| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 15 Oct 2019 14:19:03 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Igor Russkikh <Igor.Russkikh@...antia.com>
Cc: "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"davem@...emloft.net" <davem@...emloft.net>,
"richardcochran@...il.com" <richardcochran@...il.com>,
Egor Pomozov <Egor.Pomozov@...antia.com>,
Dmitry Bezrukov <Dmitry.Bezrukov@...antia.com>,
Simon Edelhaus <sedelhaus@...vell.com>,
Nikita Danilov <Nikita.Danilov@...antia.com>
Subject: Re: [PATCH v2 net-next 10/12] net: aquantia: add support for Phy
access
On Tue, Oct 08, 2019 at 10:56:56AM +0000, Igor Russkikh wrote:
> From: Dmitry Bezrukov <dmitry.bezrukov@...antia.com>
>
> GPIO PIN control and access is done by direct phy manipulation.
> Here we add an aq_phy module which is able to access phy registers
> via MDIO access mailbox.
>
> Access is controlled via HW semaphore.
>
> Co-developed-by: Nikita Danilov <nikita.danilov@...antia.com>
> Signed-off-by: Nikita Danilov <nikita.danilov@...antia.com>
> Signed-off-by: Dmitry Bezrukov <dmitry.bezrukov@...antia.com>
> Signed-off-by: Igor Russkikh <igor.russkikh@...antia.com>
Hi Igor
Is the Atlantic a combined MAC and PHY in one silicon, or are there
two devices? Could the Atlantic MAC be used in combination with for
example a Marvell PHY?
> diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_phy.c b/drivers/net/ethernet/aquantia/atlantic/aq_phy.c
> new file mode 100644
> index 000000000000..51ae921e3e1f
> --- /dev/null
> +++ b/drivers/net/ethernet/aquantia/atlantic/aq_phy.c
> @@ -0,0 +1,147 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/* aQuantia Corporation Network Driver
> + * Copyright (C) 2018-2019 aQuantia Corporation. All rights reserved
> + */
> +
> +#include "aq_phy.h"
> +
> +bool aq_mdio_busy_wait(struct aq_hw_s *aq_hw)
> +{
> + int err = 0;
> + u32 val;
> +
> + err = readx_poll_timeout_atomic(hw_atl_mdio_busy_get, aq_hw,
> + val, val == 0U, 10U, 100000U);
> +
> + if (err < 0)
> + return false;
> +
> + return true;
> +}
> +
> +u16 aq_mdio_read_word(struct aq_hw_s *aq_hw, u16 mmd, u16 addr)
> +{
> + u16 phy_addr = aq_hw->phy_id << 5 | mmd;
> +
> + /* Set Address register. */
> + hw_atl_glb_mdio_iface4_set(aq_hw, (addr & HW_ATL_MDIO_ADDRESS_MSK) <<
> + HW_ATL_MDIO_ADDRESS_SHIFT);
> + /* Send Address command. */
> + hw_atl_glb_mdio_iface2_set(aq_hw, HW_ATL_MDIO_EXECUTE_OPERATION_MSK |
> + (3 << HW_ATL_MDIO_OP_MODE_SHIFT) |
> + ((phy_addr & HW_ATL_MDIO_PHY_ADDRESS_MSK) <<
> + HW_ATL_MDIO_PHY_ADDRESS_SHIFT));
> +
> + aq_mdio_busy_wait(aq_hw);
> +
> + /* Send Read command. */
> + hw_atl_glb_mdio_iface2_set(aq_hw, HW_ATL_MDIO_EXECUTE_OPERATION_MSK |
> + (1 << HW_ATL_MDIO_OP_MODE_SHIFT) |
> + ((phy_addr & HW_ATL_MDIO_PHY_ADDRESS_MSK) <<
> + HW_ATL_MDIO_PHY_ADDRESS_SHIFT));
> + /* Read result. */
> + aq_mdio_busy_wait(aq_hw);
> +
> + return (u16)hw_atl_glb_mdio_iface5_get(aq_hw);
> +}
> +
> +void aq_mdio_write_word(struct aq_hw_s *aq_hw, u16 mmd, u16 addr, u16 data)
> +{
> + u16 phy_addr = aq_hw->phy_id << 5 | mmd;
> +
> + /* Set Address register. */
> + hw_atl_glb_mdio_iface4_set(aq_hw, (addr & HW_ATL_MDIO_ADDRESS_MSK) <<
> + HW_ATL_MDIO_ADDRESS_SHIFT);
> + /* Send Address command. */
> + hw_atl_glb_mdio_iface2_set(aq_hw, HW_ATL_MDIO_EXECUTE_OPERATION_MSK |
> + (3 << HW_ATL_MDIO_OP_MODE_SHIFT) |
> + ((phy_addr & HW_ATL_MDIO_PHY_ADDRESS_MSK) <<
> + HW_ATL_MDIO_PHY_ADDRESS_SHIFT));
> +
> + aq_mdio_busy_wait(aq_hw);
> +
> + hw_atl_glb_mdio_iface3_set(aq_hw, (data & HW_ATL_MDIO_WRITE_DATA_MSK) <<
> + HW_ATL_MDIO_WRITE_DATA_SHIFT);
> + /* Send Write command. */
> + hw_atl_glb_mdio_iface2_set(aq_hw, HW_ATL_MDIO_EXECUTE_OPERATION_MSK |
> + (2 << HW_ATL_MDIO_OP_MODE_SHIFT) |
> + ((phy_addr & HW_ATL_MDIO_PHY_ADDRESS_MSK) <<
> + HW_ATL_MDIO_PHY_ADDRESS_SHIFT));
> +
> + aq_mdio_busy_wait(aq_hw);
> +}
> +
> +u16 aq_phy_read_reg(struct aq_hw_s *aq_hw, u16 mmd, u16 address)
> +{
> + int err = 0;
> + u32 val;
> +
> + err = readx_poll_timeout_atomic(hw_atl_sem_mdio_get, aq_hw,
> + val, val == 1U, 10U, 100000U);
> +
> + if (err < 0) {
> + err = 0xffff;
> + goto err_exit;
> + }
> +
> + err = aq_mdio_read_word(aq_hw, mmd, address);
> +
> + hw_atl_reg_glb_cpu_sem_set(aq_hw, 1U, HW_ATL_FW_SM_MDIO);
> +
> +err_exit:
> + return err;
> +}
> +
> +void aq_phy_write_reg(struct aq_hw_s *aq_hw, u16 mmd, u16 address, u16 data)
> +{
> + int err = 0;
> + u32 val;
> +
> + err = readx_poll_timeout_atomic(hw_atl_sem_mdio_get, aq_hw,
> + val, val == 1U, 10U, 100000U);
> + if (err < 0)
> + return;
> +
> + aq_mdio_write_word(aq_hw, mmd, address, data);
> + hw_atl_reg_glb_cpu_sem_set(aq_hw, 1U, HW_ATL_FW_SM_MDIO);
> +}
You have here the code needed to implement a real Linux MDIO bus
driver. Are the MDIO pins exposed? Could somebody combine the chip
with say a Marvell Ethernet switch? You then need access to the MDIO
bus to control the switch. So by using a Linux MDIO bus driver, you
make it easy for somebody to do that. You can keep with your firmware
mostly driving the PHY.
> +
> +bool aq_phy_init_phy_id(struct aq_hw_s *aq_hw)
> +{
> + u16 val;
> +
> + for (aq_hw->phy_id = 0; aq_hw->phy_id < HW_ATL_PHY_ID_MAX;
> + ++aq_hw->phy_id) {
> + /* PMA Standard Device Identifier 2: Address 1.3 */
> + val = aq_phy_read_reg(aq_hw, MDIO_MMD_PMAPMD, 3);
> +
> + if (val != 0xffff)
> + return true;
> + }
> +
> + return false;
> +}
> +
> +bool aq_phy_init(struct aq_hw_s *aq_hw)
> +{
> + u32 dev_id;
> +
> + if (aq_hw->phy_id == HW_ATL_PHY_ID_MAX)
> + if (!aq_phy_init_phy_id(aq_hw))
> + return false;
> +
> + /* PMA Standard Device Identifier:
> + * Address 1.2 = MSW,
> + * Address 1.3 = LSW
> + */
> + dev_id = aq_phy_read_reg(aq_hw, MDIO_MMD_PMAPMD, 2);
> + dev_id <<= 16;
> + dev_id |= aq_phy_read_reg(aq_hw, MDIO_MMD_PMAPMD, 3);
> +
> + if (dev_id == 0xffffffff) {
> + aq_hw->phy_id = HW_ATL_PHY_ID_MAX;
> + return false;
> + }
For future proofing, should you not check it is actually one of your
PHYs?
Thanks
Andrew
Powered by blists - more mailing lists