linux-kernel - Re: [v4,7/7] thermal: mediatek: use spinlock to protect PTPCORESEL

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:   Wed, 8 Jul 2020 18:00:33 +0800
From:   Michael Kao <michael.kao@...iatek.com>
To:     Daniel Lezcano <daniel.lezcano@...aro.org>
CC:     Matthias Brugger <matthias.bgg@...il.com>,
        Zhang Rui <rui.zhang@...el.com>,
        Eduardo Valentin <edubezval@...il.com>,
        Rob Herring <robh+dt@...nel.org>,
        Mark Rutland <mark.rutland@....com>, <hsinyi@...omium.org>,
        <linux-pm@...r.kernel.org>, <srv_heupstream@...iatek.com>,
        <devicetree@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <linux-arm-kernel@...ts.infradead.org>,
        <linux-mediatek@...ts.infradead.org>
Subject: Re: [v4,7/7] thermal: mediatek: use spinlock to protect PTPCORESEL

On Tue, 2020-07-07 at 10:13 +0200, Daniel Lezcano wrote:
> On 05/06/2020 05:50, Michael Kao wrote:
> > On Fri, 2020-05-22 at 17:36 +0200, Daniel Lezcano wrote:
> >> On 23/03/2020 13:15, Michael Kao wrote:
> >>> From: "michael.kao" <michael.kao@...iatek.com>
> >>>
> >>> The driver of thermal and svs will use the
> >>> same register for the project which should select
> >>> bank before reading sensor value.
> >>
> >> Here there is a design problem AFAICT. The sensor should not be using
> >> external locks.
> >>
> > The PTPCORESEL is a common register used by svs and thermal.
> > The thermal need to ensure PTPCORESEL register will not be changed by
> > svs when thermal switch bank to read raw data of temperature.
> > So we use svs_lock to make sure there is no conflict between the two
> > drivers.
> 
> Why not use regmap ?
> 
Hi Daniel,

We are not sure the regmap can solve the problem.
The regmap can lock CORESELECT. But it can not protect 
the other sensors registers which thermal controller start to
read temperature need to access.

After internal disscussion, we don't have this kind of experience to
use regmap to protect some peice of code.
We will do some research of this kind of usage.
Thanks for your suggestion.

> >>> Signed-off-by: Michael Kao <michael.kao@...iatek.com>
> >>> ---
> >>>  drivers/thermal/mtk_thermal.c | 9 ++++-----
> >>>  1 file changed, 4 insertions(+), 5 deletions(-)
> >>>
> >>> diff --git a/drivers/thermal/mtk_thermal.c b/drivers/thermal/mtk_thermal.c
> >>> index 9eaca432920e..594ad4f0f8cd 100644
> >>> --- a/drivers/thermal/mtk_thermal.c
> >>> +++ b/drivers/thermal/mtk_thermal.c
> >>> @@ -22,6 +22,7 @@
> >>>  #include <linux/thermal.h>
> >>>  #include <linux/reset.h>
> >>>  #include <linux/types.h>
> >>> +#include <linux/power/mtk_svs.h>
> >>>  
> >>>  /* AUXADC Registers */
> >>>  #define AUXADC_CON1_SET_V	0x008
> >>> @@ -262,7 +263,7 @@ struct mtk_thermal {
> >>>  	struct clk *clk_peri_therm;
> >>>  	struct clk *clk_auxadc;
> >>>  	/* lock: for getting and putting banks */
> >>> -	struct mutex lock;
> >>> +	unsigned long flags;
> >>>  
> >>>  	/* Calibration values */
> >>>  	s32 adc_ge;
> >>> @@ -561,7 +562,7 @@ static void mtk_thermal_get_bank(struct mtk_thermal_bank *bank)
> >>>  	u32 val;
> >>>  
> >>>  	if (mt->conf->need_switch_bank) {
> >>> -		mutex_lock(&mt->lock);
> >>> +		mt->flags = claim_mtk_svs_lock();
> >>>  
> >>>  		val = readl(mt->thermal_base + PTPCORESEL);
> >>>  		val &= ~0xf;
> >>> @@ -581,7 +582,7 @@ static void mtk_thermal_put_bank(struct mtk_thermal_bank *bank)
> >>>  	struct mtk_thermal *mt = bank->mt;
> >>>  
> >>>  	if (mt->conf->need_switch_bank)
> >>> -		mutex_unlock(&mt->lock);
> >>> +		release_mtk_svs_lock(mt->flags);
> >>>  }
> >>>  
> >>>  /**
> >>> @@ -938,8 +939,6 @@ static int mtk_thermal_probe(struct platform_device *pdev)
> >>>  	if (ret)
> >>>  		return ret;
> >>>  
> >>> -	mutex_init(&mt->lock);
> >>> -
> >>>  	mt->dev = &pdev->dev;
> >>>  
> >>>  	auxadc = of_parse_phandle(np, "mediatek,auxadc", 0);
> >>>
> >>
> >>
> > 
> 
>