| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAH-L+nMDaDUp6caAqJ6nNqUd3KuXKJ7hnyOgwo3qH5iG+swn-g@mail.gmail.com>
Date: Wed, 18 Oct 2023 22:55:26 +0530
From: Kalesh Anakkur Purayil <kalesh-anakkur.purayil@...adcom.com>
To: Ajay Sharma <sharmaajay@...rosoft.com>
Cc: "sharmaajay@...uxonhyperv.com" <sharmaajay@...uxonhyperv.com>, Long Li <longli@...rosoft.com>,
Jason Gunthorpe <jgg@...pe.ca>, Leon Romanovsky <leon@...nel.org>, Dexuan Cui <decui@...rosoft.com>,
Wei Liu <wei.liu@...nel.org>, "David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
"linux-rdma@...r.kernel.org" <linux-rdma@...r.kernel.org>,
"linux-hyperv@...r.kernel.org" <linux-hyperv@...r.kernel.org>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [EXTERNAL] Re: [Patch v7 4/5] RDMA/mana_ib: Query adapter capabilities
On Wed, Oct 18, 2023 at 10:46 PM Ajay Sharma <sharmaajay@...rosoft.com>
wrote:
>
>
>
>
> *From:* Kalesh Anakkur Purayil <kalesh-anakkur.purayil@...adcom.com>
> *Sent:* Tuesday, October 17, 2023 9:11 PM
> *To:* Ajay Sharma <sharmaajay@...rosoft.com>
> *Cc:* sharmaajay@...uxonhyperv.com; Long Li <longli@...rosoft.com>; Jason
> Gunthorpe <jgg@...pe.ca>; Leon Romanovsky <leon@...nel.org>; Dexuan Cui <
> decui@...rosoft.com>; Wei Liu <wei.liu@...nel.org>; David S. Miller <
> davem@...emloft.net>; Eric Dumazet <edumazet@...gle.com>; Jakub Kicinski <
> kuba@...nel.org>; Paolo Abeni <pabeni@...hat.com>;
> linux-rdma@...r.kernel.org; linux-hyperv@...r.kernel.org;
> netdev@...r.kernel.org; linux-kernel@...r.kernel.org
> *Subject:* Re: [EXTERNAL] Re: [Patch v7 4/5] RDMA/mana_ib: Query adapter
> capabilities
>
>
>
>
>
>
>
> On Tue, Oct 17, 2023 at 11:17 AM Ajay Sharma <sharmaajay@...rosoft.com>
> wrote:
>
>
>
>
>
> On Oct 16, 2023, at 9:32 PM, Kalesh Anakkur Purayil <
> kalesh-anakkur.purayil@...adcom.com> wrote:
>
>
>
> Hi Ajay,
>
>
>
> One comment in line.
>
>
>
> Regards,
>
> Kalesh
>
>
>
> On Tue, Oct 17, 2023 at 3:42 AM <sharmaajay@...uxonhyperv.com> wrote:
>
> From: Ajay Sharma <sharmaajay@...rosoft.com>
>
> Query the adapter capabilities to expose to
> other clients and VF. This checks against
> the user supplied values and protects against
> overflows.
>
> Signed-off-by: Ajay Sharma <sharmaajay@...rosoft.com>
> ---
> drivers/infiniband/hw/mana/device.c | 4 ++
> drivers/infiniband/hw/mana/main.c | 67 ++++++++++++++++++++++------
> drivers/infiniband/hw/mana/mana_ib.h | 53 +++++++++++++++++++++-
> 3 files changed, 110 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/infiniband/hw/mana/device.c
> b/drivers/infiniband/hw/mana/device.c
> index 4077e440657a..e15da43c73a0 100644
> --- a/drivers/infiniband/hw/mana/device.c
> +++ b/drivers/infiniband/hw/mana/device.c
> @@ -97,6 +97,10 @@ static int mana_ib_probe(struct auxiliary_device *adev,
> goto free_error_eq;
> }
>
> + ret = mana_ib_query_adapter_caps(mib_dev);
> + if (ret)
> + ibdev_dbg(&mib_dev->ib_dev, "Failed to get caps, use
> defaults");
>
> [Kalesh]: You are ignoring the failure here and continuing with the IB
> register. When the FW command fails, you won't populate the
> "mib_dev->adapter_caps". Subsequent "mana_ib_query_device" may return stale
> values?
>
> Is that what you want?
>
> It will use default capabilities.
>
> [Kalesh]: Maybe I am missing something here. I could not see that code
> where you are initializing "mib_dev->adapter_caps" with default values.
>
>
>
> [Ajay] : If this call fails , then any other subsequent call to management
> will fail because this is the basis of communication with management. So
> ignoring the failure is fine in this case.
>
[Kalesh]: OK, thank you for the response.
I thought failing mana_ib_probe() would be best in this case as
mana_ib_query_device() will return some uninitialized values.
> +
> ret = ib_register_device(&mib_dev->ib_dev, "mana_%d",
> mdev->gdma_context->dev);
> if (ret)
> diff --git a/drivers/infiniband/hw/mana/main.c
> b/drivers/infiniband/hw/mana/main.c
> index 5b5d7abe79ac..82923475267d 100644
> --- a/drivers/infiniband/hw/mana/main.c
> +++ b/drivers/infiniband/hw/mana/main.c
> @@ -469,20 +469,15 @@ int mana_ib_get_port_immutable(struct ib_device
> *ibdev, u32 port_num,
> int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr
> *props,
> struct ib_udata *uhw)
> {
> - props->max_qp = MANA_MAX_NUM_QUEUES;
> - props->max_qp_wr = MAX_SEND_BUFFERS_PER_QUEUE;
> + struct mana_ib_dev *mib_dev = container_of(ibdev,
> + struct mana_ib_dev, ib_dev);
>
> - /*
> - * max_cqe could be potentially much bigger.
> - * As this version of driver only support RAW QP, set it to the
> same
> - * value as max_qp_wr
> - */
> - props->max_cqe = MAX_SEND_BUFFERS_PER_QUEUE;
> -
> - props->max_mr_size = MANA_IB_MAX_MR_SIZE;
> - props->max_mr = MANA_IB_MAX_MR;
> - props->max_send_sge = MAX_TX_WQE_SGL_ENTRIES;
> - props->max_recv_sge = MAX_RX_WQE_SGL_ENTRIES;
> + props->max_qp = mib_dev->adapter_caps.max_qp_count;
> + props->max_qp_wr = mib_dev->adapter_caps.max_requester_sq_size;
> + props->max_cqe = mib_dev->adapter_caps.max_requester_sq_size;
> + props->max_mr = mib_dev->adapter_caps.max_mr_count;
> + props->max_send_sge = mib_dev->adapter_caps.max_send_wqe_size;
> + props->max_recv_sge = mib_dev->adapter_caps.max_recv_wqe_size;
>
> return 0;
> }
> @@ -601,3 +596,49 @@ int mana_ib_create_error_eq(struct mana_ib_dev
> *mib_dev)
>
> return 0;
> }
> +
> +static void assign_caps(struct mana_ib_adapter_caps *caps,
> + struct mana_ib_query_adapter_caps_resp *resp)
> +{
> + caps->max_sq_id = resp->max_sq_id;
> + caps->max_rq_id = resp->max_rq_id;
> + caps->max_cq_id = resp->max_cq_id;
> + caps->max_qp_count = resp->max_qp_count;
> + caps->max_cq_count = resp->max_cq_count;
> + caps->max_mr_count = resp->max_mr_count;
> + caps->max_pd_count = resp->max_pd_count;
> + caps->max_inbound_read_limit = resp->max_inbound_read_limit;
> + caps->max_outbound_read_limit = resp->max_outbound_read_limit;
> + caps->mw_count = resp->mw_count;
> + caps->max_srq_count = resp->max_srq_count;
> + caps->max_requester_sq_size = resp->max_requester_sq_size;
> + caps->max_responder_sq_size = resp->max_responder_sq_size;
> + caps->max_requester_rq_size = resp->max_requester_rq_size;
> + caps->max_responder_rq_size = resp->max_responder_rq_size;
> + caps->max_send_wqe_size = resp->max_send_wqe_size;
> + caps->max_recv_wqe_size = resp->max_recv_wqe_size;
> + caps->max_inline_data_size = resp->max_inline_data_size;
> +}
> +
> +int mana_ib_query_adapter_caps(struct mana_ib_dev *mib_dev)
> +{
> + struct mana_ib_query_adapter_caps_resp resp = {};
> + struct mana_ib_query_adapter_caps_req req = {};
> + int err;
> +
> + mana_gd_init_req_hdr(&req.hdr, MANA_IB_GET_ADAPTER_CAP,
> sizeof(req),
> + sizeof(resp));
> + req.hdr.resp.msg_version = MANA_IB_GET_ADAPTER_CAP_RESPONSE_V3;
> + req.hdr.dev_id = mib_dev->gc->mana_ib.dev_id;
> +
> + err = mana_gd_send_request(mib_dev->gc, sizeof(req), &req,
> + sizeof(resp), &resp);
> +
> + if (err) {
> + ibdev_err(&mib_dev->ib_dev, "Failed to query adapter caps
> err %d", err);
> + return err;
> + }
> +
> + assign_caps(&mib_dev->adapter_caps, &resp);
> + return 0;
> +}
> diff --git a/drivers/infiniband/hw/mana/mana_ib.h
> b/drivers/infiniband/hw/mana/mana_ib.h
> index 8a652bccd978..6b9406738cb2 100644
> --- a/drivers/infiniband/hw/mana/mana_ib.h
> +++ b/drivers/infiniband/hw/mana/mana_ib.h
> @@ -20,19 +20,41 @@
>
> /* MANA doesn't have any limit for MR size */
> #define MANA_IB_MAX_MR_SIZE U64_MAX
> -
> +#define MANA_IB_GET_ADAPTER_CAP_RESPONSE_V3 3
> /*
> * The hardware limit of number of MRs is greater than maximum number of
> MRs
> * that can possibly represent in 24 bits
> */
> #define MANA_IB_MAX_MR 0xFFFFFFu
>
> +struct mana_ib_adapter_caps {
> + u32 max_sq_id;
> + u32 max_rq_id;
> + u32 max_cq_id;
> + u32 max_qp_count;
> + u32 max_cq_count;
> + u32 max_mr_count;
> + u32 max_pd_count;
> + u32 max_inbound_read_limit;
> + u32 max_outbound_read_limit;
> + u32 mw_count;
> + u32 max_srq_count;
> + u32 max_requester_sq_size;
> + u32 max_responder_sq_size;
> + u32 max_requester_rq_size;
> + u32 max_responder_rq_size;
> + u32 max_send_wqe_size;
> + u32 max_recv_wqe_size;
> + u32 max_inline_data_size;
> +};
> +
> struct mana_ib_dev {
> struct ib_device ib_dev;
> struct gdma_dev *gdma_dev;
> struct gdma_context *gc;
> struct gdma_queue *fatal_err_eq;
> mana_handle_t adapter_handle;
> + struct mana_ib_adapter_caps adapter_caps;
> };
>
> struct mana_ib_wq {
> @@ -96,6 +118,7 @@ struct mana_ib_rwq_ind_table {
> };
>
> enum mana_ib_command_code {
> + MANA_IB_GET_ADAPTER_CAP = 0x30001,
> MANA_IB_CREATE_ADAPTER = 0x30002,
> MANA_IB_DESTROY_ADAPTER = 0x30003,
> };
> @@ -120,6 +143,32 @@ struct mana_ib_destroy_adapter_resp {
> struct gdma_resp_hdr hdr;
> }; /* HW Data */
>
> +struct mana_ib_query_adapter_caps_req {
> + struct gdma_req_hdr hdr;
> +}; /*HW Data */
> +
> +struct mana_ib_query_adapter_caps_resp {
> + struct gdma_resp_hdr hdr;
> + u32 max_sq_id;
> + u32 max_rq_id;
> + u32 max_cq_id;
> + u32 max_qp_count;
> + u32 max_cq_count;
> + u32 max_mr_count;
> + u32 max_pd_count;
> + u32 max_inbound_read_limit;
> + u32 max_outbound_read_limit;
> + u32 mw_count;
> + u32 max_srq_count;
> + u32 max_requester_sq_size;
> + u32 max_responder_sq_size;
> + u32 max_requester_rq_size;
> + u32 max_responder_rq_size;
> + u32 max_send_wqe_size;
> + u32 max_recv_wqe_size;
> + u32 max_inline_data_size;
> +}; /* HW Data */
> +
> int mana_ib_gd_create_dma_region(struct mana_ib_dev *mib_dev,
> struct ib_umem *umem,
> mana_handle_t *gdma_region);
> @@ -194,4 +243,6 @@ int mana_ib_create_adapter(struct mana_ib_dev
> *mib_dev);
>
> int mana_ib_destroy_adapter(struct mana_ib_dev *mib_dev);
>
> +int mana_ib_query_adapter_caps(struct mana_ib_dev *mib_dev);
> +
> #endif
> --
> 2.25.1
>
>
>
>
> --
>
> Regards,
>
> Kalesh A P
>
>
>
>
> --
>
> Regards,
>
> Kalesh A P
>
--
Regards,
Kalesh A P
Content of type "text/html" skipped
Download attachment "smime.p7s" of type "application/pkcs7-signature" (4239 bytes)
Powered by blists - more mailing lists