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, 889 insertions, 0 deletions

diff --git a/fs/jffs2/dir.c b/fs/jffs2/dir.c
new file mode 100644
index 000000000..656c92086
--- /dev/null
+++ b/fs/jffs2/dir.c
@@ -0,0 +1,889 @@
+/*
+ * JFFS2 -- Journalling Flash File System, Version 2.
+ *
+ * Copyright © 2001-2007 Red Hat, Inc.
+ * Copyright © 2004-2010 David Woodhouse <dwmw2@infradead.org>
+ *
+ * Created by David Woodhouse <dwmw2@infradead.org>
+ *
+ * For licensing information, see the file 'LICENCE' in this directory.
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/filelock.h>
+#include <linux/crc32.h>
+#include <linux/jffs2.h>
+#include "jffs2_fs_i.h"
+#include "jffs2_fs_sb.h"
+#include <linux/time.h>
+#include "nodelist.h"
+
+static int jffs2_readdir (struct file *, struct dir_context *);
+
+static int jffs2_create (struct mnt_idmap *, struct inode *,
+		         struct dentry *, umode_t);
+static struct dentry *jffs2_lookup (struct inode *,struct dentry *,
+				    unsigned int);
+static int jffs2_link (struct dentry *,struct inode *,struct dentry *);
+static int jffs2_unlink (struct inode *,struct dentry *);
+static int jffs2_symlink (struct mnt_idmap *, struct inode *,
+			  struct dentry *, const char *);
+static struct dentry *jffs2_mkdir (struct mnt_idmap *, struct inode *,struct dentry *,
+				   umode_t);
+static int jffs2_rmdir (struct inode *,struct dentry *);
+static int jffs2_mknod (struct mnt_idmap *, struct inode *,struct dentry *,
+			umode_t,dev_t);
+static int jffs2_rename (struct mnt_idmap *, struct inode *,
+			 struct dentry *, struct inode *, struct dentry *,
+			 unsigned int);
+
+const struct file_operations jffs2_dir_operations =
+{
+	.read =		generic_read_dir,
+	.iterate_shared=jffs2_readdir,
+	.unlocked_ioctl=jffs2_ioctl,
+	.fsync =	jffs2_fsync,
+	.llseek =	generic_file_llseek,
+	.setlease =	generic_setlease,
+};
+
+
+const struct inode_operations jffs2_dir_inode_operations =
+{
+	.create =	jffs2_create,
+	.lookup =	jffs2_lookup,
+	.link =		jffs2_link,
+	.unlink =	jffs2_unlink,
+	.symlink =	jffs2_symlink,
+	.mkdir =	jffs2_mkdir,
+	.rmdir =	jffs2_rmdir,
+	.mknod =	jffs2_mknod,
+	.rename =	jffs2_rename,
+	.get_inode_acl =	jffs2_get_acl,
+	.set_acl =	jffs2_set_acl,
+	.setattr =	jffs2_setattr,
+	.listxattr =	jffs2_listxattr,
+};
+
+/***********************************************************************/
+
+
+/* We keep the dirent list sorted in increasing order of name hash,
+   and we use the same hash function as the dentries. Makes this
+   nice and simple
+*/
+static struct dentry *jffs2_lookup(struct inode *dir_i, struct dentry *target,
+				   unsigned int flags)
+{
+	struct jffs2_inode_info *dir_f;
+	struct jffs2_full_dirent *fd = NULL, *fd_list;
+	uint32_t ino = 0;
+	struct inode *inode = NULL;
+	unsigned int nhash;
+
+	jffs2_dbg(1, "jffs2_lookup()\n");
+
+	if (target->d_name.len > JFFS2_MAX_NAME_LEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	dir_f = JFFS2_INODE_INFO(dir_i);
+
+	/* The 'nhash' on the fd_list is not the same as the dentry hash */
+	nhash = full_name_hash(NULL, target->d_name.name, target->d_name.len);
+
+	mutex_lock(&dir_f->sem);
+
+	/* NB: The 2.2 backport will need to explicitly check for '.' and '..' here */
+	for (fd_list = dir_f->dents; fd_list && fd_list->nhash <= nhash; fd_list = fd_list->next) {
+		if (fd_list->nhash == nhash &&
+		    (!fd || fd_list->version > fd->version) &&
+		    strlen(fd_list->name) == target->d_name.len &&
+		    !strncmp(fd_list->name, target->d_name.name, target->d_name.len)) {
+			fd = fd_list;
+		}
+	}
+	if (fd)
+		ino = fd->ino;
+	mutex_unlock(&dir_f->sem);
+	if (ino) {
+		inode = jffs2_iget(dir_i->i_sb, ino);
+		if (IS_ERR(inode))
+			pr_warn("iget() failed for ino #%u\n", ino);
+	}
+
+	return d_splice_alias(inode, target);
+}
+
+/***********************************************************************/
+
+
+static int jffs2_readdir(struct file *file, struct dir_context *ctx)
+{
+	struct inode *inode = file_inode(file);
+	struct jffs2_inode_info *f = JFFS2_INODE_INFO(inode);
+	struct jffs2_full_dirent *fd;
+	unsigned long curofs = 1;
+
+	jffs2_dbg(1, "jffs2_readdir() for dir_i #%llu\n", inode->i_ino);
+
+	if (!dir_emit_dots(file, ctx))
+		return 0;
+
+	mutex_lock(&f->sem);
+	for (fd = f->dents; fd; fd = fd->next) {
+		curofs++;
+		/* First loop: curofs = 2; pos = 2 */
+		if (curofs < ctx->pos) {
+			jffs2_dbg(2, "Skipping dirent: \"%s\", ino #%u, type %d, because curofs %ld < offset %ld\n",
+				  fd->name, fd->ino, fd->type, curofs, (unsigned long)ctx->pos);
+			continue;
+		}
+		if (!fd->ino) {
+			jffs2_dbg(2, "Skipping deletion dirent \"%s\"\n",
+				  fd->name);
+			ctx->pos++;
+			continue;
+		}
+		jffs2_dbg(2, "Dirent %ld: \"%s\", ino #%u, type %d\n",
+			  (unsigned long)ctx->pos, fd->name, fd->ino, fd->type);
+		if (!dir_emit(ctx, fd->name, strlen(fd->name), fd->ino, fd->type))
+			break;
+		ctx->pos++;
+	}
+	mutex_unlock(&f->sem);
+	return 0;
+}
+
+/***********************************************************************/
+
+
+static int jffs2_create(struct mnt_idmap *idmap, struct inode *dir_i,
+			struct dentry *dentry, umode_t mode)
+{
+	struct jffs2_raw_inode *ri;
+	struct jffs2_inode_info *f, *dir_f;
+	struct jffs2_sb_info *c;
+	struct inode *inode;
+	int ret;
+
+	ri = jffs2_alloc_raw_inode();
+	if (!ri)
+		return -ENOMEM;
+
+	c = JFFS2_SB_INFO(dir_i->i_sb);
+
+	jffs2_dbg(1, "%s()\n", __func__);
+
+	inode = jffs2_new_inode(dir_i, mode, ri);
+
+	if (IS_ERR(inode)) {
+		jffs2_dbg(1, "jffs2_new_inode() failed\n");
+		jffs2_free_raw_inode(ri);
+		return PTR_ERR(inode);
+	}
+
+	inode->i_op = &jffs2_file_inode_operations;
+	inode->i_fop = &jffs2_file_operations;
+	inode->i_mapping->a_ops = &jffs2_file_address_operations;
+	inode->i_mapping->nrpages = 0;
+
+	f = JFFS2_INODE_INFO(inode);
+	dir_f = JFFS2_INODE_INFO(dir_i);
+
+	/* jffs2_do_create() will want to lock it, _after_ reserving
+	   space and taking c-alloc_sem. If we keep it locked here,
+	   lockdep gets unhappy (although it's a false positive;
+	   nothing else will be looking at this inode yet so there's
+	   no chance of AB-BA deadlock involving its f->sem). */
+	mutex_unlock(&f->sem);
+
+	ret = jffs2_do_create(c, dir_f, f, ri, &dentry->d_name);
+	if (ret)
+		goto fail;
+
+	inode_set_mtime_to_ts(dir_i,
+			      inode_set_ctime_to_ts(dir_i, ITIME(je32_to_cpu(ri->ctime))));
+
+	jffs2_free_raw_inode(ri);
+
+	jffs2_dbg(1, "%s(): Created ino #%llu with mode %o, nlink %d(%d). nrpages %ld\n",
+		  __func__, inode->i_ino, inode->i_mode, inode->i_nlink,
+		  f->inocache->pino_nlink, inode->i_mapping->nrpages);
+
+	d_instantiate_new(dentry, inode);
+	return 0;
+
+ fail:
+	iget_failed(inode);
+	jffs2_free_raw_inode(ri);
+	return ret;
+}
+
+/***********************************************************************/
+
+
+static int jffs2_unlink(struct inode *dir_i, struct dentry *dentry)
+{
+	struct jffs2_sb_info *c = JFFS2_SB_INFO(dir_i->i_sb);
+	struct jffs2_inode_info *dir_f = JFFS2_INODE_INFO(dir_i);
+	struct jffs2_inode_info *dead_f = JFFS2_INODE_INFO(d_inode(dentry));
+	int ret;
+	uint32_t now = JFFS2_NOW();
+
+	ret = jffs2_do_unlink(c, dir_f, dentry->d_name.name,
+			      dentry->d_name.len, dead_f, now);
+	if (dead_f->inocache)
+		set_nlink(d_inode(dentry), dead_f->inocache->pino_nlink);
+	if (!ret)
+		inode_set_mtime_to_ts(dir_i,
+				      inode_set_ctime_to_ts(dir_i, ITIME(now)));
+	return ret;
+}
+/***********************************************************************/
+
+
+static int jffs2_link (struct dentry *old_dentry, struct inode *dir_i, struct dentry *dentry)
+{
+	struct jffs2_sb_info *c = JFFS2_SB_INFO(old_dentry->d_sb);
+	struct jffs2_inode_info *f = JFFS2_INODE_INFO(d_inode(old_dentry));
+	struct jffs2_inode_info *dir_f = JFFS2_INODE_INFO(dir_i);
+	int ret;
+	uint8_t type;
+	uint32_t now;
+
+	/* Don't let people make hard links to bad inodes. */
+	if (!f->inocache)
+		return -EIO;
+
+	if (d_is_dir(old_dentry))
+		return -EPERM;
+
+	/* XXX: This is ugly */
+	type = (d_inode(old_dentry)->i_mode & S_IFMT) >> 12;
+	if (!type) type = DT_REG;
+
+	now = JFFS2_NOW();
+	ret = jffs2_do_link(c, dir_f, f->inocache->ino, type, dentry->d_name.name, dentry->d_name.len, now);
+
+	if (!ret) {
+		mutex_lock(&f->sem);
+		set_nlink(d_inode(old_dentry), ++f->inocache->pino_nlink);
+		mutex_unlock(&f->sem);
+		d_instantiate(dentry, d_inode(old_dentry));
+		inode_set_mtime_to_ts(dir_i,
+				      inode_set_ctime_to_ts(dir_i, ITIME(now)));
+		ihold(d_inode(old_dentry));
+	}
+	return ret;
+}
+
+/***********************************************************************/
+
+static int jffs2_symlink (struct mnt_idmap *idmap, struct inode *dir_i,
+			  struct dentry *dentry, const char *target)
+{
+	struct jffs2_inode_info *f, *dir_f;
+	struct jffs2_sb_info *c;
+	struct inode *inode;
+	struct jffs2_raw_inode *ri;
+	struct jffs2_raw_dirent *rd;
+	struct jffs2_full_dnode *fn;
+	struct jffs2_full_dirent *fd;
+	int namelen;
+	uint32_t alloclen;
+	int ret, targetlen = strlen(target);
+
+	/* FIXME: If you care. We'd need to use frags for the target
+	   if it grows much more than this */
+	if (targetlen > 254)
+		return -ENAMETOOLONG;
+
+	ri = jffs2_alloc_raw_inode();
+
+	if (!ri)
+		return -ENOMEM;
+
+	c = JFFS2_SB_INFO(dir_i->i_sb);
+
+	/* Try to reserve enough space for both node and dirent.
+	 * Just the node will do for now, though
+	 */
+	namelen = dentry->d_name.len;
+	ret = jffs2_reserve_space(c, sizeof(*ri) + targetlen, &alloclen,
+				  ALLOC_NORMAL, JFFS2_SUMMARY_INODE_SIZE);
+
+	if (ret) {
+		jffs2_free_raw_inode(ri);
+		return ret;
+	}
+
+	inode = jffs2_new_inode(dir_i, S_IFLNK | S_IRWXUGO, ri);
+
+	if (IS_ERR(inode)) {
+		jffs2_free_raw_inode(ri);
+		jffs2_complete_reservation(c);
+		return PTR_ERR(inode);
+	}
+
+	inode->i_op = &jffs2_symlink_inode_operations;
+
+	f = JFFS2_INODE_INFO(inode);
+
+	inode->i_size = targetlen;
+	ri->isize = ri->dsize = ri->csize = cpu_to_je32(inode->i_size);
+	ri->totlen = cpu_to_je32(sizeof(*ri) + inode->i_size);
+	ri->hdr_crc = cpu_to_je32(crc32(0, ri, sizeof(struct jffs2_unknown_node)-4));
+
+	ri->compr = JFFS2_COMPR_NONE;
+	ri->data_crc = cpu_to_je32(crc32(0, target, targetlen));
+	ri->node_crc = cpu_to_je32(crc32(0, ri, sizeof(*ri)-8));
+
+	fn = jffs2_write_dnode(c, f, ri, target, targetlen, ALLOC_NORMAL);
+
+	jffs2_free_raw_inode(ri);
+
+	if (IS_ERR(fn)) {
+		/* Eeek. Wave bye bye */
+		mutex_unlock(&f->sem);
+		jffs2_complete_reservation(c);
+		ret = PTR_ERR(fn);
+		goto fail;
+	}
+
+	/* We use f->target field to store the target path. */
+	f->target = kmemdup(target, targetlen + 1, GFP_KERNEL);
+	if (!f->target) {
+		pr_warn("Can't allocate %d bytes of memory\n", targetlen + 1);
+		mutex_unlock(&f->sem);
+		jffs2_complete_reservation(c);
+		ret = -ENOMEM;
+		goto fail;
+	}
+	inode->i_link = f->target;
+
+	jffs2_dbg(1, "%s(): symlink's target '%s' cached\n",
+		  __func__, (char *)f->target);
+
+	/* No data here. Only a metadata node, which will be
+	   obsoleted by the first data write
+	*/
+	f->metadata = fn;
+	mutex_unlock(&f->sem);
+
+	jffs2_complete_reservation(c);
+
+	ret = jffs2_init_security(inode, dir_i, &dentry->d_name);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_init_acl_post(inode);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_reserve_space(c, sizeof(*rd)+namelen, &alloclen,
+				  ALLOC_NORMAL, JFFS2_SUMMARY_DIRENT_SIZE(namelen));
+	if (ret)
+		goto fail;
+
+	rd = jffs2_alloc_raw_dirent();
+	if (!rd) {
+		/* Argh. Now we treat it like a normal delete */
+		jffs2_complete_reservation(c);
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	dir_f = JFFS2_INODE_INFO(dir_i);
+	mutex_lock(&dir_f->sem);
+
+	rd->magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
+	rd->nodetype = cpu_to_je16(JFFS2_NODETYPE_DIRENT);
+	rd->totlen = cpu_to_je32(sizeof(*rd) + namelen);
+	rd->hdr_crc = cpu_to_je32(crc32(0, rd, sizeof(struct jffs2_unknown_node)-4));
+
+	rd->pino = cpu_to_je32(dir_i->i_ino);
+	rd->version = cpu_to_je32(++dir_f->highest_version);
+	rd->ino = cpu_to_je32(inode->i_ino);
+	rd->mctime = cpu_to_je32(JFFS2_NOW());
+	rd->nsize = namelen;
+	rd->type = DT_LNK;
+	rd->node_crc = cpu_to_je32(crc32(0, rd, sizeof(*rd)-8));
+	rd->name_crc = cpu_to_je32(crc32(0, dentry->d_name.name, namelen));
+
+	fd = jffs2_write_dirent(c, dir_f, rd, dentry->d_name.name, namelen, ALLOC_NORMAL);
+
+	if (IS_ERR(fd)) {
+		/* dirent failed to write. Delete the inode normally
+		   as if it were the final unlink() */
+		jffs2_complete_reservation(c);
+		jffs2_free_raw_dirent(rd);
+		mutex_unlock(&dir_f->sem);
+		ret = PTR_ERR(fd);
+		goto fail;
+	}
+
+	inode_set_mtime_to_ts(dir_i,
+			      inode_set_ctime_to_ts(dir_i, ITIME(je32_to_cpu(rd->mctime))));
+
+	jffs2_free_raw_dirent(rd);
+
+	/* Link the fd into the inode's list, obsoleting an old
+	   one if necessary. */
+	jffs2_add_fd_to_list(c, fd, &dir_f->dents);
+
+	mutex_unlock(&dir_f->sem);
+	jffs2_complete_reservation(c);
+
+	d_instantiate_new(dentry, inode);
+	return 0;
+
+ fail:
+	iget_failed(inode);
+	return ret;
+}
+
+
+static struct dentry *jffs2_mkdir (struct mnt_idmap *idmap, struct inode *dir_i,
+				   struct dentry *dentry, umode_t mode)
+{
+	struct jffs2_inode_info *f, *dir_f;
+	struct jffs2_sb_info *c;
+	struct inode *inode;
+	struct jffs2_raw_inode *ri;
+	struct jffs2_raw_dirent *rd;
+	struct jffs2_full_dnode *fn;
+	struct jffs2_full_dirent *fd;
+	int namelen;
+	uint32_t alloclen;
+	int ret;
+
+	ri = jffs2_alloc_raw_inode();
+	if (!ri)
+		return ERR_PTR(-ENOMEM);
+
+	c = JFFS2_SB_INFO(dir_i->i_sb);
+
+	/* Try to reserve enough space for both node and dirent.
+	 * Just the node will do for now, though
+	 */
+	namelen = dentry->d_name.len;
+	ret = jffs2_reserve_space(c, sizeof(*ri), &alloclen, ALLOC_NORMAL,
+				  JFFS2_SUMMARY_INODE_SIZE);
+
+	if (ret) {
+		jffs2_free_raw_inode(ri);
+		return ERR_PTR(ret);
+	}
+
+	inode = jffs2_new_inode(dir_i, mode, ri);
+
+	if (IS_ERR(inode)) {
+		jffs2_free_raw_inode(ri);
+		jffs2_complete_reservation(c);
+		return ERR_CAST(inode);
+	}
+
+	inode->i_op = &jffs2_dir_inode_operations;
+	inode->i_fop = &jffs2_dir_operations;
+
+	f = JFFS2_INODE_INFO(inode);
+
+	/* Directories get nlink 2 at start */
+	set_nlink(inode, 2);
+	/* but ic->pino_nlink is the parent ino# */
+	f->inocache->pino_nlink = dir_i->i_ino;
+
+	ri->data_crc = cpu_to_je32(0);
+	ri->node_crc = cpu_to_je32(crc32(0, ri, sizeof(*ri)-8));
+
+	fn = jffs2_write_dnode(c, f, ri, NULL, 0, ALLOC_NORMAL);
+
+	jffs2_free_raw_inode(ri);
+
+	if (IS_ERR(fn)) {
+		/* Eeek. Wave bye bye */
+		mutex_unlock(&f->sem);
+		jffs2_complete_reservation(c);
+		ret = PTR_ERR(fn);
+		goto fail;
+	}
+	/* No data here. Only a metadata node, which will be
+	   obsoleted by the first data write
+	*/
+	f->metadata = fn;
+	mutex_unlock(&f->sem);
+
+	jffs2_complete_reservation(c);
+
+	ret = jffs2_init_security(inode, dir_i, &dentry->d_name);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_init_acl_post(inode);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_reserve_space(c, sizeof(*rd)+namelen, &alloclen,
+				  ALLOC_NORMAL, JFFS2_SUMMARY_DIRENT_SIZE(namelen));
+	if (ret)
+		goto fail;
+
+	rd = jffs2_alloc_raw_dirent();
+	if (!rd) {
+		/* Argh. Now we treat it like a normal delete */
+		jffs2_complete_reservation(c);
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	dir_f = JFFS2_INODE_INFO(dir_i);
+	mutex_lock(&dir_f->sem);
+
+	rd->magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
+	rd->nodetype = cpu_to_je16(JFFS2_NODETYPE_DIRENT);
+	rd->totlen = cpu_to_je32(sizeof(*rd) + namelen);
+	rd->hdr_crc = cpu_to_je32(crc32(0, rd, sizeof(struct jffs2_unknown_node)-4));
+
+	rd->pino = cpu_to_je32(dir_i->i_ino);
+	rd->version = cpu_to_je32(++dir_f->highest_version);
+	rd->ino = cpu_to_je32(inode->i_ino);
+	rd->mctime = cpu_to_je32(JFFS2_NOW());
+	rd->nsize = namelen;
+	rd->type = DT_DIR;
+	rd->node_crc = cpu_to_je32(crc32(0, rd, sizeof(*rd)-8));
+	rd->name_crc = cpu_to_je32(crc32(0, dentry->d_name.name, namelen));
+
+	fd = jffs2_write_dirent(c, dir_f, rd, dentry->d_name.name, namelen, ALLOC_NORMAL);
+
+	if (IS_ERR(fd)) {
+		/* dirent failed to write. Delete the inode normally
+		   as if it were the final unlink() */
+		jffs2_complete_reservation(c);
+		jffs2_free_raw_dirent(rd);
+		mutex_unlock(&dir_f->sem);
+		ret = PTR_ERR(fd);
+		goto fail;
+	}
+
+	inode_set_mtime_to_ts(dir_i,
+			      inode_set_ctime_to_ts(dir_i, ITIME(je32_to_cpu(rd->mctime))));
+	inc_nlink(dir_i);
+
+	jffs2_free_raw_dirent(rd);
+
+	/* Link the fd into the inode's list, obsoleting an old
+	   one if necessary. */
+	jffs2_add_fd_to_list(c, fd, &dir_f->dents);
+
+	mutex_unlock(&dir_f->sem);
+	jffs2_complete_reservation(c);
+
+	d_instantiate_new(dentry, inode);
+	return NULL;
+
+ fail:
+	iget_failed(inode);
+	return ERR_PTR(ret);
+}
+
+static int jffs2_rmdir (struct inode *dir_i, struct dentry *dentry)
+{
+	struct jffs2_sb_info *c = JFFS2_SB_INFO(dir_i->i_sb);
+	struct jffs2_inode_info *dir_f = JFFS2_INODE_INFO(dir_i);
+	struct jffs2_inode_info *f = JFFS2_INODE_INFO(d_inode(dentry));
+	struct jffs2_full_dirent *fd;
+	int ret;
+	uint32_t now = JFFS2_NOW();
+
+	mutex_lock(&f->sem);
+	for (fd = f->dents ; fd; fd = fd->next) {
+		if (fd->ino) {
+			mutex_unlock(&f->sem);
+			return -ENOTEMPTY;
+		}
+	}
+	mutex_unlock(&f->sem);
+
+	ret = jffs2_do_unlink(c, dir_f, dentry->d_name.name,
+			      dentry->d_name.len, f, now);
+	if (!ret) {
+		inode_set_mtime_to_ts(dir_i,
+				      inode_set_ctime_to_ts(dir_i, ITIME(now)));
+		clear_nlink(d_inode(dentry));
+		drop_nlink(dir_i);
+	}
+	return ret;
+}
+
+static int jffs2_mknod (struct mnt_idmap *idmap, struct inode *dir_i,
+		        struct dentry *dentry, umode_t mode, dev_t rdev)
+{
+	struct jffs2_inode_info *f, *dir_f;
+	struct jffs2_sb_info *c;
+	struct inode *inode;
+	struct jffs2_raw_inode *ri;
+	struct jffs2_raw_dirent *rd;
+	struct jffs2_full_dnode *fn;
+	struct jffs2_full_dirent *fd;
+	int namelen;
+	union jffs2_device_node dev;
+	int devlen = 0;
+	uint32_t alloclen;
+	int ret;
+
+	ri = jffs2_alloc_raw_inode();
+	if (!ri)
+		return -ENOMEM;
+
+	c = JFFS2_SB_INFO(dir_i->i_sb);
+
+	if (S_ISBLK(mode) || S_ISCHR(mode))
+		devlen = jffs2_encode_dev(&dev, rdev);
+
+	/* Try to reserve enough space for both node and dirent.
+	 * Just the node will do for now, though
+	 */
+	namelen = dentry->d_name.len;
+	ret = jffs2_reserve_space(c, sizeof(*ri) + devlen, &alloclen,
+				  ALLOC_NORMAL, JFFS2_SUMMARY_INODE_SIZE);
+
+	if (ret) {
+		jffs2_free_raw_inode(ri);
+		return ret;
+	}
+
+	inode = jffs2_new_inode(dir_i, mode, ri);
+
+	if (IS_ERR(inode)) {
+		jffs2_free_raw_inode(ri);
+		jffs2_complete_reservation(c);
+		return PTR_ERR(inode);
+	}
+	inode->i_op = &jffs2_file_inode_operations;
+	init_special_inode(inode, inode->i_mode, rdev);
+
+	f = JFFS2_INODE_INFO(inode);
+
+	ri->dsize = ri->csize = cpu_to_je32(devlen);
+	ri->totlen = cpu_to_je32(sizeof(*ri) + devlen);
+	ri->hdr_crc = cpu_to_je32(crc32(0, ri, sizeof(struct jffs2_unknown_node)-4));
+
+	ri->compr = JFFS2_COMPR_NONE;
+	ri->data_crc = cpu_to_je32(crc32(0, &dev, devlen));
+	ri->node_crc = cpu_to_je32(crc32(0, ri, sizeof(*ri)-8));
+
+	fn = jffs2_write_dnode(c, f, ri, (char *)&dev, devlen, ALLOC_NORMAL);
+
+	jffs2_free_raw_inode(ri);
+
+	if (IS_ERR(fn)) {
+		/* Eeek. Wave bye bye */
+		mutex_unlock(&f->sem);
+		jffs2_complete_reservation(c);
+		ret = PTR_ERR(fn);
+		goto fail;
+	}
+	/* No data here. Only a metadata node, which will be
+	   obsoleted by the first data write
+	*/
+	f->metadata = fn;
+	mutex_unlock(&f->sem);
+
+	jffs2_complete_reservation(c);
+
+	ret = jffs2_init_security(inode, dir_i, &dentry->d_name);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_init_acl_post(inode);
+	if (ret)
+		goto fail;
+
+	ret = jffs2_reserve_space(c, sizeof(*rd)+namelen, &alloclen,
+				  ALLOC_NORMAL, JFFS2_SUMMARY_DIRENT_SIZE(namelen));
+	if (ret)
+		goto fail;
+
+	rd = jffs2_alloc_raw_dirent();
+	if (!rd) {
+		/* Argh. Now we treat it like a normal delete */
+		jffs2_complete_reservation(c);
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	dir_f = JFFS2_INODE_INFO(dir_i);
+	mutex_lock(&dir_f->sem);
+
+	rd->magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
+	rd->nodetype = cpu_to_je16(JFFS2_NODETYPE_DIRENT);
+	rd->totlen = cpu_to_je32(sizeof(*rd) + namelen);
+	rd->hdr_crc = cpu_to_je32(crc32(0, rd, sizeof(struct jffs2_unknown_node)-4));
+
+	rd->pino = cpu_to_je32(dir_i->i_ino);
+	rd->version = cpu_to_je32(++dir_f->highest_version);
+	rd->ino = cpu_to_je32(inode->i_ino);
+	rd->mctime = cpu_to_je32(JFFS2_NOW());
+	rd->nsize = namelen;
+
+	/* XXX: This is ugly. */
+	rd->type = (mode & S_IFMT) >> 12;
+
+	rd->node_crc = cpu_to_je32(crc32(0, rd, sizeof(*rd)-8));
+	rd->name_crc = cpu_to_je32(crc32(0, dentry->d_name.name, namelen));
+
+	fd = jffs2_write_dirent(c, dir_f, rd, dentry->d_name.name, namelen, ALLOC_NORMAL);
+
+	if (IS_ERR(fd)) {
+		/* dirent failed to write. Delete the inode normally
+		   as if it were the final unlink() */
+		jffs2_complete_reservation(c);
+		jffs2_free_raw_dirent(rd);
+		mutex_unlock(&dir_f->sem);
+		ret = PTR_ERR(fd);
+		goto fail;
+	}
+
+	inode_set_mtime_to_ts(dir_i,
+			      inode_set_ctime_to_ts(dir_i, ITIME(je32_to_cpu(rd->mctime))));
+
+	jffs2_free_raw_dirent(rd);
+
+	/* Link the fd into the inode's list, obsoleting an old
+	   one if necessary. */
+	jffs2_add_fd_to_list(c, fd, &dir_f->dents);
+
+	mutex_unlock(&dir_f->sem);
+	jffs2_complete_reservation(c);
+
+	d_instantiate_new(dentry, inode);
+	return 0;
+
+ fail:
+	iget_failed(inode);
+	return ret;
+}
+
+static int jffs2_rename (struct mnt_idmap *idmap,
+			 struct inode *old_dir_i, struct dentry *old_dentry,
+			 struct inode *new_dir_i, struct dentry *new_dentry,
+			 unsigned int flags)
+{
+	int ret;
+	struct jffs2_sb_info *c = JFFS2_SB_INFO(old_dir_i->i_sb);
+	struct jffs2_inode_info *victim_f = NULL;
+	uint8_t type;
+	uint32_t now;
+
+	if (flags & ~RENAME_NOREPLACE)
+		return -EINVAL;
+
+	/* The VFS will check for us and prevent trying to rename a
+	 * file over a directory and vice versa, but if it's a directory,
+	 * the VFS can't check whether the victim is empty. The filesystem
+	 * needs to do that for itself.
+	 */
+	if (d_really_is_positive(new_dentry)) {
+		victim_f = JFFS2_INODE_INFO(d_inode(new_dentry));
+		if (d_is_dir(new_dentry)) {
+			struct jffs2_full_dirent *fd;
+
+			mutex_lock(&victim_f->sem);
+			for (fd = victim_f->dents; fd; fd = fd->next) {
+				if (fd->ino) {
+					mutex_unlock(&victim_f->sem);
+					return -ENOTEMPTY;
+				}
+			}
+			mutex_unlock(&victim_f->sem);
+		}
+	}
+
+	/* XXX: We probably ought to alloc enough space for
+	   both nodes at the same time. Writing the new link,
+	   then getting -ENOSPC, is quite bad :)
+	*/
+
+	/* Make a hard link */
+
+	/* XXX: This is ugly */
+	type = (d_inode(old_dentry)->i_mode & S_IFMT) >> 12;
+	if (!type) type = DT_REG;
+
+	now = JFFS2_NOW();
+	ret = jffs2_do_link(c, JFFS2_INODE_INFO(new_dir_i),
+			    d_inode(old_dentry)->i_ino, type,
+			    new_dentry->d_name.name, new_dentry->d_name.len, now);
+
+	if (ret)
+		return ret;
+
+	if (victim_f) {
+		/* There was a victim. Kill it off nicely */
+		if (d_is_dir(new_dentry))
+			clear_nlink(d_inode(new_dentry));
+		else
+			drop_nlink(d_inode(new_dentry));
+		/* Don't oops if the victim was a dirent pointing to an
+		   inode which didn't exist. */
+		if (victim_f->inocache) {
+			mutex_lock(&victim_f->sem);
+			if (d_is_dir(new_dentry))
+				victim_f->inocache->pino_nlink = 0;
+			else
+				victim_f->inocache->pino_nlink--;
+			mutex_unlock(&victim_f->sem);
+		}
+	}
+
+	/* If it was a directory we moved, and there was no victim,
+	   increase i_nlink on its new parent */
+	if (d_is_dir(old_dentry) && !victim_f)
+		inc_nlink(new_dir_i);
+
+	/* Unlink the original */
+	ret = jffs2_do_unlink(c, JFFS2_INODE_INFO(old_dir_i),
+			      old_dentry->d_name.name, old_dentry->d_name.len, NULL, now);
+
+	/* We don't touch inode->i_nlink */
+
+	if (ret) {
+		/* Oh shit. We really ought to make a single node which can do both atomically */
+		struct jffs2_inode_info *f = JFFS2_INODE_INFO(d_inode(old_dentry));
+		mutex_lock(&f->sem);
+		inc_nlink(d_inode(old_dentry));
+		if (f->inocache && !d_is_dir(old_dentry))
+			f->inocache->pino_nlink++;
+		mutex_unlock(&f->sem);
+
+		pr_notice("%s(): Link succeeded, unlink failed (err %d). You now have a hard link\n",
+			  __func__, ret);
+		/*
+		 * We can't keep the target in dcache after that.
+		 * For one thing, we can't afford dentry aliases for directories.
+		 * For another, if there was a victim, we _can't_ set new inode
+		 * for that sucker and we have to trigger mount eviction - the
+		 * caller won't do it on its own since we are returning an error.
+		 */
+		d_invalidate(new_dentry);
+		inode_set_mtime_to_ts(new_dir_i,
+				      inode_set_ctime_to_ts(new_dir_i, ITIME(now)));
+		return ret;
+	}
+
+	if (d_is_dir(old_dentry))
+		drop_nlink(old_dir_i);
+
+	inode_set_mtime_to_ts(old_dir_i,
+			      inode_set_ctime_to_ts(old_dir_i, ITIME(now)));
+	inode_set_mtime_to_ts(new_dir_i,
+			      inode_set_ctime_to_ts(new_dir_i, ITIME(now)));
+
+	return 0;
+}
+