* Izik Eidus <[EMAIL PROTECTED]> [2007-08-14 13:59]:
> On Tue, 2007-08-14 at 13:24 -0500, Ryan Harper wrote:
> > * Izik Eidus <[EMAIL PROTECTED]> [2007-08-14 10:49]:
> > > On Tue, 2007-08-14 at 10:09 -0500, Ryan Harper wrote:
> > > > * Izik Eidus <[EMAIL PROTECTED]> [2007-08-14 01:29]:
> > > > > Hey Ryan,
> > > > > thanks for the testing, i hope you didnt have too much problems to 
> > > > > get it working.
> > > > 
> > > > Sure, it wasn't too much trouble.  
> > > > 
> > > > > anyway as far as i can tell it wont have problem to drive up to 256 + 
> > > > > 3.75 giga ram for guest.
> > > > > if we want it to drive systems with even more ram we have two options 
> > > > > ( both very easily applied ):
> > > > > we can add another cmos byte to a "future reserved" byte, or we can 
> > > > > use the 3 cmos bytes that i already added and say that
> > > > > we store memory in the above bios memory at multiplier of 1 MB.  
> > > > > it is important we decide now how we want to store the memory, that 
> > > > > in the future when 256 + 3.75 giga of ram wont be enough
> > > > > we wouldn't have to change bochs bios again.
> > > > 
> > > > Yeah, I think we want to settle on a single method which gets us the
> > > > most memory as possible.  I think rather than doing the "future reserve"
> > > > we should go head an move over to 1MB multiplier.
> > > > 
> > > well this sound like a smart idea,
> > > but what we have to think about is:
> > > first in this way we have just 64 gigabyte of ram (unless we work with
> > > the extra cmos memory bytes)
> > > plus if we change the way we use the "normal cmos bytes", we arent
> > > breaking compatibility with really old stuff that check the cmos
> > > directly without doing bios interrupt? (i mean by ports)
> > 
> > Hrm, yes.  This might be an issue already.  I just booted memtest-86
> > v3.3 with 6G, and memtest says we have 528G of RAM.  Hrm, even below 2G,
> > memtest still reports bogus memory values.
> 
> memtest-86, report false memory values when the system is below 2G ?, in
> this case i guess it isnt our fault...
> memtest-86 was derived originally from linux 1.x i dont know if it even
> support above 4 giga of ram.
> 
> without the patch what result do you get from this memtest when runing
> below 2giga?

It is accurate without the patch.  It can see up to 2039M guest with no
changes to QEMU.  Also, if you apply the qemu size patch (s/int/unsigned
long) you can get up to 3831M guest to run without changing the
BIOS.  The memtest isos I've tried have no problem seeing that.

> 
> (win2003 get good memory information on 14giga guest as well, so i guess
> the e820 from the bios interrupt side, is right)

Yes, I've looked at the e820 and that looks perfectly fine.  Not sure
what the issue here is.  I'll see if I can track it down though.

> > We might need to go look at a newer BIOS spec to see how this is done in
> > newer bioses.
> > 

-- 
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
(512) 838-9253   T/L: 678-9253
[EMAIL PROTECTED]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
kvm-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Reply via email to