Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1552765 > unrolled thread

Re: [PATCH v5] PCI: add a new directory for designware core

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v5] PCI: add a new directory for designware core Arnd Bergmann <arnd@arndb.de> - 2017-01-06 14:10 +0100
    Re: [PATCH v5] PCI: add a new directory for designware core Kishon Vijay Abraham I <kishon@ti.com> - 2017-01-06 14:20 +0100
      Re: [PATCH v5] PCI: add a new directory for designware core Joao Pinto <Joao.Pinto@synopsys.com> - 2017-01-06 14:50 +0100
        Re: [PATCH v5] PCI: add a new directory for designware core Kishon Vijay Abraham I <kishon@ti.com> - 2017-01-09 10:40 +0100

#1552765 — Re: [PATCH v5] PCI: add a new directory for designware core

On Friday, January 6, 2017 6:22:48 PM CET Kishon Vijay Abraham I wrote:
> Group all the PCI drivers that use designware core in dwc directory.
> dwc IP is capable of operating in both host mode and device mode and
> keeping it inside the *host* directory is misleading.

I have no objections to the patch, it makes a lot of sense, but I think
the reasoning above is a bit questionable, as a lot of the other
host drivers (at least tegra, mvebu and xgene, probably more) equally support
endpoint mode and remain in the same directory.

Any suggestion to what we should do with them?

	Arnd

[toc] | [next] | [standalone]

#1552772

Hi Arnd,

On Friday 06 January 2017 06:31 PM, Arnd Bergmann wrote:
> On Friday, January 6, 2017 6:22:48 PM CET Kishon Vijay Abraham I wrote:
>> Group all the PCI drivers that use designware core in dwc directory.
>> dwc IP is capable of operating in both host mode and device mode and
>> keeping it inside the *host* directory is misleading.
> 
> I have no objections to the patch, it makes a lot of sense, but I think
> the reasoning above is a bit questionable, as a lot of the other
> host drivers (at least tegra, mvebu and xgene, probably more) equally support
> endpoint mode and remain in the same directory.
> 
> Any suggestion to what we should do with them?

I'm not sure of the exact IP's used by these SoCs. Now that we decided to group
the drivers by IPs used, the best would be identify the IPs used in these SoCs
and create a separate directory for each of the IPs. If the vendors own IP is
used, then create a directory with the name of the vendor?. I'm sure such a
change would increase the number of directories but I'm not sure what else to
do here.

Thanks
Kishon

[toc] | [prev] | [next] | [standalone]

#1552792

Hello,

Às 1:12 PM de 1/6/2017, Kishon Vijay Abraham I escreveu:
> Hi Arnd,
> 
> On Friday 06 January 2017 06:31 PM, Arnd Bergmann wrote:
>> On Friday, January 6, 2017 6:22:48 PM CET Kishon Vijay Abraham I wrote:
>>> Group all the PCI drivers that use designware core in dwc directory.
>>> dwc IP is capable of operating in both host mode and device mode and
>>> keeping it inside the *host* directory is misleading.
>>
>> I have no objections to the patch, it makes a lot of sense, but I think
>> the reasoning above is a bit questionable, as a lot of the other
>> host drivers (at least tegra, mvebu and xgene, probably more) equally support
>> endpoint mode and remain in the same directory.
>>
>> Any suggestion to what we should do with them?
> 
> I'm not sure of the exact IP's used by these SoCs. Now that we decided to group
> the drivers by IPs used, the best would be identify the IPs used in these SoCs
> and create a separate directory for each of the IPs. If the vendors own IP is
> used, then create a directory with the name of the vendor?. I'm sure such a
> change would increase the number of directories but I'm not sure what else to
> do here.

In case of designware based drivers you already did the clean up nicely. Next
would be, like Arnd is sugesting, to do the same to the others. If they are all
self-contained I suggest we leave them as they are and in the future if a
endpoint variant driver comes up for any of them, we can suggest to add a folder
with the vendor or SoC name in order to follow the rule. What do you think?


Joao

> 
> Thanks
> Kishon
> 

[toc] | [prev] | [next] | [standalone]

#1554170

Hi,

On Friday 06 January 2017 07:10 PM, Joao Pinto wrote:
> Hello,
> 
> Às 1:12 PM de 1/6/2017, Kishon Vijay Abraham I escreveu:
>> Hi Arnd,
>>
>> On Friday 06 January 2017 06:31 PM, Arnd Bergmann wrote:
>>> On Friday, January 6, 2017 6:22:48 PM CET Kishon Vijay Abraham I wrote:
>>>> Group all the PCI drivers that use designware core in dwc directory.
>>>> dwc IP is capable of operating in both host mode and device mode and
>>>> keeping it inside the *host* directory is misleading.
>>>
>>> I have no objections to the patch, it makes a lot of sense, but I think
>>> the reasoning above is a bit questionable, as a lot of the other
>>> host drivers (at least tegra, mvebu and xgene, probably more) equally support
>>> endpoint mode and remain in the same directory.
>>>
>>> Any suggestion to what we should do with them?
>>
>> I'm not sure of the exact IP's used by these SoCs. Now that we decided to group
>> the drivers by IPs used, the best would be identify the IPs used in these SoCs
>> and create a separate directory for each of the IPs. If the vendors own IP is
>> used, then create a directory with the name of the vendor?. I'm sure such a
>> change would increase the number of directories but I'm not sure what else to
>> do here.
> 
> In case of designware based drivers you already did the clean up nicely. Next
> would be, like Arnd is sugesting, to do the same to the others. If they are all
> self-contained I suggest we leave them as they are and in the future if a
> endpoint variant driver comes up for any of them, we can suggest to add a folder
> with the vendor or SoC name in order to follow the rule. What do you think?

That's my idea as well. Arnd, do you think it's okay not to disturb other
drivers/directories until endpoint support is added to those drivers?

Thanks
Kishon

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web