Skip to main content

Supported combinations

Mirrored from the FEMU repository

This page is hw/femu/docs/reference/combinations.md at FEMU d8b29b1e0 (2026-11-16), licensed GPL-2.0-or-later. Send corrections to the FEMU repository.

Each section is a device property and the refusals that name it: the messages realize, or a runtime setter, gives for a combination FEMU does not support. A property with no section combines with everything that its own description allows. A message keeps its format directives (%s, %d), which the running device fills in.

age_scale​

  • FEMU bbssd: age_scale must be 1 or more

ana​

  • action ana needs ana=on
  • ana group %u is not a namespace
  • ana must be the same on every controller of a femu-subsys
  • ana state must be 1, 2 or 3
  • ana takes <group>:<state>
  • ana-state needs a realized controller with ana=on
  • ana-state takes "group:state", a group of an allocated namespace and a state 1 to 4
  • ana-state: group %u is in Persistent Loss

atomic_write_lbas​

  • atomic_write_lbas: %s

awupf_lbas​

  • awupf_lbas: %s

bbssd_ns_limit​

  • bbssd_ns_limit must be between 1 and %u and cover the boot namespaces

blk_early_pe​

  • femu: blk_early_ppm needs blk_early_pe of 1 or more

blk_early_ppm​

  • femu: blk_early_ppm and erase_fail_ppm must be at most 1000000
  • femu: blk_early_ppm needs blk_early_pe of 1 or more

blk_pe_limit​

  • femu: %s is not supported with FDP and blk_pe_limit
  • femu: %s needs blk_pe_limit
  • femu: spare_lines needs blk_pe_limit

blk_pe_spread​

  • femu: blk_pe_spread must be 0 to 90, got %u

blks_per_pl​

  • FEMU bbssd: with a read-only end of life, failed_die leaves no spare, so the media would start read-only; lower devsz_mb or raise blks_per_pl
  • femu: spare_lines (%u) must be below blks_per_pl (%d)

buffer_low_pcent​

  • FEMU bbssd: buffer_low_pcent needs buffer_size, must be below buffer_thres_pcent, and does not work with shared namespaces or power_loss, got %d

buffer_size​

  • FEMU bbssd: buffer_low_pcent needs buffer_size, must be below buffer_thres_pcent, and does not work with shared namespaces or power_loss, got %d
  • FEMU bbssd: buffer_size must not be negative
  • FEMU bbssd: iu_pages is 0 to 64, and above 1 not with buffer_size, streams, mp_program, end-of-life read-only or FDP
  • power_loss requires bbssd, buffer_size > 0, and no metadata, namespace management or subsystem

buffer_thres_pcent​

  • FEMU bbssd: buffer_low_pcent needs buffer_size, must be below buffer_thres_pcent, and does not work with shared namespaces or power_loss, got %d
  • FEMU bbssd: buffer_thres_pcent must be between 1 and 100, got %d

cache_evict​

  • FEMU bbssd: unknown cache_evict "%s"; use clock, random, lru or arc

cancel​

  • femu: cancel is not supported with %s

cell_pages​

  • FEMU bbssd: cell_pages must be in [0, %d]

ch_xfer_lat​

  • FEMU bbssd: read_prio needs the plain channel model: no cmd_addr_lat, pg_xfer_lat, ch_xfer_lat or status_lat
  • OC 1.2 ch_xfer_lat must be nonnegative

cmbloc​

  • cmbloc selects base address register %u; the controller registers use 0 and 1 and the MSI-X table 4 and 5, so it must be 2

cmbsz​

  • cmbsz describes a %

cmd_addr_lat​

  • FEMU bbssd: read_prio needs the plain channel model: no cmd_addr_lat, pg_xfer_lat, ch_xfer_lat or status_lat

cqr​

  • vwc, intc, cqr and extended are single bits

csd_cp​

  • csd_cp needs a memory namespace: set csd_slm_size
  • csd_slm_size and csd_cp cannot be used with ns_mgmt, power_loss or a CXL link
  • csd_slm_size and csd_cp need a CSD namespace
  • csd_slm_size and csd_cp need femu_mode=4
  • csd_slm_size and csd_cp need room for %u more namespaces
  • ocp is not supported with the CSD memory or program namespaces (csd_slm_size, csd_cp)

csd_legacy_opcodes​

  • csd_legacy_opcodes needs a CSD namespace

csd_pause​

  • csd_pause cannot be used with shared namespaces: two clocks would time one FTL
  • csd_pause does not support icount
  • csd_pause needs KVM
  • csd_pause needs KVM_CLOCK_HOST_TSC and KVM_CAP_VM_TSC_CONTROL (Linux 6.0 or later)
  • csd_pause needs a CSD namespace
  • csd_pause needs the KVM TSC offset attribute
  • csd_pause needs the guest TSC at the host rate; do not set tsc-frequency
  • csd_pause supports KVM only on x86
  • csd_pause: namespace %d runs a mode other than bbssd or CSD
  • events_file does not work with csd_pause

csd_pause_max_ms​

  • CSD csd_pause_max_ms must be in range [1, 60000]

csd_slm_size​

  • CSD: cannot allocate csd_slm_size
  • a subsystem with csd_slm_size takes a single controller
  • csd_cp needs a memory namespace: set csd_slm_size
  • csd_slm_size and csd_cp cannot be used with ns_mgmt, power_loss or a CXL link
  • csd_slm_size and csd_cp need a CSD namespace
  • csd_slm_size and csd_cp need femu_mode=4
  • csd_slm_size and csd_cp need room for %u more namespaces
  • csd_slm_size must be at most %u MiB
  • ocp is not supported with the CSD memory or program namespaces (csd_slm_size, csd_cp)

csf_runtime_scale​

  • CSD csf_runtime_scale must be non-zero

cxl_ssd​

  • cxl_ssd must name a realized femu-cxl-ssd
  • cxl_ssd requires %s
  • cxl_ssd requires the femu-cxl-ssd to have ftl=on
  • die_slow requires bbssd or CSD with one namespace, and no ns_mgmt, namespace_modes, subsystem or cxl_ssd
  • die_stats requires bbssd, with no namespace_modes, subsystem or cxl_ssd
  • femu: onfi_mode is not supported with cxl_ssd
  • femu: the pd_* properties are not supported with cxl_ssd
  • hmb_mb is not supported with cxl_ssd

devsz_mb​

  • FEMU bbssd: with a read-only end of life, failed_die leaves no spare, so the media would start read-only; lower devsz_mb or raise blks_per_pl
  • devsz_mb must be unset or %

did​

  • the FTL did not pause in 10 s
  • the FTL queues did not drain in 10 s
  • the FTL thread did not take the query; the controller may be paused or disabled
  • the FTL thread did not take x-ftl-hold in 5 s

die_parity​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies

die_slow​

  • action slow needs die_slow=on
  • die-slow needs a realized controller with die_slow=on
  • die_slow requires bbssd or CSD with one namespace, and no ns_mgmt, namespace_modes, subsystem or cxl_ssd
  • die_slow: nchs must be 1 to %d and luns_per_ch 1 to %d

die_stats​

  • die_stats does not count plane_reads, pe_suspend, read priority, cache reads or bus timing yet
  • die_stats requires bbssd, with no namespace_modes, subsystem or cxl_ssd
  • namespace %u: kind 'dies' needs die_stats=on

dpc​

  • dpc: bit 3 (protection information in the first eight bytes of metadata) is not allowed by NVM 1.1
  • dps needs 8 bytes of metadata and a matching dpc
  • meta: protection information (dpc, dps) is not supported

dps​

  • dps needs 8 bytes of metadata and a matching dpc
  • meta: protection information (dpc, dps) is not supported

dram_hit_ns​

  • FEMU bbssd: dram_hit_ns must not be negative

durable_at_done​

  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies
  • durable_at_done is not supported with %s
  • durable_at_done needs power_loss=on

ecc_retention_sec​

  • femu: pd_open_age needs ecc_retention_sec above 0

ecc_step_ns​

  • femu: the pd_* program disturb read terms need ecc_step_ns above 0

entries​

  • entries must be in [1, 65534]
  • femu: %s needs queues x (entries + 1) of at most %d

eol_read_only​

  • FEMU bbssd: eol_read_only needs %
  • femu: erase_fail_ppm needs eol_read_only=on
  • femu: rain_read_only needs rain_parity=1 and eol_read_only=on
  • x-eol-enter requires eol_read_only on namespace 1

erase_fail_ppm​

  • femu: blk_early_ppm and erase_fail_ppm must be at most 1000000
  • femu: erase_fail_ppm is not supported with FDP
  • femu: erase_fail_ppm needs eol_read_only=on

events_file​

  • events_file does not work with csd_pause
  • events_file line %d: %s
  • events_file line %d: longer than %d bytes
  • events_file: a text file of at most %d KiB
  • events_file: more than %d events
  • events_file: more than %d lines
  • test clock does not work with events_file
  • the femu device has no events_file
  • x-events-now needs an events_file

extended​

  • meta/extended need a matching metadata capability (mc)
  • vwc, intc, cqr and extended are single bits

failed_die​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies
  • FEMU bbssd: with a read-only end of life, failed_die leaves no spare, so the media would start read-only; lower devsz_mb or raise blks_per_pl

fdm_size​

  • CSD fdm_size is too large
  • CSD mode requires fdm_size to be non-zero

fdp​

  • FEMU %s: fdp.nrg above 1 needs a logical block of whole NAND pages (%
  • FEMU bbssd: fdp.nrg (%u) must divide nchs (%d), or be a multiple of nchs that divides the %u dies of nchs x luns_per_ch: a reclaim group is a set of whole channels, or of LUNs of one channel
  • FEMU bbssd: fdp.runs must be %
  • fdp.nrg (%u) leaves the placement handle %u bits, too few for fdp.nruh (%u) handles
  • fdp.nrg (%u) must not exceed %u
  • fdp.nrg must be non-zero
  • fdp.nru (%
  • fdp.nruh (%u) must not exceed %u: the size of an FDP configuration descriptor is 16 bits
  • fdp.nruh (%u) must not exceed fdp.nru (%
  • fdp.nruh must be non-zero
  • fdp.runs (%
  • femu-subsys with fdp=on takes a single controller
  • meta: not supported with placement (fdp)
  • namespace_phls needs fdp=on on the subsystem

fdp_trim_erase_all​

  • fdp_trim_erase_all erases every namespace; it needs a single namespace

femu_mode​

  • csd_slm_size and csd_cp need femu_mode=4
  • femu_mode must be 0 (OpenChannel), 1 (black-box), 2 (no-SSD), 3 (zoned), 4 (computational storage) or 5 (key-value)
  • kv_index=lsm and the kv_lsm_* knobs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_max_keys needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_pack=off needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_timing=model and kv_model_coeffs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • mode=%s and femu_mode=%u name different modes
  • ocssd is a controller mode: femu_mode and namespace_modes must both select it, or neither
  • persist_ftl needs femu_mode=1 (bbssd)

flash_type​

  • flash_type must be 1 (SLC), 2 (MLC), 3 (TLC) or 4 (QLC)

fw_cores​

  • fw_cores must be 1 to 64, got %u

fw_slots​

  • fw_slots above 1%s does not work with a femu-subsys or a CXL device yet
  • fw_slots must be 0 (auto) to 7
  • ocp needs fw_slots of 3 or more (FWUP-6)

gc_copyback​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl
  • femu: gc_copyback is not supported with %s

gc_frontier​

  • FEMU %s: gc_frontier is not supported with %s
  • gc_frontier must match the controller that owns the shared namespaces

gc_idle_ms​

  • FEMU %s: gc_idle_ms is not supported with %s
  • FEMU %s: maint_idle needs gc_idle_ms
  • FEMU: gc_idle_ms is not supported with %s
  • FEMU: gc_idle_ms is not supported with shared namespaces
  • gc idle check requires gc_idle_ms
  • gc idle hold requires gc_idle_ms
  • gc idle kick requires gc_idle_ms
  • gc idle latch requires gc_idle_ms
  • gc idle order requires gc_idle_ms
  • maint_idle needs gc_idle_ms
  • maintenance check requires gc_idle_ms

gc_pace​

  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies

gc_policy​

  • FEMU %s: set gc_policy or gc_strategy under FDP, not both
  • FEMU bbssd: unknown gc_policy "%s"
  • FEMU kvssd: gc_policy=%s is not supported; KV mode always reclaims the emptiest line

gc_strategy​

  • FEMU %s: set gc_policy or gc_strategy under FDP, not both
  • FEMU bbssd: gc_strategy must be %d (greedy), %d (cost-benefit), %d (random) or %d (per-handle)

gc_thres_pcent​

  • FEMU bbssd: gc_thres_pcent and gc_thres_pcent_high must be in [1, 100], got %d and %d
  • FEMU bbssd: gc_thres_pcent_high (%d) must not be below gc_thres_pcent (%d); the forced watermark is reached after the background one, not before

gc_thres_pcent_high​

  • FEMU bbssd: gc_thres_pcent and gc_thres_pcent_high must be in [1, 100], got %d and %d
  • FEMU bbssd: gc_thres_pcent_high (%d) must not be below gc_thres_pcent (%d); the forced watermark is reached after the background one, not before

hmb_mb​

  • FEMU bbssd: hmb_mb is at most 65536 and needs mapping=dftl without FDP, KV or shared namespaces; hmb_ns must not be negative
  • hmb_mb %u MiB is smaller than the mapping table of this device (8 bytes per page)
  • hmb_mb is not supported with cxl_ssd
  • hmb_mb needs bbssd or CSD namespaces; namespace %d is not

hmb_ns​

  • FEMU bbssd: hmb_mb is at most 65536 and needs mapping=dftl without FDP, KV or shared namespaces; hmb_ns must not be negative

hot_cold_sep​

  • FEMU %s: hot_cold_sep is not supported under FDP: the placement handle of a write chooses its reclaim unit, and the overwrite pointer writes past it; give hot data a handle of its own

hot_k​

  • femu: hot_k needs throttle=on

inject_die​

  • FEMU bbssd: inject_die must be 0 to dies - 1

intc​

  • vwc, intc, cqr and extended are single bits

io_stages​

  • stage_ns needs io_stages=on
  • the femu device does not time I/O stages; set io_stages=on

io_trace​

  • femu-save-data: '%s' is a device's io_trace
  • io_trace must not be empty
  • io_trace: '%s' is a persist_file, pel_file or io_trace of a device
  • io_trace: cannot write '%s'
  • io_trace: no directory for '%s'

iu_pages​

  • FEMU bbssd: iu_pages is 0 to 64, and above 1 not with buffer_size, streams, mp_program, end-of-life read-only or FDP

kv_index​

  • FEMU kvssd: kv_index=lsm needs a page of at least %d bytes; the page is %d
  • kv_index=%s is not hash or lsm
  • kv_index=lsm and the kv_lsm_* knobs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_index=lsm and the kv_lsm_* knobs need a key-value namespace; shared namespaces are NoSSD or bbssd
  • kv_lsm_memtable_keys, kv_lsm_size_ratio and kv_lsm_pinned_levels need kv_index=lsm

kv_lsm_memtable_keys​

  • kv_lsm_memtable_keys, kv_lsm_size_ratio and kv_lsm_pinned_levels need kv_index=lsm
  • kv_lsm_memtable_keys=%u is not %d to %d

kv_lsm_pinned_levels​

  • kv_lsm_memtable_keys, kv_lsm_size_ratio and kv_lsm_pinned_levels need kv_index=lsm
  • kv_lsm_pinned_levels=%u is not 0 to %d

kv_lsm_size_ratio​

  • kv_lsm_memtable_keys, kv_lsm_size_ratio and kv_lsm_pinned_levels need kv_index=lsm
  • kv_lsm_size_ratio=%u is not %d to %d

kv_max_keys​

  • kv_max_keys needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_max_keys needs a key-value namespace; shared namespaces are NoSSD or bbssd
  • kv_max_keys=%u is not 0 to %u

kv_model_coeffs​

  • FEMU kvssd: kv_model_coeffs: %s; give the intercept and %d coefficients
  • kv_timing=model and kv_model_coeffs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_timing=model and kv_model_coeffs need a key-value namespace; shared namespaces are NoSSD or bbssd

kv_pack​

  • kv_pack=off needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_pack=off needs a key-value namespace; shared namespaces are NoSSD or bbssd

kv_timing​

  • FEMU kvssd: kv_timing=%s is not media or model
  • kv_timing=model and kv_model_coeffs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_timing=model and kv_model_coeffs need a key-value namespace; shared namespaces are NoSSD or bbssd

l2p_dram_ns​

  • FEMU bbssd: l2p_dram_ns must not be negative, and does not work with onfi_cache_read or FDP

lba_index​

  • femu-save-data: namespace %u is formatted to LBA format %u, not lba_index %u; a later run could not load the file
  • nlbaf must be in [1, 16] and lba_index below it
  • ocp needs lba_index 0: the factory sector size is LBA format 0 (DP-CFG-1 B, NSM-3)
  • zoned namespaces need a logical block of 4 KiB or less; lba_index selects %u bytes

lnum_ch​

  • FEMU ocssd: lnum_ch must not exceed %d and lnum_ch * lnum_lun must not exceed %d, got %u and %u
  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero

lnum_lun​

  • FEMU ocssd: lnum_ch must not exceed %d and lnum_ch * lnum_lun must not exceed %d, got %u and %u
  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero

lnum_pln​

  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero
  • FEMU ocssd: lnum_pln must be 1, 2 or 4, got %d

lockdown​

  • lockdown does not work with a femu-subsys yet

lpgs_per_blk​

  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero

lsec_size​

  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero

lsecs_per_pg​

  • FEMU ocssd: lnum_ch, lnum_lun, lnum_pln, lsecs_per_pg, lpgs_per_blk and lsec_size must all be greater than zero

luns_per_ch​

  • FEMU bbssd: fdp.nrg (%u) must divide nchs (%d), or be a multiple of nchs that divides the %u dies of nchs x luns_per_ch: a reclaim group is a set of whole channels, or of LUNs of one channel
  • die_slow: nchs must be 1 to %d and luns_per_ch 1 to %d

lver​

  • lver must be 1 (Open-Channel 1.2) or 2 (Open-Channel 2.0)

maint_idle​

  • FEMU %s: maint_idle needs gc_idle_ms
  • maint_idle needs gc_idle_ms

mapping​

  • FEMU %s: mapping=%s is not supported under FDP: the log-block mappings write and merge through lines that are not FDP-aware; use page or dftl
  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: hmb_mb is at most 65536 and needs mapping=dftl without FDP, KV or shared namespaces; hmb_ns must not be negative
  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl
  • FEMU bbssd: unknown mapping "%s"
  • Streams requires page or dftl mapping
  • femu: pd_open_read_limit and pd_open_close_sec are not supported with mapping=%s
  • hmb_mb %u MiB is smaller than the mapping table of this device (8 bytes per page)
  • persist_ftl needs mapping=page
  • the FTL section comes from other FTL properties (geometry, mapping, GC, NAND timing or aging)

max_cqes​

  • max_sqes must be %d and max_cqes %d, the entry sizes of the NVM command set

max_sqes​

  • max_sqes must be %d and max_cqes %d, the entry sizes of the NVM command set

mc​

  • meta/extended need a matching metadata capability (mc)

mdts​

  • femu: %s needs mdts 1 to %d

meta​

  • meta/extended need a matching metadata capability (mc)
  • meta: at most 8 block sizes (nlbaf), each is also offered with metadata
  • meta: namespace %d runs a mode without metadata support (block or no-SSD only)
  • meta: not supported with placement (fdp)
  • meta: protection information (dpc, dps) is not supported
  • pif is 0 or 2; 2 needs pi=on and meta of 16 or more

mode​

  • CSD mode requires fdm_size to be non-zero
  • FDP with more than one namespace needs every namespace in bbssd mode or every one in nossd mode
  • FEMU kvssd: gc_policy=%s is not supported; KV mode always reclaims the emptiest line
  • csd_pause: namespace %d runs a mode other than bbssd or CSD
  • der=cylon requires cylon-kernel-ack=on after reviewing the host dual-mode leaf lifetime fix
  • meta: namespace %d runs a mode without metadata support (block or no-SSD only)
  • mode '%s' is not one of ocssd, bbssd, nossd, zns, csd or kv
  • mode=%s and femu_mode=%u name different modes
  • namespace lifecycle is unsupported for this mode
  • namespace lifecycle requires a homogeneous mode
  • namespace_modes: unknown mode '%s'
  • ocp needs bbssd or CSD namespaces; namespace %d is mode %u
  • ocssd is a controller mode: femu_mode and namespace_modes must both select it, or neither
  • persist_file '%s' does not match this device: the mode, the backend size (%
  • shared namespace mode and capabilities must match
  • shared_ftl needs every namespace in bbssd mode; namespace %d is not
  • test-storm is threads,pages,iters,mode,seed with at most one thread per vCPU

mp_program​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: iu_pages is 0 to 64, and above 1 not with buffer_size, streams, mp_program, end-of-life read-only or FDP
  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl

mp_read​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl

mpsmax​

  • mpsmax must be in [mpsmin, 15]

mpsmin​

  • mpsmax must be in [mpsmin, 15]

ms​

  • %s is at most 60000 ms
  • a time in ms
  • a window is <ms>+<window>~<seed>, window > 0

multipoller_enabled​

  • multipoller_enabled must be 0 (one poller for all queues) or 1 (each poller owns poller_ratio queues)
  • wrr needs one poller: multipoller_enabled=0, or poller_ratio of at least queues

namespace_modes​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • die_slow requires bbssd or CSD with one namespace, and no ns_mgmt, namespace_modes, subsystem or cxl_ssd
  • die_stats requires bbssd, with no namespace_modes, subsystem or cxl_ssd
  • kv_index=lsm and the kv_lsm_* knobs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_max_keys needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_pack=off needs a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • kv_timing=model and kv_model_coeffs need a key-value namespace (femu_mode=5 or kvssd in namespace_modes)
  • namespace_modes count must equal namespaces=%u
  • namespace_modes: unknown mode '%s'
  • ocssd is a controller mode: femu_mode and namespace_modes must both select it, or neither

namespace_phls​

  • namespace_phls count must equal namespaces=%u
  • namespace_phls has no entry for namespace %u
  • namespace_phls needs fdp=on on the subsystem
  • namespace_phls: namespace %u lists handle %
  • namespace_phls: namespace %u lists more than %u handles
  • namespace_phls: namespace %u lists no handle
  • namespace_phls: namespace %u names handle '%s'; handles are 0 to %u

namespace_sizes​

  • namespace_sizes count must equal namespaces=%u
  • namespace_sizes sum (%
  • namespace_sizes: '%s' is smaller than a sector
  • namespace_sizes: invalid size '%s'

namespaces​

  • FDP needs namespaces of the NVM command set; namespace 1 is %s
  • FEMU bbssd: a reclaim group must hold every namespace, since a host may place all of its data in one: the namespaces expose %
  • FEMU bbssd: buffer_low_pcent needs buffer_size, must be below buffer_thres_pcent, and does not work with shared namespaces or power_loss, got %d
  • FEMU bbssd: hmb_mb is at most 65536 and needs mapping=dftl without FDP, KV or shared namespaces; hmb_ns must not be negative
  • FEMU bbssd: the %u namespaces expose %
  • FEMU bbssd: the %u namespaces share one FTL and expose %
  • FEMU: gc_idle_ms is not supported with shared namespaces
  • a rotational subsystem needs NVM namespaces; namespace %d is %s
  • bbssd_ns_limit must be between 1 and %u and cover the boot namespaces
  • csd_pause cannot be used with shared namespaces: two clocks would time one FTL
  • csd_slm_size and csd_cp need room for %u more namespaces
  • femu: onfi_mode is not supported with ns_mgmt or shared namespaces
  • femu: onfi_mode needs bbssd, CSD or KV namespaces; namespace %u is not one
  • femu: pd_open_read_limit and pd_open_close_sec need bbssd or CSD namespaces; namespace %u is KV
  • femu: the pd_* program disturb properties need bbssd, CSD or KV namespaces; namespace %u is not one
  • femu: the pd_* properties are not supported with ns_mgmt or shared namespaces
  • femu: throttle works with bbssd or CSD namespaces only
  • gc_frontier must match the controller that owns the shared namespaces
  • hmb_mb needs bbssd or CSD namespaces; namespace %d is not
  • kv_index=lsm and the kv_lsm_* knobs need a key-value namespace; shared namespaces are NoSSD or bbssd
  • kv_max_keys needs a key-value namespace; shared namespaces are NoSSD or bbssd
  • kv_pack=off needs a key-value namespace; shared namespaces are NoSSD or bbssd
  • kv_timing=model and kv_model_coeffs need a key-value namespace; shared namespaces are NoSSD or bbssd
  • namespace %u: kind '%s' applies only to bbssd namespaces
  • namespace %u: kind '%s' does not apply to %s namespaces
  • namespace %u: kind 'zones' applies only to znssd namespaces
  • namespace %u: query-femu does not support %s namespaces yet
  • namespace_modes count must equal namespaces=%u
  • namespace_phls count must equal namespaces=%u
  • namespace_sizes count must equal namespaces=%u
  • namespaces must be in [1, %d]
  • nand_profile needs bbssd, CSD, KV or ZNS namespaces; namespace %d is %s
  • ocp is not supported with the CSD memory or program namespaces (csd_slm_size, csd_cp)
  • ocp needs bbssd or CSD namespaces; namespace %d is mode %u
  • plane_reads works with bbssd, CSD or KV namespaces only (KV keeps the LUN rule)
  • power_loss requires page-aligned namespaces and whole LBAs per NAND page
  • power_states works with bbssd or CSD namespaces only
  • shared namespaces require homogeneous NoSSD or bbssd without Streams or default protection
  • streams requires NVM namespaces
  • the loss scan needs a controller whose namespaces are its own
  • zoned namespaces need a logical block of 4 KiB or less; lba_index selects %u bytes

nand_cell_type​

  • FEMU bbssd: nand_cell_type supports at most %d pgs_per_blk
  • femu: tier_mode=%s needs nand_cell_type 2 (MLC), 3 (TLC) or 4 (QLC): the pool programs dense blocks one bit per NAND unit
  • femu: tier_mode=%s needs pgs_per_blk of at least %d at nand_cell_type %u

nand_profile​

  • nand_profile needs a controller that owns its media; a shared or CXL linked FTL takes its timing elsewhere
  • nand_profile needs bbssd, CSD, KV or ZNS namespaces; namespace %d is %s
  • nand_profile sets the base timing, so %s must keep its default
  • nand_profile: %s
  • nand_profile: %s holds a NUL byte

nchs​

  • FEMU bbssd: fdp.nrg (%u) must divide nchs (%d), or be a multiple of nchs that divides the %u dies of nchs x luns_per_ch: a reclaim group is a set of whole channels, or of LUNs of one channel
  • die_slow: nchs must be 1 to %d and luns_per_ch 1 to %d

nlbaf​

  • meta: at most 8 block sizes (nlbaf), each is also offered with metadata
  • nlbaf must be in [1, 16] and lba_index below it

nr_cu​

  • CSD nr_cu must be in range [1, 64]

nr_thread​

  • CSD nr_thread must be non-zero

ns_mgmt​

  • csd_slm_size and csd_cp cannot be used with ns_mgmt, power_loss or a CXL link
  • die_slow requires bbssd or CSD with one namespace, and no ns_mgmt, namespace_modes, subsystem or cxl_ssd
  • femu: onfi_mode is not supported with ns_mgmt or shared namespaces
  • femu: the pd_* properties are not supported with ns_mgmt or shared namespaces
  • ns_mgmt=on does not support subsys; use a standalone controller
  • ocp is not supported with ns_mgmt
  • stage_ns is not supported with ns_mgmt

ns_wp​

  • femu: ns_wp is not supported with %s

oacs​

  • oacs may only set Format NVM (0x%x)

oc20_page_types​

  • oc20_page_types needs at most %u pages per block

ocp​

  • femu: ocp with throttle needs thermal_step_ms of 1000 or less
  • ocp is not supported with a femu-subsys
  • ocp is not supported with ns_mgmt
  • ocp is not supported with the CSD memory or program namespaces (csd_slm_size, csd_cp)
  • ocp needs bbssd or CSD namespaces; namespace %d is mode %u
  • ocp needs fw_slots of 3 or more (FWUP-6)
  • ocp needs lba_index 0: the factory sector size is LBA format 0 (DP-CFG-1 B, NSM-3)
  • ocp_profile needs ocp=on
  • plp with ocp=on needs vwc=0
  • power_states does not work with ocp yet (OCP C7h maps DSSD power states to PS0)

ocp_profile​

  • ocp_profile DP-CFG-%d A needs %s, which FEMU does not support; use B
  • ocp_profile DP-CFG-6 A needs plp=on
  • ocp_profile DP-CFG-6 B needs plp_fail_mode rom
  • ocp_profile must be %d letters, each A or B (DP-CFG-1 to DP-CFG-%d)
  • ocp_profile needs ocp=on

oncs​

  • oncs may only set Compare, Write Uncorrectable, DSM, Write Zeroes, Save/Select Feature Support, Verify and Copy

onfi_cache_read​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: l2p_dram_ns must not be negative, and does not work with onfi_cache_read or FDP
  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl
  • femu: onfi_cache_read is refused with %s
  • plane_reads needs the plain channel model: no ONFI profile, onfi_cache_read, bus phases, pe_suspend or read_prio

onfi_mode​

  • femu: %s is refused with onfi_mode, which sets the channel bus phases
  • femu: onfi_mode is not supported with cxl_ssd
  • femu: onfi_mode is not supported with ns_mgmt or shared namespaces
  • femu: onfi_mode must be -1 (off) or 0 to %d
  • femu: onfi_mode needs an even page size on the bus (secsz * secs_per_pg + onfi_spare_bytes)
  • femu: onfi_mode needs bbssd, CSD or KV namespaces; namespace %u is not one
  • femu: pe_suspend is refused with onfi_mode
  • femu: read_prio is refused with onfi_mode

onfi_mts​

  • femu: onfi_mts must be 0 to %d

onfi_param_page​

  • femu: onfi_param_page %s: %s
  • femu: onfi_param_page: %s

onfi_spare_bytes​

  • femu: onfi_mode needs an even page size on the bus (secsz * secs_per_pg + onfi_spare_bytes)
  • femu: onfi_spare_bytes and onfi_tccs_ns must not be negative

onfi_tccs_ns​

  • femu: onfi_spare_bytes and onfi_tccs_ns must not be negative

parity_xor_ns​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity

pd_edge_tiers​

  • femu: pd_edge_tiers and pd_open_tiers must be at most %d, the ECC tier cap
  • femu: pd_edge_wls and pd_edge_tiers must be set together
  • femu: pd_edge_wls, pd_edge_tiers and pd_open_tiers must not be negative

pd_edge_wls​

  • femu: pd_edge_wls and pd_edge_tiers must be set together
  • femu: pd_edge_wls must be below %d, the wordlines of a block
  • femu: pd_edge_wls, pd_edge_tiers and pd_open_tiers must not be negative

pd_open_age​

  • femu: pd_open_age needs ecc_retention_sec above 0

pd_open_close_sec​

  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies
  • femu: pd_open_read_limit and pd_open_close_sec are not supported with mapping=%s
  • femu: pd_open_read_limit and pd_open_close_sec must not be negative
  • femu: pd_open_read_limit and pd_open_close_sec need bbssd or CSD namespaces; namespace %u is KV

pd_open_read_limit​

  • femu: pd_open_read_limit and pd_open_close_sec are not supported with mapping=%s
  • femu: pd_open_read_limit and pd_open_close_sec must not be negative
  • femu: pd_open_read_limit and pd_open_close_sec need bbssd or CSD namespaces; namespace %u is KV
  • femu: pd_open_read_limit must not be above read_reclaim_limit

pd_open_tiers​

  • femu: pd_edge_tiers and pd_open_tiers must be at most %d, the ECC tier cap
  • femu: pd_edge_wls, pd_edge_tiers and pd_open_tiers must not be negative

pe_suspend​

  • FEMU bbssd: pe_suspend_max and tresume_ns need pe_suspend
  • die_stats does not count plane_reads, pe_suspend, read priority, cache reads or bus timing yet
  • femu: pe_suspend is refused with onfi_mode
  • plane_reads needs the plain channel model: no ONFI profile, onfi_cache_read, bus phases, pe_suspend or read_prio

pe_suspend_max​

  • FEMU bbssd: pe_suspend_max and tresume_ns need pe_suspend

pel_file​

  • femu-save-data: '%s' is a device's pel_file
  • io_trace: '%s' is a persist_file, pel_file or io_trace of a device
  • persist_file and pel_file name one file

perf_hints​

  • perf_hints: %s

persist_file​

  • FEMU: rrl is not supported with persist_file
  • femu: rain_loss is not supported with persist_file
  • io_trace: '%s' is a persist_file, pel_file or io_trace of a device
  • persist_file '%s' cannot be read in full
  • persist_file '%s' does not match this device: the mode, the backend size (%
  • persist_file '%s' has a short FTL section
  • persist_file '%s' has a short LBA status section
  • persist_file '%s' has bytes past its FTL state
  • persist_file '%s' has no FTL state, and an aging model needs the media history
  • persist_file '%s' is in use by another device
  • persist_file '%s', namespace %u: %s
  • persist_file '%s', namespace %u: the FTL section is longer than this FTL's
  • persist_file '%s': %s
  • persist_file '%s': the LBA status section is too long
  • persist_file and pel_file name one file
  • persist_file is not supported with %s
  • persist_file must not be empty
  • persist_file: '%s' ends in .lock, which names a lock file
  • persist_file: '%s' is a link to nothing
  • persist_ftl needs persist_file
  • power_limit is not supported with persist_file

persist_ftl​

  • persist_ftl is not supported with %s
  • persist_ftl is not supported with FDP
  • persist_ftl needs femu_mode=1 (bbssd)
  • persist_ftl needs mapping=page
  • persist_ftl needs persist_file
  • persist_ftl: a fetched request left without reaching the FTL (a queue deleted or a reset under I/O, or a drain that timed out); nothing saved
  • persist_ftl: fetched requests stayed on the FTL queue; nothing saved
  • persist_ftl: namespace %u has an FTL larger than 4 GiB to save

pg_xfer_lat​

  • FEMU bbssd: read_prio needs the plain channel model: no cmd_addr_lat, pg_xfer_lat, ch_xfer_lat or status_lat

pgs_per_blk​

  • FEMU bbssd: nand_cell_type supports at most %d pgs_per_blk
  • femu: tier_mode=%s needs pgs_per_blk of at least %d at nand_cell_type %u

pi​

  • pif is 0 or 2; 2 needs pi=on and meta of 16 or more

pif​

  • pif is 0 or 2; 2 needs pi=on and meta of 16 or more

plane_reads​

  • die_stats does not count plane_reads, pe_suspend, read priority, cache reads or bus timing yet
  • plane_reads needs the plain channel model: no ONFI profile, onfi_cache_read, bus phases, pe_suspend or read_prio
  • plane_reads works with bbssd, CSD or KV namespaces only (KV keeps the LUN rule)

plp​

  • ocp_profile DP-CFG-6 A needs plp=on
  • plp is not supported with %s
  • plp needs power_loss=on
  • plp with ocp=on needs vwc=0
  • plp-health needs plp=on and 0 to 65534
  • plp_fail_mode needs plp=on

plp_fail_mode​

  • ocp_profile DP-CFG-6 B needs plp_fail_mode rom
  • plp_fail_mode must be rom, wtm or normal
  • plp_fail_mode needs plp=on

plp_health​

  • plp_health must be 0 to 65534

poller_ratio​

  • multipoller_enabled must be 0 (one poller for all queues) or 1 (each poller owns poller_ratio queues)
  • wrr needs one poller: multipoller_enabled=0, or poller_ratio of at least queues

power_limit​

  • power_limit is not supported with persist_file
  • power_limit needs power_states

power_loss​

  • FEMU bbssd: buffer_low_pcent needs buffer_size, must be below buffer_thres_pcent, and does not work with shared namespaces or power_loss, got %d
  • action power-loss needs power_loss=on
  • csd_slm_size and csd_cp cannot be used with ns_mgmt, power_loss or a CXL link
  • durable_at_done needs power_loss=on
  • plp needs power_loss=on
  • power_loss requires bbssd, buffer_size > 0, and no metadata, namespace management or subsystem
  • power_loss requires page-aligned namespaces and whole LBAs per NAND page
  • sanitize_ms and sanitize_ops do not work with power_loss yet
  • sdc needs power_loss=on
  • simulated power loss requires a realized power_loss device

power_meas​

  • a power_meas subsystem takes a single controller
  • femu: power_meas is not supported with %s

power_states​

  • power_limit needs power_states
  • power_states does not work with a femu-subsys yet
  • power_states does not work with ocp yet (OCP C7h maps DSSD power states to PS0)
  • power_states needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000
  • power_states works with bbssd or CSD namespaces only

prog_pass2_lag​

  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl

prog_pass2_ns​

  • FEMU bbssd: prog_pass2_ns must not be negative, prog_pass2_lag is 0 to 16 and needs it, and neither works with mp_program, mp_read, onfi_cache_read, gc_copyback, KV, FDP, the pSLC tier or a mapping other than page or dftl
  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies

ps_max_watts​

  • ps_max_watts must be 1 to 127

queues​

  • femu: %s needs queues x (entries + 1) of at most %d
  • multipoller_enabled must be 0 (one poller for all queues) or 1 (each poller owns poller_ratio queues)
  • queues must be in [1, %d]
  • the FTL queues did not drain in 10 s
  • wrr needs one poller: multipoller_enabled=0, or poller_ratio of at least queues
  • x-ftl-hold holds the FTL: the loss scan needs its queues drained

rain_loss​

  • femu: rain_loss is not supported with persist_file
  • femu: rain_loss needs rain_read_only and bbssd

rain_parity​

  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies
  • FEMU bbssd: rain_rebuild needs rain_parity
  • femu: rain_read_only needs rain_parity=1 and eol_read_only=on
  • rain_parity is off or die %d is out of range
  • rain_parity keeps two members at least

rain_read_only​

  • femu: rain_loss needs rain_read_only and bbssd
  • femu: rain_read_only is not supported with %s
  • femu: rain_read_only needs rain_parity=1 and eol_read_only=on

rain_rebuild​

  • FEMU bbssd: rain_rebuild needs rain_parity

rber_ppb​

  • FEMU %s: rber_ppb is not supported with %s
  • FEMU bbssd: die_parity and rain_parity need two dies at least and mapping=page, and does not work with mp_program, mp_read, onfi_cache_read, gc_copyback, rber_ppb, KV, namespace_modes or FDP; parity_xor_ns (0 or more) and failed_die (-1 to dies - 1) need die_parity
  • FEMU: rber_ppb is not supported with %s
  • FEMU: rrl needs rber_ppb
  • femu: rber_timing_only needs rber_ppb
  • x-rber-u requires rber_ppb

rber_timing_only​

  • femu: rber_timing_only needs rber_ppb

read_prio​

  • FEMU bbssd: read_prio needs the plain channel model: no cmd_addr_lat, pg_xfer_lat, ch_xfer_lat or status_lat
  • femu: read_prio is refused with onfi_mode
  • plane_reads needs the plain channel model: no ONFI profile, onfi_cache_read, bus phases, pe_suspend or read_prio

read_reclaim_limit​

  • FEMU bbssd: read_reclaim_limit must not be negative
  • femu: pd_open_read_limit must not be above read_reclaim_limit

ready_ms​

  • ready_ms and recovery_ms must be below 7500, the CAP.TO the host reads
  • recovery_ms must be 0 or at least ready_ms

recovery_ms​

  • ready_ms and recovery_ms must be below 7500, the CAP.TO the host reads
  • recovery_ms must be 0 or at least ready_ms

reservations​

  • reservations do not work with Open-Channel
  • reservations must be the same on every controller of a femu-subsys

retention_limit_sec​

  • FEMU bbssd: retention_limit_sec must not be negative

rotational​

  • a rotational subsystem does not offer FDP
  • a rotational subsystem needs NVM namespaces; namespace %d is %s

rrl​

  • FEMU: rrl is not supported with a subsystem
  • FEMU: rrl is not supported with persist_file
  • FEMU: rrl needs rber_ppb

sanitize_ms​

  • sanitize_ms and sanitize_ops do not work with power_loss yet
  • sanitize_ms does not work with a femu-subsys yet
  • sanitize_ms is at most 3600000 (one hour)

sanitize_ns​

  • femu: sanitize_ns is not supported with %s
  • femu: sanitize_ns needs sanitize_ops

sanitize_ops​

  • femu: sanitize_ns needs sanitize_ops
  • sanitize_ms and sanitize_ops do not work with power_loss yet

sdc​

  • sdc is not supported with %s
  • sdc needs power_loss=on
  • sdc_ppm and sdc_kind need sdc=on
  • x-sdc needs sdc=on and <nsid>:<kind>:<lba> of a namespace
  • x-sdc: a fault is already waiting

sdc_kind​

  • sdc_ppm and sdc_kind need sdc=on
  • sdc_ppm must be 0 to 1000000 and sdc_kind flip, lost, misdirect or mix

sdc_ppm​

  • sdc_ppm and sdc_kind need sdc=on
  • sdc_ppm must be 0 to 1000000 and sdc_kind flip, lost, misdirect or mix

secs_per_pg​

  • femu: onfi_mode needs an even page size on the bus (secsz * secs_per_pg + onfi_spare_bytes)

secsz​

  • femu: onfi_mode needs an even page size on the bus (secsz * secs_per_pg + onfi_spare_bytes)

shared_ftl​

  • shared_ftl is not supported with %s
  • shared_ftl is not supported with a femu-subsys (FDP shares its FTL already)
  • shared_ftl needs every namespace in bbssd mode; namespace %d is not
  • shared_ftl needs namespace sizes in whole NAND pages (%

shutdown_ms​

  • shutdown_ms must be below 5000, the time a Linux host waits for a shutdown by default

spare_lines​

  • femu: spare_lines (%u) must be below blks_per_pl (%d)
  • femu: spare_lines needs blk_pe_limit

stage_ns​

  • nsid %u: stage_ns times NSID 1 to %d
  • nsid needs stage_ns=on
  • stage_ns is not supported with ns_mgmt
  • stage_ns needs io_stages=on

status_lat​

  • FEMU bbssd: read_prio needs the plain channel model: no cmd_addr_lat, pg_xfer_lat, ch_xfer_lat or status_lat

streams​

  • FEMU bbssd: iu_pages is 0 to 64, and above 1 not with buffer_size, streams, mp_program, end-of-life read-only or FDP
  • streams requires NVM namespaces
  • streams requires bbssd or NoSSD with FDP disabled
  • streams.max must be between 1 and 32

stride​

  • stride must not exceed %d

subsys​

  • ana must be the same on every controller of a femu-subsys
  • femu-subsys with fdp=on takes a single controller
  • fw_slots above 1%s does not work with a femu-subsys or a CXL device yet
  • lockdown does not work with a femu-subsys yet
  • ns_mgmt=on does not support subsys; use a standalone controller
  • ocp is not supported with a femu-subsys
  • power_states does not work with a femu-subsys yet
  • reservations must be the same on every controller of a femu-subsys
  • sanitize_ms does not work with a femu-subsys yet
  • shared_ftl is not supported with a femu-subsys (FDP shares its FTL already)
  • x-opal does not work with a femu-subsys or a CXL device yet

telemetry_da4​

  • telemetry_da4 is not supported with FDP

thermal_r​

  • femu: thermal_tau_ms needs thermal_r

thermal_step_ms​

  • femu: ocp with throttle needs thermal_step_ms of 1000 or less
  • femu: thermal_step_ms must be 1 to %d, got %u

thermal_tau_ms​

  • femu: thermal_tau_ms is not supported with %s
  • femu: thermal_tau_ms needs thermal_r
  • femu: throttle needs thermal_tau_ms

throttle​

  • femu: hot_k needs throttle=on
  • femu: ocp with throttle needs thermal_step_ms of 1000 or less
  • femu: throttle needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000
  • femu: throttle needs thermal_tau_ms
  • femu: throttle works with bbssd or CSD namespaces only

throttle_heavy_pct​

  • femu: throttle needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000
  • power_states needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000

throttle_light_pct​

  • femu: throttle needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000
  • power_states needs 100 <= throttle_light_pct <= throttle_heavy_pct <= 1000

tier_full_policy​

  • femu: tier_full_policy must be direct or through, got %s

tier_lazy_pct​

  • femu: tier_lazy_pct must be 0 to 99, got %u

tier_max_mb​

  • femu: tier_max_mb=%u holds no whole pSLC line of %

tier_mode​

  • femu: tier_mode must be off, static or dynamic, got %s
  • femu: tier_mode=%s is not supported with %s
  • femu: tier_mode=%s needs nand_cell_type 2 (MLC), 3 (TLC) or 4 (QLC): the pool programs dense blocks one bit per NAND unit
  • femu: tier_mode=%s needs pgs_per_blk of at least %d at nand_cell_type %u
  • femu: tier_mode=%s needs tier_static_mb above 0%s

tier_pslc_erase_ns​

  • femu: tier_pslc_read_ns, tier_pslc_prog_ns and tier_pslc_erase_ns must be at most %llu ns

tier_pslc_prog_ns​

  • femu: tier_pslc_read_ns, tier_pslc_prog_ns and tier_pslc_erase_ns must be at most %llu ns

tier_pslc_read_ns​

  • femu: tier_pslc_read_ns, tier_pslc_prog_ns and tier_pslc_erase_ns must be at most %llu ns

tier_shrink_pct​

  • femu: tier_shrink_pct must be 0 to 99, got %u

tier_static_mb​

  • femu: tier_mode=%s needs tier_static_mb above 0%s

trcbsy​

  • femu: trcbsy must not be negative

tresume_ns​

  • FEMU bbssd: pe_suspend_max and tresume_ns need pe_suspend
  • FEMU bbssd: tresume_ns must not be negative

tsusp_ns​

  • FEMU bbssd: tsusp_ns must not be negative

vwc​

  • plp with ocp=on needs vwc=0
  • vwc, intc, cqr and extended are single bits

wl_spread​

  • FEMU bbssd: rain_parity is 0 or 1, and does not work with die_parity, prog_pass2_ns, wl_spread, pd_open_close_sec, Streams, durable_at_done, gc_pace or the pSLC tier; with failed_die it needs three dies, and it takes at most 1024 dies
  • FEMU bbssd: wl_spread must not be negative

wrr​

  • wrr needs one poller: multipoller_enabled=0, or poller_ratio of at least queues

zns_blk_er_lat​

  • zns_flash_type %d has no built-in timing; set zns_pg_rd_lat, zns_pg_wr_lat and zns_blk_er_lat

zns_chnls_per_zone​

  • zns_chnls_per_zone %u must divide zns_num_ch %u

zns_ecc_step_ns​

  • zns_ecc_step_ns, and zns_pg_rd_lat with zns_page_types, must be at most %

zns_flash_type​

  • zns_flash_type %d has no built-in timing; set zns_pg_rd_lat, zns_pg_wr_lat and zns_blk_er_lat
  • zns_flash_type must be in [%d, %d]
  • zns_page_types needs zns_flash_type %d to %d; type %d has no page type table

zns_frl_sec​

  • zns_frl_sec needs zns_fzr_sec: an excursion follows Finish Zone Recommended

zns_fzr_sec​

  • zns_frl_sec needs zns_fzr_sec: an excursion follows Finish Zone Recommended

zns_luns_per_zone​

  • zns_luns_per_zone %u must divide zns_num_lun %u

zns_num_blk​

  • zns geometry out of range: zns_num_ch in [1, %d], zns_num_lun >= 1, zns_num_plane in [1, %d], zns_num_blk >= 1

zns_num_ch​

  • zns geometry out of range: zns_num_ch in [1, %d], zns_num_lun >= 1, zns_num_plane in [1, %d], zns_num_blk >= 1
  • zns_chnls_per_zone %u must divide zns_num_ch %u

zns_num_lun​

  • zns geometry out of range: zns_num_ch in [1, %d], zns_num_lun >= 1, zns_num_plane in [1, %d], zns_num_blk >= 1
  • zns_luns_per_zone %u must divide zns_num_lun %u

zns_num_plane​

  • zns geometry out of range: zns_num_ch in [1, %d], zns_num_lun >= 1, zns_num_plane in [1, %d], zns_num_blk >= 1

zns_num_wc​

  • zns_num_wc (%u) exceeds the %
  • zns_num_wc value %u exceeds the number of zones %u

zns_page_types​

  • zns_ecc_step_ns, and zns_pg_rd_lat with zns_page_types, must be at most %
  • zns_page_types needs zns_flash_type %d to %d; type %d has no page type table

zns_pg_rd_lat​

  • zns_ecc_step_ns, and zns_pg_rd_lat with zns_page_types, must be at most %
  • zns_flash_type %d has no built-in timing; set zns_pg_rd_lat, zns_pg_wr_lat and zns_blk_er_lat

zns_pg_wr_lat​

  • zns_flash_type %d has no built-in timing; set zns_pg_rd_lat, zns_pg_wr_lat and zns_blk_er_lat

zns_rrl_sec​

  • zns_rrl_sec needs zns_rzr_sec: the limit follows Reset Zone Recommended

zns_rzr_sec​

  • zns_rrl_sec needs zns_rzr_sec: the limit follows Reset Zone Recommended

zns_zasl_bs​

  • zns_zasl_bs %u must be a power-of-two multiple of the %u byte controller page size

zns_zone_order​

  • zns_zone_order must be ch or lun, not '%s'

zns_zrwa_num​

  • zns_zrwa_num must be non-zero when ZRWA is enabled
  • zns_zrwafg_size and zns_zrwa_num have no effect without zns_zrwa_size

zns_zrwa_size​

  • zns_zrwa_size %
  • zns_zrwafg_size and zns_zrwa_num have no effect without zns_zrwa_size

zns_zrwafg_size​

  • zns_zrwafg_size and zns_zrwa_num have no effect without zns_zrwa_size
  • zns_zrwafg_size must be between 1 and %u when ZRWA is enabled, got %