* [PATCH 2/3] scsi: Add scsi_restart_queues()
From: Bart Van Assche @ 2017-03-31 23:12 UTC (permalink / raw)
To: Jens Axboe
Cc: linux-block, Martin K . Petersen, James Bottomley,
Bart Van Assche, Christoph Hellwig, Hannes Reinecke
In-Reply-To: <20170331231205.16640-1-bart.vanassche@sandisk.com>
This patch avoids that if multiple SCSI devices are associated with
a SCSI host that a queue can get stuck if scsi_queue_rq() returns
"busy".
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
---
drivers/scsi/scsi_lib.c | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index c1519660824b..1d804e33971a 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -555,6 +555,21 @@ void scsi_run_host_queues(struct Scsi_Host *shost)
scsi_run_queue(sdev->request_queue);
}
+static void scsi_restart_queues(struct request_queue *q)
+{
+ struct scsi_device *sdev = q->queuedata;
+ struct Scsi_Host *shost = sdev->host;
+ unsigned long flags;
+
+ spin_lock_irqsave(shost->host_lock, flags);
+ __shost_for_each_device(sdev, shost) {
+ q = sdev->request_queue;
+ if (q->mq_ops && !blk_queue_dying(q))
+ blk_mq_run_hw_queues(q, true);
+ }
+ spin_unlock_irqrestore(shost->host_lock, flags);
+}
+
static void scsi_uninit_cmd(struct scsi_cmnd *cmd)
{
if (!blk_rq_is_passthrough(cmd->request)) {
@@ -2156,6 +2171,7 @@ struct request_queue *scsi_alloc_queue(struct scsi_device *sdev)
static const struct blk_mq_ops scsi_mq_ops = {
.queue_rq = scsi_queue_rq,
+ .restart_queues = scsi_restart_queues,
.complete = scsi_softirq_done,
.timeout = scsi_timeout,
.init_request = scsi_init_request,
--
2.12.0
^ permalink raw reply related
* [PATCH 1/3] blk-mq: Introduce blk_mq_ops.restart_queues
From: Bart Van Assche @ 2017-03-31 23:12 UTC (permalink / raw)
To: Jens Axboe
Cc: linux-block, Martin K . Petersen, James Bottomley,
Bart Van Assche, Christoph Hellwig, Hannes Reinecke
In-Reply-To: <20170331231205.16640-1-bart.vanassche@sandisk.com>
If a tag set is shared among multiple request queues, leave
it to the block driver to restart queues. Hence remove
QUEUE_FLAG_RESTART and introduce blk_mq_ops.restart_queues.
Remove blk_mq_sched_mark_restart_queue() because this
function has no callers.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
---
block/blk-mq-sched.c | 11 +++--------
block/blk-mq-sched.h | 14 --------------
include/linux/blk-mq.h | 4 ++++
include/linux/blkdev.h | 1 -
4 files changed, 7 insertions(+), 23 deletions(-)
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 09af8ff18719..ed9942ee0146 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -329,16 +329,11 @@ static void blk_mq_sched_restart_hctx(struct blk_mq_hw_ctx *hctx)
void blk_mq_sched_restart_queues(struct blk_mq_hw_ctx *hctx)
{
struct request_queue *q = hctx->queue;
- unsigned int i;
- if (test_bit(QUEUE_FLAG_RESTART, &q->queue_flags)) {
- if (test_and_clear_bit(QUEUE_FLAG_RESTART, &q->queue_flags)) {
- queue_for_each_hw_ctx(q, hctx, i)
- blk_mq_sched_restart_hctx(hctx);
- }
- } else {
+ if (q->mq_ops->restart_queues)
+ q->mq_ops->restart_queues(q);
+ else
blk_mq_sched_restart_hctx(hctx);
- }
}
/*
diff --git a/block/blk-mq-sched.h b/block/blk-mq-sched.h
index a75b16b123f7..fe62b1eccf4c 100644
--- a/block/blk-mq-sched.h
+++ b/block/blk-mq-sched.h
@@ -131,20 +131,6 @@ static inline void blk_mq_sched_mark_restart_hctx(struct blk_mq_hw_ctx *hctx)
set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
}
-/*
- * Mark a hardware queue and the request queue it belongs to as needing a
- * restart.
- */
-static inline void blk_mq_sched_mark_restart_queue(struct blk_mq_hw_ctx *hctx)
-{
- struct request_queue *q = hctx->queue;
-
- if (!test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
- set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
- if (!test_bit(QUEUE_FLAG_RESTART, &q->queue_flags))
- set_bit(QUEUE_FLAG_RESTART, &q->queue_flags);
-}
-
static inline bool blk_mq_sched_needs_restart(struct blk_mq_hw_ctx *hctx)
{
return test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index ea2e9dcd3aef..75160e9470cb 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -86,6 +86,7 @@ struct blk_mq_queue_data {
};
typedef int (queue_rq_fn)(struct blk_mq_hw_ctx *, const struct blk_mq_queue_data *);
+typedef void (restart_fn)(struct request_queue *q);
typedef enum blk_eh_timer_return (timeout_fn)(struct request *, bool);
typedef int (init_hctx_fn)(struct blk_mq_hw_ctx *, void *, unsigned int);
typedef void (exit_hctx_fn)(struct blk_mq_hw_ctx *, unsigned int);
@@ -108,6 +109,9 @@ struct blk_mq_ops {
*/
queue_rq_fn *queue_rq;
+ /* Called upon request completion to rerun all queues */
+ restart_fn *restart_queues;
+
/*
* Called on request timeout
*/
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index a2dc6b390d48..a80543ec8be7 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -615,7 +615,6 @@ struct request_queue {
#define QUEUE_FLAG_FLUSH_NQ 25 /* flush not queueuable */
#define QUEUE_FLAG_DAX 26 /* device supports DAX */
#define QUEUE_FLAG_STATS 27 /* track rq completion times */
-#define QUEUE_FLAG_RESTART 28 /* queue needs restart at completion */
#define QUEUE_FLAG_POLL_STATS 29 /* collecting stats for hybrid polling */
#define QUEUE_FLAG_REGISTERED 30 /* queue has been registered to a disk */
--
2.12.0
^ permalink raw reply related
* [PATCH 0/3] Avoid that scsi-mq queue processing stalls
From: Bart Van Assche @ 2017-03-31 23:12 UTC (permalink / raw)
To: Jens Axboe
Cc: linux-block, Martin K . Petersen, James Bottomley,
Bart Van Assche
Hello Jens,
The three patches in this patch series fix the queue lockup I reported
a few days ago on the linux-block mailing list. Please consider these
patches for kernel v4.11.
Thanks,
Bart.
Bart Van Assche (3):
blk-mq: Introduce blk_mq_ops.restart_queues
scsi: Add scsi_restart_queues()
scsi: Ensure that scsi_run_queue() runs all hardware queues
block/blk-mq-sched.c | 11 +++--------
block/blk-mq-sched.h | 14 --------------
drivers/scsi/scsi_lib.c | 20 +++++++++++++++++---
include/linux/blk-mq.h | 4 ++++
include/linux/blkdev.h | 1 -
5 files changed, 24 insertions(+), 26 deletions(-)
--
2.12.0
^ permalink raw reply
* Re: [PATCH 6/8] bio-integrity: add bio_integrity_setup helper
From: kbuild test robot @ 2017-03-31 22:15 UTC (permalink / raw)
To: Dmitry Monakhov
Cc: kbuild-all, linux-kernel, linux-block, martin.petersen,
Dmitry Monakhov
In-Reply-To: <1490881776-28735-7-git-send-email-dmonakhov@openvz.org>
[-- Attachment #1: Type: text/plain, Size: 1677 bytes --]
Hi Dmitry,
[auto build test ERROR on linus/master]
[also build test ERROR on v4.11-rc4]
[cannot apply to block/for-next next-20170331]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Dmitry-Monakhov/block-T10-DIF-Fixes-and-cleanups/20170401-043532
config: sparc64-defconfig (attached as .config)
compiler: sparc64-linux-gnu-gcc (Debian 6.1.1-9) 6.1.1 20160705
reproduce:
wget https://raw.githubusercontent.com/01org/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
make.cross ARCH=sparc64
All errors (new ones prefixed by >>):
In file included from include/linux/blkdev.h:20:0,
from include/linux/backing-dev.h:14,
from include/linux/nfs_fs_sb.h:5,
from include/linux/nfs_fs.h:37,
from arch/sparc/kernel/sys_sparc32.c:24:
>> include/linux/bio.h:788:12: error: 'bio_integrity_setup' defined but not used [-Werror=unused-function]
static int bio_integrity_setup(struct bio *bio)
^~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
vim +/bio_integrity_setup +788 include/linux/bio.h
782
783 static inline int bio_integrity_prep(struct bio *bio)
784 {
785 return 0;
786 }
787
> 788 static int bio_integrity_setup(struct bio *bio)
789 {
790 return 0;
791 }
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 17530 bytes --]
^ permalink raw reply
* Re: [PATCH 7/8] T10: Move opencoded contants to common header
From: kbuild test robot @ 2017-03-31 22:09 UTC (permalink / raw)
To: Dmitry Monakhov
Cc: kbuild-all, linux-kernel, linux-block, martin.petersen,
Dmitry Monakhov
In-Reply-To: <1490881776-28735-8-git-send-email-dmonakhov@openvz.org>
[-- Attachment #1: Type: text/plain, Size: 2250 bytes --]
Hi Dmitry,
[auto build test ERROR on linus/master]
[also build test ERROR on v4.11-rc4]
[cannot apply to block/for-next next-20170331]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Dmitry-Monakhov/block-T10-DIF-Fixes-and-cleanups/20170401-043532
config: x86_64-kexec (attached as .config)
compiler: gcc-6 (Debian 6.2.0-3) 6.2.0 20160901
reproduce:
# save the attached .config to linux build tree
make ARCH=x86_64
All errors (new ones prefixed by >>):
drivers/scsi//qla2xxx/qla_isr.c: In function 'qla2x00_handle_dif_error':
>> drivers/scsi//qla2xxx/qla_isr.c:1953:20: error: 'T10_APP_TAG' undeclared (first use in this function)
if ((a_app_tag == T10_APP_TAG) &&
^~~~~~~~~~~
drivers/scsi//qla2xxx/qla_isr.c:1953:20: note: each undeclared identifier is reported only once for each function it appears in
>> drivers/scsi//qla2xxx/qla_isr.c:1955:21: error: 'T10_REF_TAG' undeclared (first use in this function)
(a_ref_tag == T10_REF_TAG))) {
^~~~~~~~~~~
In file included from include/linux/blkdev.h:20:0,
from include/linux/blk-mq.h:4,
from include/scsi/scsi_host.h:10,
from drivers/scsi//qla2xxx/qla_def.h:31,
from drivers/scsi//qla2xxx/qla_isr.c:7:
At top level:
include/linux/bio.h:788:12: warning: 'bio_integrity_setup' defined but not used [-Wunused-function]
static int bio_integrity_setup(struct bio *bio)
^~~~~~~~~~~~~~~~~~~
vim +/T10_APP_TAG +1953 drivers/scsi//qla2xxx/qla_isr.c
1947
1948 /*
1949 * Ignore sector if:
1950 * For type 3: ref & app tag is all 'f's
1951 * For type 0,1,2: app tag is all 'f's
1952 */
> 1953 if ((a_app_tag == T10_APP_TAG) &&
1954 ((scsi_get_prot_type(cmd) != SCSI_PROT_DIF_TYPE3) ||
> 1955 (a_ref_tag == T10_REF_TAG))) {
1956 uint32_t blocks_done, resid;
1957 sector_t lba_s = scsi_get_lba(cmd);
1958
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 25149 bytes --]
^ permalink raw reply
* Re: [PATCH v3] blk-mq: remap queues when adding/removing hardware queues
From: Keith Busch @ 2017-03-31 21:09 UTC (permalink / raw)
To: Omar Sandoval
Cc: Jens Axboe, linux-block, Christoph Hellwig, Josef Bacik,
kernel-team
In-Reply-To: <9753ebd0c51a9d49f110a6d0d00888170905d97a.1490993257.git.osandov@fb.com>
On Fri, Mar 31, 2017 at 01:48:35PM -0700, Omar Sandoval wrote:
> From: Omar Sandoval <osandov@fb.com>
>
> blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
> behavior that drivers expect. However, commit 4e68a011428a changed
> blk_mq_queue_reinit() to not remap queues for the case of CPU
> hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
> queues as well. This breaks, for example, NBD's multi-connection mode,
> leaving the added hardware queues unused. Fix it by making
> blk_mq_update_nr_hw_queues() explicitly remap the queues.
>
> Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
> Signed-off-by: Omar Sandoval <osandov@fb.com>
This looks good to me.
Reviewed-by: Keith Busch <keith.busch@intel.com>
^ permalink raw reply
* [PATCH v3] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 20:48 UTC (permalink / raw)
To: Jens Axboe, linux-block
Cc: Christoph Hellwig, Keith Busch, Josef Bacik, kernel-team
In-Reply-To: <20170331204603.GB11462@vader>
From: Omar Sandoval <osandov@fb.com>
blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
behavior that drivers expect. However, commit 4e68a011428a changed
blk_mq_queue_reinit() to not remap queues for the case of CPU
hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
queues as well. This breaks, for example, NBD's multi-connection mode,
leaving the added hardware queues unused. Fix it by making
blk_mq_update_nr_hw_queues() explicitly remap the queues.
Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
block/blk-mq.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 061fc2cc88d3..dbc0f2f745e3 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2478,6 +2478,14 @@ static int blk_mq_alloc_rq_maps(struct blk_mq_tag_set *set)
return 0;
}
+static int blk_mq_update_queue_map(struct blk_mq_tag_set *set)
+{
+ if (set->ops->map_queues)
+ return set->ops->map_queues(set);
+ else
+ return blk_mq_map_queues(set);
+}
+
/*
* Alloc a tag set to be associated with one or more request queues.
* May fail with EINVAL for various error conditions. May adjust the
@@ -2532,10 +2540,7 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
if (!set->mq_map)
goto out_free_tags;
- if (set->ops->map_queues)
- ret = set->ops->map_queues(set);
- else
- ret = blk_mq_map_queues(set);
+ ret = blk_mq_update_queue_map(set);
if (ret)
goto out_free_mq_map;
@@ -2627,6 +2632,7 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
blk_mq_freeze_queue(q);
set->nr_hw_queues = nr_hw_queues;
+ blk_mq_update_queue_map(set);
list_for_each_entry(q, &set->tag_list, tag_set_list) {
blk_mq_realloc_hw_ctxs(set, q);
blk_mq_queue_reinit(q, cpu_online_mask);
--
2.12.1
^ permalink raw reply related
* Re: [PATCH v2] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 20:46 UTC (permalink / raw)
To: Jens Axboe, linux-block
Cc: Christoph Hellwig, Keith Busch, Josef Bacik, kernel-team
In-Reply-To: <c14ba78ad1898eef31415e86383e3b5a8ad082b9.1490992756.git.osandov@fb.com>
On Fri, Mar 31, 2017 at 01:43:41PM -0700, Omar Sandoval wrote:
> @@ -2634,6 +2640,7 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
>
> list_for_each_entry(q, &set->tag_list, tag_set_list)
> blk_mq_unfreeze_queue(q);
> +
Stupid whitespace damage...
^ permalink raw reply
* [PATCH v2] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 20:43 UTC (permalink / raw)
To: Jens Axboe, linux-block
Cc: Christoph Hellwig, Keith Busch, Josef Bacik, kernel-team
In-Reply-To: <20170331204428.GM20181@localhost.localdomain>
From: Omar Sandoval <osandov@fb.com>
blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
behavior that drivers expect. However, commit 4e68a011428a changed
blk_mq_queue_reinit() to not remap queues for the case of CPU
hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
queues as well. This breaks, for example, NBD's multi-connection mode,
leaving the added hardware queues unused. Fix it by making
blk_mq_update_nr_hw_queues() explicitly remap the queues.
Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
block/blk-mq.c | 15 +++++++++++----
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 061fc2cc88d3..70fa2b0d385a 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2478,6 +2478,14 @@ static int blk_mq_alloc_rq_maps(struct blk_mq_tag_set *set)
return 0;
}
+static int blk_mq_update_queue_map(struct blk_mq_tag_set *set)
+{
+ if (set->ops->map_queues)
+ return set->ops->map_queues(set);
+ else
+ return blk_mq_map_queues(set);
+}
+
/*
* Alloc a tag set to be associated with one or more request queues.
* May fail with EINVAL for various error conditions. May adjust the
@@ -2532,10 +2540,7 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
if (!set->mq_map)
goto out_free_tags;
- if (set->ops->map_queues)
- ret = set->ops->map_queues(set);
- else
- ret = blk_mq_map_queues(set);
+ ret = blk_mq_update_queue_map(set);
if (ret)
goto out_free_mq_map;
@@ -2627,6 +2632,7 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
blk_mq_freeze_queue(q);
set->nr_hw_queues = nr_hw_queues;
+ blk_mq_update_queue_map(set);
list_for_each_entry(q, &set->tag_list, tag_set_list) {
blk_mq_realloc_hw_ctxs(set, q);
blk_mq_queue_reinit(q, cpu_online_mask);
@@ -2634,6 +2640,7 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
list_for_each_entry(q, &set->tag_list, tag_set_list)
blk_mq_unfreeze_queue(q);
+
}
EXPORT_SYMBOL_GPL(blk_mq_update_nr_hw_queues);
--
2.12.1
^ permalink raw reply related
* Re: [PATCH] blk-mq: remap queues when adding/removing hardware queues
From: Keith Busch @ 2017-03-31 20:44 UTC (permalink / raw)
To: Omar Sandoval
Cc: Jens Axboe, linux-block, Christoph Hellwig, Josef Bacik,
kernel-team
In-Reply-To: <20170331203015.GA11462@vader>
On Fri, Mar 31, 2017 at 01:30:15PM -0700, Omar Sandoval wrote:
> On Fri, Mar 31, 2017 at 04:30:44PM -0400, Keith Busch wrote:
> > On Fri, Mar 31, 2017 at 11:59:24AM -0700, Omar Sandoval wrote:
> > > @@ -2629,11 +2639,12 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
> > > set->nr_hw_queues = nr_hw_queues;
> > > list_for_each_entry(q, &set->tag_list, tag_set_list) {
> > > blk_mq_realloc_hw_ctxs(set, q);
> > > - blk_mq_queue_reinit(q, cpu_online_mask);
> > > + blk_mq_queue_reinit(q, cpu_online_mask, true);
> >
> > I think you want to call blk_mq_update_queue_map directly outside this
> > loop rather than for each queue through blk_mq_queue_reinit. We only
> > need to map the queues once per tagset rather than per queue.
>
> Right, thanks, I'll do that. I figure you're the person to ask,
> nvme_add_dev() does want the remap to happen, right?
Yep, nvme may want to change the queue count if you alter either the CPU
topology or some device specific setting to reprovision hardware queues.
^ permalink raw reply
* Re: [PATCH] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 20:30 UTC (permalink / raw)
To: Keith Busch
Cc: Jens Axboe, linux-block, Christoph Hellwig, Josef Bacik,
kernel-team
In-Reply-To: <20170331203044.GL20181@localhost.localdomain>
On Fri, Mar 31, 2017 at 04:30:44PM -0400, Keith Busch wrote:
> On Fri, Mar 31, 2017 at 11:59:24AM -0700, Omar Sandoval wrote:
> > @@ -2629,11 +2639,12 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
> > set->nr_hw_queues = nr_hw_queues;
> > list_for_each_entry(q, &set->tag_list, tag_set_list) {
> > blk_mq_realloc_hw_ctxs(set, q);
> > - blk_mq_queue_reinit(q, cpu_online_mask);
> > + blk_mq_queue_reinit(q, cpu_online_mask, true);
>
> I think you want to call blk_mq_update_queue_map directly outside this
> loop rather than for each queue through blk_mq_queue_reinit. We only
> need to map the queues once per tagset rather than per queue.
Right, thanks, I'll do that. I figure you're the person to ask,
nvme_add_dev() does want the remap to happen, right?
^ permalink raw reply
* Re: [PATCH] blk-mq: remap queues when adding/removing hardware queues
From: Keith Busch @ 2017-03-31 20:30 UTC (permalink / raw)
To: Omar Sandoval
Cc: Jens Axboe, linux-block, Christoph Hellwig, Josef Bacik,
kernel-team
In-Reply-To: <52fb0623a8496808622c718f0f6372d37574dbe1.1490986618.git.osandov@fb.com>
On Fri, Mar 31, 2017 at 11:59:24AM -0700, Omar Sandoval wrote:
> @@ -2629,11 +2639,12 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
> set->nr_hw_queues = nr_hw_queues;
> list_for_each_entry(q, &set->tag_list, tag_set_list) {
> blk_mq_realloc_hw_ctxs(set, q);
> - blk_mq_queue_reinit(q, cpu_online_mask);
> + blk_mq_queue_reinit(q, cpu_online_mask, true);
I think you want to call blk_mq_update_queue_map directly outside this
loop rather than for each queue through blk_mq_queue_reinit. We only
need to map the queues once per tagset rather than per queue.
^ permalink raw reply
* Re: [PATCH] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 19:01 UTC (permalink / raw)
To: Jens Axboe, linux-block
Cc: Christoph Hellwig, Keith Busch, Josef Bacik, kernel-team
In-Reply-To: <52fb0623a8496808622c718f0f6372d37574dbe1.1490986618.git.osandov@fb.com>
On Fri, Mar 31, 2017 at 11:59:24AM -0700, Omar Sandoval wrote:
> From: Omar Sandoval <osandov@fb.com>
>
> blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
> behavior that drivers expect. However, commit 4e68a011428a changed
> blk_mq_queue_reinit() to not remap queues for the case of CPU
> hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
> queues as well. This breaks, for example, NBD's multi-connection mode,
> leaving the added hardware queues unused. Fix it by making
> blk_mq_queue_reinit() optionally remap queues, which we do when updating
> the number of hardware queues but not when hotplugging.
>
> Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
> Signed-off-by: Omar Sandoval <osandov@fb.com>
> ---
> The only callers of blk_mq_update_nr_hw_queues() are nbd and nbd. I *think*
> nbd_dev_add() also wants this remap behavior.
Uh, I meant nbd and nvme, and nvme_dev_add().
^ permalink raw reply
* [PATCH] blk-mq: remap queues when adding/removing hardware queues
From: Omar Sandoval @ 2017-03-31 18:59 UTC (permalink / raw)
To: Jens Axboe, linux-block
Cc: Christoph Hellwig, Keith Busch, Josef Bacik, kernel-team
From: Omar Sandoval <osandov@fb.com>
blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
behavior that drivers expect. However, commit 4e68a011428a changed
blk_mq_queue_reinit() to not remap queues for the case of CPU
hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
queues as well. This breaks, for example, NBD's multi-connection mode,
leaving the added hardware queues unused. Fix it by making
blk_mq_queue_reinit() optionally remap queues, which we do when updating
the number of hardware queues but not when hotplugging.
Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
The only callers of blk_mq_update_nr_hw_queues() are nbd and nbd. I *think*
nbd_dev_add() also wants this remap behavior.
block/blk-mq.c | 25 ++++++++++++++++++-------
1 file changed, 18 insertions(+), 7 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 061fc2cc88d3..1abbf7c83193 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2343,14 +2343,27 @@ void blk_mq_free_queue(struct request_queue *q)
blk_mq_exit_hw_queues(q, set, set->nr_hw_queues);
}
+static int blk_mq_update_queue_map(struct blk_mq_tag_set *set)
+{
+ if (set->ops->map_queues)
+ return set->ops->map_queues(set);
+ else
+ return blk_mq_map_queues(set);
+}
+
+
/* Basically redo blk_mq_init_queue with queue frozen */
static void blk_mq_queue_reinit(struct request_queue *q,
- const struct cpumask *online_mask)
+ const struct cpumask *online_mask,
+ bool remap_queues)
{
WARN_ON_ONCE(!atomic_read(&q->mq_freeze_depth));
blk_mq_sysfs_unregister(q);
+ if (remap_queues)
+ blk_mq_update_queue_map(q->tag_set);
+
/*
* redo blk_mq_init_cpu_queues and blk_mq_init_hw_queues. FIXME: maybe
* we should change hctx numa_node according to new topology (this
@@ -2387,7 +2400,7 @@ static void blk_mq_queue_reinit_work(void)
blk_mq_freeze_queue_wait(q);
list_for_each_entry(q, &all_q_list, all_q_node)
- blk_mq_queue_reinit(q, &cpuhp_online_new);
+ blk_mq_queue_reinit(q, &cpuhp_online_new, false);
list_for_each_entry(q, &all_q_list, all_q_node)
blk_mq_unfreeze_queue(q);
@@ -2532,10 +2545,7 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
if (!set->mq_map)
goto out_free_tags;
- if (set->ops->map_queues)
- ret = set->ops->map_queues(set);
- else
- ret = blk_mq_map_queues(set);
+ ret = blk_mq_update_queue_map(set);
if (ret)
goto out_free_mq_map;
@@ -2629,11 +2639,12 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
set->nr_hw_queues = nr_hw_queues;
list_for_each_entry(q, &set->tag_list, tag_set_list) {
blk_mq_realloc_hw_ctxs(set, q);
- blk_mq_queue_reinit(q, cpu_online_mask);
+ blk_mq_queue_reinit(q, cpu_online_mask, true);
}
list_for_each_entry(q, &set->tag_list, tag_set_list)
blk_mq_unfreeze_queue(q);
+
}
EXPORT_SYMBOL_GPL(blk_mq_update_nr_hw_queues);
--
2.12.1
^ permalink raw reply related
* [PATCH 25/25] block: remove the discard_zeroes_data flag
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
Now that we use the proper REQ_OP_WRITE_ZEROES operation everywhere we can
kill this hack.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
Documentation/ABI/testing/sysfs-block | 10 ++-----
Documentation/block/queue-sysfs.txt | 5 ----
block/blk-lib.c | 7 +----
block/blk-settings.c | 3 ---
block/blk-sysfs.c | 2 +-
block/compat_ioctl.c | 2 +-
block/ioctl.c | 2 +-
drivers/block/drbd/drbd_main.c | 2 --
drivers/block/drbd/drbd_nl.c | 7 +----
drivers/block/loop.c | 2 --
drivers/block/mtip32xx/mtip32xx.c | 1 -
drivers/block/nbd.c | 1 -
drivers/md/dm-cache-target.c | 1 -
drivers/md/dm-crypt.c | 1 -
drivers/md/dm-raid.c | 6 ++---
drivers/md/dm-raid1.c | 1 -
drivers/md/dm-table.c | 19 -------------
drivers/md/dm-thin.c | 2 --
drivers/md/raid5.c | 50 +++++++++++------------------------
drivers/scsi/sd.c | 5 ----
drivers/target/target_core_device.c | 2 +-
include/linux/blkdev.h | 15 -----------
include/linux/device-mapper.h | 5 ----
23 files changed, 27 insertions(+), 124 deletions(-)
diff --git a/Documentation/ABI/testing/sysfs-block b/Documentation/ABI/testing/sysfs-block
index 2da04ce6aeef..dea212db9df3 100644
--- a/Documentation/ABI/testing/sysfs-block
+++ b/Documentation/ABI/testing/sysfs-block
@@ -213,14 +213,8 @@ What: /sys/block/<disk>/queue/discard_zeroes_data
Date: May 2011
Contact: Martin K. Petersen <martin.petersen@oracle.com>
Description:
- Devices that support discard functionality may return
- stale or random data when a previously discarded block
- is read back. This can cause problems if the filesystem
- expects discarded blocks to be explicitly cleared. If a
- device reports that it deterministically returns zeroes
- when a discarded area is read the discard_zeroes_data
- parameter will be set to one. Otherwise it will be 0 and
- the result of reading a discarded area is undefined.
+ Will always return 0. Don't rely on any specific behavior
+ for discards, and don't read this file.
What: /sys/block/<disk>/queue/write_same_max_bytes
Date: January 2012
diff --git a/Documentation/block/queue-sysfs.txt b/Documentation/block/queue-sysfs.txt
index b7f6bdc96d73..2c1e67058fd3 100644
--- a/Documentation/block/queue-sysfs.txt
+++ b/Documentation/block/queue-sysfs.txt
@@ -43,11 +43,6 @@ large discards are issued, setting this value lower will make Linux issue
smaller discards and potentially help reduce latencies induced by large
discard operations.
-discard_zeroes_data (RO)
-------------------------
-When read, this file will show if the discarded block are zeroed by the
-device or not. If its value is '1' the blocks are zeroed otherwise not.
-
hw_sector_size (RO)
-------------------
This is the hardware sector size of the device, in bytes.
diff --git a/block/blk-lib.c b/block/blk-lib.c
index 7c27211570fa..268bc9e054c0 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c
@@ -37,17 +37,12 @@ int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
return -ENXIO;
if (flags & BLKDEV_DISCARD_SECURE) {
- if (flags & BLKDEV_DISCARD_ZERO)
- return -EOPNOTSUPP;
if (!blk_queue_secure_erase(q))
return -EOPNOTSUPP;
op = REQ_OP_SECURE_ERASE;
} else {
if (!blk_queue_discard(q))
return -EOPNOTSUPP;
- if ((flags & BLKDEV_DISCARD_ZERO) &&
- !q->limits.discard_zeroes_data)
- return -EOPNOTSUPP;
op = REQ_OP_DISCARD;
}
@@ -126,7 +121,7 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
&bio);
if (!ret && bio) {
ret = submit_bio_wait(bio);
- if (ret == -EOPNOTSUPP && !(flags & BLKDEV_DISCARD_ZERO))
+ if (ret == -EOPNOTSUPP)
ret = 0;
bio_put(bio);
}
diff --git a/block/blk-settings.c b/block/blk-settings.c
index 1e7174ffc9d4..4fa81ed383ca 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -103,7 +103,6 @@ void blk_set_default_limits(struct queue_limits *lim)
lim->discard_granularity = 0;
lim->discard_alignment = 0;
lim->discard_misaligned = 0;
- lim->discard_zeroes_data = 0;
lim->logical_block_size = lim->physical_block_size = lim->io_min = 512;
lim->bounce_pfn = (unsigned long)(BLK_BOUNCE_ANY >> PAGE_SHIFT);
lim->alignment_offset = 0;
@@ -127,7 +126,6 @@ void blk_set_stacking_limits(struct queue_limits *lim)
blk_set_default_limits(lim);
/* Inherit limits from component devices */
- lim->discard_zeroes_data = 1;
lim->max_segments = USHRT_MAX;
lim->max_discard_segments = 1;
lim->max_hw_sectors = UINT_MAX;
@@ -609,7 +607,6 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
t->io_opt = lcm_not_zero(t->io_opt, b->io_opt);
t->cluster &= b->cluster;
- t->discard_zeroes_data &= b->discard_zeroes_data;
/* Physical block size a multiple of the logical block size? */
if (t->physical_block_size & (t->logical_block_size - 1)) {
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 45854266e398..b65ce3c65ae8 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -208,7 +208,7 @@ static ssize_t queue_discard_max_store(struct request_queue *q,
static ssize_t queue_discard_zeroes_data_show(struct request_queue *q, char *page)
{
- return queue_var_show(queue_discard_zeroes_data(q), page);
+ return queue_var_show(0, page);
}
static ssize_t queue_write_same_max_show(struct request_queue *q, char *page)
diff --git a/block/compat_ioctl.c b/block/compat_ioctl.c
index 570021a0dc1c..04325b81c2b4 100644
--- a/block/compat_ioctl.c
+++ b/block/compat_ioctl.c
@@ -685,7 +685,7 @@ long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg)
case BLKALIGNOFF:
return compat_put_int(arg, bdev_alignment_offset(bdev));
case BLKDISCARDZEROES:
- return compat_put_uint(arg, bdev_discard_zeroes_data(bdev));
+ return compat_put_uint(arg, 0);
case BLKFLSBUF:
case BLKROSET:
case BLKDISCARD:
diff --git a/block/ioctl.c b/block/ioctl.c
index 8ea00a41be01..0de02ee67eed 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -547,7 +547,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
case BLKALIGNOFF:
return put_int(arg, bdev_alignment_offset(bdev));
case BLKDISCARDZEROES:
- return put_uint(arg, bdev_discard_zeroes_data(bdev));
+ return put_uint(arg, 0);
case BLKSECTGET:
max_sectors = min_t(unsigned int, USHRT_MAX,
queue_max_sectors(bdev_get_queue(bdev)));
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 8e62d9f65510..84455c365f57 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -931,7 +931,6 @@ void assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p, struct r
p->qlim->io_min = cpu_to_be32(queue_io_min(q));
p->qlim->io_opt = cpu_to_be32(queue_io_opt(q));
p->qlim->discard_enabled = blk_queue_discard(q);
- p->qlim->discard_zeroes_data = queue_discard_zeroes_data(q);
p->qlim->write_same_capable = !!q->limits.max_write_same_sectors;
} else {
q = device->rq_queue;
@@ -941,7 +940,6 @@ void assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p, struct r
p->qlim->io_min = cpu_to_be32(queue_io_min(q));
p->qlim->io_opt = cpu_to_be32(queue_io_opt(q));
p->qlim->discard_enabled = 0;
- p->qlim->discard_zeroes_data = 0;
p->qlim->write_same_capable = 0;
}
}
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index e4516d3b971d..02255a0d68b9 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1199,10 +1199,6 @@ static void decide_on_discard_support(struct drbd_device *device,
struct drbd_connection *connection = first_peer_device(device)->connection;
bool can_do = b ? blk_queue_discard(b) : true;
- if (can_do && b && !b->limits.discard_zeroes_data && !discard_zeroes_if_aligned) {
- can_do = false;
- drbd_info(device, "discard_zeroes_data=0 and discard_zeroes_if_aligned=no: disabling discards\n");
- }
if (can_do && connection->cstate >= C_CONNECTED && !(connection->agreed_features & DRBD_FF_TRIM)) {
can_do = false;
drbd_info(connection, "peer DRBD too old, does not support TRIM: disabling discards\n");
@@ -1484,8 +1480,7 @@ static void sanitize_disk_conf(struct drbd_device *device, struct disk_conf *dis
if (disk_conf->al_extents > drbd_al_extents_max(nbc))
disk_conf->al_extents = drbd_al_extents_max(nbc);
- if (!blk_queue_discard(q)
- || (!q->limits.discard_zeroes_data && !disk_conf->discard_zeroes_if_aligned)) {
+ if (!blk_queue_discard(q)) {
if (disk_conf->rs_discard_granularity) {
disk_conf->rs_discard_granularity = 0; /* disable feature */
drbd_info(device, "rs_discard_granularity feature disabled\n");
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 265cd2e33ff0..57f68f4ee886 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -828,7 +828,6 @@ static void loop_config_discard(struct loop_device *lo)
q->limits.discard_alignment = 0;
blk_queue_max_discard_sectors(q, 0);
blk_queue_max_write_zeroes_sectors(q, 0);
- q->limits.discard_zeroes_data = 0;
queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, q);
return;
}
@@ -837,7 +836,6 @@ static void loop_config_discard(struct loop_device *lo)
q->limits.discard_alignment = 0;
blk_queue_max_discard_sectors(q, UINT_MAX >> 9);
blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> 9);
- q->limits.discard_zeroes_data = 1;
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
}
diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c
index c96c35ab39df..0aaf8077bd19 100644
--- a/drivers/block/mtip32xx/mtip32xx.c
+++ b/drivers/block/mtip32xx/mtip32xx.c
@@ -4025,7 +4025,6 @@ static int mtip_block_initialize(struct driver_data *dd)
dd->queue->limits.discard_granularity = 4096;
blk_queue_max_discard_sectors(dd->queue,
MTIP_MAX_TRIM_ENTRY_LEN * MTIP_MAX_TRIM_ENTRIES);
- dd->queue->limits.discard_zeroes_data = 0;
}
/* Set the capacity of the device in 512 byte sectors. */
diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index 7e4287bc19e5..616e5c6d3ebd 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -1040,7 +1040,6 @@ static int nbd_dev_add(int index)
queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, disk->queue);
disk->queue->limits.discard_granularity = 512;
blk_queue_max_discard_sectors(disk->queue, UINT_MAX);
- disk->queue->limits.discard_zeroes_data = 0;
blk_queue_max_hw_sectors(disk->queue, 65536);
disk->queue->limits.max_sectors = 256;
diff --git a/drivers/md/dm-cache-target.c b/drivers/md/dm-cache-target.c
index 9c689b34e6e7..975922c8f231 100644
--- a/drivers/md/dm-cache-target.c
+++ b/drivers/md/dm-cache-target.c
@@ -2773,7 +2773,6 @@ static int cache_create(struct cache_args *ca, struct cache **result)
ti->num_discard_bios = 1;
ti->discards_supported = true;
- ti->discard_zeroes_data_unsupported = true;
ti->split_discard_bios = false;
cache->features = ca->features;
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 389a3637ffcc..ef1d836bd81b 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -2030,7 +2030,6 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
wake_up_process(cc->write_thread);
ti->num_flush_bios = 1;
- ti->discard_zeroes_data_unsupported = true;
return 0;
diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c
index f8564d63982f..468f1380de1d 100644
--- a/drivers/md/dm-raid.c
+++ b/drivers/md/dm-raid.c
@@ -2813,7 +2813,9 @@ static void configure_discard_support(struct raid_set *rs)
/* Assume discards not supported until after checks below. */
ti->discards_supported = false;
- /* RAID level 4,5,6 require discard_zeroes_data for data integrity! */
+ /*
+ * XXX: RAID level 4,5,6 require zeroing for safety.
+ */
raid456 = (rs->md.level == 4 || rs->md.level == 5 || rs->md.level == 6);
for (i = 0; i < rs->raid_disks; i++) {
@@ -2827,8 +2829,6 @@ static void configure_discard_support(struct raid_set *rs)
return;
if (raid456) {
- if (!q->limits.discard_zeroes_data)
- return;
if (!devices_handle_discard_safely) {
DMERR("raid456 discard support disabled due to discard_zeroes_data uncertainty.");
DMERR("Set dm-raid.devices_handle_discard_safely=Y to override.");
diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 2ddc2d20e62d..a95cbb80fb34 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -1124,7 +1124,6 @@ static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
ti->num_flush_bios = 1;
ti->num_discard_bios = 1;
ti->per_io_data_size = sizeof(struct dm_raid1_bio_record);
- ti->discard_zeroes_data_unsupported = true;
ms->kmirrord_wq = alloc_workqueue("kmirrord", WQ_MEM_RECLAIM, 0);
if (!ms->kmirrord_wq) {
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 5cd665c91ead..958275aca008 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1449,22 +1449,6 @@ static bool dm_table_supports_flush(struct dm_table *t, unsigned long flush)
return false;
}
-static bool dm_table_discard_zeroes_data(struct dm_table *t)
-{
- struct dm_target *ti;
- unsigned i = 0;
-
- /* Ensure that all targets supports discard_zeroes_data. */
- while (i < dm_table_get_num_targets(t)) {
- ti = dm_table_get_target(t, i++);
-
- if (ti->discard_zeroes_data_unsupported)
- return false;
- }
-
- return true;
-}
-
static int device_is_nonrot(struct dm_target *ti, struct dm_dev *dev,
sector_t start, sector_t len, void *data)
{
@@ -1620,9 +1604,6 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
}
blk_queue_write_cache(q, wc, fua);
- if (!dm_table_discard_zeroes_data(t))
- q->limits.discard_zeroes_data = 0;
-
/* Ensure that all underlying devices are non-rotational. */
if (dm_table_all_devices_attribute(t, device_is_nonrot))
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 2b266a2b5035..a5f1916f621a 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -3263,7 +3263,6 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv)
* them down to the data device. The thin device's discard
* processing will cause mappings to be removed from the btree.
*/
- ti->discard_zeroes_data_unsupported = true;
if (pf.discard_enabled && pf.discard_passdown) {
ti->num_discard_bios = 1;
@@ -4119,7 +4118,6 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv)
ti->per_io_data_size = sizeof(struct dm_thin_endio_hook);
/* In case the pool supports discards, pass them on. */
- ti->discard_zeroes_data_unsupported = true;
if (tc->pool->pf.discard_enabled) {
ti->discards_supported = true;
ti->num_discard_bios = 1;
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 8cf1f86dcd05..d6ae8d22d461 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -7229,7 +7229,6 @@ static int raid5_run(struct mddev *mddev)
if (mddev->queue) {
int chunk_size;
- bool discard_supported = true;
/* read-ahead size must cover two whole stripes, which
* is 2 * (datadisks) * chunksize where 'n' is the
* number of raid devices
@@ -7265,12 +7264,6 @@ static int raid5_run(struct mddev *mddev)
blk_queue_max_discard_sectors(mddev->queue,
0xfffe * STRIPE_SECTORS);
- /*
- * unaligned part of discard request will be ignored, so can't
- * guarantee discard_zeroes_data
- */
- mddev->queue->limits.discard_zeroes_data = 0;
-
blk_queue_max_write_same_sectors(mddev->queue, 0);
blk_queue_max_write_zeroes_sectors(mddev->queue, 0);
@@ -7279,35 +7272,24 @@ static int raid5_run(struct mddev *mddev)
rdev->data_offset << 9);
disk_stack_limits(mddev->gendisk, rdev->bdev,
rdev->new_data_offset << 9);
- /*
- * discard_zeroes_data is required, otherwise data
- * could be lost. Consider a scenario: discard a stripe
- * (the stripe could be inconsistent if
- * discard_zeroes_data is 0); write one disk of the
- * stripe (the stripe could be inconsistent again
- * depending on which disks are used to calculate
- * parity); the disk is broken; The stripe data of this
- * disk is lost.
- */
- if (!blk_queue_discard(bdev_get_queue(rdev->bdev)) ||
- !bdev_get_queue(rdev->bdev)->
- limits.discard_zeroes_data)
- discard_supported = false;
- /* Unfortunately, discard_zeroes_data is not currently
- * a guarantee - just a hint. So we only allow DISCARD
- * if the sysadmin has confirmed that only safe devices
- * are in use by setting a module parameter.
- */
- if (!devices_handle_discard_safely) {
- if (discard_supported) {
- pr_info("md/raid456: discard support disabled due to uncertainty.\n");
- pr_info("Set raid456.devices_handle_discard_safely=Y to override.\n");
- }
- discard_supported = false;
- }
}
- if (discard_supported &&
+ /*
+ * zeroing is required, otherwise data
+ * could be lost. Consider a scenario: discard a stripe
+ * (the stripe could be inconsistent if
+ * discard_zeroes_data is 0); write one disk of the
+ * stripe (the stripe could be inconsistent again
+ * depending on which disks are used to calculate
+ * parity); the disk is broken; The stripe data of this
+ * disk is lost.
+ *
+ * We only allow DISCARD if the sysadmin has confirmed that
+ * only safe devices are in use by setting a module parameter.
+ * A better idea might be to turn DISCARD into WRITE_ZEROES
+ * requests, as that is required to be safe.
+ */
+ if (devices_handle_discard_safely &&
mddev->queue->limits.max_discard_sectors >= (stripe >> 9) &&
mddev->queue->limits.discard_granularity >= stripe)
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD,
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 001593ed0444..bcb0cb020fd2 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -644,8 +644,6 @@ static void sd_config_discard(struct scsi_disk *sdkp, unsigned int mode)
unsigned int logical_block_size = sdkp->device->sector_size;
unsigned int max_blocks = 0;
- q->limits.discard_zeroes_data = 0;
-
/*
* When LBPRZ is reported, discard alignment and granularity
* must be fixed to the logical block size. Otherwise the block
@@ -681,19 +679,16 @@ static void sd_config_discard(struct scsi_disk *sdkp, unsigned int mode)
case SD_LBP_WS16:
max_blocks = min_not_zero(sdkp->max_ws_blocks,
(u32)SD_MAX_WS16_BLOCKS);
- q->limits.discard_zeroes_data = sdkp->lbprz;
break;
case SD_LBP_WS10:
max_blocks = min_not_zero(sdkp->max_ws_blocks,
(u32)SD_MAX_WS10_BLOCKS);
- q->limits.discard_zeroes_data = sdkp->lbprz;
break;
case SD_LBP_ZERO:
max_blocks = min_not_zero(sdkp->max_ws_blocks,
(u32)SD_MAX_WS10_BLOCKS);
- q->limits.discard_zeroes_data = 1;
break;
}
diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index c754ae33bf7b..d2f089cfa9ae 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -851,7 +851,7 @@ bool target_configure_unmap_from_queue(struct se_dev_attrib *attrib,
attrib->unmap_granularity = q->limits.discard_granularity / block_size;
attrib->unmap_granularity_alignment = q->limits.discard_alignment /
block_size;
- attrib->unmap_zeroes_data = q->limits.discard_zeroes_data;
+ attrib->unmap_zeroes_data = 0;
return true;
}
EXPORT_SYMBOL(target_configure_unmap_from_queue);
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index a5055d760661..d5d9dd72418a 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -339,7 +339,6 @@ struct queue_limits {
unsigned char misaligned;
unsigned char discard_misaligned;
unsigned char cluster;
- unsigned char discard_zeroes_data;
unsigned char raid_partial_stripes_expensive;
enum blk_zoned_model zoned;
};
@@ -1342,7 +1341,6 @@ extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
sector_t nr_sects, gfp_t gfp_mask, struct page *page);
#define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
-#define BLKDEV_DISCARD_ZERO (1 << 1) /* must reliably zero data */
extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
@@ -1542,19 +1540,6 @@ static inline int bdev_discard_alignment(struct block_device *bdev)
return q->limits.discard_alignment;
}
-static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
-{
- if (q->limits.max_discard_sectors && q->limits.discard_zeroes_data == 1)
- return 1;
-
- return 0;
-}
-
-static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
-{
- return queue_discard_zeroes_data(bdev_get_queue(bdev));
-}
-
static inline unsigned int bdev_write_same(struct block_device *bdev)
{
struct request_queue *q = bdev_get_queue(bdev);
diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
index 3829bee2302a..c7ea33e38fb9 100644
--- a/include/linux/device-mapper.h
+++ b/include/linux/device-mapper.h
@@ -296,11 +296,6 @@ struct dm_target {
* on max_io_len boundary.
*/
bool split_discard_bios:1;
-
- /*
- * Set if this target does not return zeroes on discarded blocks.
- */
- bool discard_zeroes_data_unsupported:1;
};
/* Each target can link one of these into the table */
--
2.11.0
^ permalink raw reply related
* [PATCH 24/25] drbd: implement REQ_OP_WRITE_ZEROES
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
It seems like DRBD assumes its on the wire TRIM request always zeroes data.
Use that fact to implement REQ_OP_WRITE_ZEROES.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/drbd/drbd_main.c | 3 ++-
drivers/block/drbd/drbd_nl.c | 2 ++
drivers/block/drbd/drbd_receiver.c | 6 +++---
drivers/block/drbd/drbd_req.c | 7 +++++--
drivers/block/drbd/drbd_worker.c | 4 +++-
5 files changed, 15 insertions(+), 7 deletions(-)
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 92c60cbd04ee..8e62d9f65510 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -1668,7 +1668,8 @@ static u32 bio_flags_to_wire(struct drbd_connection *connection,
(bio->bi_opf & REQ_FUA ? DP_FUA : 0) |
(bio->bi_opf & REQ_PREFLUSH ? DP_FLUSH : 0) |
(bio_op(bio) == REQ_OP_WRITE_SAME ? DP_WSAME : 0) |
- (bio_op(bio) == REQ_OP_DISCARD ? DP_DISCARD : 0);
+ (bio_op(bio) == REQ_OP_DISCARD ? DP_DISCARD : 0) |
+ (bio_op(bio) == REQ_OP_WRITE_ZEROES ? DP_DISCARD : 0);
else
return bio->bi_opf & REQ_SYNC ? DP_RW_SYNC : 0;
}
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index 908c704e20aa..e4516d3b971d 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1217,10 +1217,12 @@ static void decide_on_discard_support(struct drbd_device *device,
blk_queue_discard_granularity(q, 512);
q->limits.max_discard_sectors = drbd_max_discard_sectors(connection);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
+ q->limits.max_write_zeroes_sectors = drbd_max_discard_sectors(connection);
} else {
queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, q);
blk_queue_discard_granularity(q, 0);
q->limits.max_discard_sectors = 0;
+ q->limits.max_write_zeroes_sectors = 0;
}
}
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index bc1d296581f9..1b0a2be24f39 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -2285,7 +2285,7 @@ static unsigned long wire_flags_to_bio_flags(u32 dpf)
static unsigned long wire_flags_to_bio_op(u32 dpf)
{
if (dpf & DP_DISCARD)
- return REQ_OP_DISCARD;
+ return REQ_OP_WRITE_ZEROES;
else
return REQ_OP_WRITE;
}
@@ -2476,7 +2476,7 @@ static int receive_Data(struct drbd_connection *connection, struct packet_info *
op_flags = wire_flags_to_bio_flags(dp_flags);
if (pi->cmd == P_TRIM) {
D_ASSERT(peer_device, peer_req->i.size > 0);
- D_ASSERT(peer_device, op == REQ_OP_DISCARD);
+ D_ASSERT(peer_device, op == REQ_OP_WRITE_ZEROES);
D_ASSERT(peer_device, peer_req->pages == NULL);
} else if (peer_req->pages == NULL) {
D_ASSERT(device, peer_req->i.size == 0);
@@ -4789,7 +4789,7 @@ static int receive_rs_deallocated(struct drbd_connection *connection, struct pac
if (get_ldev(device)) {
struct drbd_peer_request *peer_req;
- const int op = REQ_OP_DISCARD;
+ const int op = REQ_OP_WRITE_ZEROES;
peer_req = drbd_alloc_peer_req(peer_device, ID_SYNCER, sector,
size, 0, GFP_NOIO);
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c
index 6da9ea8c48b6..b5730e17b455 100644
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -59,6 +59,7 @@ static struct drbd_request *drbd_req_new(struct drbd_device *device, struct bio
drbd_req_make_private_bio(req, bio_src);
req->rq_state = (bio_data_dir(bio_src) == WRITE ? RQ_WRITE : 0)
| (bio_op(bio_src) == REQ_OP_WRITE_SAME ? RQ_WSAME : 0)
+ | (bio_op(bio_src) == REQ_OP_WRITE_ZEROES ? RQ_UNMAP : 0)
| (bio_op(bio_src) == REQ_OP_DISCARD ? RQ_UNMAP : 0);
req->device = device;
req->master_bio = bio_src;
@@ -1180,7 +1181,8 @@ drbd_submit_req_private_bio(struct drbd_request *req)
if (get_ldev(device)) {
if (drbd_insert_fault(device, type))
bio_io_error(bio);
- else if (bio_op(bio) == REQ_OP_DISCARD)
+ else if (bio_op(bio) == REQ_OP_WRITE_ZEROES ||
+ bio_op(bio) == REQ_OP_DISCARD)
drbd_process_discard_req(req);
else
generic_make_request(bio);
@@ -1234,7 +1236,8 @@ drbd_request_prepare(struct drbd_device *device, struct bio *bio, unsigned long
_drbd_start_io_acct(device, req);
/* process discards always from our submitter thread */
- if (bio_op(bio) & REQ_OP_DISCARD)
+ if ((bio_op(bio) & REQ_OP_WRITE_ZEROES) ||
+ (bio_op(bio) & REQ_OP_DISCARD))
goto queue_for_submitter_thread;
if (rw == WRITE && req->private_bio && req->i.size
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 3bff33f21435..1afcb4e02d8d 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -174,7 +174,8 @@ void drbd_peer_request_endio(struct bio *bio)
struct drbd_peer_request *peer_req = bio->bi_private;
struct drbd_device *device = peer_req->peer_device->device;
bool is_write = bio_data_dir(bio) == WRITE;
- bool is_discard = !!(bio_op(bio) == REQ_OP_DISCARD);
+ bool is_discard = bio_op(bio) == REQ_OP_WRITE_ZEROES ||
+ bio_op(bio) == REQ_OP_DISCARD;
if (bio->bi_error && __ratelimit(&drbd_ratelimit_state))
drbd_warn(device, "%s: error=%d s=%llus\n",
@@ -249,6 +250,7 @@ void drbd_request_endio(struct bio *bio)
/* to avoid recursion in __req_mod */
if (unlikely(bio->bi_error)) {
switch (bio_op(bio)) {
+ case REQ_OP_WRITE_ZEROES:
case REQ_OP_DISCARD:
if (bio->bi_error == -EOPNOTSUPP)
what = DISCARD_COMPLETED_NOTSUPP;
--
2.11.0
^ permalink raw reply related
* [PATCH 23/25] drbd: make intelligent use of blkdev_issue_zeroout
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
drbd always wants its discard wire operations to zero the blocks, so
use blkdev_issue_zeroout with the BLKDEV_ZERO_UNMAP flag instead of
reinventing it poorly.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/drbd/drbd_debugfs.c | 3 --
drivers/block/drbd/drbd_int.h | 6 ---
drivers/block/drbd/drbd_receiver.c | 102 ++-----------------------------------
drivers/block/drbd/drbd_req.c | 6 +--
4 files changed, 7 insertions(+), 110 deletions(-)
diff --git a/drivers/block/drbd/drbd_debugfs.c b/drivers/block/drbd/drbd_debugfs.c
index de5c3ee8a790..494837e59f23 100644
--- a/drivers/block/drbd/drbd_debugfs.c
+++ b/drivers/block/drbd/drbd_debugfs.c
@@ -236,9 +236,6 @@ static void seq_print_peer_request_flags(struct seq_file *m, struct drbd_peer_re
seq_print_rq_state_bit(m, f & EE_CALL_AL_COMPLETE_IO, &sep, "in-AL");
seq_print_rq_state_bit(m, f & EE_SEND_WRITE_ACK, &sep, "C");
seq_print_rq_state_bit(m, f & EE_MAY_SET_IN_SYNC, &sep, "set-in-sync");
-
- if (f & EE_IS_TRIM)
- __seq_print_rq_state_bit(m, f & EE_IS_TRIM_USE_ZEROOUT, &sep, "zero-out", "trim");
seq_print_rq_state_bit(m, f & EE_WRITE_SAME, &sep, "write-same");
seq_putc(m, '\n');
}
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 724d1c50fc52..d5da45bb03a6 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -437,9 +437,6 @@ enum {
/* is this a TRIM aka REQ_DISCARD? */
__EE_IS_TRIM,
- /* our lower level cannot handle trim,
- * and we want to fall back to zeroout instead */
- __EE_IS_TRIM_USE_ZEROOUT,
/* In case a barrier failed,
* we need to resubmit without the barrier flag. */
@@ -482,7 +479,6 @@ enum {
#define EE_CALL_AL_COMPLETE_IO (1<<__EE_CALL_AL_COMPLETE_IO)
#define EE_MAY_SET_IN_SYNC (1<<__EE_MAY_SET_IN_SYNC)
#define EE_IS_TRIM (1<<__EE_IS_TRIM)
-#define EE_IS_TRIM_USE_ZEROOUT (1<<__EE_IS_TRIM_USE_ZEROOUT)
#define EE_RESUBMITTED (1<<__EE_RESUBMITTED)
#define EE_WAS_ERROR (1<<__EE_WAS_ERROR)
#define EE_HAS_DIGEST (1<<__EE_HAS_DIGEST)
@@ -1561,8 +1557,6 @@ extern void start_resync_timer_fn(unsigned long data);
extern void drbd_endio_write_sec_final(struct drbd_peer_request *peer_req);
/* drbd_receiver.c */
-extern int drbd_issue_discard_or_zero_out(struct drbd_device *device,
- sector_t start, unsigned int nr_sectors, bool discard);
extern int drbd_receiver(struct drbd_thread *thi);
extern int drbd_ack_receiver(struct drbd_thread *thi);
extern void drbd_send_ping_wf(struct work_struct *ws);
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index dc9a6dcd431c..bc1d296581f9 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -1448,108 +1448,14 @@ void drbd_bump_write_ordering(struct drbd_resource *resource, struct drbd_backin
drbd_info(resource, "Method to ensure write ordering: %s\n", write_ordering_str[resource->write_ordering]);
}
-/*
- * We *may* ignore the discard-zeroes-data setting, if so configured.
- *
- * Assumption is that it "discard_zeroes_data=0" is only because the backend
- * may ignore partial unaligned discards.
- *
- * LVM/DM thin as of at least
- * LVM version: 2.02.115(2)-RHEL7 (2015-01-28)
- * Library version: 1.02.93-RHEL7 (2015-01-28)
- * Driver version: 4.29.0
- * still behaves this way.
- *
- * For unaligned (wrt. alignment and granularity) or too small discards,
- * we zero-out the initial (and/or) trailing unaligned partial chunks,
- * but discard all the aligned full chunks.
- *
- * At least for LVM/DM thin, the result is effectively "discard_zeroes_data=1".
- */
-int drbd_issue_discard_or_zero_out(struct drbd_device *device, sector_t start, unsigned int nr_sectors, bool discard)
-{
- struct block_device *bdev = device->ldev->backing_bdev;
- struct request_queue *q = bdev_get_queue(bdev);
- sector_t tmp, nr;
- unsigned int max_discard_sectors, granularity;
- int alignment;
- int err = 0;
-
- if (!discard)
- goto zero_out;
-
- /* Zero-sector (unknown) and one-sector granularities are the same. */
- granularity = max(q->limits.discard_granularity >> 9, 1U);
- alignment = (bdev_discard_alignment(bdev) >> 9) % granularity;
-
- max_discard_sectors = min(q->limits.max_discard_sectors, (1U << 22));
- max_discard_sectors -= max_discard_sectors % granularity;
- if (unlikely(!max_discard_sectors))
- goto zero_out;
-
- if (nr_sectors < granularity)
- goto zero_out;
-
- tmp = start;
- if (sector_div(tmp, granularity) != alignment) {
- if (nr_sectors < 2*granularity)
- goto zero_out;
- /* start + gran - (start + gran - align) % gran */
- tmp = start + granularity - alignment;
- tmp = start + granularity - sector_div(tmp, granularity);
-
- nr = tmp - start;
- err |= blkdev_issue_zeroout(bdev, start, nr, GFP_NOIO,
- BLKDEV_ZERO_NOUNMAP);
- nr_sectors -= nr;
- start = tmp;
- }
- while (nr_sectors >= granularity) {
- nr = min_t(sector_t, nr_sectors, max_discard_sectors);
- err |= blkdev_issue_discard(bdev, start, nr, GFP_NOIO,
- BLKDEV_ZERO_NOUNMAP);
- nr_sectors -= nr;
- start += nr;
- }
- zero_out:
- if (nr_sectors) {
- err |= blkdev_issue_zeroout(bdev, start, nr_sectors, GFP_NOIO,
- BLKDEV_ZERO_NOUNMAP);
- }
- return err != 0;
-}
-
-static bool can_do_reliable_discards(struct drbd_device *device)
-{
- struct request_queue *q = bdev_get_queue(device->ldev->backing_bdev);
- struct disk_conf *dc;
- bool can_do;
-
- if (!blk_queue_discard(q))
- return false;
-
- if (q->limits.discard_zeroes_data)
- return true;
-
- rcu_read_lock();
- dc = rcu_dereference(device->ldev->disk_conf);
- can_do = dc->discard_zeroes_if_aligned;
- rcu_read_unlock();
- return can_do;
-}
-
static void drbd_issue_peer_discard(struct drbd_device *device, struct drbd_peer_request *peer_req)
{
- /* If the backend cannot discard, or does not guarantee
- * read-back zeroes in discarded ranges, we fall back to
- * zero-out. Unless configuration specifically requested
- * otherwise. */
- if (!can_do_reliable_discards(device))
- peer_req->flags |= EE_IS_TRIM_USE_ZEROOUT;
+ struct block_device *bdev = device->ldev->backing_bdev;
- if (drbd_issue_discard_or_zero_out(device, peer_req->i.sector,
- peer_req->i.size >> 9, !(peer_req->flags & EE_IS_TRIM_USE_ZEROOUT)))
+ if (blkdev_issue_zeroout(bdev, peer_req->i.sector, peer_req->i.size >> 9,
+ GFP_NOIO, 0))
peer_req->flags |= EE_WAS_ERROR;
+
drbd_endio_write_sec_final(peer_req);
}
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c
index 652114ae1a8a..6da9ea8c48b6 100644
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -1148,10 +1148,10 @@ static int drbd_process_write_request(struct drbd_request *req)
static void drbd_process_discard_req(struct drbd_request *req)
{
- int err = drbd_issue_discard_or_zero_out(req->device,
- req->i.sector, req->i.size >> 9, true);
+ struct block_device *bdev = req->device->ldev->backing_bdev;
- if (err)
+ if (blkdev_issue_zeroout(bdev, req->i.sector, req->i.size >> 9,
+ GFP_NOIO, 0))
req->private_bio->bi_error = -EIO;
bio_endio(req->private_bio);
}
--
2.11.0
^ permalink raw reply related
* [PATCH 22/25] block: stop using discards for zeroing
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
Now that we have REQ_OP_WRITE_ZEROES implemented for all devices that
support efficient zeroing of devices we can remove the call to
blkdev_issue_discard. This means we only have two ways of zeroing left
and can simply the code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
block/blk-lib.c | 11 +++++------
1 file changed, 5 insertions(+), 6 deletions(-)
diff --git a/block/blk-lib.c b/block/blk-lib.c
index 2f882e22890b..7c27211570fa 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c
@@ -279,6 +279,11 @@ static int __blkdev_issue_write_zeroes(struct block_device *bdev,
* Zero-fill a block range, either using hardware offload or by explicitly
* writing zeroes to the device.
*
+ * Note that this function may fail with -EOPNOTSUPP if the driver supports
+ * efficient zeroing operation, but the device capabilities can only be
+ * discovered by trial and error. In this case the caller should call the
+ * function again, and it will use the fallback path.
+ *
* If a device is using logical block provisioning, the underlying space will
* not be released if %flags contains BLKDEV_ZERO_NOUNMAP.
*
@@ -349,12 +354,6 @@ int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
struct bio *bio = NULL;
struct blk_plug plug;
- if (!(flags & BLKDEV_ZERO_NOUNMAP)) {
- if (!blkdev_issue_discard(bdev, sector, nr_sects, gfp_mask,
- BLKDEV_DISCARD_ZERO))
- return 0;
- }
-
blk_start_plug(&plug);
ret = __blkdev_issue_zeroout(bdev, sector, nr_sects, gfp_mask,
&bio, flags);
--
2.11.0
^ permalink raw reply related
* [PATCH 21/25] mmc: remove the discard_zeroes_data flag
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
mmc only supports discarding on large alignments, so the zeroing code
would always fall back to explicit writings of zeroes.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/mmc/core/queue.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/mmc/core/queue.c b/drivers/mmc/core/queue.c
index 493eb10ce580..4c54ad34e17a 100644
--- a/drivers/mmc/core/queue.c
+++ b/drivers/mmc/core/queue.c
@@ -167,8 +167,6 @@ static void mmc_queue_setup_discard(struct request_queue *q,
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
blk_queue_max_discard_sectors(q, max_discard);
- if (card->erased_byte == 0 && !mmc_can_discard(card))
- q->limits.discard_zeroes_data = 1;
q->limits.discard_granularity = card->pref_erase << 9;
/* granularity must not be greater than max. discard */
if (card->pref_erase > max_discard)
--
2.11.0
^ permalink raw reply related
* [PATCH 20/25] rsxx: remove the discard_zeroes_data flag
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
rsxx only supports discarding on large alignments, so the zeroing code
would always fall back to explicit writings of zeroes.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/rsxx/dev.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/block/rsxx/dev.c b/drivers/block/rsxx/dev.c
index f81d70b39d10..9c566364ac9c 100644
--- a/drivers/block/rsxx/dev.c
+++ b/drivers/block/rsxx/dev.c
@@ -300,7 +300,6 @@ int rsxx_setup_dev(struct rsxx_cardinfo *card)
RSXX_HW_BLK_SIZE >> 9);
card->queue->limits.discard_granularity = RSXX_HW_BLK_SIZE;
card->queue->limits.discard_alignment = RSXX_HW_BLK_SIZE;
- card->queue->limits.discard_zeroes_data = 1;
}
card->queue->queuedata = card;
--
2.11.0
^ permalink raw reply related
* [PATCH 19/25] rbd: remove the discard_zeroes_data flag
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
rbd only supports discarding on large alignments, so the zeroing code
would always fall back to explicit writings of zeroes.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/rbd.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 517838b65964..0ec3b430e81d 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -4380,7 +4380,6 @@ static int rbd_init_disk(struct rbd_device *rbd_dev)
q->limits.discard_granularity = segment_size;
q->limits.discard_alignment = segment_size;
blk_queue_max_discard_sectors(q, segment_size / SECTOR_SIZE);
- q->limits.discard_zeroes_data = 1;
if (!ceph_test_opt(rbd_dev->rbd_client->client, NOCRC))
q->backing_dev_info->capabilities |= BDI_CAP_STABLE_WRITES;
--
2.11.0
^ permalink raw reply related
* [PATCH 18/25] brd: remove discard support
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
It's just a in-driver reimplementation of writing zeroes to the pages,
which fails if the discards aren't page aligned.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/brd.c | 54 -----------------------------------------------------
1 file changed, 54 deletions(-)
diff --git a/drivers/block/brd.c b/drivers/block/brd.c
index 3adc32a3153b..4ec84d504780 100644
--- a/drivers/block/brd.c
+++ b/drivers/block/brd.c
@@ -134,28 +134,6 @@ static struct page *brd_insert_page(struct brd_device *brd, sector_t sector)
return page;
}
-static void brd_free_page(struct brd_device *brd, sector_t sector)
-{
- struct page *page;
- pgoff_t idx;
-
- spin_lock(&brd->brd_lock);
- idx = sector >> PAGE_SECTORS_SHIFT;
- page = radix_tree_delete(&brd->brd_pages, idx);
- spin_unlock(&brd->brd_lock);
- if (page)
- __free_page(page);
-}
-
-static void brd_zero_page(struct brd_device *brd, sector_t sector)
-{
- struct page *page;
-
- page = brd_lookup_page(brd, sector);
- if (page)
- clear_highpage(page);
-}
-
/*
* Free all backing store pages and radix tree. This must only be called when
* there are no other users of the device.
@@ -212,24 +190,6 @@ static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n)
return 0;
}
-static void discard_from_brd(struct brd_device *brd,
- sector_t sector, size_t n)
-{
- while (n >= PAGE_SIZE) {
- /*
- * Don't want to actually discard pages here because
- * re-allocating the pages can result in writeback
- * deadlocks under heavy load.
- */
- if (0)
- brd_free_page(brd, sector);
- else
- brd_zero_page(brd, sector);
- sector += PAGE_SIZE >> SECTOR_SHIFT;
- n -= PAGE_SIZE;
- }
-}
-
/*
* Copy n bytes from src to the brd starting at sector. Does not sleep.
*/
@@ -338,14 +298,6 @@ static blk_qc_t brd_make_request(struct request_queue *q, struct bio *bio)
if (bio_end_sector(bio) > get_capacity(bdev->bd_disk))
goto io_error;
- if (unlikely(bio_op(bio) == REQ_OP_DISCARD)) {
- if (sector & ((PAGE_SIZE >> SECTOR_SHIFT) - 1) ||
- bio->bi_iter.bi_size & ~PAGE_MASK)
- goto io_error;
- discard_from_brd(brd, sector, bio->bi_iter.bi_size);
- goto out;
- }
-
bio_for_each_segment(bvec, bio, iter) {
unsigned int len = bvec.bv_len;
int err;
@@ -357,7 +309,6 @@ static blk_qc_t brd_make_request(struct request_queue *q, struct bio *bio)
sector += len >> SECTOR_SHIFT;
}
-out:
bio_endio(bio);
return BLK_QC_T_NONE;
io_error:
@@ -464,11 +415,6 @@ static struct brd_device *brd_alloc(int i)
* is harmless)
*/
blk_queue_physical_block_size(brd->brd_queue, PAGE_SIZE);
-
- brd->brd_queue->limits.discard_granularity = PAGE_SIZE;
- blk_queue_max_discard_sectors(brd->brd_queue, UINT_MAX);
- brd->brd_queue->limits.discard_zeroes_data = 1;
- queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, brd->brd_queue);
#ifdef CONFIG_BLK_DEV_RAM_DAX
queue_flag_set_unlocked(QUEUE_FLAG_DAX, brd->brd_queue);
#endif
--
2.11.0
^ permalink raw reply related
* [PATCH 17/25] loop: implement REQ_OP_WRITE_ZEROES
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
It's identical to discard as hole punches will always leave us with
zeroes on reads.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/loop.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 0ecb6461ed81..265cd2e33ff0 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -528,6 +528,7 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
case REQ_OP_FLUSH:
return lo_req_flush(lo, rq);
case REQ_OP_DISCARD:
+ case REQ_OP_WRITE_ZEROES:
return lo_discard(lo, rq, pos);
case REQ_OP_WRITE:
if (lo->transfer)
@@ -826,6 +827,7 @@ static void loop_config_discard(struct loop_device *lo)
q->limits.discard_granularity = 0;
q->limits.discard_alignment = 0;
blk_queue_max_discard_sectors(q, 0);
+ blk_queue_max_write_zeroes_sectors(q, 0);
q->limits.discard_zeroes_data = 0;
queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, q);
return;
@@ -834,6 +836,7 @@ static void loop_config_discard(struct loop_device *lo)
q->limits.discard_granularity = inode->i_sb->s_blocksize;
q->limits.discard_alignment = 0;
blk_queue_max_discard_sectors(q, UINT_MAX >> 9);
+ blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> 9);
q->limits.discard_zeroes_data = 1;
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
}
@@ -1660,6 +1663,7 @@ static int loop_queue_rq(struct blk_mq_hw_ctx *hctx,
switch (req_op(cmd->rq)) {
case REQ_OP_FLUSH:
case REQ_OP_DISCARD:
+ case REQ_OP_WRITE_ZEROES:
cmd->use_aio = false;
break;
default:
--
2.11.0
^ permalink raw reply related
* [PATCH 16/25] zram: implement REQ_OP_WRITE_ZEROES
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
Just the same as discard if the block size equals the system page size.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/block/zram/zram_drv.c | 13 ++++++++-----
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
index dceb5edd1e54..1710b06f04a7 100644
--- a/drivers/block/zram/zram_drv.c
+++ b/drivers/block/zram/zram_drv.c
@@ -829,10 +829,14 @@ static void __zram_make_request(struct zram *zram, struct bio *bio)
offset = (bio->bi_iter.bi_sector &
(SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
- if (unlikely(bio_op(bio) == REQ_OP_DISCARD)) {
+ switch (bio_op(bio)) {
+ case REQ_OP_DISCARD:
+ case REQ_OP_WRITE_ZEROES:
zram_bio_discard(zram, index, offset, bio);
bio_endio(bio);
return;
+ default:
+ break;
}
bio_for_each_segment(bvec, bio, iter) {
@@ -1192,6 +1196,8 @@ static int zram_add(void)
zram->disk->queue->limits.max_sectors = SECTORS_PER_PAGE;
zram->disk->queue->limits.chunk_sectors = 0;
blk_queue_max_discard_sectors(zram->disk->queue, UINT_MAX);
+ queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zram->disk->queue);
+
/*
* zram_bio_discard() will clear all logical blocks if logical block
* size is identical with physical block size(PAGE_SIZE). But if it is
@@ -1201,10 +1207,7 @@ static int zram_add(void)
* zeroed.
*/
if (ZRAM_LOGICAL_BLOCK_SIZE == PAGE_SIZE)
- zram->disk->queue->limits.discard_zeroes_data = 1;
- else
- zram->disk->queue->limits.discard_zeroes_data = 0;
- queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zram->disk->queue);
+ blk_queue_max_write_zeroes_sectors(zram->disk->queue, UINT_MAX);
add_disk(zram->disk);
--
2.11.0
^ permalink raw reply related
* [PATCH 15/25] nvme: implement REQ_OP_WRITE_ZEROES
From: Christoph Hellwig @ 2017-03-31 16:33 UTC (permalink / raw)
To: axboe, martin.petersen, agk, snitzer, shli, philipp.reisner,
lars.ellenberg
Cc: linux-block, linux-scsi, drbd-dev, dm-devel, linux-raid
In-Reply-To: <20170331163313.31821-1-hch@lst.de>
But now for the real NVMe Write Zeroes yet, just to get rid of the
discard abuse for zeroing. Also rename the quirk flag to be a bit
more self-explanatory.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/nvme/host/core.c | 10 +++++-----
drivers/nvme/host/nvme.h | 6 +++---
drivers/nvme/host/pci.c | 6 +++---
3 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 4a6d7f408769..94b41d847b01 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -335,6 +335,8 @@ int nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
case REQ_OP_FLUSH:
nvme_setup_flush(ns, cmd);
break;
+ case REQ_OP_WRITE_ZEROES:
+ /* currently only aliased to deallocate for a few ctrls: */
case REQ_OP_DISCARD:
ret = nvme_setup_discard(ns, req, cmd);
break;
@@ -900,16 +902,14 @@ static void nvme_config_discard(struct nvme_ns *ns)
BUILD_BUG_ON(PAGE_SIZE / sizeof(struct nvme_dsm_range) <
NVME_DSM_MAX_RANGES);
- if (ctrl->quirks & NVME_QUIRK_DISCARD_ZEROES)
- ns->queue->limits.discard_zeroes_data = 1;
- else
- ns->queue->limits.discard_zeroes_data = 0;
-
ns->queue->limits.discard_alignment = logical_block_size;
ns->queue->limits.discard_granularity = logical_block_size;
blk_queue_max_discard_sectors(ns->queue, UINT_MAX);
blk_queue_max_discard_segments(ns->queue, NVME_DSM_MAX_RANGES);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, ns->queue);
+
+ if (ctrl->quirks & NVME_QUIRK_DEALLOCATE_ZEROES)
+ blk_queue_max_write_zeroes_sectors(ns->queue, UINT_MAX);
}
static int nvme_revalidate_ns(struct nvme_ns *ns, struct nvme_id_ns **id)
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 2aa20e3e5675..07ebc4a1c8fc 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -68,10 +68,10 @@ enum nvme_quirks {
NVME_QUIRK_IDENTIFY_CNS = (1 << 1),
/*
- * The controller deterministically returns O's on reads to discarded
- * logical blocks.
+ * The controller deterministically returns O's on reads to
+ * logical blocks that deallocate was called on.
*/
- NVME_QUIRK_DISCARD_ZEROES = (1 << 2),
+ NVME_QUIRK_DEALLOCATE_ZEROES = (1 << 2),
/*
* The controller needs a delay before starts checking the device
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 26a5fd05fe88..0a28787267f0 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -2135,13 +2135,13 @@ static const struct pci_error_handlers nvme_err_handler = {
static const struct pci_device_id nvme_id_table[] = {
{ PCI_VDEVICE(INTEL, 0x0953),
.driver_data = NVME_QUIRK_STRIPE_SIZE |
- NVME_QUIRK_DISCARD_ZEROES, },
+ NVME_QUIRK_DEALLOCATE_ZEROES, },
{ PCI_VDEVICE(INTEL, 0x0a53),
.driver_data = NVME_QUIRK_STRIPE_SIZE |
- NVME_QUIRK_DISCARD_ZEROES, },
+ NVME_QUIRK_DEALLOCATE_ZEROES, },
{ PCI_VDEVICE(INTEL, 0x0a54),
.driver_data = NVME_QUIRK_STRIPE_SIZE |
- NVME_QUIRK_DISCARD_ZEROES, },
+ NVME_QUIRK_DEALLOCATE_ZEROES, },
{ PCI_VDEVICE(INTEL, 0x5845), /* Qemu emulated controller */
.driver_data = NVME_QUIRK_IDENTIFY_CNS, },
{ PCI_DEVICE(0x1c58, 0x0003), /* HGST adapter */
--
2.11.0
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox