Andreas,
Okay, let's try to talk today around 6. Skype or cellphone, I'll
be on both.
Brian
Quoting Andreas Nilsson <apn2107 at columbia.edu>:
> I can meet for instance tomorrow 6-6.30 pm something.
>
>
>
> From: android-virt-bounces at lists.cs.columbia.edu
> [mailto:android-virt-bounces at lists.cs.columbia.edu] On Behalf Of Christoffer
> Dall
> Sent: den 22 april 2009 22:26
> To: Brian Smith
> Cc: android-virt at lists.cs.columbia.edu
> Subject: Re: Merge to Master
>
>
>
> Hi Brian.
>
> I am on a _very_ tight schedule as is, since I have to go to Boston with a
> very packed schedule early Friday morning and I have a report and a homework
> for tomorrow plus I have to prepare a demo for Monday.
>
> However, Andreas and I have some 50 kernel boot instructions running using
> switched page tables before we get a crash, which is good news. We are now
> investigating excactly what causes the crash and next steps will be to
> handle this after having identified what goes wrong.
>
> Best,
> Christoffer
>
>
>
> On Wed, Apr 22, 2009 at 10:12 PM, Brian Smith <bls2129 at columbia.edu> wrote:
>
> Hey guys,
> I merged in my emulation code to the master branch. The highlights
> of what was added are:
>
> * Intercept all interrupts except for FIQ (Linux doesn't use FIQ). IRQs
> interrupts are handled as we discussed (couldn't drive this while the
> guest was running though)
> * Changed kvmarm_handle_swi to be called kvmarm_handle_exit, added
> framework for handling each type of host interrupt
> * Handle UNDEFINED interrupt by driving kvmarm_emulate_undefined_instr
> * Exception emulation (not thoroughly tested, working on ways to build
> scaffolding and get this done)
> * Coprocessor emulation, currently handles the basic get/set type regs
> (CR0, CR1). Other instructions such as TLB control/cache control, etc
> not implemented
>
> Note that kvmtest will now show the final regs as if the next
> instruction to occur is the SVC interrupt handler (the PC is set to
> 0x00000008). This is because kvmtest adds an SWI at the end, and the
> exception emulation code gets driven. After the emulation code, there
> is a temporary bit of code in kvmarm_handle_exit that tells
> arm_interrupts to return control to the host so we don't get stuck in
> the guest. I would like to meet before the weekend, when are you guys free?
>
> Brian
>
> _______________________________________________
> Android-virt mailing list
> Android-virt at lists.cs.columbia.edu
> https://lists.cs.columbia.edu/cucslists/listinfo/android-virt
>
>
>
>
