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, 498 insertions, 0 deletions

diff --git a/fs/hpfs/anode.c b/fs/hpfs/anode.c
new file mode 100644
index 000000000..a4f5321ea
--- /dev/null
+++ b/fs/hpfs/anode.c
@@ -0,0 +1,498 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/fs/hpfs/anode.c
+ *
+ *  Mikulas Patocka (mikulas@artax.karlin.mff.cuni.cz), 1998-1999
+ *
+ *  handling HPFS anode tree that contains file allocation info
+ */
+
+#include "hpfs_fn.h"
+
+/* Find a sector in allocation tree */
+
+secno hpfs_bplus_lookup(struct super_block *s, struct inode *inode,
+		   struct bplus_header *btree, unsigned sec,
+		   struct buffer_head *bh)
+{
+	anode_secno a = -1;
+	struct anode *anode;
+	int i;
+	int c1, c2 = 0;
+	go_down:
+	if (hpfs_sb(s)->sb_chk) if (hpfs_stop_cycles(s, a, &c1, &c2, "hpfs_bplus_lookup")) return -1;
+	if (bp_internal(btree)) {
+		for (i = 0; i < btree->n_used_nodes; i++)
+			if (le32_to_cpu(btree->u.internal[i].file_secno) > sec) {
+				a = le32_to_cpu(btree->u.internal[i].down);
+				brelse(bh);
+				if (!(anode = hpfs_map_anode(s, a, &bh))) return -1;
+				btree = GET_BTREE_PTR(&anode->btree);
+				goto go_down;
+			}
+		hpfs_error(s, "sector %08x not found in internal anode %08x", sec, a);
+		brelse(bh);
+		return -1;
+	}
+	for (i = 0; i < btree->n_used_nodes; i++)
+		if (le32_to_cpu(btree->u.external[i].file_secno) <= sec &&
+		    le32_to_cpu(btree->u.external[i].file_secno) + le32_to_cpu(btree->u.external[i].length) > sec) {
+			a = le32_to_cpu(btree->u.external[i].disk_secno) + sec - le32_to_cpu(btree->u.external[i].file_secno);
+			if (hpfs_sb(s)->sb_chk) if (hpfs_chk_sectors(s, a, 1, "data")) {
+				brelse(bh);
+				return -1;
+			}
+			if (inode) {
+				struct hpfs_inode_info *hpfs_inode = hpfs_i(inode);
+				hpfs_inode->i_file_sec = le32_to_cpu(btree->u.external[i].file_secno);
+				hpfs_inode->i_disk_sec = le32_to_cpu(btree->u.external[i].disk_secno);
+				hpfs_inode->i_n_secs = le32_to_cpu(btree->u.external[i].length);
+			}
+			brelse(bh);
+			return a;
+		}
+	hpfs_error(s, "sector %08x not found in external anode %08x", sec, a);
+	brelse(bh);
+	return -1;
+}
+
+/* Add a sector to tree */
+
+secno hpfs_add_sector_to_btree(struct super_block *s, secno node, int fnod, unsigned fsecno)
+{
+	struct bplus_header *btree;
+	struct anode *anode = NULL, *ranode = NULL;
+	struct fnode *fnode;
+	anode_secno a, na = -1, ra, up = -1;
+	secno se;
+	struct buffer_head *bh, *bh1, *bh2;
+	int n;
+	unsigned fs;
+	int c1, c2 = 0;
+
+	if (fnod) {
+		if (!(fnode = hpfs_map_fnode(s, node, &bh))) return -1;
+		btree = GET_BTREE_PTR(&fnode->btree);
+	} else {
+		if (!(anode = hpfs_map_anode(s, node, &bh))) return -1;
+		btree = GET_BTREE_PTR(&anode->btree);
+	}
+	a = node;
+	go_down:
+	if ((n = btree->n_used_nodes - 1) < -!!fnod) {
+		hpfs_error(s, "anode %08x has no entries", a);
+		brelse(bh);
+		return -1;
+	}
+	if (bp_internal(btree)) {
+		a = le32_to_cpu(btree->u.internal[n].down);
+		btree->u.internal[n].file_secno = cpu_to_le32(-1);
+		mark_buffer_dirty(bh);
+		brelse(bh);
+		if (hpfs_sb(s)->sb_chk)
+			if (hpfs_stop_cycles(s, a, &c1, &c2, "hpfs_add_sector_to_btree #1")) return -1;
+		if (!(anode = hpfs_map_anode(s, a, &bh))) return -1;
+		btree = GET_BTREE_PTR(&anode->btree);
+		goto go_down;
+	}
+	if (n >= 0) {
+		if (le32_to_cpu(btree->u.external[n].file_secno) + le32_to_cpu(btree->u.external[n].length) != fsecno) {
+			hpfs_error(s, "allocated size %08x, trying to add sector %08x, %cnode %08x",
+				le32_to_cpu(btree->u.external[n].file_secno) + le32_to_cpu(btree->u.external[n].length), fsecno,
+				fnod?'f':'a', node);
+			brelse(bh);
+			return -1;
+		}
+		if (hpfs_alloc_if_possible(s, se = le32_to_cpu(btree->u.external[n].disk_secno) + le32_to_cpu(btree->u.external[n].length))) {
+			le32_add_cpu(&btree->u.external[n].length, 1);
+			mark_buffer_dirty(bh);
+			brelse(bh);
+			return se;
+		}
+	} else {
+		if (fsecno) {
+			hpfs_error(s, "empty file %08x, trying to add sector %08x", node, fsecno);
+			brelse(bh);
+			return -1;
+		}
+		se = !fnod ? node : (node + 16384) & ~16383;
+	}	
+	if (!(se = hpfs_alloc_sector(s, se, 1, fsecno*ALLOC_M>ALLOC_FWD_MAX ? ALLOC_FWD_MAX : fsecno*ALLOC_M<ALLOC_FWD_MIN ? ALLOC_FWD_MIN : fsecno*ALLOC_M))) {
+		brelse(bh);
+		return -1;
+	}
+	fs = n < 0 ? 0 : le32_to_cpu(btree->u.external[n].file_secno) + le32_to_cpu(btree->u.external[n].length);
+	if (!btree->n_free_nodes) {
+		up = a != node ? le32_to_cpu(anode->up) : -1;
+		if (!(anode = hpfs_alloc_anode(s, a, &na, &bh1))) {
+			brelse(bh);
+			hpfs_free_sectors(s, se, 1);
+			return -1;
+		}
+		if (a == node && fnod) {
+			anode->up = cpu_to_le32(node);
+			anode->btree.flags |= BP_fnode_parent;
+			anode->btree.n_used_nodes = btree->n_used_nodes;
+			anode->btree.first_free = btree->first_free;
+			anode->btree.n_free_nodes = 40 - anode->btree.n_used_nodes;
+			memcpy(&anode->u, &btree->u, btree->n_used_nodes * 12);
+			btree->flags |= BP_internal;
+			btree->n_free_nodes = 11;
+			btree->n_used_nodes = 1;
+			btree->first_free = cpu_to_le16((char *)&(btree->u.internal[1]) - (char *)btree);
+			btree->u.internal[0].file_secno = cpu_to_le32(-1);
+			btree->u.internal[0].down = cpu_to_le32(na);
+			mark_buffer_dirty(bh);
+		} else if (!(ranode = hpfs_alloc_anode(s, /*a*/0, &ra, &bh2))) {
+			brelse(bh);
+			brelse(bh1);
+			hpfs_free_sectors(s, se, 1);
+			hpfs_free_sectors(s, na, 1);
+			return -1;
+		}
+		brelse(bh);
+		bh = bh1;
+		btree = GET_BTREE_PTR(&anode->btree);
+	}
+	btree->n_free_nodes--; n = btree->n_used_nodes++;
+	le16_add_cpu(&btree->first_free, 12);
+	btree->u.external[n].disk_secno = cpu_to_le32(se);
+	btree->u.external[n].file_secno = cpu_to_le32(fs);
+	btree->u.external[n].length = cpu_to_le32(1);
+	mark_buffer_dirty(bh);
+	brelse(bh);
+	if ((a == node && fnod) || na == -1) return se;
+	c2 = 0;
+	while (up != (anode_secno)-1) {
+		struct anode *new_anode;
+		if (hpfs_sb(s)->sb_chk)
+			if (hpfs_stop_cycles(s, up, &c1, &c2, "hpfs_add_sector_to_btree #2")) return -1;
+		if (up != node || !fnod) {
+			if (!(anode = hpfs_map_anode(s, up, &bh))) return -1;
+			btree = GET_BTREE_PTR(&anode->btree);
+		} else {
+			if (!(fnode = hpfs_map_fnode(s, up, &bh))) return -1;
+			btree = GET_BTREE_PTR(&fnode->btree);
+		}
+		if (btree->n_free_nodes) {
+			btree->n_free_nodes--; n = btree->n_used_nodes++;
+			le16_add_cpu(&btree->first_free, 8);
+			btree->u.internal[n].file_secno = cpu_to_le32(-1);
+			btree->u.internal[n].down = cpu_to_le32(na);
+			btree->u.internal[n-1].file_secno = cpu_to_le32(fs);
+			mark_buffer_dirty(bh);
+			brelse(bh);
+			brelse(bh2);
+			hpfs_free_sectors(s, ra, 1);
+			if ((anode = hpfs_map_anode(s, na, &bh))) {
+				anode->up = cpu_to_le32(up);
+				if (up == node && fnod)
+					anode->btree.flags |= BP_fnode_parent;
+				else
+					anode->btree.flags &= ~BP_fnode_parent;
+				mark_buffer_dirty(bh);
+				brelse(bh);
+			}
+			return se;
+		}
+		up = up != node ? le32_to_cpu(anode->up) : -1;
+		btree->u.internal[btree->n_used_nodes - 1].file_secno = cpu_to_le32(/*fs*/-1);
+		mark_buffer_dirty(bh);
+		brelse(bh);
+		a = na;
+		if ((new_anode = hpfs_alloc_anode(s, a, &na, &bh))) {
+			anode = new_anode;
+			/*anode->up = cpu_to_le32(up != -1 ? up : ra);*/
+			anode->btree.flags |= BP_internal;
+			anode->btree.n_used_nodes = 1;
+			anode->btree.n_free_nodes = 59;
+			anode->btree.first_free = cpu_to_le16(16);
+			GET_BTREE_PTR(&anode->btree)->u.internal[0].down = cpu_to_le32(a);
+			GET_BTREE_PTR(&anode->btree)->u.internal[0].file_secno = cpu_to_le32(-1);
+			mark_buffer_dirty(bh);
+			brelse(bh);
+			if ((anode = hpfs_map_anode(s, a, &bh))) {
+				anode->up = cpu_to_le32(na);
+				mark_buffer_dirty(bh);
+				brelse(bh);
+			}
+		} else na = a;
+	}
+	if ((anode = hpfs_map_anode(s, na, &bh))) {
+		anode->up = cpu_to_le32(node);
+		if (fnod)
+			anode->btree.flags |= BP_fnode_parent;
+		mark_buffer_dirty(bh);
+		brelse(bh);
+	}
+	if (!fnod) {
+		if (!(anode = hpfs_map_anode(s, node, &bh))) {
+			brelse(bh2);
+			return -1;
+		}
+		btree = GET_BTREE_PTR(&anode->btree);
+	} else {
+		if (!(fnode = hpfs_map_fnode(s, node, &bh))) {
+			brelse(bh2);
+			return -1;
+		}
+		btree = GET_BTREE_PTR(&fnode->btree);
+	}
+	ranode->up = cpu_to_le32(node);
+	memcpy(&ranode->btree, btree, le16_to_cpu(btree->first_free));
+	if (fnod)
+		ranode->btree.flags |= BP_fnode_parent;
+	GET_BTREE_PTR(&ranode->btree)->n_free_nodes = (bp_internal(GET_BTREE_PTR(&ranode->btree)) ? 60 : 40) - GET_BTREE_PTR(&ranode->btree)->n_used_nodes;
+	if (bp_internal(GET_BTREE_PTR(&ranode->btree))) for (n = 0; n < GET_BTREE_PTR(&ranode->btree)->n_used_nodes; n++) {
+		struct anode *unode;
+		if ((unode = hpfs_map_anode(s, le32_to_cpu(ranode->u.internal[n].down), &bh1))) {
+			unode->up = cpu_to_le32(ra);
+			unode->btree.flags &= ~BP_fnode_parent;
+			mark_buffer_dirty(bh1);
+			brelse(bh1);
+		}
+	}
+	btree->flags |= BP_internal;
+	btree->n_free_nodes = fnod ? 10 : 58;
+	btree->n_used_nodes = 2;
+	btree->first_free = cpu_to_le16((char *)&btree->u.internal[2] - (char *)btree);
+	btree->u.internal[0].file_secno = cpu_to_le32(fs);
+	btree->u.internal[0].down = cpu_to_le32(ra);
+	btree->u.internal[1].file_secno = cpu_to_le32(-1);
+	btree->u.internal[1].down = cpu_to_le32(na);
+	mark_buffer_dirty(bh);
+	brelse(bh);
+	mark_buffer_dirty(bh2);
+	brelse(bh2);
+	return se;
+}
+
+/*
+ * Remove allocation tree. Recursion would look much nicer but
+ * I want to avoid it because it can cause stack overflow.
+ */
+
+void hpfs_remove_btree(struct super_block *s, struct bplus_header *btree)
+{
+	struct bplus_header *btree1 = btree;
+	struct anode *anode = NULL;
+	anode_secno ano = 0, oano;
+	struct buffer_head *bh;
+	int level = 0;
+	int pos = 0;
+	int i;
+	int c1, c2 = 0;
+	int d1, d2;
+	go_down:
+	d2 = 0;
+	while (bp_internal(btree1)) {
+		ano = le32_to_cpu(btree1->u.internal[pos].down);
+		if (level) brelse(bh);
+		if (hpfs_sb(s)->sb_chk)
+			if (hpfs_stop_cycles(s, ano, &d1, &d2, "hpfs_remove_btree #1"))
+				return;
+		if (!(anode = hpfs_map_anode(s, ano, &bh))) return;
+		btree1 = GET_BTREE_PTR(&anode->btree);
+		level++;
+		pos = 0;
+	}
+	for (i = 0; i < btree1->n_used_nodes; i++)
+		hpfs_free_sectors(s, le32_to_cpu(btree1->u.external[i].disk_secno), le32_to_cpu(btree1->u.external[i].length));
+	go_up:
+	if (!level) return;
+	brelse(bh);
+	if (hpfs_sb(s)->sb_chk)
+		if (hpfs_stop_cycles(s, ano, &c1, &c2, "hpfs_remove_btree #2")) return;
+	hpfs_free_sectors(s, ano, 1);
+	oano = ano;
+	ano = le32_to_cpu(anode->up);
+	if (--level) {
+		if (!(anode = hpfs_map_anode(s, ano, &bh))) return;
+		btree1 = GET_BTREE_PTR(&anode->btree);
+	} else btree1 = btree;
+	for (i = 0; i < btree1->n_used_nodes; i++) {
+		if (le32_to_cpu(btree1->u.internal[i].down) == oano) {
+			if ((pos = i + 1) < btree1->n_used_nodes)
+				goto go_down;
+			else
+				goto go_up;
+		}
+	}
+	hpfs_error(s,
+		   "reference to anode %08x not found in anode %08x "
+		   "(probably bad up pointer)",
+		   oano, level ? ano : -1);
+	if (level)
+		brelse(bh);
+}
+
+/* Just a wrapper around hpfs_bplus_lookup .. used for reading eas */
+
+static secno anode_lookup(struct super_block *s, anode_secno a, unsigned sec)
+{
+	struct anode *anode;
+	struct buffer_head *bh;
+	if (!(anode = hpfs_map_anode(s, a, &bh))) return -1;
+	return hpfs_bplus_lookup(s, NULL, GET_BTREE_PTR(&anode->btree), sec, bh);
+}
+
+int hpfs_ea_read(struct super_block *s, secno a, int ano, unsigned pos,
+	    unsigned len, char *buf)
+{
+	struct buffer_head *bh;
+	char *data;
+	secno sec;
+	unsigned l;
+	while (len) {
+		if (ano) {
+			if ((sec = anode_lookup(s, a, pos >> 9)) == -1)
+				return -1;
+		} else sec = a + (pos >> 9);
+		if (hpfs_sb(s)->sb_chk) if (hpfs_chk_sectors(s, sec, 1, "ea #1")) return -1;
+		if (!(data = hpfs_map_sector(s, sec, &bh, (len - 1) >> 9)))
+			return -1;
+		l = 0x200 - (pos & 0x1ff); if (l > len) l = len;
+		memcpy(buf, data + (pos & 0x1ff), l);
+		brelse(bh);
+		buf += l; pos += l; len -= l;
+	}
+	return 0;
+}
+
+int hpfs_ea_write(struct super_block *s, secno a, int ano, unsigned pos,
+	     unsigned len, const char *buf)
+{
+	struct buffer_head *bh;
+	char *data;
+	secno sec;
+	unsigned l;
+	while (len) {
+		if (ano) {
+			if ((sec = anode_lookup(s, a, pos >> 9)) == -1)
+				return -1;
+		} else sec = a + (pos >> 9);
+		if (hpfs_sb(s)->sb_chk) if (hpfs_chk_sectors(s, sec, 1, "ea #2")) return -1;
+		if (!(data = hpfs_map_sector(s, sec, &bh, (len - 1) >> 9)))
+			return -1;
+		l = 0x200 - (pos & 0x1ff); if (l > len) l = len;
+		memcpy(data + (pos & 0x1ff), buf, l);
+		mark_buffer_dirty(bh);
+		brelse(bh);
+		buf += l; pos += l; len -= l;
+	}
+	return 0;
+}
+
+void hpfs_ea_remove(struct super_block *s, secno a, int ano, unsigned len)
+{
+	struct anode *anode;
+	struct buffer_head *bh;
+	if (ano) {
+		if (!(anode = hpfs_map_anode(s, a, &bh))) return;
+		hpfs_remove_btree(s, GET_BTREE_PTR(&anode->btree));
+		brelse(bh);
+		hpfs_free_sectors(s, a, 1);
+	} else hpfs_free_sectors(s, a, (len + 511) >> 9);
+}
+
+/* Truncate allocation tree. Doesn't join anodes - I hope it doesn't matter */
+
+void hpfs_truncate_btree(struct super_block *s, secno f, int fno, unsigned secs)
+{
+	struct fnode *fnode;
+	struct anode *anode;
+	struct buffer_head *bh;
+	struct bplus_header *btree;
+	anode_secno node = f;
+	int i, j, nodes;
+	int c1, c2 = 0;
+	if (fno) {
+		if (!(fnode = hpfs_map_fnode(s, f, &bh))) return;
+		btree = GET_BTREE_PTR(&fnode->btree);
+	} else {
+		if (!(anode = hpfs_map_anode(s, f, &bh))) return;
+		btree = GET_BTREE_PTR(&anode->btree);
+	}
+	if (!secs) {
+		hpfs_remove_btree(s, btree);
+		if (fno) {
+			btree->n_free_nodes = 8;
+			btree->n_used_nodes = 0;
+			btree->first_free = cpu_to_le16(8);
+			btree->flags &= ~BP_internal;
+			mark_buffer_dirty(bh);
+		} else hpfs_free_sectors(s, f, 1);
+		brelse(bh);
+		return;
+	}
+	while (bp_internal(btree)) {
+		nodes = btree->n_used_nodes + btree->n_free_nodes;
+		for (i = 0; i < btree->n_used_nodes; i++)
+			if (le32_to_cpu(btree->u.internal[i].file_secno) >= secs) goto f;
+		brelse(bh);
+		hpfs_error(s, "internal btree %08x doesn't end with -1", node);
+		return;
+		f:
+		for (j = i + 1; j < btree->n_used_nodes; j++)
+			hpfs_ea_remove(s, le32_to_cpu(btree->u.internal[j].down), 1, 0);
+		btree->n_used_nodes = i + 1;
+		btree->n_free_nodes = nodes - btree->n_used_nodes;
+		btree->first_free = cpu_to_le16(8 + 8 * btree->n_used_nodes);
+		mark_buffer_dirty(bh);
+		if (btree->u.internal[i].file_secno == cpu_to_le32(secs)) {
+			brelse(bh);
+			return;
+		}
+		node = le32_to_cpu(btree->u.internal[i].down);
+		brelse(bh);
+		if (hpfs_sb(s)->sb_chk)
+			if (hpfs_stop_cycles(s, node, &c1, &c2, "hpfs_truncate_btree"))
+				return;
+		if (!(anode = hpfs_map_anode(s, node, &bh))) return;
+		btree = GET_BTREE_PTR(&anode->btree);
+	}	
+	nodes = btree->n_used_nodes + btree->n_free_nodes;
+	for (i = 0; i < btree->n_used_nodes; i++)
+		if (le32_to_cpu(btree->u.external[i].file_secno) + le32_to_cpu(btree->u.external[i].length) >= secs) goto ff;
+	brelse(bh);
+	return;
+	ff:
+	if (secs <= le32_to_cpu(btree->u.external[i].file_secno)) {
+		hpfs_error(s, "there is an allocation error in file %08x, sector %08x", f, secs);
+		if (i) i--;
+	}
+	else if (le32_to_cpu(btree->u.external[i].file_secno) + le32_to_cpu(btree->u.external[i].length) > secs) {
+		hpfs_free_sectors(s, le32_to_cpu(btree->u.external[i].disk_secno) + secs -
+			le32_to_cpu(btree->u.external[i].file_secno), le32_to_cpu(btree->u.external[i].length)
+			- secs + le32_to_cpu(btree->u.external[i].file_secno)); /* I hope gcc optimizes this :-) */
+		btree->u.external[i].length = cpu_to_le32(secs - le32_to_cpu(btree->u.external[i].file_secno));
+	}
+	for (j = i + 1; j < btree->n_used_nodes; j++)
+		hpfs_free_sectors(s, le32_to_cpu(btree->u.external[j].disk_secno), le32_to_cpu(btree->u.external[j].length));
+	btree->n_used_nodes = i + 1;
+	btree->n_free_nodes = nodes - btree->n_used_nodes;
+	btree->first_free = cpu_to_le16(8 + 12 * btree->n_used_nodes);
+	mark_buffer_dirty(bh);
+	brelse(bh);
+}
+
+/* Remove file or directory and it's eas - note that directory must
+   be empty when this is called. */
+
+void hpfs_remove_fnode(struct super_block *s, fnode_secno fno)
+{
+	struct buffer_head *bh;
+	struct fnode *fnode;
+	struct extended_attribute *ea;
+	struct extended_attribute *ea_end;
+	if (!(fnode = hpfs_map_fnode(s, fno, &bh))) return;
+	if (!fnode_is_dir(fnode)) hpfs_remove_btree(s, GET_BTREE_PTR(&fnode->btree));
+	else hpfs_remove_dtree(s, le32_to_cpu(fnode->u.external[0].disk_secno));
+	ea_end = fnode_end_ea(fnode);
+	for (ea = fnode_ea(fnode); ea < ea_end; ea = next_ea(ea))
+		if (ea_indirect(ea))
+			hpfs_ea_remove(s, ea_sec(ea), ea_in_anode(ea), ea_len(ea));
+	hpfs_ea_ext_remove(s, le32_to_cpu(fnode->ea_secno), fnode_in_anode(fnode), le32_to_cpu(fnode->ea_size_l));
+	brelse(bh);
+	hpfs_free_sectors(s, fno, 1);
+}