kernel/linux.git/drivers/target/target_core_device.c, branch linux-6.0.y

scsi: target: core: De-RCU of se_lun and se_lun acl

2022-08-01T23:36:02+00:00

se_lun and se_lun_acl are immutable pointers of struct se_dev_entry. Remove RCU usage for access to those pointers. Link: https://lore.kernel.org/r/20220727214125.19647-3-d.bogdanov@yadro.com Reviewed-by: Mike Christie Signed-off-by: Dmitry Bogdanov Signed-off-by: Martin K. Petersen

scsi: target: core: Fix race during ACL removal

2022-08-01T23:36:02+00:00

Under huge load there is a possibility of race condition in updating se_dev_entry object in ACL removal procedure: NIP [c0080000154093d0] transport_lookup_cmd_lun+0x1f8/0x3d0 [target_core_mod] LR [c00800001542ab34] target_submit_cmd_map_sgls+0x11c/0x300 [target_core_mod] Call Trace: target_submit_cmd_map_sgls+0x11c/0x300 [target_core_mod] target_submit_cmd+0x44/0x60 [target_core_mod] tcm_qla2xxx_handle_cmd+0x88/0xe0 [tcm_qla2xxx] qlt_do_work+0x2e4/0x3d0 [qla2xxx] process_one_work+0x298/0x5c Despite usage of RCU primitives with deve->se_lun pointer, it has not become dereference-safe because deve->se_lun is updated and not synchronized with a reader. That change might be in a release function called by synchronize_rcu(). But, in fact, there is no point in setting that pointer to NULL for deleting deve. All access to deve->se_lun is already under rcu_read_lock. And either deve->se_lun is always valid or deve is not valid itself and will not be found in the list_for_*. The same applicable for deve->se_lun_acl too. So a better solution is to remove that NULLing. Link: https://lore.kernel.org/r/20220727214125.19647-2-d.bogdanov@yadro.com Reviewed-by: Mike Christie Signed-off-by: Dmitry Bogdanov Signed-off-by: Martin K. Petersen

scsi: target: Add callout to configure UNMAP settings

2022-07-07T20:53:52+00:00

Add a callout to configure a backend's UNMAP settings. This will be used to allow userspace to configure UNMAP after the initial device setup, similar to how we can set up the other attributes post device configuration. Link: https://lore.kernel.org/r/20220628200230.15052-3-michael.christie@oracle.com Reviewed-by: Christoph Hellwig Signed-off-by: Mike Christie Signed-off-by: Martin K. Petersen

block: add a bdev_discard_granularity helper

2022-04-18T01:49:59+00:00

Abstract away implementation details from file systems by providing a block_device based helper to retrieve the discard granularity. Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Acked-by: Christoph Böhmwalder [drbd] Acked-by: Ryusuke Konishi Acked-by: David Sterba [btrfs] Link: https://lore.kernel.org/r/20220415045258.199825-26-hch@lst.de Signed-off-by: Jens Axboe

block: remove QUEUE_FLAG_DISCARD

2022-04-18T01:49:59+00:00

Just use a non-zero max_discard_sectors as an indicator for discard support, similar to what is done for write zeroes. The only places where needs special attention is the RAID5 driver, which must clear discard support for security reasons by default, even if the default stacking rules would allow for it. Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Acked-by: Christoph Böhmwalder [drbd] Acked-by: Jan Höppner [s390] Acked-by: Coly Li [bcache] Acked-by: David Sterba [btrfs] Reviewed-by: Chaitanya Kulkarni Link: https://lore.kernel.org/r/20220415045258.199825-25-hch@lst.de Signed-off-by: Jens Axboe

block: add a bdev_max_discard_sectors helper

2022-04-18T01:49:59+00:00

Add a helper to query the number of sectors support per each discard bio based on the block device and use this helper to stop various places from poking into the request_queue to see if discard is supported and if so how much. This mirrors what is done e.g. for write zeroes as well. Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Acked-by: Christoph Böhmwalder [drbd] Acked-by: Coly Li [bcache] Acked-by: David Sterba [btrfs] Reviewed-by: Chaitanya Kulkarni Link: https://lore.kernel.org/r/20220415045258.199825-24-hch@lst.de Signed-off-by: Jens Axboe

target: fix discard alignment on partitions

2022-04-18T01:49:58+00:00

Use the proper bdev_discard_alignment helper that accounts for partition offsets. Fixes: c66ac9db8d4a ("[SCSI] target: Add LIO target core v4.0.0-rc6") Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Reviewed-by: Chaitanya Kulkarni Link: https://lore.kernel.org/r/20220415045258.199825-4-hch@lst.de Signed-off-by: Jens Axboe

target: pass a block_device to target_configure_unmap_from_queue

2022-04-18T01:49:58+00:00

The SCSI target drivers is a consumer of the block layer and shoul d generally work on struct block_device. Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Reviewed-by: Chaitanya Kulkarni Link: https://lore.kernel.org/r/20220415045258.199825-3-hch@lst.de Signed-off-by: Jens Axboe

target: remove an incorrect unmap zeroes data deduction

2022-04-18T01:49:58+00:00

For block devices, the SCSI target drivers implements UNMAP as calls to blkdev_issue_discard, which does not guarantee zeroing just because Write Zeroes is supported. Note that this does not affect the file backed path which uses fallocate to punch holes. Fixes: 2237498f0b5c ("target/iblock: Convert WRITE_SAME to blkdev_issue_zeroout") Signed-off-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Reviewed-by: Chaitanya Kulkarni Link: https://lore.kernel.org/r/20220415045258.199825-2-hch@lst.de Signed-off-by: Jens Axboe

scsi: target: Fix ordered tag handling

2021-10-19T02:38:35+00:00

This patch fixes the following bugs: 1. If there are multiple ordered cmds queued and multiple simple cmds completing, target_restart_delayed_cmds() could be called on different CPUs and each instance could start a ordered cmd. They could then run in different orders than they were queued. 2. target_restart_delayed_cmds() and target_handle_task_attr() can race where: 1. target_handle_task_attr() has passed the simple_cmds == 0 check. 2. transport_complete_task_attr() then decrements simple_cmds to 0. 3. transport_complete_task_attr() runs target_restart_delayed_cmds() and it does not see any cmds on the delayed_cmd_list. 4. target_handle_task_attr() adds the cmd to the delayed_cmd_list. The cmd will then end up timing out. 3. If we are sent > 1 ordered cmds and simple_cmds == 0, we can execute them out of order, because target_handle_task_attr() will hit that simple_cmds check first and return false for all ordered cmds sent. 4. We run target_restart_delayed_cmds() after every cmd completion, so if there is more than 1 simple cmd running, we start executing ordered cmds after that first cmd instead of waiting for all of them to complete. 5. Ordered cmds are not supposed to start until HEAD OF QUEUE and all older cmds have completed, and not just simple. 6. It's not a bug but it doesn't make sense to take the delayed_cmd_lock for every cmd completion when ordered cmds are almost never used. Just replacing that lock with an atomic increases IOPs by up to 10% when completions are spread over multiple CPUs and there are multiple sessions/ mqs/thread accessing the same device. This patch moves the queued delayed handling to a per device work to serialze the cmd executions for each device and adds a new counter to track HEAD_OF_QUEUE and SIMPLE cmds. We can then check the new counter to determine when to run the work on the completion path. Link: https://lore.kernel.org/r/20210930020422.92578-3-michael.christie@oracle.com Signed-off-by: Mike Christie Signed-off-by: Martin K. Petersen