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

Groups > linux.kernel > #1394820 > unrolled thread

Re: [PATCH v2 03/15] MIPS: PCI: Compatibility with ARM-like PCI host drivers

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 v2 03/15] MIPS: PCI: Compatibility with ARM-like PCI host  drivers Florian Fainelli <f.fainelli@gmail.com> - 2016-05-05 03:40 +0200
    Re: [PATCH v2 03/15] MIPS: PCI: Compatibility with ARM-like PCI host  drivers Paul Burton <paul.burton@imgtec.com> - 2016-05-05 13:10 +0200
      Re: [PATCH v2 03/15] MIPS: PCI: Compatibility with ARM-like PCI host  drivers Florian Fainelli <f.fainelli@gmail.com> - 2016-05-05 19:20 +0200

#1394820 — Re: [PATCH v2 03/15] MIPS: PCI: Compatibility with ARM-like PCI host drivers

Hi Paul,

On 04/04/16 03:09, Paul Burton wrote:
> Hi Florian,
> 
> Just an FYI, the pcie-xilinx driver I wrote this for has since been
> converted away from the ARM-like pci_common_init_dev & struct hw_pci to
> use only functions provided by the core PCI subsystem[1]. As a result
> I've stopped using this patch & don't plan to continue work on it.
> Perhaps it would be cleanest to do a similar conversion for the driver
> you're using?

Yes, I did just that, but as of v4.6-rc6, I am seeing a bunch of
undefined references while doing so:

arch/mips/pci/built-in.o: In function `pcibios_enable_device':
(.text+0x550): undefined reference to `pcibios_plat_dev_init'
arch/mips/pci/built-in.o: In function `pcibios_init':
pci.c:(.init.text+0x6c): undefined reference to `pcibios_map_irq'
pci.c:(.init.text+0x78): undefined reference to `pcibios_map_irq'

and this makes perfect sense because arch/mips/pci/pci.c is referencing
those functions, while I did not add anything for BMIPS_GENERIC.

At this point, I would very much prefer that the MIPS/Linux kernel did
not rely on the different machines to provide those implementations
(though it definitively is not a big deal to add them, it just feels
unnecessary), I will try to cook a patch for that and provide dummy
fallbacks.

Thanks!
-- 
Florian

[toc] | [next] | [standalone]

#1395036

On Wed, May 04, 2016 at 06:36:32PM -0700, Florian Fainelli wrote:
> Hi Paul,
> 
> On 04/04/16 03:09, Paul Burton wrote:
> > Hi Florian,
> > 
> > Just an FYI, the pcie-xilinx driver I wrote this for has since been
> > converted away from the ARM-like pci_common_init_dev & struct hw_pci to
> > use only functions provided by the core PCI subsystem[1]. As a result
> > I've stopped using this patch & don't plan to continue work on it.
> > Perhaps it would be cleanest to do a similar conversion for the driver
> > you're using?
> 
> Yes, I did just that, but as of v4.6-rc6, I am seeing a bunch of
> undefined references while doing so:
> 
> arch/mips/pci/built-in.o: In function `pcibios_enable_device':
> (.text+0x550): undefined reference to `pcibios_plat_dev_init'
> arch/mips/pci/built-in.o: In function `pcibios_init':
> pci.c:(.init.text+0x6c): undefined reference to `pcibios_map_irq'
> pci.c:(.init.text+0x78): undefined reference to `pcibios_map_irq'
> 
> and this makes perfect sense because arch/mips/pci/pci.c is referencing
> those functions, while I did not add anything for BMIPS_GENERIC.
> 
> At this point, I would very much prefer that the MIPS/Linux kernel did
> not rely on the different machines to provide those implementations
> (though it definitively is not a big deal to add them, it just feels
> unnecessary), I will try to cook a patch for that and provide dummy
> fallbacks.

Hi Florian,

I've done much the same for Boston already - do these patches work for
you?

    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=09f91e2742fa45ec6199e2657c4302ac432b7340
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=7fa01b789d863eed17dd948266085c636d43786f
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=7b070e41a9f7fc2c8e0cfd94baf90134f27e89eb
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=c1f71dfc3de4ec018d2c4d8877e81da19e500211
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=af3c4b3ed19d556489f67f7bb46f2dc83df7a617
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=15229366d674bc4e846d5ef779f62f8bb730fe55
    https://git.linux-mips.org/cgit/linux-mti.git/commit/?id=75b59b3bdddefda806e1b82b5b3aa60741df2e73

Or you could fetch & cherry-pick them from the linux-mti.git repository:

    git remote add mti git://git.linux-mips.org/pub/scm/linux-mti.git
    git fetch mti
    git cherry-pick 75b59b3bddde~7..75b59b3bddde

They apply cleanly to v4.5, hopefully won't need much massaging to apply
to current master. When I get to resubmitting the Boston board support
these patches will be part of it.

Thanks,
    Paul

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

#1395237

On 05/05/16 04:02, Paul Burton wrote:
> On Wed, May 04, 2016 at 06:36:32PM -0700, Florian Fainelli wrote:
>> Hi Paul,
>>
>> On 04/04/16 03:09, Paul Burton wrote:
>>> Hi Florian,
>>>
>>> Just an FYI, the pcie-xilinx driver I wrote this for has since been
>>> converted away from the ARM-like pci_common_init_dev & struct hw_pci to
>>> use only functions provided by the core PCI subsystem[1]. As a result
>>> I've stopped using this patch & don't plan to continue work on it.
>>> Perhaps it would be cleanest to do a similar conversion for the driver
>>> you're using?
>>
>> Yes, I did just that, but as of v4.6-rc6, I am seeing a bunch of
>> undefined references while doing so:
>>
>> arch/mips/pci/built-in.o: In function `pcibios_enable_device':
>> (.text+0x550): undefined reference to `pcibios_plat_dev_init'
>> arch/mips/pci/built-in.o: In function `pcibios_init':
>> pci.c:(.init.text+0x6c): undefined reference to `pcibios_map_irq'
>> pci.c:(.init.text+0x78): undefined reference to `pcibios_map_irq'
>>
>> and this makes perfect sense because arch/mips/pci/pci.c is referencing
>> those functions, while I did not add anything for BMIPS_GENERIC.
>>
>> At this point, I would very much prefer that the MIPS/Linux kernel did
>> not rely on the different machines to provide those implementations
>> (though it definitively is not a big deal to add them, it just feels
>> unnecessary), I will try to cook a patch for that and provide dummy
>> fallbacks.
> 
> Hi Florian,
> 
> I've done much the same for Boston already - do these patches work for
> you?

Yep, that's exactly what I needed, this worked great with the
pcie-brcmstb.c driver, thanks!

I can re-test them once you make a formal submission for the Boston board.

Cheers!
-- 
Florian

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web