Message-ID: <CAPDyKFrgC91Z3P-5U22gjKDqs_aZiQncGUrShWpFggkWzrRMjw@mail.gmail.com>
Date:   Wed, 29 Nov 2017 16:47:07 +0100
From:   Ulf Hansson <ulf.hansson@...aro.org>
To:     Adrian Hunter <adrian.hunter@...el.com>
Cc:     linux-mmc <linux-mmc@...r.kernel.org>,
        linux-block <linux-block@...r.kernel.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Bough Chen <haibo.chen@....com>,
        Alex Lemberg <alex.lemberg@...disk.com>,
        Mateusz Nowak <mateusz.nowak@...el.com>,
        Yuliy Izrailov <Yuliy.Izrailov@...disk.com>,
        Jaehoon Chung <jh80.chung@...sung.com>,
        Dong Aisheng <dongas86@...il.com>,
        Das Asutosh <asutoshd@...eaurora.org>,
        Zhangfei Gao <zhangfei.gao@...il.com>,
        Sahitya Tummala <stummala@...eaurora.org>,
        Harjani Ritesh <riteshh@...eaurora.org>,
        Venu Byravarasu <vbyravarasu@...dia.com>,
        Linus Walleij <linus.walleij@...aro.org>,
        Shawn Lin <shawn.lin@...k-chips.com>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@...sung.com>,
        Christoph Hellwig <hch@....de>
Subject: Re: [PATCH V15 00/22] mmc: Add Command Queue support

Hi Adrian,

On 29 November 2017 at 14:40, Adrian Hunter <adrian.hunter@...el.com> wrote:
> Hi
>
> Here is V15 of the hardware command queue patches without the software
> command queue patches, now using blk-mq and now with blk-mq support for
> non-CQE I/O.

I have applied patches 1->19 for next.  Deferring patch 21->23 for a while.

For those patches that was more or less the same as in v14, I added Linus' ack.

Hopefully we get some help for the community to test this series on
different HW (and I will be checking kernelci's boot reports). I
haven't added Bartlomiej's tested-by and neither Linus' (because of
the changes that has been made), so I hoping that will happen sooner
or later.

Moreover, I will gladly add more peoples acks/reviewed-by and
tested-by tags, at any point during this release cycle.

Thanks and kind regards
Uffe

>
> V14 included a number of fixes to existing code, changes to default to
> blk-mq, and adds patches to remove legacy code.
>
> HW CMDQ offers 25% - 50% better random multi-threaded I/O.  I see a slight
> 2% drop in sequential read speed but no change to sequential write.
>
> Non-CQE blk-mq showed a 3% decrease in sequential read performance.  This
> seemed to be coming from the inferior latency of running work items compared
> with a dedicated thread.  Hacking blk-mq workqueue to be unbound reduced the
> performance degradation from 3% to 1%.
>
> While we should look at changing blk-mq to give better workqueue performance,
> a bigger gain is likely to be made by adding a new host API to enable the
> next already-prepared request to be issued directly from within ->done()
> callback of the current request.
>
> Changes since V14:
>       mmc: block: Fix missing blk_put_request()
>       mmc: block: Check return value of blk_get_request()
>       mmc: core: Do not leave the block driver in a suspended state
>       mmc: block: Ensure that debugfs files are removed
>         Dropped because they have been applied
>       mmc: block: Use data timeout in card_busy_detect()
>         Replaced by other patches
>       mmc: block: Add blk-mq support
>         Rename mmc_blk_ss_read() to mmc_blk_read_single()
>         Add more error handling to single sector read
>         Let mmc_blk_mq_complete_rq() cater for requests already "updated" by recovery
>         Rename mmc_blk_mq_acct_req_done() to mmc_blk_mq_dec_in_flight()
>         Add comments about synchronization
>         Add comment about not dispatching in parallel
>         Add comment about the queue depth
>       mmc: block: Add CQE support
>         Add coment about CQE queue depth
>       mmc: block: blk-mq: Add support for direct completion
>         Rename mmc_queue_direct_complete() to mmc_host_done_complete()
>         Rename MMC_CAP_DIRECT_COMPLETE to MMC_CAP_DONE_COMPLETE
>       mmc: block: blk-mq: Separate card polling from recovery
>         Ensure to report gen_err as an error
>       mmc: block: Make card_busy_detect() accumulate all response error bits
>         Patch moved later in the patch set and adjusted accordingly
>       mmc: block: blk-mq: Check error bits and save the exception bit when polling card busy
>         Adjusted due to patch re-ordering
>       mmc: block: Check the timeout correctly in card_busy_detect()
>         New patch.
>       mmc: block: Add timeout_clks when calculating timeout
>         New patch.
>       mmc: block: Reduce polling timeout from 10 minutes to 10 seconds
>         New patch.
>
> Changes since V13:
>       mmc: block: Fix missing blk_put_request()
>         New patch.
>       mmc: block: Check return value of blk_get_request()
>         New patch.
>       mmc: core: Do not leave the block driver in a suspended state
>         New patch.
>       mmc: block: Ensure that debugfs files are removed
>         New patch.
>       mmc: block: No need to export mmc_cleanup_queue()
>         New patch.
>       mmc: block: Simplify cleaning up the queue
>         New patch.
>       mmc: block: Use data timeout in card_busy_detect()
>         New patch.
>       mmc: block: Check for transfer state in card_busy_detect()
>         New patch.
>       mmc: block: Make card_busy_detect() accumulate all response error bits
>         New patch.
>       mmc: core: Make mmc_pre_req() and mmc_post_req() available
>         New patch.
>       mmc: core: Add parameter use_blk_mq
>         Default to y
>       mmc: block: Add blk-mq support
>         Wrap blk_mq_end_request / blk_end_request_all
>         Rename mmc_blk_rw_recovery -> mmc_blk_mq_rw_recovery
>         Additional parentheses to '==' expressions
>         Use mmc_pre_req() / mmc_post_req()
>         Fix missing tuning release on error after mmc_start_request()
>         Expand comment about timeouts
>         Allow for possibility that the queue is quiesced when removing
>         Ensure complete_work is flushed when removing
>       mmc: block: Add CQE support
>         Additional parentheses to '==' expressions
>       mmc: block: blk-mq: Check error bits and save the exception bit when polling card busy
>         Replaces patch "Stop using card_busy_detect()" retaining card_busy_detect()
>       mmc: block: blk-mq: Stop using legacy recovery
>         Allow for SPI
>       mmc: mmc_test: Do not use mmc_start_areq() anymore
>         New patch.
>       mmc: core: Remove option not to use blk-mq
>         New patch.
>       mmc: block: Remove code no longer needed after the switch to blk-mq
>         New patch.
>       mmc: core: Remove code no longer needed after the switch to blk-mq
>         New patch.
>
> Changes since V12:
>       mmc: block: Add error-handling comments
>         New patch.
>       mmc: block: Add blk-mq support
>         Use legacy error handling
>       mmc: block: Add CQE support
>         Re-base
>       mmc: block: blk-mq: Add support for direct completion
>         New patch.
>       mmc: block: blk-mq: Separate card polling from recovery
>         New patch.
>       mmc: block: blk-mq: Stop using card_busy_detect()
>         New patch.
>       mmc: block: blk-mq: Stop using legacy recovery
>         New patch.
>
> Changes since V11:
>       Split "mmc: block: Add CQE and blk-mq support" into 2 patches
>
> Changes since V10:
>       mmc: core: Remove unnecessary host claim
>       mmc: core: Introduce host claiming by context
>       mmc: core: Add support for handling CQE requests
>       mmc: mmc: Enable Command Queuing
>       mmc: mmc: Enable CQE's
>       mmc: block: Use local variables in mmc_blk_data_prep()
>       mmc: block: Prepare CQE data
>       mmc: block: Factor out mmc_setup_queue()
>       mmc: core: Add parameter use_blk_mq
>       mmc: core: Export mmc_start_bkops()
>       mmc: core: Export mmc_start_request()
>       mmc: core: Export mmc_retune_hold_now() and mmc_retune_release()
>         Dropped because they have been applied
>       mmc: block: Add CQE and blk-mq support
>         Extend blk-mq support for asynchronous read / writes to all host
>         controllers including those that require polling. The direct
>         completion path is still available but depends on a new capability
>         flag.
>         Drop blk-mq support for synchronous read / writes.
>
> Venkat Gopalakrishnan (1):
>       mmc: cqhci: support for command queue enabled host
>
> Changes since V9:
>       mmc: block: Add CQE and blk-mq support
>         - reinstate mq support for REQ_OP_DRV_IN/OUT that was removed because
>         it was incorrectly assumed to be handled by the rpmb character device
>         - don't check for rpmb block device anymore
>       mmc: cqhci: support for command queue enabled host
>         Fix cqhci_set_irqs() as per Haibo Chen
>
> Changes since V8:
>         Re-based
>       mmc: core: Introduce host claiming by context
>         Slightly simplified as per Ulf
>       mmc: core: Export mmc_retune_hold_now() and mmc_retune_release()
>         New patch.
>       mmc: block: Add CQE and blk-mq support
>         Fix missing ->post_req() on the error path
>
> Changes since V7:
>         Re-based
>       mmc: core: Introduce host claiming by context
>         Slightly simplified
>       mmc: core: Add parameter use_blk_mq
>         New patch.
>       mmc: core: Remove unnecessary host claim
>         New patch.
>       mmc: core: Export mmc_start_bkops()
>         New patch.
>       mmc: core: Export mmc_start_request()
>         New patch.
>       mmc: block: Add CQE and blk-mq support
>         Add blk-mq support for non_CQE requests
>
> Changes since V6:
>       mmc: core: Introduce host claiming by context
>         New patch.
>       mmc: core: Move mmc_start_areq() declaration
>         Dropped because it has been applied
>       mmc: block: Fix block status codes
>         Dropped because it has been applied
>       mmc: host: Add CQE interface
>         Dropped because it has been applied
>       mmc: core: Turn off CQE before sending commands
>         Dropped because it has been applied
>       mmc: block: Factor out mmc_setup_queue()
>         New patch.
>       mmc: block: Add CQE support
>         Drop legacy support and add blk-mq support
>
> Changes since V5:
>         Re-based
>       mmc: core: Add mmc_retune_hold_now()
>         Dropped because it has been applied
>       mmc: core: Add members to mmc_request and mmc_data for CQE's
>         Dropped because it has been applied
>       mmc: core: Move mmc_start_areq() declaration
>         New patch at Ulf's request
>       mmc: block: Fix block status codes
>         Another un-related patch
>       mmc: host: Add CQE interface
>         Move recovery_notifier() callback to struct mmc_request
>       mmc: core: Add support for handling CQE requests
>         Roll __mmc_cqe_request_done() into mmc_cqe_request_done()
>         Move function declarations requested by Ulf
>       mmc: core: Remove unused MMC_CAP2_PACKED_CMD
>         Dropped because it has been applied
>       mmc: block: Add CQE support
>         Add explanation to commit message
>         Adjustment for changed recovery_notifier() callback
>       mmc: cqhci: support for command queue enabled host
>         Adjustment for changed recovery_notifier() callback
>       mmc: sdhci-pci: Add CQHCI support for Intel GLK
>         Add DCMD capability for Intel controllers except GLK
>
> Changes since V4:
>       mmc: core: Add mmc_retune_hold_now()
>         Add explanation to commit message.
>       mmc: host: Add CQE interface
>         Add comments to callback declarations.
>       mmc: core: Turn off CQE before sending commands
>         Add explanation to commit message.
>       mmc: core: Add support for handling CQE requests
>         Add comments as requested by Ulf.
>       mmc: core: Remove unused MMC_CAP2_PACKED_CMD
>         New patch.
>       mmc: mmc: Enable Command Queuing
>         Adjust for removal of MMC_CAP2_PACKED_CMD.
>         Add a comment about Packed Commands.
>       mmc: mmc: Enable CQE's
>         Remove un-necessary check for MMC_CAP2_CQE
>       mmc: block: Use local variables in mmc_blk_data_prep()
>         New patch.
>       mmc: block: Prepare CQE data
>         Adjust due to "mmc: block: Use local variables in mmc_blk_data_prep()"
>         Remove priority setting.
>         Add explanation to commit message.
>       mmc: cqhci: support for command queue enabled host
>         Fix transfer descriptor setting in cqhci_set_tran_desc() for 32-bit DMA
>
> Changes since V3:
>         Adjusted ...blk_end_request...() for new block status codes
>         Fixed CQHCI transaction descriptor for "no DCMD" case
>
> Changes since V2:
>         Dropped patches that have been applied.
>         Re-based
>         Added "mmc: sdhci-pci: Add CQHCI support for Intel GLK"
>
> Changes since V1:
>
>         "Share mmc request array between partitions" is dependent
>         on changes in "Introduce queue semantics", so added that
>         and block fixes:
>
>         Added "Fix is_waiting_last_req set incorrectly"
>         Added "Fix cmd error reset failure path"
>         Added "Use local var for mqrq_cur"
>         Added "Introduce queue semantics"
>
> Changes since RFC:
>
>         Re-based on next.
>         Added comment about command queue priority.
>         Added some acks and reviews.
>
>
> Adrian Hunter (21):
>       mmc: block: No need to export mmc_cleanup_queue()
>       mmc: block: Simplify cleaning up the queue
>       mmc: core: Make mmc_pre_req() and mmc_post_req() available
>       mmc: block: Add error-handling comments
>       mmc: core: Add parameter use_blk_mq
>       mmc: block: Add blk-mq support
>       mmc: block: Add CQE support
>       mmc: sdhci-pci: Add CQHCI support for Intel GLK
>       mmc: block: blk-mq: Add support for direct completion
>       mmc: block: blk-mq: Separate card polling from recovery
>       mmc: block: Make card_busy_detect() accumulate all response error bits
>       mmc: block: blk-mq: Check error bits and save the exception bit when polling card busy
>       mmc: block: Check the timeout correctly in card_busy_detect()
>       mmc: block: Check for transfer state in card_busy_detect()
>       mmc: block: Add timeout_clks when calculating timeout
>       mmc: block: Reduce polling timeout from 10 minutes to 10 seconds
>       mmc: block: blk-mq: Stop using legacy recovery
>       mmc: mmc_test: Do not use mmc_start_areq() anymore
>       mmc: core: Remove option not to use blk-mq
>       mmc: block: Remove code no longer needed after the switch to blk-mq
>       mmc: core: Remove code no longer needed after the switch to blk-mq
>
> Venkat Gopalakrishnan (1):
>       mmc: cqhci: support for command queue enabled host
>
>  drivers/mmc/core/block.c          | 1383 +++++++++++++++++++++----------------
>  drivers/mmc/core/block.h          |   12 +-
>  drivers/mmc/core/bus.c            |    2 -
>  drivers/mmc/core/core.c           |  216 +-----
>  drivers/mmc/core/core.h           |   39 +-
>  drivers/mmc/core/host.h           |    6 +-
>  drivers/mmc/core/mmc_test.c       |  122 ++--
>  drivers/mmc/core/queue.c          |  504 +++++++++-----
>  drivers/mmc/core/queue.h          |   64 +-
>  drivers/mmc/host/Kconfig          |   14 +
>  drivers/mmc/host/Makefile         |    1 +
>  drivers/mmc/host/cqhci.c          | 1150 ++++++++++++++++++++++++++++++
>  drivers/mmc/host/cqhci.h          |  240 +++++++
>  drivers/mmc/host/sdhci-pci-core.c |  155 ++++-
>  include/linux/mmc/host.h          |    5 +-
>  15 files changed, 2835 insertions(+), 1078 deletions(-)
>  create mode 100644 drivers/mmc/host/cqhci.c
>  create mode 100644 drivers/mmc/host/cqhci.h
>
>
> Regards
> Adrian

Powered by blists - more mailing lists