| Message ID | 1523335635-10611-1-git-send-email-yanjun.zhu@oracle.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
Hi, Bart A new patch is ready. Please check it. Thanks a lot. Zhu Yanjun On 2018/4/10 12:47, Zhu Yanjun wrote: > In the function duplicate_request, the reference of skb can be increased > to replace the function skb_clone. > > This will make rxe performace better and save memory. > > CC: Srinivas Eeda <srinivas.eeda@oracle.com> > CC: Junxiao Bi <junxiao.bi@oracle.com> > Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> > --- > V1->V2: Follow Bart's suggestion, refcount_inc is replaced by skb_get. > --- > drivers/infiniband/sw/rxe/rxe_resp.c | 17 +++-------------- > 1 file changed, 3 insertions(+), 14 deletions(-) > > diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c > index a65c996..c4172ed 100644 > --- a/drivers/infiniband/sw/rxe/rxe_resp.c > +++ b/drivers/infiniband/sw/rxe/rxe_resp.c > @@ -1124,24 +1124,13 @@ static enum resp_states duplicate_request(struct rxe_qp *qp, > /* Find the operation in our list of responder resources. */ > res = find_resource(qp, pkt->psn); > if (res) { > - struct sk_buff *skb_copy; > - > - skb_copy = skb_clone(res->atomic.skb, GFP_ATOMIC); > - if (skb_copy) { > - rxe_add_ref(qp); /* for the new SKB */ > - } else { > - pr_warn("Couldn't clone atomic resp\n"); > - rc = RESPST_CLEANUP; > - goto out; > - } > - > + skb_get(res->atomic.skb); > /* Resend the result. */ > rc = rxe_xmit_packet(to_rdev(qp->ibqp.device), qp, > - pkt, skb_copy); > + pkt, res->atomic.skb); > if (rc) { > pr_err("Failed resending result. This flow is not handled - skb ignored\n"); > - rxe_drop_ref(qp); > - kfree_skb(skb_copy); > + kfree_skb(res->atomic.skb); > rc = RESPST_CLEANUP; > goto out; > } -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, 2018-04-10 at 12:47 +0800, Yanjun Zhu wrote:
> A new patch is ready. Please check it.
Hello Zhu,
I will have a look as soon as I have time. But please keep in mind that we
are in the middle of the merge window and that that is the worst possible
time for posting patches. A quote from "On submitting patches": "And during
the merge window reviewer capacity tends to be in short order and there
might be none left for you."
(https://pdfs.semanticscholar.org/aaf6/72a30c68ac457bcbaf963d89f6c42ab50f82.pdf)
Bart.
On Tue, 2018-04-10 at 00:47 -0400, Zhu Yanjun wrote: > In the function duplicate_request, the reference of skb can be increased > to replace the function skb_clone. > > This will make rxe performace better and save memory. Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
On Tue, 2018-04-10 at 23:31 +0000, Bart Van Assche wrote: > On Tue, 2018-04-10 at 00:47 -0400, Zhu Yanjun wrote: > > In the function duplicate_request, the reference of skb can be increased > > to replace the function skb_clone. > > > > This will make rxe performace better and save memory. > > Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com> > > > Thanks, applied to for-next.
diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c index a65c996..c4172ed 100644 --- a/drivers/infiniband/sw/rxe/rxe_resp.c +++ b/drivers/infiniband/sw/rxe/rxe_resp.c @@ -1124,24 +1124,13 @@ static enum resp_states duplicate_request(struct rxe_qp *qp, /* Find the operation in our list of responder resources. */ res = find_resource(qp, pkt->psn); if (res) { - struct sk_buff *skb_copy; - - skb_copy = skb_clone(res->atomic.skb, GFP_ATOMIC); - if (skb_copy) { - rxe_add_ref(qp); /* for the new SKB */ - } else { - pr_warn("Couldn't clone atomic resp\n"); - rc = RESPST_CLEANUP; - goto out; - } - + skb_get(res->atomic.skb); /* Resend the result. */ rc = rxe_xmit_packet(to_rdev(qp->ibqp.device), qp, - pkt, skb_copy); + pkt, res->atomic.skb); if (rc) { pr_err("Failed resending result. This flow is not handled - skb ignored\n"); - rxe_drop_ref(qp); - kfree_skb(skb_copy); + kfree_skb(res->atomic.skb); rc = RESPST_CLEANUP; goto out; }
In the function duplicate_request, the reference of skb can be increased to replace the function skb_clone. This will make rxe performace better and save memory. CC: Srinivas Eeda <srinivas.eeda@oracle.com> CC: Junxiao Bi <junxiao.bi@oracle.com> Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> --- V1->V2: Follow Bart's suggestion, refcount_inc is replaced by skb_get. --- drivers/infiniband/sw/rxe/rxe_resp.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)