cvs commit: src/sys/cam/scsi scsi_da.c src/sys/dev/usb umass.c usbdevs

[Scott Long]

Nate Lawson wrote:

> Matthew Jacob wrote:
> 
>>> One case this won't cover, though, is when the user changes the WCE bit
>>> after we probe.  That's obviously not a very common case, but the 
>>> only way
>>> to mostly cover it would be to do a mode sense just prior to every sync
>>> cache command.  (We could set a bit, though, if that mode page isn't
>>> supported so that we don't constantly ask for a mode page that isn't
>>> supported.)
>>>
>>> We're also assuming that the device firmware is telling the truth about
>>> whether the write cache is enabled or disabled.  (Hopefully so, but you
>>> never know.)
>>
>>
>> Changing the WCE bit is actually pretty common. I do it all the time 
>> myself. Furthermore, a common scenario is the Windows enables WCE and 
>> then uses FUA for synchronization.
>>
>> I've missed something here- other than broken devices that die 
>> spectacularly when the get one, why don't you just infer from a failed 
>> SYNCHRONIZE CACHE (e.g., "Illegal Request") that the device doesn't 
>> support it and stop doing it?
>>
>> For example, the EMC/Clariion AX100 reports Illegal Request on this 
>> command. You can't use quirks to identify this device because all 
>> Clariions have essentially the same Vendor/Product ID ("DGC", "RAID5").
> 
> 
> There are 3 kinds of devices:
> 
> * SYNC CACHE works
> * SYNC CACHE reports an error, but continues working
> * SYNC CACHE just hangs, no error
> * SYNC CACHE reports and error correctly, but then all subsequent 
> commands time out
> 
> We're talking about the 4th case and some versions with the 3rd case, if 
> it was possible to detect them without hanging (i.e. mode sense works 
> and WCE properly reported).  The problem is that we have a significant 
> SYNC CACHE quirk proliferation problem.
> 
> A few years ago, we had a 6-byte command quirk proliferation problem, 
> until I modified USB and Firewire SIMs to report "not 6-byte capable". I 
> then was able to remove dozens of quirks and we seem to have solved that 
> problem.  I'm recommending we do the same thing with SYNC CACHE now.
> 

You're suggesting that the umass and firewire SIMs universally instruct
CAM to not send a SYNC CACHE for all targets?  Easy, but I think it's
too big of a hammer.  I'd like to see the da driver get modified to
check the WCE state as has been suggested.

Scott