Re: [Qemu-devel] [PATCH v4 1/3] nvdimm acpi: introduce fit buffer

[Xiao Guangrong <guangrong.xiao@linux.intel.com>]

On 11/03/2016 06:00 PM, Stefan Hajnoczi wrote:
> On Thu, Nov 03, 2016 at 11:51:28AM +0800, Xiao Guangrong wrote:
>> +static void nvdimm_init_fit_buffer(NvdimmFitBuffer *fit_buf)
>> +{
>> +    fit_buf->fit = g_array_new(false, true /* clear */, 1);
>> +}
>> +
>> +static void nvdimm_build_fit_buffer(NvdimmFitBuffer *fit_buf)
>> +{
>> +    g_array_free(fit_buf->fit, true);
>> +    fit_buf->fit = nvdimm_build_device_structure();
>
> In the previous revision I pointed out that it's messy to inline
> g_array_new(false, true /* clear */, 1) in nvdimm_init_fit_buffer() when
> the data structure is normally created by
> nvdimm_build_device_structure().  You didn't respond.
>

Oh, sorry for that.

> Is it possible to call nvdimm_build_device_structure() in
> nvdimm_init_fit_buffer() so we don't need to duplicate the details of
> how the GArray is created?
>

Actually, i tried your suggestion however i noticed it makes the code
little confuse, as we construct the fit by calling
nvdimm_build_device_structure() whose name shows the fit will be
rebuild but do not set .dirty in the init path.

g_array_new(false, true /* clear */, 1) is a more clear way to
show this is just a empty array.