From: Ma Wupeng mawupeng1@huawei.com
hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I9UDJX
--------------------------------
L0 driver for platforms such Kunpeng 920. This provides interfaces to for user to alloc and lock memory.
Signed-off-by: Ma Wupeng mawupeng1@huawei.com --- drivers/soc/hisilicon/Kconfig | 9 ++ drivers/soc/hisilicon/Makefile | 1 + drivers/soc/hisilicon/hisi_l0.c | 172 ++++++++++++++++++++++++++++++++ 3 files changed, 182 insertions(+) create mode 100644 drivers/soc/hisilicon/hisi_l0.c
diff --git a/drivers/soc/hisilicon/Kconfig b/drivers/soc/hisilicon/Kconfig index 19c4f4602d51..90d8036ae658 100644 --- a/drivers/soc/hisilicon/Kconfig +++ b/drivers/soc/hisilicon/Kconfig @@ -63,4 +63,13 @@ config HISI_L3T
If not sure say no.
+config HISI_L0 + tristate "Add support L0 cache" + depends on ARM64 && HISI_L3T + help + L0 driver for platforms such Kunpeng 920. This provides interfaces to + for user to alloc and lock memory + + If not sure say no. + endmenu diff --git a/drivers/soc/hisilicon/Makefile b/drivers/soc/hisilicon/Makefile index 022d83eccc79..d91b8ed30306 100644 --- a/drivers/soc/hisilicon/Makefile +++ b/drivers/soc/hisilicon/Makefile @@ -6,3 +6,4 @@ obj-$(CONFIG_HISI_HBMCACHE) += hisi_hbmcache.o obj-$(CONFIG_ARM64_PBHA) += pbha.o hisi_l3t-objs := hisi_lockdown.o l3t.o obj-$(CONFIG_HISI_L3T) += hisi_l3t.o +obj-$(CONFIG_HISI_L0) += hisi_l0.o diff --git a/drivers/soc/hisilicon/hisi_l0.c b/drivers/soc/hisilicon/hisi_l0.c new file mode 100644 index 000000000000..0c5d6591eec6 --- /dev/null +++ b/drivers/soc/hisilicon/hisi_l0.c @@ -0,0 +1,172 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) Huawei Technologies Co., Ltd. 2024. All rights reserved. + */ + +#define pr_fmt(fmt) "hisi_l0: " fmt + +#include <linux/fs.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/miscdevice.h> +#include <linux/kallsyms.h> +#include <linux/mm.h> +#include <linux/gfp.h> +#include <linux/mempolicy.h> +#include <linux/pfn_t.h> + +#include "hisi_l3t.h" + +struct l0_vma_data { + struct page *page; + unsigned long size; + int nid; +}; + +static int get_node_node(struct vm_area_struct *vma) +{ + struct mempolicy *pol; + nodemask_t *nmask; + int nid; + + nid = get_vma_policy_node(vma, vma->vm_start, GFP_KERNEL, &pol, &nmask); + if (pol->mode == MPOL_BIND || pol->mode == MPOL_PREFERRED_MANY) + nid = first_node(*nmask); + + return nid; +} + +static vm_fault_t __l0_pmd_fault(struct vm_fault *vmf) +{ + unsigned long pmd_addr = vmf->address & PMD_MASK; + struct l0_vma_data *data; + unsigned long pfn; + + data = vmf->vma->vm_private_data; + if (!data) { + pr_err("%s: pid: %d invalid internal data\n", current->comm, + current->pid); + return VM_FAULT_SIGBUS; + } + + if (PAGE_ALIGN(vmf->vma->vm_end - vmf->vma->vm_start) != data->size) { + pr_err("%s: pid: %d invalid vma size, start: %#lx, end: %#lx, size: %#lx\n", + current->comm, current->pid, vmf->vma->vm_start, + vmf->vma->vm_end, data->size); + return VM_FAULT_SIGBUS; + } + + /* if we are outside of the VMA */ + if (pmd_addr < vmf->vma->vm_start || + (pmd_addr + PMD_SIZE) > vmf->vma->vm_end) + return VM_FAULT_SIGBUS; + + pfn = page_to_pfn(data->page) + + ((pmd_addr - vmf->vma->vm_start) >> PAGE_SHIFT); + + return vmf_insert_pfn_pmd(vmf, pfn_to_pfn_t(pfn), vmf->flags & FAULT_FLAG_WRITE); +} + +static vm_fault_t l0_huge_fault(struct vm_fault *vmf, + enum page_entry_size pe_size) +{ + pr_debug("%s: pid: %d %s (%#lx - %#lx) size = %d\n", current->comm, + current->pid, + (vmf->flags & FAULT_FLAG_WRITE) ? "write" : "read", + vmf->vma->vm_start, vmf->vma->vm_end, pe_size); + + if (pe_size == PE_SIZE_PMD) + return __l0_pmd_fault(vmf); + + return VM_FAULT_SIGBUS; +} + +static void l0_vma_close(struct vm_area_struct *vma) +{ + struct l0_vma_data *data; + + data = (struct l0_vma_data *)vma->vm_private_data; + if (!data) { + pr_err("invalid internal data\n"); + return; + } + + l3t_shared_unlock(data->nid, page_to_pfn(data->page), data->size); + free_contig_range(page_to_pfn(data->page), data->size >> PAGE_SHIFT); + + kfree(data); +} + +static const struct vm_operations_struct l0_vm_ops = { + .huge_fault = l0_huge_fault, + .close = l0_vma_close, +}; + +static int l0_mmap(struct file *filp, struct vm_area_struct *vma) +{ + unsigned long cont_size = PAGE_ALIGN(vma->vm_end - vma->vm_start); + struct l0_vma_data *data; + int page_cnt, nid, ret; + struct page *page; + + pr_debug("vma_start: %#lx, vma_end: %#lx vma_flags: %pGv\n", + vma->vm_start, vma->vm_end, &vma->vm_flags); + + if ((vma->vm_start % PMD_SIZE) || (vma->vm_end % PMD_SIZE)) { + pr_err("vma addr should align PMD_SIZE\n"); + return -EINVAL; + } + + data = kzalloc(sizeof(struct l0_vma_data), GFP_KERNEL); + if (!data) + return -ENOMEM; + + page_cnt = cont_size >> PAGE_SHIFT; + + nid = get_node_node(vma); + + page = alloc_contig_pages(page_cnt, GFP_KERNEL, nid, NULL); + if (!page) { + pr_err("alloc continue memory failed, size: %#lx\n", cont_size); + ret = -ENOMEM; + goto free_data; + } + + ret = l3t_shared_lock(nid, page_to_pfn(page), cont_size); + if (ret) { + pr_err("l3t lock failed, ret: %d\n", ret); + ret = -ENOMEM; + goto free_page; + } + + data->page = page; + data->size = cont_size; + data->nid = nid; + + vma->vm_ops = &l0_vm_ops; + vma->vm_flags |= (VM_MIXEDMAP | VM_DONTCOPY | VM_DONTEXPAND); + vma->vm_private_data = data; + + return 0; +free_page: + free_contig_range(page_to_pfn(page), page_cnt); +free_data: + kfree(data); + return ret; +} + +static const struct file_operations l0_fops = { + .mmap = l0_mmap, +}; + +static struct miscdevice l0_dev = { + .minor = MISC_DYNAMIC_MINOR, + .name = "hisi_l0", + .fops = &l0_fops, +}; +module_misc_device(l0_dev); + +MODULE_DESCRIPTION("HiSilicon SoC L0 driver"); +MODULE_LICENSE("GPL v2"); +MODULE_AUTHOR("Ma Wupeng mawupeng1@huawei.com");