On Fri, 2007-07-13 at 16:41 +0300, Avi Kivity wrote:
> I tested FC6 i386 about found no difference with or without the patch. > > Dave, can you diff the guest dmesg with kvm-27 and kvm-28 userspace? > Maybe that will tell us something. > I can say this: It would seem that the IDE enable bits "float" before Lucas fix. I say this because I had written/submitted a similar fix way back in kvm-teens because I found that sometimes my SLED installer guest would fail to see the IDE controller. I found it to be that the IDE controller wasn't enabled sometimes and setting the 0x80 bits as Luca has done fixed the issue. Then, I found that later versions of KVM (sans my patch) seemed to work fine so I somewhat assumed the issue had been addressed. When I saw Lucas patch come through, I realized that it must not have been fixed, but I was getting lucky that it tended to come up enabled pretty much 100%. So heres what I think is happening: I think that having the IDE controller enabled is causing FC7 to go into a different mode of operation, which is slower for whatever reason. I think that Dave has a system that relatively deterministically comes up disabled, and Avi has one that comes up enabled. Therefore, the patch has ill-effect for Dave, and no-effect for Avi. Therefore, to test this theory, Avi should modify Luca patch to *disable* the IDE and see what happens. If the theory is true, Avi will see a speed improvement. Just a shot in the dark... ;) -Greg ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ kvm-devel mailing list kvm-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kvm-devel
