[PATCH openEuler-5.10 01/21] RDMA/hns: Enable stash feature of HIP09

newer
[PATCH openEuler-5.10] BMA: Fix...

older
[PATCH OLK-5.10] BMA: Fix format...

Zheng Zengkai

8 Jan 2022 8 Jan '22

6:52 p.m.

From: Yixing Liu <liuyixing1@huawei.com> mainline inclusion from mainline-v5.15-rc1 commit 260f64a40198 category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I4O23M CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?i... ---------------------------------------------------------------------------- The stash feature is enabled by default on HIP09. Fixes: f93c39bc9547 ("RDMA/hns: Add support for QP stash") Fixes: bfefae9f108d ("RDMA/hns: Add support for CQ stash") Link: https://lore.kernel.org/r/1629539607-33217-3-git-send-email-liangwenpeng@hua... Signed-off-by: Yixing Liu <liuyixing1@huawei.com> Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> sigend-off-by: Guofeng Yue <yueguofeng@hisilicon.com> Reviewed-by: Yangyang Li <liyangyang20@huawei.com> Acked-by: Xie XiuQi <xiexiuqi@huawei.com> Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com> --- drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index aea6a6a7b238..4f8738bda0a2 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -2004,6 +2004,7 @@ static void set_default_caps(struct hns_roce_dev *hr_dev) caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM; if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) { + caps->flags |= HNS_ROCE_CAP_FLAG_STASH; caps->max_sq_inline = HNS_ROCE_V3_MAX_SQ_INLINE; } else { caps->max_sq_inline = HNS_ROCE_V2_MAX_SQ_INLINE; -- 2.20.1

Show replies by date

Zheng Zengkai

8 Jan 8 Jan

6:52 p.m.

New subject: [PATCH openEuler-5.10 14/21] etmem scan: fix memleak in vm_idle_read

From: Kemeng Shi <shikemeng@huawei.com> euleros inclusion category: feature feature: etmem bugzilla: https://gitee.com/openeuler/kernel/issues/I4OODH?from=project-issue CVE: NA ------------------------------------------------- free pic before return from vm_idle_read in etmem scan Signed-off-by: Kemeng Shi <shikemeng@huawei.com> Reviewed-by: louhongxiang <louhongxiang@huawei.com> Reviewed-by: Chen Wandun <chenwandun@huawei.com> Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com> --- fs/proc/etmem_scan.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/proc/etmem_scan.c b/fs/proc/etmem_scan.c index 9d615381e9fa..6253d4e5a556 100644 --- a/fs/proc/etmem_scan.c +++ b/fs/proc/etmem_scan.c @@ -891,6 +891,7 @@ static ssize_t vm_idle_read(struct file *file, char *buf, ret = pic->bytes_copied; *ppos = pic->next_hva; out_kvm: + kfree(pic); return ret; } -- 2.20.1

Zheng Zengkai

6:52 p.m.

New subject: [PATCH openEuler-5.10 15/21] etmem: fix concurrent access to export file operations

Zheng Zengkai

6:52 p.m.

New subject: [PATCH openEuler-5.10 16/21] etmem: fix potential UAF when walk ept page table

Zheng Zengkai

6:52 p.m.

New subject: [PATCH openEuler-5.10 17/21] etmem_scan: release CPU after scan walk_step size

From: Kemeng Shi <shikemeng@huawei.com> euleros inclusion category: feature feature: etmem bugzilla: https://gitee.com/openeuler/kernel/issues/I4OODH?from=project-issue CVE: NA ------------------------------------------------- Now, we will call cond_resched after scan a full memslot. If we scan a huge memslot, it will take long time before cond_resched. So call cond_resched after scan walk_step size memory. Signed-off-by: Kemeng Shi <shikemeng@huawei.com> Reviewed-by: louhongxiang <louhongxiang@huawei.com> Reviewed-by: Chen Wandun <chenwandun@huawei.com> Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com> --- fs/proc/etmem_scan.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/fs/proc/etmem_scan.c b/fs/proc/etmem_scan.c index e4f5664e4c89..382a87027950 100644 --- a/fs/proc/etmem_scan.c +++ b/fs/proc/etmem_scan.c @@ -106,6 +106,9 @@ static struct kernel_param_ops walk_step_ops = { static unsigned int __read_mostly walk_step = 512; // in PAGE_SIZE module_param_cb(walk_step, &walk_step_ops, &walk_step, 0644); +static unsigned int resched_step = 10; +module_param(resched_step, uint, 0644); + static unsigned long pagetype_size[16] = { [PTE_ACCESSED] = PAGE_SIZE, /* 4k page */ [PMD_ACCESSED] = PMD_SIZE, /* 2M page */ @@ -836,6 +839,7 @@ static int vm_idle_walk_hva_range(struct page_idle_ctrl *pic, unsigned long addr_range; unsigned long va_end; int ret; + int steps; #ifdef CONFIG_X86_64 ret = ept_idle_supports_cpu(pic->kvm); @@ -863,6 +867,7 @@ static int vm_idle_walk_hva_range(struct page_idle_ctrl *pic, } else { pic->gpa_to_hva = start - gpa_addr; gpa_end = gpa_addr + addr_range; + steps = 0; for (; gpa_addr < gpa_end;) { gpa_next = min(gpa_end, gpa_addr + walk_step * PAGE_SIZE); #ifdef CONFIG_ARM64 @@ -874,6 +879,11 @@ static int vm_idle_walk_hva_range(struct page_idle_ctrl *pic, if (ret) break; + + if (++steps >= resched_step) { + cond_resched(); + steps = 0; + } } va_end = pic->gpa_to_hva + gpa_end; } -- 2.20.1

6:52 p.m.

New subject: [PATCH openEuler-5.10 21/21] hugepage: add sysctl for hugepage alloc and mig

From: Kemeng Shi <shikemeng@huawei.com> euleros inclusion category: feature feature: etmem bugzilla: https://gitee.com/openeuler/kernel/issues/I4OODH?from=project-issue CVE: NA ------------------------------------------------- Add /proc/sys/kernel/hugepage_pmem_allocall switch. Set 1 to allowed all memory in pmem could alloc for hugepage. Set 0(default) hugepage alloc is limited by zone watermark as usual. Add /proc/sys/kernel/hugepage_mig_noalloc switch. Set 1 to forbid new hugepage alloc in hugepage migration when hugepage in dest node runs out. Set 0(default) to allow hugepage alloc in hugepage migration as usual. Signed-off-by: Kemeng Shi <shikemeng@huawei.com> Reviewed-by: louhongxiang <louhongxiang@huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com> Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com> --- include/linux/hugetlb.h | 5 +++++ kernel/sysctl.c | 20 ++++++++++++++++++++ mm/hugetlb.c | 7 ++++++- 3 files changed, 31 insertions(+), 1 deletion(-) diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 1b586a324b8e..50101bbdb7cf 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -176,6 +176,11 @@ struct address_space *hugetlb_page_mapping_lock_write(struct page *hpage); extern int sysctl_hugetlb_shm_group; extern struct list_head huge_boot_pages; +#ifdef CONFIG_HUGETLBFS +extern int sysctl_hugetlb_mig_noalloc; +extern int sysctl_hugetlb_pmem_allocall; +#endif + /* arch callbacks */ pte_t *huge_pte_alloc(struct mm_struct *mm, diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 8c290fcfa32f..9b0fde36de8c 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -3324,6 +3324,26 @@ static struct ctl_table vm_table[] = { .extra1 = SYSCTL_ZERO, .extra2 = SYSCTL_ONE, }, +#endif +#ifdef CONFIG_HUGETLBFS + { + .procname = "hugepage_mig_noalloc", + .data = &sysctl_hugetlb_mig_noalloc, + .maxlen = sizeof(sysctl_hugetlb_mig_noalloc), + .mode = 0600, + .proc_handler = proc_dointvec_minmax, + .extra1 = SYSCTL_ZERO, + .extra2 = SYSCTL_ONE, + }, + { + .procname = "hugepage_pmem_allocall", + .data = &sysctl_hugetlb_pmem_allocall, + .maxlen = sizeof(sysctl_hugetlb_pmem_allocall), + .mode = 0600, + .proc_handler = proc_dointvec_minmax, + .extra1 = SYSCTL_ZERO, + .extra2 = SYSCTL_ONE, + }, #endif { } }; diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 44691ce8a9b8..e553e959bda8 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -81,6 +81,9 @@ DEFINE_SPINLOCK(hugetlb_lock); static int num_fault_mutexes; struct mutex *hugetlb_fault_mutex_table ____cacheline_aligned_in_smp; +int sysctl_hugetlb_mig_noalloc; +int sysctl_hugetlb_pmem_allocall; + /* Forward declaration */ static int hugetlb_acct_memory(struct hstate *h, long delta); @@ -1805,6 +1808,8 @@ static int alloc_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, gfp_t gfp_mask = htlb_alloc_mask(h) | __GFP_THISNODE; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { + if (get_node_type(node) == NODE_TYPE_PMEM && sysctl_hugetlb_pmem_allocall) + gfp_mask |= __GFP_MEMALLOC; page = alloc_fresh_huge_page(h, gfp_mask, node, nodes_allowed, node_alloc_noretry); if (page) @@ -2064,7 +2069,7 @@ struct page *alloc_huge_page_nodemask(struct hstate *h, int preferred_nid, struct page *page; page = dequeue_huge_page_nodemask(h, gfp_mask, preferred_nid, nmask, NULL); - if (page) { + if (page || sysctl_hugetlb_mig_noalloc) { spin_unlock_irq(&hugetlb_lock); return page; } -- 2.20.1

1144

Age (days ago)

1144

Last active (days ago)

List overview

20 comments

1 participants

participants (1)

Zheng Zengkai

[PATCH openEuler-5.10 01/21] RDMA/hns: Enable stash feature of HIP09

tags

participants (1)