Have a common table of mappings from numerical transport ids to names, and
zero the transport specific area in common code in nvmet_addr_trtype_store.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
For consistancy reasons, any fabric-specific works
(e.g error recovery/reconnect) should be canceled in
nvme_stop_ctrl, as for all other NVMe pending works
(e.g. scan, keep alive).
The patch aims to simplify the logic of the code, as
we now only rely on a vague demand from any fabric
to flush its private workqueues at the beginning of
.delete_ctrl op.
Signed-off-by: Nitzan Carmi <nitzanc@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
While error recovery is ongoing, it is OK to move
ctrl to DELETING state (from concurrent delete_work).
Thus we don't need a warning for that case.
Signed-off-by: Nitzan Carmi <nitzanc@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If a task is holding a reference to a namespace on a removed controller,
the head will not be released. If the same controller is added again
later, its namespaces may not be successfully added. Instead, the user
will see kernel message "Duplicate IDs for nsid <X>".
This patch fixes that by skipping heads that don't have namespaces when
considering if a new namespace is safe to add.
Reported-by: Alex Gagniuc <Alex_Gagniuc@Dellteam.com>
Cc: stable@vger.kernel.org
Signed-off-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The .remove_one function is called for any ib_device removal.
In case the removed device has no reference in our driver, there
is no need to flush the work queue.
Reviewed-by: Israel Rukshin <israelr@mellanox.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The .remove_one function is called for any ib_device removal.
In case the removed device has no reference in our driver, there
is no need to flush the system work queue.
Reviewed-by: Israel Rukshin <israelr@mellanox.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We free nvmet rdma queues while handling rdma_cm events.
In order to avoid this we destroy the qp and the queue after destroying
the cm_id which guarantees that all rdma_cm events are done.
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When a bio completion calls back into the transport for a
back-end io device, the request completion path can free
the transport io job structure allowing it to be reused for
other operations. The transport has a defer_rcv queue which
holds temporary cmd rcv ops while waitng for io job structures.
when the job frees, if there's a cmd waiting, it is picked up
and submitted for processing, which can call back out to the
bio path if it's a read. Unfortunately, what is unknown is the
context of the original bio done call, and it may be in a state
(softirq) that is not compatible with submitting the new bio in
the same calling sequence. This is especially true when using
scsi back-end devices as scsi is in softirq when it makes the
done call.
Correct by scheduling the io to be started via workq rather
than calling the start new io path inline to the original bio
done path.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When reattaching to a removed remoteport that has not yet been
fully deleted as it's waiting for reconnect timeouts, be sure to
re-set the ports nport id and role.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Another abort race: An io request is started, becomes active,
and is attempted to be started with the lldd. At the same time
the controller is stopped/torndown and an itterator is run to
abort the ios. As the io is active, it is added to the outstanding
aborted io count. However on the original io request thread, the
driver ends up rejecting the io due to the condition that induced
the controller teardown. The driver reject path didn't check whether
it was in the outstanding io count. This left the count outstanding
stopping controller teardown.
Correct by, in the driver reject case, setting the state to
inactive and checking whether it was in the outstanding io count.
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The current nvme_fc code, when an io times out, will abort the io
on the fc link, then call the error recovery routine to reset the
controller. It is during the reset of the controller that the
transport will wait for all ios to be aborted before sending a
Disconnect LS to the target.
However, the reset routine only waits for the io which it generates
the abort for to complete. Any io that was aborted just prior to the
reset isn't in it's list to wait for. Thus the Disconnect is getting
sent before the aborts have completed.
Correct by removing the abort in the timeout handler. The reset will
generate the abort. At that point the timeout handler can be simplified
to request the reset (via the error handler) and restart the timeout
timer.
Also fixes a small typo in a comment in the reset handler.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If there are errors during initial controller create, the transport
will teardown the partially initialized controller struct and free
the ctlr memory. Trouble is - most of those errors can occur due
to asynchronous events happening such io timeouts and subsystem
connectivity failures. Those failures invoke async workq items to
reset the controller and attempt reconnect. Those may be in progress
as the main thread frees the ctrl memory, resulting in NULL ptr oops.
Prevent this from happening by having the main ctrl failure thread
changing state to DELETING followed by synchronously cancelling any
pending queued work item. The change of state will prevent the
scheduling of resets or reconnect events.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Yet another "incompatible" Samsung NVMe SSD 960 EVO and Asus motherboard
combination. 960 EVO device disappears from PCIe bus within few minutes
after boot-up when APST is in use and never gets back. Forcing
NVME_QUIRK_NO_APST is the only way to make this drive work with this
particular motherboard. NVME_QUIRK_NO_DEEPEST_PS doesn't work, upgrading
motherboard's BIOS didn't help either.
Since this is a desktop motherboard, the only drawback of not using APST
is increased device temperature.
Signed-off-by: Jarosław Janik <jaroslaw.janik@gmail.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
nvme_delete_ctrl can be called from various contexts in parallel,
and cause duplicated information prints, even though the specific
context doesn't perform the actual removal. Instead, print the
information when the actual removal occurs.
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The nvme-fabrics exports the controller address to sysfs, and we'd
like to have parity with this feature for PCIe. This patch provides
the appropiate callback and returns the controller address as the pci
domain🚌device.function.
Signed-off-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
NVMe 1.2.1 extends the get log page interface to include 64 bit
offset and increases the number of dwords to 32 bits. Implement
for future use.
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
namespaces_mutext is used to synchronize the operations on ctrl
namespaces list. Most of the time, it is a read operation.
On the other hand, there are many interfaces in nvme core that
need this lock, such as nvme_wait_freeze, and even more interfaces
will be added. If we use mutex here, circular dependency could be
introduced easily. For example:
context A context B
nvme_xxx nvme_xxx
hold namespaces_mutext require namespaces_mutext
sync context B
So it is better to change it from mutex to rwsem.
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
nvme_remove_namespaces and nvme_remove_invalid_namespaces reference
the ctrl->namespaces list w/o holding namespaces_mutext. It is ok
to invoke nvme_ns_remove there, but what if there is others.
To be safer, reference the ctrl->namespaces list under
namespaces_mutext.
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Quiesce IO queues prior to disabling device HMB accesses. A controller
using HMB may relay on it to efficiently complete IO commands.
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add examples to show how to use nvme fault injection.
Signed-off-by: Thomas Tai <thomas.tai@oracle.com>
Reviewed-by: Eric Saint-Etienne <eric.saint.etienne@oracle.com>
Signed-off-by: Karl Volz <karl.volz@oracle.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Linux's fault injection framework provides a systematic way to support
error injection via debugfs in the /sys/kernel/debug directory. This
patch uses the framework to add error injection to NVMe driver. The
fault injection source code is stored in a separate file and only linked
if CONFIG_FAULT_INJECTION_DEBUG_FS kernel config is selected.
Once the error injection is enabled, NVME_SC_INVALID_OPCODE with no
retry will be injected into the nvme_end_request. Users can change
the default status code and no retry flag via debufs. Following example
shows how to enable and inject an error. For more examples, refer to
Documentation/fault-injection/nvme-fault-injection.txt
How to enable nvme fault injection:
First, enable CONFIG_FAULT_INJECTION_DEBUG_FS kernel config,
recompile the kernel. After booting up the kernel, do the
following.
How to inject an error:
mount /dev/nvme0n1 /mnt
echo 1 > /sys/kernel/debug/nvme0n1/fault_inject/times
echo 100 > /sys/kernel/debug/nvme0n1/fault_inject/probability
cp a.file /mnt
Expected Result:
cp: cannot stat ‘/mnt/a.file’: Input/output error
Message from dmesg:
FAULT_INJECTION: forcing a failure.
name fault_inject, interval 1, probability 100, space 0, times 1
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.15.0-rc8+ #2
Hardware name: innotek GmbH VirtualBox/VirtualBox,
BIOS VirtualBox 12/01/2006
Call Trace:
<IRQ>
dump_stack+0x5c/0x7d
should_fail+0x148/0x170
nvme_should_fail+0x2f/0x50 [nvme_core]
nvme_process_cq+0xe7/0x1d0 [nvme]
nvme_irq+0x1e/0x40 [nvme]
__handle_irq_event_percpu+0x3a/0x190
handle_irq_event_percpu+0x30/0x70
handle_irq_event+0x36/0x60
handle_fasteoi_irq+0x78/0x120
handle_irq+0xa7/0x130
? tick_irq_enter+0xa8/0xc0
do_IRQ+0x43/0xc0
common_interrupt+0xa2/0xa2
</IRQ>
RIP: 0010:native_safe_halt+0x2/0x10
RSP: 0018:ffffffff82003e90 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffdd
RAX: ffffffff817a10c0 RBX: ffffffff82012480 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000000 R08: 000000008e38ce64 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff82012480
R13: ffffffff82012480 R14: 0000000000000000 R15: 0000000000000000
? __sched_text_end+0x4/0x4
default_idle+0x18/0xf0
do_idle+0x150/0x1d0
cpu_startup_entry+0x6f/0x80
start_kernel+0x4c4/0x4e4
? set_init_arg+0x55/0x55
secondary_startup_64+0xa5/0xb0
print_req_error: I/O error, dev nvme0n1, sector 9240
EXT4-fs error (device nvme0n1): ext4_find_entry:1436:
inode #2: comm cp: reading directory lblock 0
Signed-off-by: Thomas Tai <thomas.tai@oracle.com>
Reviewed-by: Eric Saint-Etienne <eric.saint.etienne@oracle.com>
Signed-off-by: Karl Volz <karl.volz@oracle.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
NVME_IDENTIFY_DATA_SIZE was added to linux/nvme.h by following commit.
commit 0add5e8e58 ("nvmet: use NVME_IDENTIFY_DATA_SIZE")
Make it use NVME_IDENTIFY_DATA_SIZE define instead of magic value
0x1000 in case of identify data size.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Minwoo Im <minwoo.im.dev@gmail.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Its perfectly valid to assign a nvmet port to listen on "any"
IP address (traddr 0.0.0.0 for ipv4 address family) for IP based
transport ports. However, we must not return this address in
discovery log entries. Instead we need to return the address
where the request was accepted on (req->port address).
Since this is nvme transport specific, introduce an optional
.disc_traddr interface that is designed to check that a
port in question is bound to "any" IP address and if so, set
the traddr from the port where the request came from.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Can be useful to check INET_ANY address for both ipv4/ipv6 addresses.
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
I'm getting a slab named "biovec-(1<<(21-12))". It is caused by unintended
expansion of the macro BIO_MAX_PAGES. This patch renames it to biovec-max.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org # v4.14+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Early alpha processors cannot write a single byte or word; they read 8
bytes, modify the value in registers and write back 8 bytes.
The type blk_status_t is defined as one byte, it is often written
asynchronously by I/O completion routines, this asynchronous modification
can corrupt content of nearby bytes if these nearby bytes can be written
simultaneously by another CPU.
- one example of such corruption is the structure dm_io where
"blk_status_t status" is written by an asynchronous completion routine
and "atomic_t io_count" is modified synchronously
- another example is the structure dm_buffer where "unsigned hold_count"
is modified synchronously from process context and "blk_status_t
write_error" is modified asynchronously from bio completion routine
This patch fixes the bug by changing the type blk_status_t to 32 bits if
we are on Alpha and if we are compiling for a processor that doesn't have
the byte-word-extension.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org # 4.13+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
scsi_device_quiesce() uses synchronize_rcu() to guarantee that the
effect of blk_set_preempt_only() will be visible for percpu_ref_tryget()
calls that occur after the queue unfreeze by using the approach
explained in https://lwn.net/Articles/573497/. The rcu read lock and
unlock calls in blk_queue_enter() form a pair with the synchronize_rcu()
call in scsi_device_quiesce(). Both scsi_device_quiesce() and
blk_queue_enter() must either use regular RCU or RCU-sched.
Since neither the RCU-protected code in blk_queue_enter() nor
blk_queue_usage_counter_release() sleeps, regular RCU protection
is sufficient. Note: scsi_device_quiesce() does not have to be
modified since it already uses synchronize_rcu().
Reported-by: Tejun Heo <tj@kernel.org>
Fixes: 3a0a529971 ("block, scsi: Make SCSI quiesce and resume work reliably")
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Oleksandr Natalenko <oleksandr@natalenko.name>
Cc: Martin Steigerwald <martin@lichtvoll.de>
Cc: stable@vger.kernel.org # v4.15
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Avoid that building with W=1 triggers the following compiler warning:
drivers/md/bcache/super.c:776:20: warning: comparison is always false due to limited range of data type [-Wtype-limits]
d->nr_stripes > SIZE_MAX / sizeof(atomic_t)) {
^
Reviewed-by: Coly Li <colyli@suse.de>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add more annotations for sparse to inform it about which functions do
not have the same number of spin_lock() and spin_unlock() calls.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This patch does not change any functionality.
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Reviewed-by: Coly Li <colyli@suse.de>
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Avoid that building with W=1 triggers warnings about the kernel-doc
headers.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This patch avoids that building with W=1 triggers complaints about
switch fall-throughs.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Make it possible for the compiler to verify the consistency of the
format string passed to __bch_check_keys() and the arguments that
should be formatted according to that format string.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This patch avoids that smatch complains about inconsistent indentation.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Reviewed-by: Coly Li <colyli@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If a bcache device is configured to writeback mode, current code does not
handle write I/O errors on backing devices properly.
In writeback mode, write request is written to cache device, and
latter being flushed to backing device. If I/O failed when writing from
cache device to the backing device, bcache code just ignores the error and
upper layer code is NOT noticed that the backing device is broken.
This patch tries to handle backing device failure like how the cache device
failure is handled,
- Add a error counter 'io_errors' and error limit 'error_limit' in struct
cached_dev. Add another io_disable to struct cached_dev to disable I/Os
on the problematic backing device.
- When I/O error happens on backing device, increase io_errors counter. And
if io_errors reaches error_limit, set cache_dev->io_disable to true, and
stop the bcache device.
The result is, if backing device is broken of disconnected, and I/O errors
reach its error limit, backing device will be disabled and the associated
bcache device will be removed from system.
Changelog:
v2: remove "bcache: " prefix in pr_error(), and use correct name string to
print out bcache device gendisk name.
v1: indeed this is new added in v2 patch set.
Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Cc: Michael Lyle <mlyle@lyle.org>
Cc: Junhui Tang <tang.junhui@zte.com.cn>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In order to catch I/O error of backing device, a separate bi_end_io
call back is required. Then a per backing device counter can record I/O
errors number and retire the backing device if the counter reaches a
per backing device I/O error limit.
This patch adds backing_request_endio() to bcache backing device I/O code
path, this is a preparation for further complicated backing device failure
handling. So far there is no real code logic change, I make this change a
separate patch to make sure it is stable and reliable for further work.
Changelog:
v2: Fix code comments typo, remove a redundant bch_writeback_add() line
added in v4 patch set.
v1: indeed this is new added in this patch set.
[mlyle: truncated commit subject]
Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Cc: Junhui Tang <tang.junhui@zte.com.cn>
Cc: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In current code closure debug file is outside of debug directory
and when unloading module there is lack of removing operation
for closure debug file, so it will cause creating error when trying
to reload module.
This patch move closure debug file into "bcache" debug direcory
so that the file can get deleted properly.
Signed-off-by: Chengguang Xu <cgxu519@gmx.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Reviewed-by: Tang Junhui <tang.junhui@zte.com.cn>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In bch_mca_scan(), There are some confusion and logical error in the use of
loop variables. In this patch, we clarify them as:
1) nr: the number of btree nodes needs to scan, which will decrease after
we scan a btree node, and should not be less than 0;
2) i: the number of btree nodes have scanned, includes both
btree_cache_freeable and btree_cache, which should not be bigger than
btree_cache_used;
3) freed: the number of btree nodes have freed.
Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In bch_mca_scan(), the return value should not be the number of freed btree
nodes, but the number of pages of freed btree nodes.
Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Stripe size is shown as zero when no strip in back end device:
[root@ceph132 ~]# cat /sys/block/sdd/bcache/stripe_size
0.0k
Actually it should be 1T Bytes (1 << 31 sectors), but in sysfs
interface, stripe_size was changed from sectors to bytes, and move
9 bits left, so the 32 bits variable overflows.
This patch change the variable to a 64 bits type before moving bits.
Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When we run IO in a detached device, and run iostat to shows IO status,
normally it will show like bellow (Omitted some fields):
Device: ... avgrq-sz avgqu-sz await r_await w_await svctm %util
sdd ... 15.89 0.53 1.82 0.20 2.23 1.81 52.30
bcache0 ... 15.89 115.42 0.00 0.00 0.00 2.40 69.60
but after IO stopped, there are still very big avgqu-sz and %util
values as bellow:
Device: ... avgrq-sz avgqu-sz await r_await w_await svctm %util
bcache0 ... 0 5326.32 0.00 0.00 0.00 0.00 100.10
The reason for this issue is that, only generic_start_io_acct() called
and no generic_end_io_acct() called for detached device in
cached_dev_make_request(). See the code:
//start generic_start_io_acct()
generic_start_io_acct(q, rw, bio_sectors(bio), &d->disk->part0);
if (cached_dev_get(dc)) {
//will callback generic_end_io_acct()
}
else {
//will not call generic_end_io_acct()
}
This patch calls generic_end_io_acct() in the end of IO for detached
devices, so we can show IO state correctly.
(Modified to use GFP_NOIO in kzalloc() by Coly Li)
Changelog:
v2: fix typo.
v1: the initial version.
Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Coly Li <colyli@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When there are too many I/O errors on cache device, current bcache code
will retire the whole cache set, and detach all bcache devices. But the
detached bcache devices are not stopped, which is problematic when bcache
is in writeback mode.
If the retired cache set has dirty data of backing devices, continue
writing to bcache device will write to backing device directly. If the
LBA of write request has a dirty version cached on cache device, next time
when the cache device is re-registered and backing device re-attached to
it again, the stale dirty data on cache device will be written to backing
device, and overwrite latest directly written data. This situation causes
a quite data corruption.
But we cannot simply stop all attached bcache devices when the cache set is
broken or disconnected. For example, use bcache to accelerate performance
of an email service. In such workload, if cache device is broken but no
dirty data lost, keep the bcache device alive and permit email service
continue to access user data might be a better solution for the cache
device failure.
Nix <nix@esperi.org.uk> points out the issue and provides the above example
to explain why it might be necessary to not stop bcache device for broken
cache device. Pavel Goran <via-bcache@pvgoran.name> provides a brilliant
suggestion to provide "always" and "auto" options to per-cached device
sysfs file stop_when_cache_set_failed. If cache set is retiring and the
backing device has no dirty data on cache, it should be safe to keep the
bcache device alive. In this case, if stop_when_cache_set_failed is set to
"auto", the device failure handling code will not stop this bcache device
and permit application to access the backing device with a unattached
bcache device.
Changelog:
[mlyle: edited to not break string constants across lines]
v3: fix typos pointed out by Nix.
v2: change option values of stop_when_cache_set_failed from 1/0 to
"auto"/"always".
v1: initial version, stop_when_cache_set_failed can be 0 (not stop) or 1
(always stop).
Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Michael Lyle <mlyle@lyle.org>
Cc: Nix <nix@esperi.org.uk>
Cc: Pavel Goran <via-bcache@pvgoran.name>
Cc: Junhui Tang <tang.junhui@zte.com.cn>
Cc: Hannes Reinecke <hare@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When too many I/Os failed on cache device, bch_cache_set_error() is called
in the error handling code path to retire whole problematic cache set. If
new I/O requests continue to come and take refcount dc->count, the cache
set won't be retired immediately, this is a problem.
Further more, there are several kernel thread and self-armed kernel work
may still running after bch_cache_set_error() is called. It needs to wait
quite a while for them to stop, or they won't stop at all. They also
prevent the cache set from being retired.
The solution in this patch is, to add per cache set flag to disable I/O
request on this cache and all attached backing devices. Then new coming I/O
requests can be rejected in *_make_request() before taking refcount, kernel
threads and self-armed kernel worker can stop very fast when flags bit
CACHE_SET_IO_DISABLE is set.
Because bcache also do internal I/Os for writeback, garbage collection,
bucket allocation, journaling, this kind of I/O should be disabled after
bch_cache_set_error() is called. So closure_bio_submit() is modified to
check whether CACHE_SET_IO_DISABLE is set on cache_set->flags. If set,
closure_bio_submit() will set bio->bi_status to BLK_STS_IOERR and
return, generic_make_request() won't be called.
A sysfs interface is also added to set or clear CACHE_SET_IO_DISABLE bit
from cache_set->flags, to disable or enable cache set I/O for debugging. It
is helpful to trigger more corner case issues for failed cache device.
Changelog
v4, add wait_for_kthread_stop(), and call it before exits writeback and gc
kernel threads.
v3, change CACHE_SET_IO_DISABLE from 4 to 3, since it is bit index.
remove "bcache: " prefix when printing out kernel message.
v2, more changes by previous review,
- Use CACHE_SET_IO_DISABLE of cache_set->flags, suggested by Junhui.
- Check CACHE_SET_IO_DISABLE in bch_btree_gc() to stop a while-loop, this
is reported and inspired from origal patch of Pavel Vazharov.
v1, initial version.
Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Cc: Junhui Tang <tang.junhui@zte.com.cn>
Cc: Michael Lyle <mlyle@lyle.org>
Cc: Pavel Vazharov <freakpv@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
struct delayed_work writeback_rate_update in struct cache_dev is a delayed
worker to call function update_writeback_rate() in period (the interval is
defined by dc->writeback_rate_update_seconds).
When a metadate I/O error happens on cache device, bcache error handling
routine bch_cache_set_error() will call bch_cache_set_unregister() to
retire whole cache set. On the unregister code path, this delayed work is
stopped by calling cancel_delayed_work_sync(&dc->writeback_rate_update).
dc->writeback_rate_update is a special delayed work from others in bcache.
In its routine update_writeback_rate(), this delayed work is re-armed
itself. That means when cancel_delayed_work_sync() returns, this delayed
work can still be executed after several seconds defined by
dc->writeback_rate_update_seconds.
The problem is, after cancel_delayed_work_sync() returns, the cache set
unregister code path will continue and release memory of struct cache set.
Then the delayed work is scheduled to run, __update_writeback_rate()
will reference the already released cache_set memory, and trigger a NULL
pointer deference fault.
This patch introduces two more bcache device flags,
- BCACHE_DEV_WB_RUNNING
bit set: bcache device is in writeback mode and running, it is OK for
dc->writeback_rate_update to re-arm itself.
bit clear:bcache device is trying to stop dc->writeback_rate_update,
this delayed work should not re-arm itself and quit.
- BCACHE_DEV_RATE_DW_RUNNING
bit set: routine update_writeback_rate() is executing.
bit clear: routine update_writeback_rate() quits.
This patch also adds a function cancel_writeback_rate_update_dwork() to
wait for dc->writeback_rate_update quits before cancel it by calling
cancel_delayed_work_sync(). In order to avoid a deadlock by unexpected
quit dc->writeback_rate_update, after time_out seconds this function will
give up and continue to call cancel_delayed_work_sync().
And here I explain how this patch stops self re-armed delayed work properly
with the above stuffs.
update_writeback_rate() sets BCACHE_DEV_RATE_DW_RUNNING at its beginning
and clears BCACHE_DEV_RATE_DW_RUNNING at its end. Before calling
cancel_writeback_rate_update_dwork() clear flag BCACHE_DEV_WB_RUNNING.
Before calling cancel_delayed_work_sync() wait utill flag
BCACHE_DEV_RATE_DW_RUNNING is clear. So when calling
cancel_delayed_work_sync(), dc->writeback_rate_update must be already re-
armed, or quite by seeing BCACHE_DEV_WB_RUNNING cleared. In both cases
delayed work routine update_writeback_rate() won't be executed after
cancel_delayed_work_sync() returns.
Inside update_writeback_rate() before calling schedule_delayed_work(), flag
BCACHE_DEV_WB_RUNNING is checked before. If this flag is cleared, it means
someone is about to stop the delayed work. Because flag
BCACHE_DEV_RATE_DW_RUNNING is set already and cancel_delayed_work_sync()
has to wait for this flag to be cleared, we don't need to worry about race
condition here.
If update_writeback_rate() is scheduled to run after checking
BCACHE_DEV_RATE_DW_RUNNING and before calling cancel_delayed_work_sync()
in cancel_writeback_rate_update_dwork(), it is also safe. Because at this
moment BCACHE_DEV_WB_RUNNING is cleared with memory barrier. As I mentioned
previously, update_writeback_rate() will see BCACHE_DEV_WB_RUNNING is clear
and quit immediately.
Because there are more dependences inside update_writeback_rate() to struct
cache_set memory, dc->writeback_rate_update is not a simple self re-arm
delayed work. After trying many different methods (e.g. hold dc->count, or
use locks), this is the only way I can find which works to properly stop
dc->writeback_rate_update delayed work.
Changelog:
v3: change values of BCACHE_DEV_WB_RUNNING and BCACHE_DEV_RATE_DW_RUNNING
to bit index, for test_bit().
v2: Try to fix the race issue which is pointed out by Junhui.
v1: The initial version for review
Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Junhui Tang <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Cc: Michael Lyle <mlyle@lyle.org>
Cc: Hannes Reinecke <hare@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
