Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Feedback-ID: i56a14606:Fastmail
User-Agent: Cyrus-JMAP/3.7.0-alpha0-1115-g8b801eadce-fm-20221102.001-g8b801ead
Mime-Version: 1.0
Message-Id: <c5e8ab50-aacb-4651-8893-a6dd9edcd155@app.fastmail.com>
In-Reply-To: <Y3QNzvAYo9y+yyTf@mail.local>
References: <20220908115337.1604277-1-arnd@kernel.org>
 <Y3QNzvAYo9y+yyTf@mail.local>
Date:   Wed, 16 Nov 2022 08:01:00 +0100
From:   "Arnd Bergmann" <arnd@arndb.de>
To:     "Alexandre Belloni" <alexandre.belloni@bootlin.com>,
        "Arnd Bergmann" <arnd@kernel.org>
Cc:     "Alessandro Zummo" <a.zummo@towertech.it>,
        "Reinier Kuipers" <kuipers.reinier@gmail.com>,
        linux-rtc@vger.kernel.org, "Russell King" <linux@armlinux.org.uk>,
        "Yang Yingliang" <yangyingliang@huawei.com>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] [RFC] rtc: y2038: remove broken RTC_HCTOSYS workaround
Content-Type: text/plain
Precedence: bulk

On Tue, Nov 15, 2022, at 23:08, Alexandre Belloni wrote:
> Hello,
>
> I'm fine with the patch and I'll probably take it, 

Ok, thanks!

> I a an observation though:
>
> On 08/09/2022 13:53:20+0200, Arnd Bergmann wrote:
>> +	 *
>> +	 * Since the kernel has no way of knowing what user space it runs,
>> +	 * warn here whenever the kernel is able to run it.
>> +	 * When CONFIG_COMPAT_32BIT_TIME is disabled, we know that the
>> +	 * system is safe, but unfortunately this this is currently not
>> +	 * supported by musl-1.2.x or most glibc based user space.
>
> I was under the impression that musl never had a 32bit time_t nor used
> the 32bit time APIs so it would not be affected by the bug.
> So I guess the only affected userspace is glibc without _TIME_BITS=64

It's actually the opposite: while new versions of musl only allow
building applications against the time64 interfaces, musl itself
uses a mix of the time32 and time64 system calls, and the musl
maintainer considers turning CONFIG_COMPAT_32BIT_TIME off a
misfeature of the kernel that he does not want to support.

      Arnd