IBV_POST_SEND(3)
π ibv_post_send β the "yeet your WRs onto the send queue" situation π
okay bestie, welcome to ibv_post_send from the Libibverbs Programmer's Manual (that's the ibv_post_send(3) page, section 3, we're doing man pages today, slay). this is the one where you take a whole little linked list of work requests and yeet them at a queue pair's send queue. it's giving β¨ main character energy β¨ for real.
πΈ NAME πΈ
ibv_post_send β post a list of work requests (WRs) to a send queue
π SYNOPSIS π
#include <infiniband/verbs.h>
int ibv_post_send(struct ibv_qp *qp, struct ibv_send_wr *wr,
struct ibv_send_wr **bad_wr);
π¦ DESCRIPTION π¦
so like, ibv_post_send() posts the linked list of work requests (WRs) starting with wr to the send queue of the queue pair qp. it stops processing WRs from this list at the first failure (the kind that can be detected immediately while requests are being posted), and returns this failing WR through bad_wr. no cap, it just bails at the first L.
the argument wr is an ibv_send_wr struct, as defined in <infiniband/verbs.h>. here's the whole thing, look at her:
struct ibv_send_wr {
uint64_t wr_id; /* User defined WR ID */
struct ibv_send_wr *next; /* Pointer to next WR in list, NULL if last WR */
struct ibv_sge *sg_list; /* Pointer to the s/g array */
int num_sge; /* Size of the s/g array */
enum ibv_wr_opcode opcode; /* Operation type */
unsigned int send_flags; /* Flags of the WR properties */
union {
__be32 imm_data; /* Immediate data (in network byte order) */
uint32_t invalidate_rkey; /* Remote rkey to invalidate */
};
union {
struct {
uint64_t remote_addr; /* Start address of remote memory buffer */
uint32_t rkey; /* Key of the remote Memory Region */
} rdma;
struct {
uint64_t remote_addr; /* Start address of remote memory buffer */
uint64_t compare_add; /* Compare operand */
uint64_t swap; /* Swap operand */
uint32_t rkey; /* Key of the remote Memory Region */
} atomic;
struct {
struct ibv_ah *ah; /* Address handle (AH) for the remote node address */
uint32_t remote_qpn; /* QP number of the destination QP */
uint32_t remote_qkey; /* Q_Key number of the destination QP */
} ud;
} wr;
union {
struct {
uint32_t remote_srqn; /* Number of the remote SRQ */
} xrc;
} qp_type;
union {
struct {
struct ibv_mw *mw; /* Memory window (MW) of type 2 to bind */
uint32_t rkey; /* The desired new rkey of the MW */
struct ibv_mw_bind_info bind_info; /* MW additional bind information */
} bind_mw;
struct {
void *hdr; /* Pointer address of inline header */
uint16_t hdr_sz; /* Inline header size */
uint16_t mss; /* Maximum segment size for each TSO fragment */
} tso;
};
};
and then there's her little sister, ibv_mw_bind_info:
struct ibv_mw_bind_info {
struct ibv_mr *mr; /* The Memory region (MR) to bind the MW to */
uint64_t addr; /* The address the MW should start at */
uint64_t length; /* The length (in bytes) the MW should span */
unsigned int mw_access_flags; /* Access flags to the MW. Use ibv_access_flags */
};
and the scatter/gather icon herself, ibv_sge:
struct ibv_sge {
uint64_t addr; /* Start address of the local memory buffer or number of bytes from the
start of the MR for MRs which are IBV_ACCESS_ZERO_BASED */
uint32_t length; /* Length of the buffer */
uint32_t lkey; /* Key of the local Memory Region */
};
π which opcodes go with which transport service type π
each QP Transport Service Type supports a specific set of opcodes, as shown in the following table. yes it's a whole grid, she's iconic:
OPCODE | IBV_QPT_UD | IBV_QPT_UC | IBV_QPT_RC | IBV_QPT_XRC_SEND | IBV_QPT_RAW_PACKET
----------------------------+------------+------------+------------+------------------+--------------------
IBV_WR_SEND | X | X | X | X | X
IBV_WR_SEND_WITH_IMM | X | X | X | X |
IBV_WR_RDMA_WRITE | | X | X | X |
IBV_WR_RDMA_WRITE_WITH_IMM | | X | X | X |
IBV_WR_RDMA_READ | | | X | X |
IBV_WR_ATOMIC_CMP_AND_SWP | | | X | X |
IBV_WR_ATOMIC_FETCH_AND_ADD | | | X | X |
IBV_WR_LOCAL_INV | | X | X | X |
IBV_WR_BIND_MW | | X | X | X |
IBV_WR_SEND_WITH_INV | | X | X | X |
IBV_WR_TSO | X | | | | X
π
the send_flags moment π
the attribute send_flags describes the properties of the WR. it is either 0 or the bitwise OR of one or more of the following flags:
IBV_SEND_FENCEβ sets the fence indicator. valid only for QPs with Transport Service TypeIBV_QPT_RC.IBV_SEND_SIGNALEDβ sets the completion notification indicator. relevant only if the QP was created withsq_sig_all=0.IBV_SEND_SOLICITEDβ sets the solicited event indicator. valid only for Send and RDMA Write with immediate.IBV_SEND_INLINEβ send data in the given gather list as inline data in a send WQE. valid only for Send and RDMA Write. the L_Key will not be checked. π«ΆIBV_SEND_IP_CSUMβ offloads the IPv4 and TCP/UDP checksum calculation. valid only whendevice_cap_flagsindevice_attrindicates the current QP is supported by checksum offload.
β RETURN VALUE β
ibv_post_send() returns 0 on success, or the value of errno on failure (which indicates the failure reason). so like, zero = slay, anything else = read your errno bestie.
π NOTES π
- the user should not alter or destroy AHs associated with WRs until the request is fully executed and a work completion has been retrieved from the corresponding completion queue (CQ), to avoid unexpected behavior. literally do not touch them, i'm begging. π
- the buffers used by a WR can only be safely reused after the WR request is fully executed and a work completion has been retrieved from the corresponding completion queue (CQ). however, if the
IBV_SEND_INLINEflag was set, the buffer can be reused immediately after the call returns. that's the little treat. β¨ IBV_WR_DRIVER1is an opcode that should be used to issue a specific driver operation.
π§ SEE ALSO π§
ibv_create_qp(3), ibv_create_ah(3), ibv_post_recv(3), ibv_post_srq_recv(3), ibv_poll_cq(3)
π AUTHORS π
- Dotan Barak
<dotanba@gmail.com> - Majd Dibbiny
<majd@mellanox.com> - Yishai Hadas
<yishaih@mellanox.com>
and that's the whole page, bestie π go forth and post responsibly.
this has been libibverbs, dated 2006-10-31 β mwah, love that for us πβ¨
π¦ where this page comes from
- alpine 3.24: rdma-core-doc 61.0-r0 Β· GPL-2.0-only OR BSD-2-Clause
- arch rolling: rdma-core 65.0-1 Β· GPL-2.0-or-later OR LicenseRef-OpenIB.org-BSD-MIT AND GPL-2.0-or-later OR LicenseRef-OpenIB.org-BSD-FreeBSD AND GPL-2.0-or-later OR LicenseRef-Intel-BSD-3-Clause AND GPL-2.0-or-later OR LicenseRef-PathScale-BSD-Patent-License AND CC0-1.0 AND MIT AND BSD-2-Clause AND CPL-1.0 OR BSD-2-Clause OR GPL-2.0-or-later
- centos-stream 10: rdma-core-devel 62.0-3.el10 Β· GPLv2 or BSD
- centos-stream 9: rdma-core-devel 61.0-3.el9 Β· GPLv2 or BSD
- centos-stream 9: rdma-core-devel 61.0-3.el9 Β· GPLv2 or BSD
- debian trixie: libibverbs-dev 56.1-1 Β· BSD-MIT OR GPL-2.0-only
- fedora 43: rdma-core-devel 58.0-4.fc43 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- fedora 43: rdma-core-devel 58.0-4.fc43 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- fedora 44: rdma-core-devel 61.0-2.fc44 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- fedora 44: rdma-core-devel 61.0-2.fc44 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- fedora 44: rdma-core-devel 61.1^20260812git975bdaf-1.fc44 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- fedora 44: rdma-core-devel 61.1^20260812git975bdaf-1.fc44 Β· GPL-2.0-only OR BSD-2-Clause AND BSD-3-Clause
- nixos 26.05: rdma-core 62.0
- nixpkgs unstable: rdma-core 65.0
- opensuse-leap 16.0: rdma-core-devel 56.1-160000.2.2 Β· BSD-2-Clause OR GPL-2.0-only
- opensuse-leap 16.0: rdma-core-devel 56.1-160000.2.2 Β· BSD-2-Clause OR GPL-2.0-only
- opensuse-leap 16.0: rdma-core-devel 56.1-160000.2.2 Β· BSD-2-Clause OR GPL-2.0-only
- opensuse-leap 16.0: rdma-core-devel 56.1-160000.2.2 Β· BSD-2-Clause OR GPL-2.0-only
- ubuntu noble: libibverbs-dev 50.0-2build2 Β· BSD-MIT OR GPL-2.0-only
- ubuntu noble: libibverbs-dev 50.0-2ubuntu0.2 Β· BSD-MIT OR GPL-2.0-only
- ubuntu questing: libibverbs-dev 56.1-1ubuntu1
- ubuntu resolute: libibverbs-dev 61.0-2ubuntu3 Β· BSD-MIT OR GPL-2.0-only
- ubuntu stonking: libibverbs-dev 64.0-1ubuntu1 Β· BSD-MIT OR GPL-2.0-only
- π girlified by deepseek-ai/DeepSeek-V4.1-Flash on : a language model's rewrite of the original, under its terms