CONFIG_DEBUG_TEST_DRIVER_REMOVE found a possible leak of q->rq_wb when a
request queue is reregistered. This has been a problem since wbt was
introduced, but the WARN_ON(!list_empty(&stats->callbacks)) in the
blk-stat rework exposed it. Fix it by cleaning up wbt when we unregister
the queue.
Fixes: 87760e5eef ("block: hook up writeback throttling")
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
blk_alloc_queue_node() already allocates q->stats, so
blk_mq_init_allocated_queue() is overwriting it with a new allocation.
Fixes: a83b576c9c ("block: fix stacked driver stats init and free")
Reviewed-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Now that the remaining drivers have been converted to one request queue
per gendisk, let's warn if a request queue gets registered more than
once. This will catch future drivers which might do it inadvertently or
any old drivers that I may have missed.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Before commit 780db2071a(blk-mq: decouble blk-mq freezing
from generic bypassing), the dying flag is checked before
entering queue, and Tejun converts the checking into .mq_freeze_depth,
and assumes the counter is increased just after dying flag
is set. Unfortunately we doesn't do that in blk_set_queue_dying().
This patch calls blk_freeze_queue_start() in blk_set_queue_dying(),
so that we can block new I/O coming once the queue is set as dying.
Given blk_set_queue_dying() is always called in remove path
of block device, and queue will be cleaned up later, we don't
need to worry about undoing the counter.
Cc: Tejun Heo <tj@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
As the .q_usage_counter is used by both legacy and
mq path, we need to block new I/O if queue becomes
dead in blk_queue_enter().
So rename it and we can use this function in both
paths.
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
Without the barrier, reading DEAD flag of .q_usage_counter
and reading .mq_freeze_depth may be reordered, then the
following wait_event_interruptible() may never return.
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
This patch adds comment on two races related with
timeout handler:
- requeue from queue busy vs. timeout
- rq free & reallocation vs. timeout
Both the races themselves and current solution aren't
explicit enough, so add comments on them.
Cc: Bart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
blkg_conf_prep() currently calls blkg_lookup_create() while holding
request queue spinlock. This means allocating memory for struct
blkcg_gq has to be made non-blocking. This causes occasional -ENOMEM
failures in call paths like below:
pcpu_alloc+0x68f/0x710
__alloc_percpu_gfp+0xd/0x10
__percpu_counter_init+0x55/0xc0
cfq_pd_alloc+0x3b2/0x4e0
blkg_alloc+0x187/0x230
blkg_create+0x489/0x670
blkg_lookup_create+0x9a/0x230
blkg_conf_prep+0x1fb/0x240
__cfqg_set_weight_device.isra.105+0x5c/0x180
cfq_set_weight_on_dfl+0x69/0xc0
cgroup_file_write+0x39/0x1c0
kernfs_fop_write+0x13f/0x1d0
__vfs_write+0x23/0x120
vfs_write+0xc2/0x1f0
SyS_write+0x44/0xb0
entry_SYSCALL_64_fastpath+0x18/0xad
In the code path above, percpu allocator cannot call vmalloc() due to
queue spinlock.
A failure in this call path gives grief to tools which are trying to
configure io weights. We see occasional failures happen shortly after
reboots even when system is not under any memory pressure. Machines
with a lot of cpus are more vulnerable to this condition.
Update blkg_create() function to temporarily drop the rcu and queue
locks when it is allowed by gfp mask.
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Tahsin Erdogan <tahsin@google.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
One hard problem adding .low limit is to detect idle cgroup. If one
cgroup doesn't dispatch enough IO against its low limit, we must have a
mechanism to determine if other cgroups dispatch more IO. We added the
think time detection mechanism before, but it doesn't work for all
workloads. Here we add a latency based approach.
We already have mechanism to calculate latency threshold for each IO
size. For every IO dispatched from a cgorup, we compare its latency
against its threshold and record the info. If most IO latency is below
threshold (in the code I use 75%), the cgroup could be treated idle and
other cgroups can dispatch more IO.
Currently this latency target check is only for SSD as we can't
calcualte the latency target for hard disk. And this is only for cgroup
leaf node so far.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
User configures latency target, but the latency threshold for each
request size isn't fixed. For a SSD, the IO latency highly depends on
request size. To calculate latency threshold, we sample some data, eg,
average latency for request size 4k, 8k, 16k, 32k .. 1M. The latency
threshold of each request size will be the sample latency (I'll call it
base latency) plus latency target. For example, the base latency for
request size 4k is 80us and user configures latency target 60us. The 4k
latency threshold will be 80 + 60 = 140us.
To sample data, we calculate the order base 2 of rounded up IO sectors.
If the IO size is bigger than 1M, it will be accounted as 1M. Since the
calculation does round up, the base latency will be slightly smaller
than actual value. Also if there isn't any IO dispatched for a specific
IO size, we will use the base latency of smaller IO size for this IO
size.
But we shouldn't sample data at any time. The base latency is supposed
to be latency where disk isn't congested, because we use latency
threshold to schedule IOs between cgroups. If disk is congested, the
latency is higher, using it for scheduling is meaningless. Hence we only
do the sampling when block throttling is in the LOW limit, with
assumption disk isn't congested in such state. If the assumption isn't
true, eg, low limit is too high, calculated latency threshold will be
higher.
Hard disk is completely different. Latency depends on spindle seek
instead of request size. Currently this feature is SSD only, we probably
can use a fixed threshold like 4ms for hard disk though.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Currently there is no way to know the request size when the request is
finished. Next patch will need this info. We could add extra field to
record the size, but blk_issue_stat has enough space to record it, so
this patch just overloads blk_issue_stat. With this, we will have 49bits
to track time, which still is very long time.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Here we introduce per-cgroup latency target. The target determines how a
cgroup can afford latency increasement. We will use the target latency
to calculate a threshold and use it to schedule IO for cgroups. If a
cgroup's bandwidth is below its low limit but its average latency is
below the threshold, other cgroups can safely dispatch more IO even
their bandwidth is higher than their low limits. On the other hand, if
the first cgroup's latency is higher than the threshold, other cgroups
are throttled to their low limits. So the target latency determines how
we efficiently utilize free disk resource without sacifice of worload's
IO latency.
For example, assume 4k IO average latency is 50us when disk isn't
congested. A cgroup sets the target latency to 30us. Then the cgroup can
accept 50+30=80us IO latency. If the cgroupt's average IO latency is
90us and its bandwidth is below low limit, other cgroups are throttled
to their low limit. If the cgroup's average IO latency is 60us, other
cgroups are allowed to dispatch more IO. When other cgroups dispatch
more IO, the first cgroup's IO latency will increase. If it increases to
81us, we then throttle other cgroups.
User will configure the interface in this way:
echo "8:16 rbps=2097152 wbps=max latency=100 idle=200" > io.low
latency is in microsecond unit
By default, latency target is 0, which means to guarantee IO latency.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Last patch introduces a way to detect idle cgroup. We use it to make
upgrade/downgrade decision. And the new algorithm can detect completely
idle cgroup too, so we can delete the corresponding code.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Add interface to configure the threshold. The io.low interface will
like:
echo "8:16 rbps=2097152 wbps=max idle=2000" > io.low
idle is in microsecond unit.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
A cgroup gets assigned a low limit, but the cgroup could never dispatch
enough IO to cross the low limit. In such case, the queue state machine
will remain in LIMIT_LOW state and all other cgroups will be throttled
according to low limit. This is unfair for other cgroups. We should
treat the cgroup idle and upgrade the state machine to lower state.
We also have a downgrade logic. If the state machine upgrades because of
cgroup idle (real idle), the state machine will downgrade soon as the
cgroup is below its low limit. This isn't what we want. A more
complicated case is cgroup isn't idle when queue is in LIMIT_LOW. But
when queue gets upgraded to lower state, other cgroups could dispatch
more IO and this cgroup can't dispatch enough IO, so the cgroup is below
its low limit and looks like idle (fake idle). In this case, the queue
should downgrade soon. The key to determine if we should do downgrade is
to detect if cgroup is truely idle.
Unfortunately it's very hard to determine if a cgroup is real idle. This
patch uses the 'think time check' idea from CFQ for the purpose. Please
note, the idea doesn't work for all workloads. For example, a workload
with io depth 8 has disk utilization 100%, hence think time is 0, eg,
not idle. But the workload can run higher bandwidth with io depth 16.
Compared to io depth 16, the io depth 8 workload is idle. We use the
idea to roughly determine if a cgroup is idle.
We treat a cgroup idle if its think time is above a threshold (by
default 1ms for SSD and 100ms for HD). The idea is think time above the
threshold will start to harm performance. HD is much slower so a longer
think time is ok.
The patch (and the latter patches) uses 'unsigned long' to track time.
We convert 'ns' to 'us' with 'ns >> 10'. This is fast but loses
precision, should not a big deal.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
When cgroups all reach low limit, cgroups can dispatch more IO. This
could make some cgroups dispatch more IO but others not, and even some
cgroups could dispatch less IO than their low limit. For example, cg1
low limit 10MB/s, cg2 limit 80MB/s, assume disk maximum bandwidth is
120M/s for the workload. Their bps could something like this:
cg1/cg2 bps: T1: 10/80 -> T2: 60/60 -> T3: 10/80
At T1, all cgroups reach low limit, so they can dispatch more IO later.
Then cg1 dispatch more IO and cg2 has no room to dispatch enough IO. At
T2, cg2 only dispatches 60M/s. Since We detect cg2 dispatches less IO
than its low limit 80M/s, we downgrade the queue from LIMIT_MAX to
LIMIT_LOW, then all cgroups are throttled to their low limit (T3). cg2
will have bandwidth below its low limit at most time.
The big problem here is we don't know the maximum bandwidth of the
workload, so we can't make smart decision to avoid the situation. This
patch makes cgroup bandwidth change smooth. After disk upgrades from
LIMIT_LOW to LIMIT_MAX, we don't allow cgroups use all bandwidth upto
their max limit immediately. Their bandwidth limit will be increased
gradually to avoid above situation. So above example will became
something like:
cg1/cg2 bps: 10/80 -> 15/105 -> 20/100 -> 25/95 -> 30/90 -> 35/85 -> 40/80
-> 45/75 -> 22/98
In this way cgroups bandwidth will be above their limit in majority
time, this still doesn't fully utilize disk bandwidth, but that's
something we pay for sharing.
Scale up is linear. The limit scales up 1/2 .low limit every
throtl_slice after upgrade. The scale up will stop if the adjusted limit
hits .max limit. Scale down is exponential. We cut the scale value half
if a cgroup doesn't hit its .low limit. If the scale becomes 0, we then
fully downgrade the queue to LIMIT_LOW state.
Note this doesn't completely avoid cgroup running under its low limit.
The best way to guarantee cgroup doesn't run under its limit is to set
max limit. For example, if we set cg1 max limit to 40, cg2 will never
run under its low limit.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
cgroup could be assigned a limit, but doesn't dispatch enough IO, eg the
cgroup is idle. When this happens, the cgroup doesn't hit its limit, so
we can't move the state machine to higher level and all cgroups will be
throttled to their lower limit, so we waste bandwidth. Detecting idle
cgroup is hard. This patch handles a simple case, a cgroup doesn't
dispatch any IO. We ignore such cgroup's limit, so other cgroups can use
the bandwidth.
Please note this will be replaced with a more sophisticated algorithm
later, but this demonstrates the idea how we handle idle cgroups, so I
leave it here.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
The throtl_slice is 100ms by default. This is a long time for SSD, a lot
of IO can run. To make cgroups have smoother throughput, we choose a
small value (20ms) for SSD.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
throtl_slice is important for blk-throttling. It's called slice
internally but it really is a time window blk-throttling samples data.
blk-throttling will make decision based on the samplings. An example is
bandwidth measurement. A cgroup's bandwidth is measured in the time
interval of throtl_slice.
A small throtl_slice meanse cgroups have smoother throughput but burn
more CPUs. It has 100ms default value, which is not appropriate for all
disks. A fast SSD can dispatch a lot of IOs in 100ms. This patch makes
it tunable.
Since throtl_slice isn't a time slice, the sysfs name
'throttle_sample_time' reflects its character better.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
cgroup could be throttled to a limit but when all cgroups cross high
limit, queue enters a higher state and so the group should be throttled
to a higher limit. It's possible the cgroup is sleeping because of
throttle and other cgroups don't dispatch IO any more. In this case,
nobody can trigger current downgrade/upgrade logic. To fix this issue,
we could either set up a timer to wakeup the cgroup if other cgroups are
idle or make sure this cgroup doesn't sleep too long. Setting up a timer
means we must change the timer very frequently. This patch chooses the
latter. Making cgroup sleep time not too big wouldn't change cgroup
bps/iops, but could make it wakeup more frequently, which isn't a big
issue because throtl_slice * 8 is already quite big.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
When queue state machine is in LIMIT_MAX state, but a cgroup is below
its low limit for some time, the queue should be downgraded to lower
state as one cgroup's low limit isn't met.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
When queue is in LIMIT_LOW state and all cgroups with low limit cross
the bps/iops limitation, we will upgrade queue's state to
LIMIT_MAX. To determine if a cgroup exceeds its limitation, we check if
the cgroup has pending request. Since cgroup is throttled according to
the limit, pending request means the cgroup reaches the limit.
If a cgroup has limit set for both read and write, we consider the
combination of them for upgrade. The reason is read IO and write IO can
interfere with each other. If we do the upgrade based in one direction
IO, the other direction IO could be severly harmed.
For a cgroup hierarchy, there are two cases. Children has lower low
limit than parent. Parent's low limit is meaningless. If children's
bps/iops cross low limit, we can upgrade queue state. The other case is
children has higher low limit than parent. Children's low limit is
meaningless. As long as parent's bps/iops (which is a sum of childrens
bps/iops) cross low limit, we can upgrade queue state.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
each queue will have a state machine. Initially queue is in LIMIT_LOW
state, which means all cgroups will be throttled according to their low
limit. After all cgroups with low limit cross the limit, the queue state
gets upgraded to LIMIT_MAX state.
For max limit, cgroup will use the limit configured by user.
For low limit, cgroup will use the minimal value between low limit and
max limit configured by user. If the minimal value is 0, which means the
cgroup doesn't configure low limit, we will use max limit to throttle
the cgroup and the cgroup is ready to upgrade to LIMIT_MAX
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Add low limit for cgroup and corresponding cgroup interface. To be
consistent with memcg, we allow users configure .low limit higher than
.max limit. But the internal logic always assumes .low limit is lower
than .max limit. So we add extra bps/iops_conf fields in throtl_grp for
userspace configuration. Old bps/iops fields in throtl_grp will be the
actual limit we use for throttling.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
As discussed in LSF, add configure option for the interface and mark it
as experimental, so people can try/test.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
We are going to support low/max limit, each cgroup will have 2 limits
after that. This patch prepares for the multiple limits change.
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
blk_integrity_profile's are never modified, so mark them 'const' so that
they are placed in .rodata and benefit from memory protection.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
blkdev_issue_flush() is now always synchronous, and it no longer has a
flags argument. So remove the part of the comment about the WAIT flag.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
There isn't a bug here, but Smatch is not smart enough to know that
"nr_iovecs" can't be negative so it complains about underflows.
Really, it's slightly cleaner to make this parameter unsigned.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
Turn the different ways of merging or issuing I/O into a series of if/else
statements instead of the current maze of gotos. Note that this means we
pin the CPU a little longer for some cases as the CTX put is moved to
common code at the end of the function.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
Now that we have a nice direct issue heper this helps simplifying
the code a bit, and also gets rid of the old_rq variable.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
Rename blk_mq_try_issue_directly to __blk_mq_try_issue_directly and add a
new wrapper that takes care of RCU / SRCU locking to avoid having
boileplate code in the caller which would get duplicated with new callers.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
They are mostly the same code anyway - this just one small conditional
for the plug case that is different for both variants.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
This flag was never used since it was introduced.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
When device open races with device shutdown, we can get the following
oops in scsi_disk_get():
[11863.044351] general protection fault: 0000 [#1] SMP
[11863.045561] Modules linked in: scsi_debug xfs libcrc32c netconsole btrfs raid6_pq zlib_deflate lzo_compress xor [last unloaded: loop]
[11863.047853] CPU: 3 PID: 13042 Comm: hald-probe-stor Tainted: G W 4.10.0-rc2-xen+ #35
[11863.048030] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[11863.048030] task: ffff88007f438200 task.stack: ffffc90000fd0000
[11863.048030] RIP: 0010:scsi_disk_get+0x43/0x70
[11863.048030] RSP: 0018:ffffc90000fd3a08 EFLAGS: 00010202
[11863.048030] RAX: 6b6b6b6b6b6b6b6b RBX: ffff88007f56d000 RCX: 0000000000000000
[11863.048030] RDX: 0000000000000001 RSI: 0000000000000004 RDI: ffffffff81a8d880
[11863.048030] RBP: ffffc90000fd3a18 R08: 0000000000000000 R09: 0000000000000001
[11863.059217] R10: 0000000000000000 R11: 0000000000000000 R12: 00000000fffffffa
[11863.059217] R13: ffff880078872800 R14: ffff880070915540 R15: 000000000000001d
[11863.059217] FS: 00007f2611f71800(0000) GS:ffff88007f0c0000(0000) knlGS:0000000000000000
[11863.059217] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[11863.059217] CR2: 000000000060e048 CR3: 00000000778d4000 CR4: 00000000000006e0
[11863.059217] Call Trace:
[11863.059217] ? disk_get_part+0x22/0x1f0
[11863.059217] sd_open+0x39/0x130
[11863.059217] __blkdev_get+0x69/0x430
[11863.059217] ? bd_acquire+0x7f/0xc0
[11863.059217] ? bd_acquire+0x96/0xc0
[11863.059217] ? blkdev_get+0x350/0x350
[11863.059217] blkdev_get+0x126/0x350
[11863.059217] ? _raw_spin_unlock+0x2b/0x40
[11863.059217] ? bd_acquire+0x7f/0xc0
[11863.059217] ? blkdev_get+0x350/0x350
[11863.059217] blkdev_open+0x65/0x80
...
As you can see RAX value is already poisoned showing that gendisk we got
is already freed. The problem is that get_gendisk() looks up device
number in ext_devt_idr and then does get_disk() which does kobject_get()
on the disks kobject. However the disk gets removed from ext_devt_idr
only in disk_release() (through blk_free_devt()) at which moment it has
already 0 refcount and is already on its way to be freed. Indeed we've
got a warning from kobject_get() about 0 refcount shortly before the
oops.
We fix the problem by using kobject_get_unless_zero() in get_disk() so
that get_disk() cannot get reference on a disk that is already being
freed.
Tested-by: Lekshmi Pillai <lekshmicpillai@in.ibm.com>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <axboe@fb.com>
If a driver allocates a queue for stacked usage, then it does
not currently get stats allocated. This causes the later init
of, eg, writeback throttling to blow up. Move the init to the
queue allocation instead.
Additionally, allow a NULL callback unregistration. This avoids
having the caller check for that, fixing another oops on
removal of a block device that doesn't have poll stats allocated.
Fixes: 34dbad5d26 ("blk-stat: convert to callback-based statistics reporting")
Signed-off-by: Jens Axboe <axboe@fb.com>
Currently, statistics are gathered in ~0.13s windows, and users grab the
statistics whenever they need them. This is not ideal for both in-tree
users:
1. Writeback throttling wants its own dynamically sized window of
statistics. Since the blk-stats statistics are reset after every
window and the wbt windows don't line up with the blk-stats windows,
wbt doesn't see every I/O.
2. Polling currently grabs the statistics on every I/O. Again, depending
on how the window lines up, we may miss some I/Os. It's also
unnecessary overhead to get the statistics on every I/O; the hybrid
polling heuristic would be just as happy with the statistics from the
previous full window.
This reworks the blk-stats infrastructure to be callback-based: users
register a callback that they want called at a given time with all of
the statistics from the window during which the callback was active.
Users can dynamically bucketize the statistics. wbt and polling both
currently use read vs. write, but polling can be extended to further
subdivide based on request size.
The callbacks are kept on an RCU list, and each callback has percpu
stats buffers. There will only be a few users, so the overhead on the
I/O completion side is low. The stats flushing is also simplified
considerably: since the timer function is responsible for clearing the
statistics, we don't have to worry about stale statistics.
wbt is a trivial conversion. After the conversion, the windowing problem
mentioned above is fixed.
For polling, we register an extra callback that caches the previous
window's statistics in the struct request_queue for the hybrid polling
heuristic to use.
Since we no longer have a single stats buffer for the request queue,
this also removes the sysfs and debugfs stats entries. To replace those,
we add a debugfs entry for the poll statistics.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
This is an implementation detail that no-one outside of blk-stat.c uses.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
The stats buckets will become generic soon, so make the existing users
use the common READ and WRITE definitions instead of one internal to
blk-stat.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
We always call wbt_exit() from blk_release_queue(), so these are
unnecessary.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
We need to flush the batch _before_ we check the number of samples,
otherwise we'll miss all of the batched samples.
Fixes: cf43e6b ("block: add scalable completion tracking of requests")
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
If we have scheduling enabled, we jump directly to insert-and-run.
That's fine, but we run the queue async and we don't pass in information
on whether we can block from this context or not. Fixup both these
cases.
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
In case cpu was unplugged, we need to make sure not to assume
that the tags for that cpu are still allocated. so check
for null tags when reinitializing a tagset.
Reported-by: Yi Zhang <yizhan@redhat.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@fb.com>
Commit 79bd99596b ("blk: improve order of bio handling in generic_make_request()")
changed current->bio_list so that it did not contain *all* of the
queued bios, but only those submitted by the currently running
make_request_fn.
There are two places which walk the list and requeue selected bios,
and others that check if the list is empty. These are no longer
correct.
So redefine current->bio_list to point to an array of two lists, which
contain all queued bios, and adjust various code to test or walk both
lists.
Signed-off-by: NeilBrown <neilb@suse.com>
Fixes: 79bd99596b ("blk: improve order of bio handling in generic_make_request()")
Signed-off-by: Jens Axboe <axboe@fb.com>
To avoid recursion on the kernel stack when stacked block devices
are in use, generic_make_request() will, when called recursively,
queue new requests for later handling. They will be handled when the
make_request_fn for the current bio completes.
If any bios are submitted by a make_request_fn, these will ultimately
be handled seqeuntially. If the handling of one of those generates
further requests, they will be added to the end of the queue.
This strict first-in-first-out behaviour can lead to deadlocks in
various ways, normally because a request might need to wait for a
previous request to the same device to complete. This can happen when
they share a mempool, and can happen due to interdependencies
particular to the device. Both md and dm have examples where this happens.
These deadlocks can be erradicated by more selective ordering of bios.
Specifically by handling them in depth-first order. That is: when the
handling of one bio generates one or more further bios, they are
handled immediately after the parent, before any siblings of the
parent. That way, when generic_make_request() calls make_request_fn
for some particular device, we can be certain that all previously
submited requests for that device have been completely handled and are
not waiting for anything in the queue of requests maintained in
generic_make_request().
An easy way to achieve this would be to use a last-in-first-out stack
instead of a queue. However this will change the order of consecutive
bios submitted by a make_request_fn, which could have unexpected consequences.
Instead we take a slightly more complex approach.
A fresh queue is created for each call to a make_request_fn. After it completes,
any bios for a different device are placed on the front of the main queue, followed
by any bios for the same device, followed by all bios that were already on
the queue before the make_request_fn was called.
This provides the depth-first approach without reordering bios on the same level.
This, by itself, it not enough to remove all deadlocks. It just makes
it possible for drivers to take the extra step required themselves.
To avoid deadlocks, drivers must never risk waiting for a request
after submitting one to generic_make_request. This includes never
allocing from a mempool twice in the one call to a make_request_fn.
A common pattern in drivers is to call bio_split() in a loop, handling
the first part and then looping around to possibly split the next part.
Instead, a driver that finds it needs to split a bio should queue
(with generic_make_request) the second part, handle the first part,
and then return. The new code in generic_make_request will ensure the
requests to underlying bios are processed first, then the second bio
that was split off. If it splits again, the same process happens. In
each case one bio will be completely handled before the next one is attempted.
With this is place, it should be possible to disable the
punt_bios_to_recover() recovery thread for many block devices, and
eventually it may be possible to remove it completely.
Ref: http://www.spinics.net/lists/raid/msg54680.html
Tested-by: Jinpu Wang <jinpu.wang@profitbricks.com>
Inspired-by: Lars Ellenberg <lars.ellenberg@linbit.com>
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
This reverts commit 0dba1314d4. It causes
leaking of device numbers for SCSI when SCSI registers multiple gendisks
for one request_queue in succession. It can be easily reproduced using
Omar's script [1] on kernel with CONFIG_DEBUG_TEST_DRIVER_REMOVE.
Furthermore the protection provided by this commit is not needed anymore
as the problem it was fixing got also fixed by commit 165a5e22fa
"block: Move bdi_unregister() to del_gendisk()".
[1]: http://marc.info/?l=linux-block&m=148554717109098&w=2
Signed-off-by: Jan Kara <jack@suse.cz>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
