Updated nroff-generated man pages

Signed-off-by: OFIWG Bot <[email protected]>

man/man3/fi_endpoint.3

@@ -1,6 +1,6 @@
 .\" Automatically generated by Pandoc 2.9.2.1
 .\"
-.TH "fi_endpoint" "3" "2024\-03\-07" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
+.TH "fi_endpoint" "3" "2024\-03\-21" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
 .hy
 .SH NAME
 .PP
@@ -1779,7 +1779,7 @@ A resource domain was not bound to the endpoint or an attempt was made
 to bind multiple domains.
 .TP
 \f[I]-FI_ENOCQ\f[R]
-The endpoint has not been configured with necessary event queue.
+The endpoint has not been configured with necessary completion queue.
 .TP
 \f[I]-FI_EOPBADSTATE\f[R]
 The endpoint\[cq]s state does not permit the requested operation.

man/man3/fi_errno.3

@@ -1,6 +1,6 @@
 .\" Automatically generated by Pandoc 2.9.2.1
 .\"
-.TH "fi_errno" "3" "2022\-12\-09" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
+.TH "fi_errno" "3" "2024\-03\-21" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
 .hy
 .SH NAME
 .PP
@@ -57,12 +57,18 @@ No space left on device
 \f[I]FI_ENOSYS\f[R]
 Function not implemented
 .TP
+\f[I]FI_EWOULDBLOCK\f[R]
+Operation would block
+.TP
 \f[I]FI_ENOMSG\f[R]
 No message of desired type
 .TP
 \f[I]FI_ENODATA\f[R]
 No data available
 .TP
+\f[I]FI_EOVERFLOW\f[R]
+Value too large for defined data type
+.TP
 \f[I]FI_EMSGSIZE\f[R]
 Message too long
 .TP
@@ -90,6 +96,9 @@ Software caused connection abort
 \f[I]FI_ECONNRESET\f[R]
 Connection reset by peer
 .TP
+\f[I]FI_ENOBUFS\f[R]
+No buffer space available
+.TP
 \f[I]FI_EISCONN\f[R]
 Transport endpoint is already connected
 .TP
@@ -105,6 +114,9 @@ Operation timed out
 \f[I]FI_ECONNREFUSED\f[R]
 Connection refused
 .TP
+\f[I]FI_EHOSTDOWN\f[R]
+Host is down
+.TP
 \f[I]FI_EHOSTUNREACH\f[R]
 No route to host
 .TP
@@ -149,6 +161,27 @@ Invalid resource domain
 .TP
 \f[I]FI_ENOCQ\f[R]
 Missing or unavailable completion queue
+.TP
+\f[I]FI_ECRC\f[R]
+CRC error
+.TP
+\f[I]FI_ETRUNC\f[R]
+Truncation error
+.TP
+\f[I]FI_ENOKEY\f[R]
+Required key not available
+.TP
+\f[I]FI_ENOAV\f[R]
+Missing or unavailable address vector
+.TP
+\f[I]FI_EOVERRUN\f[R]
+Queue has been overrun
+.TP
+\f[I]FI_ENORX\f[R]
+Receiver not ready, no receive buffers available
+.TP
+\f[I]FI_ENOMR\f[R]
+Memory registration limit exceeded
 .SH SEE ALSO
 .PP
 \f[C]fabric\f[R](7)

man/man7/fi_cxi.7

@@ -1,7 +1,7 @@
 .\"t
 .\" Automatically generated by Pandoc 2.9.2.1
 .\"
-.TH "fi_cxi" "7" "2024\-02\-01" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
+.TH "fi_cxi" "7" "2024\-03\-21" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
 .hy
 .SH NAME
 .PP
@@ -269,6 +269,19 @@ application processes.
 .IP "7." 3
 Application processes select from the list of available service IDs and
 VNIs to form an authorization key to use for Endpoint allocation.
+.SS Endpoint Protocols
+.PP
+The provider supports multiple endpoint protocols.
+The default protocol is FI_PROTO_CXI and fully supports the messaging
+requirements of parallel applicaitons.
+.PP
+The FI_PROTO_CXI_RNR endpoint protocol is an optional protocol that
+targets client/server environments where send-after-send ordering is not
+required and messaging is generally to pre-posted buffers; FI_MULTI_RECV
+is recommended.
+It utilizes a receiver-not-ready implementation where
+\f[I]FI_CXI_RNR_MAX_TIMEOUT_US\f[R] can be tuned to control the maximum
+retry duration.
 .SS Address Vectors
 .PP
 The CXI provider supports both \f[I]FI_AV_TABLE\f[R] and
@@ -514,6 +527,16 @@ Using Pinned mode avoids any overhead due to network page faults but
 requires all buffers to be backed by physical memory.
 Copy-on-write semantics are broken when using pinned memory.
 See the Fork section for more information.
+.PP
+The CXI provider supports DMABUF for device memory registration.
+If the ROCR and CUDA libraries support it, the CXI provider will default
+to use DMA-buf.
+There may be situations with CUDA that may double the BAR consumption.
+Until this is fixed in the CUDA stack, the environment variable
+\f[I]FI_CXI_DISABLE_DMABUF_CUDA\f[R] can be used to fall back to the
+nvidia peer-memory interface.
+Also, \f[I]FI_CXI_DISABLE_DMABUF_ROCR\f[R] can be used to fall back to
+the amdgpu peer-memory interface.
 .SS Translation Cache
 .PP
 Mapping a buffer for use by the NIC is an expensive operation.
@@ -1300,6 +1323,13 @@ queue becomes empty.
 Default VNI value used only for service IDs where the VNI is not
 restricted.
 .TP
+\f[I]FI_CXI_RNR_MAX_TIMEOUT_US\f[R]
+When using the endpoint FI_PROTO_CXI_RNR protocol, this setting is used
+to control the maximum time from the original posting of the message
+that the message should be retried.
+A value of 0 will return an error completion on the first RNR ack
+status.
+.TP
 \f[I]FI_CXI_EQ_ACK_BATCH_SIZE\f[R]
 Number of EQ events to process before writing an acknowledgement to HW.
 Batching ACKs amortizes the cost of event acknowledgement over multiple

man/man7/fi_setup.7

@@ -1,6 +1,6 @@
 .\" Automatically generated by Pandoc 2.9.2.1
 .\"
-.TH "fi_setup" "7" "2023\-01\-02" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
+.TH "fi_setup" "7" "2024\-03\-21" "Libfabric Programmer\[cq]s Manual" "#VERSION#"
 .hy
 .SH NAME
 .PP
@@ -1118,8 +1118,12 @@ struct fi_cq_err_entry {
 /* Sample error handling */
 struct fi_cq_msg_entry entry;
 struct fi_cq_err_entry err_entry;
+char err_data[256];
 int ret;
 
+err_entry.err_data = err_data;
+err_entry.err_data_size = 256;
+
 ret = fi_cq_read(cq, &entry, 1);
 if (ret == -FI_EAVAIL)
     ret = fi_cq_readerr(cq, &err_entry, 0);