Re: [Qemu-devel] [PATCH v4 04/20] hw/acpi/aml-build: Add aml_memory32_fixed() term

[Alex Bennée]

Michael S. Tsirkin <address@hidden> writes:

> On Wed, Apr 08, 2015 at 03:54:45PM +0100, Alex Bennée wrote:
>> > diff --git a/hw/acpi/aml-build.c b/hw/acpi/aml-build.c
>> > index 8d01959..fefe7c7 100644
>> > --- a/hw/acpi/aml-build.c
>> > +++ b/hw/acpi/aml-build.c
>> > @@ -505,6 +505,28 @@ Aml *aml_call4(const char *method, Aml *arg1, Aml 
>> > *arg2, Aml *arg3, Aml *arg4)
>> >      return var;
>> >  }
>> >  
>> > +/*
>> > + * ACPI 1.0: 6.4.3.4 Memory32Fixed (Memory Resource Descriptor Macro)
>> > + */
>> > +Aml *aml_memory32_fixed(uint64_t addr, uint64_t size, uint8_t rw_flag)
>> > +{
>> > +    Aml *var = aml_alloc();
>> 
>> This is more aimed at the ACPI maintainers but I wonder if there should
>> be an aml_alloc_sized that pre-allocates the GArray? Otherwise we spend
>> a lot of time realloc'ing while building these entries up. Or even a
>> varidac build_append_bytes?
>
> Can you show measureable VM boot speedup from this?
> If not, it's not worth bothering with.

Fair enough, this is a start-up thing and I guess you could probably
only measure a difference in a highly targeted pathaological test case.

-- 
Alex Bennée