Re: [RFC] Build with arm64 configuration

[Jaehoon Chung <jh80.chung@samsung.com>]

Dear, Liviu.

On 03/19/2015 12:16 AM, Liviu Dudau wrote:
> On Wed, Mar 11, 2015 at 08:00:11AM +0000, Jaehoon Chung wrote:
>> Dear, Bjorn.
>>
>> On 03/10/2015 11:17 AM, Bjorn Helgaas wrote:
>>> [+cc Jingoo, Mohit]
>>>
>>> On Mon, Mar 9, 2015 at 8:36 PM, Jaehoon Chung <jh80.chung@samsung.com> wrote:
>>>> Dear, Bjorn.
>>>>
>>>> Thanks for reply.
>>>>
>>>> On 03/10/2015 08:35 AM, Bjorn Helgaas wrote:
>>>>> On Mon, Mar 9, 2015 at 7:18 AM, Jaehoon Chung <jh80.chung@samsung.com> wrote:
>>>>>> Hi,
>>>>>>
>>>>>> I have a question for building arm64.
>>>>>> (Actually, i didn't have many knowledges for PCIe.)
>>>>>> When i built with arm64 configuration, then i always found the compiler errors.
>>>>>
>>>>> The kernel does build for arm64, so you'll have to be more specific
>>>>> about the problem you're seeing.
>>>>
>>>> I built the pcie-designware.c and pci-exynos.c.
>>>>
>>>> When pcie-designware.c is built, i found the below error message.
>>>> drivers/pci/host/pcie-designware.c:74:52: warning: Ëœstruct pci_sys_dataâ„¢ declared inside parameter list
>>>>  static inline struct pcie_port *sys_to_pcie(struct pci_sys_data *sys)
>>>>                                                     ^
>>>> drivers/pci/host/pcie-designware.c:74:52: warning: its scope is only this definition or declaration, which is probably not what you want
>>>> In file included from include/uapi/linux/stddef.h:1:0,
>>>>                  from include/linux/stddef.h:4,
>>>>                  from ./include/uapi/linux/posix_types.h:4,
>>>>                  from include/uapi/linux/types.h:13,
>>>>                  from include/linux/types.h:5,
>>>>                  from include/linux/smp.h:10,
>>>>                  from include/linux/irq.h:12,
>>>>                  from drivers/pci/host/pcie-designware.c:14:
>>>> drivers/pci/host/pcie-designware.c: In function Ëœsys_to_pcie":
>>>> drivers/pci/host/pcie-designware.c:76:13: error: dereferencing pointer to incomplete type
>>>>   BUG_ON(!sys->private_data);
>>>>
>>>> I found it's located at arch/arm/include/asm/mach/pci.h.
>>>> It can't refer when do build for arm64.
>>>> So I think that i missed something..I have also searched the patches relevant to this at patchwork.
>>>> Some patch have copied header files relevant to "pci" to "arch/arm64/include/".
>>>> (This is not solution.)
>>>>
>>>> And i don't want to add #ifdef CONFIG_ARM64 into pcie-designware.c
>>>> I think mailing should know the graceful solution.
>>>>
>>>> If i can solve this problem, i will contribute for pci-exynos.c.
>>>
>>> Oh, I think pcie-designware.c and pci-exynos.c currently only work on
>>> arm, not on arm64.  I don't know enough about them to know whether
>>> they could be made to work on arm64.  But Jingoo and Mohit probably
>>> do.
> 
> Jaehoon,
> 
> pci-exynos and pcie-designware need to be converted to the generic PCIe framework and then
> they will work on both arm and arm64. For an example of a driver that successfully works
> in both worlds look at pci-versatile.c.

Thanks a lot! I'm looking at pci-versatile.c and comparing with pcie-designware and pci-exynos.

One question..
At pci-versatile.c, i found the resource_list_for_each_entry() into versatile_pci_parse_request_of_pci_ranges().
resource_list_for_each_entry needs two arguments, but it's passed three arguments.
It seems to want using "resource_list_for_each_entry_safe()", doesn't it? (+CC'd Rob Herring)

diff --git a/drivers/pci/host/pci-versatile.c b/drivers/pci/host/pci-versatile.c
index 1ec694a..8eb3b80 100644
--- a/drivers/pci/host/pci-versatile.c
+++ b/drivers/pci/host/pci-versatile.c
@@ -80,7 +80,7 @@ static int versatile_pci_parse_request_of_pci_ranges(struct device *dev,
        if (err)
                return err;
 
-       resource_list_for_each_entry(win, res, list) {
+       resource_list_for_each_entry_safe(win, res, list) {
                struct resource *parent, *res = win->res;
 
                switch (resource_type(res)) {


Best Regards,
Jaehoon Chung

> 
> Currently, PCI host bridges that work on arch/arm depend on pci_sys_data, like you have
> discovered. There is no reason to do that nowadays, and Lorenzo Pieralisi, Yijing Wang
> and others are trying to get rid of that structure from arch/arm.
> 
> Best regards,
> Liviu
> 
>>
>> I'm checking the code..
>> If my understanding is right, pci_sys_data is just used to take data relevant to pice_port.
>> So I think it can get from device-tree or other.
>> I don't exactly know why needs "pci_sys_data".
>>
>> It seems to discuss about pcie (struct pci_sys_data and hw_pci) on arm64 at mailing..right?
>> ([RFC 2/4] PCI: generic: Add support for ARM64 and MSI(x))
>>
>> If need to work, i have interesting for contributing pcie on arm64.(pcie-designware.c and pci-exynos.c)
>> But i need to gain the knowledge of pcie..
>>
>> Best Regards,
>> Jaehoon Chung
>>
>>>
>>> Bjorn
>>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>