Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma updates from Jason Gunthorpe:
 "Quite a small cycle this time, even with the rc8. I suppose everyone
  went to sleep over xmas.

   - Minor driver updates for hfi1, cxgb4, erdma, hns, irdma, mlx5, siw,
     mana

   - inline CQE support for hns

   - Have mlx5 display device error codes

   - Pinned DMABUF support for irdma

   - Continued rxe cleanups, particularly converting the MRs to use
     xarray

   - Improvements to what can be cached in the mlx5 mkey cache"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (61 commits)
  IB/mlx5: Extend debug control for CC parameters
  IB/hfi1: Fix sdma.h tx->num_descs off-by-one errors
  IB/hfi1: Fix math bugs in hfi1_can_pin_pages()
  RDMA/irdma: Add support for dmabuf pin memory regions
  RDMA/mlx5: Use query_special_contexts for mkeys
  net/mlx5e: Use query_special_contexts for mkeys
  net/mlx5: Change define name for 0x100 lkey value
  net/mlx5: Expose bits for querying special mkeys
  RDMA/rxe: Fix missing memory barriers in rxe_queue.h
  RDMA/mana_ib: Fix a bug when the PF indicates more entries for registering memory on first packet
  RDMA/rxe: Remove rxe_alloc()
  RDMA/cma: Distinguish between sockaddr_in and sockaddr_in6 by size
  Subject: RDMA/rxe: Handle zero length rdma
  iw_cxgb4: Fix potential NULL dereference in c4iw_fill_res_cm_id_entry()
  RDMA/mlx5: Use rdma_umem_for_each_dma_block()
  RDMA/umem: Remove unused 'work' member from struct ib_umem
  RDMA/irdma: Cap MSIX used to online CPUs + 1
  RDMA/mlx5: Check reg_create() create for errors
  RDMA/restrack: Correct spelling
  RDMA/cxgb4: Fix potential null-ptr-deref in pass_establish()
  ...

1 files changed, 40 insertions, 21 deletions

diff --git a/drivers/infiniband/hw/hfi1/user_pages.c b/drivers/infiniband/hw/hfi1/user_pages.c
index 7bce963e2ae6..36aaedc65145 100644
--- a/drivers/infiniband/hw/hfi1/user_pages.c
+++ b/drivers/infiniband/hw/hfi1/user_pages.c
@@ -29,33 +29,52 @@ MODULE_PARM_DESC(cache_size, "Send and receive side cache size limit (in MB)");
 bool hfi1_can_pin_pages(struct hfi1_devdata *dd, struct mm_struct *mm,
 			u32 nlocked, u32 npages)
 {
-	unsigned long ulimit = rlimit(RLIMIT_MEMLOCK), pinned, cache_limit,
-		size = (cache_size * (1UL << 20)); /* convert to bytes */
-	unsigned int usr_ctxts =
-			dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt;
-	bool can_lock = capable(CAP_IPC_LOCK);
+	unsigned long ulimit_pages;
+	unsigned long cache_limit_pages;
+	unsigned int usr_ctxts;
 
 	/*
-	 * Calculate per-cache size. The calculation below uses only a quarter
-	 * of the available per-context limit. This leaves space for other
-	 * pinning. Should we worry about shared ctxts?
+	 * Perform RLIMIT_MEMLOCK based checks unless CAP_IPC_LOCK is present.
 	 */
-	cache_limit = (ulimit / usr_ctxts) / 4;
-
-	/* If ulimit isn't set to "unlimited" and is smaller than cache_size. */
-	if (ulimit != (-1UL) && size > cache_limit)
-		size = cache_limit;
-
-	/* Convert to number of pages */
-	size = DIV_ROUND_UP(size, PAGE_SIZE);
-
-	pinned = atomic64_read(&mm->pinned_vm);
+	if (!capable(CAP_IPC_LOCK)) {
+		ulimit_pages =
+			DIV_ROUND_DOWN_ULL(rlimit(RLIMIT_MEMLOCK), PAGE_SIZE);
+
+		/*
+		 * Pinning these pages would exceed this process's locked memory
+		 * limit.
+		 */
+		if (atomic64_read(&mm->pinned_vm) + npages > ulimit_pages)
+			return false;
+
+		/*
+		 * Only allow 1/4 of the user's RLIMIT_MEMLOCK to be used for HFI
+		 * caches.  This fraction is then equally distributed among all
+		 * existing user contexts.  Note that if RLIMIT_MEMLOCK is
+		 * 'unlimited' (-1), the value of this limit will be > 2^42 pages
+		 * (2^64 / 2^12 / 2^8 / 2^2).
+		 *
+		 * The effectiveness of this check may be reduced if I/O occurs on
+		 * some user contexts before all user contexts are created.  This
+		 * check assumes that this process is the only one using this
+		 * context (e.g., the corresponding fd was not passed to another
+		 * process for concurrent access) as there is no per-context,
+		 * per-process tracking of pinned pages.  It also assumes that each
+		 * user context has only one cache to limit.
+		 */
+		usr_ctxts = dd->num_rcv_contexts - dd->first_dyn_alloc_ctxt;
+		if (nlocked + npages > (ulimit_pages / usr_ctxts / 4))
+			return false;
+	}
 
-	/* First, check the absolute limit against all pinned pages. */
-	if (pinned + npages >= ulimit && !can_lock)
+	/*
+	 * Pinning these pages would exceed the size limit for this cache.
+	 */
+	cache_limit_pages = cache_size * (1024 * 1024) / PAGE_SIZE;
+	if (nlocked + npages > cache_limit_pages)
 		return false;
 
-	return ((nlocked + npages) <= size) || can_lock;
+	return true;
 }
 
 int hfi1_acquire_user_pages(struct mm_struct *mm, unsigned long vaddr, size_t npages,