Re: [PATCH v1 7/9] memory: introduce RAM_NORESERVE and wire it up in qemu_ram_mmap()

[Thomas Huth <thuth@redhat.com>]

On 03/03/2021 12.35, Cornelia Huck wrote:
> On Tue, 2 Mar 2021 20:02:34 +0100
> David Hildenbrand <david@redhat.com> wrote:
> 
>> On 02.03.21 18:32, Peter Xu wrote:
>>> On Tue, Feb 09, 2021 at 02:49:37PM +0100, David Hildenbrand wrote:
>>>> @@ -899,13 +899,17 @@ int kvm_s390_mem_op_pv(S390CPU *cpu, uint64_t offset, void *hostbuf,
>>>>     * to grow. We also have to use MAP parameters that avoid
>>>>     * read-only mapping of guest pages.
>>>>     */
>>>> -static void *legacy_s390_alloc(size_t size, uint64_t *align, bool shared)
>>>> +static void *legacy_s390_alloc(size_t size, uint64_t *align, bool shared,
>>>> +                               bool noreserve)
>>>>    {
>>>>        static void *mem;
>>>>    
>>>>        if (mem) {
>>>>            /* we only support one allocation, which is enough for initial ram */
>>>>            return NULL;
>>>> +    } else if (noreserve) {
>>>> +        error_report("Skipping reservation of swap space is not supported.");
>>>> +        return NULL
>>>
>>> Semicolon missing.
>>
>> Thanks for catching that!
> 
> Regardless of that (and this patch set), can we finally get rid of
> legacy_s390_alloc? We already fence off running with a kernel prior to
> 3.15, and KVM_CAP_S390_COW depends on ESOP -- are non-ESOP kvm hosts
> still relevant? This seems to be a generation 10 feature; do we
> realistically expect anyone running this on e.g. a z/VM host that
> doesn't provide ESOP?

Looking at the support charts ( 
https://www.ibm.com/support/pages/ibm-mainframe-life-cycle-history ), the 
z10 is already unsupported. So if all newer mainframes have ESOP, I guess it 
should be fine to get rid of this code now.

  Thomas