linux-kernel - Re: [PATCH] regulator: ab8500: Fix get

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:	Mon, 15 Apr 2013 13:34:06 +0200
From:	Bengt Jönsson <bengt.g.jonsson@...ricsson.com>
To:	Axel Lin <axel.lin@...ics.com>
Cc:	Mark Brown <broonie@...nsource.wolfsonmicro.com>,
	Lee Jones <lee.jones@...aro.org>,
	Yvan FILLION <yvan.fillion@...ricsson.com>,
	Liam Girdwood <lgirdwood@...il.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] regulator: ab8500: Fix get_mode for shared mode regulators

On 04/15/2013 10:50 AM, Axel Lin wrote:
>> My understanding is for shared mode regulators:
>> It can be in LP mode only when *BOTH* are in LP mode.
>> If only one of the regulator in HP mode, then *BOTH* should be in HP mode.
>> Did I misunderstand something?
Your understanding is correct.
> Let me put this issue this way:
>
> Current code behavior:
> get_mode() returns IDLE if only one lp_mode_req flag is true, but mode
> register value is HP.
>
> AB8540_LDO_ANAMIC1      AB8540_LDO_ANAMIC2      mode register
> get_mode() returns
> lp_mode_req=true        lp_mode_req=true        HP
> REGULATOR_MODE_NORMAL
> lp_mode_req=true        lp_mode_req=false       HP
> REGULATOR_MODE_IDLE
> lp_mode_req=false       lp_mode_req=true        HP
> REGULATOR_MODE_IDLE
> lp_mode_req=false       lp_mode_req=false       LP
> REGULATOR_MODE_IDLE
I think it looks like this:

AB8540_LDO_ANAMIC1      AB8540_LDO_ANAMIC2      mode register
get_mode() returns
lp_mode_req=true        lp_mode_req=true        LP
REGULATOR_MODE_IDLE     REGULATOR_MODE_IDLE
lp_mode_req=true        lp_mode_req=false       HP
REGULATOR_MODE_IDLE     REGULATOR_MODE_NORMAL
lp_mode_req=false       lp_mode_req=true        HP
REGULATOR_MODE_NORMAL   REGULATOR_MODE_IDLE
lp_mode_req=false       lp_mode_req=false       HP
REGULATOR_MODE_NORMAL   REGULATOR_MODE_NORMAL

> with this path:
> mode register value is consistent with get_mode().
>
> AB8540_LDO_ANAMIC1      AB8540_LDO_ANAMIC2      mode register
> get_mode() returns
> lp_mode_req=true        lp_mode_req=true        HP
> REGULATOR_MODE_NORMAL
> lp_mode_req=true        lp_mode_req=false       HP
> REGULATOR_MODE_NORMAL
> lp_mode_req=false       lp_mode_req=true        HP
> REGULATOR_MODE_NORMAL
> lp_mode_req=false       lp_mode_req=false       LP
> REGULATOR_MODE_IDLE

And like this:

AB8540_LDO_ANAMIC1      AB8540_LDO_ANAMIC2      mode register
get_mode() returns
lp_mode_req=true        lp_mode_req=true        LP
REGULATOR_MODE_IDLE     REGULATOR_MODE_IDLE
lp_mode_req=true        lp_mode_req=false       HP
REGULATOR_MODE_NORMAL   REGULATOR_MODE_NORMAL
lp_mode_req=false       lp_mode_req=true        HP
REGULATOR_MODE_NORMAL   REGULATOR_MODE_NORMAL
lp_mode_req=false       lp_mode_req=false       HP
REGULATOR_MODE_NORMAL   REGULATOR_MODE_NORMAL

I guess what you don't like with the current approach is that the driver returns REGULATOR_MODE_IDLE in some cases where the mode register is set to LP. But I think, with patch applied, the control may be wrong in some cases because the regulator framework will call get_mode and see that the mode is already correct and not call set_mode so lp_mode_req will not get updated. I realised my previous example was incorrect so here I describe another example:

0. Start condition:
ANAMIC1 is requested in LP mode (lp_mode_req=true) and ANAMIC2 is requested in HP mode (lp_mode_req=false).
So the mode register is set to HP.

1. Now ANAMIC1 is requested to HP mode from consumer side. The regulator framework checks the current mode with get_mode which returns HP. So the regulator framework returns without calling set_mode because the mode is already correct.
For ANAMIC1 lp_mode_req=true and for ANAMIC2 lp_mode_req=false (still).
The mode register will be correct at HP.

2. If ANAMIC2 is now requested to LP mode from consumer side, the regulator framework calls get_mode which returns HP, so the framework calls set_mode.
In set_mode, the function checks the other regulator's status which is lp_mode_req=true. So the function will continue and set the regulator in LP mode even if it should not be.

Bengt

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/