diff options
author | Jason Gunthorpe <jgg@nvidia.com> | 2020-11-15 13:43:05 +0200 |
---|---|---|
committer | Jason Gunthorpe <jgg@nvidia.com> | 2020-11-16 16:53:29 -0400 |
commit | b045db62f6f61c2f0f993696abe620379db34163 (patch) | |
tree | 6da3391b0773d818eb9039608f1df1ca86a73bbc /drivers/infiniband/hw/mlx5/mlx5_ib.h | |
parent | RDMA/efa: Remove .create_ah callback assignment (diff) | |
download | linux-dev-b045db62f6f61c2f0f993696abe620379db34163.tar.xz linux-dev-b045db62f6f61c2f0f993696abe620379db34163.zip |
RDMA/mlx5: Use ib_umem_find_best_pgoff() for SRQ
SRQ uses a quantized and scaled page_offset, which is another variation of
ib_umem_find_best_pgsz(). Add mlx5_umem_find_best_quantized_pgoff() to
perform this calculation for each mailbox. A macro shows how the
calculation is directly connected to the mailbox format.
This new routine replaces the limited mlx5_ib_cont_pages() and
mlx5_ib_get_buf_offset() pairing which would reject valid configurations
rather than adjust the page_size to make it work.
In turn this is much more aggressive about choosing large page sizes for
these objects and when THP is enabled it will now often find a single page
solution.
Link: https://lore.kernel.org/r/20201115114311.136250-2-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Diffstat (limited to 'drivers/infiniband/hw/mlx5/mlx5_ib.h')
-rw-r--r-- | drivers/infiniband/hw/mlx5/mlx5_ib.h | 31 |
1 files changed, 31 insertions, 0 deletions
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index bb44080170be..2f08a5b4a438 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -69,6 +69,37 @@ __mlx5_log_page_size_to_bitmap(unsigned int log_pgsz_bits, pgsz_shift), \ iova) +static __always_inline unsigned long +__mlx5_page_offset_to_bitmask(unsigned int page_offset_bits, + unsigned int offset_shift) +{ + unsigned int largest_offset_shift = + min_t(unsigned long, page_offset_bits - 1 + offset_shift, + BITS_PER_LONG - 1); + + return GENMASK(largest_offset_shift, offset_shift); +} + +/* + * QP/CQ/WQ/etc type commands take a page offset that satisifies: + * page_offset_quantized * (page_size/scale) = page_offset + * Which restricts allowed page sizes to ones that satisify the above. + */ +unsigned long __mlx5_umem_find_best_quantized_pgoff( + struct ib_umem *umem, unsigned long pgsz_bitmap, + unsigned int page_offset_bits, u64 pgoff_bitmask, unsigned int scale, + unsigned int *page_offset_quantized); +#define mlx5_umem_find_best_quantized_pgoff(umem, typ, log_pgsz_fld, \ + pgsz_shift, page_offset_fld, \ + scale, page_offset_quantized) \ + __mlx5_umem_find_best_quantized_pgoff( \ + umem, \ + __mlx5_log_page_size_to_bitmap( \ + __mlx5_bit_sz(typ, log_pgsz_fld), pgsz_shift), \ + __mlx5_bit_sz(typ, page_offset_fld), \ + GENMASK(31, order_base_2(scale)), scale, \ + page_offset_quantized) + enum { MLX5_IB_MMAP_OFFSET_START = 9, MLX5_IB_MMAP_OFFSET_END = 255, |