From fbdcc8a2d4c2b7f145b034ee7bde7105e66e0a4e Mon Sep 17 00:00:00 2001 From: David Nieto Date: Thu, 7 Dec 2017 16:48:09 -0800 Subject: gpu: nvgpu: Initial Nvlink driver skeleton Adds the skeleton and integration of the GV100 endpoint driver to NVGPU (1) Adds a OS abstraction layer for the internal nvlink structure. (2) Adds linux specific integration with Nvlink core driver. (3) Adds function pointers for nvlink api, initialization and isr process. (4) Adds initial support for minion. (5) Adds new GPU enable properties to handle NVLINK presence (6) Adds new GPU enable properties for SG_PHY bypass (required for NVLINK over PCI) (7) Adds parsing of nvlink vbios structures. (8) Adds logging defines for NVGPU JIRA: EVLR-2328 Change-Id: I0720a165a15c7187892c8c1a0662ec598354ac06 Signed-off-by: David Nieto Reviewed-on: https://git-master.nvidia.com/r/1644708 Reviewed-by: mobile promotions Tested-by: mobile promotions --- drivers/gpu/nvgpu/common/linux/nvlink.c | 531 ++++++++++++++++++++++++++++++++ 1 file changed, 531 insertions(+) create mode 100644 drivers/gpu/nvgpu/common/linux/nvlink.c (limited to 'drivers/gpu/nvgpu/common/linux/nvlink.c') diff --git a/drivers/gpu/nvgpu/common/linux/nvlink.c b/drivers/gpu/nvgpu/common/linux/nvlink.c new file mode 100644 index 00000000..eea6eda7 --- /dev/null +++ b/drivers/gpu/nvgpu/common/linux/nvlink.c @@ -0,0 +1,531 @@ +/* + * Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#include +#include +#include +#include "module.h" + +#ifdef CONFIG_TEGRA_NVLINK +#include +#endif + +#ifdef CONFIG_TEGRA_NVLINK + +/* + * WAR: use this function to find detault link, as only one is supported + * on the library for now + * Returns NVLINK_MAX_LINKS_SW on failure + */ +static u32 __nvgpu_nvlink_get_link(struct nvlink_device *ndev) +{ + u32 link_id; + struct gk20a *g = (struct gk20a *) ndev->priv; + + if (!g) + return NVLINK_MAX_LINKS_SW; + + /* Lets find the detected link */ + if (g->nvlink.initialized_links) + link_id = fls(g->nvlink.initialized_links); + else + return NVLINK_MAX_LINKS_SW; + + if (g->nvlink.links[link_id].remote_info.is_connected) + return link_id; + + return NVLINK_MAX_LINKS_SW; +} +static int nvgpu_nvlink_early_init(struct nvlink_device *ndev) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + int err; + + /* For now master topology is the only one supported */ + if (!ndev->is_master) { + nvgpu_log(g, gpu_dbg_info | gpu_dbg_nvlink, + "dGPU is not master of Nvlink link"); + return -EINVAL; + } + + err = g->ops.nvlink.early_init(g); + return err; +} + +static int nvgpu_nvlink_link_early_init(struct nvlink_device *ndev) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + int err; + u32 link_id; + + /* + * First check the topology and setup connectivity + * HACK: we are only enabling one link for now!!! + */ + link_id = fls(g->nvlink.discovered_links); + g->nvlink.links[link_id].remote_info.is_connected = true; + + err = g->ops.nvlink.link_early_init(g, BIT(link_id)); + + if (err == 0) { + g->nvlink.links[link_id].priv = (void *) &(ndev->link); + ndev->link.priv = (void *) g; + } + return err; +} + +static int nvgpu_nvlink_interface_init(struct nvlink_device *ndev) +{ + int err; + struct gk20a *g = (struct gk20a *) ndev->priv; + + err = g->ops.nvlink.interface_init(g); + return err; +} + +static int nvgpu_nvlink_shutdown(struct nvlink_device *ndev) +{ + int err; + struct gk20a *g = (struct gk20a *) ndev->priv; + + err = g->ops.nvlink.shutdown(g); + return 0; +} + +static int nvgpu_nvlink_reg_init(struct nvlink_device *ndev) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + int err; + + err = g->ops.nvlink.reg_init(g); + + return err; +} + +static u32 nvgpu_nvlink_get_link_mode(struct nvlink_device *ndev) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + u32 mode; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return -EINVAL; + + mode = g->ops.nvlink.link_get_mode(g, link_id); + + switch (mode) { + case nvgpu_nvlink_link_off: + return NVLINK_LINK_OFF; + case nvgpu_nvlink_link_hs: + return NVLINK_LINK_HS; + case nvgpu_nvlink_link_safe: + return NVLINK_LINK_SAFE; + case nvgpu_nvlink_link_fault: + return NVLINK_LINK_FAULT; + case nvgpu_nvlink_link_recovery: + return NVLINK_LINK_RECOVERY; + case nvgpu_nvlink_link_detect: + return NVLINK_LINK_DETECT; + case nvgpu_nvlink_link_reset: + return NVLINK_LINK_RESET; + case nvgpu_nvlink_link_enable_pm: + return NVLINK_LINK_ENABLE_PM; + case nvgpu_nvlink_link_disable_pm: + return NVLINK_LINK_DISABLE_PM; + case nvgpu_nvlink_link_disable_err_detect: + return NVLINK_LINK_DISABLE_ERR_DETECT; + case nvgpu_nvlink_link_lane_disable: + return NVLINK_LINK_LANE_DISABLE; + case nvgpu_nvlink_link_lane_shutdown: + return NVLINK_LINK_LANE_SHUTDOWN; + default: + nvgpu_log(g, gpu_dbg_info | gpu_dbg_nvlink, + "unsupported mode %u", mode); + } + + return NVLINK_LINK_OFF; +} + +static u32 nvgpu_nvlink_get_link_state(struct nvlink_device *ndev) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return -EINVAL; + + return g->ops.nvlink.link_get_state(g, link_id); +} + +static int nvgpu_nvlink_set_link_mode(struct nvlink_device *ndev, u32 mode) +{ + + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + u32 mode_sw; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return -EINVAL; + + switch (mode) { + case NVLINK_LINK_OFF: + mode_sw = nvgpu_nvlink_link_off; + break; + case NVLINK_LINK_HS: + mode_sw = nvgpu_nvlink_link_hs; + break; + case NVLINK_LINK_SAFE: + mode_sw = nvgpu_nvlink_link_safe; + break; + case NVLINK_LINK_FAULT: + mode_sw = nvgpu_nvlink_link_fault; + break; + case NVLINK_LINK_RECOVERY: + mode_sw = nvgpu_nvlink_link_recovery; + break; + case NVLINK_LINK_DETECT: + mode_sw = nvgpu_nvlink_link_detect; + break; + case NVLINK_LINK_RESET: + mode_sw = nvgpu_nvlink_link_reset; + break; + case NVLINK_LINK_ENABLE_PM: + mode_sw = nvgpu_nvlink_link_enable_pm; + break; + case NVLINK_LINK_DISABLE_PM: + mode_sw = nvgpu_nvlink_link_disable_pm; + break; + case NVLINK_LINK_LANE_DISABLE: + mode_sw = nvgpu_nvlink_link_lane_disable; + break; + case NVLINK_LINK_LANE_SHUTDOWN: + mode_sw = nvgpu_nvlink_link_lane_shutdown; + break; + default: + mode_sw = nvgpu_nvlink_link_off; + } + + return g->ops.nvlink.link_set_mode(g, link_id, mode_sw); +} + +static void nvgpu_nvlink_get_tx_sublink_state(struct nvlink_device *ndev, + u32 *state) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return; + if (state) + *state = g->ops.nvlink.get_tx_sublink_state(g, link_id); +} + +static void nvgpu_nvlink_get_rx_sublink_state(struct nvlink_device *ndev, + u32 *state) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return; + if (state) + *state = g->ops.nvlink.get_rx_sublink_state(g, link_id); +} + +static u32 nvgpu_nvlink_get_sublink_mode(struct nvlink_device *ndev, + bool is_rx_sublink) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + u32 mode; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return -EINVAL; + + mode = g->ops.nvlink.get_sublink_mode(g, link_id, is_rx_sublink); + switch (mode) { + case nvgpu_nvlink_sublink_tx_hs: + return NVLINK_TX_HS; + case nvgpu_nvlink_sublink_tx_off: + return NVLINK_TX_OFF; + case nvgpu_nvlink_sublink_tx_single_lane: + return NVLINK_TX_SINGLE_LANE; + case nvgpu_nvlink_sublink_tx_safe: + return NVLINK_TX_SAFE; + case nvgpu_nvlink_sublink_tx_enable_pm: + return NVLINK_TX_ENABLE_PM; + case nvgpu_nvlink_sublink_tx_disable_pm: + return NVLINK_TX_DISABLE_PM; + case nvgpu_nvlink_sublink_tx_common: + return NVLINK_TX_COMMON; + case nvgpu_nvlink_sublink_tx_common_disable: + return NVLINK_TX_COMMON_DISABLE; + case nvgpu_nvlink_sublink_tx_data_ready: + return NVLINK_TX_DATA_READY; + case nvgpu_nvlink_sublink_tx_prbs_en: + return NVLINK_TX_PRBS_EN; + case nvgpu_nvlink_sublink_rx_hs: + return NVLINK_RX_HS; + case nvgpu_nvlink_sublink_rx_enable_pm: + return NVLINK_RX_ENABLE_PM; + case nvgpu_nvlink_sublink_rx_disable_pm: + return NVLINK_RX_DISABLE_PM; + case nvgpu_nvlink_sublink_rx_single_lane: + return NVLINK_RX_SINGLE_LANE; + case nvgpu_nvlink_sublink_rx_safe: + return NVLINK_RX_SAFE; + case nvgpu_nvlink_sublink_rx_off: + return NVLINK_RX_OFF; + case nvgpu_nvlink_sublink_rx_rxcal: + return NVLINK_RX_RXCAL; + default: + nvgpu_log(g, gpu_dbg_nvlink, "Unsupported mode: %u", mode); + break; + } + + if (is_rx_sublink) + return NVLINK_RX_OFF; + return NVLINK_TX_OFF; +} + +static int nvgpu_nvlink_set_sublink_mode(struct nvlink_device *ndev, + bool is_rx_sublink, u32 mode) +{ + struct gk20a *g = (struct gk20a *) ndev->priv; + u32 link_id; + u32 mode_sw; + + link_id = __nvgpu_nvlink_get_link(ndev); + if (link_id == NVLINK_MAX_LINKS_SW) + return -EINVAL; + + if (!is_rx_sublink) { + switch (mode) { + case NVLINK_TX_HS: + mode_sw = nvgpu_nvlink_sublink_tx_hs; + break; + case NVLINK_TX_ENABLE_PM: + mode_sw = nvgpu_nvlink_sublink_tx_enable_pm; + break; + case NVLINK_TX_DISABLE_PM: + mode_sw = nvgpu_nvlink_sublink_tx_disable_pm; + break; + case NVLINK_TX_SINGLE_LANE: + mode_sw = nvgpu_nvlink_sublink_tx_single_lane; + break; + case NVLINK_TX_SAFE: + mode_sw = nvgpu_nvlink_sublink_tx_safe; + break; + case NVLINK_TX_OFF: + mode_sw = nvgpu_nvlink_sublink_tx_off; + break; + case NVLINK_TX_COMMON: + mode_sw = nvgpu_nvlink_sublink_tx_common; + break; + case NVLINK_TX_COMMON_DISABLE: + mode_sw = nvgpu_nvlink_sublink_tx_common_disable; + break; + case NVLINK_TX_DATA_READY: + mode_sw = nvgpu_nvlink_sublink_tx_data_ready; + break; + case NVLINK_TX_PRBS_EN: + mode_sw = nvgpu_nvlink_sublink_tx_prbs_en; + break; + default: + return -EINVAL; + } + } else { + switch (mode) { + case NVLINK_RX_HS: + mode_sw = nvgpu_nvlink_sublink_rx_hs; + break; + case NVLINK_RX_ENABLE_PM: + mode_sw = nvgpu_nvlink_sublink_rx_enable_pm; + break; + case NVLINK_RX_DISABLE_PM: + mode_sw = nvgpu_nvlink_sublink_rx_disable_pm; + break; + case NVLINK_RX_SINGLE_LANE: + mode_sw = nvgpu_nvlink_sublink_rx_single_lane; + break; + case NVLINK_RX_SAFE: + mode_sw = nvgpu_nvlink_sublink_rx_safe; + break; + case NVLINK_RX_OFF: + mode_sw = nvgpu_nvlink_sublink_rx_off; + break; + case NVLINK_RX_RXCAL: + mode_sw = nvgpu_nvlink_sublink_rx_rxcal; + break; + default: + return -EINVAL; + } + } + + return g->ops.nvlink.set_sublink_mode(g, link_id, is_rx_sublink, + mode_sw); +} +#endif + +u32 nvgpu_nvlink_enumerate(struct gk20a *g) +{ + u32 err = -ENODEV; +#ifdef CONFIG_TEGRA_NVLINK + struct nvlink_device *ndev; + + ndev = (struct nvlink_device *) g->nvlink.priv; + + if (ndev) + err = nvlink_enumerate(ndev); +#endif + return err; +} + +u32 nvgpu_nvlink_train(struct gk20a *g, u32 link_id, bool from_off) +{ + u32 err = -ENODEV; +#ifdef CONFIG_TEGRA_NVLINK + struct nvlink_device *ndev; + + ndev = (struct nvlink_device *) g->nvlink.priv; + + if (!ndev) + return -ENODEV; + + /* Check if the link is connected */ + if (!g->nvlink.links[link_id].remote_info.is_connected) + return -ENODEV; + + if (from_off) + return nvlink_transition_intranode_conn_off_to_safe(ndev); + return nvlink_train_intranode_conn_safe_to_hs(ndev); + +#endif + return err; +} + +u32 nvgpu_nvlink_probe(struct gk20a *g) +{ +#ifdef CONFIG_TEGRA_NVLINK + u32 err = 0; + struct device_node *np = nvgpu_get_node(g); + struct device_node *nvlink_np = NULL, *endp_np = NULL; + struct nvlink_device *ndev; + u32 phys_link_id; + + /* Parse DT */ + if (np) { + nvlink_np = of_get_child_by_name(np, "nvidia,nvlink"); + if (nvlink_np) + endp_np = of_get_child_by_name(np, "endpoint"); + } + + if (!endp_np) { + nvgpu_log(g, gpu_dbg_info | gpu_dbg_nvlink, + "No Nvlink DT detected"); + return -ENODEV; + } + + /* Allocating structures */ + ndev = nvgpu_kzalloc(g, sizeof(struct nvlink_device)); + if (!ndev) { + nvgpu_err(g, "OOM while allocating nvlink device struct"); + return -ENOMEM; + } + + ndev->priv = (void *) g; + g->nvlink.priv = (void *) ndev; + + /* Parse DT structure to detect endpoint topology */ + of_property_read_u32(endp_np, "local_dev_id", &ndev->device_id); + of_property_read_u32(endp_np, "local_link_id", &ndev->link.link_id); + ndev->is_master = of_property_read_bool(endp_np, "is_master"); + of_property_read_u32(endp_np, "remote_dev_id", + &ndev->link.remote_dev_info.device_id); + of_property_read_u32(endp_np, "remote_link_id", + &ndev->link.remote_dev_info.link_id); + of_property_read_u32(endp_np, "physical_link", + &phys_link_id); + + g->nvlink.topology_connected_links = BIT(phys_link_id); + + mutex_init(&ndev->init_state_mutex); + + /* Check that we are in dGPU mode */ + if (ndev->device_id != NVLINK_ENDPT_GV100) { + nvgpu_err(g, "Local nvlink device is not dGPU"); + err = -EINVAL; + goto free_nvlink; + } + + err = nvlink_set_init_state(ndev, NVLINK_DEV_OFF); + if (err) { + nvgpu_err(g, "Error initalizing device state to OFF"); + goto free_nvlink; + } + + /* Fill in device struct */ + ndev->dev_ops.dev_early_init = nvgpu_nvlink_early_init; + ndev->dev_ops.dev_interface_init = nvgpu_nvlink_interface_init; + ndev->dev_ops.dev_reg_init = nvgpu_nvlink_reg_init; + ndev->dev_ops.dev_shutdown = nvgpu_nvlink_shutdown; + + /* Fill in the link struct */ + ndev->link.device_id = ndev->device_id; + ndev->link.mode = NVLINK_LINK_OFF; + ndev->link.link_ops.get_link_mode = nvgpu_nvlink_get_link_mode; + ndev->link.link_ops.set_link_mode = nvgpu_nvlink_set_link_mode; + ndev->link.link_ops.get_sublink_mode = nvgpu_nvlink_get_sublink_mode; + ndev->link.link_ops.set_sublink_mode = nvgpu_nvlink_set_sublink_mode; + ndev->link.link_ops.get_link_state = nvgpu_nvlink_get_link_state; + ndev->link.link_ops.get_tx_sublink_state = + nvgpu_nvlink_get_tx_sublink_state; + ndev->link.link_ops.get_rx_sublink_state = + nvgpu_nvlink_get_rx_sublink_state; + ndev->link.link_ops.link_early_init = + nvgpu_nvlink_link_early_init; + + /* Register device with core driver*/ + err = nvlink_register_device(ndev); + if (err) { + nvgpu_err(g, "failed on nvlink device registration"); + goto free_nvlink; + } + + /* Register link with core driver */ + err = nvlink_register_link(&ndev->link); + if (err) { + nvgpu_err(g, "failed on nvlink link registration"); + goto free_nvlink; + } + + /* Enable NVLINK support */ + __nvgpu_set_enabled(g, NVGPU_SUPPORT_NVLINK, true); +free_nvlink: + nvgpu_kfree(g, ndev); + return err; + +#else + return -ENODEV; +#endif +} + -- cgit v1.2.2