1================= 2Queue sysfs files 3================= 4 5This text file will detail the queue files that are located in the sysfs tree 6for each block device. Note that stacked devices typically do not export 7any settings, since their queue merely functions are a remapping target. 8These files are the ones found in the /sys/block/xxx/queue/ directory. 9 10Files denoted with a RO postfix are readonly and the RW postfix means 11read-write. 12 13add_random (RW) 14--------------- 15This file allows to turn off the disk entropy contribution. Default 16value of this file is '1'(on). 17 18chunk_sectors (RO) 19------------------ 20This has different meaning depending on the type of the block device. 21For a RAID device (dm-raid), chunk_sectors indicates the size in 512B sectors 22of the RAID volume stripe segment. For a zoned block device, either host-aware 23or host-managed, chunk_sectors indicates the size in 512B sectors of the zones 24of the device, with the eventual exception of the last zone of the device which 25may be smaller. 26 27dax (RO) 28-------- 29This file indicates whether the device supports Direct Access (DAX), 30used by CPU-addressable storage to bypass the pagecache. It shows '1' 31if true, '0' if not. 32 33discard_granularity (RO) 34------------------------ 35This shows the size of internal allocation of the device in bytes, if 36reported by the device. A value of '0' means device does not support 37the discard functionality. 38 39discard_max_hw_bytes (RO) 40------------------------- 41Devices that support discard functionality may have internal limits on 42the number of bytes that can be trimmed or unmapped in a single operation. 43The discard_max_bytes parameter is set by the device driver to the maximum 44number of bytes that can be discarded in a single operation. Discard 45requests issued to the device must not exceed this limit. A discard_max_bytes 46value of 0 means that the device does not support discard functionality. 47 48discard_max_bytes (RW) 49---------------------- 50While discard_max_hw_bytes is the hardware limit for the device, this 51setting is the software limit. Some devices exhibit large latencies when 52large discards are issued, setting this value lower will make Linux issue 53smaller discards and potentially help reduce latencies induced by large 54discard operations. 55 56discard_zeroes_data (RO) 57------------------------ 58Obsolete. Always zero. 59 60fua (RO) 61-------- 62Whether or not the block driver supports the FUA flag for write requests. 63FUA stands for Force Unit Access. If the FUA flag is set that means that 64write requests must bypass the volatile cache of the storage device. 65 66hw_sector_size (RO) 67------------------- 68This is the hardware sector size of the device, in bytes. 69 70io_poll (RW) 71------------ 72When read, this file shows whether polling is enabled (1) or disabled 73(0). Writing '0' to this file will disable polling for this device. 74Writing any non-zero value will enable this feature. 75 76io_poll_delay (RW) 77------------------ 78If polling is enabled, this controls what kind of polling will be 79performed. It defaults to -1, which is classic polling. In this mode, 80the CPU will repeatedly ask for completions without giving up any time. 81If set to 0, a hybrid polling mode is used, where the kernel will attempt 82to make an educated guess at when the IO will complete. Based on this 83guess, the kernel will put the process issuing IO to sleep for an amount 84of time, before entering a classic poll loop. This mode might be a 85little slower than pure classic polling, but it will be more efficient. 86If set to a value larger than 0, the kernel will put the process issuing 87IO to sleep for this amount of microseconds before entering classic 88polling. 89 90io_timeout (RW) 91--------------- 92io_timeout is the request timeout in milliseconds. If a request does not 93complete in this time then the block driver timeout handler is invoked. 94That timeout handler can decide to retry the request, to fail it or to start 95a device recovery strategy. 96 97iostats (RW) 98------------- 99This file is used to control (on/off) the iostats accounting of the 100disk. 101 102logical_block_size (RO) 103----------------------- 104This is the logical block size of the device, in bytes. 105 106max_discard_segments (RO) 107------------------------- 108The maximum number of DMA scatter/gather entries in a discard request. 109 110max_hw_sectors_kb (RO) 111---------------------- 112This is the maximum number of kilobytes supported in a single data transfer. 113 114max_integrity_segments (RO) 115--------------------------- 116Maximum number of elements in a DMA scatter/gather list with integrity 117data that will be submitted by the block layer core to the associated 118block driver. 119 120max_sectors_kb (RW) 121------------------- 122This is the maximum number of kilobytes that the block layer will allow 123for a filesystem request. Must be smaller than or equal to the maximum 124size allowed by the hardware. 125 126max_segments (RO) 127----------------- 128Maximum number of elements in a DMA scatter/gather list that is submitted 129to the associated block driver. 130 131max_segment_size (RO) 132--------------------- 133Maximum size in bytes of a single element in a DMA scatter/gather list. 134 135minimum_io_size (RO) 136-------------------- 137This is the smallest preferred IO size reported by the device. 138 139nomerges (RW) 140------------- 141This enables the user to disable the lookup logic involved with IO 142merging requests in the block layer. By default (0) all merges are 143enabled. When set to 1 only simple one-hit merges will be tried. When 144set to 2 no merge algorithms will be tried (including one-hit or more 145complex tree/hash lookups). 146 147nr_requests (RW) 148---------------- 149This controls how many requests may be allocated in the block layer for 150read or write requests. Note that the total allocated number may be twice 151this amount, since it applies only to reads or writes (not the accumulated 152sum). 153 154To avoid priority inversion through request starvation, a request 155queue maintains a separate request pool per each cgroup when 156CONFIG_BLK_CGROUP is enabled, and this parameter applies to each such 157per-block-cgroup request pool. IOW, if there are N block cgroups, 158each request queue may have up to N request pools, each independently 159regulated by nr_requests. 160 161nr_zones (RO) 162------------- 163For zoned block devices (zoned attribute indicating "host-managed" or 164"host-aware"), this indicates the total number of zones of the device. 165This is always 0 for regular block devices. 166 167optimal_io_size (RO) 168-------------------- 169This is the optimal IO size reported by the device. 170 171physical_block_size (RO) 172------------------------ 173This is the physical block size of device, in bytes. 174 175read_ahead_kb (RW) 176------------------ 177Maximum number of kilobytes to read-ahead for filesystems on this block 178device. 179 180rotational (RW) 181--------------- 182This file is used to stat if the device is of rotational type or 183non-rotational type. 184 185rq_affinity (RW) 186---------------- 187If this option is '1', the block layer will migrate request completions to the 188cpu "group" that originally submitted the request. For some workloads this 189provides a significant reduction in CPU cycles due to caching effects. 190 191For storage configurations that need to maximize distribution of completion 192processing setting this option to '2' forces the completion to run on the 193requesting cpu (bypassing the "group" aggregation logic). 194 195scheduler (RW) 196-------------- 197When read, this file will display the current and available IO schedulers 198for this block device. The currently active IO scheduler will be enclosed 199in [] brackets. Writing an IO scheduler name to this file will switch 200control of this block device to that new IO scheduler. Note that writing 201an IO scheduler name to this file will attempt to load that IO scheduler 202module, if it isn't already present in the system. 203 204write_cache (RW) 205---------------- 206When read, this file will display whether the device has write back 207caching enabled or not. It will return "write back" for the former 208case, and "write through" for the latter. Writing to this file can 209change the kernels view of the device, but it doesn't alter the 210device state. This means that it might not be safe to toggle the 211setting from "write back" to "write through", since that will also 212eliminate cache flushes issued by the kernel. 213 214write_same_max_bytes (RO) 215------------------------- 216This is the number of bytes the device can write in a single write-same 217command. A value of '0' means write-same is not supported by this 218device. 219 220wbt_lat_usec (RW) 221----------------- 222If the device is registered for writeback throttling, then this file shows 223the target minimum read latency. If this latency is exceeded in a given 224window of time (see wb_window_usec), then the writeback throttling will start 225scaling back writes. Writing a value of '0' to this file disables the 226feature. Writing a value of '-1' to this file resets the value to the 227default setting. 228 229throttle_sample_time (RW) 230------------------------- 231This is the time window that blk-throttle samples data, in millisecond. 232blk-throttle makes decision based on the samplings. Lower time means cgroups 233have more smooth throughput, but higher CPU overhead. This exists only when 234CONFIG_BLK_DEV_THROTTLING_LOW is enabled. 235 236write_zeroes_max_bytes (RO) 237--------------------------- 238For block drivers that support REQ_OP_WRITE_ZEROES, the maximum number of 239bytes that can be zeroed at once. The value 0 means that REQ_OP_WRITE_ZEROES 240is not supported. 241 242zoned (RO) 243---------- 244This indicates if the device is a zoned block device and the zone model of the 245device if it is indeed zoned. The possible values indicated by zoned are 246"none" for regular block devices and "host-aware" or "host-managed" for zoned 247block devices. The characteristics of host-aware and host-managed zoned block 248devices are described in the ZBC (Zoned Block Commands) and ZAC 249(Zoned Device ATA Command Set) standards. These standards also define the 250"drive-managed" zone model. However, since drive-managed zoned block devices 251do not support zone commands, they will be treated as regular block devices 252and zoned will report "none". 253 254Jens Axboe <jens.axboe@oracle.com>, February 2009 255