Re: [PCI/MSI] a4fc4cf388: dmesg.genirq:Flags_mismatch_irq##(mei_me)vs.#(xhci_hcd)

From: Marc Zyngier
Date: Wed Sep 01 2021 - 09:09:18 EST


On 2021-08-31 22:36, Barry Song wrote:
On Tue, Aug 31, 2021 at 8:08 PM Marc Zyngier <maz@xxxxxxxxxx> wrote:

On 2021-08-31 02:21, Barry Song wrote:
> On Mon, Aug 30, 2021 at 2:38 AM kernel test robot
> <oliver.sang@xxxxxxxxx> wrote:
>>
>>
>>
>> Greeting,
>>
>> FYI, we noticed the following commit (built with gcc-9):
>>
>> commit: a4fc4cf388319ea957ffbdab5073bdd267de9082 ("[PATCH v3 3/3]
>> PCI/MSI: remove msi_attrib.default_irq in msi_desc")
>> url:
>> https://github.com/0day-ci/linux/commits/Barry-Song/PCI-MSI-Clarify-the-IRQ-sysfs-ABI-for-PCI-devices/20210825-183018
>> base: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git
>> 6e764bcd1cf72a2846c0e53d3975a09b242c04c9
>>
>> in testcase: kernel-selftests
>> version: kernel-selftests-x86_64-ebaa603b-1_20210825
>> with following parameters:
>>
>> group: pidfd
>> ucode: 0xe2
>>
>> test-description: The kernel contains a set of "self tests" under the
>> tools/testing/selftests/ directory. These are intended to be small
>> unit tests to exercise individual code paths in the kernel.
>> test-url: https://www.kernel.org/doc/Documentation/kselftest.txt
>>
>>
>> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz
>> with 32G memory
>>
>> caused below changes (please refer to attached dmesg/kmsg for entire
>> log/backtrace):
>>
>>
>>
>> If you fix the issue, kindly add following tag
>> Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
>>
>>
>>
>> [ 179.602028][ T34] genirq: Flags mismatch irq 16. 00002000
>> (mei_me) vs. 00000000 (xhci_hcd)
>> [ 179.614073][ T34] CPU: 2 PID: 34 Comm: kworker/u8:2 Not tainted
>> 5.14.0-rc7-00014-ga4fc4cf38831 #1
>> [ 179.623225][ T34] Hardware name: Dell Inc. OptiPlex 7040/0Y7WYT,
>> BIOS 1.8.1 12/05/2017
>> [ 179.631432][ T34] Workqueue: events_unbound async_run_entry_fn
>> [ 179.637543][ T34] Call Trace:
>> [ 179.640789][ T34] dump_stack_lvl+0x45/0x59
>> [ 179.645253][ T34] __setup_irq.cold+0x50/0xd4
>> [ 179.649893][ T34] ? mei_me_pg_exit_sync+0x480/0x480 [mei_me]
>> [ 179.655923][ T34] request_threaded_irq+0x10c/0x180
>> [ 179.661073][ T34] ? mei_me_irq_quick_handler+0x240/0x240
>> [mei_me]
>> [ 179.667528][ T34] mei_me_probe+0x131/0x300 [mei_me]
>> [ 179.672767][ T34] local_pci_probe+0x42/0x80
>> [ 179.677313][ T34] pci_device_probe+0x107/0x1c0
>> [ 179.682118][ T34] really_probe+0xb6/0x380
>> [ 179.687094][ T34] __driver_probe_device+0xfe/0x180
>> [ 179.692242][ T34] driver_probe_device+0x1e/0xc0
>> [ 179.697133][ T34] __driver_attach_async_helper+0x2b/0x80
>> [ 179.702802][ T34] async_run_entry_fn+0x30/0x140
>> [ 179.707693][ T34] process_one_work+0x274/0x5c0
>> [ 179.712503][ T34] worker_thread+0x50/0x3c0
>> [ 179.716959][ T34] ? process_one_work+0x5c0/0x5c0
>> [ 179.721936][ T34] kthread+0x14f/0x180
>> [ 179.725958][ T34] ? set_kthread_struct+0x40/0x40
>> [ 179.730935][ T34] ret_from_fork+0x22/0x30
>> [ 179.735699][ T34] mei_me 0000:00:16.0: request_threaded_irq
>> failure. irq = 16
>> [ 179.743125][ T34] mei_me 0000:00:16.0: initialization failed.
>> [ 179.749399][ T34] mei_me: probe of 0000:00:16.0 failed with error
>> -16
>>
>>
>
> it seems there is a direct reference to pdev->irq.
> Hi Oliver, would you try if the below patch can fix the problem:
>
> diff --git a/drivers/misc/mei/pci-me.c b/drivers/misc/mei/pci-me.c
> index c3393b383e59..a45a2d4257a6 100644
> --- a/drivers/misc/mei/pci-me.c
> +++ b/drivers/misc/mei/pci-me.c
> @@ -216,7 +216,7 @@ static int mei_me_probe(struct pci_dev *pdev,
> const struct pci_device_id *ent)
>
> pci_enable_msi(pdev);
>
> - hw->irq = pdev->irq;
> + hw->irq = pci_irq_vector(pdev, 0);
>
> /* request and enable interrupt */
> irqflags = pci_dev_msi_enabled(pdev) ? IRQF_ONESHOT :
> IRQF_SHARED;
>

Ah! one victim, 3000 to go! :D


yep.

That's exactly the kind of stuff I was mentioning when we
discussed this patch. Exposing the MSI vector as the INTx
IRQ has led to all sorts of broken drivers.

I guess drivers should depend on int pci_irq_vector(struct pci_dev
*dev, unsigned int nr)
rather than hardcodely use pdev->irq.

pci_irq_vector() supports all cases(intx, msi, msi-x)

Yes, that'd be a sensible approach. Feels like a job for
a coccinelle script!

M.
--
Jazz is not dead. It just smells funny...