Hello!
I am new to marss and I am trying to play with some statistics of a test
program(printf "Hello World") on marss. I am using a single atom core
configuration(single thread).

Here are some of my queries on the statistics on simulation of this Hello
World program:
1. The number of system calls (base_machine:: atom1_0:: thread_0::
assists:: syscalls)  = 168 (user) and 0 (kernel)
Does it mean that the hello world program invoked system calls 168 times?
Is that a reasonable figure? How come there are so many number of system
calls invoked for such a simple program?
2. If syscall gives the no.of times system calls are called, then what do
counters sysret and sysenter convey?
3. I see that there are multiple instances of statistics for a single
simulation, separated by --- in the stats file. (3 consecutive hyphens) (6
for the above test program and config). What does it mean?
4. Which counter provides the number of disk memory accesses?
decode::pagecache::invalidates::dma or exec_page_fault? What is the
difference between these two?
5. How can I determine the number of CPU context switches? I read in a
thread that the contents of CR3 can be used to determine which page
directory it is pointing to, and having a mapping of these addresses as
user space and kernel space can help provide the number of context
switches. Is it the correct way to follow or there is an easier, intuitive
way?
A follow up question is can I access the value of control registers during
simulation?
6. The int counter refers to the number of interrupts, how can we further
determine the no.of exceptions/traps/faults among those interrupts?
Basically, can I determine the no.of
hardware/software/traps/exceptions/faults interrupts etc individually?

Sorry if the questions appear too trivial. I am just beginning to
understand marss and use it for my research projects.
Thanks in advance.

Deepak K
_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to