linux-kernel - RE: [PATCH 2/3] scsi: ufs: Keep device power on only fWriteBoosterBufferFlushDuringHibernate == 1

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]

Date:   Thu, 3 Dec 2020 12:15:40 +0000
From:   Avri Altman <Avri.Altman@....com>
To:     Bean Huo <huobean@...il.com>,
        "alim.akhtar@...sung.com" <alim.akhtar@...sung.com>,
        "asutoshd@...eaurora.org" <asutoshd@...eaurora.org>,
        "jejb@...ux.ibm.com" <jejb@...ux.ibm.com>,
        "martin.petersen@...cle.com" <martin.petersen@...cle.com>,
        "stanley.chu@...iatek.com" <stanley.chu@...iatek.com>,
        "beanhuo@...ron.com" <beanhuo@...ron.com>,
        "bvanassche@....org" <bvanassche@....org>,
        "tomas.winkler@...el.com" <tomas.winkler@...el.com>,
        "cang@...eaurora.org" <cang@...eaurora.org>
CC:     "linux-scsi@...r.kernel.org" <linux-scsi@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH 2/3] scsi: ufs: Keep device power on only
 fWriteBoosterBufferFlushDuringHibernate == 1

> On Thu, 2020-12-03 at 10:46 +0000, Avri Altman wrote:
> > > > > From: Bean Huo <beanhuo@...ron.com>
> > > > >
> > > > > Keep device power mode as active power mode and VCC supply only
> > > > > if
> > > > > fWriteBoosterBufferFlushDuringHibernate setting 1 is
> > > > > successful.
> > >
> > > Hi Avri
> > > Thanks so much taking time reiew.
> > >
> > > > Why would it fail?
> > >
> > > During the reliability testing in harsh environments, such as:
> > > EMS testing, in the high/low-temperature environment. The system
> > > would
> > > reboot itself, there will be programming failure very likely.
> > > If we assume failure will never hit, why we capture its result
> > > following with dev_err(). If you keep using your phone in a harsh
> > > environment, you will see this print message.
> > >
> > > Of course, in a normal environment, the chance of failure likes you
> > > to
> > > win a lottery, but the possibility still exists.
> >
> > Exactly.
> 
> so, you agree the possiblity of failure  exists.
I was more relating to the lottery part.

> 
> > Hence we need-not any extra logic protecting device management
> > command failures.
> 
> what extra logic?
> 
> >
> > if reading the configuration pass correctly, and UFSHCD_CAP_WB_EN is
> > set,
> 
> 
> UFSHCD_CAP_WB_EN set is DRAM level. still in the cache.
> 
> > one should expect that any other functionality would work.
> >
> No,  The programming will consume more power than reading, the
> later setting will more possbile fail than reading.
> 
> > Otherwise, any non-standard behavior should be added with a quirk.
> >
> 
> NO, this is not what is standard or non-standard. This is independent
> of UFS device/controller. It is a software design. IMO, we didn't deal
> with programming status that is a potential bug. If having to impose to
> a component, do you think should be controller or device? Instead of
> addin a quirk, I prefer dropping this patch.
It seems you are adding some special treatment in case some device management command failed,
A vanishingly unlikely event but a one that has significant impact over power consumption.
If a device is not responding properly to some specific device management command,
It should be treated accordingly.

Thanks,
Avri

> 
> 
> 
> 
> > Thanks,
> > Avri
> > >
> > >
> > > > Since UFSHCD_CAP_WB_EN is toggled off on ufshcd_wb_probe If the
> > > > device doesn't support wb,
> > > > The check ufshcd_is_wb_allowed should suffice, isn't it?
> > > >
> > >
> > > No, UFSHCD_CAP_WB_EN only tells us if the platform supports WB,
> > > doesn't tell us fWriteBoosterBufferFlushDuringHibernate status.
> > >
> > > Thanks,
> > > Bean