I agree that the "Control Flow View" name is not immediately obvious to all. I believe that it was first used by Karim Yaghmour in LTT and came from a context where you had only one CPU core and the interesting action was the control flow between a few tasks, the operating system and interrupts. I believe that a similar naming was used in related systems at the time like the IRIX and the VXWorks trace viewers.
Indeed, I agree that it should be simpler for most users to think about views for different objects / resources:
- CPU cores view (what each CPU core is doing over time)
- Threads view (the state of each thread over time)
- Disks view (the state of each disk over time)
- Mutex view (the state and holder of each mutex over time)
Things get a bit more complicated when you have processes (running several threads) or Virtual Machines (with several vCPUs). Sometimes, it is simpler to show a CPU core view filtered for a specific process or virtual machine.
----- Le 9 Mai 18, à 15:42, Matthew Khouzam <matthew.khouzam@xxxxxxxxxxxx> a écrit :
The Control flow view is a view that has some legacy to it. I am wondering if we should rename it to "Threads" view. It effectively shows the threads running on a system, so it would be more intuitive IMO.
In case you're wondering, it is not a process view, it shows more threads than there are processes. Proof: make a trace with tid and pid.
tracecompass-dev mailing list
To change your delivery options, retrieve your password, or unsubscribe from this list, visit