Not in the way you describe, however there are various profiling tools for Linux, like oprofile, SystemTap and perf events, I'm not sure how well they'd fit into embedded development though.
EDIT: perf is probably best (if you are running a recent enough kernel to use it) since it's in the mainline so you just have to turn it on, and it's really basic.
EDIT: if none of those work for you you can always modify the kernel context switch code...
EDIT: I missed one of the tracing frameworks, there is also LTTng
If you're using busybox and can compile your own kernel perf is probably the most minimal way to go, it's consists of turning on perf events in the kernel and compiling the perf tool that comes with the kernel source (it's in tools/perf)
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…