Re: [PATCH v4 2/3] dwc: PCI: intel: PCIe RC controller driver

[Dilip Kota <eswara.kota@linux.intel.com>]

Hi Bjorn Helgaas,

On 10/22/2019 1:17 AM, Bjorn Helgaas wrote:
> On Mon, Oct 21, 2019 at 02:39:19PM +0800, Dilip Kota wrote:
>> Add support to PCIe RC controller on Intel Gateway SoCs.
>> PCIe controller is based of Synopsys DesignWare pci core.
>>
>> Intel PCIe driver requires Upconfig support, fast training
>> sequence configuration and link speed change. So adding the
>> respective helper functions in the pcie DesignWare framework.
>> It also programs hardware autonomous speed during speed
>> configuration so defining it in pci_regs.h.
>>
>> +static void intel_pcie_link_setup(struct intel_pcie_port *lpp)
>> +{
>> +	u32 val;
>> +
>> +	val = pcie_rc_cfg_rd(lpp, PCIE_CAP_OFST + PCI_EXP_LNKCAP);
>> +	lpp->max_speed = FIELD_GET(PCI_EXP_LNKCAP_SLS, val);
>> +	lpp->max_width = FIELD_GET(PCI_EXP_LNKCAP_MLW, val);
>> +
>> +	val = pcie_rc_cfg_rd(lpp, PCIE_CAP_OFST + PCI_EXP_LNKCTL);
>> +
>> +	val &= ~(PCI_EXP_LNKCTL_LD | PCI_EXP_LNKCTL_ASPMC);
>> +	val |= (PCI_EXP_LNKSTA_SLC << 16) | PCI_EXP_LNKCTL_CCC |
>> +	       PCI_EXP_LNKCTL_RCB;
> Link Control is only 16 bits wide, so "PCI_EXP_LNKSTA_SLC << 16"
> wouldn't make sense.  But I guess you're writing a device-specific
> register that is not actually the Link Control as documented in PCIe
> r5.0, sec 7.5.3.7, even though the bits are similar?
It is not device specific.
You are correct about register size that pcie spec mentions 
PCIE_EXP_LNKCTL at 0x10 and PCIE_EXP_LNKSTS at 0x12 each of 2bytes. 
Accessing 4bytes at offset 0x10 ends up accessing LNK control and status 
register.
In Synopsys PCIe controller LINK_CONTROL_LINK_STATUS_REG is of 4bytes 
size at offset 0x10,
In both the cases everything is same except the size of the register, so 
i used PCIE_EXP_LNKCTL macro which is already defined in pci_regs.h


>
> Likewise, PCI_EXP_LNKCTL_RCB is RO for Root Ports, but maybe you're
> telling the device what it should advertise in its Link Control?
You are correct, PCI_EXP_LNKCTL_RCB is RO. I will remove it.
>
> PCI_EXP_LNKCTL_CCC is RW.  But doesn't it depend on the components on
> both ends of the link?  Do you know what device is at the other end?
> I would have assumed that you'd have to start with CCC==0, which
> should be most conservative, then set CCC=1 only if you know both ends
> have a common clock.
PCIe RC and endpoint device are having the common clock so set the CCC=1.
>
>> +	pcie_rc_cfg_wr(lpp, val, PCIE_CAP_OFST + PCI_EXP_LNKCTL);
>> +}
>> +
>> +static void intel_pcie_max_speed_setup(struct intel_pcie_port *lpp)
>> +{
>> +	u32 reg, val;
>> +
>> +	reg = pcie_rc_cfg_rd(lpp, PCIE_CAP_OFST + PCI_EXP_LNKCTL2);
>> +	switch (lpp->link_gen) {
>> +	case PCIE_LINK_SPEED_GEN1:
>> +		reg &= ~PCI_EXP_LNKCTL2_TLS;
>> +		reg |= PCI_EXP_LNKCTL2_HASD|
>> +			PCI_EXP_LNKCTL2_TLS_2_5GT;
>> +		break;
>> +	case PCIE_LINK_SPEED_GEN2:
>> +		reg &= ~PCI_EXP_LNKCTL2_TLS;
>> +		reg |= PCI_EXP_LNKCTL2_HASD|
>> +			PCI_EXP_LNKCTL2_TLS_5_0GT;
>> +		break;
>> +	case PCIE_LINK_SPEED_GEN3:
>> +		reg &= ~PCI_EXP_LNKCTL2_TLS;
>> +		reg |= PCI_EXP_LNKCTL2_HASD|
>> +			PCI_EXP_LNKCTL2_TLS_8_0GT;
>> +		break;
>> +	case PCIE_LINK_SPEED_GEN4:
>> +		reg &= ~PCI_EXP_LNKCTL2_TLS;
>> +		reg |= PCI_EXP_LNKCTL2_HASD|
>> +			PCI_EXP_LNKCTL2_TLS_16_0GT;
>> +		break;
>> +	default:
>> +		/* Use hardware capability */
>> +		val = pcie_rc_cfg_rd(lpp, PCIE_CAP_OFST + PCI_EXP_LNKCAP);
>> +		val = FIELD_GET(PCI_EXP_LNKCAP_SLS, val);
>> +		reg &= ~PCI_EXP_LNKCTL2_HASD;
>> +		reg |= val;
>> +		break;
>> +	}
>> +
>> +	pcie_rc_cfg_wr(lpp, reg, PCIE_CAP_OFST + PCI_EXP_LNKCTL2);
>> +	dw_pcie_link_set_n_fts(&lpp->pci, lpp->n_fts);
> There are other users of of_pci_get_max_link_speed() that look sort of
> similar to this (dra7xx_pcie_establish_link(),
> ks_pcie_set_link_speed(), tegra_pcie_prepare_host()).  Do these *need*
> to be different, or is there something that could be factored out?
dra7xx_pcie_establish_link() and ks_pcie_set_link_speed() are doing 
speed configuration for GEN1 and GEN1 &2 respectively.

intel_pcie_max_speed_setup() can be moved to dwc framework and dra7xx and ks_pcie drivers can call.

>
>> +}
>> +
>> +
>> +
> Remove extra blank lines here.
i will remove it.
>
>> +static void intel_pcie_port_logic_setup(struct intel_pcie_port *lpp)
>> ...
>> +	/* Intel PCIe doesn't configure IO region, so configure
>> +	 * viewport to not to access IO region during register
>> +	 * read write operations.
>> +	 */
> This comment doesn't describe the code.  Is there supposed to be some
> code here that configures the viewport?  Where do we tell the viewport
> not to access IO?
>
> I guess maybe this means something like "tell
> dw_pcie_access_other_conf() not to program an outbound ATU for I/O"?
> I don't know that structure well enough to write this in a way that
> makes sense, but this code doesn't look like it's configuring any
> viewports.
Yes, you are correct. Telling not to program an outbound ATU for IO.
I will update the description.
> Please use usual multi-line comment style, i.e.,
>
>    /*
>     * Intel PCIe ...
>     */
Sure, i will correct it.

Thanks for reviewing and providing the valuable inputs!
Regards,
Dilip

>> +	pci->num_viewport = data->num_viewport;
>> +	dev_info(dev, "Intel PCIe Root Complex Port %d init done\n", lpp->id);
>> +
>> +	return ret;
>> +}