[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20251222080046.2312024-1-liujing40@xiaomi.com>
Date: Mon, 22 Dec 2025 16:00:46 +0800
From: liujing40 <liujing.root@...il.com>
To: rostedt@...dmis.org
Cc: alexei.starovoitov@...il.com,
andrii@...nel.org,
ast@...nel.org,
bpf@...r.kernel.org,
daniel@...earbox.net,
eddyz87@...il.com,
haoluo@...gle.com,
john.fastabend@...il.com,
jolsa@...nel.org,
kpsingh@...nel.org,
linux-kernel@...r.kernel.org,
linux-trace-kernel@...r.kernel.org,
liujing.root@...il.com,
liujing40@...omi.com,
martin.lau@...ux.dev,
mhiramat@...nel.org,
sdf@...ichev.me,
song@...nel.org,
yonghong.song@...ux.dev
Subject: Re: [PATCH 2/2] bpf: Implement kretprobe fallback for kprobe multi link
On Thu, 18 Dec 2025 16:09:25 -0500
Steven Rostedt <rostedt@...dmis.org> wrote:
> On Thu, 18 Dec 2025 09:53:16 -0800
> Alexei Starovoitov <alexei.starovoitov@...il.com> wrote:
>
> > > +static void bpf_kprobe_unregister(struct bpf_kprobe *kps, u32 cnt)
> > > +{
> > > + for (int i = 0; i < cnt; i++)
> > > + unregister_kretprobe(&kps[i].rp);
> > > +}
> >
> > Nack.
> > This is not a good idea.
> > unregister_kretprobe() calls synchronize_rcu().
> > So the above loop will cause soft lockups for sure.
>
> Looks like it could be replaced with:
>
> unregister_kretprobes(kps, cnt);
>
> Which unregisters an array of kreptrobes and does a single
> synchronize_rcu().
>
> -- Steve
Thanks for the suggestion. I will refactor the loop to use
unregister_kretprobes() for a single RCU sync.
Powered by blists - more mailing lists