linux-kernel - Re: [PATCH 3/3] ASoC: omap-mcbsp: Enable/disable sidetone block auto clock gating for omap3

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, 21 Mar 2016 10:44:38 +0200
From:	Peter Ujfalusi <peter.ujfalusi@...com>
To:	Paul Walmsley <paul@...an.com>, Mark Brown <broonie@...nel.org>
CC:	Tony Lindgren <tony@...mide.com>,
	Liam Girdwood <lgirdwood@...il.com>,
	Jarkko Nikula <jarkko.nikula@...mer.com>,
	Tero Kristo <t-kristo@...com>, <linux-omap@...r.kernel.org>,
	<linux-arm-kernel@...ts.infradead.org>,
	<linux-kernel@...r.kernel.org>, <devicetree@...r.kernel.org>,
	<alsa-devel@...a-project.org>
Subject: Re: [PATCH 3/3] ASoC: omap-mcbsp: Enable/disable sidetone block auto
 clock gating for omap3

Paul,

On 03/19/16 21:37, Paul Walmsley wrote:
> On Fri, 18 Mar 2016, Peter Ujfalusi wrote:
> 
>> OMAP3's McBSP2 and McBSP3 module have integrated sidetone block with
>> dedicated SYSCONFIG register. The sidetone is operating from the maain
>> McBSP module's ICLK. For normal operation the sidetone clock auto idle
>> support needs to be disabled when it is activated.
>> Note: This is not enough to avoid choppy sidetone because this AUTOIDLE
>> bit is controlling only the clock auto idle from the McBSP to the sidetone
>> block. If the McBSP_ICLK is idling, the sidetone clock is going to do the
>> same.
>>
>> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@...com>
> 
> Mark, please drop this patch for the time being, until the SoC integration 
> issues can be sorted out first.  It's best to wait a little while before 
> applying patches like these so folks have a chance to comment on their 
> correctness first.
> 
> We used to handle this problem in the OMAP hwmod SoC integration layer 
> with a flag that forced the interface clock to stay active as long as the 
> underlying IP blocks were active.  However I can't find that flag right 
> now in the current data, so maybe it got accidentally or inadvertently 
> removed at some point in time in the past.  The right way to fix this 
> would be to add that flag back in, rather than messing with the SoC 
> integration registers from the McBSP drivers.

I can not recall such a flag. We had both hwmods attached to the given McBSP
mkodule via dev_attr and we dealt with the McBSP_ICLK autoidle enable/disable
via callbacks provided to the driver via platform data.
arch/arm/mach-omap2/mcbsp.c: omap3_enable_st_clock() In there we use
omap2_clk_deny_idle()/omap2_clk_allow_idle() to make sure that the ICLK is not
gated when the ST is enabled in the given McBSP module.
But this only works when we boot in legacy mode. The DT boot is broken in this
regards as long we have first booted OMAP3 with DT.

-- 
Péter