linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Miklos Szeredi <miklos@szeredi.hu>
To: Bernd Schubert <bschubert@ddn.com>
Cc: "Vivek Goyal" <vgoyal@redhat.com>,
	"Stefan Hajnoczi" <stefanha@redhat.com>,
	"Eugenio Pérez" <eperezma@redhat.com>,
	linux-fsdevel@vger.kernel.org,
	"Joanne Koong" <joannelkoong@gmail.com>,
	"Josef Bacik" <josef@toxicpanda.com>
Subject: Re: [PATCH v3 1/4] fuse: Make the fuse unique value a per-cpu counter
Date: Fri, 4 Apr 2025 14:43:32 +0200	[thread overview]
Message-ID: <CAJfpegtBinmb_D=R0zYWF3AoXscwFugRhCMQKP_aRehq5Y_Wfg@mail.gmail.com> (raw)
In-Reply-To: <20250403-fuse-io-uring-trace-points-v3-1-35340aa31d9c@ddn.com>

On Thu, 3 Apr 2025 at 22:23, Bernd Schubert <bschubert@ddn.com> wrote:

> +/**
> + * Get the next unique ID for a request
> + */
> +static inline u64 fuse_get_unique(struct fuse_iqueue *fiq)
> +{
> +       int step = FUSE_REQ_ID_STEP * (task_cpu(current));
> +       u64 cntr = this_cpu_inc_return(*fiq->reqctr);
> +
> +       return cntr * FUSE_REQ_ID_STEP * NR_CPUS + step;

Thinking a bit... this looks wrong.

The reason is that the task could be migrated to a different CPU
between the task_cpu() and the this_cpu_inc_return(), resulting in a
possibly duplicated value.

This could be fixed with a preempt_disable()/preempt_enable() pair,
but I think it would be cleaner to go with my original idea and
initialize the percpu counters to  CPUID and increment by NR_CPU *
FUSE_REQ_ID_STEP when fetching a new value.

Thanks,
Miklos

  reply	other threads:[~2025-04-04 12:43 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-03 20:22 [PATCH v3 0/4] fuse: Improve ftraces, per-cpu req unique and code dup removal Bernd Schubert
2025-04-03 20:22 ` [PATCH v3 1/4] fuse: Make the fuse unique value a per-cpu counter Bernd Schubert
2025-04-04 12:43   ` Miklos Szeredi [this message]
2025-04-04 13:19     ` Bernd Schubert
2025-04-03 20:22 ` [PATCH v3 2/4] fuse: Set request unique on allocation Bernd Schubert
2025-04-03 20:22 ` [PATCH v3 3/4] fuse: {io-uring} Avoid _send code dup Bernd Schubert
2025-04-03 20:22 ` [PATCH v3 4/4] fuse: fine-grained request ftraces Bernd Schubert

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='CAJfpegtBinmb_D=R0zYWF3AoXscwFugRhCMQKP_aRehq5Y_Wfg@mail.gmail.com' \
    --to=miklos@szeredi.hu \
    --cc=bschubert@ddn.com \
    --cc=eperezma@redhat.com \
    --cc=joannelkoong@gmail.com \
    --cc=josef@toxicpanda.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=stefanha@redhat.com \
    --cc=vgoyal@redhat.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).