Apparatus and method for deadlock avoidance

A non-transitory computer-readable medium having instructions stored thereon for causing a suitably programmed information processor to perform a deadlock-avoidance method in a parallel-processor system,

• wherein the parallel-processor system includes a plurality of nodes,
• wherein each one of the plurality of nodes includes a node buffer, a processor and local memory,
• wherein the plurality of nodes includes a first node having a first node buffer, a second node having a second node buffer, and a third node having a third node buffer, wherein each one of the plurality of nodes is operatively coupled to a plurality of other nodes over a network, and
• wherein at least one software process executes on each one of the plurality of nodes, the deadlock-avoidance method comprising:
• receiving, in the first node, a first allocation command from a first software process executing in the processor of the first node to reserve a first space in the first node buffer for communication between the first software process executing in the processor of the first network node and other software processes executing in processors of other network nodes;
• reserving the first space for use by the first software process;
• entering a first remote memory access (RMA) request from the first software process into the first node buffer;
• performing a first RMA operation by sending the first RMA request over the network to the second network node, and communicating data between the second network node and the first network node based on the first RMA request;
• removing the first RMA request from the first node buffer;
• un-reserving the first space in the node buffer of the first network node;
• receiving, in the first network node, a second allocation command from a second software process executing in the processor of the first network node, to reserve a second space in the first node buffer for communication between the second software process executing in the processor of the first network node and other software processes executing in processors of other network nodes;
• reserving the second space for use by the second software process;
• entering a second RMA request from the second software process into the first node buffer;
• performing a second RMA operation by sending the second RMA request over the network to the third network node, and communicating data between the third network node and the first network node based on the second RMA request;
• removing the second RMA request from the first node buffer;
• and
• un-reserving the second space in the node buffer of the first node.

A parallel-processor system comprising:

• a plurality of nodes, wherein each one of the plurality of nodes includes a node buffer, a processor and local memory, wherein the plurality of nodes includes a first node that has a first node buffer, a second node that has a second node buffer, and a third node that has a third node buffer, wherein each one of the plurality of nodes is operatively coupled to a plurality of other nodes, and wherein at least one software process executes on each one of the plurality of nodes,
• a reservation unit in the first node configured to receive a first allocation command from a first software process that executes in the processor of the first node, to reserve a first space in the first node buffer for communication between the first software process and other software processes that execute in processors of other network nodes, and based on the first allocation command, to reserve the first space for use by the first software process;
• an RMA unit configured to enter a first remote memory access (RMA) request from the first software process into the first node buffer, to send the first RMA request over the network to the second network node, to communicate data between the second network node and the first network node based on the first RMA request, and to remove the first RMA request from the first node buffer;
• wherein the reservation unit is configured to later un-reserve the first space in the node buffer of the first network node;
• wherein the reservation unit is also configured to receive a second allocation command from a second software process that executes in the processor of the first network node, to reserve a second space in the first node buffer for communication between the second software process of the first network node and other software processes that execute in processors of other network nodes, and based on the second allocation command, to reserve the second space for use by the second software process;
• wherein the RMA unit is also configured to enter a second RMA request from the second software process into the first node buffer, to send the second RMA request over the network to the third network node, to communicate data between the third network node and the first network node based on the second RMA request, and to remove the second RMA request from the first node buffer;
• wherein the reservation unit is configured to later un-reserve the second space in the node buffer of the first network node.