OK, but I’m still missing something (obvious) here. Isn’t there memory bus
snooping going on that will observe the memory write and invalidate the
cache line(s) of the other cpus?
-----Original Message-----
From: Dave Harvey [mailto:xxxxx@syssoftsol.com]
Sent: Friday, June 21, 2002 9:50 AM
To: NT Developers Interest List
Subject: [ntdev] Re: posible bug in XP kernel - was
MmMapLockedPages… . on XP vs. 2000The CPU, upon writing uncached memory, will not shoot down
the cache line on another CPU. -DH“Roddy, Mark” wrote in message
news:xxxxx@ntdev…
>
>
>
> >
> > Remember that one of the changes from Win2K to XP/.NET was a fix
> > whereby memory mappings must be consistent with regard to cache
> > type. In other words, if the memory is mapped one place non-cached,
> > all other mappings of those pages must also be non-cached.
> >
>
> We’ve been wondering here exactly why this is a ‘fix’. Intuitively
> it sounds right, if memeory is cached one place it ought to be cached
> everywhere, but can somebody explain exactly how the cache coherency
> protocol gets violated in this case and memory access gets potentially
> incoherent?
>
>
—
You are currently subscribed to ntdev as: xxxxx@stratus.com To
unsubscribe send a blank email to %%email.unsub%%