linux-kernel - RE: [EXT] Re: [v2] rtc: pcf85363/pcf85263: fix error that failed to run hwclock -w

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]

Message-ID: <DB7PR04MB4490A1949E964EAD1F539F2E8FA10@DB7PR04MB4490.eurprd04.prod.outlook.com>
Date:   Mon, 26 Aug 2019 10:40:37 +0000
From:   Biwen Li <biwen.li@....com>
To:     Alexandre Belloni <alexandre.belloni@...tlin.com>
CC:     Nandor Han <nandor.han@...sala.com>, Leo Li <leoyang.li@....com>,
        Mark Brown <broonie@...nel.org>,
        "a.zummo@...ertech.it" <a.zummo@...ertech.it>,
        "linux-rtc@...r.kernel.org" <linux-rtc@...r.kernel.org>,
        lkml <linux-kernel@...r.kernel.org>
Subject: RE: [EXT] Re: [v2] rtc: pcf85363/pcf85263: fix error that failed to
 run hwclock -w

> 
> Hi,
> 
> On 26/08/2019 09:49:49+0000, Biwen Li wrote:
> > >
> > > On 8/26/19 7:29 AM, Biwen Li wrote:
> > > >>
> > > >> On 8/16/19 10:40 PM, Li Yang wrote:
> > > >>> On Fri, Aug 16, 2019 at 11:30 AM Alexandre Belloni
> > > >>> <alexandre.belloni@...tlin.com> wrote:
> > > >>>>
> > > >>>> On 16/08/2019 10:50:49-0500, Li Yang wrote:
> > > >>>>> On Fri, Aug 16, 2019 at 3:05 AM Alexandre Belloni
> > > >>>>> <alexandre.belloni@...tlin.com> wrote:
> > > >>>>>>
> > > >>>>>> On 16/08/2019 10:46:36+0800, Biwen Li wrote:
> > > >>>>>>> Issue:
> > > >>>>>>>       - # hwclock -w
> > > >>>>>>>         hwclock: RTC_SET_TIME: Invalid argument
> > > >>>>>>>
> > > >>>>>>> Why:
> > > >>>>>>>       - Relative patch:
> > > >> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2
> > > >>
> Flkml.org&amp;data=02%7C01%7Cbiwen.li%40nxp.com%7C03141ff7858343
> 3
> > > >>
> 20be408d72a0d1e10%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%
> 7C63
> > > >>
> 7024108138794294&amp;sdata=QrALkFN6heF%2B7S73FQ9c%2FyKNRHyBuL
> %2B6
> > > >> %2B4PDM9hYRyM%3D&amp;reserved=0
> > > >> %2Flkml%2F2019%2F4%2F3%2F55&amp;data=02%7C01%7Cbiwen.li%
> 40n
> > > xp.
> > > >>
> > >
> com%7Cff8cebc3f1034ae3fa9608d725ff9e5e%7C686ea1d3bc2b4c6fa92cd99
> > > >>
> > >
> c5c301635%7C0%7C0%7C637019652111923736&amp;sdata=spY6e22YOkOF
> > > >>
> 3%2BF7crSM0M6xPmOhgULDqMZLQw%2BAmdI%3D&amp;reserved=0 ,
> > > this patch
> > > >>>>>>>         will always check for unwritable registers, it will compare
> reg
> > > >>>>>>>         with max_register in regmap_writeable.
> > > >>>>>>>
> > > >>>>>>>       - In drivers/rtc/rtc-pcf85363.c, CTRL_STOP_EN is 0x2e,
> > > >>>>>>> but
> > > >> DT_100THS
> > > >>>>>>>         is 0, max_regiter is 0x2f, then reg will be equal to 0x30,
> > > >>>>>>>         '0x30 < 0x2f' is false,so regmap_writeable will return
> false.
> > > >>>>>>>
> > > >>>>>>>       - Root cause: the buf[] was written to a wrong place in the
> file
> > > >>>>>>>         drivers/rtc/rtc-pcf85363.c
> > > >>>>>>>
> > > >>>>>>
> > > >>>>>> This is not true, the RTC wraps the register accesses
> > > >>>>>> properly and this
> > > >>>>>
> > > >>>>> This performance hack probably deserve some explanation in the
> > > >>>>> code comment.  :)
> > > >>>>>
> > > >>>>>> is probably something that should be handled by regmap_writable.
> > > >>>>>
> > > >>>>> The address wrapping is specific to this RTC chip.  Is it also
> > > >>>>> commonly used by other I2C devices?  I'm not sure if
> > > >>>>> regmap_writable should handle the wrapping case if it is too special.
> > > >>>>>
> > > >>>>
> > > >>>> Most of the i2c RTCs do address wrapping which is sometimes the
> > > >>>> only way to properly set the time.
> > > >>>
> > > >>> Adding Mark and Nandor to the loop.
> > > >>>
> > > >>> Regards,
> > > >>> Leo
> > > >>>
> > > >>
> > > >> Hi,
> > > >>     `regmap` provides couple of ways to validate the registers:
> > > >> max_register, callback function and write table. All of these are
> > > >> optional, so it gives you the freedom to customize it as needed.
> > > >>
> > > >> In this situation probably you could:
> > > >>     1. Avoid using the wrapping feature of pcf85363 (you can just
> > > >> provide separate calls for stop, reset and time confguration). In
> > > >> this way the `max_register` validation method will work fine.
> > > > Yes, I use this way. Path as follows:
> > > > Stop and reset - > set time > stop
> > > >
> > >
> > > Some of the concerns regarding this method was that it might not be
> > > precise enough. That because you need 2 I2C operations (one for stop
> > > and one for time configuration). Not sure about your case if this is a problem
> or not.
> > Ok, got it, thanks.
> 
> To be clear, for this RTC it is fine to separate both writes. Want I want is a
> corrected commit message with a proper reference to
> 8b9f9d4dc511309918c4f6793bae7387c0c638af instead of a link to lkml.org
> and a proper explanation.
Ok, got it, thanks.I will replace link to lkml.org with
8b9f9d4dc511309918c4f6793bae7387c0c638af and add a proper explanation
to the commit message in v4.
> 
> --
> Alexandre Belloni, Bootlin
> Embedded Linux and Kernel engineering
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbootlin.
> com&amp;data=02%7C01%7Cbiwen.li%40nxp.com%7C03141ff7858343320b
> e408d72a0d1e10%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C6
> 37024108138794294&amp;sdata=XnAxJmOkh1VVA9ed%2FLr%2BbvWbVpLD
> bwLjJrdaFidRtDk%3D&amp;reserved=0