1.. SPDX-License-Identifier: GPL-2.0
  2
  3========================
  4Null block device driver
  5========================
  6
  7Overview
  8========
  9
 10The null block device (``/dev/nullb*``) is used for benchmarking the various
 11block-layer implementations. It emulates a block device of X gigabytes in size.
 12It does not execute any read/write operation, just mark them as complete in
 13the request queue. The following instances are possible:
 14
 15  Multi-queue block-layer
 16
 17    - Request-based.
 18    - Configurable submission queues per device.
 19
 20  No block-layer (Known as bio-based)
 21
 22    - Bio-based. IO requests are submitted directly to the device driver.
 23    - Directly accepts bio data structure and returns them.
 24
 25All of them have a completion queue for each core in the system.
 26
 27Module parameters
 28=================
 29
 30queue_mode=[0-2]: Default: 2-Multi-queue
 31  Selects which block-layer the module should instantiate with.
 32
 33  =  ============
 34  0  Bio-based
 35  1  Single-queue (deprecated)
 36  2  Multi-queue
 37  =  ============
 38
 39home_node=[0--nr_nodes]: Default: NUMA_NO_NODE
 40  Selects what CPU node the data structures are allocated from.
 41
 42gb=[Size in GB]: Default: 250GB
 43  The size of the device reported to the system.
 44
 45bs=[Block size (in bytes)]: Default: 512 bytes
 46  The block size reported to the system.
 47
 48nr_devices=[Number of devices]: Default: 1
 49  Number of block devices instantiated. They are instantiated as /dev/nullb0,
 50  etc.
 51
 52irqmode=[0-2]: Default: 1-Soft-irq
 53  The completion mode used for completing IOs to the block-layer.
 54
 55  =  ===========================================================================
 56  0  None.
 57  1  Soft-irq. Uses IPI to complete IOs across CPU nodes. Simulates the overhead
 58     when IOs are issued from another CPU node than the home the device is
 59     connected to.
 60  2  Timer: Waits a specific period (completion_nsec) for each IO before
 61     completion.
 62  =  ===========================================================================
 63
 64completion_nsec=[ns]: Default: 10,000ns
 65  Combined with irqmode=2 (timer). The time each completion event must wait.
 66
 67submit_queues=[1..nr_cpus]: Default: 1
 68  The number of submission queues attached to the device driver. If unset, it
 69  defaults to 1. For multi-queue, it is ignored when use_per_node_hctx module
 70  parameter is 1.
 71
 72hw_queue_depth=[0..qdepth]: Default: 64
 73  The hardware queue depth of the device.
 74
 75memory_backed=[0/1]: Default: 0
 76  Whether or not to use a memory buffer to respond to IO requests
 77
 78  =  =============================================
 79  0  Transfer no data in response to IO requests
 80  1  Use a memory buffer to respond to IO requests
 81  =  =============================================
 82
 83discard=[0/1]: Default: 0
 84  Support discard operations (requires memory-backed null_blk device).
 85
 86  =  =====================================
 87  0  Do not support discard operations
 88  1  Enable support for discard operations
 89  =  =====================================
 90
 91cache_size=[Size in MB]: Default: 0
 92  Cache size in MB for memory-backed device.
 93
 94mbps=[Maximum bandwidth in MB/s]: Default: 0 (no limit)
 95  Bandwidth limit for device performance.
 96
 97Multi-queue specific parameters
 98-------------------------------
 99
100use_per_node_hctx=[0/1]: Default: 0
101  Number of hardware context queues.
102
103  =  =====================================================================
104  0  The number of submit queues are set to the value of the submit_queues
105     parameter.
106  1  The multi-queue block layer is instantiated with a hardware dispatch
107     queue for each CPU node in the system.
108  =  =====================================================================
109
110no_sched=[0/1]: Default: 0
111  Enable/disable the io scheduler.
112
113  =  ======================================
114  0  nullb* use default blk-mq io scheduler
115  1  nullb* doesn't use io scheduler
116  =  ======================================
117
118blocking=[0/1]: Default: 0
119  Blocking behavior of the request queue.
120
121  =  ===============================================================
122  0  Register as a non-blocking blk-mq driver device.
123  1  Register as a blocking blk-mq driver device, null_blk will set
124     the BLK_MQ_F_BLOCKING flag, indicating that it sometimes/always
125     needs to block in its ->queue_rq() function.
126  =  ===============================================================
127
128shared_tags=[0/1]: Default: 0
129  Sharing tags between devices.
130
131  =  ================================================================
132  0  Tag set is not shared.
133  1  Tag set shared between devices for blk-mq. Only makes sense with
134     nr_devices > 1, otherwise there's no tag set to share.
135  =  ================================================================
136
137zoned=[0/1]: Default: 0
138  Device is a random-access or a zoned block device.
139
140  =  ======================================================================
141  0  Block device is exposed as a random-access block device.
142  1  Block device is exposed as a host-managed zoned block device. Requires
143     CONFIG_BLK_DEV_ZONED.
144  =  ======================================================================
145
146zone_size=[MB]: Default: 256
147  Per zone size when exposed as a zoned block device. Must be a power of two.
148
149zone_nr_conv=[nr_conv]: Default: 0
150  The number of conventional zones to create when block device is zoned.  If
151  zone_nr_conv >= nr_zones, it will be reduced to nr_zones - 1.