Re: [RFC PATCH v1 03/10] vfio-ccw: Use subchannel lpm in the orb

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




On 11/19/19 10:38 AM, Cornelia Huck wrote:
> On Tue, 19 Nov 2019 10:16:30 -0500
> Eric Farman <farman@xxxxxxxxxxxxx> wrote:
> 
>> On 11/19/19 8:00 AM, Cornelia Huck wrote:
>>> On Fri, 15 Nov 2019 03:56:13 +0100
>>> Eric Farman <farman@xxxxxxxxxxxxx> wrote:
>>>   
>>>> From: Farhan Ali <alifm@xxxxxxxxxxxxx>
>>>>
>>>> The subchannel logical path mask (lpm) would have the most
>>>> up to date information of channel paths that are logically
>>>> available for the subchannel.
>>>>
>>>> Signed-off-by: Farhan Ali <alifm@xxxxxxxxxxxxx>
>>>> Signed-off-by: Eric Farman <farman@xxxxxxxxxxxxx>
>>>> ---
>>>>
>>>> Notes:
>>>>     v0->v1: [EF]
>>>>      - None; however I am greatly confused by this one.  Thoughts?  
>>>
>>> I think it's actually wrong.
>>>   
>>>>
>>>>  drivers/s390/cio/vfio_ccw_cp.c | 4 +---
>>>>  1 file changed, 1 insertion(+), 3 deletions(-)
>>>>
>>>> diff --git a/drivers/s390/cio/vfio_ccw_cp.c b/drivers/s390/cio/vfio_ccw_cp.c
>>>> index 3645d1720c4b..d4a86fb9d162 100644
>>>> --- a/drivers/s390/cio/vfio_ccw_cp.c
>>>> +++ b/drivers/s390/cio/vfio_ccw_cp.c
>>>> @@ -779,9 +779,7 @@ union orb *cp_get_orb(struct channel_program *cp, u32 intparm, u8 lpm)
>>>>  	orb->cmd.intparm = intparm;
>>>>  	orb->cmd.fmt = 1;
>>>>  	orb->cmd.key = PAGE_DEFAULT_KEY >> 4;
>>>> -
>>>> -	if (orb->cmd.lpm == 0)
>>>> -		orb->cmd.lpm = lpm;  
>>>
>>> In the end, the old code will use the lpm from the subchannel
>>> structure, if userspace did not supply anything to be used.
>>>   
>>>> +	orb->cmd.lpm = lpm;  
>>>
>>> The new code will always discard any lpm userspace has supplied and
>>> replace it with the lpm from the subchannel structure. This feels
>>> wrong; what if the I/O is supposed to be restricted to a subset of the
>>> paths?  
>>
>> I had the same opinion, but didn't want to flat-out discard it from his
>> series without a second look.  :)
> 
> :)
> 
>>
>>>
>>> If we want to include the current value of the subchannel lpm in the
>>> requests, we probably want to AND the masks instead.  
>>
>> Then we'd be on the hook to return some sort of error if the result is
>> zero.  Is it better to just send it to hw as-is, and let the response
>> come back naturally?  (Which is what we do today.)
> 
> But if a chpid is logically varied off, it is removed from the lpm,
> right? Therefore, the caller really should get a 'no path' indication
> back, shouldn't it?

Agreed, just don't know whether we should be kicking it out here, versus
just doing what we do today and sending the guest LPM to the hardware.

The routine being modified here is cp_get_orb(), which doesn't seem like
the right place for such a check.  It does currently return NULL when
the cp is not initialized, claiming an error in the caller.  Not sure
what happens if we start doing that when a path goes away unexpectedly.

I'll poke around and see what happens if we do that, and if I can drive
some path-directed I/O easily.

> 
>>
>>>   
>>>>  
>>>>  	chain = list_first_entry(&cp->ccwchain_list, struct ccwchain, next);
>>>>  	cpa = chain->ch_ccw;  
>>>   
>>
> 



[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]

  Powered by Linux