All of lore.kernel.org
 help / color / mirror / Atom feed
* RBD Kernel: write_same_max_bytes and SCSI XCOPY
@ 2015-12-09  7:08 Timofey Titovets
  2015-12-09 21:37 ` Mike Christie
  0 siblings, 1 reply; 3+ messages in thread
From: Timofey Titovets @ 2015-12-09  7:08 UTC (permalink / raw)
  To: Ceph Development

Hi list,
As i found write_same_max_bytes are equal to zero
if i understand correctly, it's means what RBD not support offloading
copy operations

Is this possible to add this feature, as example for LIO with SCSI XCOPY?
-- 
Have a nice day,
Timofey.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: RBD Kernel: write_same_max_bytes and SCSI XCOPY
  2015-12-09  7:08 RBD Kernel: write_same_max_bytes and SCSI XCOPY Timofey Titovets
@ 2015-12-09 21:37 ` Mike Christie
  2015-12-09 21:44   ` Timofey Titovets
  0 siblings, 1 reply; 3+ messages in thread
From: Mike Christie @ 2015-12-09 21:37 UTC (permalink / raw)
  To: Timofey Titovets, Ceph Development

On 12/09/2015 01:08 AM, Timofey Titovets wrote:
> Hi list,
> As i found write_same_max_bytes are equal to zero
> if i understand correctly, it's means what RBD not support offloading
> copy operations
> 

Not yet.

I did patches to support commands like write same and compare and write
in this patchset where I made a lio rbd backend driver:

http://www.spinics.net/lists/target-devel/msg10330.html

but they were not accepted upstream.

I am currently reworking the patches to be acceptable. Some distros did
pick them up though so you might see them in there.


> Is this possible to add this feature, as example for LIO with SCSI XCOPY?


The linux kernel does yet not have xcopy support like how it supports
write same or discard. There are patches, but it has not been merged
upstream.

However, LIO does support this command by what I guess you could call
emulating it where if the underlying device does not support the command
natively then LIO will perform the operations needed. It does this for
other commands like write same too. So, even though rbd does not
currently support these commands, you can use LIO and rbd and it will
still execute them although it would not be offloaded like how you would
want. For example, for write same LIO will just do multiple writes to
the rbd device.


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: RBD Kernel: write_same_max_bytes and SCSI XCOPY
  2015-12-09 21:37 ` Mike Christie
@ 2015-12-09 21:44   ` Timofey Titovets
  0 siblings, 0 replies; 3+ messages in thread
From: Timofey Titovets @ 2015-12-09 21:44 UTC (permalink / raw)
  To: Mike Christie; +Cc: Ceph Development

Thank you Mike, for your work & answers

2015-12-10 0:37 GMT+03:00 Mike Christie <mchristi@redhat.com>:
> On 12/09/2015 01:08 AM, Timofey Titovets wrote:
>> Hi list,
>> As i found write_same_max_bytes are equal to zero
>> if i understand correctly, it's means what RBD not support offloading
>> copy operations
>>
>
> Not yet.
>
> I did patches to support commands like write same and compare and write
> in this patchset where I made a lio rbd backend driver:
>
> http://www.spinics.net/lists/target-devel/msg10330.html
>
> but they were not accepted upstream.
>
> I am currently reworking the patches to be acceptable. Some distros did
> pick them up though so you might see them in there.
>
>
>> Is this possible to add this feature, as example for LIO with SCSI XCOPY?
>
>
> The linux kernel does yet not have xcopy support like how it supports
> write same or discard. There are patches, but it has not been merged
> upstream.
>
> However, LIO does support this command by what I guess you could call
> emulating it where if the underlying device does not support the command
> natively then LIO will perform the operations needed. It does this for
> other commands like write same too. So, even though rbd does not
> currently support these commands, you can use LIO and rbd and it will
> still execute them although it would not be offloaded like how you would
> want. For example, for write same LIO will just do multiple writes to
> the rbd device.
>



-- 
Have a nice day,
Timofey.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2015-12-09 21:44 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-12-09  7:08 RBD Kernel: write_same_max_bytes and SCSI XCOPY Timofey Titovets
2015-12-09 21:37 ` Mike Christie
2015-12-09 21:44   ` Timofey Titovets

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.