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, 738 insertions, 0 deletions

diff --git a/fs/cachefiles/io.c b/fs/cachefiles/io.c
new file mode 100644
index 000000000..9540ec25b
--- /dev/null
+++ b/fs/cachefiles/io.c
@@ -0,0 +1,738 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* kiocb-using read/write
+ *
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/mount.h>
+#include <linux/slab.h>
+#include <linux/file.h>
+#include <linux/uio.h>
+#include <linux/bio.h>
+#include <linux/falloc.h>
+#include <linux/sched/mm.h>
+#include <trace/events/fscache.h>
+#include <trace/events/netfs.h>
+#include "internal.h"
+
+struct cachefiles_kiocb {
+	struct kiocb		iocb;
+	refcount_t		ki_refcnt;
+	loff_t			start;
+	union {
+		size_t		skipped;
+		size_t		len;
+	};
+	struct cachefiles_object *object;
+	netfs_io_terminated_t	term_func;
+	void			*term_func_priv;
+	bool			was_async;
+	unsigned int		inval_counter;	/* Copy of cookie->inval_counter */
+	u64			b_writing;
+};
+
+static inline void cachefiles_put_kiocb(struct cachefiles_kiocb *ki)
+{
+	if (refcount_dec_and_test(&ki->ki_refcnt)) {
+		cachefiles_put_object(ki->object, cachefiles_obj_put_ioreq);
+		fput(ki->iocb.ki_filp);
+		kfree(ki);
+	}
+}
+
+/*
+ * Handle completion of a read from the cache.
+ */
+static void cachefiles_read_complete(struct kiocb *iocb, long ret)
+{
+	struct cachefiles_kiocb *ki = container_of(iocb, struct cachefiles_kiocb, iocb);
+	struct inode *inode = file_inode(ki->iocb.ki_filp);
+
+	_enter("%ld", ret);
+
+	if (ret < 0)
+		trace_cachefiles_io_error(ki->object, inode, ret,
+					  cachefiles_trace_read_error);
+
+	if (ki->term_func) {
+		if (ret >= 0) {
+			if (ki->object->cookie->inval_counter == ki->inval_counter)
+				ki->skipped += ret;
+			else
+				ret = -ESTALE;
+		}
+
+		ki->term_func(ki->term_func_priv, ret);
+	}
+
+	cachefiles_put_kiocb(ki);
+}
+
+/*
+ * Initiate a read from the cache.
+ */
+static int cachefiles_read(struct netfs_cache_resources *cres,
+			   loff_t start_pos,
+			   struct iov_iter *iter,
+			   enum netfs_read_from_hole read_hole,
+			   netfs_io_terminated_t term_func,
+			   void *term_func_priv)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_kiocb *ki;
+	struct file *file;
+	unsigned int old_nofs;
+	ssize_t ret = -ENOBUFS;
+	size_t len = iov_iter_count(iter), skipped = 0;
+
+	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_READ))
+		goto presubmission_error;
+
+	fscache_count_read();
+	object = cachefiles_cres_object(cres);
+	file = cachefiles_cres_file(cres);
+
+	_enter("%pD,%llu,%llx,%zx/%llx",
+	       file, file_inode(file)->i_ino, start_pos, len,
+	       i_size_read(file_inode(file)));
+
+	/* If the caller asked us to seek for data before doing the read, then
+	 * we should do that now.  If we find a gap, we fill it with zeros.
+	 */
+	if (read_hole != NETFS_READ_HOLE_IGNORE) {
+		loff_t off = start_pos, off2;
+
+		off2 = cachefiles_inject_read_error();
+		if (off2 == 0)
+			off2 = vfs_llseek(file, off, SEEK_DATA);
+		if (off2 < 0 && off2 >= (loff_t)-MAX_ERRNO && off2 != -ENXIO) {
+			skipped = 0;
+			ret = off2;
+			goto presubmission_error;
+		}
+
+		if (off2 == -ENXIO || off2 >= start_pos + len) {
+			/* The region is beyond the EOF or there's no more data
+			 * in the region, so clear the rest of the buffer and
+			 * return success.
+			 */
+			ret = -ENODATA;
+			if (read_hole == NETFS_READ_HOLE_FAIL)
+				goto presubmission_error;
+
+			iov_iter_zero(len, iter);
+			skipped = len;
+			ret = 0;
+			goto presubmission_error;
+		}
+
+		skipped = off2 - off;
+		iov_iter_zero(skipped, iter);
+	}
+
+	ret = -ENOMEM;
+	ki = kzalloc_obj(struct cachefiles_kiocb);
+	if (!ki)
+		goto presubmission_error;
+
+	refcount_set(&ki->ki_refcnt, 2);
+	ki->iocb.ki_filp	= file;
+	ki->iocb.ki_pos		= start_pos + skipped;
+	ki->iocb.ki_flags	= IOCB_DIRECT;
+	ki->iocb.ki_ioprio	= get_current_ioprio();
+	ki->skipped		= skipped;
+	ki->object		= object;
+	ki->inval_counter	= cres->inval_counter;
+	ki->term_func		= term_func;
+	ki->term_func_priv	= term_func_priv;
+	ki->was_async		= true;
+
+	if (ki->term_func)
+		ki->iocb.ki_complete = cachefiles_read_complete;
+
+	get_file(ki->iocb.ki_filp);
+	cachefiles_grab_object(object, cachefiles_obj_get_ioreq);
+
+	trace_cachefiles_read(object, file_inode(file), ki->iocb.ki_pos, len - skipped);
+	old_nofs = memalloc_nofs_save();
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		ret = vfs_iocb_iter_read(file, &ki->iocb, iter);
+	memalloc_nofs_restore(old_nofs);
+	switch (ret) {
+	case -EIOCBQUEUED:
+		goto in_progress;
+
+	case -ERESTARTSYS:
+	case -ERESTARTNOINTR:
+	case -ERESTARTNOHAND:
+	case -ERESTART_RESTARTBLOCK:
+		/* There's no easy way to restart the syscall since other AIO's
+		 * may be already running. Just fail this IO with EINTR.
+		 */
+		ret = -EINTR;
+		fallthrough;
+	default:
+		ki->was_async = false;
+		cachefiles_read_complete(&ki->iocb, ret);
+		if (ret > 0)
+			ret = 0;
+		break;
+	}
+
+in_progress:
+	cachefiles_put_kiocb(ki);
+	_leave(" = %zd", ret);
+	return ret;
+
+presubmission_error:
+	if (term_func)
+		term_func(term_func_priv, ret < 0 ? ret : skipped);
+	return ret;
+}
+
+/*
+ * Query the occupancy of the cache in a region, returning where the next chunk
+ * of data starts and how long it is.
+ */
+static int cachefiles_query_occupancy(struct netfs_cache_resources *cres,
+				      loff_t start, size_t len, size_t granularity,
+				      loff_t *_data_start, size_t *_data_len)
+{
+	struct cachefiles_object *object;
+	struct file *file;
+	loff_t off, off2;
+
+	*_data_start = -1;
+	*_data_len = 0;
+
+	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_READ))
+		return -ENOBUFS;
+
+	object = cachefiles_cres_object(cres);
+	file = cachefiles_cres_file(cres);
+	granularity = max_t(size_t, object->volume->cache->bsize, granularity);
+
+	_enter("%pD,%llu,%llx,%zx/%llx",
+	       file, file_inode(file)->i_ino, start, len,
+	       i_size_read(file_inode(file)));
+
+	off = cachefiles_inject_read_error();
+	if (off == 0)
+		off = vfs_llseek(file, start, SEEK_DATA);
+	if (off == -ENXIO)
+		return -ENODATA; /* Beyond EOF */
+	if (off < 0 && off >= (loff_t)-MAX_ERRNO)
+		return -ENOBUFS; /* Error. */
+	if (round_up(off, granularity) >= start + len)
+		return -ENODATA; /* No data in range */
+
+	off2 = cachefiles_inject_read_error();
+	if (off2 == 0)
+		off2 = vfs_llseek(file, off, SEEK_HOLE);
+	if (off2 == -ENXIO)
+		return -ENODATA; /* Beyond EOF */
+	if (off2 < 0 && off2 >= (loff_t)-MAX_ERRNO)
+		return -ENOBUFS; /* Error. */
+
+	/* Round away partial blocks */
+	off = round_up(off, granularity);
+	off2 = round_down(off2, granularity);
+	if (off2 <= off)
+		return -ENODATA;
+
+	*_data_start = off;
+	if (off2 > start + len)
+		*_data_len = len;
+	else
+		*_data_len = off2 - off;
+	return 0;
+}
+
+/*
+ * Handle completion of a write to the cache.
+ */
+static void cachefiles_write_complete(struct kiocb *iocb, long ret)
+{
+	struct cachefiles_kiocb *ki = container_of(iocb, struct cachefiles_kiocb, iocb);
+	struct cachefiles_object *object = ki->object;
+	struct inode *inode = file_inode(ki->iocb.ki_filp);
+
+	_enter("%ld", ret);
+
+	if (ki->was_async)
+		kiocb_end_write(iocb);
+
+	if (ret < 0)
+		trace_cachefiles_io_error(object, inode, ret,
+					  cachefiles_trace_write_error);
+
+	atomic_long_sub(ki->b_writing, &object->volume->cache->b_writing);
+	set_bit(FSCACHE_COOKIE_HAVE_DATA, &object->cookie->flags);
+	if (ki->term_func)
+		ki->term_func(ki->term_func_priv, ret);
+	cachefiles_put_kiocb(ki);
+}
+
+/*
+ * Initiate a write to the cache.
+ */
+int __cachefiles_write(struct cachefiles_object *object,
+		       struct file *file,
+		       loff_t start_pos,
+		       struct iov_iter *iter,
+		       netfs_io_terminated_t term_func,
+		       void *term_func_priv)
+{
+	struct cachefiles_cache *cache;
+	struct cachefiles_kiocb *ki;
+	unsigned int old_nofs;
+	ssize_t ret;
+	size_t len = iov_iter_count(iter);
+
+	fscache_count_write();
+	cache = object->volume->cache;
+
+	_enter("%pD,%llu,%llx,%zx/%llx",
+	       file, file_inode(file)->i_ino, start_pos, len,
+	       i_size_read(file_inode(file)));
+
+	ki = kzalloc_obj(struct cachefiles_kiocb);
+	if (!ki) {
+		if (term_func)
+			term_func(term_func_priv, -ENOMEM);
+		return -ENOMEM;
+	}
+
+	refcount_set(&ki->ki_refcnt, 2);
+	ki->iocb.ki_filp	= file;
+	ki->iocb.ki_pos		= start_pos;
+	ki->iocb.ki_flags	= IOCB_DIRECT | IOCB_WRITE;
+	ki->iocb.ki_ioprio	= get_current_ioprio();
+	ki->object		= object;
+	ki->start		= start_pos;
+	ki->len			= len;
+	ki->term_func		= term_func;
+	ki->term_func_priv	= term_func_priv;
+	ki->was_async		= true;
+	ki->b_writing		= (len + (1 << cache->bshift) - 1) >> cache->bshift;
+
+	if (ki->term_func)
+		ki->iocb.ki_complete = cachefiles_write_complete;
+	atomic_long_add(ki->b_writing, &cache->b_writing);
+
+	get_file(ki->iocb.ki_filp);
+	cachefiles_grab_object(object, cachefiles_obj_get_ioreq);
+
+	trace_cachefiles_write(object, file_inode(file), ki->iocb.ki_pos, len);
+	old_nofs = memalloc_nofs_save();
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		ret = vfs_iocb_iter_write(file, &ki->iocb, iter);
+	memalloc_nofs_restore(old_nofs);
+	switch (ret) {
+	case -EIOCBQUEUED:
+		goto in_progress;
+
+	case -ERESTARTSYS:
+	case -ERESTARTNOINTR:
+	case -ERESTARTNOHAND:
+	case -ERESTART_RESTARTBLOCK:
+		/* There's no easy way to restart the syscall since other AIO's
+		 * may be already running. Just fail this IO with EINTR.
+		 */
+		ret = -EINTR;
+		fallthrough;
+	default:
+		ki->was_async = false;
+		cachefiles_write_complete(&ki->iocb, ret);
+		break;
+	}
+
+in_progress:
+	cachefiles_put_kiocb(ki);
+	_leave(" = %zd", ret);
+	return ret;
+}
+
+static int cachefiles_write(struct netfs_cache_resources *cres,
+			    loff_t start_pos,
+			    struct iov_iter *iter,
+			    netfs_io_terminated_t term_func,
+			    void *term_func_priv)
+{
+	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE)) {
+		if (term_func)
+			term_func(term_func_priv, -ENOBUFS);
+		trace_netfs_sreq(term_func_priv, netfs_sreq_trace_cache_nowrite);
+		return -ENOBUFS;
+	}
+
+	return __cachefiles_write(cachefiles_cres_object(cres),
+				  cachefiles_cres_file(cres),
+				  start_pos, iter,
+				  term_func, term_func_priv);
+}
+
+static inline enum netfs_io_source
+cachefiles_do_prepare_read(struct netfs_cache_resources *cres,
+			   loff_t start, size_t *_len, loff_t i_size,
+			   unsigned long *_flags, ino_t netfs_ino)
+{
+	enum cachefiles_prepare_read_trace why;
+	struct cachefiles_object *object = NULL;
+	struct cachefiles_cache *cache;
+	struct fscache_cookie *cookie = fscache_cres_cookie(cres);
+	const struct cred *saved_cred;
+	struct file *file = cachefiles_cres_file(cres);
+	enum netfs_io_source ret = NETFS_DOWNLOAD_FROM_SERVER;
+	size_t len = *_len;
+	loff_t off, to;
+	ino_t ino = file ? file_inode(file)->i_ino : 0;
+
+	_enter("%zx @%llx/%llx", len, start, i_size);
+
+	if (start >= i_size) {
+		ret = NETFS_FILL_WITH_ZEROES;
+		why = cachefiles_trace_read_after_eof;
+		goto out_no_object;
+	}
+
+	if (test_bit(FSCACHE_COOKIE_NO_DATA_TO_READ, &cookie->flags)) {
+		__set_bit(NETFS_SREQ_COPY_TO_CACHE, _flags);
+		why = cachefiles_trace_read_no_data;
+		goto out_no_object;
+	}
+
+	/* The object and the file may be being created in the background. */
+	if (!file) {
+		why = cachefiles_trace_read_no_file;
+		if (!fscache_wait_for_operation(cres, FSCACHE_WANT_READ))
+			goto out_no_object;
+		file = cachefiles_cres_file(cres);
+		if (!file)
+			goto out_no_object;
+		ino = file_inode(file)->i_ino;
+	}
+
+	object = cachefiles_cres_object(cres);
+	cache = object->volume->cache;
+	cachefiles_begin_secure(cache, &saved_cred);
+	off = cachefiles_inject_read_error();
+	if (off == 0)
+		off = vfs_llseek(file, start, SEEK_DATA);
+	if (off < 0 && off >= (loff_t)-MAX_ERRNO) {
+		if (off == (loff_t)-ENXIO) {
+			why = cachefiles_trace_read_seek_nxio;
+			goto download_and_store;
+		}
+		trace_cachefiles_io_error(object, file_inode(file), off,
+					  cachefiles_trace_seek_error);
+		why = cachefiles_trace_read_seek_error;
+		goto out;
+	}
+
+	if (off >= start + len) {
+		why = cachefiles_trace_read_found_hole;
+		goto download_and_store;
+	}
+
+	if (off > start) {
+		off = round_up(off, cache->bsize);
+		len = off - start;
+		*_len = len;
+		why = cachefiles_trace_read_found_part;
+		goto download_and_store;
+	}
+
+	to = cachefiles_inject_read_error();
+	if (to == 0)
+		to = vfs_llseek(file, start, SEEK_HOLE);
+	if (to < 0 && to >= (loff_t)-MAX_ERRNO) {
+		trace_cachefiles_io_error(object, file_inode(file), to,
+					  cachefiles_trace_seek_error);
+		why = cachefiles_trace_read_seek_error;
+		goto out;
+	}
+
+	if (to < start + len) {
+		if (start + len >= i_size)
+			to = round_up(to, cache->bsize);
+		else
+			to = round_down(to, cache->bsize);
+		len = to - start;
+		*_len = len;
+	}
+
+	why = cachefiles_trace_read_have_data;
+	ret = NETFS_READ_FROM_CACHE;
+	goto out;
+
+download_and_store:
+	__set_bit(NETFS_SREQ_COPY_TO_CACHE, _flags);
+out:
+	cachefiles_end_secure(cache, saved_cred);
+out_no_object:
+	trace_cachefiles_prep_read(object, start, len, *_flags, ret, why, ino, netfs_ino);
+	return ret;
+}
+
+/*
+ * Prepare a read operation, shortening it to a cached/uncached
+ * boundary as appropriate.
+ */
+static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *subreq,
+						    unsigned long long i_size)
+{
+	return cachefiles_do_prepare_read(&subreq->rreq->cache_resources,
+					  subreq->start, &subreq->len, i_size,
+					  &subreq->flags, subreq->rreq->inode->i_ino);
+}
+
+/*
+ * Prepare for a write to occur.
+ */
+int __cachefiles_prepare_write(struct cachefiles_object *object,
+			       struct file *file,
+			       loff_t *_start, size_t *_len, size_t upper_len,
+			       bool no_space_allocated_yet)
+{
+	struct cachefiles_cache *cache = object->volume->cache;
+	loff_t start = *_start, pos;
+	size_t len = *_len;
+	int ret;
+
+	/* Round to DIO size */
+	start = round_down(*_start, PAGE_SIZE);
+	if (start != *_start || *_len > upper_len) {
+		/* Probably asked to cache a streaming write written into the
+		 * pagecache when the cookie was temporarily out of service to
+		 * culling.
+		 */
+		fscache_count_dio_misfit();
+		return -ENOBUFS;
+	}
+
+	*_len = round_up(len, PAGE_SIZE);
+
+	/* We need to work out whether there's sufficient disk space to perform
+	 * the write - but we can skip that check if we have space already
+	 * allocated.
+	 */
+	if (no_space_allocated_yet)
+		goto check_space;
+
+	pos = cachefiles_inject_read_error();
+	if (pos == 0)
+		pos = vfs_llseek(file, start, SEEK_DATA);
+	if (pos < 0 && pos >= (loff_t)-MAX_ERRNO) {
+		if (pos == -ENXIO)
+			goto check_space; /* Unallocated tail */
+		trace_cachefiles_io_error(object, file_inode(file), pos,
+					  cachefiles_trace_seek_error);
+		return pos;
+	}
+	if ((u64)pos >= (u64)start + *_len)
+		goto check_space; /* Unallocated region */
+
+	/* We have a block that's at least partially filled - if we're low on
+	 * space, we need to see if it's fully allocated.  If it's not, we may
+	 * want to cull it.
+	 */
+	if (cachefiles_has_space(cache, 0, *_len / PAGE_SIZE,
+				 cachefiles_has_space_check) == 0)
+		return 0; /* Enough space to simply overwrite the whole block */
+
+	pos = cachefiles_inject_read_error();
+	if (pos == 0)
+		pos = vfs_llseek(file, start, SEEK_HOLE);
+	if (pos < 0 && pos >= (loff_t)-MAX_ERRNO) {
+		trace_cachefiles_io_error(object, file_inode(file), pos,
+					  cachefiles_trace_seek_error);
+		return pos;
+	}
+	if ((u64)pos >= (u64)start + *_len)
+		return 0; /* Fully allocated */
+
+	/* Partially allocated, but insufficient space: cull. */
+	fscache_count_no_write_space();
+	ret = cachefiles_inject_remove_error();
+	if (ret == 0)
+		ret = vfs_fallocate(file, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
+				    start, *_len);
+	if (ret < 0) {
+		trace_cachefiles_io_error(object, file_inode(file), ret,
+					  cachefiles_trace_fallocate_error);
+		cachefiles_io_error_obj(object,
+					"CacheFiles: fallocate failed (%d)\n", ret);
+		ret = -EIO;
+	}
+
+	return ret;
+
+check_space:
+	return cachefiles_has_space(cache, 0, *_len / PAGE_SIZE,
+				    cachefiles_has_space_for_write);
+}
+
+static int cachefiles_prepare_write(struct netfs_cache_resources *cres,
+				    loff_t *_start, size_t *_len, size_t upper_len,
+				    loff_t i_size, bool no_space_allocated_yet)
+{
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+	struct cachefiles_cache *cache = object->volume->cache;
+	const struct cred *saved_cred;
+	int ret;
+
+	if (!cachefiles_cres_file(cres)) {
+		if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE))
+			return -ENOBUFS;
+		if (!cachefiles_cres_file(cres))
+			return -ENOBUFS;
+	}
+
+	cachefiles_begin_secure(cache, &saved_cred);
+	ret = __cachefiles_prepare_write(object, cachefiles_cres_file(cres),
+					 _start, _len, upper_len,
+					 no_space_allocated_yet);
+	cachefiles_end_secure(cache, saved_cred);
+	return ret;
+}
+
+static void cachefiles_prepare_write_subreq(struct netfs_io_subrequest *subreq)
+{
+	struct netfs_io_request *wreq = subreq->rreq;
+	struct netfs_cache_resources *cres = &wreq->cache_resources;
+	struct netfs_io_stream *stream = &wreq->io_streams[subreq->stream_nr];
+
+	_enter("W=%x[%x] %llx", wreq->debug_id, subreq->debug_index, subreq->start);
+
+	stream->sreq_max_len = MAX_RW_COUNT;
+	stream->sreq_max_segs = BIO_MAX_VECS;
+
+	if (!cachefiles_cres_file(cres)) {
+		if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE))
+			return netfs_prepare_write_failed(subreq);
+		if (!cachefiles_cres_file(cres))
+			return netfs_prepare_write_failed(subreq);
+	}
+}
+
+static void cachefiles_issue_write(struct netfs_io_subrequest *subreq)
+{
+	struct netfs_io_request *wreq = subreq->rreq;
+	struct netfs_cache_resources *cres = &wreq->cache_resources;
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct netfs_io_stream *stream = &wreq->io_streams[subreq->stream_nr];
+	const struct cred *saved_cred;
+	size_t off, pre, post, len = subreq->len;
+	loff_t start = subreq->start;
+	int ret;
+
+	_enter("W=%x[%x] %llx-%llx",
+	       wreq->debug_id, subreq->debug_index, start, start + len - 1);
+
+	/* We need to start on the cache granularity boundary */
+	off = start & (CACHEFILES_DIO_BLOCK_SIZE - 1);
+	if (off) {
+		pre = CACHEFILES_DIO_BLOCK_SIZE - off;
+		if (pre >= len) {
+			fscache_count_dio_misfit();
+			netfs_write_subrequest_terminated(subreq, len);
+			return;
+		}
+		subreq->transferred += pre;
+		start += pre;
+		len -= pre;
+		iov_iter_advance(&subreq->io_iter, pre);
+	}
+
+	/* We also need to end on the cache granularity boundary */
+	if (start + len == wreq->i_size) {
+		size_t part = len % CACHEFILES_DIO_BLOCK_SIZE;
+		size_t need = CACHEFILES_DIO_BLOCK_SIZE - part;
+
+		if (part && stream->submit_extendable_to >= need) {
+			len += need;
+			subreq->len += need;
+			subreq->io_iter.count += need;
+		}
+	}
+
+	post = len & (CACHEFILES_DIO_BLOCK_SIZE - 1);
+	if (post) {
+		len -= post;
+		if (len == 0) {
+			fscache_count_dio_misfit();
+			netfs_write_subrequest_terminated(subreq, post);
+			return;
+		}
+		iov_iter_truncate(&subreq->io_iter, len);
+	}
+
+	trace_netfs_sreq(subreq, netfs_sreq_trace_cache_prepare);
+	cachefiles_begin_secure(cache, &saved_cred);
+	ret = __cachefiles_prepare_write(object, cachefiles_cres_file(cres),
+					 &start, &len, len, true);
+	cachefiles_end_secure(cache, saved_cred);
+	if (ret < 0) {
+		netfs_write_subrequest_terminated(subreq, ret);
+		return;
+	}
+
+	trace_netfs_sreq(subreq, netfs_sreq_trace_cache_write);
+	cachefiles_write(&subreq->rreq->cache_resources,
+			 subreq->start, &subreq->io_iter,
+			 netfs_write_subrequest_terminated, subreq);
+}
+
+/*
+ * Clean up an operation.
+ */
+static void cachefiles_end_operation(struct netfs_cache_resources *cres)
+{
+	struct file *file = cachefiles_cres_file(cres);
+
+	if (file)
+		fput(file);
+	fscache_end_cookie_access(fscache_cres_cookie(cres), fscache_access_io_end);
+}
+
+static const struct netfs_cache_ops cachefiles_netfs_cache_ops = {
+	.end_operation		= cachefiles_end_operation,
+	.read			= cachefiles_read,
+	.write			= cachefiles_write,
+	.issue_write		= cachefiles_issue_write,
+	.prepare_read		= cachefiles_prepare_read,
+	.prepare_write		= cachefiles_prepare_write,
+	.prepare_write_subreq	= cachefiles_prepare_write_subreq,
+	.query_occupancy	= cachefiles_query_occupancy,
+};
+
+/*
+ * Open the cache file when beginning a cache operation.
+ */
+bool cachefiles_begin_operation(struct netfs_cache_resources *cres,
+				enum fscache_want_state want_state)
+{
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+
+	if (!cachefiles_cres_file(cres)) {
+		cres->ops = &cachefiles_netfs_cache_ops;
+		if (object->file) {
+			spin_lock(&object->lock);
+			if (!cres->cache_priv2 && object->file)
+				cres->cache_priv2 = get_file(object->file);
+			spin_unlock(&object->lock);
+		}
+	}
+
+	if (!cachefiles_cres_file(cres) && want_state != FSCACHE_WANT_PARAMS) {
+		pr_err("failed to get cres->file\n");
+		return false;
+	}
+
+	return true;
+}