Re: [PATCH 1/2] Xen: Delay remapping memory of pv-domain

[Juergen Gross <jgross@suse.com>]

On 10/29/2014 06:30 AM, Juergen Gross wrote:
> On 10/28/2014 06:34 PM, David Vrabel wrote:
>> Can these patches be split up?  They're hard to review as-is.  e.g., the
>> changes to page allocation look they could be split out.
>>
>> On 27/10/14 14:52, Juergen Gross wrote:
>>> Early in the boot process the memory layout of a pv-domain is changed
>>> to match the E820 map (either the host one for Dom0 or the Xen one)
>>> regarding placement of RAM and PCI holes. This requires removing memory
>>> pages initially located at positions not suitable for RAM and adding
>>> them later at higher addresses where no restrictions apply.
>>
>> How does this impact Matt Rushton's recent change to ensure the
>> relocated frames are contiguous?
>
> It is simplifying it. :-)
>
> The relocated frames are in the same sequence as before, just at
> another location. As the remapping is done later, I don't have to be
> careful not to cross a p2m leaf page boundary (allocating a new leaf
> page is no problem now).
>
>>
>> The change was "xen/setup: Remap Xen Identity Mapped RAM" (4fbb67e3c87).
>>
>>> +    /* Update kernel mapping */
>>> +    if (HYPERVISOR_update_va_mapping((unsigned long)__va(pfn <<
>>> PAGE_SHIFT),
>>> +                     mfn_pte(mfn, PAGE_KERNEL), 0)) {
>>> +        WARN(1, "Failed to update kernel mapping for mfn=%ld
>>> pfn=%ld\n",
>>> +              mfn, pfn);
>>> +        return 0;
>>
>> I think you need to check if this is a PFN for a high page before
>> updating the mapping.
>
> Really? I'm not aware of having removed such a check. If it is really
> necessary it must have been missing before...

Wait, now I see my fault: by doing the remapping later I've skipped the
highmem check during adding the memory to the system. Okay, I'll update
the patch.


Juergen