Re: [Gems-users] ASI handlers?!!?


Date: Tue, 13 Feb 2007 19:52:41 -0600
From: Dan Gibson <degibson@xxxxxxxx>
Subject: Re: [Gems-users] ASI handlers?!!?
Ruby attaches itself to phys_mem0, which is a Simics memory object. Memory transactions passed to Ruby from simics have an "ASI" field, (see v9_transaction_t in the Simics Reference Manual), though I am not sure if *all* accesses (eg ASI!=ASI_P) come to Ruby or not. I would suspect that only "memory" accesses come to Ruby (you might want to verify this empirically). Some of these non-ASI_P accesses will have may_stall set to 0, and Ruby will filter these accesses, so work at low-levels in Ruby's code (ie. SimicsDriver.C or lower).

Some other "memory" objects map to other ASIs -- the TLB for instance maps to several different ASIs. I have not tried to install ruby to time these events, but it might be possible. Any Simics object that allows the "timing interface" (see phys_mem0) will work with Ruby. Code to install Ruby on a memory object can be found in ruby.c and again in interface.C.

Regards,
Dan

mehmetderin.harmanci@xxxxxxx wrote:

Hello,

I know this sounds more like a Simics issue, but I thought either Ruby
takes care of the problem or there are some ways to do what I like
in Ruby.

OK. Here is the problem: I need to define a new ASI and to be able to
handle accesses to an alternate space and I would like to attribute a
delay for this alternate space access. Since Ruby is meant to be
providing timing information to Simics my first question is the following:
Can Ruby receive requests for an ASI that is not that of the main memory
(like the ASI_PRIMARY etc.)? If it can not is it possible to make it
sensitive to that alternate ASI by installing a kind of ASI handler
(I have seen such handlers on Simics reference manual but I do not know
how to use them)? Finally, if anybody knows how to deal with ASI handlers,
especially how to install them, that would be of great help to me.

Sorry to disturb with such a question, but I'm quite desperate about
the problem.

Best Regards,

Derin Harmanci.

_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.


[← Prev in Thread] Current Thread [Next in Thread→]