Method and apparatus for the context switching of devices

A multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space, said system further comprising a means for providing the time sharing of the user defined device, said means comprising:

• means for mapping the device in each process address space in a manner that each device is mapped into at least one page;
• means for resetting the corresponding valid bits of the pages of each of the process address space to which the device is mapped;
• means for determining that a page fault occurs because a process has attempted to access a device mapped to a page in a process address space and the corresponding valid bit is reset;
• means for performing a context switch of the device when a page fault occurs due to a process attempting to access a device and the corresponding valid bit is reset, said means comprising;
• means for restoring on the device the context of the device with respect to the process attempting to access the device;
• means for setting the corresponding valid bit of the process attempting to access the device such that subsequent attempts to access the device by the process do not cause a page fault;
• whereby the process attempting to access the device can access the device and the device resumes execution in the context of the process which attempted to access the device.

A multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one user-defined device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising segments, said segments comprising at least one page, each page having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space each of said segments having a page fault handler associated with it, said system further comprising a means for providing the time sharing of the user defined device, said means comprising:

• means for controlling the mapping of the context of the user-defined device with respect to a particular process, such that each mapping of a device with respect to a process is mapped into separate segments of the respective process address space;
• means for controlling the state of the valid bits of the pages of each of the processes address space to which the device is mapped, such that the valid bits are initially reset;
• means for determining that a page fault occurs because a process has attempted to access a device mapped to a page;
• if a page fault occurs, means for enabling the page fault handler of the segment within which the page fault occurred, said page fault handler indicating the steps to be taken to perform a context switch;
• means for performing a context switch as indicated by page fault handler comprising;
• means for setting the valid bits of the pages of the process address space to which the device is mapped with respect to the process which is attempting to access the device, whereby subsequent attempts to access by the process do not cause a page fault;
• means for restoring the context of the process attemping to access the device whereby the device resumes execution in the context of the process attempting to access the device;
• and
• means for controlling the means for restoring, such that the context of the device with respect to the process attempting to access the device is restored on the device only if the device context on the device is not the device context with respect to the process attempting to access the device;
• means for saving the context of the device with respect to the process the next time a process switch occurs and the process is no longer executed by the CPU;
• and
• when the next process switch occurs, means for resetting the valid bits of the pages of the address space to which the device is mapped with respect to the process the CPU executes prior to the process switch, so that subsequent attempts to access by the process cause a page fault;
• whereby multiple processes can share a single device.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one user-defined device which is connected to the system through a memory management unit (MMU), said device executing commands issued by the CPU, said being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a corresponding valid bit that invokes a page fault if the bit is not set an attempt is made to access the address space, a process for providing the time sharing of the user defined device, a method for performing context switching comprising the steps of:

• mapping the device in each process address space in a manner that each device is mapped into at least one page;
• resetting the valid bits of the pages of each of the process address space to which the device is mapped;
• determining that a page fault occurs because a process has attempted to access the device mapped to a page in the process address space;
• performing a context switch of the device when a page fault occurs due to a process attempting to access the device, comprising the steps of:
• restoring on the device the context of the device with respect to the process attempting to access the device whereby the device resumes execution in the context of the process which attemped to access the device;
• permanently setting the corresponding valid bits in the pages of the process address space to which the device is mapped such that subsequent attempts by the process to access the device do not cause a page fault;
• performing a context switch when a process switch occurs comprising the steps of:
• saving the context of the device with respect to the process, executed by the CPU prior to the process switch;
• restoring on the device the context of the device with respect to the process, which is executed by CPU subsequent to the process switch;
• whereby after a page fault occurs, a context switch will be performed with respect to the process each time a process switch occurs.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one user-defined device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a valid bit that involes a page fault if the bit is not set and an attempt is made to access the address space, a process for providing the time sharing of the user defined device, said means comprising the steps of:

• mapping the device in each process address space in a manner that each device is mapped into at least one page;
• resetting the corresponding valid bits of the pages of each of the process address space to which the device is mapped;
• determining that a page fault occurs because a process has attempted to access a device mapped to a page in a process address space and the corresponding valid bit is reset;
• performing a context switch of the device when a page fault occurs due to a process attempting to access a device and the corresponding valid bit is reset, comprising the steps of:
• restoring on the device the context of the device with respect to the process attemping to access the device whereby the device resumes execution in the context of the process which attempted to access the device;
• setting the corresponding valid bits of the page of the process address space to which the device is mapped with respect to the process attempting to access the device such that subsequent attempts to access the device do not cause a page fault;
• when the next process switch occurs;
• saving the context of the device with respect to the process executed by the CPU prior to the process switch;
• and
• resetting the valid bits of the pages of the process address space to which the device is mapped with respect to the process executed by the CPU prior to the process switch, such that subsequent attempts to access by the process cause a page fault.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one user-defined device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space, a process for providing the time sharing of the user defined device, said means comprising the steps of:

• mapping the device in each process address space in a manner that each device is mapped into at least one page;
• resetting the corresponding valid bits of the pages of each of the process address space to which the device is mapped;
• determining that a page fault occurs because a process has attempted to access a device mapped to a page in a process address space and the corresponding valid bit is reset;
• performing a context switch of the device when a page fault occurs due to a process attempting to access a device, comprising the steps of:
• determining if the context on the device is the device context with respect to the process attempting to access the device;
• if the device context on the device is not the device context with respect to the process attempting to access the device, restoring on the device the context of the device with respect to the process attempting to access the device;
• setting the valid bits of the pages of the process address space to which the device is mapped with respect to the process attempting to access the device such that subsequent attempts to access by the process do not cause a page fault;
• resuming device execution in the context of the process which attempted to access the device;
• when a process switch occurs, saving the context of the device with respect to the process which is executed by the CPU prior to the process switch and resetting the valid bits of the pages of the process address space to which the device is mapped with respect to the process attempting to access the device such that subsequent attempts to access the device by the process causes a page fault.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one user-defined device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space, a process for providing the time sharing of the user defined device, said means comprising the steps of:

• mapping the device in each process address in a manner that each device is mapped into at least one page;
• resetting the corresponding valid bits of the pages of each of the process address space to which the device is mapped;
• determining that a page fault occurs because a process has attempted to access the device mapped to a page in the process address space;
• performing a context switch of the device when a page fault occurs due to a process attempting to access a device, comprising the steps of:
• saving the context of the device currently in the device with respect to the process which las accessed the device;
• restoring on the device the context of the device with respect to the process attempting to access the device whereby the device resumes execution in the context of the process which attempted to access the device;
• setting the valid bits of the pages of the process address space to which the device is mapped with respect to the process attempting to access the device;
• and
• resetting the valid bits of the pages of the process address space to which the device is mapped with respect to the process which last accessed the device.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages which are grouped into segments, each of said pages having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space, each of said segments having a page fault handler associated with it which is invoked when a page fault occurs in the segment, a method for controlling said device comprising the steps of:

• mapping the device in each process address space such that each device is mapped into a separate segment whereby processes access the device through the address the device is mapped to in its process address space;
• initially resetting the valid bits of the pages of each of the segments located in the process address space to which the device is mapped;
• initializing the page fault handler to perform a context switch of the device, said context switch comprising the steps of:
• saving the context currently in the device;
• restoring on the device the context of the device with respect to the process attempting to access the device, and
• setting the corresponding valid bits of the process address space with respect to the process attempting to access the device
• whereby the device resumes execution in the context of the process attempting to access the device;
• when a process switch occurs, resetting the corresponding valid bits of a process executed by the CPU prior to the process switch whereby the process will cause a page fault when an attempt is made to access the device;
• determining that a page fault occurs because a process has attempted to access a device mapped in its address space and the corresponding valid bits are reset, whereby the page fault handler is invoked to perform the context switch of the device;
• whereby the device and access to the device are controlled by setting and resetting the valid bits of the pages to which the device is mapped such that when a process attempts to access the device and the corresponding valid bits are reset, a page fault occurs and the page fault handler is invoked which executes predetermined tasks to perform the context switching of the device.

In a multi-tasking system in which multiple processes are executed by a central processing unit (CPU) of the system on a time-shared basis using context switching, the context of the CPU being the state of the CPU with respect to a particular process, the switching of the CPU from one process to another process referred to as process switching, said system further comprising at least one device which is connected to the system through a memory management unit (MMU), said device being mapped into the process address space of each process which may access the device, said process address space comprising pages, each page having a corresponding valid bit that invokes a page fault if the bit is not set and an attempt is made to access the address space, a method for providing the time sharing of the user defined device, said method comprising the steps of:

• mapping the device in each process address space in a manner that each device is mapped into at least one page;
• resetting the corresponding valid bits of the pages of each of the process address space to which the device is mapped;
• determining that a page fault occurred because a process has attempted to access the device mapped to a page in a process address space and the corresponding valid bit is reset;
• performing a context switch of the device when a page fault occurs due to a process attempting to access the device and the corresponding valid bit is reset, comprising the steps of:
• restoring on the device the context of the device with respect to the process attempting to access the device;
• setting the corresponding valid bit of the process attempting to access the device such that subsequent attemps to access the device by the process do not cause a page fault;
• whereby the process attempting to access the device can access the device and the device resumes execution in the context of the process which attempted to access the device.