Re: [Qemu-devel] [PATCH v4 01/16] memory: Allow subregions to not be printed by info mtree

[Alistair Francis]

On Fri, Feb 26, 2016 at 8:54 AM, Peter Maydell <address@hidden> wrote:
> On 26 February 2016 at 16:51, Alex Bennée <address@hidden> wrote:
>> I think this needlessly complicates the memory region code and I'm not
>> sure what is too be gained for the register code. The only usage of the
>> code is inside a loop in register_init_block32. In each case the region
>> has the same set of ops. Why isn't a single region being created with an
>> indirect handler which can dispatch to the individual register handling
>> code?
>>
>> While its true some drivers create individual IO regions by an large
>> most are creating a block with a common handler.
>
> Yeah, I have to say I'm not really convinced about having one MR
> per register -- the MR code was never intended to be used that way,
> and it seems like a good way to find nasty performance or memory
> usage surprises.

I have removed the one memory region per register.

Thanks,

Alistair

>
> thanks
> -- PMM
>