Running a PFS in a colony address space

By default, PFSs are initialized in the kernel address space. An installation may choose to run a PFS in a separate colony address space by specifying an ASNAME parameter on its FILESYSTYPE statement. You may want to have a PFS run in a colony address space if:

The PFS is constrained by kernel address space resources, such as:
- Storage
- Data set allocations
- Lock contention
The PFS needs to request callable services itself, in order to:
- Use sockets
- Make remote procedure calls
- Obtain POSIX file I/O

When a PFS runs in a colony address space, an extra address space is created, and each PFS operation has a slightly longer path length.

Any PFS can run in a colony address space unchanged. PFSs that are running in colony address spaces can use the osi_thread service, which is not available to PFSs that are running in the kernel address space. Any PFS that uses this service must document to its users that the PFS must be initialized in a colony address space. See Using daemon tasks within a PFS for more information about the osi_thread service.

The writer of a PFS cannot assume that the PFS will run in the kernel, nor that it will run under the task that calls it.