On 04/27/2015 06:58 PM, Christopher Covington wrote:
Hi Alex,
On 04/27/2015 09:41 AM, Alexander Graf wrote:
--- a/hw/arm/virt.c
+++ b/hw/arm/virt.c
@@ -131,6 +138,7 @@ static const MemMapEntry a15memmap[] = {
[VIRT_FW_CFG] = { 0x09020000, 0x0000000a },
[VIRT_MMIO] = { 0x0a000000, 0x00000200 },
/* ...repeating for a total of NUM_VIRTIO_TRANSPORTS, each of that
size */
+ [VIRT_PLATFORM_BUS] = { 0x0c000000, 0x02000000 },
Peter, would you have a hard time if we just get rid of VIRT_MMIO completely
and allow users to create the mmio-virtio bridges using -device for -M
virt-2.4 and above?
At the end of the day, I'm fairly sure people will end up virtio-pci anyway
and it's just a big waste of address space to keep VIRT_MMIO around, no?
I'm not sure I have an opinion one way or the other, but I would like to
understand the "big waste" argument. Is there something that users are eager
to reuse this address space for, like more RAM?
It will get used for platform device assignment. If we only reserve 32MB
we will quickly run out of address space to map devices into.
Alex
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
