IB/mlx5: Add contiguous ODP support

Currenlty ODP supports only regular MMU pages.
Add ODP support for regions consisting of physically contiguous chunks
of arbitrary order (huge pages for instance) to improve performance.

Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c
index a0c2af96424916510107381c0f17c31e0dc9bc0c..914f212e7ef60652a655682a2d668c9cec19b64a 100644 (file)
--- a/drivers/infiniband/hw/mlx5/mem.c
+++ b/drivers/infiniband/hw/mlx5/mem.c
@@ -61,13 +61,12 @@ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr,
        int entry;
        unsigned long page_shift = umem->page_shift;
 
-       /* With ODP we must always match OS page size. */
        if (umem->odp_data) {
-               *count = ib_umem_page_count(umem);
-               *shift = PAGE_SHIFT;
-               *ncont = *count;
+               *ncont = ib_umem_page_count(umem);
+               *count = *ncont << (page_shift - PAGE_SHIFT);
+               *shift = page_shift;
                if (order)
-                       *order = ilog2(roundup_pow_of_two(*count));
+                       *order = ilog2(roundup_pow_of_two(*ncont));
 
                return;
        }

diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 0d52b72ff99bb4f149213e5fc471ded659f4ff6f..eddabd6e6596fcf228232a9cc9b1d7870021d870 100644 (file)
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -200,7 +200,7 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
         */
 
        for (addr = start; addr < end; addr += BIT(umem->page_shift)) {
-               idx = (addr - ib_umem_start(umem)) / PAGE_SIZE;
+               idx = (addr - ib_umem_start(umem)) >> umem->page_shift;
                /*
                 * Strive to write the MTTs in chunks, but avoid overwriting
                 * non-existing MTTs. The huristic here can be improved to
@@ -218,8 +218,7 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 
                        if (in_block && umr_offset == 0) {
                                mlx5_ib_update_xlt(mr, blk_start_idx,
-                                                  idx - blk_start_idx,
-                                                  PAGE_SHIFT,
+                                                  idx - blk_start_idx, 0,
                                                   MLX5_IB_UPD_XLT_ZAP |
                                                   MLX5_IB_UPD_XLT_ATOMIC);
                                in_block = 0;
@@ -228,8 +227,7 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
        }
        if (in_block)
                mlx5_ib_update_xlt(mr, blk_start_idx,
-                                  idx - blk_start_idx + 1,
-                                  PAGE_SHIFT,
+                                  idx - blk_start_idx + 1, 0,
                                   MLX5_IB_UPD_XLT_ZAP |
                                   MLX5_IB_UPD_XLT_ATOMIC);
        /*
@@ -516,7 +514,7 @@ void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr)
 /*
  * Handle a single data segment in a page-fault WQE or RDMA region.
  *
- * Returns number of pages retrieved on success. The caller may continue to
+ * Returns number of OS pages retrieved on success. The caller may continue to
  * the next data segment.
  * Can return the following error codes:
  * -EAGAIN to designate a temporary error. The caller will abort handling the
@@ -531,13 +529,14 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *dev,
 {
        int srcu_key;
        unsigned int current_seq = 0;
-       u64 start_idx;
+       u64 start_idx, page_mask;
        int npages = 0, ret = 0;
        struct mlx5_ib_mr *mr;
        u64 access_mask = ODP_READ_ALLOWED_BIT;
        struct ib_umem_odp *odp;
        int implicit = 0;
        size_t size;
+       int page_shift;
 
        srcu_key = srcu_read_lock(&dev->mr_srcu);
        mr = mlx5_ib_odp_find_mr_lkey(dev, key);
@@ -583,6 +582,9 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *dev,
                odp = mr->umem->odp_data;
        }
 
+       page_shift = mr->umem->page_shift;
+       page_mask = ~(BIT(page_shift) - 1);
+
 next_mr:
        current_seq = READ_ONCE(odp->notifiers_seq);
        /*
@@ -592,7 +594,7 @@ next_mr:
        smp_rmb();
 
        size = min_t(size_t, bcnt, ib_umem_end(odp->umem) - io_virt);
-       start_idx = (io_virt - (mr->mmkey.iova & PAGE_MASK)) >> PAGE_SHIFT;
+       start_idx = (io_virt - (mr->mmkey.iova & page_mask)) >> page_shift;
 
        if (mr->umem->writable)
                access_mask |= ODP_WRITE_ALLOWED_BIT;
@@ -614,7 +616,7 @@ next_mr:
                         * checks this.
                         */
                        ret = mlx5_ib_update_xlt(mr, start_idx, np,
-                                                PAGE_SHIFT,
+                                                page_shift,
                                                 MLX5_IB_UPD_XLT_ATOMIC);
                } else {
                        ret = -EAGAIN;
@@ -625,14 +627,14 @@ next_mr:
                                mlx5_ib_err(dev, "Failed to update mkey page tables\n");
                        goto srcu_unlock;
                }
-
                if (bytes_mapped) {
-                       u32 new_mappings = np * PAGE_SIZE -
-                               (io_virt - round_down(io_virt, PAGE_SIZE));
+                       u32 new_mappings = (np << page_shift) -
+                               (io_virt - round_down(io_virt,
+                                                     1 << page_shift));
                        *bytes_mapped += min_t(u32, new_mappings, size);
                }
 
-               npages += np;
+               npages += np << (page_shift - PAGE_SHIFT);
        }
 
        bcnt -= size;