Received-SPF: pass (google.com: domain of linux-kernel+bounces-211245-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Message-ID: <c0b6ad83-b9d4-43e7-8c1a-14b71a2060f8@prolan.hu>
Date: Wed, 12 Jun 2024 11:16:02 +0200
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [RFC PATCH v2] rtc: pcf2127: Add PPS capability through Seconds
 Interrupt
To: Miroslav Lichvar <mlichvar@redhat.com>, Richard Cochran
	<richardcochran@gmail.com>
CC: <linux-rtc@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	=?UTF-8?Q?Szentendrei=2C_Tam=C3=A1s?= <szentendrei.tamas@prolan.hu>,
	Alexandre Belloni <alexandre.belloni@bootlin.com>
References: <20240611150458.684349-1-csokas.bence@prolan.hu>
 <Zmks31shpsnoLQ3k@hoboy.vegasvil.org> <ZmlTQsgRiW9fmYcB@localhost>
Content-Language: en-US
From: =?UTF-8?B?Q3PDs2vDoXMgQmVuY2U=?= <csokas.bence@prolan.hu>
In-Reply-To: <ZmlTQsgRiW9fmYcB@localhost>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit

On 6/12/24 09:50, Miroslav Lichvar wrote:
> On Tue, Jun 11, 2024 at 10:06:39PM -0700, Richard Cochran wrote:
>> On Tue, Jun 11, 2024 at 05:04:57PM +0200, Csókás, Bence wrote:
>>
>>> PCF2127/29/31 is capable of generating an interrupt on every
>>> second (SI) or minute (MI) change. It signals this through
>>> the Minute/Second Flag (MSF) as well, which needs to be cleared.
>>
>> This is a RFC, and my comment is that a PPS from an RTC is not useful
>> to the Linux kernel.
> 
> I think a TCXO-based RTC can be useful to user space to improve
> holdover performance with NTP/PTP.

Exactly.

> There already is the RTC_UIE_ON
> ioctl to enable interrupts and receive them in user space.
> 
> The advantage of the PPS device over the ioctl would be more accurate
> timestamping (kernel vs user-space). Should PPS be supported, it would
> be nice if it worked generally with all drivers that support RTC_UIE_ON.

As we've discussed in v1, UIE hardware support is being removed from the 
RTC subsystem, which I tried to optionally re-introduce. Since there was 
no response since then, I assumed that there is no willingness to do 
that, so I chose the next best option, the PPS subsystem.

On 5/28/24 19:56, Alexandre Belloni wrote:
 > This has been removed from the kernel 13 years ago. What is your use
 > case to reintroduce it?

I also agree that multiple RTCs would benefit from this feature. 
However, we should only add it to those which *have* hardware support 
for a "one second has elapsed" signal. UIE is currently implemented by 
setting an alarm to the next second, which didn't work well with the 
PCF2129.

Bence