[dpdk-dev] [RFC PATCH v3 2/3] eal: add mask and unmask interrupt APIs

[Nithin Kumar Dabilpuram]

On 7/17/2019 4:46 PM, Jerin Jacob Kollanukkaran wrote:
>> -----Original Message-----
>> From: Hyong Youb Kim (hyonkim) <hyonkim at cisco.com>
>> Sent: Wednesday, July 17, 2019 4:36 PM
>> To: Jerin Jacob Kollanukkaran <jerinj at marvell.com>; Nithin Kumar
>> Dabilpuram <ndabilpuram at marvell.com>; David Marchand
>> <david.marchand at redhat.com>; Thomas Monjalon
>> <thomas at monjalon.net>; Ferruh Yigit <ferruh.yigit at intel.com>; Bruce
>> Richardson <bruce.richardson at intel.com>
>> Cc: John Daley (johndale) <johndale at cisco.com>; Shahed Shaikh
>> <shshaikh at marvell.com>; dev at dpdk.org
>> Subject: RE: [RFC PATCH v3 2/3] eal: add mask and unmask interrupt APIs
>>
>>> -----Original Message-----
>>> From: Jerin Jacob Kollanukkaran <jerinj at marvell.com>
>>> Sent: Wednesday, July 17, 2019 7:44 PM
>>> To: Hyong Youb Kim (hyonkim) <hyonkim at cisco.com>; Nithin Kumar
>>> Dabilpuram <ndabilpuram at marvell.com>; David Marchand
>>> <david.marchand at redhat.com>; Thomas Monjalon
>> <thomas at monjalon.net>;
>>> Ferruh Yigit <ferruh.yigit at intel.com>; Bruce Richardson
>>> <bruce.richardson at intel.com>
>>> Cc: John Daley (johndale) <johndale at cisco.com>; Shahed Shaikh
>>> <shshaikh at marvell.com>; dev at dpdk.org
>>> Subject: RE: [RFC PATCH v3 2/3] eal: add mask and unmask interrupt
>>> APIs
>>>
>>>>> I think, it vary from the perspective of IRQ Chip(or controller)
>>>>> vs NIC
>>>>> register(Source) PoV.
>>>>> Since the API starts from rte_intr_* it is more of controller so
>>>>> _ack_ make sense Other reason for ack:
>>>>> 1) It will enforce that it needs to be called form ISR
>>>>> 2) It would be have been really correct to unmask if
>>>>> VFIO+MSIx+Linux supports it
>>>>> 3) if it is ack, no need to add unmask counterpart, the _mask_ API
>>>>>
>>>> Just curious, what you mean by irq controller? Ack/mask/unmask PIOs
>>>> all
>>> go
>>>
>>> Programmable Interrupt Controller. Like Intel 8259A, GIC from ARM etc
>>> The drivers in linux/drivers/irqchip/
>>>
>>>> to the NIC. It is the NIC that asserts/de-asserts irq..
>>>>
>>>>>> Besides the name, are we agreeing that we want these?
>>>>>> - Unmask if INTx
>>>>> Yes
>>>>>
>>>>>> - Nothing if MSI/MSI-X
>>>>> Yes for MSI over VFIO
>>>>> No for MSI over UIO/igb_uio
>>>>>
>>>> I guess I was not clear. For MSI/MSI-X, we do not want to do
>>>> mask/unmask regardless of vfio-pci/igb_uio.  Below is my comment
>>>> about linux/windows/freebsd from an earlier email. Do you disagree?
>>>> I am sure there are plenty of kernel NIC driver guys here. Please
>>>> correct me if I am mistaken...
>>>
>>> For some reason, igb_uio kernel driver mask the interrupt for MSIx.
>>> We need to ack or unmask if needs to work with MSIX + IGB_UIO.
>>>
>>> See
>>> pci_uio_alloc_resource()
>>>          if (dev->kdrv == RTE_KDRV_IGB_UIO)
>>>                  dev->intr_handle.type = RTE_INTR_HANDLE_UIO;
>>>          else {
>>>                  dev->intr_handle.type = RTE_INTR_HANDLE_UIO_INTX;
>>>
>>> igbuio_pci_irqcontrol() for masking in kernel.
>>>
>> igb_uio does not auto-mask MSI/MSI-X.
> I have not tested igbuio as we don't specific NIC + IGB_UIO platform.
>
> The observation based on following code. see code under HAVE_PCI_MSI_MASK_IRQ
>
> static int
> igbuio_pci_irqcontrol(struct uio_info *info, s32 irq_state)
> {
>          struct rte_uio_pci_dev *udev = info->priv;
>          struct pci_dev *pdev = udev->pdev;
>
> #ifdef HAVE_PCI_MSI_MASK_IRQ
>          struct irq_data *irq = irq_get_irq_data(udev->info.irq);
> #endif
>
>          pci_cfg_access_lock(pdev);
>
>          if (udev->mode == RTE_INTR_MODE_MSIX || udev->mode == RTE_INTR_MODE_MSI) {
> #ifdef HAVE_PCI_MSI_MASK_IRQ
>                  if (irq_state == 1)
>                          pci_msi_unmask_irq(irq);
>                  else
>                          pci_msi_mask_irq(irq);
> #else
>                  igbuio_mask_irq(pdev, udev->mode, irq_state);
> #endif
>          }
>
>          if (udev->mode == RTE_INTR_MODE_LEGACY)
>                  pci_intx(pdev, !!irq_state);
>
>          pci_cfg_access_unlock(pdev);
>
>          return 0;
> }
>
>> static irqreturn_t
>> igbuio_pci_irqhandler(int irq, void *dev_id) {
>>          struct rte_uio_pci_dev *udev = (struct rte_uio_pci_dev *)dev_id;
>>          struct uio_info *info = &udev->info;
>>
>>          /* Legacy mode need to mask in hardware */
>>          if (udev->mode == RTE_INTR_MODE_LEGACY &&
>>              !pci_check_and_mask_intx(udev->pdev))
>>                  return IRQ_NONE;
>>
>>          uio_event_notify(info);
>>
>>          /* Message signal mode, no share IRQ and automasked */
>>          return IRQ_HANDLED;
>> }
>>
>> Also tested just now with igb_uio. The driver does not need to call
>> rte_intr_enable(), and it keeps getting interrupts without any issues.
>   If you are sure, we can make MSIX+IGB_UIO as NOP in rte_intr_ack()

Ok. Another problem is that we might not be able to distinguish in case 
of IGB_UIO
at rte_intr_ack() level if underlying interrupt is a INTx or MSIx. See 
igbuio_pci_enable_interrupts() that
finds and stores that mode in uio->mode.

So we think leaving the behavior as earlier is needed and simpler as it 
meets the current expectation.

>
>> Am I missing something?
>>
>> -Hyong
>>
>>> So it is more of making inline with igb_uio kernel driver AND not
>>> break The existing drivers which was using rte_intr_enable in ISR with
>>> MSIX+IGB_UIO
>>>
>>> I do agree with that for edge trigged interrupt mask may not require
>>> from kernel.
>>> But I am not sure why it is added in igb_uio kernel driver. May  be it
>>> is just legacy.
>>> Anyway this wont change schematics, when igb_uio kenrel fixed then the
>>> counter Part can be changed in rte_intr_ack(). Ie. it is transparent
>>> to drivers.
>>>
>>>>> I don't  have very strong opinion unmask vs ack. I prefer to have
>>>>> ack due the reasons stated above.
>>>>> If you really have strong opinion on using unmask, we will stick
>>>>> with that to make forward progress.
>>>>> Let us know.
>>>>>
>>>> I have no strong opinion either.
>>> OK. Lets stick with rte_intr_ack().
>>>
>>>> Thanks..
>>>> -Hyong