This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


[Xen-devel] Re: [PATCH v2 4/5] exec.c: refactor cpu_physical_memory_map

On 12.07.2011, at 00:17, Jan Kiszka wrote:

> On 2011-05-19 19:35, stefano.stabellini@xxxxxxxxxxxxx wrote:
>> From: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
>> Introduce qemu_ram_ptr_length that takes an address and a size as
>> parameters rather than just an address.
>> Refactor cpu_physical_memory_map so that we call qemu_ram_ptr_length only
>> once rather than calling qemu_get_ram_ptr one time per page.
>> This is not only more efficient but also tries to simplify the logic of
>> the function.
>> Currently we are relying on the fact that all the pages are mapped
>> contiguously in qemu's address space: we have a check to make sure that
>> the virtual address returned by qemu_get_ram_ptr from the second call on
>> is consecutive. Now we are making this more explicit replacing all the
>> calls to qemu_get_ram_ptr with a single call to qemu_ram_ptr_length
>> passing a size argument.
> This breaks cpu_physical_memory_map for >4G addresses on PC.
> Effectively, it doesn't account for the PCI gap, ie. that the RAM block
> is actually mapped in two chunks into the guest physical memory. One
> outcome is that QEMU aborts when we try to process an address that is
> now "outside" RAM. Simple to reproduce with a virtio NIC and 5G guest
> memory, even without KVM.

Do you have some reliable test case? I can't seem to reproduce the issue. It 
works just fine for me with -m 10G, virtio nic and disk, polluting all the 
guest page cache.


Xen-devel mailing list