aboutsummaryrefslogtreecommitdiff
path: root/drivers/net/ethernet/mellanox/mlx4/icm.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/net/ethernet/mellanox/mlx4/icm.c')
-rw-r--r--drivers/net/ethernet/mellanox/mlx4/icm.c88
1 files changed, 59 insertions, 29 deletions
diff --git a/drivers/net/ethernet/mellanox/mlx4/icm.c b/drivers/net/ethernet/mellanox/mlx4/icm.c
index a9ade1c3cad..97c9b1db1d2 100644
--- a/drivers/net/ethernet/mellanox/mlx4/icm.c
+++ b/drivers/net/ethernet/mellanox/mlx4/icm.c
@@ -93,13 +93,17 @@ void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm, int coherent)
kfree(icm);
}
-static int mlx4_alloc_icm_pages(struct scatterlist *mem, int order, gfp_t gfp_mask)
+static int mlx4_alloc_icm_pages(struct scatterlist *mem, int order,
+ gfp_t gfp_mask, int node)
{
struct page *page;
- page = alloc_pages(gfp_mask, order);
- if (!page)
- return -ENOMEM;
+ page = alloc_pages_node(node, gfp_mask, order);
+ if (!page) {
+ page = alloc_pages(gfp_mask, order);
+ if (!page)
+ return -ENOMEM;
+ }
sg_set_page(mem, page, PAGE_SIZE << order, 0);
return 0;
@@ -130,9 +134,15 @@ struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
/* We use sg_set_buf for coherent allocs, which assumes low memory */
BUG_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
- icm = kmalloc(sizeof *icm, gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
- if (!icm)
- return NULL;
+ icm = kmalloc_node(sizeof(*icm),
+ gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN),
+ dev->numa_node);
+ if (!icm) {
+ icm = kmalloc(sizeof(*icm),
+ gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+ if (!icm)
+ return NULL;
+ }
icm->refcount = 0;
INIT_LIST_HEAD(&icm->chunk_list);
@@ -141,10 +151,17 @@ struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
while (npages > 0) {
if (!chunk) {
- chunk = kmalloc(sizeof *chunk,
- gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
- if (!chunk)
- goto fail;
+ chunk = kmalloc_node(sizeof(*chunk),
+ gfp_mask & ~(__GFP_HIGHMEM |
+ __GFP_NOWARN),
+ dev->numa_node);
+ if (!chunk) {
+ chunk = kmalloc(sizeof(*chunk),
+ gfp_mask & ~(__GFP_HIGHMEM |
+ __GFP_NOWARN));
+ if (!chunk)
+ goto fail;
+ }
sg_init_table(chunk->mem, MLX4_ICM_CHUNK_LEN);
chunk->npages = 0;
@@ -161,7 +178,8 @@ struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
cur_order, gfp_mask);
else
ret = mlx4_alloc_icm_pages(&chunk->mem[chunk->npages],
- cur_order, gfp_mask);
+ cur_order, gfp_mask,
+ dev->numa_node);
if (ret) {
if (--cur_order < 0)
@@ -227,9 +245,11 @@ int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev)
MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
}
-int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj,
+ gfp_t gfp)
{
- int i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+ u32 i = (obj & (table->num_obj - 1)) /
+ (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
int ret = 0;
mutex_lock(&table->mutex);
@@ -240,7 +260,7 @@ int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
}
table->icm[i] = mlx4_alloc_icm(dev, MLX4_TABLE_CHUNK_SIZE >> PAGE_SHIFT,
- (table->lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
+ (table->lowmem ? gfp : GFP_HIGHUSER) |
__GFP_NOWARN, table->coherent);
if (!table->icm[i]) {
ret = -ENOMEM;
@@ -262,16 +282,18 @@ out:
return ret;
}
-void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj)
{
- int i;
+ u32 i;
+ u64 offset;
i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
mutex_lock(&table->mutex);
if (--table->icm[i]->refcount == 0) {
- mlx4_UNMAP_ICM(dev, table->virt + i * MLX4_TABLE_CHUNK_SIZE,
+ offset = (u64) i * MLX4_TABLE_CHUNK_SIZE;
+ mlx4_UNMAP_ICM(dev, table->virt + offset,
MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
mlx4_free_icm(dev, table->icm[i], table->coherent);
table->icm[i] = NULL;
@@ -280,9 +302,11 @@ void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
mutex_unlock(&table->mutex);
}
-void *mlx4_table_find(struct mlx4_icm_table *table, int obj, dma_addr_t *dma_handle)
+void *mlx4_table_find(struct mlx4_icm_table *table, u32 obj,
+ dma_addr_t *dma_handle)
{
- int idx, offset, dma_offset, i;
+ int offset, dma_offset, i;
+ u64 idx;
struct mlx4_icm_chunk *chunk;
struct mlx4_icm *icm;
struct page *page = NULL;
@@ -292,7 +316,7 @@ void *mlx4_table_find(struct mlx4_icm_table *table, int obj, dma_addr_t *dma_han
mutex_lock(&table->mutex);
- idx = (obj & (table->num_obj - 1)) * table->obj_size;
+ idx = (u64) (obj & (table->num_obj - 1)) * table->obj_size;
icm = table->icm[idx / MLX4_TABLE_CHUNK_SIZE];
dma_offset = offset = idx % MLX4_TABLE_CHUNK_SIZE;
@@ -326,13 +350,14 @@ out:
}
int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
- int start, int end)
+ u32 start, u32 end)
{
int inc = MLX4_TABLE_CHUNK_SIZE / table->obj_size;
- int i, err;
+ int err;
+ u32 i;
for (i = start; i <= end; i += inc) {
- err = mlx4_table_get(dev, table, i);
+ err = mlx4_table_get(dev, table, i, GFP_KERNEL);
if (err)
goto fail;
}
@@ -349,22 +374,23 @@ fail:
}
void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
- int start, int end)
+ u32 start, u32 end)
{
- int i;
+ u32 i;
for (i = start; i <= end; i += MLX4_TABLE_CHUNK_SIZE / table->obj_size)
mlx4_table_put(dev, table, i);
}
int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
- u64 virt, int obj_size, int nobj, int reserved,
+ u64 virt, int obj_size, u32 nobj, int reserved,
int use_lowmem, int use_coherent)
{
int obj_per_chunk;
int num_icm;
unsigned chunk_size;
int i;
+ u64 size;
obj_per_chunk = MLX4_TABLE_CHUNK_SIZE / obj_size;
num_icm = (nobj + obj_per_chunk - 1) / obj_per_chunk;
@@ -380,10 +406,12 @@ int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
table->coherent = use_coherent;
mutex_init(&table->mutex);
+ size = (u64) nobj * obj_size;
for (i = 0; i * MLX4_TABLE_CHUNK_SIZE < reserved * obj_size; ++i) {
chunk_size = MLX4_TABLE_CHUNK_SIZE;
- if ((i + 1) * MLX4_TABLE_CHUNK_SIZE > nobj * obj_size)
- chunk_size = PAGE_ALIGN(nobj * obj_size - i * MLX4_TABLE_CHUNK_SIZE);
+ if ((i + 1) * MLX4_TABLE_CHUNK_SIZE > size)
+ chunk_size = PAGE_ALIGN(size -
+ i * MLX4_TABLE_CHUNK_SIZE);
table->icm[i] = mlx4_alloc_icm(dev, chunk_size >> PAGE_SHIFT,
(use_lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
@@ -413,6 +441,8 @@ err:
mlx4_free_icm(dev, table->icm[i], use_coherent);
}
+ kfree(table->icm);
+
return -ENOMEM;
}
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-10 17:53:32 +0000