Messages in this thread | | | Date | Sun, 14 Mar 1999 21:48:53 +1100 | From | Richard Gooch <> | Subject | Re: MTRR & NVTNT |
| |
Alan Cox writes: > > Is this really needed? It's getting more complicated (can you say > > "bloat") for what appears to be a fairly uninteresting and uncommon > > problem. > > Maybe. Its just there are two ways to deal with the smart situation > > 1. You do it per request > > 2. You keep a "wishlist mtrr" set and you attempt to fulfill it as best > you can. This also means that if an mtrr is deleted you just > run mtrr_recalc() and it may reassigned mtrrs to drivers that it > couldnt fit in before > > Im not convinced tht the smart per request handling is any easier > than resolving the generic "here is the wanted mtrr list" case.
I think (2) *is* going to be more complex than (1). More code and more data. And for what benefit? Not that many people have multiple video cards, and those who do probably have CPUs with >2 MTRRs. Even the el cheapo Centaurs have 8. Time for AMD to pull their finger out...
Anyway, once my current patch gets applied, I can ponder this more closely. One step at a time.
Regards,
Richard....
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |