Re: [Patch V1 0/6] Refine generic/PCI MSI irqodmian interfaces

[Jiang Liu <jiang.liu@linux.intel.com>]

On 2014/11/13 20:55, Yijing Wang wrote:
> On 2014/11/13 20:39, Jiang Liu wrote:
>>
>>
>> On 2014/11/13 20:28, Yijing Wang wrote:
>>> On 2014/11/13 19:43, Jiang Liu wrote:
>>>> This patch set is based on tip/irq/irqdomain and tries to refine
>>>> interfaces to support irqdomain for generic MSI and PCI MSI.
>>>>
>>>> Patch 1 is just minor fixes for tip/irq/irqdomain.
>>>>
>>>> Patch 2 introduces some helpers to hide struct msi_desc implementation
>>>> details, so later we could move msi_list from struct pci_dev into
>>>> struct device to enable generic MSI support.
>>>
>>> Hi Gerry,
>>>    I tried to move msi info(eg. msi_list) into struct device, but I found
>>> DMAR fault interrupt is initialized before the driver core init. And I don't
>>> know whether there are other devices like ARM consolidator(introduced in GIC v3)
>>> need to be initialized before driver core. What do you think about this ?
>> Hi Yijing,
>> 	DMAR interrupt doesn't make use of msi_desc, so has no
>> dependency on msi_list.
> 
> OK, I thought we could use msi_desc to describe DMAR/HPET irq, so they could
> share the mask/unmask, write_msg/read_etc.. But maybe it's not a right direction. :)
It doesn't help us much because there's no common way to implement
mask/unmask operations. Read/write can't be shared too.

HT_IRQ may be switched to use msi_desc, but it's not so popular:)

> 
>> 	Actually there are two levels of generic MSI sharing. The first
>> level is to share common irq_chip/irqdomain code, such as HPET, DMAR and
>> HT_IRQ. The second level is to share msi_desc, such as some device
>> on ARM side as you have mentioned.
>> 	With this patch set applied, we achieve level one sharing. For
>> level two sharing, we need to move msi_list into struct device, refactor
>> struct msi_desc, and provide a generic pci_enable_msix_range alike
>> user interfaces. I'm still working on this part, so we could cooperate
>> with each other.
> 
> That's good :)
> 
> 
>> Thanks!
>> Gerry
>>
>>>
>>> Thanks!
>>> Yijing.
>>>
>>>>
>>>> Patch 3 introduces msi_domain_{alloc|free}_irqs() which generalize
>>>> pci_msi_domain_alloc_irqs() to support generic MSI.
>>>>
>>>> Patch 4 introduces default data structures and callback implementations
>>>> to support msi_domain_alloc_irqs(), so reduce burden on generic MSI
>>>> users.
>>>>
>>>> Patch 5 converts PCI MSI to use generic MSI interfaces, and also
>>>> implement default callbacks for PCI MSI.
>>>>
>>>> Patch 6 introduces a mechanism to replace arch_setup_msi_irq()/
>>>> arch_setup_msi_irqs()/arch_teardown_msi_irq()/arch_teardown_msi_irqs().
>>>>
>>>> With this patch set applied, the generic MSI and PCI MSI interfaces
>>>> are much easier to use. For extreme case, you only need to define
>>>> a "struct msi_domain_info" and don't need to implement any callbacks,
>>>> just using the default callbacks is OK:)
>>>>
>>>> This patch set is also a preparation for:
>>>> 1) Kill all weak functions in drivers/pci/msi.c
>>>> 2) Implement support for non-PCI-compliant MSI device
>>>>
>>>> It has been tested on x86 platforms, comments are welcomed!
>>>>
>>>> Jiang Liu (6):
>>>>   PCI, MSI: Fix errors caused by commit e5f1a59c4e12
>>>>   PCI, MSI: Introduce helpers to hide struct msi_desc implemenation
>>>>     details
>>>>   genirq: Introduce msi_domain_{alloc|free}_irqs()
>>>>   genirq: Provide default callbacks for msi_domain_ops
>>>>   PCI, MSI: Refine irqdomain interfaces to simplify its usage
>>>>   PCI, MSI: Provide mechanism to alloc/free MSI/MSIX interrupt from
>>>>     irqdomain
>>>>
>>>>  drivers/pci/msi.c   |  170 ++++++++++++++++++++++++++++++++++++++++-----------
>>>>  include/linux/msi.h |   90 ++++++++++++++++++++++++---
>>>>  kernel/irq/msi.c    |  150 ++++++++++++++++++++++++++++++++++++++++++++-
>>>>  3 files changed, 366 insertions(+), 44 deletions(-)
>>>>
>>>
>>>
>>
>> .
>>
> 
>