[Fuego] PC_WRITE_RETIRED event not implemented on A57?

Liu, Wenlong liuwl.fnst at cn.fujitsu.com
Wed May 17 12:11:02 UTC 2017



> -----Original Message-----
> From: Bird, Timothy [mailto:Tim.Bird at sony.com]
> Sent: Wednesday, May 17, 2017 1:15 PM
> To: Liu, Wenlong/刘 文龙 <liuwl.fnst at cn.fujitsu.com>;
> fuego at lists.linuxfoundation.org
> Subject: RE: PC_WRITE_RETIRED event not implemented on A57?
> 
> 
> 
> > -----Original Message-----
> > From Liu, Wenlong on Tuesday, May 16, 2017 6:08 AM
> >
> > Now I am doing the LTP test on the renesas board.
> >
> >
> >
> > The syscalls testcase“perf_event_open0” responded TCONF and show me
> > that “PERF_COUNT_HW_BRANCH_INSTRUCTIONS” not supported.
> >
> > I checked the kernel sourcecode, and find that PC_WRITE was not
> > supported on A57, but supported on A53.
> >
> > The kernel I used is 4.6.0-yocto-standard.
> >
> >
> >
> > It made me felt confused, I think A57 should support all implemented
> > on A53^_^(I just know a little about the archs).
> >
> > So I also checked the CPU Technical Reference Manual of A57 and A53.
> >
> > Indeed, the event not implemented on A57, and it can also learn from
> > the event name “PC_WRITE_RETIRED” in A53 Manual.
> >
> >
> >
> > Sorry for my long description.
> >
> > Now, the issue is coming:
> >
> > 1.  What replaced the PC_WRITE event on A57. It just disappeared?
> I don't know.  This is something architecture-specific.  You might want
> to ask that question on the ARM linux kernel list, and see if the ARM experts
> there have some information about this.
OK, I will. 
Thanks for your advice.

> >
> > 2.  What should I do to pass the testcase, not TCONF.
> 
> I think TCONF is the correct result in this case.  My understanding is that
> LTP uses TCONF to indicate that a test could not be run because it was not
> applicable to a particular situation, or the current configuration of
> hardware or software didn't support it.  That sounds like the correct result
> for this situation.
>
> If you want to avoid running the test, to remove it from your result list
> so that you have a "perfect" (100% success) result, then you may want
> blacklist the test.  We just added support for blacklisting an LTP sub-test
> to the next release of the Fuego.  However, that release will likely not
> be finalized for another month or two.  LTP itself includes a "skip" option,
> that you might want to use yourself to customize LTP for this issue.
Yeah, the "skip" option seems to be useful for me and looking forward to the next release.
Thanks for your time.

> 
> > Thanks in advance.
> >
> >
> >
> > One more thing,
> >
> > the A53 CPU’s were disabled on my board. Thanks for the mail named “Re:
> > [agl-discussions] A53 CPU on renesas board?” before.
> >
> >
> >
> > B.R.
> >
> > Liu
> >
> >
> 
> 





More information about the Fuego mailing list