| Message ID | 1501704648-20159-20-git-send-email-longli@exchange.microsoft.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
> -----Original Message----- > From: linux-cifs-owner@vger.kernel.org [mailto:linux-cifs- > owner@vger.kernel.org] On Behalf Of Long Li > Sent: Wednesday, August 2, 2017 4:11 PM > To: Steve French <sfrench@samba.org>; linux-cifs@vger.kernel.org; samba- > technical@lists.samba.org; linux-kernel@vger.kernel.org > Cc: Long Li <longli@microsoft.com> > Subject: [[PATCH v1] 19/37] [CIFS] SMBD: Manage credits on SMBD client and > server > > /* > + * Extend the credits to remote peer > + * This implements [MS-SMBD] 3.1.5.9 > + * The idea is that we should extend credits to remote peer as quickly as > + * it's allowed, to maintain data flow. We allocate as much as receive > + * buffer as possible, and extend the receive credits to remote peer > + * return value: the new credtis being granted. > + */ > +static int manage_credits_prior_sending(struct cifs_rdma_info *info) > +{ > + int ret = 0; > + struct cifs_rdma_response *response; > + int rc; > + > + if (atomic_read(&info->receive_credit_target) > When does the receive_credit_target value change? It seems wasteful to perform an atomic_read() on this local value each time. Tom. -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
> -----Original Message----- > From: Tom Talpey > Sent: Monday, August 14, 2017 1:47 PM > To: Long Li <longli@microsoft.com>; Steve French <sfrench@samba.org>; > linux-cifs@vger.kernel.org; samba-technical@lists.samba.org; linux- > kernel@vger.kernel.org > Subject: RE: [[PATCH v1] 19/37] [CIFS] SMBD: Manage credits on SMBD client > and server > > > -----Original Message----- > > From: linux-cifs-owner@vger.kernel.org [mailto:linux-cifs- > > owner@vger.kernel.org] On Behalf Of Long Li > > Sent: Wednesday, August 2, 2017 4:11 PM > > To: Steve French <sfrench@samba.org>; linux-cifs@vger.kernel.org; > > samba- technical@lists.samba.org; linux-kernel@vger.kernel.org > > Cc: Long Li <longli@microsoft.com> > > Subject: [[PATCH v1] 19/37] [CIFS] SMBD: Manage credits on SMBD client > > and server > > > > /* > > + * Extend the credits to remote peer > > + * This implements [MS-SMBD] 3.1.5.9 > > + * The idea is that we should extend credits to remote peer as > > +quickly as > > + * it's allowed, to maintain data flow. We allocate as much as > > +receive > > + * buffer as possible, and extend the receive credits to remote peer > > + * return value: the new credtis being granted. > > + */ > > +static int manage_credits_prior_sending(struct cifs_rdma_info *info) > > +{ > > + int ret = 0; > > + struct cifs_rdma_response *response; > > + int rc; > > + > > + if (atomic_read(&info->receive_credit_target) > > > When does the receive_credit_target value change? It seems wasteful to > perform an atomic_read() on this local value each time. It could be potentially changed while receiving a SMBD packet, as specified in MS-SMBD 3.1.5.8. I agree with you there is no need to use atomic since this value is not increased or decreased, just being set. Will change it. > > Tom. -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/cifs/cifsrdma.c b/fs/cifs/cifsrdma.c index eb48651..97cde3f 100644 --- a/fs/cifs/cifsrdma.c +++ b/fs/cifs/cifsrdma.c @@ -575,6 +575,46 @@ static int cifs_rdma_post_send_negotiate_req(struct cifs_rdma_info *info) } /* + * Extend the credits to remote peer + * This implements [MS-SMBD] 3.1.5.9 + * The idea is that we should extend credits to remote peer as quickly as + * it's allowed, to maintain data flow. We allocate as much as receive + * buffer as possible, and extend the receive credits to remote peer + * return value: the new credtis being granted. + */ +static int manage_credits_prior_sending(struct cifs_rdma_info *info) +{ + int ret = 0; + struct cifs_rdma_response *response; + int rc; + + if (atomic_read(&info->receive_credit_target) > + atomic_read(&info->receive_credits)) { + while (true) { + response = get_receive_buffer(info); + if (!response) + break; + + response->type = SMBD_TRANSFER_DATA; + response->first_segment = false; + rc = cifs_rdma_post_recv(info, response); + if (rc) { + log_rdma_recv("post_recv failed rc=%d\n", rc); + put_receive_buffer(info, response); + break; + } + + ret++; + } + } + + atomic_add(ret, &info->receive_credits); + log_transport_credit(info); + + return ret; +} + +/* * Send a page * page: the page to send * offset: offset in the page to send @@ -607,6 +647,8 @@ static int cifs_rdma_post_send_page(struct cifs_rdma_info *info, struct page *pa packet = (struct smbd_data_transfer *) request->packet; packet->credits_requested = cpu_to_le16(info->send_credit_target); + packet->credits_granted = + cpu_to_le16(manage_credits_prior_sending(info)); packet->flags = cpu_to_le16(0); packet->reserved = cpu_to_le16(0); @@ -718,6 +760,8 @@ static int cifs_rdma_post_send_empty(struct cifs_rdma_info *info) request->info = info; packet = (struct smbd_data_transfer_no_data *) request->packet; + credits_granted = manage_credits_prior_sending(info); + /* nothing to do? */ if (credits_granted==0 && flags==0) { mempool_free(request, info->request_mempool); @@ -827,6 +871,7 @@ static int cifs_rdma_post_send_data( packet = (struct smbd_data_transfer *) request->packet; packet->credits_requested = cpu_to_le16(info->send_credit_target); + packet->credits_granted = cpu_to_le16(manage_credits_prior_sending(info)); packet->flags = cpu_to_le16(0); packet->reserved = cpu_to_le16(0);