Re: [PATCH v1 0/9] s390: implement and optimize vmemmap_free()

[David Hildenbrand <david@redhat.com>]

On 08.07.20 08:50, David Hildenbrand wrote:
> On 07.07.20 14:13, David Hildenbrand wrote:
>> On 07.07.20 14:08, Heiko Carstens wrote:
>>> On Fri, Jul 03, 2020 at 03:39:08PM +0200, David Hildenbrand wrote:
>>>> This series is based on the latest s390/features branch [1]. It implements
>>>> vmemmap_free(), consolidating it with vmem_add_range(), and optimizes it by
>>>> - Freeing empty page tables (now also done for idendity mapping).
>>>> - Handling cases where the vmemmap of a section does not fill huge pages
>>>>   completely.
>>>>
>>>> vmemmap_free() is currently never used, unless adiing standby memory fails
>>>> (unlikely). This is relevant for virtio-mem, which adds/removes memory
>>>> in memory block/section granularity (always removes memory in the same
>>>> granularity it added it).
>>>>
>>>> I gave this a proper test with my virtio-mem prototype (which I will share
>>>> once the basic QEMU implementation is upstream), both with 56 byte memmap
>>>> per page and 64 byte memmap per page, with and without huge page support.
>>>> In both cases, removing memory (routed through arch_remove_memory()) will
>>>> result in
>>>> - all populated vmemmap pages to get removed/freed
>>>> - all applicable page tables for the vmemmap getting removed/freed
>>>> - all applicable page tables for the idendity mapping getting removed/freed
>>>> Unfortunately, I don't have access to bigger and z/VM (esp. dcss)
>>>> environments.
>>>>
>>>> This is the basis for real memory hotunplug support for s390x and should
>>>> complete my journey to s390x vmem/vmemmap code for now :)
>>>>
>>>> What needs double-checking is tlb flushing. AFAIKS, as there are no valid
>>>> accesses, doing a single range flush at the end is sufficient, both when
>>>> removing vmemmap pages and the idendity mapping.
>>>>
>>>> Along, some minor cleanups.
>>>
>>> Hmm.. I really would like to see if there would be only a single page
>>> table walker left in vmem.c, which handles both adding and removing
>>> things.
>>> Now we end up with two different page table walk implementations
>>> within the same file. However not sure if it is worth the effort to
>>> unify them though.
>>
>> I tried to unify vmemmap_populate() and vmem_add_range() already and
>> didn't like the end result ... so, unifying these along with the removal
>> part won't be any better - most probably. Open for suggestions :)
>>
>> (at least arm64 and x86-64 handle it similarly)
>>
> 
> I'll play with something like
> 
> static void modify_pagetable(unsigned long start, unsigned long end,
> 			     bool direct, bool add)
> 
> and see how it turns out.
> 

Did a quick hack. With a single walker (modify_pagetable) I get

 arch/s390/mm/vmem.c | 628 ++++++++++++++++++++++++++++++--------------
 1 file changed, 434 insertions(+), 194 deletions(-)

Overall looks cleaner, only modify_pte_table() and modify_pmd_table()
are a little more involved ...

-- 
Thanks,

David / dhildenb