Re: [Qemu-devel] Executing from a rom device - Re: [PATCH 2/4] pc: Support system flash memory with pflash

[Jordan Justen]

Avi, all,

Would it be acceptable to implement the pflash support for qemu only
at this time?  If kvm is enabled, I could install ram (in read-only
mode) as is done today for the firmware to execute from.

Later, when the additional kvm feature you described below is
implemented, then I can update the qemu kvm path to enable this
support.  (I will also attempt to implement the kvm kernel+qemu pieces
of this.)

Thanks,

-Jordan

On Tue, Oct 25, 2011 at 00:47, Avi Kivity <address@hidden> wrote:
> On 10/25/2011 01:00 AM, Jordan Justen wrote:
>> Avi,
>>
>> Alex pointed out that my patch below should not work for kvm, because
>> kvm currently does not support executing from a rom region.  This
>> surprised me, because I thought I had been testing with kvm enabled.
>> But, it turns out I wasn't, and in fact this patch does not work with
>> kvm enabled.  (Sorry all for this big mistake in my testing.)
>>
>> Alex also suggested that you might be able to answer whether it would
>> be possible to execute from a flash device (ie, qemu 'rom_device'
>> device).  Is this something that would be possible with kvm?
>>
>> If so, would it require changes to kvm on the kernel side?  Or, the
>> qemu side?  Perhaps just within the pflash_cfi01 device?
>>
>> Would implementing it require the flash based execution to be very
>> slow under kvm?
>
> The core issue that kvm (the kernel part) supports two styles of memory:
> read/write RAM, and read/write MMIO.  ROM wants writes to be ignored,
> and rom/device wants reads serviced from memory and writes serviced by
> userspace (as MMIO).
>
> It should not be too hard to patch kvm to support these additional two
> styles.  The entry point is the KVM_SET_USER_MEMORY_REGION ioctl to
> define the new attributes for the region, and kvm_mmu_page_fault() to
> map these pages as read-only and emulate writes (for ROM/device regions).
>
> --
> I have a truly marvellous patch that fixes the bug which this
> signature is too narrow to contain.
>
>