On 04/21/2015 02:22 PM, Mark Cave-Ayland wrote:
On 21/04/15 12:36, Alexander Graf wrote:
Hi Mark,
Does this ring a bell with you? I'm getting the error below with latest QEMU+KVM on a G5.
Alex
2015-04-15 23:54:36: C>> annot manage 'ethernet controller' PCI device type 'network': 2015-04-15 23:54:36: >> 8086 100e (2 0 0) 2015-04-15 23:54:36: 2015-04-15 23:54:36: >> ============================================================= 2015-04-15 23:54:36: >> OpenBIOS 1.1 [Mar 12 2015 08:09] 2015-04-15 23:54:36: >> Configuration device id QEMU version 1 machine id 1 2015-04-15 23:54:36: >> CPUs: 1 2015-04-15 23:54:36: >> Memory: 512M 2015-04-15 23:54:36: >> UUID: 00000000-0000-0000-0000-000000000000 2015-04-15 23:54:36: >> CPU type PowerPC,G4 2015-04-15 23:54:36: milliseconds isn't unique. 2015-04-16 00:06:57: (Process terminated with status 0)
Yeah, it's harmless. It's there because I moved SPARC over to use an interrupt-driven counter for milliseconds with the eventual aim of adding alarms if required. However since this isn't yet complete, the fallback is for PPC to redefine its own milliseconds word based on the CPU tick register which is why the warning appears on the serial console.
Well, this is the last thing happening on any console. VGA doesn't even get initialized ;).
Alex
At some point I'll switch it over to using a decrementer interrupt at which point the warning will go away, a lot of timer cruft can be removed, and there is potential for adding alarms later if required.
ATB,
Mark.