linux-kernel - Re: [PATCH] rtnetlink: add rtnl

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] [day] [month] [year] [list]

Date:   Fri, 30 Apr 2021 20:48:02 +0900
From:   Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>
To:     Rocco Yue <rocco.yue@...iatek.com>
Cc:     Peter Enderborg <peter.enderborg@...y.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Anshuman Khandual <anshuman.khandual@....com>,
        Vitor Massaru Iha <vitor@...saru.org>,
        Sedat Dilek <sedat.dilek@...il.com>,
        Wei Yang <richard.weiyang@...il.com>,
        Cong Wang <cong.wang@...edance.com>,
        Di Zhu <zhudi21@...wei.com>,
        Stephen Hemminger <stephen@...workplumber.org>,
        Francis Laniel <laniel_francis@...vacyrequired.com>,
        Roopa Prabhu <roopa@...ulusnetworks.com>,
        Andrii Nakryiko <andrii@...nel.org>,
        linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-mediatek@...ts.infradead.org, wsd_upsream@...iatek.com,
        "David S . Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Masahiro Yamada <masahiroy@...nel.org>,
        Matthias Brugger <matthias.bgg@...il.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        peterz@...radead.org
Subject: Re: [PATCH] rtnetlink: add rtnl_lock debug log

On 2021/04/29 16:02, Rocco Yue wrote:
> +static void rtnl_print_btrace(struct timer_list *unused)
> +{
> +	pr_info("----------- %s start -----------\n", __func__);
> +	pr_info("%s[%d][%c] hold rtnl_lock more than 2 sec, start time: %llu\n",
> +		rtnl_instance.task->comm,
> +		rtnl_instance.pid,
> +		task_state_to_char(rtnl_instance.task),
> +		rtnl_instance.start_time);
> +	stack_trace_print(rtnl_instance.addrs, rtnl_instance.nr_entries, 0);

Do we want to print same traces every 2 seconds? 

Since it is possible to stall for e.g. 30 seconds, printing either only upon
first call to rtnl_print_btrace() for each stalling duration or only upon
end of stalling duration (i.e. from rtnl_relase_btrace()) is better?

> +	show_stack(rtnl_instance.task, NULL, KERN_DEBUG);

Why KERN_DEBUG ?

If you retrieve the output via dmesg, KERN_DEBUG would be fine.
But for syzkaller (which counts on printk() messages being printed to
consoles), KERN_INFO (or default) is expected.

> +	pr_info("------------ %s end -----------\n", __func__);
> +}
> +
> +static void rtnl_relase_btrace(void)
> +{
> +	rtnl_instance.end_time = sched_clock();
> +

You should del_timer_sync() here than

> +	if (rtnl_instance.end_time - rtnl_instance.start_time > 2000000000ULL) {
> +		pr_info("rtnl_lock is held by [%d] from [%llu] to [%llu]\n",
> +			rtnl_instance.pid,
> +			rtnl_instance.start_time,
> +			rtnl_instance.end_time);
> +	}
> +
> +	del_timer(&rtnl_chk_timer);

here in order to make sure that end message is printed only after
rtnl_print_btrace() messages are printed.

> +}
> +#endif
> +