On Wed, 6 Nov 2013 11:40:15 +0800, David Crayford wrote:
>I know you can trace syscalls etc but do any of those traces compare to
>"system trace"? Are there any traces for zLinux which report
>hardware/software interrupts?
All the tools I mentioned are (predominantly) kernel-space.
ftrace (nominally {kernel} function trace, but more than just that) exports a
ring buffer via a debugfs mountpoint - so you don't even need to get a
failure/dump to look at it. The probe points are instrumented in the kernel but
dormant until activated.
perf can reach down to the counters in the hardware - TLB purges, cache misses
that sort of thing. Think HIS data but documented. Not sure how much is
implemented in z; I did see mention a while back but it slipped by. Not
something you'd run all the time.
Now for my wishlist, I'd like something (closely) akin to IPCS ... as if.
Shane ...
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
