On 13/04/16 12:23, Matthias Brugger wrote:
>
>
> On 13/04/16 12:56, Marc Zyngier wrote:
>> On 13/04/16 11:30, Matthias Brugger wrote:
>>> From: Matthias Brugger <matthias.bgg@xxxxxxxxx>
>>>
>>> The fsl-mc driver can't be build as a module because it uses msi_*
>>> functions directly. Port the driver to use the platform_msi_*
>>> infrastructure instead, to allow it to be build as a module.
>>>
>>> Signed-off-by: Matthias Brugger <mbrugger@xxxxxxxx>
>>> ---
>>> .../staging/fsl-mc/bus/irq-gic-v3-its-fsl-mc-msi.c | 5 +-
>>> drivers/staging/fsl-mc/bus/mc-allocator.c | 9 +-
>>> drivers/staging/fsl-mc/bus/mc-msi.c | 169 +--------------------
>>> drivers/staging/fsl-mc/include/mc-sys.h | 3 +
>>> 4 files changed, 14 insertions(+), 172 deletions(-)
>>>
>>> diff --git a/drivers/staging/fsl-mc/bus/irq-gic-v3-its-fsl-mc-msi.c b/drivers/staging/fsl-mc/bus/irq-gic-v3-its-fsl-mc-msi.c
>>> index 720e2b0..0eecb7e 100644
>>> --- a/drivers/staging/fsl-mc/bus/irq-gic-v3-its-fsl-mc-msi.c
>>> +++ b/drivers/staging/fsl-mc/bus/irq-gic-v3-its-fsl-mc-msi.c
>>> @@ -25,7 +25,6 @@ static struct irq_chip its_msi_irq_chip = {
>>> .irq_mask = irq_chip_mask_parent,
>>> .irq_unmask = irq_chip_unmask_parent,
>>> .irq_eoi = irq_chip_eoi_parent,
>>> - .irq_set_affinity = msi_domain_set_affinity
>>> };
>>>
>>> static int its_fsl_mc_msi_prepare(struct irq_domain *msi_domain,
>>> @@ -86,7 +85,7 @@ int __init its_fsl_mc_msi_init(void)
>>> continue;
>>> }
>>>
>>> - mc_msi_domain = fsl_mc_msi_create_irq_domain(
>>> + mc_msi_domain = platform_msi_create_irq_domain(
>>> of_node_to_fwnode(np),
>>> &its_fsl_mc_msi_domain_info,
>>> parent);
>>
>> What? We are already creating a platform MSI domain for the ITS. How is
>> that going to work? If you want to convert this set of drivers to
>> platform ITS, fine. But you can't randomly hack in the ITS code and pray
>> for things not to fall apart.
>>
>
> From what I see, the difference between irq-gic-v3-its-fsl-mc-msi and
> the irq-gic-v3-its-platform-msi is the way ITS specific DeviceID is
> created in msi_prepare.
It is not "created". It is extracted from the HW, either by looking at
the RequesterID (PCI), at the DT (platform MSI), or a bus-specific method.
> German, is there a reason why you use the ICID read from the DPRC as dev_id?
Because that's what is presented to the ITS as a DevID. This is a HW
constraint, and you can't just change it.
If you want to use platform MSI for that, you also need to describe the
DevID in the DT (breaking the existing platforms in the process).
M.
--
Jazz is not dead. It just smells funny...
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
