From: Miklos Szeredi <mszeredi@redhat.com>
To: linux-fsdevel@vger.kernel.org
Cc: Peter-Jan Gootzen <pgootzen@nvidia.com>,
Jingbo Xu <jefflexu@linux.alibaba.com>,
Stefan Hajnoczi <stefanha@redhat.com>,
Yoray Zack <yorayz@nvidia.com>, Vivek Goyal <vgoyal@redhat.com>,
virtualization@lists.linux.dev
Subject: [PATCH] fuse: cleanup request queuing towards virtiofs
Date: Wed, 29 May 2024 17:52:07 +0200 [thread overview]
Message-ID: <20240529155210.2543295-1-mszeredi@redhat.com> (raw)
Virtiofs has its own queing mechanism, but still requests are first queued
on fiq->pending to be immediately dequeued and queued onto the virtio
queue.
The queuing on fiq->pending is unnecessary and might even have some
performance impact due to being a contention point.
Forget requests are handled similarly.
Move the queuing of requests and forgets into the fiq->ops->*.
fuse_iqueue_ops are renamed to reflect the new semantics.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
---
fs/fuse/dev.c | 159 ++++++++++++++++++++++++--------------------
fs/fuse/fuse_i.h | 19 ++----
fs/fuse/virtio_fs.c | 41 ++++--------
3 files changed, 106 insertions(+), 113 deletions(-)
diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index 9eb191b5c4de..a4f510f1b1a4 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -192,10 +192,22 @@ unsigned int fuse_len_args(unsigned int numargs, struct fuse_arg *args)
}
EXPORT_SYMBOL_GPL(fuse_len_args);
-u64 fuse_get_unique(struct fuse_iqueue *fiq)
+static u64 fuse_get_unique_locked(struct fuse_iqueue *fiq)
{
fiq->reqctr += FUSE_REQ_ID_STEP;
return fiq->reqctr;
+
+}
+
+u64 fuse_get_unique(struct fuse_iqueue *fiq)
+{
+ u64 ret;
+
+ spin_lock(&fiq->lock);
+ ret = fuse_get_unique_locked(fiq);
+ spin_unlock(&fiq->lock);
+
+ return ret;
}
EXPORT_SYMBOL_GPL(fuse_get_unique);
@@ -215,22 +227,67 @@ __releases(fiq->lock)
spin_unlock(&fiq->lock);
}
+static void fuse_dev_queue_forget(struct fuse_iqueue *fiq, struct fuse_forget_link *forget)
+{
+ spin_lock(&fiq->lock);
+ if (fiq->connected) {
+ fiq->forget_list_tail->next = forget;
+ fiq->forget_list_tail = forget;
+ fuse_dev_wake_and_unlock(fiq);
+ } else {
+ kfree(forget);
+ spin_unlock(&fiq->lock);
+ }
+}
+
+static void fuse_dev_queue_interrupt(struct fuse_iqueue *fiq, struct fuse_req *req)
+{
+ spin_lock(&fiq->lock);
+ if (list_empty(&req->intr_entry)) {
+ list_add_tail(&req->intr_entry, &fiq->interrupts);
+ /*
+ * Pairs with smp_mb() implied by test_and_set_bit()
+ * from fuse_request_end().
+ */
+ smp_mb();
+ if (test_bit(FR_FINISHED, &req->flags)) {
+ list_del_init(&req->intr_entry);
+ spin_unlock(&fiq->lock);
+ }
+ fuse_dev_wake_and_unlock(fiq);
+ } else {
+ spin_unlock(&fiq->lock);
+ }
+}
+
+static void fuse_dev_queue_req(struct fuse_iqueue *fiq, struct fuse_req *req)
+{
+ spin_lock(&fiq->lock);
+ if (fiq->connected) {
+ if (req->in.h.opcode != FUSE_NOTIFY_REPLY)
+ req->in.h.unique = fuse_get_unique_locked(fiq);
+ list_add_tail(&req->list, &fiq->pending);
+ fuse_dev_wake_and_unlock(fiq);
+ } else {
+ spin_unlock(&fiq->lock);
+ req->out.h.error = -ENOTCONN;
+ fuse_request_end(req);
+ }
+}
+
const struct fuse_iqueue_ops fuse_dev_fiq_ops = {
- .wake_forget_and_unlock = fuse_dev_wake_and_unlock,
- .wake_interrupt_and_unlock = fuse_dev_wake_and_unlock,
- .wake_pending_and_unlock = fuse_dev_wake_and_unlock,
+ .send_forget = fuse_dev_queue_forget,
+ .send_interrupt = fuse_dev_queue_interrupt,
+ .send_req = fuse_dev_queue_req,
};
EXPORT_SYMBOL_GPL(fuse_dev_fiq_ops);
-static void queue_request_and_unlock(struct fuse_iqueue *fiq,
- struct fuse_req *req)
-__releases(fiq->lock)
+static void fuse_send_one(struct fuse_iqueue *fiq, struct fuse_req *req)
{
req->in.h.len = sizeof(struct fuse_in_header) +
fuse_len_args(req->args->in_numargs,
(struct fuse_arg *) req->args->in_args);
- list_add_tail(&req->list, &fiq->pending);
- fiq->ops->wake_pending_and_unlock(fiq);
+ fiq->ops->send_req(fiq, req);
}
void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
@@ -241,15 +298,7 @@ void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
forget->forget_one.nodeid = nodeid;
forget->forget_one.nlookup = nlookup;
- spin_lock(&fiq->lock);
- if (fiq->connected) {
- fiq->forget_list_tail->next = forget;
- fiq->forget_list_tail = forget;
- fiq->ops->wake_forget_and_unlock(fiq);
- } else {
- kfree(forget);
- spin_unlock(&fiq->lock);
- }
+ fiq->ops->send_forget(fiq, forget);
}
static void flush_bg_queue(struct fuse_conn *fc)
@@ -263,9 +312,7 @@ static void flush_bg_queue(struct fuse_conn *fc)
req = list_first_entry(&fc->bg_queue, struct fuse_req, list);
list_del(&req->list);
fc->active_background++;
- spin_lock(&fiq->lock);
- req->in.h.unique = fuse_get_unique(fiq);
- queue_request_and_unlock(fiq, req);
+ fuse_send_one(fiq, req);
}
}
@@ -335,29 +382,12 @@ static int queue_interrupt(struct fuse_req *req)
{
struct fuse_iqueue *fiq = &req->fm->fc->iq;
- spin_lock(&fiq->lock);
/* Check for we've sent request to interrupt this req */
- if (unlikely(!test_bit(FR_INTERRUPTED, &req->flags))) {
- spin_unlock(&fiq->lock);
+ if (unlikely(!test_bit(FR_INTERRUPTED, &req->flags)))
return -EINVAL;
- }
- if (list_empty(&req->intr_entry)) {
- list_add_tail(&req->intr_entry, &fiq->interrupts);
- /*
- * Pairs with smp_mb() implied by test_and_set_bit()
- * from fuse_request_end().
- */
- smp_mb();
- if (test_bit(FR_FINISHED, &req->flags)) {
- list_del_init(&req->intr_entry);
- spin_unlock(&fiq->lock);
- return 0;
- }
- fiq->ops->wake_interrupt_and_unlock(fiq);
- } else {
- spin_unlock(&fiq->lock);
- }
+ fiq->ops->send_interrupt(fiq, req);
+
return 0;
}
@@ -412,21 +442,15 @@ static void __fuse_request_send(struct fuse_req *req)
struct fuse_iqueue *fiq = &req->fm->fc->iq;
BUG_ON(test_bit(FR_BACKGROUND, &req->flags));
- spin_lock(&fiq->lock);
- if (!fiq->connected) {
- spin_unlock(&fiq->lock);
- req->out.h.error = -ENOTCONN;
- } else {
- req->in.h.unique = fuse_get_unique(fiq);
- /* acquire extra reference, since request is still needed
- after fuse_request_end() */
- __fuse_get_request(req);
- queue_request_and_unlock(fiq, req);
- request_wait_answer(req);
- /* Pairs with smp_wmb() in fuse_request_end() */
- smp_rmb();
- }
+ /* acquire extra reference, since request is still needed after
+ fuse_request_end() */
+ __fuse_get_request(req);
+ fuse_send_one(fiq, req);
+
+ request_wait_answer(req);
+ /* Pairs with smp_wmb() in fuse_request_end() */
+ smp_rmb();
}
static void fuse_adjust_compat(struct fuse_conn *fc, struct fuse_args *args)
@@ -581,7 +605,6 @@ static int fuse_simple_notify_reply(struct fuse_mount *fm,
{
struct fuse_req *req;
struct fuse_iqueue *fiq = &fm->fc->iq;
- int err = 0;
req = fuse_get_req(fm, false);
if (IS_ERR(req))
@@ -592,16 +615,9 @@ static int fuse_simple_notify_reply(struct fuse_mount *fm,
fuse_args_to_req(req, args);
- spin_lock(&fiq->lock);
- if (fiq->connected) {
- queue_request_and_unlock(fiq, req);
- } else {
- err = -ENODEV;
- spin_unlock(&fiq->lock);
- fuse_put_request(req);
- }
+ fuse_send_one(fiq, req);
- return err;
+ return 0;
}
/*
@@ -1076,9 +1092,9 @@ __releases(fiq->lock)
return err ? err : reqsize;
}
-struct fuse_forget_link *fuse_dequeue_forget(struct fuse_iqueue *fiq,
- unsigned int max,
- unsigned int *countp)
+static struct fuse_forget_link *fuse_dequeue_forget(struct fuse_iqueue *fiq,
+ unsigned int max,
+ unsigned int *countp)
{
struct fuse_forget_link *head = fiq->forget_list_head.next;
struct fuse_forget_link **newhead = &head;
@@ -1097,7 +1113,6 @@ struct fuse_forget_link *fuse_dequeue_forget(struct fuse_iqueue *fiq,
return head;
}
-EXPORT_SYMBOL(fuse_dequeue_forget);
static int fuse_read_single_forget(struct fuse_iqueue *fiq,
struct fuse_copy_state *cs,
@@ -1112,7 +1127,7 @@ __releases(fiq->lock)
struct fuse_in_header ih = {
.opcode = FUSE_FORGET,
.nodeid = forget->forget_one.nodeid,
- .unique = fuse_get_unique(fiq),
+ .unique = fuse_get_unique_locked(fiq),
.len = sizeof(ih) + sizeof(arg),
};
@@ -1143,7 +1158,7 @@ __releases(fiq->lock)
struct fuse_batch_forget_in arg = { .count = 0 };
struct fuse_in_header ih = {
.opcode = FUSE_BATCH_FORGET,
- .unique = fuse_get_unique(fiq),
+ .unique = fuse_get_unique_locked(fiq),
.len = sizeof(ih) + sizeof(arg),
};
@@ -1822,7 +1837,7 @@ static void fuse_resend(struct fuse_conn *fc)
spin_lock(&fiq->lock);
/* iq and pq requests are both oldest to newest */
list_splice(&to_queue, &fiq->pending);
- fiq->ops->wake_pending_and_unlock(fiq);
+ fuse_dev_wake_and_unlock(fiq);
}
static int fuse_notify_resend(struct fuse_conn *fc)
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index f23919610313..33b21255817e 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -449,22 +449,19 @@ struct fuse_iqueue;
*/
struct fuse_iqueue_ops {
/**
- * Signal that a forget has been queued
+ * Send one forget
*/
- void (*wake_forget_and_unlock)(struct fuse_iqueue *fiq)
- __releases(fiq->lock);
+ void (*send_forget)(struct fuse_iqueue *fiq, struct fuse_forget_link *link);
/**
- * Signal that an INTERRUPT request has been queued
+ * Send interrupt for request
*/
- void (*wake_interrupt_and_unlock)(struct fuse_iqueue *fiq)
- __releases(fiq->lock);
+ void (*send_interrupt)(struct fuse_iqueue *fiq, struct fuse_req *req);
/**
- * Signal that a request has been queued
+ * Send one request
*/
- void (*wake_pending_and_unlock)(struct fuse_iqueue *fiq)
- __releases(fiq->lock);
+ void (*send_req)(struct fuse_iqueue *fiq, struct fuse_req *req);
/**
* Clean up when fuse_iqueue is destroyed
@@ -1053,10 +1050,6 @@ void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
struct fuse_forget_link *fuse_alloc_forget(void);
-struct fuse_forget_link *fuse_dequeue_forget(struct fuse_iqueue *fiq,
- unsigned int max,
- unsigned int *countp);
-
/*
* Initialize READ or READDIR request
*/
diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c
index 1a52a51b6b07..690e508dbc4d 100644
--- a/fs/fuse/virtio_fs.c
+++ b/fs/fuse/virtio_fs.c
@@ -1089,22 +1089,13 @@ static struct virtio_driver virtio_fs_driver = {
#endif
};
-static void virtio_fs_wake_forget_and_unlock(struct fuse_iqueue *fiq)
-__releases(fiq->lock)
+static void virtio_fs_send_forget(struct fuse_iqueue *fiq, struct fuse_forget_link *link)
{
- struct fuse_forget_link *link;
struct virtio_fs_forget *forget;
struct virtio_fs_forget_req *req;
- struct virtio_fs *fs;
- struct virtio_fs_vq *fsvq;
- u64 unique;
-
- link = fuse_dequeue_forget(fiq, 1, NULL);
- unique = fuse_get_unique(fiq);
-
- fs = fiq->priv;
- fsvq = &fs->vqs[VQ_HIPRIO];
- spin_unlock(&fiq->lock);
+ struct virtio_fs *fs = fiq->priv;
+ struct virtio_fs_vq *fsvq = &fs->vqs[VQ_HIPRIO];
+ u64 unique = fuse_get_unique(fiq);
/* Allocate a buffer for the request */
forget = kmalloc(sizeof(*forget), GFP_NOFS | __GFP_NOFAIL);
@@ -1124,8 +1115,7 @@ __releases(fiq->lock)
kfree(link);
}
-static void virtio_fs_wake_interrupt_and_unlock(struct fuse_iqueue *fiq)
-__releases(fiq->lock)
+static void virtio_fs_send_interrupt(struct fuse_iqueue *fiq, struct fuse_req *req)
{
/*
* TODO interrupts.
@@ -1134,7 +1124,6 @@ __releases(fiq->lock)
* Exceptions are blocking lock operations; for example fcntl(F_SETLKW)
* with shared lock between host and guest.
*/
- spin_unlock(&fiq->lock);
}
/* Count number of scatter-gather elements required */
@@ -1339,21 +1328,17 @@ static int virtio_fs_enqueue_req(struct virtio_fs_vq *fsvq,
return ret;
}
-static void virtio_fs_wake_pending_and_unlock(struct fuse_iqueue *fiq)
-__releases(fiq->lock)
+static void virtio_fs_send_req(struct fuse_iqueue *fiq, struct fuse_req *req)
{
unsigned int queue_id;
struct virtio_fs *fs;
- struct fuse_req *req;
struct virtio_fs_vq *fsvq;
int ret;
- WARN_ON(list_empty(&fiq->pending));
- req = list_last_entry(&fiq->pending, struct fuse_req, list);
+ if (req->in.h.opcode != FUSE_NOTIFY_REPLY)
+ req->in.h.unique = fuse_get_unique(fiq);
+
clear_bit(FR_PENDING, &req->flags);
- list_del_init(&req->list);
- WARN_ON(!list_empty(&fiq->pending));
- spin_unlock(&fiq->lock);
fs = fiq->priv;
queue_id = VQ_REQUEST + fs->mq_map[raw_smp_processor_id()];
@@ -1393,10 +1378,10 @@ __releases(fiq->lock)
}
static const struct fuse_iqueue_ops virtio_fs_fiq_ops = {
- .wake_forget_and_unlock = virtio_fs_wake_forget_and_unlock,
- .wake_interrupt_and_unlock = virtio_fs_wake_interrupt_and_unlock,
- .wake_pending_and_unlock = virtio_fs_wake_pending_and_unlock,
- .release = virtio_fs_fiq_release,
+ .send_forget = virtio_fs_send_forget,
+ .send_interrupt = virtio_fs_send_interrupt,
+ .send_req = virtio_fs_send_req,
+ .release = virtio_fs_fiq_release,
};
static inline void virtio_fs_ctx_set_defaults(struct fuse_fs_context *ctx)
--
2.45.1
next reply other threads:[~2024-05-29 15:52 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-05-29 15:52 Miklos Szeredi [this message]
2024-05-29 18:32 ` [PATCH] fuse: cleanup request queuing towards virtiofs Stefan Hajnoczi
2024-05-30 9:06 ` Miklos Szeredi
2024-05-30 13:38 ` Peter-Jan Gootzen
2024-06-05 10:40 ` Peter-Jan Gootzen
2024-06-05 11:04 ` Stefan Hajnoczi
2024-05-30 3:20 ` Jingbo Xu
2024-05-30 9:00 ` Miklos Szeredi
2024-05-30 15:36 ` Jingbo Xu
2024-05-30 17:07 ` Bernd Schubert
2024-09-23 10:33 ` Lai, Yi
2024-09-23 22:48 ` Joanne Koong
2024-09-23 23:47 ` Joanne Koong
2024-09-24 8:58 ` Miklos Szeredi
2024-09-24 9:52 ` Lai, Yi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240529155210.2543295-1-mszeredi@redhat.com \
--to=mszeredi@redhat.com \
--cc=jefflexu@linux.alibaba.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=pgootzen@nvidia.com \
--cc=stefanha@redhat.com \
--cc=vgoyal@redhat.com \
--cc=virtualization@lists.linux.dev \
--cc=yorayz@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).