Merge patch series "netfs: Miscellaneous fixes"

David Howells <[e-mail]> says:

Here are some miscellaneous fixes and changes for netfslib, if you could
pull them:

(1) Fix an oops in write-retry due to mis-resetting the I/O iterator.

(2) Fix the recording of transferred bytes for short DIO reads.

(3) Fix a request's work item to not require a reference, thereby avoiding
the need to get rid of it in BH/IRQ context.

(4) Fix waiting and waking to be consistent about the waitqueue used.

* patches from https://lore.kernel.org/[e-mail]:
netfs: Fix wait/wake to be consistent about the waitqueue used
netfs: Fix the request's work item to not require a ref
netfs: Fix setting of transferred bytes with short DIO reads
netfs: Fix oops in write-retry from mis-resetting the subreq iterator

Link: https://lore.kernel.org/[e-mail]
Signed-off-by: Christian Brauner <[e-mail]>

fs/9p/vfs_addr.c

···
 	len = p9_client_write(fid, subreq->start, &subreq->io_iter, &err);
 	if (len > 0)
 		__set_bit(NETFS_SREQ_MADE_PROGRESS, &subreq->flags);
-	netfs_write_subrequest_terminated(subreq, len ?: err, false);
+	netfs_write_subrequest_terminated(subreq, len ?: err);
 }
 
 /**

fs/afs/write.c

···
 
 #if 0 // Error injection
 	if (subreq->debug_index == 3)
-		return netfs_write_subrequest_terminated(subreq, -ENOANO, false);
+		return netfs_write_subrequest_terminated(subreq, -ENOANO);
 
 	if (!subreq->retry_count) {
 		set_bit(NETFS_SREQ_NEED_RETRY, &subreq->flags);
-		return netfs_write_subrequest_terminated(subreq, -EAGAIN, false);
+		return netfs_write_subrequest_terminated(subreq, -EAGAIN);
 	}
 #endif
 
 	op = afs_alloc_operation(wreq->netfs_priv, vnode->volume);
 	if (IS_ERR(op))
-		return netfs_write_subrequest_terminated(subreq, -EAGAIN, false);
+		return netfs_write_subrequest_terminated(subreq, -EAGAIN);
 
 	afs_op_set_vnode(op, 0, vnode);
 	op->file[0].dv_delta	= 1;
···
 		break;
 	}
 
-	netfs_write_subrequest_terminated(subreq, ret < 0 ? ret : subreq->len, false);
+	netfs_write_subrequest_terminated(subreq, ret < 0 ? ret : subreq->len);
 }
 
 void afs_issue_write(struct netfs_io_subrequest *subreq)

fs/cachefiles/io.c

···
 				ret = -ESTALE;
 		}
 
-		ki->term_func(ki->term_func_priv, ret, ki->was_async);
+		ki->term_func(ki->term_func_priv, ret);
 	}
 
 	cachefiles_put_kiocb(ki);
···
 
 presubmission_error:
 	if (term_func)
-		term_func(term_func_priv, ret < 0 ? ret : skipped, false);
+		term_func(term_func_priv, ret < 0 ? ret : skipped);
 	return ret;
 }
 
···
 	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, ki->was_async);
+		ki->term_func(ki->term_func_priv, ret);
 	cachefiles_put_kiocb(ki);
 }
 
···
 	ki = kzalloc(sizeof(struct cachefiles_kiocb), GFP_KERNEL);
 	if (!ki) {
 		if (term_func)
-			term_func(term_func_priv, -ENOMEM, false);
+			term_func(term_func_priv, -ENOMEM);
 		return -ENOMEM;
 	}
 
···
 {
 	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE)) {
 		if (term_func)
-			term_func(term_func_priv, -ENOBUFS, false);
+			term_func(term_func_priv, -ENOBUFS);
 		trace_netfs_sreq(term_func_priv, netfs_sreq_trace_cache_nowrite);
 		return -ENOBUFS;
 	}
···
 		pre = CACHEFILES_DIO_BLOCK_SIZE - off;
 		if (pre >= len) {
 			fscache_count_dio_misfit();
-			netfs_write_subrequest_terminated(subreq, len, false);
+			netfs_write_subrequest_terminated(subreq, len);
 			return;
 		}
 		subreq->transferred += pre;
···
 		len -= post;
 		if (len == 0) {
 			fscache_count_dio_misfit();
-			netfs_write_subrequest_terminated(subreq, post, false);
+			netfs_write_subrequest_terminated(subreq, post);
 			return;
 		}
 		iov_iter_truncate(&subreq->io_iter, len);
···
 					 &start, &len, len, true);
 	cachefiles_end_secure(cache, saved_cred);
 	if (ret < 0) {
-		netfs_write_subrequest_terminated(subreq, ret, false);
+		netfs_write_subrequest_terminated(subreq, ret);
 		return;
 	}
 

fs/ceph/addr.c

···
 	folio_start_private_2(page_folio(page)); /* [DEPRECATED] */
 }
 
-static void ceph_fscache_write_terminated(void *priv, ssize_t error, bool was_async)
+static void ceph_fscache_write_terminated(void *priv, ssize_t error)
 {
 	struct inode *inode = priv;
 

fs/erofs/fscache.c

···
 		erofs_fscache_req_put(req);
 }
 
-static void erofs_fscache_req_end_io(void *priv,
-		ssize_t transferred_or_error, bool was_async)
+static void erofs_fscache_req_end_io(void *priv, ssize_t transferred_or_error)
 {
 	struct erofs_fscache_io *io = priv;
 	struct erofs_fscache_rq *req = io->private;
···
 	struct bio_vec bvecs[BIO_MAX_VECS];
 };
 
-static void erofs_fscache_bio_endio(void *priv,
-		ssize_t transferred_or_error, bool was_async)
+static void erofs_fscache_bio_endio(void *priv, ssize_t transferred_or_error)
 {
 	struct erofs_fscache_bio *io = priv;
 

fs/netfs/buffered_read.c

···
 				if (ret < 0) {
 					subreq->error = ret;
 					/* Not queued - release both refs. */
-					netfs_put_subrequest(subreq, false,
+					netfs_put_subrequest(subreq,
 							     netfs_sreq_trace_put_cancel);
-					netfs_put_subrequest(subreq, false,
+					netfs_put_subrequest(subreq,
 							     netfs_sreq_trace_put_cancel);
 					break;
 				}
···
 			subreq->error = ret;
 			trace_netfs_sreq(subreq, netfs_sreq_trace_cancel);
 			/* Not queued - release both refs. */
-			netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_cancel);
-			netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_cancel);
+			netfs_put_subrequest(subreq, netfs_sreq_trace_put_cancel);
+			netfs_put_subrequest(subreq, netfs_sreq_trace_put_cancel);
 			break;
 		}
 		size -= slice;
···
 	if (unlikely(size > 0)) {
 		smp_wmb(); /* Write lists before ALL_QUEUED. */
 		set_bit(NETFS_RREQ_ALL_QUEUED, &rreq->flags);
-		netfs_wake_read_collector(rreq);
+		netfs_wake_collector(rreq);
 	}
 
 	/* Defer error return as we may need to wait for outstanding I/O. */
···
 		goto cleanup_free;
 	netfs_read_to_pagecache(rreq, ractl);
 
-	netfs_put_request(rreq, true, netfs_rreq_trace_put_return);
-	return;
+	return netfs_put_request(rreq, netfs_rreq_trace_put_return);
 
 cleanup_free:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_failed);
-	return;
+	return netfs_put_request(rreq, netfs_rreq_trace_put_failed);
 }
 EXPORT_SYMBOL(netfs_readahead);
 
···
 		folio_mark_uptodate(folio);
 	}
 	folio_unlock(folio);
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 	return ret < 0 ? ret : 0;
 
 discard:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_discard);
+	netfs_put_request(rreq, netfs_rreq_trace_put_discard);
 alloc_error:
 	folio_unlock(folio);
 	return ret;
···
 
 	netfs_read_to_pagecache(rreq, NULL);
 	ret = netfs_wait_for_read(rreq);
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 	return ret < 0 ? ret : 0;
 
 discard:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_discard);
+	netfs_put_request(rreq, netfs_rreq_trace_put_discard);
 alloc_error:
 	folio_unlock(folio);
 	return ret;
···
 	ret = netfs_wait_for_read(rreq);
 	if (ret < 0)
 		goto error;
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 
 have_folio:
 	ret = folio_wait_private_2_killable(folio);
···
 	return 0;
 
 error_put:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_failed);
+	netfs_put_request(rreq, netfs_rreq_trace_put_failed);
 error:
 	if (folio) {
 		folio_unlock(folio);
···
 
 	netfs_read_to_pagecache(rreq, NULL);
 	ret = netfs_wait_for_read(rreq);
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 	return ret < 0 ? ret : 0;
 
 error_put:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_discard);
+	netfs_put_request(rreq, netfs_rreq_trace_put_discard);
 error:
 	_leave(" = %d", ret);
 	return ret;

fs/netfs/buffered_write.c

···
 		wbc_detach_inode(&wbc);
 		if (ret2 == -EIOCBQUEUED)
 			return ret2;
-		if (ret == 0)
+		if (ret == 0 && ret2 < 0)
 			ret = ret2;
 	}
 

fs/netfs/direct_read.c

···
 		if (rreq->netfs_ops->prepare_read) {
 			ret = rreq->netfs_ops->prepare_read(subreq);
 			if (ret < 0) {
-				netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_cancel);
+				netfs_put_subrequest(subreq, netfs_sreq_trace_put_cancel);
 				break;
 			}
 		}
···
 		rreq->netfs_ops->issue_read(subreq);
 
 		if (test_bit(NETFS_RREQ_PAUSE, &rreq->flags))
-			netfs_wait_for_pause(rreq);
+			netfs_wait_for_paused_read(rreq);
 		if (test_bit(NETFS_RREQ_FAILED, &rreq->flags))
 			break;
 		cond_resched();
···
 	if (unlikely(size > 0)) {
 		smp_wmb(); /* Write lists before ALL_QUEUED. */
 		set_bit(NETFS_RREQ_ALL_QUEUED, &rreq->flags);
-		netfs_wake_read_collector(rreq);
+		netfs_wake_collector(rreq);
 	}
 
 	return ret;
···
 	ret = netfs_dispatch_unbuffered_reads(rreq);
 
 	if (!rreq->submitted) {
-		netfs_put_request(rreq, false, netfs_rreq_trace_put_no_submit);
+		netfs_put_request(rreq, netfs_rreq_trace_put_no_submit);
 		inode_dio_end(rreq->inode);
 		ret = 0;
 		goto out;
···
 	}
 
 out:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 	if (ret > 0)
 		orig_count -= ret;
 	return ret;

fs/netfs/direct_write.c

···
 	}
 
 	__set_bit(NETFS_RREQ_USE_IO_ITER, &wreq->flags);
+	if (async)
+		__set_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &wreq->flags);
 
 	/* Copy the data into the bounce buffer and encrypt it. */
 	// TODO
···
 
 	if (!async) {
 		trace_netfs_rreq(wreq, netfs_rreq_trace_wait_ip);
-		wait_on_bit(&wreq->flags, NETFS_RREQ_IN_PROGRESS,
-			    TASK_UNINTERRUPTIBLE);
-		ret = wreq->error;
-		if (ret == 0) {
-			ret = wreq->transferred;
+		ret = netfs_wait_for_write(wreq);
+		if (ret > 0)
 			iocb->ki_pos += ret;
-		}
 	} else {
 		ret = -EIOCBQUEUED;
 	}
 
 out:
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(wreq, netfs_rreq_trace_put_return);
 	return ret;
 }
 EXPORT_SYMBOL(netfs_unbuffered_write_iter_locked);

fs/netfs/fscache_io.c

···
 /*
  * Deal with the completion of writing the data to the cache.
  */
-static void fscache_wreq_done(void *priv, ssize_t transferred_or_error,
-			      bool was_async)
+static void fscache_wreq_done(void *priv, ssize_t transferred_or_error)
 {
 	struct fscache_write_request *wreq = priv;
 
···
 					wreq->set_bits);
 
 	if (wreq->term_func)
-		wreq->term_func(wreq->term_func_priv, transferred_or_error,
-				was_async);
+		wreq->term_func(wreq->term_func_priv, transferred_or_error);
 	fscache_end_operation(&wreq->cache_resources);
 	kfree(wreq);
 }
···
 	return;
 
 abandon_end:
-	return fscache_wreq_done(wreq, ret, false);
+	return fscache_wreq_done(wreq, ret);
 abandon_free:
 	kfree(wreq);
 abandon:
 	if (using_pgpriv2)
 		fscache_clear_page_bits(mapping, start, len, cond);
 	if (term_func)
-		term_func(term_func_priv, ret, false);
+		term_func(term_func_priv, ret);
 }
 EXPORT_SYMBOL(__fscache_write_to_cache);
 

fs/netfs/internal.h

···
 /*
  * buffered_read.c
  */
-void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error, bool was_async);
+void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error);
 int netfs_prefetch_for_write(struct file *file, struct folio *folio,
 			     size_t offset, size_t len);
 
···
 struct folio_queue *netfs_buffer_make_space(struct netfs_io_request *rreq,
 					    enum netfs_folioq_trace trace);
 void netfs_reset_iter(struct netfs_io_subrequest *subreq);
+void netfs_wake_collector(struct netfs_io_request *rreq);
+void netfs_subreq_clear_in_progress(struct netfs_io_subrequest *subreq);
+void netfs_wait_for_in_progress_stream(struct netfs_io_request *rreq,
+				       struct netfs_io_stream *stream);
+ssize_t netfs_wait_for_read(struct netfs_io_request *rreq);
+ssize_t netfs_wait_for_write(struct netfs_io_request *rreq);
+void netfs_wait_for_paused_read(struct netfs_io_request *rreq);
+void netfs_wait_for_paused_write(struct netfs_io_request *rreq);
 
 /*
  * objects.c
···
 					     loff_t start, size_t len,
 					     enum netfs_io_origin origin);
 void netfs_get_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what);
-void netfs_clear_subrequests(struct netfs_io_request *rreq, bool was_async);
-void netfs_put_request(struct netfs_io_request *rreq, bool was_async,
-		       enum netfs_rreq_ref_trace what);
+void netfs_clear_subrequests(struct netfs_io_request *rreq);
+void netfs_put_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what);
 struct netfs_io_subrequest *netfs_alloc_subrequest(struct netfs_io_request *rreq);
 
 static inline void netfs_see_request(struct netfs_io_request *rreq,
···
 /*
  * read_collect.c
  */
+bool netfs_read_collection(struct netfs_io_request *rreq);
 void netfs_read_collection_worker(struct work_struct *work);
-void netfs_wake_read_collector(struct netfs_io_request *rreq);
-void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error, bool was_async);
-ssize_t netfs_wait_for_read(struct netfs_io_request *rreq);
-void netfs_wait_for_pause(struct netfs_io_request *rreq);
+void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error);
 
 /*
  * read_pgpriv2.c
···
  * write_collect.c
  */
 int netfs_folio_written_back(struct folio *folio);
+bool netfs_write_collection(struct netfs_io_request *wreq);
 void netfs_write_collection_worker(struct work_struct *work);
-void netfs_wake_write_collector(struct netfs_io_request *wreq, bool was_async);
 
 /*
  * write_issue.c
···
 int netfs_advance_writethrough(struct netfs_io_request *wreq, struct writeback_control *wbc,
 			       struct folio *folio, size_t copied, bool to_page_end,
 			       struct folio **writethrough_cache);
-int netfs_end_writethrough(struct netfs_io_request *wreq, struct writeback_control *wbc,
-			   struct folio *writethrough_cache);
+ssize_t netfs_end_writethrough(struct netfs_io_request *wreq, struct writeback_control *wbc,
+			       struct folio *writethrough_cache);
 int netfs_unbuffered_write(struct netfs_io_request *wreq, bool may_wait, size_t len);
 
 /*
···
 	    netfs_group != NETFS_FOLIO_COPY_TO_CACHE &&
 	    refcount_sub_and_test(nr, &netfs_group->ref))
 		netfs_group->free(netfs_group);
+}
+
+/*
+ * Clear and wake up a NETFS_RREQ_* flag bit on a request.
+ */
+static inline void netfs_wake_rreq_flag(struct netfs_io_request *rreq,
+					unsigned int rreq_flag,
+					enum netfs_rreq_trace trace)
+{
+	if (test_bit(rreq_flag, &rreq->flags)) {
+		trace_netfs_rreq(rreq, trace);
+		clear_bit_unlock(rreq_flag, &rreq->flags);
+		smp_mb__after_atomic(); /* Set flag before task state */
+		wake_up(&rreq->waitq);
+	}
 }
 
 /*

fs/netfs/misc.c

···
 	return true;
 }
 EXPORT_SYMBOL(netfs_release_folio);
+
+/*
+ * Wake the collection work item.
+ */
+void netfs_wake_collector(struct netfs_io_request *rreq)
+{
+	if (test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags) &&
+	    !test_bit(NETFS_RREQ_RETRYING, &rreq->flags)) {
+		queue_work(system_unbound_wq, &rreq->work);
+	} else {
+		trace_netfs_rreq(rreq, netfs_rreq_trace_wake_queue);
+		wake_up(&rreq->waitq);
+	}
+}
+
+/*
+ * Mark a subrequest as no longer being in progress and, if need be, wake the
+ * collector.
+ */
+void netfs_subreq_clear_in_progress(struct netfs_io_subrequest *subreq)
+{
+	struct netfs_io_request *rreq = subreq->rreq;
+	struct netfs_io_stream *stream = &rreq->io_streams[subreq->stream_nr];
+
+	clear_bit_unlock(NETFS_SREQ_IN_PROGRESS, &subreq->flags);
+	smp_mb__after_atomic(); /* Clear IN_PROGRESS before task state */
+
+	/* If we are at the head of the queue, wake up the collector. */
+	if (list_is_first(&subreq->rreq_link, &stream->subrequests) ||
+	    test_bit(NETFS_RREQ_RETRYING, &rreq->flags))
+		netfs_wake_collector(rreq);
+}
+
+/*
+ * Wait for all outstanding I/O in a stream to quiesce.
+ */
+void netfs_wait_for_in_progress_stream(struct netfs_io_request *rreq,
+				       struct netfs_io_stream *stream)
+{
+	struct netfs_io_subrequest *subreq;
+	DEFINE_WAIT(myself);
+
+	list_for_each_entry(subreq, &stream->subrequests, rreq_link) {
+		if (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags))
+			continue;
+
+		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
+		for (;;) {
+			prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
+
+			if (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags))
+				break;
+
+			trace_netfs_sreq(subreq, netfs_sreq_trace_wait_for);
+			schedule();
+			trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
+		}
+	}
+
+	finish_wait(&rreq->waitq, &myself);
+}
+
+/*
+ * Perform collection in app thread if not offloaded to workqueue.
+ */
+static int netfs_collect_in_app(struct netfs_io_request *rreq,
+				bool (*collector)(struct netfs_io_request *rreq))
+{
+	bool need_collect = false, inactive = true;
+
+	for (int i = 0; i < NR_IO_STREAMS; i++) {
+		struct netfs_io_subrequest *subreq;
+		struct netfs_io_stream *stream = &rreq->io_streams[i];
+
+		if (!stream->active)
+			continue;
+		inactive = false;
+		trace_netfs_collect_stream(rreq, stream);
+		subreq = list_first_entry_or_null(&stream->subrequests,
+						  struct netfs_io_subrequest,
+						  rreq_link);
+		if (subreq &&
+		    (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags) ||
+		     test_bit(NETFS_SREQ_MADE_PROGRESS, &subreq->flags))) {
+			need_collect = true;
+			break;
+		}
+	}
+
+	if (!need_collect && !inactive)
+		return 0; /* Sleep */
+
+	__set_current_state(TASK_RUNNING);
+	if (collector(rreq)) {
+		/* Drop the ref from the NETFS_RREQ_IN_PROGRESS flag. */
+		netfs_put_request(rreq, netfs_rreq_trace_put_work_ip);
+		return 1; /* Done */
+	}
+
+	if (inactive) {
+		WARN(true, "Failed to collect inactive req R=%08x\n",
+		     rreq->debug_id);
+		cond_resched();
+	}
+	return 2; /* Again */
+}
+
+/*
+ * Wait for a request to complete, successfully or otherwise.
+ */
+static ssize_t netfs_wait_for_request(struct netfs_io_request *rreq,
+				      bool (*collector)(struct netfs_io_request *rreq))
+{
+	DEFINE_WAIT(myself);
+	ssize_t ret;
+
+	for (;;) {
+		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
+		prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
+
+		if (!test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags)) {
+			switch (netfs_collect_in_app(rreq, collector)) {
+			case 0:
+				break;
+			case 1:
+				goto all_collected;
+			case 2:
+				continue;
+			}
+		}
+
+		if (!test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags))
+			break;
+
+		schedule();
+		trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
+	}
+
+all_collected:
+	finish_wait(&rreq->waitq, &myself);
+
+	ret = rreq->error;
+	if (ret == 0) {
+		ret = rreq->transferred;
+		switch (rreq->origin) {
+		case NETFS_DIO_READ:
+		case NETFS_DIO_WRITE:
+		case NETFS_READ_SINGLE:
+		case NETFS_UNBUFFERED_WRITE:
+			break;
+		default:
+			if (rreq->submitted < rreq->len) {
+				trace_netfs_failure(rreq, NULL, ret, netfs_fail_short_read);
+				ret = -EIO;
+			}
+			break;
+		}
+	}
+
+	return ret;
+}
+
+ssize_t netfs_wait_for_read(struct netfs_io_request *rreq)
+{
+	return netfs_wait_for_request(rreq, netfs_read_collection);
+}
+
+ssize_t netfs_wait_for_write(struct netfs_io_request *rreq)
+{
+	return netfs_wait_for_request(rreq, netfs_write_collection);
+}
+
+/*
+ * Wait for a paused operation to unpause or complete in some manner.
+ */
+static void netfs_wait_for_pause(struct netfs_io_request *rreq,
+				 bool (*collector)(struct netfs_io_request *rreq))
+{
+	DEFINE_WAIT(myself);
+
+	trace_netfs_rreq(rreq, netfs_rreq_trace_wait_pause);
+
+	for (;;) {
+		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
+		prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
+
+		if (!test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags)) {
+			switch (netfs_collect_in_app(rreq, collector)) {
+			case 0:
+				break;
+			case 1:
+				goto all_collected;
+			case 2:
+				continue;
+			}
+		}
+
+		if (!test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags) ||
+		    !test_bit(NETFS_RREQ_PAUSE, &rreq->flags))
+			break;
+
+		schedule();
+		trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
+	}
+
+all_collected:
+	finish_wait(&rreq->waitq, &myself);
+}
+
+void netfs_wait_for_paused_read(struct netfs_io_request *rreq)
+{
+	return netfs_wait_for_pause(rreq, netfs_read_collection);
+}
+
+void netfs_wait_for_paused_write(struct netfs_io_request *rreq)
+{
+	return netfs_wait_for_pause(rreq, netfs_write_collection);
+}

fs/netfs/objects.c

···
 #include <linux/delay.h>
 #include "internal.h"
 
+static void netfs_free_request(struct work_struct *work);
+
 /*
  * Allocate an I/O request and initialise it.
  */
···
 	}
 
 	memset(rreq, 0, kmem_cache_size(cache));
+	INIT_WORK(&rreq->cleanup_work, netfs_free_request);
 	rreq->start	= start;
 	rreq->len	= len;
 	rreq->origin	= origin;
···
 	INIT_LIST_HEAD(&rreq->io_streams[0].subrequests);
 	INIT_LIST_HEAD(&rreq->io_streams[1].subrequests);
 	init_waitqueue_head(&rreq->waitq);
-	refcount_set(&rreq->ref, 1);
+	refcount_set(&rreq->ref, 2);
 
 	if (origin == NETFS_READAHEAD ||
 	    origin == NETFS_READPAGE ||
···
 	}
 
 	atomic_inc(&ctx->io_count);
-	trace_netfs_rreq_ref(rreq->debug_id, 1, netfs_rreq_trace_new);
+	trace_netfs_rreq_ref(rreq->debug_id, refcount_read(&rreq->ref), netfs_rreq_trace_new);
 	netfs_proc_add_rreq(rreq);
 	netfs_stat(&netfs_n_rh_rreq);
 	return rreq;
···
 	trace_netfs_rreq_ref(rreq->debug_id, r + 1, what);
 }
 
-void netfs_clear_subrequests(struct netfs_io_request *rreq, bool was_async)
+void netfs_clear_subrequests(struct netfs_io_request *rreq)
 {
 	struct netfs_io_subrequest *subreq;
 	struct netfs_io_stream *stream;
···
 			subreq = list_first_entry(&stream->subrequests,
 						  struct netfs_io_subrequest, rreq_link);
 			list_del(&subreq->rreq_link);
-			netfs_put_subrequest(subreq, was_async,
-					     netfs_sreq_trace_put_clear);
+			netfs_put_subrequest(subreq, netfs_sreq_trace_put_clear);
 		}
 	}
 }
···
 static void netfs_free_request(struct work_struct *work)
 {
 	struct netfs_io_request *rreq =
-		container_of(work, struct netfs_io_request, work);
+		container_of(work, struct netfs_io_request, cleanup_work);
 	struct netfs_inode *ictx = netfs_inode(rreq->inode);
 	unsigned int i;
 
 	trace_netfs_rreq(rreq, netfs_rreq_trace_free);
+
+	/* Cancel/flush the result collection worker.  That does not carry a
+	 * ref of its own, so we must wait for it somewhere.
+	 */
+	cancel_work_sync(&rreq->work);
+
 	netfs_proc_del_rreq(rreq);
-	netfs_clear_subrequests(rreq, false);
+	netfs_clear_subrequests(rreq);
 	if (rreq->netfs_ops->free_request)
 		rreq->netfs_ops->free_request(rreq);
 	if (rreq->cache_resources.ops)
···
 	call_rcu(&rreq->rcu, netfs_free_request_rcu);
 }
 
-void netfs_put_request(struct netfs_io_request *rreq, bool was_async,
-		       enum netfs_rreq_ref_trace what)
+void netfs_put_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what)
 {
 	unsigned int debug_id;
 	bool dead;
···
 		debug_id = rreq->debug_id;
 		dead = __refcount_dec_and_test(&rreq->ref, &r);
 		trace_netfs_rreq_ref(debug_id, r - 1, what);
-		if (dead) {
-			if (was_async) {
-				rreq->work.func = netfs_free_request;
-				if (!queue_work(system_unbound_wq, &rreq->work))
-					WARN_ON(1);
-			} else {
-				netfs_free_request(&rreq->work);
-			}
-		}
+		if (dead)
+			WARN_ON(!queue_work(system_unbound_wq, &rreq->cleanup_work));
 	}
 }
 
···
 			     what);
 }
 
-static void netfs_free_subrequest(struct netfs_io_subrequest *subreq,
-				  bool was_async)
+static void netfs_free_subrequest(struct netfs_io_subrequest *subreq)
 {
 	struct netfs_io_request *rreq = subreq->rreq;
 
···
 		rreq->netfs_ops->free_subrequest(subreq);
 	mempool_free(subreq, rreq->netfs_ops->subrequest_pool ?: &netfs_subrequest_pool);
 	netfs_stat_d(&netfs_n_rh_sreq);
-	netfs_put_request(rreq, was_async, netfs_rreq_trace_put_subreq);
+	netfs_put_request(rreq, netfs_rreq_trace_put_subreq);
 }
 
-void netfs_put_subrequest(struct netfs_io_subrequest *subreq, bool was_async,
+void netfs_put_subrequest(struct netfs_io_subrequest *subreq,
 			  enum netfs_sreq_ref_trace what)
 {
 	unsigned int debug_index = subreq->debug_index;
···
 	dead = __refcount_dec_and_test(&subreq->ref, &r);
 	trace_netfs_sreq_ref(debug_id, debug_index, r - 1, what);
 	if (dead)
-		netfs_free_subrequest(subreq, was_async);
+		netfs_free_subrequest(subreq);
 }

fs/netfs/read_collect.c

···
 			stream->need_retry = true;
 			notes |= NEED_RETRY | MADE_PROGRESS;
 			break;
+		} else if (test_bit(NETFS_RREQ_SHORT_TRANSFER, &rreq->flags)) {
+			notes |= MADE_PROGRESS;
 		} else {
 			if (!stream->failed)
-				stream->transferred = stream->collected_to - rreq->start;
+				stream->transferred += transferred;
+			if (front->transferred < front->len)
+				set_bit(NETFS_RREQ_SHORT_TRANSFER, &rreq->flags);
 			notes |= MADE_PROGRESS;
 		}
 
···
 						 struct netfs_io_subrequest, rreq_link);
 		stream->front = front;
 		spin_unlock(&rreq->lock);
-		netfs_put_subrequest(remove, false,
+		netfs_put_subrequest(remove,
 				     notes & ABANDON_SREQ ?
 				     netfs_sreq_trace_put_abandon :
 				     netfs_sreq_trace_put_done);
···
 
 	if (notes & NEED_RETRY)
 		goto need_retry;
-	if ((notes & MADE_PROGRESS) && test_bit(NETFS_RREQ_PAUSE, &rreq->flags)) {
-		trace_netfs_rreq(rreq, netfs_rreq_trace_unpause);
-		clear_bit_unlock(NETFS_RREQ_PAUSE, &rreq->flags);
-		smp_mb__after_atomic(); /* Set PAUSE before task state */
-		wake_up(&rreq->waitq);
-	}
-
 	if (notes & MADE_PROGRESS) {
+		netfs_wake_rreq_flag(rreq, NETFS_RREQ_PAUSE, netfs_rreq_trace_unpause);
 		//cond_resched();
 		goto reassess;
 	}
···
  */
 static void netfs_rreq_assess_dio(struct netfs_io_request *rreq)
 {
-	struct netfs_io_subrequest *subreq;
-	struct netfs_io_stream *stream = &rreq->io_streams[0];
 	unsigned int i;
-
-	/* Collect unbuffered reads and direct reads, adding up the transfer
-	 * sizes until we find the first short or failed subrequest.
-	 */
-	list_for_each_entry(subreq, &stream->subrequests, rreq_link) {
-		rreq->transferred += subreq->transferred;
-
-		if (subreq->transferred < subreq->len ||
-		    test_bit(NETFS_SREQ_FAILED, &subreq->flags)) {
-			rreq->error = subreq->error;
-			break;
-		}
-	}
 
 	if (rreq->origin == NETFS_DIO_READ) {
 		for (i = 0; i < rreq->direct_bv_count; i++) {
···
  * Note that we're in normal kernel thread context at this point, possibly
  * running on a workqueue.
  */
-static void netfs_read_collection(struct netfs_io_request *rreq)
+bool netfs_read_collection(struct netfs_io_request *rreq)
 {
 	struct netfs_io_stream *stream = &rreq->io_streams[0];
 
···
 	 * queue is empty.
 	 */
 	if (!test_bit(NETFS_RREQ_ALL_QUEUED, &rreq->flags))
-		return;
+		return false;
 	smp_rmb(); /* Read ALL_QUEUED before subreq lists. */
 
 	if (!list_empty(&stream->subrequests))
-		return;
+		return false;
 
 	/* Okay, declare that all I/O is complete. */
 	rreq->transferred = stream->transferred;
···
 	}
 	task_io_account_read(rreq->transferred);
 
-	trace_netfs_rreq(rreq, netfs_rreq_trace_wake_ip);
-	clear_and_wake_up_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags);
+	netfs_wake_rreq_flag(rreq, NETFS_RREQ_IN_PROGRESS, netfs_rreq_trace_wake_ip);
+	/* As we cleared NETFS_RREQ_IN_PROGRESS, we acquired its ref. */
 
 	trace_netfs_rreq(rreq, netfs_rreq_trace_done);
-	netfs_clear_subrequests(rreq, false);
+	netfs_clear_subrequests(rreq);
 	netfs_unlock_abandoned_read_pages(rreq);
 	if (unlikely(rreq->copy_to_cache))
 		netfs_pgpriv2_end_copy_to_cache(rreq);
+	return true;
 }
 
 void netfs_read_collection_worker(struct work_struct *work)
···
 	struct netfs_io_request *rreq = container_of(work, struct netfs_io_request, work);
 
 	netfs_see_request(rreq, netfs_rreq_trace_see_work);
-	if (test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags))
-		netfs_read_collection(rreq);
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_work);
-}
-
-/*
- * Wake the collection work item.
- */
-void netfs_wake_read_collector(struct netfs_io_request *rreq)
-{
-	if (test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags) &&
-	    !test_bit(NETFS_RREQ_RETRYING, &rreq->flags)) {
-		if (!work_pending(&rreq->work)) {
-			netfs_get_request(rreq, netfs_rreq_trace_get_work);
-			if (!queue_work(system_unbound_wq, &rreq->work))
-				netfs_put_request(rreq, true, netfs_rreq_trace_put_work_nq);
-		}
-	} else {
-		trace_netfs_rreq(rreq, netfs_rreq_trace_wake_queue);
-		wake_up(&rreq->waitq);
+	if (test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags)) {
+		if (netfs_read_collection(rreq))
+			/* Drop the ref from the IN_PROGRESS flag. */
+			netfs_put_request(rreq, netfs_rreq_trace_put_work_ip);
+		else
+			netfs_see_request(rreq, netfs_rreq_trace_see_work_complete);
 	}
 }
 
···
 	    list_is_first(&subreq->rreq_link, &stream->subrequests)
 	    ) {
 		__set_bit(NETFS_SREQ_MADE_PROGRESS, &subreq->flags);
-		netfs_wake_read_collector(rreq);
+		netfs_wake_collector(rreq);
 	}
 }
 EXPORT_SYMBOL(netfs_read_subreq_progress);
···
 void netfs_read_subreq_terminated(struct netfs_io_subrequest *subreq)
 {
 	struct netfs_io_request *rreq = subreq->rreq;
-	struct netfs_io_stream *stream = &rreq->io_streams[0];
 
 	switch (subreq->source) {
 	case NETFS_READ_FROM_CACHE:
···
 	}
 
 	trace_netfs_sreq(subreq, netfs_sreq_trace_terminated);
-
-	clear_bit_unlock(NETFS_SREQ_IN_PROGRESS, &subreq->flags);
-	smp_mb__after_atomic(); /* Clear IN_PROGRESS before task state */
-
-	/* If we are at the head of the queue, wake up the collector. */
-	if (list_is_first(&subreq->rreq_link, &stream->subrequests) ||
-	    test_bit(NETFS_RREQ_RETRYING, &rreq->flags))
-		netfs_wake_read_collector(rreq);
-
-	netfs_put_subrequest(subreq, true, netfs_sreq_trace_put_terminated);
+	netfs_subreq_clear_in_progress(subreq);
+	netfs_put_subrequest(subreq, netfs_sreq_trace_put_terminated);
 }
 EXPORT_SYMBOL(netfs_read_subreq_terminated);
 
 /*
  * Handle termination of a read from the cache.
  */
-void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error, bool was_async)
+void netfs_cache_read_terminated(void *priv, ssize_t transferred_or_error)
 {
 	struct netfs_io_subrequest *subreq = priv;
 
···
 		subreq->error = transferred_or_error;
 	}
 	netfs_read_subreq_terminated(subreq);
-}
-
-/*
- * Wait for the read operation to complete, successfully or otherwise.
- */
-ssize_t netfs_wait_for_read(struct netfs_io_request *rreq)
-{
-	struct netfs_io_subrequest *subreq;
-	struct netfs_io_stream *stream = &rreq->io_streams[0];
-	DEFINE_WAIT(myself);
-	ssize_t ret;
-
-	for (;;) {
-		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
-		prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
-
-		subreq = list_first_entry_or_null(&stream->subrequests,
-						  struct netfs_io_subrequest, rreq_link);
-		if (subreq &&
-		    (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags) ||
-		     test_bit(NETFS_SREQ_MADE_PROGRESS, &subreq->flags))) {
-			__set_current_state(TASK_RUNNING);
-			netfs_read_collection(rreq);
-			continue;
-		}
-
-		if (!test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags))
-			break;
-
-		schedule();
-		trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
-	}
-
-	finish_wait(&rreq->waitq, &myself);
-
-	ret = rreq->error;
-	if (ret == 0) {
-		ret = rreq->transferred;
-		switch (rreq->origin) {
-		case NETFS_DIO_READ:
-		case NETFS_READ_SINGLE:
-			ret = rreq->transferred;
-			break;
-		default:
-			if (rreq->submitted < rreq->len) {
-				trace_netfs_failure(rreq, NULL, ret, netfs_fail_short_read);
-				ret = -EIO;
-			}
-			break;
-		}
-	}
-
-	return ret;
-}
-
-/*
- * Wait for a paused read operation to unpause or complete in some manner.
- */
-void netfs_wait_for_pause(struct netfs_io_request *rreq)
-{
-	struct netfs_io_subrequest *subreq;
-	struct netfs_io_stream *stream = &rreq->io_streams[0];
-	DEFINE_WAIT(myself);
-
-	trace_netfs_rreq(rreq, netfs_rreq_trace_wait_pause);
-
-	for (;;) {
-		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
-		prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
-
-		if (!test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags)) {
-			subreq = list_first_entry_or_null(&stream->subrequests,
-							  struct netfs_io_subrequest, rreq_link);
-			if (subreq &&
-			    (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags) ||
-			     test_bit(NETFS_SREQ_MADE_PROGRESS, &subreq->flags))) {
-				__set_current_state(TASK_RUNNING);
-				netfs_read_collection(rreq);
-				continue;
-			}
-		}
-
-		if (!test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags) ||
-		    !test_bit(NETFS_RREQ_PAUSE, &rreq->flags))
-			break;
-
-		schedule();
-		trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
-	}
-
-	finish_wait(&rreq->waitq, &myself);
 }

fs/netfs/read_pgpriv2.c

···
 	return creq;
 
 cancel_put:
-	netfs_put_request(creq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(creq, netfs_rreq_trace_put_return);
 cancel:
 	rreq->copy_to_cache = ERR_PTR(-ENOBUFS);
 	clear_bit(NETFS_RREQ_FOLIO_COPY_TO_CACHE, &rreq->flags);
···
 	smp_wmb(); /* Write lists before ALL_QUEUED. */
 	set_bit(NETFS_RREQ_ALL_QUEUED, &creq->flags);
 
-	netfs_put_request(creq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(creq, netfs_rreq_trace_put_return);
 	creq->copy_to_cache = NULL;
 }
 

fs/netfs/read_retry.c

···
 						      &stream->subrequests, rreq_link) {
 				trace_netfs_sreq(subreq, netfs_sreq_trace_superfluous);
 				list_del(&subreq->rreq_link);
-				netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_done);
+				netfs_put_subrequest(subreq, netfs_sreq_trace_put_done);
 				if (subreq == to)
 					break;
 			}
···
  */
 void netfs_retry_reads(struct netfs_io_request *rreq)
 {
-	struct netfs_io_subrequest *subreq;
 	struct netfs_io_stream *stream = &rreq->io_streams[0];
-	DEFINE_WAIT(myself);
 
 	netfs_stat(&netfs_n_rh_retry_read_req);
-
-	set_bit(NETFS_RREQ_RETRYING, &rreq->flags);
 
 	/* Wait for all outstanding I/O to quiesce before performing retries as
 	 * we may need to renegotiate the I/O sizes.
 	 */
-	list_for_each_entry(subreq, &stream->subrequests, rreq_link) {
-		if (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags))
-			continue;
-
-		trace_netfs_rreq(rreq, netfs_rreq_trace_wait_queue);
-		for (;;) {
-			prepare_to_wait(&rreq->waitq, &myself, TASK_UNINTERRUPTIBLE);
-
-			if (!test_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags))
-				break;
-
-			trace_netfs_sreq(subreq, netfs_sreq_trace_wait_for);
-			schedule();
-			trace_netfs_rreq(rreq, netfs_rreq_trace_woke_queue);
-		}
-
-		finish_wait(&rreq->waitq, &myself);
-	}
+	set_bit(NETFS_RREQ_RETRYING, &rreq->flags);
+	netfs_wait_for_in_progress_stream(rreq, stream);
 	clear_bit(NETFS_RREQ_RETRYING, &rreq->flags);
 
 	trace_netfs_rreq(rreq, netfs_rreq_trace_resubmit);

fs/netfs/read_single.c

···
 	set_bit(NETFS_RREQ_ALL_QUEUED, &rreq->flags);
 	return ret;
 cancel:
-	netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_cancel);
+	netfs_put_subrequest(subreq, netfs_sreq_trace_put_cancel);
 	return ret;
 }
 
···
 	netfs_single_dispatch_read(rreq);
 
 	ret = netfs_wait_for_read(rreq);
-	netfs_put_request(rreq, true, netfs_rreq_trace_put_return);
+	netfs_put_request(rreq, netfs_rreq_trace_put_return);
 	return ret;
 
 cleanup_free:
-	netfs_put_request(rreq, false, netfs_rreq_trace_put_failed);
+	netfs_put_request(rreq, netfs_rreq_trace_put_failed);
 	return ret;
 }
 EXPORT_SYMBOL(netfs_read_single);

fs/netfs/write_collect.c

···
 							 struct netfs_io_subrequest, rreq_link);
 			stream->front = front;
 			spin_unlock(&wreq->lock);
-			netfs_put_subrequest(remove, false,
+			netfs_put_subrequest(remove,
 					     notes & SAW_FAILURE ?
 					     netfs_sreq_trace_put_cancel :
 					     netfs_sreq_trace_put_done);
···
 
 	if (notes & NEED_RETRY)
 		goto need_retry;
-	if ((notes & MADE_PROGRESS) && test_bit(NETFS_RREQ_PAUSE, &wreq->flags)) {
-		trace_netfs_rreq(wreq, netfs_rreq_trace_unpause);
-		clear_bit_unlock(NETFS_RREQ_PAUSE, &wreq->flags);
-		smp_mb__after_atomic(); /* Set PAUSE before task state */
-		wake_up(&wreq->waitq);
-	}
 
-	if (notes & NEED_REASSESS) {
+	if (notes & MADE_PROGRESS) {
+		netfs_wake_rreq_flag(wreq, NETFS_RREQ_PAUSE, netfs_rreq_trace_unpause);
 		//cond_resched();
 		goto reassess_streams;
 	}
-	if (notes & MADE_PROGRESS) {
+
+	if (notes & NEED_REASSESS) {
 		//cond_resched();
 		goto reassess_streams;
 	}
···
 /*
  * Perform the collection of subrequests, folios and encryption buffers.
  */
-void netfs_write_collection_worker(struct work_struct *work)
+bool netfs_write_collection(struct netfs_io_request *wreq)
 {
-	struct netfs_io_request *wreq = container_of(work, struct netfs_io_request, work);
 	struct netfs_inode *ictx = netfs_inode(wreq->inode);
 	size_t transferred;
 	int s;
 
 	_enter("R=%x", wreq->debug_id);
 
-	netfs_see_request(wreq, netfs_rreq_trace_see_work);
-	if (!test_bit(NETFS_RREQ_IN_PROGRESS, &wreq->flags)) {
-		netfs_put_request(wreq, false, netfs_rreq_trace_put_work);
-		return;
-	}
-
 	netfs_collect_write_results(wreq);
 
 	/* We're done when the app thread has finished posting subreqs and all
 	 * the queues in all the streams are empty.
 	 */
-	if (!test_bit(NETFS_RREQ_ALL_QUEUED, &wreq->flags)) {
-		netfs_put_request(wreq, false, netfs_rreq_trace_put_work);
-		return;
-	}
+	if (!test_bit(NETFS_RREQ_ALL_QUEUED, &wreq->flags))
+		return false;
 	smp_rmb(); /* Read ALL_QUEUED before lists. */
 
 	transferred = LONG_MAX;
···
 		struct netfs_io_stream *stream = &wreq->io_streams[s];
 		if (!stream->active)
 			continue;
-		if (!list_empty(&stream->subrequests)) {
-			netfs_put_request(wreq, false, netfs_rreq_trace_put_work);
-			return;
-		}
+		if (!list_empty(&stream->subrequests))
+			return false;
 		if (stream->transferred < transferred)
 			transferred = stream->transferred;
 	}
···
 		inode_dio_end(wreq->inode);
 
 	_debug("finished");
-	trace_netfs_rreq(wreq, netfs_rreq_trace_wake_ip);
-	clear_and_wake_up_bit(NETFS_RREQ_IN_PROGRESS, &wreq->flags);
+	netfs_wake_rreq_flag(wreq, NETFS_RREQ_IN_PROGRESS, netfs_rreq_trace_wake_ip);
+	/* As we cleared NETFS_RREQ_IN_PROGRESS, we acquired its ref. */
 
 	if (wreq->iocb) {
 		size_t written = min(wreq->transferred, wreq->len);
···
 		wreq->iocb = VFS_PTR_POISON;
 	}
 
-	netfs_clear_subrequests(wreq, false);
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_work_complete);
+	netfs_clear_subrequests(wreq);
+	return true;
 }
 
-/*
- * Wake the collection work item.
- */
-void netfs_wake_write_collector(struct netfs_io_request *wreq, bool was_async)
+void netfs_write_collection_worker(struct work_struct *work)
 {
-	if (!work_pending(&wreq->work)) {
-		netfs_get_request(wreq, netfs_rreq_trace_get_work);
-		if (!queue_work(system_unbound_wq, &wreq->work))
-			netfs_put_request(wreq, was_async, netfs_rreq_trace_put_work_nq);
+	struct netfs_io_request *rreq = container_of(work, struct netfs_io_request, work);
+
+	netfs_see_request(rreq, netfs_rreq_trace_see_work);
+	if (test_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags)) {
+		if (netfs_write_collection(rreq))
+			/* Drop the ref from the IN_PROGRESS flag. */
+			netfs_put_request(rreq, netfs_rreq_trace_put_work_ip);
+		else
+			netfs_see_request(rreq, netfs_rreq_trace_see_work_complete);
 	}
 }
 
···
  * netfs_write_subrequest_terminated - Note the termination of a write operation.
  * @_op: The I/O request that has terminated.
  * @transferred_or_error: The amount of data transferred or an error code.
- * @was_async: The termination was asynchronous
  *
  * This tells the library that a contributory write I/O operation has
  * terminated, one way or another, and that it should collect the results.
···
  * negative error code.  The library will look after reissuing I/O operations
  * as appropriate and writing downloaded data to the cache.
  *
- * If @was_async is true, the caller might be running in softirq or interrupt
- * context and we can't sleep.
- *
  * When this is called, ownership of the subrequest is transferred back to the
  * library, along with a ref.
  *
  * Note that %_op is a void* so that the function can be passed to
  * kiocb::term_func without the need for a casting wrapper.
  */
-void netfs_write_subrequest_terminated(void *_op, ssize_t transferred_or_error,
-				       bool was_async)
+void netfs_write_subrequest_terminated(void *_op, ssize_t transferred_or_error)
 {
 	struct netfs_io_subrequest *subreq = _op;
 	struct netfs_io_request *wreq = subreq->rreq;
-	struct netfs_io_stream *stream = &wreq->io_streams[subreq->stream_nr];
 
 	_enter("%x[%x] %zd", wreq->debug_id, subreq->debug_index, transferred_or_error);
 
···
 	}
 
 	trace_netfs_sreq(subreq, netfs_sreq_trace_terminated);
-
-	clear_and_wake_up_bit(NETFS_SREQ_IN_PROGRESS, &subreq->flags);
-
-	/* If we are at the head of the queue, wake up the collector,
-	 * transferring a ref to it if we were the ones to do so.
-	 */
-	if (list_is_first(&subreq->rreq_link, &stream->subrequests))
-		netfs_wake_write_collector(wreq, was_async);
-
-	netfs_put_subrequest(subreq, was_async, netfs_sreq_trace_put_terminated);
+	netfs_subreq_clear_in_progress(subreq);
+	netfs_put_subrequest(subreq, netfs_sreq_trace_put_terminated);
 }
 EXPORT_SYMBOL(netfs_write_subrequest_terminated);

fs/netfs/write_issue.c

···
 	return wreq;
 nomem:
 	wreq->error = -ENOMEM;
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_failed);
+	netfs_put_request(wreq, netfs_rreq_trace_put_failed);
 	return ERR_PTR(-ENOMEM);
 }
 
···
 	_enter("R=%x[%x],%zx", wreq->debug_id, subreq->debug_index, subreq->len);
 
 	if (test_bit(NETFS_SREQ_FAILED, &subreq->flags))
-		return netfs_write_subrequest_terminated(subreq, subreq->error, false);
+		return netfs_write_subrequest_terminated(subreq, subreq->error);
 
 	trace_netfs_sreq(subreq, netfs_sreq_trace_submit);
 	stream->issue_write(subreq);
···
 	}
 
 	if (needs_poke)
-		netfs_wake_write_collector(wreq, false);
+		netfs_wake_collector(wreq);
 }
 
 /*
···
 		goto couldnt_start;
 	}
 
+	__set_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &wreq->flags);
 	trace_netfs_write(wreq, netfs_write_trace_writeback);
 	netfs_stat(&netfs_n_wh_writepages);
 
···
 	netfs_end_issue_write(wreq);
 
 	mutex_unlock(&ictx->wb_lock);
+	netfs_wake_collector(wreq);
 
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(wreq, netfs_rreq_trace_put_return);
 	_leave(" = %d", error);
 	return error;
 
···
 /*
  * End a write operation used when writing through the pagecache.
  */
-int netfs_end_writethrough(struct netfs_io_request *wreq, struct writeback_control *wbc,
-			   struct folio *writethrough_cache)
+ssize_t netfs_end_writethrough(struct netfs_io_request *wreq, struct writeback_control *wbc,
+			       struct folio *writethrough_cache)
 {
 	struct netfs_inode *ictx = netfs_inode(wreq->inode);
-	int ret;
+	ssize_t ret;
 
 	_enter("R=%x", wreq->debug_id);
 
···
 
 	mutex_unlock(&ictx->wb_lock);
 
-	if (wreq->iocb) {
+	if (wreq->iocb)
 		ret = -EIOCBQUEUED;
-	} else {
-		wait_on_bit(&wreq->flags, NETFS_RREQ_IN_PROGRESS, TASK_UNINTERRUPTIBLE);
-		ret = wreq->error;
-	}
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_return);
+	else
+		ret = netfs_wait_for_write(wreq);
+	netfs_put_request(wreq, netfs_rreq_trace_put_return);
 	return ret;
 }
 
···
 		start += part;
 		len -= part;
 		rolling_buffer_advance(&wreq->buffer, part);
-		if (test_bit(NETFS_RREQ_PAUSE, &wreq->flags)) {
-			trace_netfs_rreq(wreq, netfs_rreq_trace_wait_pause);
-			wait_event(wreq->waitq, !test_bit(NETFS_RREQ_PAUSE, &wreq->flags));
-		}
+		if (test_bit(NETFS_RREQ_PAUSE, &wreq->flags))
+			netfs_wait_for_paused_write(wreq);
 		if (test_bit(NETFS_RREQ_FAILED, &wreq->flags))
 			break;
 	}
···
 		goto couldnt_start;
 	}
 
-	trace_netfs_write(wreq, netfs_write_trace_writeback);
+	__set_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &wreq->flags);
+	trace_netfs_write(wreq, netfs_write_trace_writeback_single);
 	netfs_stat(&netfs_n_wh_writepages);
 
 	if (__test_and_set_bit(NETFS_RREQ_UPLOAD_TO_SERVER, &wreq->flags))
···
 	set_bit(NETFS_RREQ_ALL_QUEUED, &wreq->flags);
 
 	mutex_unlock(&ictx->wb_lock);
+	netfs_wake_collector(wreq);
 
-	netfs_put_request(wreq, false, netfs_rreq_trace_put_return);
+	netfs_put_request(wreq, netfs_rreq_trace_put_return);
 	_leave(" = %d", ret);
 	return ret;
 

fs/netfs/write_retry.c

···
 			if (test_bit(NETFS_SREQ_FAILED, &subreq->flags))
 				break;
 			if (__test_and_clear_bit(NETFS_SREQ_NEED_RETRY, &subreq->flags)) {
-				struct iov_iter source = subreq->io_iter;
+				struct iov_iter source;
 
-				iov_iter_revert(&source, subreq->len - source.count);
+				netfs_reset_iter(subreq);
+				source = subreq->io_iter;
 				netfs_get_subrequest(subreq, netfs_sreq_trace_get_resubmit);
 				netfs_reissue_write(stream, subreq, &source);
 			}
···
 						      &stream->subrequests, rreq_link) {
 				trace_netfs_sreq(subreq, netfs_sreq_trace_discard);
 				list_del(&subreq->rreq_link);
-				netfs_put_subrequest(subreq, false, netfs_sreq_trace_put_done);
+				netfs_put_subrequest(subreq, netfs_sreq_trace_put_done);
 				if (subreq == to)
 					break;
 			}
···
  */
 void netfs_retry_writes(struct netfs_io_request *wreq)
 {
-	struct netfs_io_subrequest *subreq;
 	struct netfs_io_stream *stream;
 	int s;
 
···
 	/* Wait for all outstanding I/O to quiesce before performing retries as
 	 * we may need to renegotiate the I/O sizes.
 	 */
+	set_bit(NETFS_RREQ_RETRYING, &wreq->flags);
 	for (s = 0; s < NR_IO_STREAMS; s++) {
 		stream = &wreq->io_streams[s];
-		if (!stream->active)
-			continue;
-
-		list_for_each_entry(subreq, &stream->subrequests, rreq_link) {
-			wait_on_bit(&subreq->flags, NETFS_SREQ_IN_PROGRESS,
-				    TASK_UNINTERRUPTIBLE);
-		}
+		if (stream->active)
+			netfs_wait_for_in_progress_stream(wreq, stream);
 	}
+	clear_bit(NETFS_RREQ_RETRYING, &wreq->flags);
 
 	// TODO: Enc: Fetch changed partial pages
 	// TODO: Enc: Reencrypt content if needed.

fs/smb/client/cifsproto.h

···
 				   bool from_readdir);
 extern void cifs_update_eof(struct cifsInodeInfo *cifsi, loff_t offset,
 			    unsigned int bytes_written);
-void cifs_write_subrequest_terminated(struct cifs_io_subrequest *wdata, ssize_t result,
-				      bool was_async);
+void cifs_write_subrequest_terminated(struct cifs_io_subrequest *wdata, ssize_t result);
 extern struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *, int);
 extern int cifs_get_writable_file(struct cifsInodeInfo *cifs_inode,
 				  int flags,

fs/smb/client/cifssmb.c

···
 			      server->credits, server->in_flight,
 			      0, cifs_trace_rw_credits_write_response_clear);
 	wdata->credits.value = 0;
-	cifs_write_subrequest_terminated(wdata, result, true);
+	cifs_write_subrequest_terminated(wdata, result);
 	release_mid(mid);
 	trace_smb3_rw_credits(credits.rreq_debug_id, credits.rreq_debug_index, 0,
 			      server->credits, server->in_flight,
···
 out:
 	if (rc) {
 		add_credits_and_wake_if(wdata->server, &wdata->credits, 0);
-		cifs_write_subrequest_terminated(wdata, rc, false);
+		cifs_write_subrequest_terminated(wdata, rc);
 	}
 }
 

fs/smb/client/file.c

···
 	else
 		trace_netfs_sreq(subreq, netfs_sreq_trace_fail);
 	add_credits_and_wake_if(wdata->server, &wdata->credits, 0);
-	cifs_write_subrequest_terminated(wdata, rc, false);
+	cifs_write_subrequest_terminated(wdata, rc);
 	goto out;
 }
 
···
 	return rc;
 }
 
-void cifs_write_subrequest_terminated(struct cifs_io_subrequest *wdata, ssize_t result,
-				      bool was_async)
+void cifs_write_subrequest_terminated(struct cifs_io_subrequest *wdata, ssize_t result)
 {
 	struct netfs_io_request *wreq = wdata->rreq;
 	struct netfs_inode *ictx = netfs_inode(wreq->inode);
···
 			netfs_resize_file(ictx, wrend, true);
 	}
 
-	netfs_write_subrequest_terminated(&wdata->subreq, result, was_async);
+	netfs_write_subrequest_terminated(&wdata->subreq, result);
 }
 
 struct cifsFileInfo *find_readable_file(struct cifsInodeInfo *cifs_inode,

fs/smb/client/smb2pdu.c

···
 			      0, cifs_trace_rw_credits_write_response_clear);
 	wdata->credits.value = 0;
 	trace_netfs_sreq(&wdata->subreq, netfs_sreq_trace_io_progress);
-	cifs_write_subrequest_terminated(wdata, result ?: written, true);
+	cifs_write_subrequest_terminated(wdata, result ?: written);
 	release_mid(mid);
 	trace_smb3_rw_credits(rreq_debug_id, subreq_debug_index, 0,
 			      server->credits, server->in_flight,
···
 				      -(int)wdata->credits.value,
 				      cifs_trace_rw_credits_write_response_clear);
 		add_credits_and_wake_if(wdata->server, &wdata->credits, 0);
-		cifs_write_subrequest_terminated(wdata, rc, true);
+		cifs_write_subrequest_terminated(wdata, rc);
 	}
 }
 

include/linux/fscache.h

···
 					 term_func, term_func_priv,
 					 using_pgpriv2, caching);
 	else if (term_func)
-		term_func(term_func_priv, -ENOBUFS, false);
+		term_func(term_func_priv, -ENOBUFS);
 
 }
 

include/linux/netfs.h

···
 	NETFS_WRITE_TO_CACHE,
 } __mode(byte);
 
-typedef void (*netfs_io_terminated_t)(void *priv, ssize_t transferred_or_error,
-				      bool was_async);
+typedef void (*netfs_io_terminated_t)(void *priv, ssize_t transferred_or_error);
 
 /*
  * Per-inode context.  This wraps the VFS inode.
···
  */
 struct netfs_io_request {
 	union {
-		struct work_struct work;
+		struct work_struct cleanup_work; /* Deferred cleanup work */
 		struct rcu_head rcu;
 	};
+	struct work_struct	work;		/* Result collector work */
 	struct inode		*inode;		/* The file being accessed */
 	struct address_space	*mapping;	/* The mapping being accessed */
 	struct kiocb		*iocb;		/* AIO completion vector */
···
 #define NETFS_RREQ_OFFLOAD_COLLECTION	0	/* Offload collection to workqueue */
 #define NETFS_RREQ_NO_UNLOCK_FOLIO	2	/* Don't unlock no_unlock_folio on completion */
 #define NETFS_RREQ_FAILED		4	/* The request failed */
-#define NETFS_RREQ_IN_PROGRESS		5	/* Unlocked when the request completes */
+#define NETFS_RREQ_IN_PROGRESS		5	/* Unlocked when the request completes (has ref) */
 #define NETFS_RREQ_FOLIO_COPY_TO_CACHE	6	/* Copy current folio to cache from read */
 #define NETFS_RREQ_UPLOAD_TO_SERVER	8	/* Need to write to the server */
 #define NETFS_RREQ_PAUSE		11	/* Pause subrequest generation */
 #define NETFS_RREQ_USE_IO_ITER		12	/* Use ->io_iter rather than ->i_pages */
 #define NETFS_RREQ_ALL_QUEUED		13	/* All subreqs are now queued */
 #define NETFS_RREQ_RETRYING		14	/* Set if we're in the retry path */
+#define NETFS_RREQ_SHORT_TRANSFER	15	/* Set if we have a short transfer */
 #define NETFS_RREQ_USE_PGPRIV2		31	/* [DEPRECATED] Use PG_private_2 to mark
 						 * write to cache on read */
 	const struct netfs_request_ops *netfs_ops;
···
 void netfs_get_subrequest(struct netfs_io_subrequest *subreq,
 			  enum netfs_sreq_ref_trace what);
 void netfs_put_subrequest(struct netfs_io_subrequest *subreq,
-			  bool was_async, enum netfs_sreq_ref_trace what);
+			  enum netfs_sreq_ref_trace what);
 ssize_t netfs_extract_user_iter(struct iov_iter *orig, size_t orig_len,
 				struct iov_iter *new,
 				iov_iter_extraction_t extraction_flags);
 size_t netfs_limit_iter(const struct iov_iter *iter, size_t start_offset,
 			size_t max_size, size_t max_segs);
 void netfs_prepare_write_failed(struct netfs_io_subrequest *subreq);
-void netfs_write_subrequest_terminated(void *_op, ssize_t transferred_or_error,
-				       bool was_async);
+void netfs_write_subrequest_terminated(void *_op, ssize_t transferred_or_error);
 void netfs_queue_write_request(struct netfs_io_subrequest *subreq);
 
 int netfs_start_io_read(struct inode *inode);

include/trace/events/netfs.h

···
 	EM(netfs_write_trace_dio_write,		"DIO-WRITE")	\
 	EM(netfs_write_trace_unbuffered_write,	"UNB-WRITE")	\
 	EM(netfs_write_trace_writeback,		"WRITEBACK")	\
+	EM(netfs_write_trace_writeback_single,	"WB-SINGLE") \
 	E_(netfs_write_trace_writethrough,	"WRITETHRU")
 
 #define netfs_rreq_origins					\
···
 #define netfs_rreq_ref_traces					\
 	EM(netfs_rreq_trace_get_for_outstanding,"GET OUTSTND")	\
 	EM(netfs_rreq_trace_get_subreq,		"GET SUBREQ ")	\
-	EM(netfs_rreq_trace_get_work,		"GET WORK   ")	\
 	EM(netfs_rreq_trace_put_complete,	"PUT COMPLT ")	\
 	EM(netfs_rreq_trace_put_discard,	"PUT DISCARD")	\
 	EM(netfs_rreq_trace_put_failed,		"PUT FAILED ")	\
 	EM(netfs_rreq_trace_put_no_submit,	"PUT NO-SUBM")	\
 	EM(netfs_rreq_trace_put_return,		"PUT RETURN ")	\
 	EM(netfs_rreq_trace_put_subreq,		"PUT SUBREQ ")	\
-	EM(netfs_rreq_trace_put_work,		"PUT WORK   ")	\
-	EM(netfs_rreq_trace_put_work_complete,	"PUT WORK CP")	\
-	EM(netfs_rreq_trace_put_work_nq,	"PUT WORK NQ")	\
+	EM(netfs_rreq_trace_put_work_ip,	"PUT WORK IP ")	\
 	EM(netfs_rreq_trace_see_work,		"SEE WORK   ")	\
+	EM(netfs_rreq_trace_see_work_complete,	"SEE WORK CP")	\
 	E_(netfs_rreq_trace_new,		"NEW        ")
 
 #define netfs_sreq_ref_traces					\

net/9p/client.c

···
 				    start, len, &subreq->io_iter);
 	}
 	if (IS_ERR(req)) {
-		netfs_write_subrequest_terminated(subreq, PTR_ERR(req), false);
+		netfs_write_subrequest_terminated(subreq, PTR_ERR(req));
 		return;
 	}
 
···
 	if (err) {
 		trace_9p_protocol_dump(clnt, &req->rc);
 		p9_req_put(clnt, req);
-		netfs_write_subrequest_terminated(subreq, err, false);
+		netfs_write_subrequest_terminated(subreq, err);
 		return;
 	}
 
···
 	p9_debug(P9_DEBUG_9P, "<<< RWRITE count %d\n", len);
 
 	p9_req_put(clnt, req);
-	netfs_write_subrequest_terminated(subreq, written, false);
+	netfs_write_subrequest_terminated(subreq, written);
 }
 EXPORT_SYMBOL(p9_client_write_subreq);