PCI MSI API support in backports

PCI MSI API support in backports

[Omer Efrat]

Our wireless driver uses pci_alloc_irq_vectors function in order to allocat=
e the pci irq vectors.
(hardware supports MSI, with no MSIx support)

When trying to backport from linux-stable v4.16 to v3.10.104, compilation f=
ails.
It seems that backports didn't backported pci_alloc_irq_vectors function (i=
ntroduced at aff171641d181ea573380efc3f559c9de4741fc5 PCI: Provide sensible=
 IRQ vector alloc/free routines), although MSIx APIs are supported in backp=
orts.

Is that makes sense?
Are changes to PCI driver being included in backports by their owners as a =
workflow?
Is pci_alloc_irq_vectors function planned to be included in backports?
Are we missing something?

Best regards,
Omer Efrat.=
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: PCI MSI API support in backports

[Arend van Spriel]

On 5/23/2018 1:39 PM, Omer Efrat wrote:
> Our wireless driver uses pci_alloc_irq_vectors function in order to allocate the pci irq vectors.
> (hardware supports MSI, with no MSIx support)
>
> When trying to backport from linux-stable v4.16 to v3.10.104, compilation fails.
> It seems that backports didn't backported pci_alloc_irq_vectors function (introduced at aff171641d181ea573380efc3f559c9de4741fc5 PCI: Provide sensible IRQ vector alloc/free routines), although MSIx APIs are supported in backports.
>
> Is that makes sense?
> Are changes to PCI driver being included in backports by their owners as a workflow?
> Is pci_alloc_irq_vectors function planned to be included in backports?
> Are we missing something?

Well. I think what you are missing is a wireless driver in linux 
upstream. We backport on a as-needed basis so the reason is probably 
that none of the upstream drivers require the pci_alloc_irq_vectors() 
function. So what wireless driver are we talking about here. Is it a 
proprietary driver?

Regards,
Arend

--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: PCI MSI API support in backports

[Omer Efrat]

On 5/23/2018 3:21 PM, Arend van Spriel wrote:
> Well. I think what you are missing is a wireless driver in linux=20
> upstream. We backport on a as-needed basis so the reason is probably=20
> that none of the upstream drivers require the pci_alloc_irq_vectors()=20
> function. So what wireless driver are we talking about here. Is it a=20
> proprietary driver?

Yes, we are talking about a proprietary driver.
Does that mean I should just add pci_alloc_irq_vectors function myself and =
send it here as a patch?

Best regards,
Omer Efrat.=
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: PCI MSI API support in backports

[Arend van Spriel]

On 5/24/2018 3:40 PM, Omer Efrat wrote:
> On 5/23/2018 3:21 PM, Arend van Spriel wrote:
>> Well. I think what you are missing is a wireless driver in linux
>> upstream. We backport on a as-needed basis so the reason is probably
>> that none of the upstream drivers require the pci_alloc_irq_vectors()
>> function. So what wireless driver are we talking about here. Is it a
>> proprietary driver?
>
> Yes, we are talking about a proprietary driver.
> Does that mean I should just add pci_alloc_irq_vectors function myself and send it here as a patch?

You could, but I asked about your driver because the backports project 
exports its symbols using EXPORT_SYMBOL_GPL(). This means that only 
GPL-compatible drivers can use it. This has been a conscious decision 
(see [1]).

Regards,
Arend

[1] https://backports.wiki.kernel.org/index.php/License
--
To unsubscribe from this list: send the line "unsubscribe backports" in