[GIT PULL] nvme updates for Linux 4.15

[GIT PULL] nvme updates for Linux 4.15

[Christoph Hellwig]

Hi Jens,

below are the currently queue nvme updates for Linux 4.15.  There are
a few more things that could make it for this merge window, but I'd
like to get things into linux-next, especially for the unlikely case
that Linus decided to cut -rc8.

Highlights:
 - support for SGLs in the PCIe driver (Chaitanya Kulkarni)
 - disable I/O schedulers for the admin queue (Israel Rukshin)
 - various Fibre Channel fixes and enhancements (James Smart)
 - various refactoring for better code sharing between transports
   (Sagi Grimberg and me)

as well as lots of little bits from various contributors.

The following changes since commit 9c9883744dda1cc38339a448dd8435140537027e:

  block: move __elv_next_request to blk-core.c (2017-10-03 08:43:04 -0600)

are available in the git repository at:

  git://git.infradead.org/nvme.git nvme-4.15

for you to fetch changes up to a806c6c81e6c0d07c8a8b2643bad4a37a196d681:

  nvme: comment typo fixed in clearing AER (2017-11-03 10:02:20 +0300)

----------------------------------------------------------------
Chaitanya Kulkarni (1):
      nvme-pci: add SGL support

Christoph Hellwig (11):
      nvme: simplify compat_ioctl handling
      nvme: use ida_simple_{get,remove} for the controller instance
      nvme: use kref_get_unless_zero in nvme_find_get_ns
      nvme: simplify nvme_open
      nvme: switch controller refcounting to use struct device
      nvme: get rid of nvme_ctrl_list
      nvme: check for a live controller in nvme_dev_open
      nvme-fc: merge __nvme_fc_schedule_delete_work into __nvme_fc_del_ctrl
      nvme: move controller deletion to common code
      nvme-rdma: remove nvme_rdma_remove_ctrl
      nvme: consolidate common code from ->reset_work

Israel Rukshin (3):
      nvme-rdma: Add BLK_MQ_F_NO_SCHED flag to admin tag set
      nvme-fc: Add BLK_MQ_F_NO_SCHED flag to admin tag set
      nvme-loop: Add BLK_MQ_F_NO_SCHED flag to admin tag set

James Smart (18):
      nvmet: bump NVMET_NR_QUEUES to 128
      nvme-fc: add uevent for auto-connect
      nvme-fc: create fc class and transport device
      nvme-fc: move remote port get/put/free location
      nvme-fc: correct io termination handling
      nvme-fc: correct io timeout behavior
      nvme: add duplicate_connect option
      nvme: add helper to compare options to controller
      nvme-rdma: add support for duplicate_connect option
      nvme-fc: add support for duplicate_connect option
      nvme-fc: remove NVME_FC_MAX_SEGMENTS
      nvme-fc: avoid workqueue flush stalls
      nvme-fc: change ctlr state assignments during reset/reconnect
      nvme-fc: add a dev_loss_tmo field to the remoteport
      nvme-fc: check connectivity before initiating reconnects
      nvme: allow controller RESETTING to RECONNECTING transition
      nvme-fc: add dev_loss_tmo timeout and remoteport resume support
      nvmet: fix fatal_err_work deadlock

Keith Busch (1):
      nvme: Remove unused headers

Marc Olson (1):
      nvme: update timeout module parameter type

Max Gurtovoy (1):
      nvme-rdma: align nvme_rdma_device structure

Minwoo Im (2):
      nvme-pci: fix typos in comments
      nvme: comment typo fixed in clearing AER

Nitzan Carmi (1):
      nvme-rdma: Add debug message when reaches timeout

Randy Dunlap (1):
      nvme: use menu Kconfig interface

Roy Shterman (1):
      nvmet: Change max_nsid in subsystem due to ns_disable if needed

Sagi Grimberg (14):
      nvme-fabrics: request transport module
      block: introduce blk_mq_tagset_iter
      nvme: introduce nvme_reinit_tagset
      block: remove blk_mq_reinit_tagset
      nvme-rdma: pass tagset to directly nvme_rdma_free_tagset
      nvme-rdma: fix wrong logging message
      nvme-rdma: move assignment to declaration
      nvme-rdma: Check that reinit_request got a proper mr
      nvme-rdma: teardown admin/io queues once on error recovery
      nvme-rdma: Don't local invalidate if the queue is not live
      nvme-rdma: change queue flag semantics DELETING -> ALLOCATED
      nvme-rdma: stop controller reset if the controller is deleting
      nvme-rdma: reuse nvme_delete_ctrl when reconnect attempts expire
      nvme: flush reset_work before safely continuing with delete operation

 block/blk-mq-tag.c             |  11 +-
 drivers/nvme/Kconfig           |   4 +
 drivers/nvme/host/core.c       | 260 +++++++---------
 drivers/nvme/host/fabrics.c    |  12 +-
 drivers/nvme/host/fabrics.h    |  14 +
 drivers/nvme/host/fc.c         | 656 ++++++++++++++++++++++++++++++-----------
 drivers/nvme/host/nvme.h       |  26 +-
 drivers/nvme/host/pci.c        | 224 +++++++++++---
 drivers/nvme/host/rdma.c       | 225 ++++++++------
 drivers/nvme/target/core.c     |  13 +
 drivers/nvme/target/fc.c       |  16 +-
 drivers/nvme/target/loop.c     |  47 +--
 drivers/nvme/target/nvmet.h    |   2 +-
 include/linux/blk-mq.h         |   4 +-
 include/linux/nvme-fc-driver.h |  15 +-
 15 files changed, 1026 insertions(+), 503 deletions(-)

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/03/2017 06:17 AM, Christoph Hellwig wrote:
> Hi Jens,
> 
> below are the currently queue nvme updates for Linux 4.15.  There are
> a few more things that could make it for this merge window, but I'd
> like to get things into linux-next, especially for the unlikely case
> that Linus decided to cut -rc8.
> 
> Highlights:
>  - support for SGLs in the PCIe driver (Chaitanya Kulkarni)
>  - disable I/O schedulers for the admin queue (Israel Rukshin)
>  - various Fibre Channel fixes and enhancements (James Smart)
>  - various refactoring for better code sharing between transports
>    (Sagi Grimberg and me)
> 
> as well as lots of little bits from various contributors.

Pulled, thanks.

-- 
Jens Axboe

[GIT PULL] nvme updates for Linux 4.15

[Christoph Hellwig]

Hi Jens,

another round of NVMe updates for 4.15.

The biggest items are:

 - native multipath support (me + Hannes)
 - userspace notifications for AENs (Keith)
 - obey the command effects log page (Keith)

in addition to that lots of fixes from a wide variety of people.

The following changes since commit 7d1dff993252366cc5e21e07b80072d323628363:

  blktrace: fix unlocked registration of tracepoints (2017-11-07 09:42:10 -0700)

are available in the git repository at:

  git://git.infradead.org/nvme.git nvme-4.15

for you to fetch changes up to b119d90862842a055ff78ca54953d430da314808:

  nvme: expose subsys attribute to sysfs (2017-11-10 12:16:19 +0100)

----------------------------------------------------------------
Christoph Hellwig (14):
      nvme: move the dying queue check from cancel to completion
      nvme: always unregister the integrity profile in __nvme_revalidate_disk
      nvme: don't pass struct nvme_ns to nvme_init_integrity
      nvme: don't pass struct nvme_ns to nvme_config_discard
      nvme: set the chunk size before freezing the queue
      nvme: split __nvme_revalidate_disk
      nvme: fix and clarify the check for missing metadata
      nvmet: better data length validation
      nvmet: kill nvmet_inline_bio_init
      nvme: track subsystems
      nvme: introduce a nvme_ns_ids structure
      nvme: track shared namespaces
      nvme: implement multipath access to nvme subsystems
      nvme: also expose the namespace identification sysfs files for mpath nodes

Hannes Reinecke (3):
      block: create 'slaves' and 'holders' entries for hidden gendisks
      nvme: create 'slaves' and 'holders' entries for hidden controllers
      nvme: expose subsys attribute to sysfs

Israel Rukshin (1):
      nvmet-rdma: update queue list during ib_device removal

James Smart (3):
      nvme-fc: fix localport resume using stale values
      nvme-fc: decouple ns references from lldd references
      lpfc: tie in to new dev_loss_tmo interface in nvme transport

Javier Gonz�lez (2):
      nvme: compare NQN string with right size
      nvme: fix eui_show() print format

Keith Busch (7):
      nvme: factor get log into a helper
      nvme: check admin passthru command effects
      nvme: centralize AEN defines
      nvme-fc: remove unused "queue_size" field
      nvme: remove handling of multiple AEN requests
      nvme: unexport starting async event work
      nvme: send uevent for some asynchronous events

Max Gurtovoy (1):
      nvme-rdma: fix nvme_rdma_create_queue_ib error flow

Ming Lei (1):
      nvme-pci: avoid dereference of symbol from unloaded module

Minwoo Im (1):
      nvmet: fix comment typos in admin-cmd.c

Sagi Grimberg (2):
      nvmet: remove redundant memset if failed to get_smart_log failed
      nvmet: remove redundant local variable

 block/genhd.c                   |   14 +-
 drivers/nvme/host/Kconfig       |    9 +
 drivers/nvme/host/Makefile      |    1 +
 drivers/nvme/host/core.c        | 1050 ++++++++++++++++++++++++++++++---------
 drivers/nvme/host/fabrics.c     |    4 +-
 drivers/nvme/host/fc.c          |  143 ++++--
 drivers/nvme/host/lightnvm.c    |   14 +-
 drivers/nvme/host/multipath.c   |  291 +++++++++++
 drivers/nvme/host/nvme.h        |  138 ++++-
 drivers/nvme/host/pci.c         |   21 +-
 drivers/nvme/host/rdma.c        |   21 +-
 drivers/nvme/target/admin-cmd.c |   21 +-
 drivers/nvme/target/core.c      |   10 +
 drivers/nvme/target/fc.c        |   32 +-
 drivers/nvme/target/io-cmd.c    |   18 +-
 drivers/nvme/target/loop.c      |   19 +-
 drivers/nvme/target/nvmet.h     |    4 +
 drivers/nvme/target/rdma.c      |   16 +-
 drivers/scsi/lpfc/lpfc_attr.c   |    5 +
 include/linux/nvme.h            |   30 ++
 20 files changed, 1456 insertions(+), 405 deletions(-)
 create mode 100644 drivers/nvme/host/multipath.c

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/10/2017 07:34 AM, Christoph Hellwig wrote:
> Hi Jens,
> 
> another round of NVMe updates for 4.15.
> 
> The biggest items are:
> 
>  - native multipath support (me + Hannes)
>  - userspace notifications for AENs (Keith)
>  - obey the command effects log page (Keith)
> 
> in addition to that lots of fixes from a wide variety of people.

Pulled. It had two conflicts with master, but both pretty trivial.

-- 
Jens Axboe

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/10/2017 08:18 AM, Jens Axboe wrote:
> On 11/10/2017 07:34 AM, Christoph Hellwig wrote:
>> Hi Jens,
>>
>> another round of NVMe updates for 4.15.
>>
>> The biggest items are:
>>
>>  - native multipath support (me + Hannes)
>>  - userspace notifications for AENs (Keith)
>>  - obey the command effects log page (Keith)
>>
>> in addition to that lots of fixes from a wide variety of people.
> 
> Pulled. It had two conflicts with master, but both pretty trivial.

Actually, double checking, one of them is a little suspicious:

        if (a == &dev_attr_uuid.attr) {
<<<<<<< HEAD
                if (uuid_is_null(&ns->uuid) &&
                    !memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
=======
                if (uuid_is_null(&ids->uuid) ||
                    !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
>>>>>>> 34a9690712aa386fb5fa9e3b8fb44a22f5f2aec6
                        return 0;
        }

with the previous copy using is_null && memchr_inv, the change to ids
makes it an OR instead. I'm going with the latter, but thought I'd bring
it up.

-- 
Jens Axboe

Re: [GIT PULL] nvme updates for Linux 4.15

[Christoph Hellwig]

On Fri, Nov 10, 2017 at 08:22:32AM -0700, Jens Axboe wrote:
> Actually, double checking, one of them is a little suspicious:
> 
>         if (a == &dev_attr_uuid.attr) {
> <<<<<<< HEAD
>                 if (uuid_is_null(&ns->uuid) &&
>                     !memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
> =======
>                 if (uuid_is_null(&ids->uuid) ||
>                     !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
> >>>>>>> 34a9690712aa386fb5fa9e3b8fb44a22f5f2aec6
>                         return 0;
>         }
> 
> with the previous copy using is_null && memchr_inv, the change to ids
> makes it an OR instead. I'm going with the latter, but thought I'd bring
> it up.

The && from master is the right one, it comes from:

ommit 007a61ae2f35c7fcf767313285c4924e81f11983
Author: Martin Wilck <mwilck@suse.com>
Date:   Thu Sep 28 21:33:23 2017 +0200

    nvme: fix visibility of "uuid" ns attribute

and needs to be logically applied to the new code as well.

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/10/2017 10:22 AM, Christoph Hellwig wrote:
> On Fri, Nov 10, 2017 at 08:22:32AM -0700, Jens Axboe wrote:
>> Actually, double checking, one of them is a little suspicious:
>>
>>         if (a == &dev_attr_uuid.attr) {
>> <<<<<<< HEAD
>>                 if (uuid_is_null(&ns->uuid) &&
>>                     !memchr_inv(ns->nguid, 0, sizeof(ns->nguid)))
>> =======
>>                 if (uuid_is_null(&ids->uuid) ||
>>                     !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
>>>>>>>>> 34a9690712aa386fb5fa9e3b8fb44a22f5f2aec6
>>                         return 0;
>>         }
>>
>> with the previous copy using is_null && memchr_inv, the change to ids
>> makes it an OR instead. I'm going with the latter, but thought I'd bring
>> it up.
> 
> The && from master is the right one, it comes from:
> 
> ommit 007a61ae2f35c7fcf767313285c4924e81f11983
> Author: Martin Wilck <mwilck@suse.com>
> Date:   Thu Sep 28 21:33:23 2017 +0200
> 
>     nvme: fix visibility of "uuid" ns attribute
> 
> and needs to be logically applied to the new code as well.

That makes for a bit of an awkward merge, why wasn't this fixed up
in your tree?

In any case, I just fixed it up in my for-next.

-- 
Jens Axboe

Re: [GIT PULL] nvme updates for Linux 4.15

[Christoph Hellwig]

On Fri, Nov 10, 2017 at 10:27:24AM -0700, Jens Axboe wrote:
> That makes for a bit of an awkward merge, why wasn't this fixed up
> in your tree?

Because you asked me to always base on for-4.15/block last time?

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/10/2017 10:33 AM, Christoph Hellwig wrote:
> On Fri, Nov 10, 2017 at 10:27:24AM -0700, Jens Axboe wrote:
>> That makes for a bit of an awkward merge, why wasn't this fixed up
>> in your tree?
> 
> Because you asked me to always base on for-4.15/block last time?

That's not what I meant. It's conflicting because of a patch, that's
fine. But the code in your tree is:

if (a == &dev_attr_uuid.attr) {
        if (uuid_is_null(&ids->uuid) ||
            !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
                return 0;
}

and you're saying the right resolution is:

if (a == &dev_attr_uuid.attr) {
        if (uuid_is_null(&ids->uuid) ||
            !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
                return 0;
}

My point is that this should have been fixed up in your tree. Yes, it'd
still cause a conflict, that's fine, but at least the merge would not
end up also containing a fix for the above code. That part is not
logical.

-- 
Jens Axboe

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/10/2017 10:38 AM, Jens Axboe wrote:
> On 11/10/2017 10:33 AM, Christoph Hellwig wrote:
>> On Fri, Nov 10, 2017 at 10:27:24AM -0700, Jens Axboe wrote:
>>> That makes for a bit of an awkward merge, why wasn't this fixed up
>>> in your tree?
>>
>> Because you asked me to always base on for-4.15/block last time?
> 
> That's not what I meant. It's conflicting because of a patch, that's
> fine. But the code in your tree is:
> 
> if (a == &dev_attr_uuid.attr) {
>         if (uuid_is_null(&ids->uuid) ||
>             !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
>                 return 0;
> }
> 
> and you're saying the right resolution is:
> 
> if (a == &dev_attr_uuid.attr) {
>         if (uuid_is_null(&ids->uuid) ||
>             !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
>                 return 0;
> }

This one should have had the && instead of course...

-- 
Jens Axboe

Re: [GIT PULL] nvme updates for Linux 4.15

[Christoph Hellwig]

Do you want to apply this forward ported version of the patch
to your for-4.15/block tree?  That should at least make the conflict
go away for Linus when he pulls it.

---
>From 9a687c554af721a7a4d77a82d6d8be4e66c1e1a0 Mon Sep 17 00:00:00 2001
From: Martin Wilck <mwilck@suse.com>
Date: Thu, 28 Sep 2017 21:33:23 +0200
Subject: nvme: fix visibility of "uuid" ns attribute

"uuid" must be invisible if both ns->uuid and ns->nguid are unset,
not if either one is.

Fixes: d934f9848a77 "nvme: provide UUID value to userspace"
Signed-off-by: Martin Wilck <mwilck@suse.com>
[hch: rebased to the nvme-4.15 tree to help resolving a conflict]
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 drivers/nvme/host/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index b4d649039665..592fa6af227f 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -2524,7 +2524,7 @@ static umode_t nvme_ns_id_attrs_are_visible(struct kobject *kobj,
 	struct nvme_ns_ids *ids = &dev_to_ns_head(dev)->ids;
 
 	if (a == &dev_attr_uuid.attr) {
-		if (uuid_is_null(&ids->uuid) ||
+		if (uuid_is_null(&ids->uuid) &&
 		    !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
 			return 0;
 	}
-- 
2.14.2

Re: [GIT PULL] nvme updates for Linux 4.15

[Jens Axboe]

On 11/11/2017 01:43 AM, Christoph Hellwig wrote:
> Do you want to apply this forward ported version of the patch
> to your for-4.15/block tree?  That should at least make the conflict
> go away for Linus when he pulls it.

Yeah I will probably add that. I won't kill the conflict, but it'll
make it more logically solvable.

-- 
Jens Axboe