On 10/16/2015 05:09 PM, Aneesh Kumar K.V wrote:
> Anshuman Khandual <khand...@linux.vnet.ibm.com> writes:
>
>> On 10/15/2015 09:16 PM, Anshuman Khandual wrote:
>>> This adds two tests for memory page migration. One for normal page
>>> migration which works for both 4K or 64K base page size kernel and
>>> the other one is for 16MB huge page migration which will work both
>>> 4K or 64K base page sized 16MB huge pages as and when we support
>>> huge page migration.
>>>
>>> Signed-off-by: Anshuman Khandual <khand...@linux.vnet.ibm.com>
>>> ---
>>> - Works for normal page migration on both 64K and 4K base pages
>>> - Works for 16MB huge page migration (64K) on Aneesh's V2 PTE changes
>>>
>>> +
>>> +int test_migration(unsigned long length)
>>> +{
>>> + unsigned long failed;
>>> + void *addr;
>>> + int ret;
>>> +
>>> + addr = mmap(MMAP_ADDR, length, MMAP_PROT, MMAP_FLAGS, -1, 0);
>>> + if (addr == MAP_FAILED) {
>>> + perror("mmap() failed");
>>> + exit(-1);
>>> + }
>>
>> Will add a mlock() call here as well. Some times soft offline
>> fails while trying to move a huge chunk of memory on a system
>> because of swapping. With a mlock() in there the problem gets
>> resolved.
>
> Is that with respect to hugetlb pages ? I doubt, because hugetlb pages
> are not swapped out.
Thats correct. It is related to normal pages (4K or 64K). Have not
seen this swap problem of HugeTLB pages.
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev
