On 18 April 2013 21:29, Marc Zyngier <marc.zyngier@xxxxxxx> wrote: > On Thu, 18 Apr 2013 11:25:56 -0400, Christopher Covington > <cov@xxxxxxxxxxxxxx> wrote: >> Hi Pranavkumar, >> >> On 04/18/2013 01:52 AM, PranavkumarSawargaonkar wrote: >>> From: Pranavkumar Sawargaonkar <pranavkumar@xxxxxxxxxx> >>> >>> This patch implements early printk support for virtio-mmio console >>> devices >>> without using any hypercalls. >> >> Is it possible that using DCC might be an easier solution? > > You would end up with the exact same problem as with a hypercall based > solution: once you're in the hypervisor, what do you do with the data? You > end up having to invent another channel to forward it down to your platform > emulation in order to have it printed where the user expects it. > > Using a MMIO based solution is probably the best solution, as it uses the > existing infrastructure. Completely agree with Marc that instead of writing hypercalls (which will have hypervisor and arch specific implementation) use of existing MMIO and virtio console layer is easier and good option. - Pranav > > M. > -- > Fast, cheap, reliable. Pick two. _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
