[PATCH 1/7] PCI: imx6: Make reset-gpio optional

[marex@denx.de (Marek Vasut)]

On Thursday, December 12, 2013 at 06:10:31 AM, Tim Harvey wrote:
> On Wed, Dec 11, 2013 at 2:30 AM, Marek Vasut <marex@denx.de> wrote:
> > Some boards do not have a PCIe reset GPIO. To avoid probe
> > failure on these boards, make the reset GPIO optional as
> > well.
> > 
> > Signed-off-by: Marek Vasut <marex@denx.de>
> > Cc: Bjorn Helgaas <bhelgaas@google.com>
> > Cc: Frank Li <lznuaa@gmail.com>
> > Cc: Harro Haan <hrhaan@gmail.com>
> > Cc: Jingoo Han <jg1.han@samsung.com>
> > Cc: Mohit KUMAR <Mohit.KUMAR@st.com>
> > Cc: Pratyush Anand <pratyush.anand@st.com>
> > Cc: Richard Zhu <r65037@freescale.com>
> > Cc: Sascha Hauer <s.hauer@pengutronix.de>
> > Cc: Sean Cross <xobs@kosagi.com>
> > Cc: Shawn Guo <shawn.guo@linaro.org>
> > Cc: Siva Reddy Kallam <siva.kallam@samsung.com>
> > Cc: Srikanth T Shivanand <ts.srikanth@samsung.com>
> > Cc: Tim Harvey <tharvey@gateworks.com>
> > Cc: Troy Kisky <troy.kisky@boundarydevices.com>
> > Cc: Yinghai Lu <yinghai@kernel.org>
> > ---
> > 
> >  .../devicetree/bindings/pci/designware-pcie.txt    |  2 +-
> >  drivers/pci/host/pci-imx6.c                        | 29
> >  +++++++++++----------- 2 files changed, 16 insertions(+), 15
> >  deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/pci/designware-pcie.txt
> > b/Documentation/devicetree/bindings/pci/designware-pcie.txt index
> > d5d26d4..b7a2279 100644
> > --- a/Documentation/devicetree/bindings/pci/designware-pcie.txt
> > +++ b/Documentation/devicetree/bindings/pci/designware-pcie.txt
> > 
> > @@ -19,9 +19,9 @@ Required properties:
> >         to define the mapping of the PCIe interface to interrupt
> >         numbers.
> >  
> >  - num-lanes: number of lanes to use
> > 
> > -- reset-gpio: gpio pin number of power good signal
> > 
> >  Optional properties for fsl,imx6q-pcie
> > 
> > +- reset-gpio: gpio pin number of power good signal
> > 
> >  - power-on-gpio: gpio pin number of power-enable signal
> >  - wake-up-gpio: gpio pin number of incoming wakeup signal
> >  - disable-gpio: gpio pin number of outgoing rfkill/endpoint disable
> >  signal
> > 
> > diff --git a/drivers/pci/host/pci-imx6.c b/drivers/pci/host/pci-imx6.c
> > index bd70af8..52027ad 100644
> > --- a/drivers/pci/host/pci-imx6.c
> > +++ b/drivers/pci/host/pci-imx6.c
> > @@ -214,9 +214,12 @@ static int imx6_pcie_assert_core_reset(struct
> > pcie_port *pp)
> > 
> >         regmap_update_bits(imx6_pcie->iomuxc_gpr, IOMUXC_GPR1,
> >         
> >                         IMX6Q_GPR1_PCIE_REF_CLK_EN, 0 << 16);
> > 
> > -       gpio_set_value(imx6_pcie->reset_gpio, 0);
> > -       msleep(100);
> > -       gpio_set_value(imx6_pcie->reset_gpio, 1);
> > +       /* Some boards don't have PCIe reset GPIO. */
> > +       if (gpio_is_valid(imx6_pcie->reset_gpio)) {
> > +               gpio_set_value(imx6_pcie->reset_gpio, 0);
> > +               msleep(100);
> > +               gpio_set_value(imx6_pcie->reset_gpio, 1);
> > +       }
> > 
> >         return 0;
> >  
> >  }
> 
> Marek,
> 
> Though not the fault of your patch, I noticed while looking at this
> that the PCI Express specification is not being properly met with
> regards to PERST# and the reference clock.  The spec states that
> PERST# must be kept asserted until after the reference clock is stable
> (I'm not entirely clear how long of a delay is needed for the clock to
> become stable but I think the value is typically the 100ms).  I see in
> the current pci-imx6.c code that imx6_pcie_host_init calls
> imx6_pcie_assert_core_reset first, then imx6_pcie_init_phy, followed
> by imx6_pcie_deassert_core_reset.  Despite the function names,
> imx6_pcie_assert_core_reset as shown above asserts then de-asserts
> PERST# before the clock is enabled in imx6_pcie_deassert_core_reset.
> This seems to me to be a violation of the spec and I believe the
> msleep(100) and de-assertion of the option reset_gpio should be done
> in imx6_pcie_deassert_core reset after the clock is brought up.
> 
> If you agree with my assessment, would you mind resolving this issue
> at the same time?  If not, I'm happy to follow-up with a patch to
> resolve it after your patch is accepted.

Is this not resolved by patch 0006 in this series please?