diff --git a/man/man3/fi_domain.3 b/man/man3/fi_domain.3 index 5306dfe98f4..c452d1eee23 100644 --- a/man/man3/fi_domain.3 +++ b/man/man3/fi_domain.3 @@ -15,7 +15,7 @@ . ftr VB CB . ftr VBI CBI .\} -.TH "fi_domain" "3" "2025\-01\-30" "Libfabric Programmer\[cq]s Manual" "#VERSION#" +.TH "fi_domain" "3" "2025\-03\-06" "Libfabric Programmer\[cq]s Manual" "#VERSION#" .hy .SH NAME .PP @@ -444,7 +444,7 @@ The table assumes that all peers enable or disable RM the same. .PP .TS tab(@); -cw(7.7n) cw(16.2n) cw(16.2n) cw(15.4n) cw(14.6n). +cw(5.4n) cw(11.3n) cw(11.3n) cw(10.8n) cw(10.2n) cw(10.8n) cw(10.2n). T{ Resource T}@T{ @@ -452,9 +452,13 @@ DGRAM EP-no RM T}@T{ DGRAM EP-with RM T}@T{ -RDM/MSG EP-no RM +MSG EP-no RM T}@T{ -RDM/MSG EP-with RM +MSG EP-with RM +T}@T{ +RDM EP-no RM +T}@T{ +RDM EP-with RM T} _ T{ @@ -467,6 +471,10 @@ T}@T{ undefined error T}@T{ EAGAIN +T}@T{ +undefined error +T}@T{ +EAGAIN T} T{ Rx Ctx @@ -478,6 +486,10 @@ T}@T{ undefined error T}@T{ EAGAIN +T}@T{ +undefined error +T}@T{ +EAGAIN T} T{ Tx CQ @@ -489,6 +501,10 @@ T}@T{ undefined error T}@T{ EAGAIN +T}@T{ +undefined error +T}@T{ +EAGAIN T} T{ Rx CQ @@ -500,6 +516,10 @@ T}@T{ undefined error T}@T{ EAGAIN +T}@T{ +undefined error +T}@T{ +EAGAIN T} T{ Target EP @@ -511,6 +531,10 @@ T}@T{ transmit error T}@T{ retried +T}@T{ +transmit error +T}@T{ +retried T} T{ No Rx Buffer @@ -522,6 +546,10 @@ T}@T{ transmit error T}@T{ retried +T}@T{ +transmit error +T}@T{ +retried T} T{ Rx Buf Overrun @@ -533,6 +561,10 @@ T}@T{ truncate or error T}@T{ truncate or error +T}@T{ +truncate or error +T}@T{ +truncate or error T} T{ Unmatched RMA @@ -544,6 +576,10 @@ T}@T{ transmit error T}@T{ transmit error +T}@T{ +transmit error +T}@T{ +transmit error T} T{ RMA Overrun @@ -555,6 +591,25 @@ T}@T{ transmit error T}@T{ transmit error +T}@T{ +transmit error +T}@T{ +transmit error +T} +T{ +Unreachable EP +T}@T{ +dropped +T}@T{ +dropped +T}@T{ +not applicable +T}@T{ +not applicable +T}@T{ +transmit error +T}@T{ +transmit error T} .TE .PP @@ -627,15 +682,30 @@ Unlike send operations, RMA operations that attempt to access a memory address that is either not registered for such operations, or attempt to access outside of the target memory region will fail, resulting in a transmit error. -.PP -When a resource management error occurs on an endpoint, the endpoint is -transitioned into a disabled state. -Any operations which have not already completed will fail and be -discarded. -For connectionless endpoints, the endpoint must be re-enabled before it -will accept new data transfer operations. -For connected endpoints, the connection is torn down and must be -re-established. +.TP +\f[I]Unreachable EP\f[R] +Unreachable endpoint is a connectionless specific scenario where +transmit operations are issued to unreachable target endpoints. +Such scenarios include no-route-to-host or down target NIC. +For FI_EP_DGRAM endpoints, transmit operations targeting an unreachable +endpoint will have operation dropped. +For FI_EP_RDM, target operations targeting an unreachable endpoint will +result in a transmit error. +.PP +When a resource management error occurs on an a connected endpoint, the +endpoint will transition into a disabled state and the connection torn +down. +A disabled endpoint will drop any queued or inflight operations. +.PP +The behavior of resource management errors on connectionless endpoints +depends on the type of error. +If RM is disabled and one of the following errors occur, the endpoint +will be disabled: Tx Ctx, Rx Ctx, Tx CQ, or Rx CQ. +For other errors (Target EP, No Rx Buffer, etc.), the operation may +fail, but the endpoint will remain enabled. +A disabled endpoint will drop or fail any queued or inflight operations. +In addition, a disabled endpoint must be re-enabled before it will +accept new data transfer operations. .PP There is one notable restriction on the protections offered by resource management.