb, vboxsf_inode_cachep, GFP_NOFS);
+	if (!sf_i)
+		return NULL;
+
+	sf_i->force_restat = 0;
+	INIT_LIST_HEAD(&sf_i->handle_list);
+
+	return &sf_i->vfs_inode;
+}
+
+static void vboxsf_free_inode(struct inode *inode)
+{
+	struct vboxsf_sbi *sbi = VBOXSF_SBI(inode->i_sb);
+	unsigned long flags;
+
+	spin_lock_irqsave(&sbi->ino_idr_lock, flags);
+	idr_remove(&sbi->ino_idr, inode->i_ino);
+	spin_unlock_irqrestore(&sbi->ino_idr_lock, flags);
+	kmem_cache_free(vboxsf_inode_cachep, VBOXSF_I(inode));
+}
+
+static void vboxsf_put_super(struct super_block *sb)
+{
+	struct vboxsf_sbi *sbi = VBOXSF_SBI(sb);
+
+	vboxsf_unmap_folder(sbi->root);
+	if (sbi->bdi_id >= 0)
+		ida_free(&vboxsf_bdi_ida, sbi->bdi_id);
+	if (sbi->nls)
+		unload_nls(sbi->nls);
+
+	/*
+	 * vboxsf_free_inode uses the idr, make sure all delayed rcu free
+	 * inodes are flushed.
+	 */
+	rcu_barrier();
+	idr_destroy(&sbi->ino_idr);
+	kfree(sbi);
+}
+
+static int vboxsf_statfs(struct dentry *dentry, struct kstatfs *stat)
+{
+	struct super_block *sb = dentry->d_sb;
+	struct shfl_volinfo shfl_volinfo;
+	struct vboxsf_sbi *sbi;
+	u32 buf_len;
+	int err;
+
+	sbi = VBOXSF_SBI(sb);
+	buf_len = sizeof(shfl_volinfo);
+	err = vboxsf_fsinfo(sbi->root, 0, SHFL_INFO_GET | SHFL_INFO_VOLUME,
+			    &buf_len, &shfl_volinfo);
+	if (err)
+		return err;
+
+	stat->f_type = VBOXSF_SUPER_MAGIC;
+	stat->f_bsize = shfl_volinfo.bytes_per_allocation_unit;
+
+	do_div(shfl_volinfo.total_allocation_bytes,
+	       shfl_volinfo.bytes_per_allocation_unit);
+	stat->f_blocks = shfl_volinfo.total_allocation_bytes;
+
+	do_div(shfl_volinfo.available_allocation_bytes,
+	       shfl_volinfo.bytes_per_allocation_unit);
+	stat->f_bfree  = shfl_volinfo.available_allocation_bytes;
+	stat->f_bavail = shfl_volinfo.available_allocation_bytes;
+
+	stat->f_files = 1000;
+	/*
+	 * Don't return 0 here since the guest may then think that it is not
+	 * possible to create any more files.
+	 */
+	stat->f_ffree = 1000000;
+	stat->f_fsid.val[0] = 0;
+	stat->f_fsid.val[1] = 0;
+	stat->f_namelen = 255;
+	return 0;
+}
+
+static struct super_operations vboxsf_super_ops = {
+	.alloc_inode	= vboxsf_alloc_inode,
+	.free_inode	= vboxsf_free_inode,
+	.put_super	= vboxsf_put_super,
+	.statfs		= vboxsf_statfs,
+};
+
+static int vboxsf_setup(void)
+{
+	int err;
+
+	mutex_lock(&vboxsf_setup_mutex);
+
+	if (vboxsf_setup_done)
+		goto success;
+
+	vboxsf_inode_cachep =
+		kmem_cache_create("vboxsf_inode_cache",
+				  sizeof(struct vboxsf_inode), 0,
+				  SLAB_RECLAIM_ACCOUNT | SLAB_ACCOUNT,
+				  vboxsf_inode_init_once);
+	if (!vboxsf_inode_cachep) {
+		err = -ENOMEM;
+		goto fail_nomem;
+	}
+
+	err = vboxsf_connect();
+	if (err) {
+		vbg_err("vboxsf: err %d connecting to guest PCI-device\n", err);
+		vbg_err("vboxsf: make sure you are inside a VirtualBox VM\n");
+		vbg_err("vboxsf: and check dmesg for vboxguest errors\n");
+		goto fail_free_cache;
+	}
+
+	err = vboxsf_set_utf8();
+	if (err) {
+		vbg_err("vboxsf_setutf8 error %d\n", err);
+		goto fail_disconnect;
+	}
+
+	if (!follow_symlinks) {
+		err = vboxsf_set_symlinks();
+		if (err)
+			vbg_warn("vboxsf: Unable to show symlinks: %d\n", err);
+	}
+
+	vboxsf_setup_done = true;
+success:
+	mutex_unlock(&vboxsf_setup_mutex);
+	return 0;
+
+fail_disconnect:
+	vboxsf_disconnect();
+fail_free_cache:
+	kmem_cache_destroy(vboxsf_inode_cachep);
+fail_nomem:
+	mutex_unlock(&vboxsf_setup_mutex);
+	return err;
+}
+
+static int vboxsf_parse_monolithic(struct fs_context *fc, void *data)
+{
+	if (data && !memcmp(data, VBSF_MOUNT_SIGNATURE, 4)) {
+		vbg_err("vboxsf: Old binary mount data not supported, remove obsolete mount.vboxsf and/or update your VBoxService.\n");
+		return -EINVAL;
+	}
+
+	return generic_parse_monolithic(fc, data);
+}
+
+static int vboxsf_get_tree(struct fs_context *fc)
+{
+	int err;
+
+	err = vboxsf_setup();
+	if (err)
+		return err;
+
+	return get_tree_nodev(fc, vboxsf_fill_super);
+}
+
+static int vboxsf_reconfigure(struct fs_context *fc)
+{
+	struct vboxsf_sbi *sbi = VBOXSF_SBI(fc->root->d_sb);
+	struct vboxsf_fs_context *ctx = fc->fs_private;
+	struct inode *iroot = fc->root->d_sb->s_root->d_inode;
+
+	/* Apply changed options to the root inode */
+	sbi->o = ctx->o;
+	vboxsf_init_inode(sbi, iroot, &sbi->root_info, true);
+
+	return 0;
+}
+
+static void vboxsf_free_fc(struct fs_context *fc)
+{
+	struct vboxsf_fs_context *ctx = fc->fs_private;
+
+	kfree(ctx->nls_name);
+	kfree(ctx);
+}
+
+static const struct fs_context_operations vboxsf_context_ops = {
+	.free			= vboxsf_free_fc,
+	.parse_param		= vboxsf_parse_param,
+	.parse_monolithic	= vboxsf_parse_monolithic,
+	.get_tree		= vboxsf_get_tree,
+	.reconfigure		= vboxsf_reconfigure,
+};
+
+static int vboxsf_init_fs_context(struct fs_context *fc)
+{
+	struct vboxsf_fs_context *ctx;
+
+	ctx = kzalloc_obj(*ctx);
+	if (!ctx)
+		return -ENOMEM;
+
+	current_uid_gid(&ctx->o.uid, &ctx->o.gid);
+
+	fc->fs_private = ctx;
+	fc->ops = &vboxsf_context_ops;
+	return 0;
+}
+
+static struct file_system_type vboxsf_fs_type = {
+	.owner			= THIS_MODULE,
+	.name			= "vboxsf",
+	.init_fs_context	= vboxsf_init_fs_context,
+	.kill_sb		= kill_anon_super
+};
+
+/* Module initialization/finalization handlers */
+static int __init vboxsf_init(void)
+{
+	return register_filesystem(&vboxsf_fs_type);
+}
+
+static void __exit vboxsf_fini(void)
+{
+	unregister_filesystem(&vboxsf_fs_type);
+
+	mutex_lock(&vboxsf_setup_mutex);
+	if (vboxsf_setup_done) {
+		vboxsf_disconnect();
+		/*
+		 * Make sure all delayed rcu free inodes are flushed
+		 * before we destroy the cache.
+		 */
+		rcu_barrier();
+		kmem_cache_destroy(vboxsf_inode_cachep);
+	}
+	mutex_unlock(&vboxsf_setup_mutex);
+}
+
+module_init(vboxsf_init);
+module_exit(vboxsf_fini);
+
+MODULE_DESCRIPTION("Oracle VM VirtualBox Module for Host File System Access");
+MODULE_AUTHOR("Oracle Corporation");
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS_FS("vboxsf");