linux-kernel - Re: [PATCH 4/5] perf/hw_breakpoint: Set breakpoint as disabled in modify_user_hw

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20180809141713.GB22636@redhat.com>
Date:   Thu, 9 Aug 2018 16:17:13 +0200
From:   Oleg Nesterov <oleg@...hat.com>
To:     Jiri Olsa <jolsa@...nel.org>
Cc:     Arnaldo Carvalho de Melo <acme@...nel.org>,
        lkml <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        David Ahern <dsahern@...il.com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Peter Zijlstra <a.p.zijlstra@...llo.nl>,
        Milind Chabbi <chabbi.milind@...il.com>,
        Frederic Weisbecker <frederic@...nel.org>
Subject: Re: [PATCH 4/5] perf/hw_breakpoint: Set breakpoint as disabled in
 modify_user_hw_breakpoint error path

On 08/09, Jiri Olsa wrote:
>
> @@ -523,8 +523,10 @@ int modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *att
>  		perf_event_disable(bp);
>
>  	err = modify_user_hw_breakpoint_check(bp, attr, false);
> -	if (err)
> +	if (err) {
> +		bp->attr.disabled = 1;
>  		return err;

Yes, but on the second thought... Can't we simply do

	int modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr)
	{
		int err;

		/*
		 * modify_user_hw_breakpoint can be invoked with IRQs disabled and hence it
		 * will not be possible to raise IPIs that invoke __perf_event_disable.
		 * So call the function directly after making sure we are targeting the
		 * current task.
		 */
		if (irqs_disabled() && bp->ctx && bp->ctx->task == current)
			perf_event_disable_local(bp);
		else
			perf_event_disable(bp);

		err = modify_user_hw_breakpoint_check(bp, attr, false);

		if (!bp.attr->disabled)
			perf_event_enable(bp);

		return err;
	}

instead of this and the next patch?

We can do this because (as you pointed out) validate_hw_breakpoint() has already
gone in -tip tree, and (afaics) modify_user_hw_breakpoint_check() never changes
perf_event's state on failure, so we can safely "restart" this bp if it was enabled
before.

1. This is what we had before the recent f67b15037a7a50c57f72e69a6d59941ad90a0f0f
   "perf/hwbp: Simplify the perf-hwbp code, fix documentation".

   Note that this commit was actually the bug fix which introduced another problem
   fixed by your 2/5.

   But see above, perf_event_enable() is no longer unsafe after
   modify_user_hw_breakpoint_check(), we can restore the previous semantics.

2. This matches perf_event_modify_breakpoint(). Which btw can be simplified a bit,
   it too can simply do

		if (!bp->attr.disabled)
			_perf_event_enable(bp);

		return err;

Oleg.