Automated thread stack management for resource-constrained sensor operating systems
Adam Torgerson · CU Scholar (University of Colorado Boulder) · 2005
As low-power microcontrollers for wireless sensor nodes and other embedded devices continue to be manufactured with more memory, multi-threaded operating systems are becoming available on such devices.Each thread must be associated with an independent stack and part of the system memory must be dedicated to these stacks.This paper proposes a method of bounding worst-case thread stack sizes from assembly code by constructing a limited control ow graph and statically examining all possible execution ow within each thread, accounting for any stack space used by non-reentrant interrupts.This technique eliminates any guesswork from determining a thread's stack requirement and allows thread stack sizes to be managed automatically.