From mboxrd@z Thu Jan  1 00:00:00 1970
From: jg1.han@samsung.com (Jingoo Han)
Date: Fri, 13 Dec 2013 13:57:06 +0900
Subject: [PATCH 09/12] pcie: SPEAr13xx: Add designware pcie support
In-Reply-To: <2CC2A0A4A178534D93D5159BF3BCB66189F95F66A2@EAPEX1MAIL1.st.com>
References: <cover.1386752447.git.mohit.kumar@st.com>
 <f5c9cbbd116d462c428afc2e8afefd95aad8c6ec.1386752447.git.mohit.kumar@st.com>
 <201312120000.54545.arnd@arndb.de>
 <2CC2A0A4A178534D93D5159BF3BCB66189F95F66A2@EAPEX1MAIL1.st.com>
Message-ID: <002c01cef7bf$c5a7b640$50f722c0$%han@samsung.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Friday, December 13, 2013 1:30 PM, Mohit KUMAR DCG wrote:
> On Thursday, December 12, 2013 4:31 AM, Arnd Bergmann wrote:
> > On Wednesday 11 December 2013, Mohit Kumar wrote:

[.....]

> > > +struct spear13xx_pcie {
> > > +	void __iomem		*phy_base;
> > > +	void __iomem		*app_base;
> > > +	struct clk		*clk;
> > > +	struct pcie_port	pp;
> > > +	int			id;
> > > +	int			is_gen1;
> > > +};
> >
> > The pcie driver shouldn't have direct access to the phy registers, use a phy
> > driver for that.
> 
> - OK. In few workarounds we have to access controller as well as phy registers,
> How we should handle such cases if we separate out phy driver from here? Should it be
> through phy global functions called from pcie driver?

Arnd, do you mean the following?

1. Implement Spear PCIe PHY driver using General PHY framework.
  ./drivers/phy/phy-spear13xx-pcie.c

2. Call General PHY APIs
  ./drivers/pci/host/pcie-spear13xx.c
  #include <linux/phy/phy.h>

  pp->phy = devm_phy_get(pp->dev, ...);

  phy_power_on(pp->phy);

  phy_power_off(pp->phy);


Best regards,
Jingoo Han