Re: [PATCH rdma-next 14/15] RDMA/mlx5: Support new Unordered access flag
From: Jason Gunthorpe <jgg@nvidia.com>
Date: 2026-08-11 19:31:57
Also in:
linux-rdma, lkml
quoted hunk ↗ jump to hunk
@@ -71,9 +71,14 @@ static void set_mkc_access_pd_addr_fields(void *mkc, int acc, u64 start_addr, MLX5_SET(mkc, mkc, lw, !!(acc & IB_ACCESS_LOCAL_WRITE)); MLX5_SET(mkc, mkc, lr, 1); - if (acc & IB_ACCESS_RELAXED_ORDERING) + if (acc & (IB_ACCESS_RELAXED_ORDERING | IB_ACCESS_UNORDERED)) mlx5_core_mkey_set_relaxed_ordering(dev->mdev, mkc); + if ((acc & IB_ACCESS_UNORDERED) && + MLX5_CAP_GEN(dev->mdev, mkc_order_read_after_write)) + MLX5_SET(mkc, mkc, order_read_after_write, + MLX5_MKC_ORDER_READ_AFTER_WRITE_RO);
So we have this nice helper that sets all the bits then we open code other bits? Let's have the helper do everything? Since the new bit is not umr able I would have expected this patch to also introduce some logic in the mkey key system so we are checking that the cached mkeys have the right mkc? (even though on this HW we expect them to all be uniform) -- Jason