svn commit: r204648 - head/sys/dev/ciss

[Attila Nagy]

Alexander Motin wrote:
> Scott Long wrote:
>   
>> On Mar 3, 2010, at 1:43 PM, Alexander Motin wrote:
>>     
>>> Attila Nagy wrote:
>>>       
>>>> Alexander Motin wrote:
>>>>         
>>>>> Author: mav
>>>>> Date: Wed Mar  3 17:58:41 2010
>>>>> New Revision: 204648
>>>>> URL: http://svn.freebsd.org/changeset/base/204648
>>>>>
>>>>> Log:
>>>>>  Several changes to fix livelock under high load, introduced by r203489:
>>>>>  - change the way in which command queue overflow is handled;
>>>>>  - do not expose to CAM two command slots, used for driver's internal purposes;
>>>>>  - allow driver to use up to 1024 command slots, instead of 256 before.
>>>>>
>>>>>           
>>>> Thank you for hunting this bug down. Do you plan to increase the maximum
>>>> logical device limit as well?
>>>>         
>>> Not sure. I just have no idea why this limitation was enforced. Even if
>>> I personally don't see problems from increasing it, it would be nice if
>>> somebody more aware of this driver tell whether it is correct for all
>>> supported hardware. I just don't have any. Thank you for provided
>>> access. If nobody responds, we may try our luck in CURRENT.
>>>       
>> If you don't know, then please ask the people who maintain the code.
>>     
>
> For last two years CVS shows mostly you. And you are already here. So do
> you have some opinion?
>   
I think 15 is too low in 2010 and it was even that in 2008 (first asked
this question), when SAS controllers and ZFS became mainstream.

Sad that HP doesn't really gives an option to work around these ancient
RAID stuff with simple SAS HBAs in their blades.

BTW, I'm also curious why there is such a limit there. Maybe not so
relevant, but Linux seems to have a 1024 limit, hence works out of the
box with external SAS boxes like MSA70 (25 disks per box) or MDS600 (70
disks per box).

I hope you can make a compromise. :)

Thanks,