Re: blk-mq flush fix

[Jens Axboe <axboe@kernel.dk>]

On 10/28/2013 01:30 PM, Christoph Hellwig wrote:
> On Mon, Oct 28, 2013 at 10:59:32AM -0600, Jens Axboe wrote:
>>> The problem with it is that it will pass a tag number to the low level
>>> driver which it doesn't expect.  In case the tags are used 1:1 as
>>> hardware tags that would lead to nasy bugs.
>>>
>>> At vefy least we'd need to mess with ->tag for this special request.
>>
>> Yes indeed. Actually the more I think about it, the better I like just
>> using the normal tagging infrastructure and punting to a reserved tag
>> for the flush. Just needs adding of the check whether it actually needs
>> it or not.
> 
> That issue happens when using reserved tags as-is: e.g. the device
> supports 32 hardware tag, first reserved one is 33, something the
> hardware can't handle.
> 
> So either way we'll have to copy over the tag from the original request.
> 
> I don't think I can come up with something sane for that ASAP, so I'd
> suggest you actually do take my patch for now, and we'll sort something
> out less in a hurry.

Agree, I will queue it up.

-- 
Jens Axboe