IB/mlx5: Implement fragmented completion queue (CQ)

The current implementation of create CQ requires contiguous
memory, such requirement is problematic once the memory is
fragmented or the system is low in memory, it causes for
failures in dma_zalloc_coherent().

This patch implements new scheme of fragmented CQ to overcome
this issue by introducing new type: 'struct mlx5_frag_buf_ctrl'
to allocate fragmented buffers, rather than contiguous ones.

Base the Completion Queues (CQs) on this new fragmented buffer.

It fixes following crashes:
kworker/29:0: page allocation failure: order:6, mode:0x80d0
CPU: 29 PID: 8374 Comm: kworker/29:0 Tainted: G OE 3.10.0
Workqueue: ib_cm cm_work_handler [ib_cm]
Call Trace:
[<>] dump_stack+0x19/0x1b
[<>] warn_alloc_failed+0x110/0x180
[<>] __alloc_pages_slowpath+0x6b7/0x725
[<>] __alloc_pages_nodemask+0x405/0x420
[<>] dma_generic_alloc_coherent+0x8f/0x140
[<>] x86_swiotlb_alloc_coherent+0x21/0x50
[<>] mlx5_dma_zalloc_coherent_node+0xad/0x110 [mlx5_core]
[<>] ? mlx5_db_alloc_node+0x69/0x1b0 [mlx5_core]
[<>] mlx5_buf_alloc_node+0x3e/0xa0 [mlx5_core]
[<>] mlx5_buf_alloc+0x14/0x20 [mlx5_core]
[<>] create_cq_kernel+0x90/0x1f0 [mlx5_ib]
[<>] mlx5_ib_create_cq+0x3b0/0x4e0 [mlx5_ib]

Signed-off-by: Yonatan Cohen <yonatanc@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

1 files changed, 25 insertions, 12 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
index 47239bf7bf43..323ffe8bf7e4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
@@ -71,19 +71,24 @@ static void *mlx5_dma_zalloc_coherent_node(struct mlx5_core_dev *dev,
 }
 
 int mlx5_buf_alloc_node(struct mlx5_core_dev *dev, int size,
-			struct mlx5_buf *buf, int node)
+			struct mlx5_frag_buf *buf, int node)
 {
 	dma_addr_t t;
 
 	buf->size = size;
 	buf->npages       = 1;
 	buf->page_shift   = (u8)get_order(size) + PAGE_SHIFT;
-	buf->direct.buf   = mlx5_dma_zalloc_coherent_node(dev, size,
-							  &t, node);
-	if (!buf->direct.buf)
+
+	buf->frags = kzalloc(sizeof(*buf->frags), GFP_KERNEL);
+	if (!buf->frags)
 		return -ENOMEM;
 
-	buf->direct.map = t;
+	buf->frags->buf   = mlx5_dma_zalloc_coherent_node(dev, size,
+							  &t, node);
+	if (!buf->frags->buf)
+		goto err_out;
+
+	buf->frags->map = t;
 
 	while (t & ((1 << buf->page_shift) - 1)) {
 		--buf->page_shift;
@@ -91,18 +96,24 @@ int mlx5_buf_alloc_node(struct mlx5_core_dev *dev, int size,
 	}
 
 	return 0;
+err_out:
+	kfree(buf->frags);
+	return -ENOMEM;
 }
 
-int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, struct mlx5_buf *buf)
+int mlx5_buf_alloc(struct mlx5_core_dev *dev,
+		   int size, struct mlx5_frag_buf *buf)
 {
 	return mlx5_buf_alloc_node(dev, size, buf, dev->priv.numa_node);
 }
-EXPORT_SYMBOL_GPL(mlx5_buf_alloc);
+EXPORT_SYMBOL(mlx5_buf_alloc);
 
-void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_buf *buf)
+void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_frag_buf *buf)
 {
-	dma_free_coherent(&dev->pdev->dev, buf->size, buf->direct.buf,
-			  buf->direct.map);
+	dma_free_coherent(&dev->pdev->dev, buf->size, buf->frags->buf,
+			  buf->frags->map);
+
+	kfree(buf->frags);
 }
 EXPORT_SYMBOL_GPL(mlx5_buf_free);
 
@@ -147,6 +158,7 @@ err_free_buf:
 err_out:
 	return -ENOMEM;
 }
+EXPORT_SYMBOL_GPL(mlx5_frag_buf_alloc_node);
 
 void mlx5_frag_buf_free(struct mlx5_core_dev *dev, struct mlx5_frag_buf *buf)
 {
@@ -162,6 +174,7 @@ void mlx5_frag_buf_free(struct mlx5_core_dev *dev, struct mlx5_frag_buf *buf)
 	}
 	kfree(buf->frags);
 }
+EXPORT_SYMBOL_GPL(mlx5_frag_buf_free);
 
 static struct mlx5_db_pgdir *mlx5_alloc_db_pgdir(struct mlx5_core_dev *dev,
 						 int node)
@@ -275,13 +288,13 @@ void mlx5_db_free(struct mlx5_core_dev *dev, struct mlx5_db *db)
 }
 EXPORT_SYMBOL_GPL(mlx5_db_free);
 
-void mlx5_fill_page_array(struct mlx5_buf *buf, __be64 *pas)
+void mlx5_fill_page_array(struct mlx5_frag_buf *buf, __be64 *pas)
 {
 	u64 addr;
 	int i;
 
 	for (i = 0; i < buf->npages; i++) {
-		addr = buf->direct.map + (i << buf->page_shift);
+		addr = buf->frags->map + (i << buf->page_shift);
 
 		pas[i] = cpu_to_be64(addr);
 	}