Re: [RFC PATCH v1 2/3] PCI: hisi: Add PCIe host support for Hisilicon Soc Hip05

[Zhou Wang <wangzhou1@hisilicon.com>]

On 2015/5/21 20:09, Bjorn Helgaas wrote:
> On Wed, May 20, 2015 at 8:41 PM, Zhou Wang <wangzhou1@hisilicon.com> wrote:
>> On 2015/5/20 20:50, Jingoo Han wrote:
>>> On Wed, 20 May 2015 14:21:40 +0800, Zhou Wang wrote:
> 
>>>> +/* Configure vmid/asid table in PCIe host */
>>>> +static void hisi_pcie_config_context(struct hisi_pcie *pcie)
>>>> +{
>>>> +    int i;
>>>> +
>>>> +    hisi_pcie_change_apb_mode(pcie, PCIE_SLV_CONTENT_MODE);
>>>> +
>>>> +    for (i = 0; i < 0x400; i++)
>>>> +            hisi_pcie_apb_writel(pcie, 0x0, i * 4);
>>>> +
>>>> +    for (i = 0x400; i < 0x800; i++)
>>>> +            hisi_pcie_apb_writel(pcie, 0x0, i * 4);
>>>
>>> How about the following?
>>>
>>> +     for (i = 0; i < 0x800; i++)
>>> +             hisi_pcie_apb_writel(pcie, 0x0, i * 4);
>>>
>>
>> This is to configure init value of vmid and asid of each pcie device.
> 
> But the combined loop looks functionally equivalent, i.e., I don't
> think the device can tell the difference.  Personally I wouldn't mind
> if you kept the two loops separate, but if you do, each one needs a
> comment immediately before it so the reader has a clue that they are
> doing semantically different things.  Otherwise it just looks like a
> mistake and the reader wonders why they are split.
> 
> Bjorn

Got it. There is a vmid table and a asid table in host controll
registers. They are just adjoining as above. I will combine above loops
with more comments together.

Thanks and Regards,
Zhou

> 
> .
>