On 09/03/2020 13:57, Hannes Reinecke wrote:
On 3/9/20 1:48 PM, John Garry wrote:
On 23/08/2019 14:26, John Garry wrote:
quite some drivers use internal commands for various purposes, most
commonly sending TMFs or querying the HBA status.
While these commands use the same submission mechanism than normal
I/O commands, they will not be counted as outstanding commands,
requiring those drivers to implement their own mechanism to figure
out outstanding commands.
This patchset enables the use of reserved tags for the SCSI midlayer,
enabling LLDDs to rely on the block layer for tracking outstanding
commands.
More importantly, it allows LLDD to request a valid tag from the block
layer without having to implement some tracking mechanism within the
driver. This removes quite some hacks which were required for some
drivers (eg. fnic or snic).
As usual, comments and reviews are welcome.
Hi Hannes,
JFYI, I have rebased this series to 5.6-rc4 here:
https://github.com/hisilicon/kernel-dev/tree/private-topic-sas-5.6-resv-commands-v1
I am interested in enabling this for libsas and associated HBAs, so
there are some patches on top for that.
No review comments have been addressed, apart from removing "block:
disable elevator for reserved tags"
Please let me know your plan for this series.
The plan is to wait for the shared tagset patchset to go in first :-)
I think that one should go last. Or, to be more specific, switching the
drivers to expose multiple queues - which is part of that series -
should go last.
Without it we'll be running on one tag bitmap per queue, and we risk
duplicate tags when sending down reserved commands :-(
Until we expose multiple queues, we still have a single HW queue from
blk-mq perspective, so will only have a single tagset per HBA, so no
chance of duplicate tags.
This series should be able to go in without dependencies.
Thanks,
John
