- 11 Feb, 2019 6 commits
-
-
Hans Holmberg authored
pblk stripes writes of minimal write size across all non-offline chunks in a line, which means that the maximum write pointer delta should not exceed the minimal write size. Extend the line write pointer balance check to cover this case, and ignore the offline chunk wps. This will render us a warning during recovery if something unexpected has happened to the chunk write pointers (i.e. powerloss, a spurious chunk reset, ..). Reported-by: Zhoujie Wu <zjwu@marvell.com> Tested-by: Zhoujie Wu <zjwu@marvell.com> Reviewed-by: Javier González <javier@javigon.com> Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Masahiro Yamada authored
As the comment block in include/trace/define_trace.h says, TRACE_INCLUDE_PATH should be a relative path to the define_trace.h ../../drivers/lightnvm is the correct relative path. ../../../drivers/lightnvm is working by coincidence because the top Makefile adds -I$(srctree)/arch/$(SRCARCH)/include as a header search path, but we should not rely on it. Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Reviewed-by: Hans Holmberg <hans.holmberg@cnexlabs.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Andy Shevchenko authored
There are new types and helpers that are supposed to be used in new code. As a preparation to get rid of legacy types and API functions do the conversion here. Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Javier González <javier@javigon.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Andy Shevchenko authored
Sparse complains about using strict data types: drivers/lightnvm/pblk-read.c:254:43: warning: incorrect type in assignment (different base types) drivers/lightnvm/pblk-read.c:254:43: expected restricted __le64 <noident> drivers/lightnvm/pblk-read.c:254:43: got unsigned long long [unsigned] [usertype] <noident> drivers/lightnvm/pblk-read.c:255:29: warning: cast from restricted __le64 drivers/lightnvm/pblk-read.c:268:29: warning: cast from restricted __le64 drivers/lightnvm/pblk-read.c:328:41: warning: incorrect type in assignment (different base types) drivers/lightnvm/pblk-read.c:328:41: expected restricted __le64 <noident> drivers/lightnvm/pblk-read.c:328:41: got unsigned long long [unsigned] [usertype] <noident> In the code it seems explicit that lba_list_mem and lba_list_media members of struct pblk_pr_ctx are used on CPU side, which means they should not be of strict types. Change types of lba_list_mem and lba_list_media members to be u64. Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Javier González <javier@javigon.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Hans Holmberg authored
As chunk metadata is allocated using vmalloc, we need to free it using vfree. Fixes: 090ee26f ("lightnvm: use internal allocation for chunk log page") Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Hans Holmberg authored
pblk_line_meta_free might sleep (it can end up calling vfree, depending on how we allocate lba lists), and this can lead to a BUG() if we wake up on a different cpu and release the lock. As there is no point of grabbing the free lock when pblk has shut down, remove the lock. Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com> Signed-off-by: Matias Bjørling <mb@lightnvm.io> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
- 10 Feb, 2019 2 commits
-
-
Marcos Paulo de Souza authored
The Notes section of the comment was removed, because now blk_release_queue can only be executed from blk_cleanup_queue (being called when the q->kobj reaches zero), and also blk_init_queue was removed in a1ce35fa. Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Marcos Paulo de Souza authored
Since 4cf6324b, a portion of function blk_cleanup_queue was moved to a newly created function called blk_exit_queue, including the call of blkcg_exit_queue. So, adjust the documenation according. Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
- 09 Feb, 2019 21 commits
-
-
Jens Axboe authored
We have QUEUE_FLAG_DEFAULT defined, but it's not used anymore since the legacy IO stack is gone. Kill it. Sanitize the queue flags in general, they use spaces (for some reason), and the space is pretty sparse. With the flags renumbered, we can more clearly see how many we have available. Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Jens Axboe authored
We have various helpers for setting/clearing this flag, and also a helper to check if the queue supports queueable flushes or not. But nobody uses them anymore, kill it with fire. Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
In 'commit 752f66a7 ("bcache: use REQ_PRIO to indicate bio for metadata")' REQ_META is replaced by REQ_PRIO to indicate metadata bio. This assumption is not always correct, e.g. XFS uses REQ_META to mark metadata bio other than REQ_PRIO. This is why Nix noticed that bcache does not cache metadata for XFS after the above commit. Thanks to Dave Chinner, he explains the difference between REQ_META and REQ_PRIO from view of file system developer. Here I quote part of his explanation from mailing list, REQ_META is used for metadata. REQ_PRIO is used to communicate to the lower layers that the submitter considers this IO to be more important that non REQ_PRIO IO and so dispatch should be expedited. IOWs, if the filesystem considers metadata IO to be more important that user data IO, then it will use REQ_PRIO | REQ_META rather than just REQ_META. Then it seems bios with REQ_META or REQ_PRIO should both be cached for performance optimation, because they are all probably low I/O latency demand by upper layer (e.g. file system). So in this patch, when we want to decide whether to bypass the cache, REQ_META and REQ_PRIO are both checked. Then both metadata and high priority I/O requests will be handled properly. Reported-by: Nix <nix@esperi.org.uk> Signed-off-by: Coly Li <colyli@suse.de> Reviewed-by: Andre Noll <maan@tuebingen.mpg.de> Tested-by: Nix <nix@esperi.org.uk> Cc: stable@vger.kernel.org Cc: Dave Chinner <david@fromorbit.com> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
Cache set sysfs entry io_error_halflife is used to set c->error_decay. c->error_decay is in type unsigned int, and it is converted by strtoul_or_return(), therefore overflow to c->error_decay is possible for a large input value. This patch fixes the overflow by using strtoul_safe_clamp() to convert input string to an unsigned long value in range [0, UINT_MAX], then divides by 88 and set it to c->error_decay. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
c->error_limit is in type unsigned int, it is set via cache set sysfs file io_error_limit. Inside the bcache code, input string is converted by strtoul_or_return() and set the converted value to c->error_limit. Because the converted value is unsigned long, and c->error_limit is unsigned int, if the input is large enought, overflow will happen to c->error_limit. This patch uses sysfs_strtoul_clamp() to convert input string, and set the range in [0, UINT_MAX] to avoid the potential overflow. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
c->journal_delay_ms is in type unsigned short, it is set via sysfs interface and converted by sysfs_strtoul() from input string to unsigned short value. Therefore overflow to unsigned short might be happen when the converted value exceed USHRT_MAX. e.g. writing 65536 into sysfs file journal_delay_ms, c->journal_delay_ms is set to 0. This patch uses sysfs_strtoul_clamp() to convert the input string and limit value range in [0, USHRT_MAX], to avoid the input overflow. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
dc->writeback_rate_minimum is type unsigned integer variable, it is set via sysfs interface, and converte from input string to unsigned integer by d_strtoul_nonzero(). When the converted input value is larger than UINT_MAX, overflow to unsigned integer happens. This patch fixes the overflow by using sysfs_strotoul_clamp() to convert input string and limit the value in range [1, UINT_MAX], then the overflow can be avoided. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
Current code already uses d_strtoul_nonzero() to convert input string to an unsigned integer, to make sure writeback_rate_p_term_inverse won't be zero value. But overflow may happen when converting input string to an unsigned integer value by d_strtoul_nonzero(), then dc->writeback_rate_p_term_inverse can still be set to 0 even if the sysfs file input value is not zero, e.g. 4294967296 (a.k.a UINT_MAX+1). If dc->writeback_rate_p_term_inverse is set to 0, it might cause a dev-zero error in following code from __update_writeback_rate(), int64_t proportional_scaled = div_s64(error, dc->writeback_rate_p_term_inverse); This patch replaces d_strtoul_nonzero() by sysfs_strtoul_clamp() and limit the value range in [1, UINT_MAX]. Then the unsigned integer overflow and dev-zero error can be avoided. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
dc->writeback_rate_i_term_inverse can be set via sysfs interface. It is in type unsigned int, and convert from input string by d_strtoul(). The problem is d_strtoul() does not check valid range of the input, if 4294967296 is written into sysfs file writeback_rate_i_term_inverse, an overflow of unsigned integer will happen and value 0 is set to dc->writeback_rate_i_term_inverse. In writeback.c:__update_writeback_rate(), there are following lines of code, integral_scaled = div_s64(dc->writeback_rate_integral, dc->writeback_rate_i_term_inverse); If dc->writeback_rate_i_term_inverse is set to 0 via sysfs interface, a div-zero error might be triggered in the above code. Therefore we need to add a range limitation in the sysfs interface, this is what this patch does, use sysfs_stroul_clamp() to replace d_strtoul() and restrict the input range in [1, UINT_MAX]. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
Sysfs file writeback_delay is used to configure dc->writeback_delay which is type unsigned int. But bcache code uses sysfs_strtoul() to convert the input string, therefore it might be overflowed if the input value is too large. E.g. input value is 4294967296 but indeed 0 is set to dc->writeback_delay. This patch uses sysfs_strtoul_clamp() to convert the input string and set the result value range in [0, UINT_MAX] to avoid such unsigned integer overflow. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
When setting bcache parameters via sysfs, there are some variables are defined as bit-field value. Current bcache code in sysfs.c uses either d_strtoul() or sysfs_strtoul() to convert the input string to unsigned integer value and set it to the corresponded bit-field value. The problem is, the bit-field value only takes the lowest bit of the converted value. If input is 2, the expected value (like bool value) of the bit-field value should be 1, but indeed it is 0. The following sysfs files for bit-field variables have such problem, bypass_torture_test, for dc->bypass_torture_test writeback_metadata, for dc->writeback_metadata writeback_running, for dc->writeback_running verify, for c->verify key_merging_disabled, for c->key_merging_disabled gc_always_rewrite, for c->gc_always_rewrite btree_shrinker_disabled,for c->shrinker_disabled copy_gc_enabled, for c->copy_gc_enabled This patch uses sysfs_strtoul_bool() to set such bit-field variables, then if the converted value is non-zero, the bit-field variables will be set to 1, like setting a bool value like expensive_debug_checks. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
When setting bool values via sysfs interface, e.g. writeback_metadata, if writing 1 into writeback_metadata file, dc->writeback_metadata is set to 1, but if writing 2 into the file, dc->writeback_metadata is 0. This is misleading, a better result should be 1 for all non-zero input value. It is because dc->writeback_metadata is a bit-field variable, and current code simply use d_strtoul() to convert a string into integer and takes the lowest bit value. To fix such error, we need a routine to convert the input string into unsigned integer, and set target variable to 1 if the converted integer is non-zero. This patch introduces a new macro called sysfs_strtoul_bool(), it can be used to convert input string into bool value, we can use it to set bool value for bit-field vairables. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
People may set sequential_cutoff of a cached device via sysfs file, but current code does not check input value overflow. E.g. if value 4294967295 (UINT_MAX) is written to file sequential_cutoff, its value is 4GB, but if 4294967296 (UINT_MAX + 1) is written into, its value will be 0. This is an unexpected behavior. This patch replaces d_strtoi_h() by sysfs_strtoul_clamp() to convert input string to unsigned integer value, and limit its range in [0, UINT_MAX]. Then the input overflow can be fixed. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
Cache set congested threshold values congested_read_threshold_us and congested_write_threshold_us can be set via sysfs interface. These two values are 'unsigned int' type, but sysfs interface uses strtoul to convert input string. So if people input a large number like 9999999999, the value indeed set is 1410065407, which is not expected behavior. This patch replaces sysfs_strtoul() by sysfs_strtoul_clamp() when convert input string to unsigned int value, and set value range in [0, UINT_MAX], to avoid the above integer overflow errors. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
Currently sysfs_strtoul_clamp() is defined as, 82 #define sysfs_strtoul_clamp(file, var, min, max) \ 83 do { \ 84 if (attr == &sysfs_ ## file) \ 85 return strtoul_safe_clamp(buf, var, min, max) \ 86 ?: (ssize_t) size; \ 87 } while (0) The problem is, if bit width of var is less then unsigned long, min and max may not protect var from integer overflow, because overflow happens in strtoul_safe_clamp() before checking min and max. To fix such overflow in sysfs_strtoul_clamp(), to make min and max take effect, this patch adds an unsigned long variable, and uses it to macro strtoul_safe_clamp() to convert an unsigned long value in range defined by [min, max]. Then assign this value to var. By this method, if bit width of var is less than unsigned long, integer overflow won't happen before min and max are checking. Now sysfs_strtoul_clamp() can properly handle smaller data type like unsigned int, of cause min and max should be defined in range of unsigned int too. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Tang Junhui authored
Stale && dirty keys can be produced in the follow way: After writeback in write_dirty_finish(), dirty keys k1 will replace by clean keys k2 ==>ret = bch_btree_insert(dc->disk.c, &keys, NULL, &w->key); ==>btree_insert_fn(struct btree_op *b_op, struct btree *b) ==>static int bch_btree_insert_node(struct btree *b, struct btree_op *op, struct keylist *insert_keys, atomic_t *journal_ref, Then two steps: A) update k1 to k2 in btree node memory; bch_btree_insert_keys(b, op, insert_keys, replace_key) B) Write the bset(contains k2) to cache disk by a 30s delay work bch_btree_leaf_dirty(b, journal_ref). But before the 30s delay work write the bset to cache device, these things happened: A) GC works, and reclaim the bucket k2 point to; B) Allocator works, and invalidate the bucket k2 point to, and increase the gen of the bucket, and place it into free_inc fifo; C) Until now, the 30s delay work still does not finish work, so in the disk, the key still is k1, it is dirty and stale (its gen is smaller than the gen of the bucket). and then the machine power off suddenly happens; D) When the machine power on again, after the btree reconstruction, the stale dirty key appear. In bch_extent_bad(), when expensive_debug_checks is off, it would treat the dirty key as good even it is stale keys, and it would cause bellow probelms: A) In read_dirty() it would cause machine crash: BUG_ON(ptr_stale(dc->disk.c, &w->key, 0)); B) It could be worse when reads hits stale dirty keys, it would read old incorrect data. This patch tolerate the existence of these stale && dirty keys, and treat them as bad key in bch_extent_bad(). (Coly Li: fix indent which was modified by sender's email client) Signed-off-by: Tang Junhui <tang.junhui.linux@gmail.com> Cc: stable@vger.kernel.org Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Colin Ian King authored
There is a hunk of code that is indented one level too deep, fix this by removing the extra tabs. Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
When there are multiple bcache devices, after a reboot the name of bcache devices may change (e.g. current /dev/bcache1 was /dev/bcache0 before reboot). Therefore we need the backing device UUID (sb.uuid) to identify each bcache device. Backing device uuid can be found by program bcache-super-show, but directly exporting backing_dev_uuid by sysfs file /sys/block/bcache<?>/bcache/backing_dev_uuid is a much simpler method. With backing_dev_uuid, and partition uuids from /dev/disk/by-partuuid/, now we can identify each bcache device and its partitions conveniently. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
This patch export dc->backing_dev_name to sysfs file /sys/block/bcache<?>/bcache/backing_dev_name, then people or user space tools may know the backing device name of this bcache device. Of cause it can be done by parsing sysfs links, but this method can be much simpler to find the link between bcache device and backing device. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Coly Li authored
In stats.c:bch_cache_accounting_clear(), a hard coded number '7' is used in memset(). It is because in struct cache_stats, there are 7 atomic_t type members. This is not good when new members added into struct stats, the hard coded number will only clear part of memory. This patch replaces 'sizeof(unsigned long) * 7' by more generic 'sizeof(struct cache_stats))', to avoid potential error if new member added into struct cache_stats. Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Daniel Axtens authored
Some users see panics like the following when performing fstrim on a bcached volume: [ 529.803060] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 530.183928] #PF error: [normal kernel read fault] [ 530.412392] PGD 8000001f42163067 P4D 8000001f42163067 PUD 1f42168067 PMD 0 [ 530.750887] Oops: 0000 [#1] SMP PTI [ 530.920869] CPU: 10 PID: 4167 Comm: fstrim Kdump: loaded Not tainted 5.0.0-rc1+ #3 [ 531.290204] Hardware name: HP ProLiant DL360 Gen9/ProLiant DL360 Gen9, BIOS P89 12/27/2015 [ 531.693137] RIP: 0010:blk_queue_split+0x148/0x620 [ 531.922205] Code: 60 38 89 55 a0 45 31 db 45 31 f6 45 31 c9 31 ff 89 4d 98 85 db 0f 84 7f 04 00 00 44 8b 6d 98 4c 89 ee 48 c1 e6 04 49 03 70 78 <8b> 46 08 44 8b 56 0c 48 8b 16 44 29 e0 39 d8 48 89 55 a8 0f 47 c3 [ 532.838634] RSP: 0018:ffffb9b708df39b0 EFLAGS: 00010246 [ 533.093571] RAX: 00000000ffffffff RBX: 0000000000046000 RCX: 0000000000000000 [ 533.441865] RDX: 0000000000000200 RSI: 0000000000000000 RDI: 0000000000000000 [ 533.789922] RBP: ffffb9b708df3a48 R08: ffff940d3b3fdd20 R09: 0000000000000000 [ 534.137512] R10: ffffb9b708df3958 R11: 0000000000000000 R12: 0000000000000000 [ 534.485329] R13: 0000000000000000 R14: 0000000000000000 R15: ffff940d39212020 [ 534.833319] FS: 00007efec26e3840(0000) GS:ffff940d1f480000(0000) knlGS:0000000000000000 [ 535.224098] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 535.504318] CR2: 0000000000000008 CR3: 0000001f4e256004 CR4: 00000000001606e0 [ 535.851759] Call Trace: [ 535.970308] ? mempool_alloc_slab+0x15/0x20 [ 536.174152] ? bch_data_insert+0x42/0xd0 [bcache] [ 536.403399] blk_mq_make_request+0x97/0x4f0 [ 536.607036] generic_make_request+0x1e2/0x410 [ 536.819164] submit_bio+0x73/0x150 [ 536.980168] ? submit_bio+0x73/0x150 [ 537.149731] ? bio_associate_blkg_from_css+0x3b/0x60 [ 537.391595] ? _cond_resched+0x1a/0x50 [ 537.573774] submit_bio_wait+0x59/0x90 [ 537.756105] blkdev_issue_discard+0x80/0xd0 [ 537.959590] ext4_trim_fs+0x4a9/0x9e0 [ 538.137636] ? ext4_trim_fs+0x4a9/0x9e0 [ 538.324087] ext4_ioctl+0xea4/0x1530 [ 538.497712] ? _copy_to_user+0x2a/0x40 [ 538.679632] do_vfs_ioctl+0xa6/0x600 [ 538.853127] ? __do_sys_newfstat+0x44/0x70 [ 539.051951] ksys_ioctl+0x6d/0x80 [ 539.212785] __x64_sys_ioctl+0x1a/0x20 [ 539.394918] do_syscall_64+0x5a/0x110 [ 539.568674] entry_SYSCALL_64_after_hwframe+0x44/0xa9 We have observed it where both: 1) LVM/devmapper is involved (bcache backing device is LVM volume) and 2) writeback cache is involved (bcache cache_mode is writeback) On one machine, we can reliably reproduce it with: # echo writeback > /sys/block/bcache0/bcache/cache_mode (not sure whether above line is required) # mount /dev/bcache0 /test # for i in {0..10}; do file="$(mktemp /test/zero.XXX)" dd if=/dev/zero of="$file" bs=1M count=256 sync rm $file done # fstrim -v /test Observing this with tracepoints on, we see the following writes: fstrim-18019 [022] .... 91107.302026: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 4260112 + 196352 hit 0 bypass 1 fstrim-18019 [022] .... 91107.302050: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 4456464 + 262144 hit 0 bypass 1 fstrim-18019 [022] .... 91107.302075: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 4718608 + 81920 hit 0 bypass 1 fstrim-18019 [022] .... 91107.302094: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 5324816 + 180224 hit 0 bypass 1 fstrim-18019 [022] .... 91107.302121: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 5505040 + 262144 hit 0 bypass 1 fstrim-18019 [022] .... 91107.302145: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 5767184 + 81920 hit 0 bypass 1 fstrim-18019 [022] .... 91107.308777: bcache_write: 73f95583-561c-408f-a93a-4cbd2498f5c8 inode 0 DS 6373392 + 180224 hit 1 bypass 0 <crash> Note the final one has different hit/bypass flags. This is because in should_writeback(), we were hitting a case where the partial stripe condition was returning true and so should_writeback() was returning true early. If that hadn't been the case, it would have hit the would_skip test, and as would_skip == s->iop.bypass == true, should_writeback() would have returned false. Looking at the git history from 'commit 72c27061 ("bcache: Write out full stripes")', it looks like the idea was to optimise for raid5/6: * If a stripe is already dirty, force writes to that stripe to writeback mode - to help build up full stripes of dirty data To fix this issue, make sure that should_writeback() on a discard op never returns true. More details of debugging: https://www.spinics.net/lists/linux-bcache/msg06996.html Previous reports: - https://bugzilla.kernel.org/show_bug.cgi?id=201051 - https://bugzilla.kernel.org/show_bug.cgi?id=196103 - https://www.spinics.net/lists/linux-bcache/msg06885.html (Coly Li: minor modification to follow maximum 75 chars per line rule) Cc: Kent Overstreet <koverstreet@google.com> Cc: stable@vger.kernel.org Fixes: 72c27061 ("bcache: Write out full stripes") Signed-off-by: Daniel Axtens <dja@axtens.net> Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
- 08 Feb, 2019 2 commits
-
-
Aleksei Zakharov authored
There's no reason to freeze queue and set nr_requests value if current value is the same. Signed-off-by: Aleksei Zakharov <zakharov.a.g@yandex.ru> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Guenter Roeck authored
The following traceback is sometimes seen when booting an image in qemu: [ 54.608293] cdrom: Uniform CD-ROM driver Revision: 3.20 [ 54.611085] Fusion MPT base driver 3.04.20 [ 54.611877] Copyright (c) 1999-2008 LSI Corporation [ 54.616234] Fusion MPT SAS Host driver 3.04.20 [ 54.635139] sysctl duplicate entry: /dev/cdrom//info [ 54.639578] CPU: 0 PID: 266 Comm: kworker/u4:5 Not tainted 5.0.0-rc5 #1 [ 54.639578] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 [ 54.641273] Workqueue: events_unbound async_run_entry_fn [ 54.641273] Call Trace: [ 54.641273] dump_stack+0x67/0x90 [ 54.641273] __register_sysctl_table+0x50b/0x570 [ 54.641273] ? rcu_read_lock_sched_held+0x6f/0x80 [ 54.641273] ? kmem_cache_alloc_trace+0x1c7/0x1f0 [ 54.646814] __register_sysctl_paths+0x1c8/0x1f0 [ 54.646814] cdrom_sysctl_register.part.7+0xc/0x5f [ 54.646814] register_cdrom.cold.24+0x2a/0x33 [ 54.646814] sr_probe+0x4bd/0x580 [ 54.646814] ? __driver_attach+0xd0/0xd0 [ 54.646814] really_probe+0xd6/0x260 [ 54.646814] ? __driver_attach+0xd0/0xd0 [ 54.646814] driver_probe_device+0x4a/0xb0 [ 54.646814] ? __driver_attach+0xd0/0xd0 [ 54.646814] bus_for_each_drv+0x73/0xc0 [ 54.646814] __device_attach+0xd6/0x130 [ 54.646814] bus_probe_device+0x9a/0xb0 [ 54.646814] device_add+0x40c/0x670 [ 54.646814] ? __pm_runtime_resume+0x4f/0x80 [ 54.646814] scsi_sysfs_add_sdev+0x81/0x290 [ 54.646814] scsi_probe_and_add_lun+0x888/0xc00 [ 54.646814] ? scsi_autopm_get_host+0x21/0x40 [ 54.646814] __scsi_add_device+0x116/0x130 [ 54.646814] ata_scsi_scan_host+0x93/0x1c0 [ 54.646814] async_run_entry_fn+0x34/0x100 [ 54.646814] process_one_work+0x237/0x5e0 [ 54.646814] worker_thread+0x37/0x380 [ 54.646814] ? rescuer_thread+0x360/0x360 [ 54.646814] kthread+0x118/0x130 [ 54.646814] ? kthread_create_on_node+0x60/0x60 [ 54.646814] ret_from_fork+0x3a/0x50 The only sensible explanation is that cdrom_sysctl_register() is called twice, once from the module init function and once from register_cdrom(). cdrom_sysctl_register() is not mutex protected and may happily execute twice if the second call is made before the first call is complete. Use a static atomic to ensure that the function is executed exactly once. Signed-off-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
- 04 Feb, 2019 6 commits
-
-
https://github.com/liu-song-6/linuxJens Axboe authored
Pull MD changes for 5.1 from Song. * 'md-next' of https://github.com/liu-song-6/linux: raid1: simplify raid1_error function md-linear: use struct_size() in kzalloc()
-
Yufen Yu authored
Remove redundance set_bit and let code simplify. Signed-off-by: Yufen Yu <yuyufen@huawei.com> Signed-off-by: Song Liu <songliubraving@fb.com>
-
Gustavo A. R. Silva authored
One of the more common cases of allocation size calculations is finding the size of a structure that has a zero-sized array at the end, along with memory for some number of elements for that array. For example: struct foo { int stuff; void *entry[]; }; instance = kzalloc(sizeof(struct foo) + sizeof(void *) * count, GFP_KERNEL); Instead of leaving these open-coded and prone to type mistakes, we can now use the new struct_size() helper: instance = kzalloc(struct_size(instance, entry, count), GFP_KERNEL); This code was detected with the help of Coccinelle. Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com> Signed-off-by: Song Liu <songliubraving@fb.com>
-
git://git.infradead.org/nvmeJens Axboe authored
Pull 5.1 NVMe material from Christoph: "Below is our current (small) queue of NVMe patches for Linux 5.1. We want the re-addition of the Write Zeroes support to be in linu-next for a few weeks as it caused some problems last time. The only other patch is a cleanup from Sagi." * 'nvme-5.1' of git://git.infradead.org/nvme: nvme: remove the .stop_ctrl callout nvme: add support for the Write Zeroes command
-
Sagi Grimberg authored
It is used now just to flush error recovery and reconnect work items in the RDMA and TCP transports, which can simply be moved to the corresponding teardown routines. Signed-off-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de>
-
Chaitanya Kulkarni authored
Allow write zeroes operations (REQ_OP_WRITE_ZEROES) on the block device, if the device supports an optional command bit set for write zeroes. Add support to setup write zeroes command. Set maximum possible write zeroes sectors in one write zeroes command according to nvme write zeroes command definition. This patch was posted as a part of block-write-zeroes support implementation (https://patchwork.kernel.org/patch/9454859/), but did not make into mainline kernel as it got reverted due to failure on the Linus's machine. In this patch in order to be more cautious, we use NVMe controller's maximum hardware sector size which is calculated based on the controller's MDTS (Maximum Data Transfer Size) field to calculate the maximum sectors for the write zeroes request. Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> [folded a fix from Keith Busch to properly respect NVME_QUIRK_DEALLOCATE_ZEROES] Signed-off-by: Christoph Hellwig <hch@lst.de>
-
- 01 Feb, 2019 2 commits
-
-
Jianchao Wang authored
Currently, we check whether the hctx type is supported every time in hot path. Actually, this is not necessary, we could save the default hctx into ctx->hctxs if the type is not supported when map swqueues and use it directly with ctx->hctxs[type]. We also needn't check whether the poll is enabled or not, because the caller would clear the REQ_HIPRI in that case. Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
Jianchao Wang authored
Currently, the queue mapping result is saved in a two-dimensional array. In the hot path, to get a hctx, we need do following: q->queue_hw_ctx[q->tag_set->map[type].mq_map[cpu]] This isn't very efficient. We could save the queue mapping result into ctx directly with different hctx type, like, ctx->hctxs[type] Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-
- 31 Jan, 2019 1 commit
-
-
Paolo Valente authored
When a new I/O request arrives for a bfq_queue, say Q, bfq checks whether that request is close to (a) the head request of some other queue waiting to be served, or (b) the last request dispatched for the in-service queue (in case Q itself is not the in-service queue) If a queue, say Q2, is found for which the above condition holds, then bfq merges Q and Q2, to hopefully get a more sequential I/O in the resulting merged queue, and thus a possibly higher throughput. Case (b) is checked by comparing the new request for Q with the last request dispatched, assuming that the latter necessarily belonged to the in-service queue. Unfortunately, this assumption is no longer always correct, since commit d0edc247 ("block, bfq: inject other-queue I/O into seeky idle queues on NCQ flash"). When the assumption does not hold, queues that must not be merged may be merged, causing unexpected loss of control on per-queue service guarantees. This commit solves this problem by adding an extra field, which stores the actual last request dispatched for the in-service queue, and by using this new field to correctly check case (b). Signed-off-by: Paolo Valente <paolo.valente@linaro.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
-