Hi, We want to be able to profile the time taken by each major page fault (the time spent in the OS and the time spent waiting for the disk) occurring in the FS simulation of linux on gem5. Originally we thought we could just get/time each disk access in gem5, get/time each page fault in the linux kernel running in the gem5 simulation, and then afterwards match up the disk accesses with the page faults using the physical address. The problem is that when a page fault happens, the CPU and thus the OS don't know the physical address.
Is there a place in the Linux source that shows the address passed to the disk to lookup that we can look at to match page faults with appropriate disk accesses that we capture from the disk object in gem5? Or is there a better way to do this inside of gem5 and not touching the kernel? What function/method is called in the simulator to handle a page fault? It would be ideal to capture this directly from gem5 without having to modify the linux source code. -Zaid _______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
