Re: Oops in UHCI when encountering "host controller process error"

[Jeremy Fitzhardinge <jeremy@goop.org>]

Alan Stern wrote:
> On Tue, 9 Dec 2008, Jeremy Fitzhardinge wrote:
>
>   
>> Yep, there's something very odd going on in there.  By contrast, does 
>> this look OK?  The allocations are fine, but I'm wondering if the 
>> skeleton QH stuff is all correct; it seems to work OK in this state.
>>
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4000 handle=7e212000
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4080 handle=7e212080
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4100 handle=7e212100
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4180 handle=7e212180
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4200 handle=7e212200
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4280 handle=7e212280
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4300 handle=7e212300
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4380 handle=7e212380
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4400 handle=7e212400
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4480 handle=7e212480
>> uhci_alloc_qh: uhci=ffff88002e5de3d0 qh=ffff88002e5f4500 handle=7e212500
>> Root-hub state: reset   FSBR: 0
>> HC status
>>   usbcmd    =     0000   Maxp32 
>>   usbstat   =     0020   HCHalted 
>>   usbint    =     0000
>>   usbfrnum  =   (0)000
>>   flbaseadd = 7ffd5000
>>   sof       =       40
>>   stat1     =     0080  
>>   stat2     =     0080  
>> Most recent frame: 0 (0)   Last ISO frame: 0 (0)
>> Periodic load table
>>         0       0       0       0       0       0       0       0
>>         0       0       0       0       0       0       0       0
>>         0       0       0       0       0       0       0       0
>>         0       0       0       0       0       0       0       0
>> Total: 0, #INT: 0, #ISO: 0
>> Frame List
>> Skeleton QHs
>> - skel_unlink_qh
>>     [ffff88002e5f4000] Skel QH link (00000001) element (00000001)
>>       queue is empty
>> - skel_iso_qh
>>     [ffff88002e5f4080] Skel QH link (00000001) element (00000001)
>>       queue is empty
>> - skel_int128_qh
>>     [ffff88002e5f4100] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int64_qh
>>     [ffff88002e5f4180] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int32_qh
>>     [ffff88002e5f4200] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int16_qh
>>     [ffff88002e5f4280] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int8_qh
>>     [ffff88002e5f4300] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int4_qh
>>     [ffff88002e5f4380] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_int2_qh
>>     [ffff88002e5f4400] Skel QH link (7e212482) element (00000001)
>>       queue is empty
>> - skel_async_qh
>>     [ffff88002e5f4480] Skel QH link (00000001) element (7e215000)
>>       queue is empty
>> [ffff88002e5f3000] link (00000001) e0 Length=0 MaxLen=7ff DT0 EndPt=0 Dev=7f, PID=69(IN) (buf=00000000)
>> - skel_term_qh
>>     [ffff88002e5f4500] Skel QH link (7e212502) element (7e215000)
>>       queue is empty
>>     
>
> Yes, that looks normal for a controller with no activity.
>   

OK.  If the controller is idle and the kernel wants to submit it some 
new work, where does that happen?  Can I stick a uhci_sprint_schedule() 
there to see the state of the structures before it starts to work on them?


Thanks,
    J