group driver that manages life cycle of a bunch of leaf driver instances and bridges them with root. Signed-off-by: Sonal Santan <sonal.santan@xxxxxxxxxx> Signed-off-by: Max Zhen <max.zhen@xxxxxxxxxx> Signed-off-by: Lizhi Hou <lizhi.hou@xxxxxxxxxx> Reviewed-by: Tom Rix <trix@xxxxxxxxxx> --- drivers/fpga/xrt/include/group.h | 25 +++ drivers/fpga/xrt/lib/group.c | 278 +++++++++++++++++++++++++++++++ 2 files changed, 303 insertions(+) create mode 100644 drivers/fpga/xrt/include/group.h create mode 100644 drivers/fpga/xrt/lib/group.c diff --git a/drivers/fpga/xrt/include/group.h b/drivers/fpga/xrt/include/group.h new file mode 100644 index 000000000000..09e9d03f53fe --- /dev/null +++ b/drivers/fpga/xrt/include/group.h @@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Copyright (C) 2020-2021 Xilinx, Inc. + * + * Authors: + * Cheng Zhen <maxz@xxxxxxxxxx> + */ + +#ifndef _XRT_GROUP_H_ +#define _XRT_GROUP_H_ + +#include "xleaf.h" + +/* + * Group driver leaf calls. + */ +enum xrt_group_leaf_cmd { + XRT_GROUP_GET_LEAF = XRT_XLEAF_CUSTOM_BASE, /* See comments in xleaf.h */ + XRT_GROUP_PUT_LEAF, + XRT_GROUP_INIT_CHILDREN, + XRT_GROUP_FINI_CHILDREN, + XRT_GROUP_TRIGGER_EVENT, +}; + +#endif /* _XRT_GROUP_H_ */ diff --git a/drivers/fpga/xrt/lib/group.c b/drivers/fpga/xrt/lib/group.c new file mode 100644 index 000000000000..b45f05449e0b --- /dev/null +++ b/drivers/fpga/xrt/lib/group.c @@ -0,0 +1,278 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Xilinx Alveo FPGA Group Driver + * + * Copyright (C) 2020-2021 Xilinx, Inc. + * + * Authors: + * Cheng Zhen <maxz@xxxxxxxxxx> + */ + +#include <linux/mod_devicetable.h> +#include "xleaf.h" +#include "subdev_pool.h" +#include "group.h" +#include "metadata.h" +#include "lib-drv.h" + +#define XRT_GRP "xrt_group" + +struct xrt_group { + struct xrt_device *xdev; + struct xrt_subdev_pool leaves; + bool leaves_created; + struct mutex lock; /* lock for group */ +}; + +static int xrt_grp_root_cb(struct device *dev, void *parg, + enum xrt_root_cmd cmd, void *arg) +{ + int rc; + struct xrt_device *xdev = + container_of(dev, struct xrt_device, dev); + struct xrt_group *xg = (struct xrt_group *)parg; + + switch (cmd) { + case XRT_ROOT_GET_LEAF_HOLDERS: { + struct xrt_root_get_holders *holders = + (struct xrt_root_get_holders *)arg; + rc = xrt_subdev_pool_get_holders(&xg->leaves, + holders->xpigh_xdev, + holders->xpigh_holder_buf, + holders->xpigh_holder_buf_len); + break; + } + default: + /* Forward parent call to root. */ + rc = xrt_subdev_root_request(xdev, cmd, arg); + break; + } + + return rc; +} + +/* + * Cut subdev's dtb from group's dtb based on passed-in endpoint descriptor. + * Return the subdev's dtb through dtbp, if found. + */ +static int xrt_grp_cut_subdev_dtb(struct xrt_group *xg, struct xrt_dev_endpoints *eps, + char *grp_dtb, char **dtbp) +{ + int ret, i, ep_count = 0; + char *dtb = NULL; + + ret = xrt_md_create(DEV(xg->xdev), &dtb); + if (ret) + return ret; + + for (i = 0; eps->xse_names[i].ep_name || eps->xse_names[i].compat; i++) { + const char *ep_name = eps->xse_names[i].ep_name; + const char *compat = eps->xse_names[i].compat; + + if (!ep_name) + xrt_md_get_compatible_endpoint(DEV(xg->xdev), grp_dtb, compat, &ep_name); + if (!ep_name) + continue; + + ret = xrt_md_copy_endpoint(DEV(xg->xdev), dtb, grp_dtb, ep_name, compat, NULL); + if (ret) + continue; + xrt_md_del_endpoint(DEV(xg->xdev), grp_dtb, ep_name, compat); + ep_count++; + } + /* Found enough endpoints, return the subdev's dtb. */ + if (ep_count >= eps->xse_min_ep) { + *dtbp = dtb; + return 0; + } + + /* Cleanup - Restore all endpoints that has been deleted, if any. */ + if (ep_count > 0) { + xrt_md_copy_endpoint(DEV(xg->xdev), grp_dtb, dtb, + XRT_MD_NODE_ENDPOINTS, NULL, NULL); + } + vfree(dtb); + *dtbp = NULL; + return 0; +} + +static int xrt_grp_create_leaves(struct xrt_group *xg) +{ + struct xrt_subdev_platdata *pdata = DEV_PDATA(xg->xdev); + struct xrt_dev_endpoints *eps = NULL; + int ret = 0, failed = 0; + enum xrt_subdev_id did; + char *grp_dtb = NULL; + unsigned long mlen; + + if (!pdata) + return -EINVAL; + + mlen = xrt_md_size(DEV(xg->xdev), pdata->xsp_dtb); + if (mlen == XRT_MD_INVALID_LENGTH) { + xrt_err(xg->xdev, "invalid dtb, len %ld", mlen); + return -EINVAL; + } + + mutex_lock(&xg->lock); + + if (xg->leaves_created) { + /* + * This is expected since caller does not keep track of the state of the group + * and may, in some cases, still try to create leaves after it has already been + * created. This special error code will let the caller know what is going on. + */ + mutex_unlock(&xg->lock); + return -EEXIST; + } + + grp_dtb = vmalloc(mlen); + if (!grp_dtb) { + mutex_unlock(&xg->lock); + return -ENOMEM; + } + + /* Create all leaves based on dtb. */ + xrt_info(xg->xdev, "bringing up leaves..."); + memcpy(grp_dtb, pdata->xsp_dtb, mlen); + for (did = 0; did < XRT_SUBDEV_NUM; did++) { + eps = xrt_drv_get_endpoints(did); + while (eps && eps->xse_names) { + char *dtb = NULL; + + ret = xrt_grp_cut_subdev_dtb(xg, eps, grp_dtb, &dtb); + if (ret) { + failed++; + xrt_err(xg->xdev, "failed to cut subdev dtb for drv %s: %d", + xrt_drv_name(did), ret); + } + if (!dtb) { + /* + * No more dtb to cut or bad things happened for this instance, + * switch to the next one. + */ + eps++; + continue; + } + + /* Found a dtb for this instance, let's add it. */ + ret = xrt_subdev_pool_add(&xg->leaves, did, xrt_grp_root_cb, xg, dtb); + if (ret < 0) { + /* + * It is not a fatal error here. Some functionality is not usable + * due to this missing device, but the error can be handled + * when the functionality is used. + */ + failed++; + xrt_err(xg->xdev, "failed to add %s: %d", xrt_drv_name(did), ret); + } + vfree(dtb); + /* Continue searching for the same instance from grp_dtb. */ + } + } + + xg->leaves_created = true; + vfree(grp_dtb); + mutex_unlock(&xg->lock); + return failed == 0 ? 0 : -ECHILD; +} + +static void xrt_grp_remove_leaves(struct xrt_group *xg) +{ + mutex_lock(&xg->lock); + + if (!xg->leaves_created) { + mutex_unlock(&xg->lock); + return; + } + + xrt_info(xg->xdev, "tearing down leaves..."); + xrt_subdev_pool_fini(&xg->leaves); + xg->leaves_created = false; + + mutex_unlock(&xg->lock); +} + +static int xrt_grp_probe(struct xrt_device *xdev) +{ + struct xrt_group *xg; + + xrt_info(xdev, "probing..."); + + xg = devm_kzalloc(&xdev->dev, sizeof(*xg), GFP_KERNEL); + if (!xg) + return -ENOMEM; + + xg->xdev = xdev; + mutex_init(&xg->lock); + xrt_subdev_pool_init(DEV(xdev), &xg->leaves); + xrt_set_drvdata(xdev, xg); + + return 0; +} + +static void xrt_grp_remove(struct xrt_device *xdev) +{ + struct xrt_group *xg = xrt_get_drvdata(xdev); + + xrt_info(xdev, "leaving..."); + xrt_grp_remove_leaves(xg); +} + +static int xrt_grp_leaf_call(struct xrt_device *xdev, u32 cmd, void *arg) +{ + int rc = 0; + struct xrt_group *xg = xrt_get_drvdata(xdev); + + switch (cmd) { + case XRT_XLEAF_EVENT: + /* Simply forward to every child. */ + xrt_subdev_pool_handle_event(&xg->leaves, + (struct xrt_event *)arg); + break; + case XRT_GROUP_GET_LEAF: { + struct xrt_root_get_leaf *get_leaf = + (struct xrt_root_get_leaf *)arg; + + rc = xrt_subdev_pool_get(&xg->leaves, get_leaf->xpigl_match_cb, + get_leaf->xpigl_match_arg, + DEV(get_leaf->xpigl_caller_xdev), + &get_leaf->xpigl_tgt_xdev); + break; + } + case XRT_GROUP_PUT_LEAF: { + struct xrt_root_put_leaf *put_leaf = + (struct xrt_root_put_leaf *)arg; + + rc = xrt_subdev_pool_put(&xg->leaves, put_leaf->xpipl_tgt_xdev, + DEV(put_leaf->xpipl_caller_xdev)); + break; + } + case XRT_GROUP_INIT_CHILDREN: + rc = xrt_grp_create_leaves(xg); + break; + case XRT_GROUP_FINI_CHILDREN: + xrt_grp_remove_leaves(xg); + break; + case XRT_GROUP_TRIGGER_EVENT: + xrt_subdev_pool_trigger_event(&xg->leaves, (enum xrt_events)(uintptr_t)arg); + break; + default: + xrt_err(xdev, "unknown IOCTL cmd %d", cmd); + rc = -EINVAL; + break; + } + return rc; +} + +static struct xrt_driver xrt_group_driver = { + .driver = { + .name = XRT_GRP, + }, + .subdev_id = XRT_SUBDEV_GRP, + .probe = xrt_grp_probe, + .remove = xrt_grp_remove, + .leaf_call = xrt_grp_leaf_call, +}; + +XRT_LEAF_INIT_FINI_FUNC(group); -- 2.27.0