Hi Jens,
Thanks for your comment!
On Thu, Apr 06, 2017 at 01:29:26PM -0600, Jens Axboe wrote:
> On 04/06/2017 02:23 AM, Ming Lei wrote:
> > On Thu, Apr 06, 2017 at 12:57:51AM -0700, Omar Sandoval wrote:
> >> On Thu, Apr 06, 2017 at 12:31:18PM +0800, Ming Lei wrote:
> >>> On Wed, Apr 05, 2017 at
Currently only dm and md/raid5 bios trigger
trace_block_bio_complete(). Now that we have bio_chain() and
bio_inc_remaining(), it is not possible, in general, for a driver to
know when the bio is really complete. Only bio_endio() knows that.
So move the trace_block_bio_complete() call to
The comment for the 'flags' field of 'bio' mentions
"command" which is no longer stored there, and doesn't
mention the bvec pool number, which is.
BIO_RESET_BITS is set in such a way that it would need to be
updated if new bits were added, which is easy to miss.
BVEC_POOL_BITS is larger than
Hi Jens,
I think all objections to this patch have been answered so I'm
resending.
I've added a small cleanup patch first which makes some small
enhancements to the documentation and #defines for the bio->flags
field.
Thanks,
NeilBrown
---
NeilBrown (2):
block: simple improvements
When a filesystem is mounted from a loop device, writes are
throttled by balance_dirty_pages() twice: once when writing
to the filesystem and once when the loop_handle_cmd() writes
to the backing file. This double-throttling can trigger
positive feedback loops that create significant delays.
On Mon, Apr 03, 2017 at 11:52:11PM -0700, Christoph Hellwig wrote:
> > + if (unaligned_io) {
> > + /* If we are going to wait for other DIO to finish, bail */
> > + if ((iocb->ki_flags & IOCB_NOWAIT) &&
> > +atomic_read(>i_dio_count))
> > +
> On Apr 6, 2017, at 11:39 AM, Christoph Hellwig wrote:
>
> Add a nbd-specific field instead.
>
> Signed-off-by: Christoph Hellwig
This is fine with me, you can add,
Reviewed-by: Josef Bacik
Thanks,
Josef
> On Apr 6, 2017, at 5:01 PM, Josef Bacik wrote:
>
> This patchset adds a new netlink configuration interface to NBD as well as a
> bunch of enhancments around path failures. The patches provide the following
> enhancemnts to NBD
>
> - Netlink configuration interface
In order to support deleting the device on disconnect we need to
refcount the actual nbd_device struct. So add the refcounting framework
and change how we free the normal devices at rmmod time so we can catch
reference leaks.
Signed-off-by: Josef Bacik
---
drivers/block/nbd.c |
For ease of management it would be nice for users to specify that the
device node for a nbd device is destroyed once it is disconnected and
there are no more users. Add a client flag and enable this operation to
happen.
Signed-off-by: Josef Bacik
---
drivers/block/nbd.c |
Sometimes we like to upgrade our server without making all of our
clients freak out and reconnect. This patch provides a way to specify a
dead connection timeout to allow us to pause all requests and wait for
new connections to be opened. With this in place I can take down the
nbd server for
Allow users to query the status of existing nbd devices. Right now this
only returns whether or not the device is connected, but could be
extended in the future to include more information.
Signed-off-by: Josef Bacik
---
drivers/block/nbd.c | 108
This patchset adds a new netlink configuration interface to NBD as well as a
bunch of enhancments around path failures. The patches provide the following
enhancemnts to NBD
- Netlink configuration interface that doesn't leave a userspace application
waiting in kernel space for the device to
The existing ioctl interface for configuring NBD devices is a bit
cumbersome and hard to extend. The other problem is we leave a
userspace app sitting in it's syscall until the device disconnects,
which is less than ideal.
This patch introduces a netlink interface for adding and disconnecting
In preparation for the upcoming netlink interface we need to not rely on
already having the bdev for the NBD device we are doing operations on.
Instead of passing the bdev around, just use it in places where we know
we already have the bdev.
Signed-off-by: Josef Bacik
---
In order to properly refcount the various aspects of a NBD device we
need to separate out the configuration elements of the nbd device. The
configuration of a NBD device has a different lifetime from the actual
device, so it doesn't make sense to bundle these two concepts. Add a
config_refs to
Provide a mechanism to notify userspace that there's been a link problem
on a NBD device. This will allow userspace to re-establish a connection
and provide the new socket to the device without disrupting the device.
Signed-off-by: Josef Bacik
---
drivers/block/nbd.c
When adding a new socket we look it up and then try to add it to our
configuration. If any of those steps fail we need to make sure we put
the socket so we don't leak them.
Signed-off-by: Josef Bacik
---
drivers/block/nbd.c | 9 +++--
1 file changed, 7 insertions(+), 2
When running a disconnect torture test I noticed that sometimes we would
crash with a negative ref count on our queue. This was because we were
ending the same request twice. Turns out we were racing with
NBD_CLEAR_SOCK clearing the requests as well as the teardown of the
device clearing the
Currently if we have multiple connections and one of them goes down we will tear
down the whole device. However there's no reason we need to do this as we
could have other connections that are working fine. Deal with this by keeping
track of the state of the different connections, and if we lose
On Thu, Apr 06, 2017 at 05:39:19PM +0200, Christoph Hellwig wrote:
> Currently the request structure has an errors field that is used in
> various different ways. The oldest drivers use it as an error count,
> blk-mq and the generic timeout code assume that it holds a Linux
> errno for block
> Il giorno 04 apr 2017, alle ore 17:28, Bart Van Assche
> ha scritto:
>
> On Tue, 2017-04-04 at 12:42 +0200, Paolo Valente wrote:
>>> Il giorno 31 mar 2017, alle ore 17:31, Bart Van Assche
>>> ha scritto:
>>>
>>> On Fri, 2017-03-31 at
On 04/06/2017 02:23 AM, Ming Lei wrote:
> On Thu, Apr 06, 2017 at 12:57:51AM -0700, Omar Sandoval wrote:
>> On Thu, Apr 06, 2017 at 12:31:18PM +0800, Ming Lei wrote:
>>> On Wed, Apr 05, 2017 at 12:01:29PM -0700, Omar Sandoval wrote:
From: Omar Sandoval
While
On 04/06/2017 12:10 PM, Bart Van Assche wrote:
> + for (i = 0; i < queue->nr_hw_queues; i++) {
> + j = (i + hctx->queue_num + 1) % queue->nr_hw_queues;
> + h = queue->queue_hw_ctx[j];
> + if (h->tags == tags &&
On Thu, Apr 06, 2017 at 05:39:26PM +0200, Christoph Hellwig wrote:
> Remove passing req->errors (which at that point is always 0) to
> blk_mq_complete_requestq, and rely on the virtio status code for the
blk_mq_complete_request ^
> serial number passthrough request.
>
> Signed-off-by: Christoph
On Thu, Apr 06, 2017 at 05:39:25PM +0200, Christoph Hellwig wrote:
> Signed-off-by: Christoph Hellwig
> ---
Fair enough,
Reviewed-by: Johannes Thumshirn
--
Johannes Thumshirn Storage
jthumsh...@suse.de
On Thu, Apr 06, 2017 at 05:39:22PM +0200, Christoph Hellwig wrote:
> nvme_complete_async_event expects the little endian status code
> including the phase bit, and a new completion handler I plan to
> introduce will do so as well.
>
> Change the status variable into the little endian format with
On Thu, Apr 06, 2017 at 05:39:23PM +0200, Christoph Hellwig wrote:
> We want our own clearly defined error field for NVMe passthrough commands,
> and the request errors field is going away in its current form.
>
> Just store the status and result field in the nvme_request field from
> hardirq
On Thu, Apr 06, 2017 at 05:39:21PM +0200, Christoph Hellwig wrote:
> The function only returns -EIO if rq->errors is non-zero, which is not
> very useful and lets a large number of callers ignore the return value.
>
> Just let the callers figure out their error themselves.
>
> Signed-off-by:
On Thu, Apr 06, 2017 at 12:19:46PM -0600, Jens Axboe wrote:
> We share that goal. The block related tests will have _fewer_ dependencies.
> And it'll also be Linux only, so we don't have to worry about cross platform.
xfstests effectively has been Linux-only for many years. We've just
been to
On 04/06/2017 12:15 PM, Johannes Thumshirn wrote:
> On Thu, Apr 06, 2017 at 10:32:43AM -0500, Eric Sandeen wrote:
>> Full-blown xfstests doesn't have /that/ much in the way of deps,
>> on debian it's ostensibly just:
>>
>> sudo apt-get install xfslibs-dev uuid-dev libtool-bin \
>>
On Thu, Apr 06, 2017 at 10:32:43AM -0500, Eric Sandeen wrote:
> Full-blown xfstests doesn't have /that/ much in the way of deps,
> on debian it's ostensibly just:
>
> sudo apt-get install xfslibs-dev uuid-dev libtool-bin \
> e2fsprogs automake gcc libuuid1 quota
To improve scalability, if hardware queues are shared, restart
a single hardware queue in round-robin fashion. Rename
blk_mq_sched_restart_queues() to reflect the new semantics.
Remove blk_mq_sched_mark_restart_queue() because this function
has no callers. Remove flag QUEUE_FLAG_RESTART because
Introduce a function that runs a hardware queue unconditionally
after a delay. Note: there is already a function that stops and
restarts a hardware queue after a delay, namely blk_mq_delay_queue().
Signed-off-by: Bart Van Assche
Cc: Christoph Hellwig
Cc:
commit 52d7f1b5c2f3 ("blk-mq: Avoid that requeueing starts stopped
queues") removed the blk_mq_stop_hw_queue() call from scsi_queue_rq()
for the BLK_MQ_RQ_QUEUE_BUSY case. Hence change all calls to functions
that are intended to rerun a busy queue such that these examine all
hardware queues
The blk_mq_dispatch_rq_list() implementation got modified several
times but the comments in that function were not updated every
time. Since it is nontrivial what is going on, update the comments
in blk_mq_dispatch_rq_list().
Signed-off-by: Bart Van Assche
Cc: Omar
Since the next patch in this series will use RCU to iterate over
tag_list, make this safe. Add lockdep_assert_held() statements
in functions that iterate over tag_list to make clear that using
list_for_each_entry() instead of list_for_each_entry_rcu() is
fine in these functions.
Signed-off-by:
Hello Jens,
The five patches in this patch series fix the queue lockup I reported
recently on the linux-block mailing list. Please consider these patches
for inclusion in the upstream kernel.
Thanks,
Bart.
Changes between v2 and v3:
- Removed the blk_mq_ops.restart_hctx function pointer again.
On Thu, Apr 06, 2017 at 10:52:01AM +0200, Hannes Reinecke wrote:
> On 04/06/2017 10:19 AM, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > The request queue created when we create a loop device has the default
> > logical block size of 512. When we associate the device with
Signed-off-by: Christoph Hellwig
---
drivers/block/ataflop.c | 12 +++-
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c
index 2104b1b4ccda..fa69ecd52cb5 100644
--- a/drivers/block/ataflop.c
+++
This is for the legacy floppy and ataflop drivers that currently abuse
->errors for this purpose. It's stashed away in a union to not grow
the struct size, the other fields are either used by modern drivers
for different purposes or the I/O scheduler before queing the I/O
to drivers.
Signed-off-by: Christoph Hellwig
---
block/blk-core.c | 14 +-
block/blk-exec.c | 3 +--
block/blk-mq.c | 10 +++---
block/blk-timeout.c | 1 -
include/linux/blkdev.h | 2 --
include/trace/events/block.h | 17
xen-blkfron is the last users using rq->errros for passing back error to
blk-mq, and I'd like to get rid of that. In the longer run the driver
should be moving more of the completion processing into .complete, but
this is the minimal change to move forward for now.
Signed-off-by: Christoph
Signed-off-by: Christoph Hellwig
---
include/trace/events/block.h | 44 ++--
kernel/trace/blktrace.c | 9 -
2 files changed, 10 insertions(+), 43 deletions(-)
diff --git a/include/trace/events/block.h
Signed-off-by: Christoph Hellwig
---
drivers/block/floppy.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c
index ce102ec47ef2..60d4c7653178 100644
--- a/drivers/block/floppy.c
+++ b/drivers/block/floppy.c
@@
The driver never sets req->errors
---
drivers/block/paride/pd.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c
index 82c6d02193ae..3b0ab214fe74 100644
--- a/drivers/block/paride/pd.c
+++ b/drivers/block/paride/pd.c
@@
Now that we always have a ->complete callback we can remove the direct
call to blk_mq_end_request, as well as the error argument to
blk_mq_complete_request.
Signed-off-by: Christoph Hellwig
---
block/blk-mq.c| 14 +++---
drivers/block/loop.c
Signed-off-by: Christoph Hellwig
---
drivers/block/swim3.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/block/swim3.c b/drivers/block/swim3.c
index 61b3ffa4f458..ba4809c9bdba 100644
--- a/drivers/block/swim3.c
+++ b/drivers/block/swim3.c
@@ -343,8
Instead of using req->errors, which will go away.
Signed-off-by: Christoph Hellwig
---
drivers/block/mtip32xx/mtip32xx.c | 16 +---
drivers/block/mtip32xx/mtip32xx.h | 1 +
2 files changed, 10 insertions(+), 7 deletions(-)
diff --git
Merge blk_mq_ipi_complete_request and blk_mq_stat_add into their only
caller.
Signed-off-by: Christoph Hellwig
---
block/blk-mq.c | 21 ++---
1 file changed, 6 insertions(+), 15 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index
This passes on the scsi_cmnd result field to users of passthrough
requests. Currently we abuse req->errors for this purpose, but that
field will go away in its current form.
Note that the old IDE code abuses the errors field in very creative
ways and stores all kinds of different values in it.
dm never uses rq->errors, so there is no need to pass an error argument
to blk_mq_complete_request.
Signed-off-by: Christoph Hellwig
---
drivers/md/dm-rq.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/md/dm-rq.c b/drivers/md/dm-rq.c
index
We'll get all proper errors reported through ->end_io and ->errors will
go away soon.
Signed-off-by: Christoph Hellwig
---
drivers/md/dm-mpath.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index
In thruth I've just audited which blk-mq drivers don't currently have a
complete callback, but I think this change is at least borderline useful.
Signed-off-by: Christoph Hellwig
---
drivers/block/loop.c | 30 ++
drivers/block/loop.h | 1 +
2 files
Signed-off-by: Christoph Hellwig
---
drivers/block/null_blk.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c
index f93906ff31e8..24ca85a70fd8 100644
--- a/drivers/block/null_blk.c
+++ b/drivers/block/null_blk.c
Remove passing req->errors (which at that point is always 0) to
blk_mq_complete_requestq, and rely on the virtio status code for the
serial number passthrough request.
Signed-off-by: Christoph Hellwig
---
drivers/block/virtio_blk.c | 10 +++---
1 file changed, 3 insertions(+),
nvme_complete_async_event expects the little endian status code
including the phase bit, and a new completion handler I plan to
introduce will do so as well.
Change the status variable into the little endian format with the
phase bit used in the NVMe CQE to fix / enable this.
Signed-off-by:
Currently the request structure has an errors field that is used in
various different ways. The oldest drivers use it as an error count,
blk-mq and the generic timeout code assume that it holds a Linux
errno for block completions, and various drivers use it for internal
status values, often
Signed-off-by: Christoph Hellwig
---
drivers/block/virtio_blk.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index eaf99022bdc6..dbc4e80680b1 100644
--- a/drivers/block/virtio_blk.c
+++
Currently it's used by the lighnvm passthrough ioctl, but we'd like to make
it private in preparation of block layer specific error code. Lighnvm already
returns the real NVMe status anyway, so I think we can just limit it to
returning -EIO for any status set.
This will need a careful audit from
Currently error is always 0 for non-passthrough requests when reaching the
scsi_noretry_cmd check in scsi_io_completion, which effectively disables
all fastfail logic. Fix this by having a single call to
__scsi_error_from_host_byte at the beginning of the function and always
having a valid error
The function only returns -EIO if rq->errors is non-zero, which is not
very useful and lets a large number of callers ignore the return value.
Just let the callers figure out their error themselves.
Signed-off-by: Christoph Hellwig
---
block/blk-exec.c | 8 +---
We want our own clearly defined error field for NVMe passthrough commands,
and the request errors field is going away in its current form.
Just store the status and result field in the nvme_request field from
hardirq completion context (using a new helper) and then generate a
Linux errno for the
This drivers was added in 2008, but as far as a I can tell we never had a
single platform that actually registered resources for the platform driver.
It's also been unmaintained for a long time and apparently has a ATA mode
that can be driven using the IDE/libata subsystem.
Signed-off-by:
On 4/6/17 10:20 AM, Johannes Thumshirn wrote:
> On Thu, Apr 06, 2017 at 08:51:36AM -0600, Jens Axboe wrote:
>> On 04/06/2017 08:47 AM, Christoph Hellwig wrote:
>>> On Thu, Apr 06, 2017 at 08:33:55AM -0600, Jens Axboe wrote:
That is exactly what my recommendation was at lsfmm as well - fork
On Thu, Apr 06, 2017 at 08:51:36AM -0600, Jens Axboe wrote:
> On 04/06/2017 08:47 AM, Christoph Hellwig wrote:
> > On Thu, Apr 06, 2017 at 08:33:55AM -0600, Jens Axboe wrote:
> >> That is exactly what my recommendation was at lsfmm as well - fork
> >> xfstest, prune bits we don't need, and off we
On Thu, Apr 06, 2017 at 02:21:56PM +0200, Hannes Reinecke wrote:
> On 04/06/2017 02:03 PM, Ming Lei wrote:
> > On Thu, Apr 06, 2017 at 04:43:18PM +0800, Ming Lei wrote:
> >> On Thu, Apr 06, 2017 at 01:19:45AM -0700, Omar Sandoval wrote:
> >>> From: Omar Sandoval
> >>>
> >>> The
On 04/06/2017 08:47 AM, Christoph Hellwig wrote:
> On Thu, Apr 06, 2017 at 08:33:55AM -0600, Jens Axboe wrote:
>> That is exactly what my recommendation was at lsfmm as well - fork
>> xfstest, prune bits we don't need, and off we go. I'll get around to
>> it soonish.
>
> So you volunteer to do
On Thu, Apr 06, 2017 at 08:33:55AM -0600, Jens Axboe wrote:
> That is exactly what my recommendation was at lsfmm as well - fork
> xfstest, prune bits we don't need, and off we go. I'll get around to
> it soonish.
So you volunteer to do it? I was -><- this close to offering it myself,
and than
On 04/06/2017 07:55 AM, Christoph Hellwig wrote:
> I still disagree with using xfstests for this. Even if we were going
> to treat the block devices nodes as yet another file system that's not
> what the patches do - they create specific virtual devices to test
> for each test.
>
> I think the
On Thu, Apr 06, 2017 at 01:36:36PM +0300, Sagi Grimberg wrote:
> Simply refer to the generic affinity mask helper.
>
> Reviewed-by: Christoph Hellwig
> Signed-off-by: Sagi Grimberg
> ---
Thanks,
Acked-by: Leon Romanovsky
signature.asc
On Thu, Apr 06, 2017 at 01:36:34PM +0300, Sagi Grimberg wrote:
> generic api takes care of spreading affinity similar to
> what mlx5 open coded (and even handles better asymmetric
> configurations). Ask the generic API to spread affinity
> for us, and feed him pre_vectors that do not participate
>
On Thu, Apr 06, 2017 at 01:36:33PM +0300, Sagi Grimberg wrote:
> Now that we have a generic code to allocate an array
> of irq vectors and even correctly spread their affinity,
> correctly handle cpu hotplug events and more, were much
> better off using it.
>
> Reviewed-by: Christoph Hellwig
I still disagree with using xfstests for this. Even if we were going
to treat the block devices nodes as yet another file system that's not
what the patches do - they create specific virtual devices to test
for each test.
I think the right way is to keep your patches as-is and copy the few
bits
same as blockdev/004 but with add more stress on corrupted fs.
Signed-off-by: Dmitry Monakhov
---
tests/blockdev/005 | 92 ++
tests/blockdev/005.out | 2 ++
tests/blockdev/group | 1 +
tests/generic/390 | 1 +
4
Signed-off-by: Dmitry Monakhov
---
tests/blockdev/004 | 81 ++
tests/blockdev/004.out | 2 ++
tests/blockdev/group | 1 +
3 files changed, 84 insertions(+)
create mode 100755 tests/blockdev/004
create mode 100644
Regression test for fix https://lkml.org/lkml/2017/3/22/789
Page cache should be dropped after successful discard
Signed-off-by: Dmitry Monakhov
---
common/config | 1 +
tests/blockdev/001 | 74 ++
During LSFMM we have discussed how to test lower-backend of linux IO-stack.
Main idea was that xfstests is the most obvious solution which cover
most use-cases filesystem do care about. There were reasonable objections
that xfstests is not ideal place for that, especially when they mixed
with pure
Test create virtual block device via lio-targed infastructure and
perform basic IO operations with data corruption detection.
Temprorally mark is as dangerous, because currently it trigger BUG_ON
inside blkdev_issue_flush
BTW: I use 'dd' to test read from corrupted image instead of xfs_io
Linux-IO Target is very good framework for testing block backend.
It is more flexible than scsi_debug.
http://linux-iscsi.org/wiki/LIO
Signed-off-by: Dmitry Monakhov
---
common/config| 2 +
common/liotarget | 111
On 04/06/2017 02:03 PM, Ming Lei wrote:
> On Thu, Apr 06, 2017 at 04:43:18PM +0800, Ming Lei wrote:
>> On Thu, Apr 06, 2017 at 01:19:45AM -0700, Omar Sandoval wrote:
>>> From: Omar Sandoval
>>>
>>> The request queue created when we create a loop device has the default
>>> logical
On Thu, Apr 06, 2017 at 04:43:18PM +0800, Ming Lei wrote:
> On Thu, Apr 06, 2017 at 01:19:45AM -0700, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > The request queue created when we create a loop device has the default
> > logical block size of 512. When we associate the
Stephen added the first two includes after a linux-next build
failure. Did this go through a long buildbot cycle to verify
it's really not needed anymore?
Umm no. I just compiled it and ran and it worked...
If these are actually needed then we can safely discard
this patch.
Stephen added the first two includes after a linux-next build
failure. Did this go through a long buildbot cycle to verify
it's really not needed anymore?
On Thu, Apr 06, 2017 at 01:33:36PM +0300, Sagi Grimberg wrote:
> Signed-off-by: Sagi Grimberg
> ---
> block/blk-mq-pci.c
Looks good,
Reviewed-by: Christoph Hellwig
Looks good,
Reviewed-by: Christoph Hellwig
On Tue, Apr 04, 2017 at 10:56:40PM +0400, Dmitry Monakhov wrote:
> Some ->bi_end_io handlers (for example: pi_verify or decrypt handlers)
Just curious: are you planning to use it for the latter as well in
future patches? If not can you
> if (bio_no_advance_iter(bio))
> iter->bi_size -= bytes;
> - else
> - bvec_iter_advance(bio->bi_io_vec, iter, bytes);
> + else {
> + int err;
> + err = bvec_iter_advance(bio->bi_io_vec, iter, bytes);
> + if
Looks good,
Reviewed-by: Christoph Hellwig
> --- a/Documentation/block/data-integrity.txt
> +++ b/Documentation/block/data-integrity.txt
> @@ -202,9 +202,6 @@ will require extra work due to the application tag.
>added. It is up to the caller to ensure that the bio does not
>change while I/O is in progress.
>
> -
Please drop the v2 from the subject line and move your changelog either
into the cover letter (preferred) or at least below the --- line.
Otherwise looks good:
Reviewed-by: Christoph Hellwig
Looks good,
Reviewed-by: Christoph Hellwig
On Thu, Apr 06, 2017 at 08:38:10AM +0200, Wouter Verhelst wrote:
> On Wed, Apr 05, 2017 at 01:30:31PM +0200, Michal Hocko wrote:
> > On Wed 05-04-17 09:46:59, Vlastimil Babka wrote:
> > > We now have memalloc_noreclaim_{save,restore} helpers for robust setting
> > > and
> > > clearing of
This drivers was added in 2008, but as far as a I can tell we never had a
single platform that actually registered resources for the platform driver.
It's also been unmaintained for a long time and apparently has a ATA mode
that can be driven using the IDE/libata subsystem.
Signed-off-by:
Use the generic block layer affinity mapping helper. Also,
limit nr_hw_queues to the rdma device number of irq vectors
as we don't really need more.
Reviewed-by: Christoph Hellwig
Signed-off-by: Sagi Grimberg
---
drivers/nvme/host/rdma.c | 29
Like pci and virtio, we add a rdma helper for affinity
spreading. This achieves optimal mq affinity assignments
according to the underlying rdma device affinity maps.
Reviewed-by: Jens Axboe
Reviewed-by: Christoph Hellwig
Reviewed-by: Max Gurtovoy
This patch set is aiming to automatically find the optimal
queue <-> irq multi-queue assignments in storage ULPs (demonstrated
on nvme-rdma) based on the underlying rdma device irq affinity
settings.
Changes from v1:
- Removed mlx5e_get_cpu as Christoph suggested
- Fixed up nvme-rdma queue
Now that we have a generic code to allocate an array
of irq vectors and even correctly spread their affinity,
correctly handle cpu hotplug events and more, were much
better off using it.
Reviewed-by: Christoph Hellwig
Signed-off-by: Sagi Grimberg
---
Signed-off-by: Sagi Grimberg
---
block/blk-mq-pci.c | 3 ---
1 file changed, 3 deletions(-)
diff --git a/block/blk-mq-pci.c b/block/blk-mq-pci.c
index 0c3354cf3552..70b8a6765cc7 100644
--- a/block/blk-mq-pci.c
+++ b/block/blk-mq-pci.c
@@ -10,12 +10,9 @@
* FITNESS FOR A
shouldn't you include and like in
commit 8ec2ef2b66ea2f that fixes blk-mq-pci.c ?
Not really. We can lose these from blk-mq-pci.c as well.
+#include
+#include
+#include
+#include
+#include "blk-mq.h"
Is this include needed ?
You're right, I can just keep:
+#include
+#include
On Thu, Apr 06, 2017 at 01:19:45AM -0700, Omar Sandoval wrote:
> From: Omar Sandoval
>
> The request queue created when we create a loop device has the default
> logical block size of 512. When we associate the device with an fd, we
> set the block size on the block_device but
1 - 100 of 117 matches
Mail list logo