QNX - Technology

Technology

The QNX kernel contains only CPU scheduling, interprocess communication, interrupt redirection and timers. Everything else runs as a user process, including a special process known as proc which performs process creation, and memory management by operating in conjunction with the microkernel. This is made possible by two key mechanisms — subroutine-call type interprocess communication, and a boot loader which can load an image containing not only the kernel but any desired collection of user programs and shared libraries. There are no device drivers in the kernel. The network stack is based on NetBSD code. Along with its support for its own, native, device drivers, QNX supports its legacy, io-net manager server, and the network drivers ported from NetBSD.

QNX interprocess communication consists of sending a message from one process to another and waiting for a reply. This is a single operation, called MsgSend. The message is copied, by the kernel, from the address space of the sending process to that of the receiving process. If the receiving process is waiting for the message, control of the CPU is transferred at the same time, without a pass through the CPU scheduler. Thus, sending a message to another process and waiting for a reply does not result in "losing one's turn" for the CPU. This tight integration between message passing and CPU scheduling is one of the key mechanisms that makes QNX message passing broadly usable. Most Unix and Linux interprocess communication mechanisms lack this tight integration, although an implementation of QNX-type messaging for Linux does exist. Mishandling of this subtle issue is a primary reason for the disappointing performance of some other microkernel systems such as early versions of Mach.

Due to the microkernel architecture QNX is also a distributed operating system. Dan Dodge and Peter van der Veen hold US patent 6,697,876 based on the QNX operating system's distributed processing features known commercially as Transparent Distributed Processing.

All I/O operations, file system operations, and network operations were meant to work through this mechanism, and the data transferred was copied during message passing. Later versions of QNX reduce the number of separate processes and integrate the network stack and other function blocks into single applications for performance reasons.

Message handling is prioritized by thread priority. Since I/O requests are performed using message passing, high priority threads receive I/O service before low priority threads, an essential feature in a hard real-time system.

The boot loader, although seldom discussed, is the other key component of the minimal microkernel system. Because user programs can be built into the boot image, the set of device drivers and support libraries needed for startup need not be, and are not, in the kernel. Even such functions as program loading are not in the kernel, but instead are in shared user-space libraries loaded as part of the boot image. It is possible to put an entire boot image into ROM, which is used for diskless embedded systems.

Neutrino supports symmetric multiprocessing and processor affinity, called bound multiprocessing (BMP) in QNX terminology. BMP is used to improve cache hitting and to ease the migration of non-SMP safe applications to multi-processor computers.

Neutrino supports strict priority-preemptive scheduling and adaptive partition scheduling (APS). APS guarantees minimum CPU percentages to selected groups of threads, even though others may have higher priority. The adaptive partition scheduler is still strictly priority-preemptive when the system is underloaded. It can also be configured to run a selected set of critical threads strictly realtime, even when the system is overloaded.