linux-trace-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Oleg Nesterov <oleg@redhat.com>
To: "Liao, Chang" <liaochang1@huawei.com>
Cc: mhiramat@kernel.org, peterz@infradead.org, mingo@redhat.com,
	acme@kernel.org, namhyung@kernel.org, mark.rutland@arm.com,
	alexander.shishkin@linux.intel.com, jolsa@kernel.org,
	irogers@google.com, adrian.hunter@intel.com,
	kan.liang@linux.intel.com, linux-kernel@vger.kernel.org,
	linux-trace-kernel@vger.kernel.org,
	linux-perf-users@vger.kernel.org
Subject: Re: [PATCH] uprobes: Improve scalability by reducing the contention on siglock
Date: Tue, 6 Aug 2024 19:25:29 +0200	[thread overview]
Message-ID: <20240806172529.GC20881@redhat.com> (raw)
In-Reply-To: <0c69ef28-26d8-4b6e-fa78-2211a7b84eca@huawei.com>

On 08/06, Liao, Chang wrote:
>
> You're absolutely right. handle_signlestep() has chance to handle _DENY_SIGANL
> unless it followed by setting TIF_UPROBE in uprobe_deny_signal(). This means
> _DENY_SIGNAL is likey replaced during next uprobe single-stepping.
>
> I believe introducing _DENY_SIGNAL as the immediate state between UTASK_SSTEP
> and UTASK_SSTEP_ACK is still necessary. This allow uprobe_post_sstep_notifier()
> to correctly restore TIF_SIGPENDING upon the completion of single-step.
>
> A revised implementation would look like this:

Still looks "obviously wrong" to me... even the approach itself.

Perhaps I am wrong, yet another day when I can't even read emails on lkml
carefully, sorry.

But can you please send the patch which I could actually apply? This one
looks white-space damaged...

I'll try to reply with more details as soon I convince myself I fully
understand what does your patch actually do, but most probably not tomorrow.

Thanks,

Oleg.

> ------------------%<------------------
> --- a/kernel/events/uprobes.c
> +++ b/kernel/events/uprobes.c
> @@ -1980,6 +1980,7 @@ bool uprobe_deny_signal(void)
> 
>         if (task_sigpending(t)) {
>                 clear_tsk_thread_flag(t, TIF_SIGPENDING);
> +               utask->state = UTASK_SSTEP_DENY_SIGNAL;
> 
>                 if (__fatal_signal_pending(t) || arch_uprobe_xol_was_trapped(t)) {
>                         utask->state = UTASK_SSTEP_TRAPPED;
> @@ -2276,22 +2277,23 @@ static void handle_singlestep(struct uprobe_task *utask, struct pt_regs *regs)
>         int err = 0;
> 
>         uprobe = utask->active_uprobe;
> -       if (utask->state == UTASK_SSTEP_ACK)
> +       switch (utask->state) {
> +       case UTASK_SSTEP_ACK:
>                 err = arch_uprobe_post_xol(&uprobe->arch, regs);
> -       else if (utask->state == UTASK_SSTEP_TRAPPED)
> +               break;
> +       case UTASK_SSTEP_TRAPPED:
>                 arch_uprobe_abort_xol(&uprobe->arch, regs);
> -       else
> +               set_thread_flag(TIF_SIGPENDING);
> +               break;
> +       default:
>                 WARN_ON_ONCE(1);
> +       }
> 
>         put_uprobe(uprobe);
>         utask->active_uprobe = NULL;
>         utask->state = UTASK_RUNNING;
>         xol_free_insn_slot(current);
> 
> -       spin_lock_irq(&current->sighand->siglock);
> -       recalc_sigpending(); /* see uprobe_deny_signal() */
> -       spin_unlock_irq(&current->sighand->siglock);
> -
>         if (unlikely(err)) {
>                 uprobe_warn(current, "execute the probed insn, sending SIGILL.");
>                 force_sig(SIGILL);
> @@ -2351,6 +2353,8 @@ int uprobe_post_sstep_notifier(struct pt_regs *regs)
>                 /* task is currently not uprobed */
>                 return 0;
> 
> +       if (utask->state == UTASK_SSTEP_DENY_SIGNAL)
> +               set_thread_flag(TIF_SIGPENDING);
>         utask->state = UTASK_SSTEP_ACK;
>         set_thread_flag(TIF_UPROBE);
>         return 1;
> 
> ------------------>%------------------
> 
> > 
> > Oleg.
> > 
> > 
> 
> -- 
> BR
> Liao, Chang
> 


  reply	other threads:[~2024-08-06 17:25 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-01  8:24 [PATCH] uprobes: Improve scalability by reducing the contention on siglock Liao Chang
2024-08-01 14:06 ` Oleg Nesterov
2024-08-02  1:38   ` Liao, Chang
2024-08-02  9:24     ` Oleg Nesterov
2024-08-06  3:06       ` Liao, Chang
2024-08-06 17:25         ` Oleg Nesterov [this message]
2024-08-07 10:17           ` Oleg Nesterov
2024-08-08  7:30             ` Liao, Chang
2024-08-08 10:28               ` Oleg Nesterov
2024-08-08 12:31                 ` Liao, Chang
2024-08-08 13:17                   ` Oleg Nesterov

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=20240806172529.GC20881@redhat.com \
    --to=oleg@redhat.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=irogers@google.com \
    --cc=jolsa@kernel.org \
    --cc=kan.liang@linux.intel.com \
    --cc=liaochang1@huawei.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=linux-trace-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=mhiramat@kernel.org \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    /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).