On 6/27/24 00:15, Bjorn Helgaas wrote:
>>> Yes, I am talking about the PCI "Function Level Reset"
>>>
>>>> FLR and disk/controller suspend execution timing are unrelated. FLR can be
>>>> triggered at any time through sysfs. So please give details here. Why is FLR
>>>> done when the system is being suspended ?
>>>
>>> Yes, it is because FLR can be triggered at any time that we are testing the
>>> reliability of executing FLR commands after disk/controller suspended.
>>
>> "can be triggered" ? FLR is not a random asynchronous event. It is an action
>> that is *issued* by a user with sys admin rights. And such users can do a lot
>> of things that can break a machine...
>>
>> I fail to see the point of doing a function reset while the device is
>> suspended. But granted, I guess the device should comeback up in such case,
>> though I would like to hear what the PCI guys have to say about this.
>>
>> Bjorn,
>>
>> Is reseting a suspended PCI device something that should be/is supported ?
>
> I doubt it. The PCI core should be preserving all the generic PCI
> state across suspend/resume. The driver should only need to
> save/restore device-specific things the PCI core doesn't know about.
>
> A reset will clear out most state, and the driver doesn't know the
> reset happened, so it will expect most device state to have been
> preserved.
That is what I suspected. However, checking the code, reset_store() in
pci-sysfs.c does:
pm_runtime_get_sync(dev);
result = pci_reset_function(pdev);
pm_runtime_put(dev);
and pm_runtime_get_sync() calls __pm_runtime_resume() which will resume a
suspended device.
So while I still think it is not a good idea to reset a suspended device, things
should still work as execpected and not cause any problem with the device state,
right ?
Yihang,
I think that the issue at hand here is that once the reset finishes, the
controller goes back to suspended state, and I suspect that is because of the
"auto" setting for its power/control. That triggers because the FLR is done
after the controller resumed but *before* the revalidation of the drives
connected to it completes. So FLR makes the revalidation fail (scsi
scan/revalidation is asynchronous...).
This seems to me to be the expected behavior for what you are doing and I fail
to see how that ever worked correctly, even before 0c76106cb975 and 626b13f015e0.
Could you try this: add a call to msleep(30000) at the end of _resume_v3_hw(). I.e.:
diff --git a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
index feda9b54b443..54224568d749 100644
--- a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
@@ -5104,6 +5104,8 @@ static int _resume_v3_hw(struct device *device)
dev_warn(dev, "end of resuming controller\n");
+ msleep(30000);
+
return 0;
}
To see if it makes any difference to actually wait for the connected disks to
resume correctly before doing the FLR.
--
Damien Le Moal
Western Digital Research
