On Tue, 26 Aug 2003, Egbert Eich wrote:
> John Dennis writes:
> > I just received information from HP (thank you Bjorn Helgaas) that the
> > memory attribute does not direct access between RAM and IO, I was in
> > error. But what is critical is that the EFI MDT (Memory Descriptor
> > Table?) which is set up at boot time is the arbiter of which memory
> > attributes are used for various memory regions. User space mappings that
> > try to force cached vs. non-cached accesses are inappropriate, its not
> > their decision, rather it must be consistent with the EFI table which is
> > set up at boot time. There is a kernel patch that ignores the user
> > request for cached vs. uncached mappings and instead (indirectly)
> > consults the EFI MDT such that the memory attribute field in the TLB is
> > consistent with how that memory is referenced in the system, as
> > determined by the firmware (EFI). Beta RH kernel's were missing this
> > patch that had been present in earlier kernels. Bjorn tells me that with
> > the patch applied no changes need to be made to X for proper operation
> > and that the patch is in the upstream kernel sources.
> thanks for following up on this. Does this patch also take care of
> the SlowCopy() access to the VGA framebuffer memory at 0xa0000?
> > FYI, the patch follows for those interested, note that O_SYNC is no
> > longer used as a trigger for non-cached access.
> I have dropped the O_SYNC from open("/dev/mem") for all platforms
> a while ago. However on IA64 we pass MAP_NONCACHED or
> MAP_WRITECOMBINED to mmap() depending on the type of region.
> I assume we don't need this any more, either.
> Now, one question remains: does the EFI MDT distingquish between
> (cacheable) framebuffer memory and MMIO registers which should
> neither be cached nor set WC?
> Uncached framebuffer memory doesn't cause failures but gives a huge
> performance penalty.
Frankly, I don't see how this EFI MDT can be accurate given that, in
general, whether or not a particular PCI memory assignment will tolerate
caching and/or write-combining is highly device-specific. That would be a
horrific PCI device database for EFI to maintain.
Marc.
+----------------------------------+-----------------------------------+
| Marc Aurele La France | work: 1-780-492-9310 |
| Computing and Network Services | fax: 1-780-492-1729 |
| 352 General Services Building | email: [EMAIL PROTECTED] |
| University of Alberta +-----------------------------------+
| Edmonton, Alberta | |
| T6G 2H1 | Standard disclaimers apply |
| CANADA | |
+----------------------------------+-----------------------------------+
XFree86 Core Team member. ATI driver and X server internals.
_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel
