Re: [Qemu-block] [PATCH 4/4] block: Cater to iscsi with non-power-of-2 discard

[Eric Blake]

On 10/25/2016 09:20 AM, Peter Lieven wrote:
>> The firmware is probably technically buggy for advertising too large of
>> a minimum granularity, if it can piece together smaller requests into a
>> larger discard.  If discards need to happen at a smaller granularity,
>> the firmware (or kernel quirk system) should fix the advertisement to
>> the actual granularity that it will honor.  I don't see a reason to
>> change qemu's current behavior.
>>
> 
> The issue is that the optimal unmap granularity is only a hint.
> There is no evidence what happens with unaligned requests or requests
> that are smaller. They could still lead to a discard operation in the
> storage device. It just says if you can send me discards of that size
> thats optimal for me. Thats not said that smaller or unaligned requests
> have no effect.

So your argument is that we should always pass down every unaligned
less-than-optimum discard request all the way to the hardware, rather
than dropping it higher in the stack, even though discard requests are
already advisory, in order to leave the hardware as the ultimate
decision on whether to ignore the unaligned request?

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature