Methods and apparatus for allocating access to a host device buffer
View Patent ↗Methods and apparatus for allocating access to a buffer of a host device to buffer data transferred between a controller of the host device and one or more remote devices are disclosed. The host device is configured to couple to each of the one or more remote devices through one or more corresponding dedicated lanes. Buffer access is allocated by determining, for each of one or more remote devices coupled to the host device, a number of dedicated lanes between the host device and each of the one or more remote devices and allocating access to the buffer of the host device for each of the one or more remote devices responsive to the determined number of dedicated lanes.
1. An apparatus for use in a host device to buffer data transferred between a controller of the host device and one or more remote devices, the host device being configured to couple to each of the one or more remote devices through one or more corresponding dedicated lanes, the apparatus comprising:
a buffer that buffers data transferred between one or more remote devices and the controller of the host device when the one or more remote devices are coupled to the host device, wherein each of the one or more remote devices couple to the host device through a corresponding link including one or more dedicated lanes;
a buffer controller coupled to the buffer, the buffer controller configured to allocate access to the buffer by the data transferred between the one or more remote devices and the controller of the host device based on a number of dedicated lanes in each link corresponding to each of the one or more remote devices;
a plurality of link cores coupled to the buffer controller, the data transferred between the controller of the host device and each of the one or more remote devices when coupled to the host device passing through an associated one of the plurality of link cores; and
a plurality of link trainers, each of the plurality of link trainers corresponding to one of the link cores, the link trainers configured to negotiate with the one or more remote devices when coupled to the host device to determine the number of dedicated lanes in each link corresponding to the one or more remote devices.
2. The apparatus of claim 1 , wherein the buffer is a receive buffer comprising an address port and a data port and wherein the buffer controller comprises:
a plurality of pipeline registers, each of the plurality of pipeline registers corresponding to one of the plurality of link cores, the plurality of pipeline registers configured to store data from the link cores;
a receive data multiplexer coupled between the plurality of pipeline registers and the receive buffer, the receive data multiplexer selectively passing data from the pipeline registers responsive to data selection signals; and
a buffer access controller coupled to the plurality of pipeline registers, the receive data multiplexer, and the plurality of link cores, the buffer access controller generating the data selection signals based on the number of dedicated lanes in each link corresponding to each of the one or more remote devices.
3. The apparatus of claim 2 , wherein the buffer controller further comprises:
a plurality of address pointers, each of the plurality of address pointers corresponding to one of the plurality of link cores; and
a receive address multiplexer coupled between the plurality of address pointers and the receive buffer, the receive address multiplexer selectively passing addresses from the plurality of address pointers responsive the address selection signals; wherein the buffer access controller is further coupled to the plurality of address pointers and the receive address multiplexer and is further configured to generate the address selection signals based on the number of dedicated lanes In each link corresponding to each of the one or more remote devices.
4. The apparatus of claim 1 , wherein the buffer is a transmit buffer comprising an address port and a data port and wherein the buffer controller comprises:
a plurality of pipeline registers, each of the plurality of pipeline registers corresponding to one of the plurality of link cores, the plurality of pipeline registers configured to store data from the transmit buffer; and
a buffer access controller coupled to the plurality of pipeline registers and the plurality of link cores, the buffer access controller generating load signals for loading data Into the plurality of pipeline registers based on the number of dedicated lanes in each link corresponding to each of the one or more remote devices.
5. The apparatus of claim 4 , wherein the buffer controller further comprises:
a plurality of address pointers, each of the plurality of address pointers corresponding to one of the plurality of link cores; and
a transmit address multiplexer coupled between the plurality of address pointers and the transmit buffer, the transmit address multiplexer selectively passing addresses from the plurality of address pointers responsive to address selection signals; wherein the buffer access controller is further coupled to the plurality of address pointers and the transmit address multiplexer and is further configured to generate the address selection signals based on the number of dedicated lanes in the links corresponding to each of the one or more remote devices.
6. In a computer architecture comprising a host device to buffer data transferred between a controller of the host device and one or more remote devices, wherein the host device is configured to couple to each of the one or more remote devices through one or more corresponding dedicated lanes, the computer architecture implementing a method comprising the steps of:
buffering data transferred between one or more remote devices and the controller of the host device when the one or more remote devices are coupled to the host device, wherein each of the one or more remote devices couple to the host device through a corresponding link including one or more dedicated lanes;
controlling the buffering so as to allocate access to the data transferred between the one or more remote devices and the controller of the host device based on a number of dedicated lanes in each link corresponding to each of the one or more remote devices;
transferring, via a plurality of link cores, the data transferred between the controller of the host device and each of the one or more remote devices; and
negotiating with the one or more remote devices coupled to the host device, via a plurality of link trainers wherein each link trainer corresponds to one of the link cores, to determine the number of dedicated lanes in each link corresponding to the one or more remote devices.
7. The method of claim 6 , further comprising the steps of:
configuring a plurality of pipeline registers to store data received from the link cores, wherein each of the plurality of pipeline registers corresponds to one of the plurality of link cores;
selectively passing data from the pipeline registers for buffering in response to data selection signals; and
controlling the buffering of the passed data based upon the number of dedicated lanes in each link corresponding to each of the one or more remote devices.