- 29 Dec, 2023 3 commits
-
-
Christoph Hellwig authored
Current the discard granularity defaults to 0 and must be initialized by any driver that wants to support discard. Default to the sector size instead, which is the smallest possible value, and a very useful default. Signed-off-by:
Christoph Hellwig <hch@lst.de> Link: https://lore.kernel.org/r/20231228075545.362768-4-hch@lst.deSigned-off-by:
Jens Axboe <axboe@kernel.dk>
-
Christoph Hellwig authored
Just like all block I/O, discards are in units of sectors. Thus setting a smaller than sector size discard limit in case of > 512 byte sectors in bcache doesn't make sense. Always set the discard granularity to 512 bytes instead. Signed-off-by:
-
Christoph Hellwig authored
A zero discard_granularity is not treated the same as a single-block one, and not having any segments after taking alignment is perfectly fine and does not need a warning. Signed-off-by:
-
- 27 Dec, 2023 5 commits
-
-
Christoph Hellwig authored
Give BLK_DEF_MAX_SECTORS a _CAP postfix and document what it is used for. Signed-off-by:
-
Christoph Hellwig authored
BLK_DEF_MAX_SECTORS despite the confusing name is the default cap for the max_sectors limits. Don't use it to initialize max_hw_setors, which is a hardware / driver capacility. Signed-off-by:
-
Christoph Hellwig authored
BLK_DEF_MAX_SECTORS despite the confusing name is the default cap for the max_sectors limits. Don't use it to initialize max_hw_setors, which is a hardware / driver capacility. Signed-off-by:
-
Christoph Hellwig authored
null_blk has some rather odd capping of the max_hw_sectors value to BLK_DEF_MAX_SECTORS, which doesn't make sense - max_hw_sector is the hardware limit, and BLK_DEF_MAX_SECTORS despite the confusing name is the default cap for the max_sectors field used for normal file system I/O. Remove all the capping, and simply leave it to the block layer or user to take up or not all of that for file system I/O. Fixes: ea17fd35 ("null_blk: Allow controlling max_hw_sectors limit") Signed-off-by:
-
Christoph Hellwig authored
loop_set_status doesn't change anything relevant to the discard and write_zeroes setting, so don't bother calling loop_config_discard. Signed-off-by:
-
- 26 Dec, 2023 2 commits
-
-
Christoph Hellwig authored
Use the queue wide write back cache tracking insted of duplicating the value in strut rq_wb. Signed-off-by:
-
Christoph Hellwig authored
submit_bio_noacct allows completely invalid operations, or operations that are not supported in the bio path. Extent the existing switch statement to rejcect all invalid types. Move the code point for REQ_OP_ZONE_APPEND so that it's not right in the middle of the zone management operations and the switch statement can follow the numerical order of the operations. Signed-off-by:
-
- 22 Dec, 2023 2 commits
-
-
Randy Dunlap authored
Fix all kernel-doc warnings in drbd_actlog.c: drbd_actlog.c:963: warning: No description found for return value of 'drbd_rs_begin_io' drbd_actlog.c:1015: warning: Function parameter or member 'peer_device' not described in 'drbd_try_rs_begin_io' drbd_actlog.c:1015: warning: Excess function parameter 'device' description in 'drbd_try_rs_begin_io' drbd_actlog.c:1015: warning: No description found for return value of 'drbd_try_rs_begin_io' drbd_actlog.c:1197: warning: No description found for return value of 'drbd_rs_del_all' Fix one spelling error (s/ore/or/). Signed-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Philipp Reisner <philipp.reisner@linbit.com> Cc: Lars Ellenberg <lars.ellenberg@linbit.com> Cc: Christoph Böhmwalder <christoph.boehmwalder@linbit.com> Cc: <drbd-dev@lists.linbit.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: <linux-block@vger.kernel.org> Link: https://lore.kernel.org/r/20231222061909.8791-1-rdunlap@infradead.orgSigned-off-by:
-
Kundan Kumar authored
commit 41fa7222 ("blk-mq: do not include passthrough requests in I/O accounting")' disables I/O accounting for passthrough requests. Since tools like 'iostat' do not show anything useful for passthrough I/O, it's wasteful to do start/end time-stamping. So do away with that. Avoiding the time-stamping improves the I/O performance by ~7% Signed-off-by:
Kundan Kumar <kundan.kumar@samsung.com> Signed-off-by:
Kanchan Joshi <joshi.k@samsung.com> Link: https://lore.kernel.org/r/20231222101707.6921-1-kundan.kumar@samsung.comSigned-off-by:
-
- 21 Dec, 2023 4 commits
-
-
git://git.infradead.org/nvmeJens Axboe authored
Pull NVMe updates from Keith: "nvme updates for Linux 6.8 - nvme fabrics spec updates (Guixin, Max) - nvme target udpates (Guixin, Evan) - nvme attribute refactoring (Daniel) - nvme-fc numa fix (Keith)" * tag 'nvme-6.8-2023-12-21' of git://git.infradead.org/nvme: nvme-fc: set numa_node after nvme_init_ctrl nvme-fabrics: don't check discovery ioccsz/iorcsz nvmet: configfs: use ctrl->instance to track passthru subsystems nvme: repack struct nvme_ns_head nvme: add csi, ms and nuse to sysfs nvme: rename ns attribute group nvme: refactor ns info setup function nvme: refactor ns info helpers nvme: move ns id info to struct nvme_ns_head nvmet: remove cntlid_min and cntlid_max check in nvmet_alloc_ctrl nvmet: allow identical cntlid_min and cntlid_max settings nvme-fabrics: check ioccsz and iorcsz nvme: introduce nvme_check_ctrl_fabric_info helper
-
Keith Busch authored
nvme_init_ctrl() resets numa_node to NUMA_NO_NODE, so be sure to set the desired value after that function call so it won't be overwritten. Reviewed-by:
Sagi Grimberg <sagi@grimberg.me> Reviewed-by:
Max Gurtovoy <mgurtovoy@nvidia.com> Signed-off-by:
Keith Busch <kbusch@kernel.org>
-
Max Gurtovoy authored
IOCCSZ and IORCSZ are reserved for discovery controllers. Avoid checking their values during identify controller phase. Fixes: 2fcd3ab3 ("nvme-fabrics: check ioccsz and iorcsz") Reported-by:
Daniel Wagner <dwagner@suse.de> Tested-by:
-
Jens Axboe authored
A previous commit split disk_set_zoned(..., bool) into not taking an argument for whether to set or clear, and instead added disk_clear_zoned() as the counterpart. However, that commit neglected to export the new symbol, causing failures for modular drivers that used it. Reported-by:
Stephen Rothwell <sfr@canb.auug.org.au> Fixes: d73e93b4 ("block: simplify disk_set_zoned") Signed-off-by:
-
- 20 Dec, 2023 5 commits
-
-
Christoph Hellwig authored
disk_clear_zoned only needs to be called when a device reported zone managed mode first and we clear it. Add a check so that disk_clear_zoned isn't called on devices that were never zoned. This avoids a fairly expensive queue freezing when revalidating conventional devices. Signed-off-by:
Damien Le Moal <dlemoal@kernel.org> Reviewed-by:
Martin K. Petersen <martin.petersen@oracle.com> Link: https://lore.kernel.org/r/20231217165359.604246-6-hch@lst.deSigned-off-by:
-
Christoph Hellwig authored
Only use disk_set_zoned to actually enable zoned device support. For clearing it, call disk_clear_zoned, which is renamed from disk_clear_zone_settings and now directly clears the zoned flag as well. Signed-off-by:
-
Christoph Hellwig authored
When zones were first added the SCSI and ATA specs, two different models were supported (in addition to the drive managed one that is invisible to the host): - host managed where non-conventional zones there is strict requirement to write at the write pointer, or else an error is returned - host aware where a write point is maintained if writes always happen at it, otherwise it is left in an under-defined state and the sequential write preferred zones behave like conventional zones (probably very badly performing ones, though) Not surprisingly this lukewarm model didn't prove to be very useful and was finally removed from the ZBC and SBC specs (NVMe never implemented it). Due to to the easily disappearing write pointer host software could never rely on the write pointer to actually be useful for say recovery. Fortunately only a few HDD prototypes shipped using this model which never made it to mass production. Drop the support before it is too late. Note that any such host aware prototype HDD can still be used with Linux as we'll now treat it as a conventional HDD. Signed-off-by:
-
Christoph Hellwig authored
virtblk_revalidate_zones is called unconditionally from virtblk_config_changed_work from the virtio config_changed callback. virtblk_revalidate_zones is a bit odd in that it re-clears the zoned state for host aware or non-zoned devices, which isn't needed unless the zoned mode changed - but a zone mode change to a host managed model isn't handled at all, and virtio_blk also doesn't handle any other config change except for a capacity change is handled (and even if it was the upper layers above virtio_blk wouldn't handle it very well). But even the useful case of a size change that would add or remove zones isn't handled properly as blk_revalidate_disk_zones expects the device capacity to cover all zones, but the capacity is only updated after virtblk_revalidate_zones. As this code appears to be entirely untested and is getting in the way remove it for now, but it can be readded in a fixed version with proper test coverage if needed. Fixes: 95bfec41 ("virtio-blk: add support for zoned block devices") Fixes: f1ba4e67 ("virtio-blk: fix to match virtio spec") Signed-off-by:
Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by:
-
Christoph Hellwig authored
Move reading and checking the zoned model from virtblk_probe_zoned_device into the caller, leaving only the code to perform the actual setup for host managed zoned devices in virtblk_probe_zoned_device. This allows to share the model reading and sharing between builds with and without CONFIG_BLK_DEV_ZONED, and improve it for the !CONFIG_BLK_DEV_ZONED case. Signed-off-by:
-
- 19 Dec, 2023 12 commits
-
-
Jens Axboe authored
Merge tag 'md-next-20231219' of https://git.kernel.org/pub/scm/linux/kernel/git/song/md into for-6.8/block Pull MD updates from Song: "1. Remove deprecated flavors, by Song Liu; 2. raid1 read error check support, by Li Nan; 3. Better handle events off-by-1 case, by Alex Lyakas." * tag 'md-next-20231219' of https://git.kernel.org/pub/scm/linux/kernel/git/song/md: md: Remove deprecated CONFIG_MD_FAULTY md: Remove deprecated CONFIG_MD_MULTIPATH md: Remove deprecated CONFIG_MD_LINEAR md/raid1: support read error check md: factor out a helper exceed_read_errors() to check read_errors md: Whenassemble the array, consult the superblock of the freshest device md/raid1: remove unnecessary null checking
-
Song Liu authored
md-faulty has been marked as deprecated for 2.5 years. Remove it. Cc: Christoph Hellwig <hch@lst.de> Cc: Jens Axboe <axboe@kernel.dk> Cc: Neil Brown <neilb@suse.de> Cc: Guoqing Jiang <guoqing.jiang@linux.dev> Cc: Mateusz Grzonka <mateusz.grzonka@intel.com> Cc: Jes Sorensen <jes@trained-monkey.org> Signed-off-by:
Song Liu <song@kernel.org> Reviewed-by:
Hannes Reinecke <hare@suse.de> Link: https://lore.kernel.org/r/20231214222107.2016042-4-song@kernel.org
-
Song Liu authored
md-multipath has been marked as deprecated for 2.5 years. Remove it. Cc: Christoph Hellwig <hch@lst.de> Cc: Jens Axboe <axboe@kernel.dk> Cc: Neil Brown <neilb@suse.de> Cc: Guoqing Jiang <guoqing.jiang@linux.dev> Cc: Mateusz Grzonka <mateusz.grzonka@intel.com> Cc: Jes Sorensen <jes@trained-monkey.org> Signed-off-by:
-
Song Liu authored
md-linear has been marked as deprecated for 2.5 years. Remove it. Cc: Christoph Hellwig <hch@lst.de> Cc: Jens Axboe <axboe@kernel.dk> Cc: Neil Brown <neilb@suse.de> Cc: Guoqing Jiang <guoqing.jiang@linux.dev> Cc: Mateusz Grzonka <mateusz.grzonka@intel.com> Cc: Jes Sorensen <jes@trained-monkey.org> Signed-off-by:
-
Evan Burgess authored
To prevent enabling more than one passthrough subsystem per NVMe controller, passthru.c maintains an xarray indexed by cntlid values. Passthrough for a given nvmet subsystem cannot be enabled by configfs if the subsystem's passthru_ctrl->cntlid value is already accounted for in the xarray. However, according to the NVMe spec (rev 2.0c, p.145), "The Controller ID (CNTLID) value returned in the Identify Controller data structure may be used to uniquely identify a controller within an NVM subsystem," meaning that cntlid values are not guaranteed to be globally unique across multiple subsystems. Instead, the cntlid only uniquely identifies multiple controllers _within_ a subsystem. As a result, multiple unique & valid NVMe targets can be blocked from enabling passthrough at the same time if their controllers share cntlid values, a behavior allowed by the spec. Fix this by indexing the xarray with passthru_ctrl->instance values, which are allocated per controller by IDA and thus should be truly unique. Reviewed-by:
Evan Burgess <evan.burgess@seagate.com> Signed-off-by:
-
Daniel Wagner authored
ns_id, lba_shift and ms are always accessed for every read/write I/O in nvme_setup_rw. By grouping these variables into one cacheline we can safe some cycles. 4k sequential reads: baseline patched Bandwidth: 1620 1634 IOPs 66345579 66910939 Reviewed-by: -
Daniel Wagner authored
libnvme is using the sysfs for enumarating the nvme resources. Though there are few missing attritbutes in the sysfs. For these libnvme issues commands during discovering. As the kernel already knows all these attributes and we would like to avoid libnvme to issue commands all the time, expose these missing attributes. The nuse value is updated on request because the nuse is a volatile value. Since any user can read the sysfs attribute, a very simple rate limit is added (update once every 5 seconds). A more sophisticated update strategy can be added later if there is actually a need for it. Reviewed-by:
-
Daniel Wagner authored
Drop the 'id' part of the attribute group name because we want to expose non 'id' related attributes via the ns attribute group. Reviewed-by:
-
Daniel Wagner authored
Use nvme_ns_head instead of nvme_ns where possible. This reduces the coupling between the different data structures. Reviewed-by:
-
Daniel Wagner authored
Pass in the nvme_ns_head pointer directly. This reduces the necessity on the caller side have the nvme_ns data structure present. Thus we can refactor the caller side in the next step as well. Reviewed-by:
-
Daniel Wagner authored
Move the namesapce info to struct nvme_ns_head, because it's the same for all associated namespaces. Note: with multipathing enabled the PI information is shared between all paths. If a path is using a different PI configuration it will overwrite the previous settings. This is obviously not correct and such configuration will be rejected in future. For the time being we expect a correctly configured storage. Reviewed-by:
-
Li Nan authored
'first_minor' represents the starting minor number of disks, and 'minors' represents the number of partitions in the device. Neither of them can be greater than MINORMASK + 1. Commit e338924b ("block: check minor range in device_add_disk()") only added the check of 'first_minor + minors'. However, their sum might be less than MINORMASK but their values are wrong. Complete the checks now. Fixes: e338924b ("block: check minor range in device_add_disk()") Signed-off-by:
Li Nan <linan122@huawei.com> Reviewed-by:
-
- 18 Dec, 2023 1 commit
-
-
Kundan Kumar authored
Even if BLK_CGROUP is enabled, it does not work for passthrough io. So skip setting up blkg for passthrough bio. Reduced processing gives ~5% hike in peak-performance workload. Signed-off-by:
-
- 15 Dec, 2023 6 commits
-
-
Li Nan authored
After commit 1e50915f ("raid: improve MD/raid10 handling of correctable read errors."), rdev will be set to faulty if it reads data error to many times in raid10. Add this mechanism to raid1 now. Signed-off-by:
-
Li Nan authored
Move check_decay_read_errors() to raid1-10.c and factor out a helper exceed_read_errors() to check if read_errors exceeds the limit, so that raid1 can also use it. There are no functional changes. Signed-off-by:
-
Alex Lyakas authored
Upon assembling the array, both kernel and mdadm allow the devices to have event counter difference of 1, and still consider them as up-to-date. However, a device whose event count is behind by 1, may in fact not be up-to-date, and array resync with such a device may cause data corruption. To avoid this, consult the superblock of the freshest device about the status of a device, whose event counter is behind by 1. Signed-off-by:
Alex Lyakas <alex.lyakas@zadara.com> Signed-off-by:
-
Jens Axboe authored
It's clearly been a while since someone looked at this, so I gave it a quick shot. There are few issues in here: - Random bundling of members that are mostly read-only and often written - Random holes that need not be there This moves the most frequently used bits into cacheline 1 and 2, with the 2nd one being more write intensive than the first one, which is basically read-only. Outside of making this work a bit more efficiently, it also reduces the size of struct request_queue for my test setup from 864 bytes (spanning 14 cachelines!) to 832 bytes and 13 cachelines. Reviewed-by:
-
Christoph Hellwig authored
bio_add_hw_page currently always fails or succeeds. This is fine for the existing callers that always add PAGE_SIZE worth given that the max_segment_size and max_sectors must always allow at least a page worth of data. But when we want to add it for bigger amounts of data this means it can also fail when adding the data to a bio, and creating a fallback for that becomes really annoying in the callers. Make use of the existing API design that allows to return a smaller length than the one passed in and add up to max_segment_size worth of data from a larger input. All the existing callers are fine with this - not because they handle this return correctly, but because they never pass more than a page in. Signed-off-by:
Johannes Thumshirn <johannes.thumshirn@wdc.com> Link: https://lore.kernel.org/r/20231204173419.782378-3-hch@lst.deSigned-off-by:
-
Christoph Hellwig authored
Reordered a check to avoid a possible overflow when adding len to bv_len. Signed-off-by:
-
