From: Bernd Schubert <bernd@bsbernd.com>
To: Joanne Koong <joannelkoong@gmail.com>,
miklos@szeredi.hu, linux-fsdevel@vger.kernel.org
Cc: kernel-team@meta.com
Subject: Re: [PATCH v2] fuse: optimize over-io-uring request expiration check
Date: Mon, 17 Feb 2025 00:16:18 +0100 [thread overview]
Message-ID: <bdb3980b-65e3-4bd5-aa5b-0a48d6d6e7a0@bsbernd.com> (raw)
In-Reply-To: <CAJnrk1a9zdW2GfcEHmM=QMouMV8m_huUzZao+SsMgtK7Anx=BQ@mail.gmail.com>
On 2/3/25 20:37, Joanne Koong wrote:
> On Mon, Feb 3, 2025 at 11:30 AM Joanne Koong <joannelkoong@gmail.com> wrote:
>>
>> Currently, when checking whether a request has timed out, we check
>> fpq processing, but fuse-over-io-uring has one fpq per core and 256
>> entries in the processing table. For systems where there are a
>> large number of cores, this may be too much overhead.
>>
>> Instead of checking the fpq processing list, check ent_w_req_queue
>> and ent_in_userspace.
>>
>> Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
>> ---
>> fs/fuse/dev.c | 2 +-
>> fs/fuse/dev_uring.c | 26 +++++++++++++++++++++-----
>> fs/fuse/fuse_dev_i.h | 1 -
>> 3 files changed, 22 insertions(+), 7 deletions(-)
>>
>
> v1: https://lore.kernel.org/linux-fsdevel/20250123235251.1139078-1-joannelkoong@gmail.com/
> Changes from v1 -> v2:
> * Remove commit queue check, which should be fine since if the request
> has expired while on this queue, it will be shortly processed anyways
>
>> diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
>> index 3c03aac480a4..80a11ef4b69a 100644
>> --- a/fs/fuse/dev.c
>> +++ b/fs/fuse/dev.c
>> @@ -45,7 +45,7 @@ bool fuse_request_expired(struct fuse_conn *fc, struct list_head *list)
>> return time_is_before_jiffies(req->create_time + fc->timeout.req_timeout);
>> }
>>
>> -bool fuse_fpq_processing_expired(struct fuse_conn *fc, struct list_head *processing)
>> +static bool fuse_fpq_processing_expired(struct fuse_conn *fc, struct list_head *processing)
>> {
>> int i;
>>
>> diff --git a/fs/fuse/dev_uring.c b/fs/fuse/dev_uring.c
>> index ab8c26042aa8..50f5b4e32ed5 100644
>> --- a/fs/fuse/dev_uring.c
>> +++ b/fs/fuse/dev_uring.c
>> @@ -140,6 +140,21 @@ void fuse_uring_abort_end_requests(struct fuse_ring *ring)
>> }
>> }
>>
>> +static bool ent_list_request_expired(struct fuse_conn *fc, struct list_head *list)
>> +{
>> + struct fuse_ring_ent *ent;
>> + struct fuse_req *req;
>> +
>> + ent = list_first_entry_or_null(list, struct fuse_ring_ent, list);
>> + if (!ent)
>> + return false;
>> +
>> + req = ent->fuse_req;
>> +
>> + return time_is_before_jiffies(req->create_time +
>> + fc->timeout.req_timeout);
>> +}
>> +
>> bool fuse_uring_request_expired(struct fuse_conn *fc)
>> {
>> struct fuse_ring *ring = fc->ring;
>> @@ -157,7 +172,8 @@ bool fuse_uring_request_expired(struct fuse_conn *fc)
>> spin_lock(&queue->lock);
>> if (fuse_request_expired(fc, &queue->fuse_req_queue) ||
>> fuse_request_expired(fc, &queue->fuse_req_bg_queue) ||
>> - fuse_fpq_processing_expired(fc, queue->fpq.processing)) {
>> + ent_list_request_expired(fc, &queue->ent_w_req_queue) ||
>> + ent_list_request_expired(fc, &queue->ent_in_userspace)) {
>> spin_unlock(&queue->lock);
>> return true;
>> }
>> @@ -495,7 +511,7 @@ static void fuse_uring_cancel(struct io_uring_cmd *cmd,
>> spin_lock(&queue->lock);
>> if (ent->state == FRRS_AVAILABLE) {
>> ent->state = FRRS_USERSPACE;
>> - list_move(&ent->list, &queue->ent_in_userspace);
>> + list_move_tail(&ent->list, &queue->ent_in_userspace);
>> need_cmd_done = true;
>> ent->cmd = NULL;
>> }
>> @@ -715,7 +731,7 @@ static int fuse_uring_send_next_to_ring(struct fuse_ring_ent *ent,
>> cmd = ent->cmd;
>> ent->cmd = NULL;
>> ent->state = FRRS_USERSPACE;
>> - list_move(&ent->list, &queue->ent_in_userspace);
>> + list_move_tail(&ent->list, &queue->ent_in_userspace);
>> spin_unlock(&queue->lock);
>>
>> io_uring_cmd_done(cmd, 0, 0, issue_flags);
>> @@ -769,7 +785,7 @@ static void fuse_uring_add_req_to_ring_ent(struct fuse_ring_ent *ent,
>> spin_unlock(&fiq->lock);
>> ent->fuse_req = req;
>> ent->state = FRRS_FUSE_REQ;
>> - list_move(&ent->list, &queue->ent_w_req_queue);
>> + list_move_tail(&ent->list, &queue->ent_w_req_queue);
>> fuse_uring_add_to_pq(ent, req);
>> }
>>
>> @@ -1185,7 +1201,7 @@ static void fuse_uring_send(struct fuse_ring_ent *ent, struct io_uring_cmd *cmd,
>>
>> spin_lock(&queue->lock);
>> ent->state = FRRS_USERSPACE;
>> - list_move(&ent->list, &queue->ent_in_userspace);
>> + list_move_tail(&ent->list, &queue->ent_in_userspace);
>> ent->cmd = NULL;
>> spin_unlock(&queue->lock);
>>
>> diff --git a/fs/fuse/fuse_dev_i.h b/fs/fuse/fuse_dev_i.h
>> index 3c4ae4d52b6f..19c29c6000a7 100644
>> --- a/fs/fuse/fuse_dev_i.h
>> +++ b/fs/fuse/fuse_dev_i.h
>> @@ -63,7 +63,6 @@ void fuse_dev_queue_forget(struct fuse_iqueue *fiq,
>> void fuse_dev_queue_interrupt(struct fuse_iqueue *fiq, struct fuse_req *req);
>>
>> bool fuse_request_expired(struct fuse_conn *fc, struct list_head *list);
>> -bool fuse_fpq_processing_expired(struct fuse_conn *fc, struct list_head *processing);
>>
>> #endif
>>
>> --
>> 2.43.5
>>
>
Reviewed-by: Bernd Schubert <bernd@bsbernd.com>
next prev parent reply other threads:[~2025-02-16 23:16 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-03 19:30 [PATCH v2] fuse: optimize over-io-uring request expiration check Joanne Koong
2025-02-03 19:37 ` Joanne Koong
2025-02-16 23:16 ` Bernd Schubert [this message]
2025-03-18 0:58 ` Joanne Koong
2025-04-15 11:46 ` Miklos Szeredi
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=bdb3980b-65e3-4bd5-aa5b-0a48d6d6e7a0@bsbernd.com \
--to=bernd@bsbernd.com \
--cc=joannelkoong@gmail.com \
--cc=kernel-team@meta.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=miklos@szeredi.hu \
/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).