Re: [PATCH v3 1/1] Add pci_hole_min_size

[Don Slutz <dslutz@verizon.com>]

On 07/01/14 08:48, Jan Beulich wrote:
>>>> On 01.07.14 at 13:56, <dslutz@verizon.com> wrote:
>> On 06/27/14 04:34, Jan Beulich wrote:
>>> Furthermore the whole concept collides with the auto-growing of
>>> the hole immediately following this code. Both logically (both having
>>> a similar purpose) and, what's worse, functionally in that the
>>> shifting done there requires that ((1ULL << 32) - pci_mem_start)
>>> be a power of 2. An option might be to skip this auto-sizing if the
>>> option was given.
>> I am fine with disabling the auto-growing (since as far as I know
>> this is bug #28 (
>>
>>
>>    #28 - support PCI hole resize in qemu-xen
>>
>> )
> But that's only with upstream qemu I suppose; the auto-resizing
> was added for a reason (see the history of the file). So this is
> definitely not going to be an unconditional disabling (other than
> what I kind of read from your reply).

I was not clear.  I am thinking that if pci_hole_min_size (or what ever
the name is) is non-zero (i.e. not default (could be exists also)), I would
disable the auto-resizing code.  As far as I know, the auto-resizing only
works with qemu-xen-traditional.  I was not planning to add
max-ram-below-4g to qemu-xen-traditional (but it would not be a big
patch nor as complex as what is in qemu-xen (upstream master)).

so currently it would be:

+   }
+    else
+   {
/*
      * At the moment qemu-xen can't deal with relocated memory regions.
      * It's too close to the release to make a proper fix; for now,
      * only allow the MMIO hole to grow large enough to move guest memory
      * if we're running qemu-traditional.  Items that don't fit will be
      * relocated into the 64-bit address space.
*
      * This loop now does the following:
      * - If allow_memory_relocate, increase the MMIO hole until it's
      *   big enough, or until it's 2GiB
      * - If !allow_memory_relocate, increase the MMIO hole until it's
      *   big enough, or until it's 2GiB, or until it overlaps guest
      * memory
      */
     while ( (mmio_total > (pci_mem_end - pci_mem_start))
             && ((pci_mem_start << 1) != 0)
             && (allow_memory_relocate
                 || (((pci_mem_start << 1) >> PAGE_SHIFT)
                     >= hvm_info->low_mem_pgend)) )
         pci_mem_start <<= 1;
+  }



But looking closer I could just set:

HVM_XS_ALLOW_MEMORY_RELOCATE to 0

to stop part of this.  And the other part should be false do to
xc_hvm_build_with_hole()

but this will need testing.

      -Don Slutz


> Jan
>