Date: Fri, 1 Apr 2016 10:25:07 +0800
From: Shawn Guo <shawnguo@kernel.org>
To: Stefan Agner <stefan@agner.ch>
Cc: mturquette@baylibre.com, sboyd@codeaurora.org, kernel@pengutronix.de,
        sergeimir@emcraft.com, tglx@linutronix.de, jason@lakedaemon.net,
        marc.zyngier@arm.com, robh+dt@kernel.org, pawel.moll@arm.com,
        mark.rutland@arm.com, ijc+devicetree@hellion.org.uk,
        galak@codeaurora.org, devicetree@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        linux-clk@vger.kernel.org
Subject: Re: [PATCH 15/18] ARM: vf610: PM: initial suspend/resume support
Message-ID: <20160401022507.GF18833@tiger>
References: <1457576219-7971-1-git-send-email-stefan@agner.ch>
 <1457576219-7971-16-git-send-email-stefan@agner.ch>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1457576219-7971-16-git-send-email-stefan@agner.ch>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1998
Lines: 38

On Wed, Mar 09, 2016 at 06:16:56PM -0800, Stefan Agner wrote:
> Add system suspend and resume support for Vybrid SoC. The standby
> sleep state puts the SoC in STOP mode. The SoC can be woken through
> an interrupt from GPC (Global Power Controller). The GPC can use any
> interrupt as wake-up source. To save power the main PLL1 is bypassed
> and uses the 24MHz on-chip oscillator. However, memory clock need to
> be at full speed, hence the PLL2 needs to be on to keep the memory
> clocked. The mode is completely implemented in C since we can access
> the full memory at all times. The mode provides most power saving
> while being able to be woken by any IRQ as wake-up source.
> 
> The mem sleep state (Suspend-to-RAM) uses Vybrid's LPSTOP2 mode. This
> mode powergates most parts of the SoC expect some peripherials such as
> Wake-Up controller (WKPU) or LP RTC. Parts of the internal SRAM is
> retained too. The suspend code written in assembly runs from this SRAM.
> The code puts the main memory (DDR3) into self-refresh mode and takes
> it out of self-refresh mode on resume. Verified with Colibri VF50/VF61
> V1.2A.
> 
> Signed-off-by: Stefan Agner <stefan@agner.ch>
> ---
>  arch/arm/mach-imx/Makefile        |   3 +
>  arch/arm/mach-imx/common.h        |  10 +
>  arch/arm/mach-imx/mach-vf610.c    |   8 +
>  arch/arm/mach-imx/pm-vf610.c      | 634 ++++++++++++++++++++++++++++++++++++++
>  arch/arm/mach-imx/suspend-vf610.S | 437 ++++++++++++++++++++++++++
>  drivers/clk/imx/clk-vf610.c       |  17 +
>  6 files changed, 1109 insertions(+)
>  create mode 100644 arch/arm/mach-imx/pm-vf610.c
>  create mode 100644 arch/arm/mach-imx/suspend-vf610.S

I know this is how we implemented suspend for i.MX6.  But this is not
the direction moving forward.  When people was pushing a pile of code
adding suspend for i.MX7D, I refused to take it and asked them to push
those hardware details into firmware and use PSCI to implement suspend.
I would like to suggest the same for Vybrid.

Shawn