Re: [PATCH v2 3/3] misc: pci_endpoint_test: Fix irq_type to convey the correct type

[Niklas Cassel <cassel@kernel.org>]

On Tue, Jan 28, 2025 at 08:02:31PM +0530, Manivannan Sadhasivam wrote:
> On Wed, Jan 22, 2025 at 11:24:46AM +0900, Kunihiko Hayashi wrote:
> > There are two variables that indicate the interrupt type to be used
> > in the next test execution, "irq_type" as global and test->irq_type.
> > 
> > The global is referenced from pci_endpoint_test_get_irq() to preserve
> > the current type for ioctl(PCITEST_GET_IRQTYPE).
> > 
> > The type set in this function isn't reflected in the global "irq_type",
> > so ioctl(PCITEST_GET_IRQTYPE) returns the previous type.
> > As a result, the wrong type will be displayed in "pcitest" as follows:
> > 
> >     # pcitest -i 0
> >     SET IRQ TYPE TO LEGACY:         OKAY
> >     # pcitest -I
> >     GET IRQ TYPE:           MSI
> > 
> > Fix this issue by propagating the current type to the global "irq_type".
> > 
> 
> This is becoming a nuisance. I think we should get rid of the global 'irq_type'
> and just stick to the one that is configurable using IOCTL command. Even if the
> user has configured the global 'irq_type' it is bound to change with IOCTL
> command.

+1


But I also don't like how since we migrated to selftests:
READ_TEST / WRITE_TEST / COPY_TEST unconditionally call
ioctl(PCITEST_SET_IRQTYPE, MSI) before doing their thing.

Will this cause the test case to fail for platforms that only support MSI-X?
(See e.g. dwc/pci-layerscape-ep.c where this could be the case.)


Sure, before, in pcitest.sh, we would do:


pcitest -i 2
        pcitest -x $msix


pcitest -i 1

pcitest -r -s 1
pcitest -r -s 1024
pcitest -r -s 1025
pcitest -r -s 1024000
pcitest -r -s 1024001


Which would probably print an error if:
pcitest -i 1
failed.

but the READ_TEST / WRITE_TEST / COPY_TEST tests themselves
would not fail.


Perhaps we should rethink this, and introduce a new
PCITEST_SET_IRQTYPE, AUTO

I would be fine if
READ_TEST / WRITE_TEST / COPY_TEST
called PCITEST_SET_IRQTYPE, AUTO
before doing their thing.



How I suggest PCITEST_SET_IRQTYPE, AUTO
would work:

Since we now have capabilties merged:
https://lore.kernel.org/linux-pci/20241203063851.695733-4-cassel@kernel.org/

Add epc_features->msi_capable and epc->features->msix_capable
as two new bits in the PCI_ENDPOINT_TEST_CAPS register.

If PCITEST_SET_IRQTYPE, AUTO:
if EP CAP has msi_capable set: set IRQ type MSI
else if EP CAP has msix_capable set: set IRQ type MSI-X
else: set legacy/INTx


Kind regards,
Niklas