linux-next/fs/netfs/objects.c
David Howells b4f239c91f
netfs: Change the read result collector to only use one work item
Change the way netfslib collects read results to do all the collection for
a particular read request using a single work item that walks along the
subrequest queue as subrequests make progress or complete, unlocking folios
progressively rather than doing the unlock in parallel as parallel requests
come in.

The code is remodelled to be more like the write-side code, though only
using a single stream.  This makes it more directly comparable and thus
easier to duplicate fixes between the two sides.

This has a number of advantages:

 (1) It's simpler.  There doesn't need to be a complex donation mechanism
     to handle mismatches between the size and alignment of subrequests and
     folios.  The collector unlocks folios as the subrequests covering each
     complete.

 (2) It should cause less scheduler overhead as there's a single work item
     in play unlocking pages in parallel when a read gets split up into a
     lot of subrequests instead of one per subrequest.

     Whilst the parallellism is nice in theory, in practice, the vast
     majority of loads are sequential reads of the whole file, so
     committing a bunch of threads to unlocking folios out of order doesn't
     help in those cases.

 (3) It should make it easier to implement content decryption.  A folio
     cannot be decrypted until all the requests that contribute to it have
     completed - and, again, most loads are sequential and so, most of the
     time, we want to begin decryption sequentially (though it's great if
     the decryption can happen in parallel).

There is a disadvantage in that we're losing the ability to decrypt and
unlock things on an as-things-arrive basis which may affect some
applications.

Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://lore.kernel.org/r/20241108173236.1382366-29-dhowells@redhat.com
cc: Jeff Layton <jlayton@kernel.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
2024-12-02 11:21:23 +01:00

235 lines
6.5 KiB
C

// SPDX-License-Identifier: GPL-2.0-only
/* Object lifetime handling and tracing.
*
* Copyright (C) 2022 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*/
#include <linux/slab.h>
#include <linux/mempool.h>
#include <linux/delay.h>
#include "internal.h"
/*
* Allocate an I/O request and initialise it.
*/
struct netfs_io_request *netfs_alloc_request(struct address_space *mapping,
struct file *file,
loff_t start, size_t len,
enum netfs_io_origin origin)
{
static atomic_t debug_ids;
struct inode *inode = file ? file_inode(file) : mapping->host;
struct netfs_inode *ctx = netfs_inode(inode);
struct netfs_io_request *rreq;
mempool_t *mempool = ctx->ops->request_pool ?: &netfs_request_pool;
struct kmem_cache *cache = mempool->pool_data;
int ret;
for (;;) {
rreq = mempool_alloc(mempool, GFP_KERNEL);
if (rreq)
break;
msleep(10);
}
memset(rreq, 0, kmem_cache_size(cache));
rreq->start = start;
rreq->len = len;
rreq->origin = origin;
rreq->netfs_ops = ctx->ops;
rreq->mapping = mapping;
rreq->inode = inode;
rreq->i_size = i_size_read(inode);
rreq->debug_id = atomic_inc_return(&debug_ids);
rreq->wsize = INT_MAX;
rreq->io_streams[0].sreq_max_len = ULONG_MAX;
rreq->io_streams[0].sreq_max_segs = 0;
spin_lock_init(&rreq->lock);
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);
if (origin == NETFS_READAHEAD ||
origin == NETFS_READPAGE ||
origin == NETFS_READ_GAPS ||
origin == NETFS_READ_SINGLE ||
origin == NETFS_READ_FOR_WRITE ||
origin == NETFS_DIO_READ) {
INIT_WORK(&rreq->work, netfs_read_collection_worker);
rreq->io_streams[0].avail = true;
} else {
INIT_WORK(&rreq->work, netfs_write_collection_worker);
}
__set_bit(NETFS_RREQ_IN_PROGRESS, &rreq->flags);
if (file && file->f_flags & O_NONBLOCK)
__set_bit(NETFS_RREQ_NONBLOCK, &rreq->flags);
if (rreq->netfs_ops->init_request) {
ret = rreq->netfs_ops->init_request(rreq, file);
if (ret < 0) {
mempool_free(rreq, rreq->netfs_ops->request_pool ?: &netfs_request_pool);
return ERR_PTR(ret);
}
}
atomic_inc(&ctx->io_count);
trace_netfs_rreq_ref(rreq->debug_id, 1, netfs_rreq_trace_new);
netfs_proc_add_rreq(rreq);
netfs_stat(&netfs_n_rh_rreq);
return rreq;
}
void netfs_get_request(struct netfs_io_request *rreq, enum netfs_rreq_ref_trace what)
{
int r;
__refcount_inc(&rreq->ref, &r);
trace_netfs_rreq_ref(rreq->debug_id, r + 1, what);
}
void netfs_clear_subrequests(struct netfs_io_request *rreq, bool was_async)
{
struct netfs_io_subrequest *subreq;
struct netfs_io_stream *stream;
int s;
for (s = 0; s < ARRAY_SIZE(rreq->io_streams); s++) {
stream = &rreq->io_streams[s];
while (!list_empty(&stream->subrequests)) {
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);
}
}
}
static void netfs_free_request_rcu(struct rcu_head *rcu)
{
struct netfs_io_request *rreq = container_of(rcu, struct netfs_io_request, rcu);
mempool_free(rreq, rreq->netfs_ops->request_pool ?: &netfs_request_pool);
netfs_stat_d(&netfs_n_rh_rreq);
}
static void netfs_free_request(struct work_struct *work)
{
struct netfs_io_request *rreq =
container_of(work, struct netfs_io_request, work);
struct netfs_inode *ictx = netfs_inode(rreq->inode);
unsigned int i;
trace_netfs_rreq(rreq, netfs_rreq_trace_free);
netfs_proc_del_rreq(rreq);
netfs_clear_subrequests(rreq, false);
if (rreq->netfs_ops->free_request)
rreq->netfs_ops->free_request(rreq);
if (rreq->cache_resources.ops)
rreq->cache_resources.ops->end_operation(&rreq->cache_resources);
if (rreq->direct_bv) {
for (i = 0; i < rreq->direct_bv_count; i++) {
if (rreq->direct_bv[i].bv_page) {
if (rreq->direct_bv_unpin)
unpin_user_page(rreq->direct_bv[i].bv_page);
}
}
kvfree(rreq->direct_bv);
}
rolling_buffer_clear(&rreq->buffer);
if (atomic_dec_and_test(&ictx->io_count))
wake_up_var(&ictx->io_count);
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)
{
unsigned int debug_id;
bool dead;
int r;
if (rreq) {
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);
}
}
}
}
/*
* Allocate and partially initialise an I/O request structure.
*/
struct netfs_io_subrequest *netfs_alloc_subrequest(struct netfs_io_request *rreq)
{
struct netfs_io_subrequest *subreq;
mempool_t *mempool = rreq->netfs_ops->subrequest_pool ?: &netfs_subrequest_pool;
struct kmem_cache *cache = mempool->pool_data;
for (;;) {
subreq = mempool_alloc(rreq->netfs_ops->subrequest_pool ?: &netfs_subrequest_pool,
GFP_KERNEL);
if (subreq)
break;
msleep(10);
}
memset(subreq, 0, kmem_cache_size(cache));
INIT_WORK(&subreq->work, NULL);
INIT_LIST_HEAD(&subreq->rreq_link);
refcount_set(&subreq->ref, 2);
subreq->rreq = rreq;
subreq->debug_index = atomic_inc_return(&rreq->subreq_counter);
netfs_get_request(rreq, netfs_rreq_trace_get_subreq);
netfs_stat(&netfs_n_rh_sreq);
return subreq;
}
void netfs_get_subrequest(struct netfs_io_subrequest *subreq,
enum netfs_sreq_ref_trace what)
{
int r;
__refcount_inc(&subreq->ref, &r);
trace_netfs_sreq_ref(subreq->rreq->debug_id, subreq->debug_index, r + 1,
what);
}
static void netfs_free_subrequest(struct netfs_io_subrequest *subreq,
bool was_async)
{
struct netfs_io_request *rreq = subreq->rreq;
trace_netfs_sreq(subreq, netfs_sreq_trace_free);
if (rreq->netfs_ops->free_subrequest)
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);
}
void netfs_put_subrequest(struct netfs_io_subrequest *subreq, bool was_async,
enum netfs_sreq_ref_trace what)
{
unsigned int debug_index = subreq->debug_index;
unsigned int debug_id = subreq->rreq->debug_id;
bool dead;
int r;
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);
}