[PATCH rdma-next v1 1/2] RDMA/umem: Don't hold mmap_sem for too long

[Leon Romanovsky <leon@xxxxxxxxxx>]

From: Leon Romanovsky <leonro@xxxxxxxxxxxx>

DMA mapping is time consuming operation and doesn't need to be
performed with mmap_sem semaphore is held on. Limit the scope of
that semaphore for accounting part only.

Signed-off-by: Huy Nguyen <huyn@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
---
 drivers/infiniband/core/umem.c | 22 +++++++++++++---------
 1 file changed, 13 insertions(+), 9 deletions(-)

diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c
index 498f59bb4989..5c7ba60848d5 100644
--- a/drivers/infiniband/core/umem.c
+++ b/drivers/infiniband/core/umem.c
@@ -84,7 +84,6 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 	struct ib_umem *umem;
 	struct page **page_list;
 	struct vm_area_struct **vma_list;
-	unsigned long locked;
 	unsigned long lock_limit;
 	unsigned long cur_base;
 	unsigned long npages;
@@ -149,15 +148,16 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 
 	npages = ib_umem_num_pages(umem);
 
-	down_write(&current->mm->mmap_sem);
-
-	locked     = npages + current->mm->pinned_vm;
 	lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
 
-	if ((locked > lock_limit) && !capable(CAP_IPC_LOCK)) {
+	down_write(&current->mm->mmap_sem);
+	current->mm->pinned_vm += npages;
+	if ((current->mm->pinned_vm > lock_limit) && !capable(CAP_IPC_LOCK)) {
+		up_write(&current->mm->mmap_sem);
 		ret = -ENOMEM;
 		goto out;
 	}
+	up_write(&current->mm->mmap_sem);
 
 	cur_base = addr & PAGE_MASK;
 
@@ -177,11 +177,12 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 	sg_list_start = umem->sg_head.sgl;
 
 	while (npages) {
+		down_read(&current->mm->mmap_sem);
 		ret = get_user_pages_longterm(cur_base,
 				     min_t(unsigned long, npages,
 					   PAGE_SIZE / sizeof (struct page *)),
 				     gup_flags, page_list, vma_list);
-
+		up_read(&current->mm->mmap_sem);
 		if (ret < 0)
 			goto out;
 
@@ -189,12 +190,14 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 		cur_base += ret * PAGE_SIZE;
 		npages   -= ret;
 
+		down_read(&current->mm->mmap_sem);
 		for_each_sg(sg_list_start, sg, ret, i) {
 			if (vma_list && !is_vm_hugetlb_page(vma_list[i]))
 				umem->hugetlb = 0;
 
 			sg_set_page(sg, page_list[i], PAGE_SIZE, 0);
 		}
+		up_read(&current->mm->mmap_sem);
 
 		/* preparing for next loop */
 		sg_list_start = sg;
@@ -215,13 +218,14 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 
 out:
 	if (ret < 0) {
+		down_write(&current->mm->mmap_sem);
+		current->mm->pinned_vm -= ib_umem_num_pages(umem);
+		up_write(&current->mm->mmap_sem);
 		if (need_release)
 			__ib_umem_release(context->device, umem, 0);
 		kfree(umem);
-	} else
-		current->mm->pinned_vm = locked;
+	}
 
-	up_write(&current->mm->mmap_sem);
 	if (vma_list)
 		free_page((unsigned long) vma_list);
 	free_page((unsigned long) page_list);
-- 
2.14.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html