| Message ID | 20220203214444.1508-1-kabel@kernel.org |
|---|---|
| Headers | show |
| Series | Armada 3720 comphy native implementation | expand |
On 03-02-22, 22:44, Marek Behún wrote: > Hello Vinod, > > this is v4 of series converting A3720 comphy to native kernel > implementation. > > Tested on Turris MOX and ESPRESSObin (tested all supported PHY modes: > ethernet, PCIe, USB3, SATA). Applied all, thanks
On Friday 25 February 2022 19:13:10 Vinod Koul wrote: > On 03-02-22, 22:44, Marek Behún wrote: > > Hello Vinod, > > > > this is v4 of series converting A3720 comphy to native kernel > > implementation. > > > > Tested on Turris MOX and ESPRESSObin (tested all supported PHY modes: > > ethernet, PCIe, USB3, SATA). > > Applied all, thanks > > -- > ~Vinod Thank you very much!
Hello Vinod, this is v4 of series converting A3720 comphy to native kernel implementation. Tested on Turris MOX and ESPRESSObin (tested all supported PHY modes: ethernet, PCIe, USB3, SATA). Changes since v3: - addressed Vinod's comments, see https://lore.kernel.org/linux-phy/20220203223913.41e9f375@thinkpad/T/#u - changed comment of step 12 in mvebu_a3700_comphy_ethernet_power_on() a little Changes since v2: - removed device-tree changing patch, Gregory applied it on mvebu/dt64 - changed raw spin lock to ordinary spin lock - fixed indentation of gbe_phy_init_fix - use FIELD_PREP() macro Changes since v1 (all changes in patch 2/6): - refactored to use GENMASK() - fixed various register names to conform to Functional Specification - refactored various register names to have uniform naming - refactored register access into one function, which decides whether the access should be direct or indirect, so as to not have all these checks done everywhere in the code - moved around one register definition to have the registers sorted according to their offset - added comments to some register definitions - added support for PHY reset() method - when changing PHY mode, the driver now resets the PHY and its registers to put the PHY in known state - fixed the power_off() method - added spinlock guard when accessing PHY Selector register, since it is shared between the PHYs and so the PHY mutex is insufficient - upon driver probe changed the calls to power_off() to reset() The original cover letter said: The RPC calling to ARM Trusted Firmware implementation of Armada 3720 comphy driver caused quite a number of issues (explained more in patch 2). This series fixes those issues by reimplementing the driver to native kernel implementation. The last 3 commits revert commits in AHCI, USB, and PCI drivers, that were introduced there because of those issues. Pali Rohár (5): phy: marvell: phy-mvebu-a3700-comphy: Remove port from driver configuration phy: marvell: phy-mvebu-a3700-comphy: Add native kernel implementation Revert "ata: ahci: mvebu: Make SATA PHY optional for Armada 3720" Revert "usb: host: xhci: mvebu: make USB 3.0 PHY optional for Armada 3720" Revert "PCI: aardvark: Fix initialization with old Marvell's Arm Trusted Firmware" drivers/ata/ahci.h | 2 - drivers/ata/ahci_mvebu.c | 2 +- drivers/ata/libahci_platform.c | 2 +- drivers/pci/controller/pci-aardvark.c | 4 +- drivers/phy/marvell/phy-mvebu-a3700-comphy.c | 1350 ++++++++++++++++-- drivers/usb/host/xhci-mvebu.c | 42 - drivers/usb/host/xhci-mvebu.h | 6 - drivers/usb/host/xhci-plat.c | 20 +- drivers/usb/host/xhci-plat.h | 1 - 9 files changed, 1226 insertions(+), 203 deletions(-)