The field relates to transfers of information between various address spaces.
Modern computers are extremely sophisticated, complicated and powerful. To improve the processing capabilities, many modern computers include intelligent input output (I/O) devices. These intelligent I/O devices often include a system-on-a-chip, which includes a microprocessor and has a connected memory. Of course, the computer includes its own main microprocessor and memory. The presence of multiple microprocessors often creates a situation of disparate address spaces between the various microprocessors. The intelligent I/O devices are often connected to the computer using advanced interfaces, such as Peripheral Component Interconnect Express (PCIe). PCIe includes its own address space, which is in addition to the address spaces of the various microprocessors. The existence of these plurality of address spaces in the past has required address translations at each step and in each direction. The many address translations complicated overall computer system design, often unnecessarily.
An address space field is used in conjunction with the normal address field to allow indication of an address space for the particular address value. In one instance, one address space value is used to indicate the bypassing of the address translation used between address spaces. A different address space value is designated for conventional operation, where address translations are performed. Other address space values are used to designate different transformations of the address values or the data. This technique provides a simplified format for handling address values and the like between different devices having different address spaces, simplifying overall computer system design and operation.
For a detailed description of various examples, reference will now be made to the accompanying drawings in which:
Referring now to
The I/O card 104 includes a semiconductor device that is a system on a chip (SOC) 116. The SOC 116 includes a microprocessor 118, which may include an MMU 119 (
Referring now to
If the data is to be transferred from the buffer 232 to the buffer 230, then the data is retrieved from the buffer 232 and provided to a transmit block 242 in the PCIe port 220. The transmit block 242 includes a translate block 244 to translate from the physical memory address used with memory 224 to the PCIe address. The data is received at a receive block 246 in the PCIe root port 210. A translate block 248 translates the address from the PCIe address to the physical address of the memory 214 and then the data is deposited in the buffer 230.
As can be seen, multiple address translations are required even though the processors on each device are highly sophisticated and can manage very complicated memory maps. One difficulty is the need to program each of the particular translations into the various translate blocks. On the PCIe receive side, this is done through base address registers (BARs), of which there are a limited number and must be programmed or reprogrammed as needed. On the PCIe transmit side, the direct address translations are programmed or reprogrammed as necessary.
Referring now to
In operation, the DMA controller 122 is programmed for a particular data transfer. As an example, for a transfer of data from a buffer in the memory 124 to the PCIe port 120, the DMA controller 122 is programmed to provide an ASEL value of zero when reading the data from the memory 124 and is programmed to use an ASEL value of one when writing the data to the PCIe port 120 if no address translation is to be performed. If address translation is to be performed, then the DMA controller 122 is programmed to use an ASEL value of zero on the write operation. If the microprocessor 118 is operating directly with the PCIe port 120, then an ASEL value of zero is used as in the example the microprocessor 118 only provides memory operations using a zero ASEL value. In that instance, the PCIe port 120 performs address translations as normally done.
Referring now to
In the illustrated example, a receive (RX) address path 414 is provided from the PCIe core 404 to the input of a demultiplexer 416. The output control of the demultiplexer 416 is provided by an address map block 418 which examines the address bits, such as Axx-A00, of the transaction and appropriately determines whether the addresses need to be translated or can be used in an untranslated format. If translation is required, the zero output from the demultiplexer 416 is provided to a translate block 420, which performs the necessary address translation. The output of the translate block 420 is provided to a zero input of a multiplexer 422, with a zero value ASEL bit pre-pended to the address. The output of the multiplexer 422 is provided as the receive address to the bus master module 402.
If no address translation is required, the one output of the demultiplexer 416 provides the address directly to the one input of the multiplexer 422, with a one value provided as the pre-pended ASEL bit. By providing the ASEL bit at the multiplexer 422, the internal routing inside the SOC 116 can be done using the ASEL bit as illustrated in
In another example, the untranslated or bypass path is not present in the receive path and address translations based on the base address registers (BARs) are performed. In this example the demultiplexer 416, multiplexer 422 and address map 418 used with the demultiplexer 416 are not present and an ASEL value of zero is always used.
Referring now to
The above description uses a single ASEL bit to simplify the description. In another example shown in
Data paths were not illustrated in
While
The above description has focused on an endpoint PCIe port configuration. It is understood that the use of the additional bits to provide address space value indications which are used to indicate different functions can be used in a PCIe root port configuration as well.
The above description has described the ASEL bits as pre-pended to the normal address bits. The pre-pending is done for illustrative purposes. It is understood that the ASEL bits can be post-pended or can be provided in a side channel if desired.
While configuration and operation is described using a PCIe port as exemplary, it is understood that selections based on address space value bits can be done for many different functions, functions that need some processing of address or data values between the source and destination. As one example, inclusion in and operation with a HyperTransport interface is similar to the exemplary PCIe port. Other address translation examples include various remote DMA (RDMA) protocols. Endian conversion and compression were described as data operations but other data operations can also be done, such as encryption and the like.
It is understood that different ports or functional blocks can use different address and data transformations in the same system. For example, address translation bypassing can be used with a PCIe port and data encryption can be used with a disk array function by assigning each operation to a different address space, such as zero for normal PCIe port and serial attached SCSI (SAS) port operation, one for PCIe port address translation bypassing and two for SAS port data encryption. In such a case, the multiplexers and demultiplexers at each port do not need to be complete, that is include all outputs or inputs for the number of address space value selection bits, but can only decode for address space values relevant to the particular function, with any other address space value at that port generating an error condition.
While the above description has focused on configurations and operation internal to an SOC, it is understood that the configurations and operation apply equally to external connections.
While only the DMA controller and the PCIe port have been described as providing and using different ASEL bit values in their operations, it is understood that any other device can use address space values if desired. For example, the MMU in the processor can be extended to add ASEL bits to its operation to allow the processor to perform address space indication as well.
By providing address space value selection bits in addition to normal address bits, different functionalities for the different address spaces can be specified, allowing improved operation of the resulting computer.
The above description is intended to be illustrative, and not restrictive. For example, the above-described examples may be used in combination with each other. Many other examples will be upon reviewing the above description. The scope should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled. In the appended claims, the terms “including” and “in which” are used as the plain-English equivalents of the respective terms “comprising” and “wherein.”
This application is a continuation of U.S. patent application Ser. No. 17/946,270, filed Sep. 16, 2022, which is a continuation of U.S. patent application Ser. No. 16/559,154, filed Sep. 3, 2019, now U.S. Pat. No. 11,449,444, which is a continuation of U.S. patent application Ser. No. 15/890,558, filed Feb. 7, 2018, now U.S. Pat. No. 10,402,355, which claims the benefit of U.S. Provisional Patent Application Ser. No. 62/456,162, filed Feb. 8, 2017, each of which is incorporated by reference herein in its entirety.
Number | Date | Country | |
---|---|---|---|
62456162 | Feb 2017 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 17946270 | Sep 2022 | US |
Child | 18794472 | US | |
Parent | 16559154 | Sep 2019 | US |
Child | 17946270 | US | |
Parent | 15890558 | Feb 2018 | US |
Child | 16559154 | US |