Re: [PATCH v2 2/2] phy: cadence-torrent: Add PCIe multilink + USB with same SSC register config for 100 MHz refclk

[Siddharth Vadapalli]

On Tue, Jan 07, 2025 at 10:49:38AM +0530, Siddharth Vadapalli wrote:
> On Mon, Jan 06, 2025 at 11:05:35PM +0100, Christophe JAILLET wrote:
> > Le 06/01/2025 à 14:19, Siddharth Vadapalli a écrit :
> > > From: Swapnil Jakhade <sjakhade@xxxxxxxxxxx>
> > > 
> > > Add register sequences and support for PCIe multilink + USB configuration
> > > for 100MHz reference clock. The same SSC is used for both PCIe and USB.
> > > 
> > > Signed-off-by: Swapnil Jakhade <sjakhade@xxxxxxxxxxx>
> > > Co-developed-by: Siddharth Vadapalli <s-vadapalli@xxxxxx>
> > > Signed-off-by: Siddharth Vadapalli <s-vadapalli@xxxxxx>
> > > ---
> > ...
> > 
> > > @@ -2513,6 +2517,44 @@ int cdns_torrent_phy_configure_multilink(struct cdns_torrent_phy *cdns_phy)
> > >   		phy_t1 = fns(cdns_phy->protocol_bitmask, 0);
> > >   		phy_t2 = fns(cdns_phy->protocol_bitmask, 1);
> > > +
> > > +		/**
> > 
> > Why kernel-doc like start of comment?
> > Only /* ?
> > 
> > > +		 * PCIe Multilink configuration can be supported along with a
> > > +		 * non-PCIe protocol. The existing limitation associated with
> > > +		 * the standalone PCIe Multilink configuration still remains,
> > > +		 * implying that there can be only two links (subnodes) of the
> > > +		 * PHY type PCIe which constitute the PCIe Multilink.
> > > +		 *
> > > +		 * Such configurations are handled by introducing a new protocol
> > > +		 * namely TYPE_PCIE_ML. Both of the PCIe links which have the
> > > +		 * protocol as TYPE_PCIE shall be treated as though the protocol
> > > +		 * corresponding to them is TYPE_PCIE_ML only for the sake of
> > > +		 * configuring the SERDES.
> > > +		 *
> > > +		 * PCIe Multilink configuration can be identified by checking if
> > > +		 * there are exactly two links with phy_type set to TYPE_PCIE.
> > > +		 * phy_t1 and phy_t2 are modified in such cases to support the
> > > +		 * PCIe Multilink configuration with a non-PCIe protocol.
> > > +		 */
> > ...
> > 
> > > @@ -3134,6 +3180,37 @@ static void cdns_torrent_phy_remove(struct platform_device *pdev)
> > >   	cdns_torrent_clk_cleanup(cdns_phy);
> > >   }
> > > +/* Multilink PCIe and USB Same SSC link configuration */
> > > +static struct cdns_reg_pairs ml_pcie_usb_link_cmn_regs[] = {
> > > +	{0x0002, PHY_PLL_CFG},
> > > +	{0x8600, CMN_PDIAG_PLL0_CLK_SEL_M0}
> > > +};
> > 
> > Maybe const could be added here and below?
> > 
> > > +
> > > +static struct cdns_reg_pairs ml_pcie_usb_xcvr_diag_ln_regs[] = {
> > > +	{0x0100, XCVR_DIAG_HSCLK_SEL},
> > > +	{0x0013, XCVR_DIAG_HSCLK_DIV},
> > > +	{0x0812, XCVR_DIAG_PLLDRC_CTRL}
> > > +};
> 
> I will update this patch based on your feedback in the v3 series. Thank
> you for the review.

I have posted the v3 series at:
https://lore.kernel.org/r/20250109121614.754466-1-s-vadapalli@xxxxxx/

Regards,
Siddharth.