Register-transfer-level (RTL) code, formal analysis, RTL simulation, and logic synthesis have all raised the abstraction level of electronic design and verification. Today’s designers operate very differently than their predecessors who drew circuit-level schematics and ran only SPICE. However, underneath all this abstraction the physical properties of electronic devices remain unchanged, and these must be considered during design. One well-known example is metastability, which can occur wherever a signal crosses between flip-flops running on asynchronous clocks, known as a clock domain crossing (CDC).

The most common example of metastability happens when the output value of a flip-flop on the sending clock changes during the setup and hold time of the clock for the receiving flip-flop. The output of the flip-flop on the receiving clock can take on an indeterminate value that requires some time to settle to a one or zero. If the output of the receiving flip-flop is used immediately, an invalid value may be fed into the downstream logic and produce incorrect results. Unfortunately, there is no way to design a flip-flop that does not have a risk of metastability.

The most common way to handle the metastability problem is to use a synchronizer consisting of two flip-flops on the receiving clock. The goal is for the signal to settle to its final value before the second flip-flop is clocked and its output is used by downstream logic. Theoretically it is possible that the output from the first flip-flop is still sufficiently metastable when it is clocked into the second flip-flop to cause problems. However, a mean time between failures (MTBF) calculation typically shows a high probability that the metastability will settle before the second flip-flop is clocked and the signal is used.

This approach works well for control signals, but it’s insufficient for multi-bit signals such as data buses. Different bits may have different delays due to logic and routing, so that they arrive at the receiving flip-flops at different times. An inconsistent data value can be clocked in and used by downstream logic. There are a few ways of addressing this issue, but perhaps the most sophisticated is the handshake synchronizer. This approach establishes a multi-cycle path (MCP) for the data bits so the outputs of the sending flip-flops are held stable long enough that they will not be changing when the values are clocked into the receiving flip-flop. Thus, there is no risk of metastability.

Control and status registers (CSRs) are one part of many designs where asynchronous clocks are often used and metastability is a very real problem. Most register blocks have a “software” interface on the system bus from which programs running on host systems or embedded processors can write or read the registers. Operating system boot routines include setting the registers to their initial values, and many other types of code, including interrupt service routines and applications, will read and write registers as well. Register blocks are usually in the system bus clock domain.

Register blocks also have a “hardware” interface from which the rest of the design can read or write the registers. The operation of the hardware is often highly dependent upon the settings of control and configuration registers, so the read path is essential. Both the read and write capabilities are vital when register bits or fields are used for handshaking between the hardware and software. Data can also be exchanged using registers. Since the hardware is almost always on a different clock than the system bus, there is a CDC that must be handled properly.

As the industry leader in the automation of register design and verification, Agnisys provides a pushbutton solution for CDC crossings related to register blocks. IDesignSpec™ (IDS) helps IP/SoC architects and engineers create an executable specification for registers and automatically generate outputs for software and hardware teams. Supported input formats include Microsoft Word and Excel, SystemRDL, and IP-XACT. IDS generates a UVM verification model complete with constraints and coverage, and customizable documentation in such file formats as Word, PDF, and HTML.

Most importantly, IDS automatically generates the RTL code for the register design (in VHDL, Verilog, SystemVerilog, or SystemC) plus a bus slave and decode logic specific to the system bus protocol (AHB, APB, AXI, AXI-Lite, TileLink, or proprietary). If the registers are in a different clock domain than the system bus, IDS also generates RTL for all the synchronization logic plus assertions for use in formal CDC verification. Agnisys has enhanced SystemRDL 2.0 so the designers can specify their desired synchronization method in the register specification.

IDS offers other synchronization options, including two-level flip-flops and mux synchronizers. It can also handle cases in which the register block is in its own clock domain distinct from the system bus and the rest of the hardware design. This level of automation provides great benefit to designers, who do not have to worry about CDCs at all, and provides help for the verification team as well. With a well-proven track record and excellent support, Agnisys and IDesignSpec are an essential part of ensuring success in designs with asynchronous clock domains.

To learn much more about CDCs and the CDC support within IDE, please view our recent webinar.

Anupam Bakshi

Leave a Reply