Hi, Steve,

I got it. Thanks a lot!

Leonard




On Sun, Mar 14, 2010 at 9:37 PM, Steve Reinhardt <[email protected]> wrote:

> It's part of the implementation of the Alpha "store conditional" (SC)
> instruction, used for synchronization.  It's returning true because in
> this case the access is "satisfied" in the sense that the instruction
> can complete.
>
> If you have questions about how SC works, please look it up (e.g.,
> it's covered in section 6.7 of the 3rd edition of H&P under
> "synchronization").  If you have more questions about the code
> specifically, please ask.
>
> Steve
>
> On Sun, Mar 14, 2010 at 3:25 PM, Sage <[email protected]> wrote:
> > Hi, everyone,
> >
> > I have question about a code block of the function  "bool
> > Cache<TagStore>::access(PacketPtr pkt, BlkType *&blk, int &lat,
> PacketList
> > &writebacks)" in the file "src/mem/cache/cache_impl.hh". The following is
> > the code block to be ased about.
> >
> > if (blk == NULL && pkt->isLLSC() && pkt->isWrite()) {
> >        // complete miss on store conditional... just give up now
> >        pkt->req->setExtraData(0);
> >        return true;
> > }
> >
> > Although there is a sentence of comments there, the meaning of the code
> > block is still not clear to me at all. What is the significance of the
> code
> > block? What is it used for? Since "blk" is NULL, why is a "true" still
> > returned finally?
> >
> > Any help would be greatly appreciated!
> >
> > Thanks,
> > Leonard
> >
> >
> >
> > --
> > Give our ability to our work, but our genius to our life!
> >
> > _______________________________________________
> > m5-users mailing list
> > [email protected]
> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
> >
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>



-- 
Give our ability to our work, but our genius to our life!
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to