From: Eli Cohen <eli-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
To: Roland Dreier <rdreier-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org>
Cc: RDMA list <linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>
Subject: [PATCH] libmlx4: fix possible inline size
Date: Thu, 16 Sep 2010 14:57:14 +0200 [thread overview]
Message-ID: <20100916125714.GA10857@mtldesk30> (raw)
The current driver checks required inline size by making sure it does not
exceed 1024. This is wrong since the whole WQE is limited to 1008 bytes.
Moreover, a more careful claculation is required to avoid cases where the
application requests inline support in a certain size that when used later
could cause connections to stall due to bad WQEs. This patch takes into account
the size of the WQE, the segements used to create a WQE and the overhead
incured by the inline segments themselves.
Signed-off-by: Eli Cohen <eli-VPRAkNaXOzVS1MOuV/RT9w@public.gmane.org>
---
src/verbs.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++-----
1 files changed, 46 insertions(+), 5 deletions(-)
diff --git a/src/verbs.c b/src/verbs.c
index 1ac1362..69f5c63 100644
--- a/src/verbs.c
+++ b/src/verbs.c
@@ -384,6 +384,51 @@ int mlx4_destroy_srq(struct ibv_srq *srq)
return 0;
}
+static int verify_sizes(struct ibv_qp_init_attr *attr)
+{
+ int size;
+ int nsegs;
+
+ if (attr->cap.max_send_wr > 65536 ||
+ attr->cap.max_recv_wr > 65536 ||
+ attr->cap.max_send_sge > 64 ||
+ attr->cap.max_recv_sge > 64)
+ return -1;
+
+ /*
+ * basic numbers needed to understand the calculation
+ * 1008 is max size of a WQE.
+ * 64 is a cache line
+ * 4 bytes for inline header
+ */
+ if (attr->cap.max_inline_data) {
+ switch (attr->qp_type) {
+ case IBV_QPT_UD:
+ size = 1008 - sizeof(struct mlx4_wqe_ctrl_seg) -
+ sizeof(struct mlx4_wqe_datagram_seg);
+ nsegs = (attr->cap.max_inline_data + 59) / 60;
+ break;
+
+ case IBV_QPT_RC:
+ case IBV_QPT_UC:
+ case IBV_QPT_XRC:
+ size = 1008 - sizeof(struct mlx4_wqe_ctrl_seg) -
+ sizeof(struct mlx4_wqe_raddr_seg) - 4;
+ nsegs = (attr->cap.max_inline_data - 28 + 59) / 60;
+ break;
+
+ default:
+ return 0;
+ }
+
+ size -= nsegs * 4;
+ if (attr->cap.max_inline_data > size)
+ return -1;
+ }
+
+ return 0;
+}
+
struct ibv_qp *mlx4_create_qp(struct ibv_pd *pd, struct ibv_qp_init_attr *attr)
{
struct mlx4_create_qp cmd;
@@ -392,11 +437,7 @@ struct ibv_qp *mlx4_create_qp(struct ibv_pd *pd, struct ibv_qp_init_attr *attr)
int ret;
/* Sanity check QP size before proceeding */
- if (attr->cap.max_send_wr > 65536 ||
- attr->cap.max_recv_wr > 65536 ||
- attr->cap.max_send_sge > 64 ||
- attr->cap.max_recv_sge > 64 ||
- attr->cap.max_inline_data > 1024)
+ if (verify_sizes(attr))
return NULL;
qp = malloc(sizeof *qp);
--
1.7.2.2
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
reply other threads:[~2010-09-16 12:57 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100916125714.GA10857@mtldesk30 \
--to=eli-ldsdmyg8hgv8yrgs2mwiifqbs+8scbdb@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=rdreier-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox