mbox series

[v8,00/18] blk-mq/scsi: Provide hostwide shared tags for SCSI HBAs

Message ID 1597850436-116171-1-git-send-email-john.garry@huawei.com
Headers show
Series blk-mq/scsi: Provide hostwide shared tags for SCSI HBAs | expand

Message

John Garry Aug. 19, 2020, 3:20 p.m. UTC
Hi all,

Here is v8 of the patchset.

In this version of the series, we keep the shared sbitmap for driver tags,
and introduce changes to fix up the tag budgeting across request queues.
We also have a change to count requests per-hctx for when an elevator is
enabled, as an optimisation. I also dropped the debugfs changes - more on
that below.

Some performance figures:

Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,
but it is not always an appropriate scheduler to use.

Tag depth 		4000 (default)			260**

Baseline (v5.9-rc1):
none sched:		2094K IOPS			513K
mq-deadline sched:	2145K IOPS			1336K

Final, host_tagset=0 in LLDD *, ***:
none sched:		2120K IOPS			550K
mq-deadline sched:	2121K IOPS			1309K

Final ***:
none sched:		2132K IOPS			1185			
mq-deadline sched:	2145K IOPS			2097	

* this is relevant as this is the performance in supporting but not
  enabling the feature
** depth=260 is relevant as some point where we are regularly waiting for
   tags to be available. Figures were are a bit unstable here.
*** Included "[PATCH V4] scsi: core: only re-run queue in
    scsi_end_request() if device queue is busy"

A copy of the patches can be found here:
https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

The hpsa patch depends on:
https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

And the smartpqi patch is not to be accepted.

Comments (and testing) welcome, thanks!

Differences to v7:
- Add null_blk and scsi_debug support
- Drop debugfs tags patch - it's too difficult to be the same between
hostwide and non-hostwide, as discussed:
https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#mb3eb462d8be40273718505989abd12f8228c15fd
And from commit 6bf0eb550452 ("sbitmap: Consider cleared bits in
sbitmap_bitmap_show()"), I guess not many used this anyway...
- Add elevator per-hctx request count for optimisation
- Break up "blk-mq: rename blk_mq_update_tag_set_depth()" into 2x patches
- Pass flags for avoid per-hq queue tags init/free for hostwide tags
- Add Don's reviewed-tag and tested-by tags to appropiate patches
	- (@Don, please let me know if issue with how I did this)
- Add "scsi: core: Show nr_hw_queues in sysfs"
- Rework megaraid SAS patch to have module param (Kashyap)
- rebase

V7 is here for more info:
https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#t

Hannes Reinecke (5):
  blk-mq: Rename blk_mq_update_tag_set_depth()
  blk-mq: Free tags in blk_mq_init_tags() upon error
  scsi: Add host and host template flag 'host_tagset'
  hpsa: enable host_tagset and switch to MQ
  smartpqi: enable host tagset

John Garry (10):
  blk-mq: Pass flags for tag init/free
  blk-mq: Use pointers for blk_mq_tags bitmap tags
  blk-mq: Facilitate a shared sbitmap per tagset
  blk-mq: Relocate hctx_may_queue()
  blk-mq: Record nr_active_requests per queue for when using shared
    sbitmap
  blk-mq: Record active_queues_shared_sbitmap per tag_set for when using
    shared sbitmap
  null_blk: Support shared tag bitmap
  scsi: core: Show nr_hw_queues in sysfs
  scsi: hisi_sas: Switch v3 hw to MQ
  scsi: scsi_debug: Support host tagset

Kashyap Desai (2):
  blk-mq, elevator: Count requests per hctx to improve performance
  scsi: megaraid_sas: Added support for shared host tagset for
    cpuhotplug

Ming Lei (1):
  blk-mq: Rename BLK_MQ_F_TAG_SHARED as BLK_MQ_F_TAG_QUEUE_SHARED

 block/bfq-iosched.c                         |   9 +-
 block/blk-core.c                            |   2 +
 block/blk-mq-debugfs.c                      |  10 +-
 block/blk-mq-sched.c                        |  13 +-
 block/blk-mq-tag.c                          | 149 ++++++++++++++------
 block/blk-mq-tag.h                          |  56 +++-----
 block/blk-mq.c                              |  81 +++++++----
 block/blk-mq.h                              |  76 +++++++++-
 block/kyber-iosched.c                       |   4 +-
 block/mq-deadline.c                         |   6 +
 drivers/block/null_blk_main.c               |   6 +
 drivers/block/rnbd/rnbd-clt.c               |   2 +-
 drivers/scsi/hisi_sas/hisi_sas.h            |   3 +-
 drivers/scsi/hisi_sas/hisi_sas_main.c       |  36 ++---
 drivers/scsi/hisi_sas/hisi_sas_v3_hw.c      |  87 +++++-------
 drivers/scsi/hosts.c                        |   1 +
 drivers/scsi/hpsa.c                         |  44 +-----
 drivers/scsi/hpsa.h                         |   1 -
 drivers/scsi/megaraid/megaraid_sas_base.c   |  39 +++++
 drivers/scsi/megaraid/megaraid_sas_fusion.c |  29 ++--
 drivers/scsi/scsi_debug.c                   |  28 ++--
 drivers/scsi/scsi_lib.c                     |   2 +
 drivers/scsi/scsi_sysfs.c                   |  11 ++
 drivers/scsi/smartpqi/smartpqi_init.c       |  45 ++++--
 include/linux/blk-mq.h                      |  13 +-
 include/linux/blkdev.h                      |   3 +
 include/scsi/scsi_host.h                    |   9 +-
 27 files changed, 484 insertions(+), 281 deletions(-)

-- 
2.26.2

Comments

John Garry Aug. 27, 2020, 8:53 a.m. UTC | #1
Hi Jens,

I was wondering if you could kindly consider the block changes in this 
series, since I have now dropped the RFC flag?

I guess patch 5/18 (using pointers to bitmaps) would be of first 
concern. We did discuss this previously, and I think what we're doing 
now could be considered satisfactory.

Thanks,
john

> 

> Here is v8 of the patchset.

> 

> In this version of the series, we keep the shared sbitmap for driver tags,

> and introduce changes to fix up the tag budgeting across request queues.

> We also have a change to count requests per-hctx for when an elevator is

> enabled, as an optimisation. I also dropped the debugfs changes - more on

> that below.

> 

> Some performance figures:

> 

> Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,

> but it is not always an appropriate scheduler to use.

> 

> Tag depth 		4000 (default)			260**

> 

> Baseline (v5.9-rc1):

> none sched:		2094K IOPS			513K

> mq-deadline sched:	2145K IOPS			1336K

> 

> Final, host_tagset=0 in LLDD *, ***:

> none sched:		2120K IOPS			550K

> mq-deadline sched:	2121K IOPS			1309K

> 

> Final ***:

> none sched:		2132K IOPS			1185			

> mq-deadline sched:	2145K IOPS			2097	

> 

> * this is relevant as this is the performance in supporting but not

>    enabling the feature

> ** depth=260 is relevant as some point where we are regularly waiting for

>     tags to be available. Figures were are a bit unstable here.

> *** Included "[PATCH V4] scsi: core: only re-run queue in

>      scsi_end_request() if device queue is busy"

> 

> A copy of the patches can be found here:

> https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

> 

> The hpsa patch depends on:

> https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

> 

> And the smartpqi patch is not to be accepted.

> 

> Comments (and testing) welcome, thanks!

> 

> Differences to v7:

> - Add null_blk and scsi_debug support

> - Drop debugfs tags patch - it's too difficult to be the same between

> hostwide and non-hostwide, as discussed:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#mb3eb462d8be40273718505989abd12f8228c15fd

> And from commit 6bf0eb550452 ("sbitmap: Consider cleared bits in

> sbitmap_bitmap_show()"), I guess not many used this anyway...

> - Add elevator per-hctx request count for optimisation

> - Break up "blk-mq: rename blk_mq_update_tag_set_depth()" into 2x patches

> - Pass flags for avoid per-hq queue tags init/free for hostwide tags

> - Add Don's reviewed-tag and tested-by tags to appropiate patches

> 	- (@Don, please let me know if issue with how I did this)

> - Add "scsi: core: Show nr_hw_queues in sysfs"

> - Rework megaraid SAS patch to have module param (Kashyap)

> - rebase

> 

> V7 is here for more info:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#t

> 

> Hannes Reinecke (5):

>    blk-mq: Rename blk_mq_update_tag_set_depth()

>    blk-mq: Free tags in blk_mq_init_tags() upon error

>    scsi: Add host and host template flag 'host_tagset'

>    hpsa: enable host_tagset and switch to MQ

>    smartpqi: enable host tagset

> 

> John Garry (10):

>    blk-mq: Pass flags for tag init/free

>    blk-mq: Use pointers for blk_mq_tags bitmap tags

>    blk-mq: Facilitate a shared sbitmap per tagset

>    blk-mq: Relocate hctx_may_queue()

>    blk-mq: Record nr_active_requests per queue for when using shared

>      sbitmap

>    blk-mq: Record active_queues_shared_sbitmap per tag_set for when using

>      shared sbitmap

>    null_blk: Support shared tag bitmap

>    scsi: core: Show nr_hw_queues in sysfs

>    scsi: hisi_sas: Switch v3 hw to MQ

>    scsi: scsi_debug: Support host tagset

> 

> Kashyap Desai (2):

>    blk-mq, elevator: Count requests per hctx to improve performance

>    scsi: megaraid_sas: Added support for shared host tagset for

>      cpuhotplug

> 

> Ming Lei (1):

>    blk-mq: Rename BLK_MQ_F_TAG_SHARED as BLK_MQ_F_TAG_QUEUE_SHARED

> 

>   block/bfq-iosched.c                         |   9 +-

>   block/blk-core.c                            |   2 +

>   block/blk-mq-debugfs.c                      |  10 +-

>   block/blk-mq-sched.c                        |  13 +-

>   block/blk-mq-tag.c                          | 149 ++++++++++++++------

>   block/blk-mq-tag.h                          |  56 +++-----

>   block/blk-mq.c                              |  81 +++++++----

>   block/blk-mq.h                              |  76 +++++++++-

>   block/kyber-iosched.c                       |   4 +-

>   block/mq-deadline.c                         |   6 +

>   drivers/block/null_blk_main.c               |   6 +

>   drivers/block/rnbd/rnbd-clt.c               |   2 +-

>   drivers/scsi/hisi_sas/hisi_sas.h            |   3 +-

>   drivers/scsi/hisi_sas/hisi_sas_main.c       |  36 ++---

>   drivers/scsi/hisi_sas/hisi_sas_v3_hw.c      |  87 +++++-------

>   drivers/scsi/hosts.c                        |   1 +

>   drivers/scsi/hpsa.c                         |  44 +-----

>   drivers/scsi/hpsa.h                         |   1 -

>   drivers/scsi/megaraid/megaraid_sas_base.c   |  39 +++++

>   drivers/scsi/megaraid/megaraid_sas_fusion.c |  29 ++--

>   drivers/scsi/scsi_debug.c                   |  28 ++--

>   drivers/scsi/scsi_lib.c                     |   2 +

>   drivers/scsi/scsi_sysfs.c                   |  11 ++

>   drivers/scsi/smartpqi/smartpqi_init.c       |  45 ++++--

>   include/linux/blk-mq.h                      |  13 +-

>   include/linux/blkdev.h                      |   3 +

>   include/scsi/scsi_host.h                    |   9 +-

>   27 files changed, 484 insertions(+), 281 deletions(-)

>
Douglas Gilbert Sept. 3, 2020, 7:28 p.m. UTC | #2
On 2020-08-19 11:20 a.m., John Garry wrote:
> Hi all,

> 

> Here is v8 of the patchset.

> 

> In this version of the series, we keep the shared sbitmap for driver tags,

> and introduce changes to fix up the tag budgeting across request queues.

> We also have a change to count requests per-hctx for when an elevator is

> enabled, as an optimisation. I also dropped the debugfs changes - more on

> that below.

> 

> Some performance figures:

> 

> Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,

> but it is not always an appropriate scheduler to use.

> 

> Tag depth 		4000 (default)			260**

> 

> Baseline (v5.9-rc1):

> none sched:		2094K IOPS			513K

> mq-deadline sched:	2145K IOPS			1336K

> 

> Final, host_tagset=0 in LLDD *, ***:

> none sched:		2120K IOPS			550K

> mq-deadline sched:	2121K IOPS			1309K

> 

> Final ***:

> none sched:		2132K IOPS			1185			

> mq-deadline sched:	2145K IOPS			2097	

> 

> * this is relevant as this is the performance in supporting but not

>    enabling the feature

> ** depth=260 is relevant as some point where we are regularly waiting for

>     tags to be available. Figures were are a bit unstable here.

> *** Included "[PATCH V4] scsi: core: only re-run queue in

>      scsi_end_request() if device queue is busy"

> 

> A copy of the patches can be found here:

> https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

> 

> The hpsa patch depends on:

> https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

> 

> And the smartpqi patch is not to be accepted.

> 

> Comments (and testing) welcome, thanks!


I tested this v8 patchset on MKP's 5.10/scsi-queue branch together with
my rewritten sg driver on my laptop and a Ryzen 5 3600 machine. Since I
don't have same hardware, I use the scsi_debug driver as the target:

    modprobe scsi_debug dev_size_mb=1024 sector_size=512 add_host=7 
per_host_store=1 ndelay=1000 random=1 submit_queues=12

My test is a script which runs these three commands many times with
differing parameters:

sg_mrq_dd iflag=random bs=512 of=/dev/sg8 thr=64 time=2
time to transfer data was 0.312705 secs, 3433.72 MB/sec
2097152+0 records in
2097152+0 records out

sg_mrq_dd bpt=256 thr=64 mrq=36 time=2 if=/dev/sg8 bs=512 of=/dev/sg9
time to transfer data was 0.212090 secs, 5062.67 MB/sec
2097152+0 records in
2097152+0 records out

sg_mrq_dd --verify if=/dev/sg8 of=/dev/sg9 bs=512 bpt=256 thr=64 mrq=36 time=2
Doing verify/cmp rather than copy
time to transfer data was 0.184563 secs, 5817.75 MB/sec
2097152+0 records in
2097152+0 records verified

The above is the output from last section of the my script run on the Ryzen 5.

So the three steps are:
    1) produce random data on /dev/sg8
    2) copy /dev/sg8 to /dev/sg9
    3) verify /dev/sg8 and /dev/sg9 are the same.

The latter step is done with a sequence of READ(/dev/sg8) and
VERIFY(BYTCHK=1 on /dev/sg9). The "mrq" stands for multiple requests (in
one invocation; the bsg driver did that before its write(2) command was
removed.
The SCSI devices on the Ryzen 5 machine are:

# lsscsi -gs
[2:0:0:0]  disk    IBM-207x HUSMM8020ASS20   J4B6  /dev/sda   /dev/sg0   200GB
[2:0:1:0]  disk    SEAGATE  ST200FM0073      0007  /dev/sdb   /dev/sg1   200GB
[2:0:2:0]  enclosu Areca Te ARC-802801.37.69 0137  -          /dev/sg2       -
[3:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdc   /dev/sg3  1.07GB
[4:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdd   /dev/sg4  1.07GB
[5:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sde   /dev/sg5  1.07GB
[6:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdf   /dev/sg6  1.07GB
[7:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdg   /dev/sg7  1.07GB
[8:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdh   /dev/sg8  1.07GB
[9:0:0:0]  disk    Linux    scsi_debug       0190  /dev/sdi   /dev/sg9  1.07GB
[N:0:1:1]  disk    WDC WDS250G2B0C-00PXH0__1       /dev/nvme0n1  -      250GB

My script took 17m12 and the highest throughput (on a copy) was 7.5 GB/sec.
Then I reloaded the scsi_debug module, this time with an additional
'host_max_queue=128' parameter. The script run time was 5 seconds shorter
and the maximum throughput was around 7.6 GB/sec. [Average throughput is
around 4 GB/sec.]

For comparison:

# time liburing/examples/io_uring-cp /dev/sdh /dev/sdi
real	0m1.542s
user	0m0.004s
sys	0m1.027s

Umm, that's less then 1 GB/sec. In its defence io_uring-cp is an
extremely simple, single threaded, proof-of-concept copy program,
at least compared to sg_mrq_dd . As used by the sg_mrq_dd the
rewritten sg driver bypasses moving 1 GB to and from the _user_
space while doing the above copy and verify steps.

So:

Tested-by: Douglas Gilbert <dgilbert@interlog.com>


> Differences to v7:

> - Add null_blk and scsi_debug support

> - Drop debugfs tags patch - it's too difficult to be the same between

> hostwide and non-hostwide, as discussed:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#mb3eb462d8be40273718505989abd12f8228c15fd

> And from commit 6bf0eb550452 ("sbitmap: Consider cleared bits in

> sbitmap_bitmap_show()"), I guess not many used this anyway...

> - Add elevator per-hctx request count for optimisation

> - Break up "blk-mq: rename blk_mq_update_tag_set_depth()" into 2x patches

> - Pass flags for avoid per-hq queue tags init/free for hostwide tags

> - Add Don's reviewed-tag and tested-by tags to appropiate patches

> 	- (@Don, please let me know if issue with how I did this)

> - Add "scsi: core: Show nr_hw_queues in sysfs"

> - Rework megaraid SAS patch to have module param (Kashyap)

> - rebase

> 

> V7 is here for more info:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#t

> 

> Hannes Reinecke (5):

>    blk-mq: Rename blk_mq_update_tag_set_depth()

>    blk-mq: Free tags in blk_mq_init_tags() upon error

>    scsi: Add host and host template flag 'host_tagset'

>    hpsa: enable host_tagset and switch to MQ

>    smartpqi: enable host tagset

> 

> John Garry (10):

>    blk-mq: Pass flags for tag init/free

>    blk-mq: Use pointers for blk_mq_tags bitmap tags

>    blk-mq: Facilitate a shared sbitmap per tagset

>    blk-mq: Relocate hctx_may_queue()

>    blk-mq: Record nr_active_requests per queue for when using shared

>      sbitmap

>    blk-mq: Record active_queues_shared_sbitmap per tag_set for when using

>      shared sbitmap

>    null_blk: Support shared tag bitmap

>    scsi: core: Show nr_hw_queues in sysfs

>    scsi: hisi_sas: Switch v3 hw to MQ

>    scsi: scsi_debug: Support host tagset

> 

> Kashyap Desai (2):

>    blk-mq, elevator: Count requests per hctx to improve performance

>    scsi: megaraid_sas: Added support for shared host tagset for

>      cpuhotplug

> 

> Ming Lei (1):

>    blk-mq: Rename BLK_MQ_F_TAG_SHARED as BLK_MQ_F_TAG_QUEUE_SHARED

> 

>   block/bfq-iosched.c                         |   9 +-

>   block/blk-core.c                            |   2 +

>   block/blk-mq-debugfs.c                      |  10 +-

>   block/blk-mq-sched.c                        |  13 +-

>   block/blk-mq-tag.c                          | 149 ++++++++++++++------

>   block/blk-mq-tag.h                          |  56 +++-----

>   block/blk-mq.c                              |  81 +++++++----

>   block/blk-mq.h                              |  76 +++++++++-

>   block/kyber-iosched.c                       |   4 +-

>   block/mq-deadline.c                         |   6 +

>   drivers/block/null_blk_main.c               |   6 +

>   drivers/block/rnbd/rnbd-clt.c               |   2 +-

>   drivers/scsi/hisi_sas/hisi_sas.h            |   3 +-

>   drivers/scsi/hisi_sas/hisi_sas_main.c       |  36 ++---

>   drivers/scsi/hisi_sas/hisi_sas_v3_hw.c      |  87 +++++-------

>   drivers/scsi/hosts.c                        |   1 +

>   drivers/scsi/hpsa.c                         |  44 +-----

>   drivers/scsi/hpsa.h                         |   1 -

>   drivers/scsi/megaraid/megaraid_sas_base.c   |  39 +++++

>   drivers/scsi/megaraid/megaraid_sas_fusion.c |  29 ++--

>   drivers/scsi/scsi_debug.c                   |  28 ++--

>   drivers/scsi/scsi_lib.c                     |   2 +

>   drivers/scsi/scsi_sysfs.c                   |  11 ++

>   drivers/scsi/smartpqi/smartpqi_init.c       |  45 ++++--

>   include/linux/blk-mq.h                      |  13 +-

>   include/linux/blkdev.h                      |   3 +

>   include/scsi/scsi_host.h                    |   9 +-

>   27 files changed, 484 insertions(+), 281 deletions(-)

>
Jens Axboe Sept. 3, 2020, 9:23 p.m. UTC | #3
On 8/19/20 9:20 AM, John Garry wrote:
> Hi all,

> 

> Here is v8 of the patchset.

> 

> In this version of the series, we keep the shared sbitmap for driver tags,

> and introduce changes to fix up the tag budgeting across request queues.

> We also have a change to count requests per-hctx for when an elevator is

> enabled, as an optimisation. I also dropped the debugfs changes - more on

> that below.

> 

> Some performance figures:

> 

> Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,

> but it is not always an appropriate scheduler to use.

> 

> Tag depth 		4000 (default)			260**

> 

> Baseline (v5.9-rc1):

> none sched:		2094K IOPS			513K

> mq-deadline sched:	2145K IOPS			1336K

> 

> Final, host_tagset=0 in LLDD *, ***:

> none sched:		2120K IOPS			550K

> mq-deadline sched:	2121K IOPS			1309K

> 

> Final ***:

> none sched:		2132K IOPS			1185			

> mq-deadline sched:	2145K IOPS			2097	

> 

> * this is relevant as this is the performance in supporting but not

>   enabling the feature

> ** depth=260 is relevant as some point where we are regularly waiting for

>    tags to be available. Figures were are a bit unstable here.

> *** Included "[PATCH V4] scsi: core: only re-run queue in

>     scsi_end_request() if device queue is busy"

> 

> A copy of the patches can be found here:

> https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

> 

> The hpsa patch depends on:

> https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

> 

> And the smartpqi patch is not to be accepted.

> 

> Comments (and testing) welcome, thanks!


I applied 1-11, leaving the SCSI core bits and drivers to Martin. I can
also carry them, just let me know.

-- 
Jens Axboe
John Garry Sept. 4, 2020, 9:09 a.m. UTC | #4
On 03/09/2020 22:23, Jens Axboe wrote:
> On 8/19/20 9:20 AM, John Garry wrote:

>> Hi all,

>>

>> Here is v8 of the patchset.

>>

>> In this version of the series, we keep the shared sbitmap for driver tags,

>> and introduce changes to fix up the tag budgeting across request queues.

>> We also have a change to count requests per-hctx for when an elevator is

>> enabled, as an optimisation. I also dropped the debugfs changes - more on

>> that below.

>>

>> Some performance figures:

>>

>> Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,

>> but it is not always an appropriate scheduler to use.

>>

>> Tag depth 		4000 (default)			260**

>>

>> Baseline (v5.9-rc1):

>> none sched:		2094K IOPS			513K

>> mq-deadline sched:	2145K IOPS			1336K

>>

>> Final, host_tagset=0 in LLDD *, ***:

>> none sched:		2120K IOPS			550K

>> mq-deadline sched:	2121K IOPS			1309K

>>

>> Final ***:

>> none sched:		2132K IOPS			1185			

>> mq-deadline sched:	2145K IOPS			2097	

>>

>> * this is relevant as this is the performance in supporting but not

>>    enabling the feature

>> ** depth=260 is relevant as some point where we are regularly waiting for

>>     tags to be available. Figures were are a bit unstable here.

>> *** Included "[PATCH V4] scsi: core: only re-run queue in

>>      scsi_end_request() if device queue is busy"

>>

>> A copy of the patches can be found here:

>> https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

>>

>> The hpsa patch depends on:

>> https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

>>

>> And the smartpqi patch is not to be accepted.

>>

>> Comments (and testing) welcome, thanks!

> 

> I applied 1-11, leaving the SCSI core bits and drivers to Martin. I can

> also carry them, just let me know.

> 


Great, thanks!

So the SCSI parts depend on the block parts for building, so I guess it 
makes sense if you could carry them also.

hpsa and smartpqi patches are pending for now, but the rest could be 
picked up. Martin/James may want more review of the SCSI core bits, though.

Thanks again,
John
Martin K. Petersen Sept. 4, 2020, 12:44 p.m. UTC | #5
John,

> Martin/James may want more review of the SCSI core bits, though.


I'll take a look later today.

-- 
Martin K. Petersen	Oracle Linux Engineering
Hannes Reinecke Sept. 8, 2020, 12:46 p.m. UTC | #6
On 8/19/20 5:20 PM, John Garry wrote:
> Hi all,

> 

> Here is v8 of the patchset.

> 

> In this version of the series, we keep the shared sbitmap for driver tags,

> and introduce changes to fix up the tag budgeting across request queues.

> We also have a change to count requests per-hctx for when an elevator is

> enabled, as an optimisation. I also dropped the debugfs changes - more on

> that below.

> 

> Some performance figures:

> 

> Using 12x SAS SSDs on hisi_sas v3 hw. mq-deadline results are included,

> but it is not always an appropriate scheduler to use.

> 

> Tag depth 		4000 (default)			260**

> 

> Baseline (v5.9-rc1):

> none sched:		2094K IOPS			513K

> mq-deadline sched:	2145K IOPS			1336K

> 

> Final, host_tagset=0 in LLDD *, ***:

> none sched:		2120K IOPS			550K

> mq-deadline sched:	2121K IOPS			1309K

> 

> Final ***:

> none sched:		2132K IOPS			1185			

> mq-deadline sched:	2145K IOPS			2097	

> 

> * this is relevant as this is the performance in supporting but not

>   enabling the feature

> ** depth=260 is relevant as some point where we are regularly waiting for

>    tags to be available. Figures were are a bit unstable here.

> *** Included "[PATCH V4] scsi: core: only re-run queue in

>     scsi_end_request() if device queue is busy"

> 

> A copy of the patches can be found here:

> https://github.com/hisilicon/kernel-dev/tree/private-topic-blk-mq-shared-tags-v8

> 

> The hpsa patch depends on:

> https://lore.kernel.org/linux-scsi/20200430131904.5847-1-hare@suse.de/

> 

> And the smartpqi patch is not to be accepted.

> 

> Comments (and testing) welcome, thanks!

> 

> Differences to v7:

> - Add null_blk and scsi_debug support

> - Drop debugfs tags patch - it's too difficult to be the same between

> hostwide and non-hostwide, as discussed:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#mb3eb462d8be40273718505989abd12f8228c15fd

> And from commit 6bf0eb550452 ("sbitmap: Consider cleared bits in

> sbitmap_bitmap_show()"), I guess not many used this anyway...

> - Add elevator per-hctx request count for optimisation

> - Break up "blk-mq: rename blk_mq_update_tag_set_depth()" into 2x patches

> - Pass flags for avoid per-hq queue tags init/free for hostwide tags

> - Add Don's reviewed-tag and tested-by tags to appropiate patches

> 	- (@Don, please let me know if issue with how I did this)

> - Add "scsi: core: Show nr_hw_queues in sysfs"

> - Rework megaraid SAS patch to have module param (Kashyap)

> - rebase

> 

> V7 is here for more info:

> https://lore.kernel.org/linux-scsi/1591810159-240929-1-git-send-email-john.garry@huawei.com/T/#t

> 

> Hannes Reinecke (5):

>   blk-mq: Rename blk_mq_update_tag_set_depth()

>   blk-mq: Free tags in blk_mq_init_tags() upon error

>   scsi: Add host and host template flag 'host_tagset'

>   hpsa: enable host_tagset and switch to MQ

>   smartpqi: enable host tagset

> 

> John Garry (10):

>   blk-mq: Pass flags for tag init/free

>   blk-mq: Use pointers for blk_mq_tags bitmap tags

>   blk-mq: Facilitate a shared sbitmap per tagset

>   blk-mq: Relocate hctx_may_queue()

>   blk-mq: Record nr_active_requests per queue for when using shared

>     sbitmap

>   blk-mq: Record active_queues_shared_sbitmap per tag_set for when using

>     shared sbitmap

>   null_blk: Support shared tag bitmap

>   scsi: core: Show nr_hw_queues in sysfs

>   scsi: hisi_sas: Switch v3 hw to MQ

>   scsi: scsi_debug: Support host tagset

> 

> Kashyap Desai (2):

>   blk-mq, elevator: Count requests per hctx to improve performance

>   scsi: megaraid_sas: Added support for shared host tagset for

>     cpuhotplug

> 

> Ming Lei (1):

>   blk-mq: Rename BLK_MQ_F_TAG_SHARED as BLK_MQ_F_TAG_QUEUE_SHARED

> 

Now that Jens merged the block bits in his tree, wouldn't it be better
to re-send the SCSI bits only, thereby avoiding a potential merge error
later on?

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		           Kernel Storage Architect
hare@suse.de			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
John Garry Sept. 8, 2020, 1:38 p.m. UTC | #7
On 08/09/2020 13:46, Hannes Reinecke wrote:
> Now that Jens merged the block bits in his tree, wouldn't it be better

> to re-send the SCSI bits only, thereby avoiding a potential merge error

> later on?

> 


Anything which I resend would need to be against Jens' tree (and not 
Martin's), assuming Jens will carry them also. So I am not sure how that 
will help.

JFYI, I just tested against today's linux-next, and the SCSI parts (hpsa 
and smartpqi omitted) still apply there without conflict.

Thanks,
John
John Garry Sept. 16, 2020, 7:21 a.m. UTC | #8
On 04/09/2020 13:44, Martin K. Petersen wrote:
>> Martin/James may want more review of the SCSI core bits, though.
> I'll take a look later today.


Hi Martin,

Have you had a chance to check these outstanding SCSI patches?

scsi: megaraid_sas: Added support for shared host tagset for cpuhotplug
scsi: scsi_debug: Support host tagset
scsi: hisi_sas: Switch v3 hw to MQ
scsi: core: Show nr_hw_queues in sysfs
scsi: Add host and host template flag 'host_tagset'

Cheers,
John
Martin K. Petersen Sept. 17, 2020, 1:10 a.m. UTC | #9
John,

> Have you had a chance to check these outstanding SCSI patches?
>
> scsi: megaraid_sas: Added support for shared host tagset for cpuhotplug
> scsi: scsi_debug: Support host tagset
> scsi: hisi_sas: Switch v3 hw to MQ
> scsi: core: Show nr_hw_queues in sysfs
> scsi: Add host and host template flag 'host_tagset'

These look good to me.

Jens, feel free to merge.

Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
John Garry Sept. 17, 2020, 6:48 a.m. UTC | #10
> 
>> Have you had a chance to check these outstanding SCSI patches?
>>
>> scsi: megaraid_sas: Added support for shared host tagset for cpuhotplug
>> scsi: scsi_debug: Support host tagset
>> scsi: hisi_sas: Switch v3 hw to MQ
>> scsi: core: Show nr_hw_queues in sysfs
>> scsi: Add host and host template flag 'host_tagset'
> 
> These look good to me.
> 
> Jens, feel free to merge.

Hi Jens,

I'm waiting on the hpsa and smartpqi patches update, so please kindly 
merge only those patches, above.

Thanks!

> 
> Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
> 

Cheers Martin
Don Brace Sept. 21, 2020, 9:35 p.m. UTC | #11
Subject: Re: [PATCH v8 00/18] blk-mq/scsi: Provide hostwide shared tags for SCSI HBAs

>>Hi Jens,


>>I'm waiting on the hpsa and smartpqi patches >>update, so please kindly merge only those >>patches, above.


>>Thanks!


John, the hpsa driver crashes, the  or more patches to allow internal commands from Hannas seem to be missing.

I'll let you know exactly which ones soon.

Thanks,
Don
John Garry Sept. 21, 2020, 10:15 p.m. UTC | #12
On 21/09/2020 22:35, Don.Brace@microchip.com wrote:
>>> I'm waiting on the hpsa and smartpqi patches >>update, so please kindly merge only those >>patches, above.
>>> Thanks!
> John, the hpsa driver crashes, the  or more patches to allow internal commands from Hannas seem to be missing.
> 
> I'll let you know exactly which ones soon.
> 

Hi Don,

Right, that branch did not include Hannes patches as they did not apply 
cleanly, but I think that you need the same patches as before. I can 
create a branch for you to test which does include those tomorrow - let 
me know.

Alternatively I think that we could create a hpsa patch which does not 
rely on that series, like I mentioned here [0], but it would not be as 
clean.

Cheers,
John

https://lore.kernel.org/linux-scsi/dc0e72d8-7076-060c-3cd3-3d51ac7e6de8@huawei.com/
John Garry Sept. 22, 2020, 9:03 a.m. UTC | #13
On 21/09/2020 23:15, John Garry wrote:
> On 21/09/2020 22:35, Don.Brace@microchip.com wrote:
>>>> I'm waiting on the hpsa and smartpqi patches >>update, so please 
>>>> kindly merge only those >>patches, above.
>>>> Thanks!
>> John, the hpsa driver crashes, the  or more patches to allow internal 
>> commands from Hannas seem to be missing.
>>
>> I'll let you know exactly which ones soon.
>>
> 
> Hi Don,
> 
> Right, that branch did not include Hannes patches as they did not apply 
> cleanly, but I think that you need the same patches as before. I can 
> create a branch for you to test which does include those tomorrow - let 
> me know.
> 
> Alternatively I think that we could create a hpsa patch which does not 
> rely on that series, like I mentioned here [0], but it would not be as 
> clean.
> 
> Cheers,
> John
> 
> https://lore.kernel.org/linux-scsi/dc0e72d8-7076-060c-3cd3-3d51ac7e6de8@huawei.com/ 
> 
> .

JFYI, I put the reserved command v6 patchset and $subject patchset on 
this branch:

https://github.com/hisilicon/kernel-dev/commits/private-topic-blk-mq-shared-tags-rfc-v8-resv-v6-baseline

I don't have HW to test hpsa or smartpqi.

Thanks,
John
Kashyap Desai Sept. 28, 2020, 4:11 p.m. UTC | #14
>
> John,
>
> > Have you had a chance to check these outstanding SCSI patches?
> >
> > scsi: megaraid_sas: Added support for shared host tagset for
cpuhotplug
> > scsi: scsi_debug: Support host tagset
> > scsi: hisi_sas: Switch v3 hw to MQ
> > scsi: core: Show nr_hw_queues in sysfs
> > scsi: Add host and host template flag 'host_tagset'
>
> These look good to me.
>
> Jens, feel free to merge.

Hi Jens, Gentle ping. I am not able to find commits for above listed scsi
patches. I want to use your repo which has above mentioned patch for
<scsi:io_uring iopoll> patch submission.  Martin has Acked the scsi
patches.

>
> Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
>
> --
> Martin K. Petersen	Oracle Linux Engineering
John Garry Oct. 6, 2020, 2:24 p.m. UTC | #15
On 28/09/2020 17:11, Kashyap Desai wrote:
>>

>> John,

>>

>>> Have you had a chance to check these outstanding SCSI patches?

>>>

>>> scsi: megaraid_sas: Added support for shared host tagset for

> cpuhotplug

>>> scsi: scsi_debug: Support host tagset

>>> scsi: hisi_sas: Switch v3 hw to MQ

>>> scsi: core: Show nr_hw_queues in sysfs

>>> scsi: Add host and host template flag 'host_tagset'

>>

>> These look good to me.

>>

>> Jens, feel free to merge.

> 

> Hi Jens, Gentle ping. I am not able to find commits for above listed scsi

> patches. I want to use your repo which has above mentioned patch for

> <scsi:io_uring iopoll> patch submission.  Martin has Acked the scsi

> patches.

> 

>>

>> Acked-by: Martin K. Petersen <martin.petersen@oracle.com>

>>

>> --

>> Martin K. Petersen	Oracle Linux Engineering



Hi Jens,

Could you kindly pick up the following patches, to go along with the 
blk-mq changes:

scsi: megaraid_sas: Added support for shared host tagset for
cpuhotplug
scsi: scsi_debug: Support host tagset
scsi: hisi_sas: Switch v3 hw to MQ
scsi: core: Show nr_hw_queues in sysfs
scsi: Add host and host template flag 'host_tagset'

Thanks,
John
Jens Axboe Oct. 6, 2020, 2:42 p.m. UTC | #16
On 10/6/20 8:24 AM, John Garry wrote:
> On 28/09/2020 17:11, Kashyap Desai wrote:
>>>
>>> John,
>>>
>>>> Have you had a chance to check these outstanding SCSI patches?
>>>>
>>>> scsi: megaraid_sas: Added support for shared host tagset for
>> cpuhotplug
>>>> scsi: scsi_debug: Support host tagset
>>>> scsi: hisi_sas: Switch v3 hw to MQ
>>>> scsi: core: Show nr_hw_queues in sysfs
>>>> scsi: Add host and host template flag 'host_tagset'
>>>
>>> These look good to me.
>>>
>>> Jens, feel free to merge.
>>
>> Hi Jens, Gentle ping. I am not able to find commits for above listed scsi
>> patches. I want to use your repo which has above mentioned patch for
>> <scsi:io_uring iopoll> patch submission.  Martin has Acked the scsi
>> patches.
>>
>>>
>>> Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
>>>
>>> --
>>> Martin K. Petersen	Oracle Linux Engineering
> 
> 
> Hi Jens,
> 
> Could you kindly pick up the following patches, to go along with the 
> blk-mq changes:
> 
> scsi: megaraid_sas: Added support for shared host tagset for
> cpuhotplug
> scsi: scsi_debug: Support host tagset
> scsi: hisi_sas: Switch v3 hw to MQ
> scsi: core: Show nr_hw_queues in sysfs
> scsi: Add host and host template flag 'host_tagset'

Sorry about the delay, picked up these 5 patches.