solidigm: Eliminate <linux/limits.h>

ARG_MAX is defined in <limits.h>, per POSIX, but it's defined to be a
variable in glibc. Instead, get rid of it entirely by using asprintf to
construct the commands. This prevents us from trying a partially
constructed command that might do something unintentional. For
solidigm-market-log.c, there's no constants needed from
<linux/limits.h>. With this we can delete it both places.

Signed-off-by: Warner Losh <imp@bsdimp.com>

nvme: remove double free in persistent-event-log

The pevent_collected structure uses the buffer address which is allocated
using nvme_alloc_huge(). So, pevent_collected and mh.p has same address.
Hence, remove _cleanup_free_ from pevent_collected to prevent double free.

Signed-off-by: Francis Pravin <francis.p@samsung.com>
Signed-off-by: Sathyavathi M <sathya.m@samsung.com>

ocp: Add Get DSSD Power State Feature (FID: C7h)

Implements the Get DSSD Power State Feature (FID C7h). Also added an
optional flag that you can pass to the plugin command to print out
all three values at once.

Signed-off-by: Arthur Shau <arthurshau@meta.com>

nvme: initialize default library logging

The library will only print errors unless we initialize default logging.
By doing so, all nvme commands are enabled to also print debug
information.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme: use cleanup helper for nvme_root_t objects

Use a cleanup helper for the nvme_root_t objects which simplifies the
error paths and ensures we are always removing all the allocated
resources.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

build: bump libnvme wrap

Import nvme_init_default_logging.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

doc: add nvme connet ctrl-loss-tmo description

Describe the kernel retry a connection behavior.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

plugins/solidigm: Automatic enabling Data Area 4 when retrieving Telemetry.

vs-internal-log, stopped extracting Host Initiated Telemetry previous
snapshot, simplified the type name of Telemetry snapshots to CIT and HIT,
removed command options for setting namespace, and file prefix, and added
command option to set output folder.

Signed-off-by: Leonardo da Cunha <leonardo.da.cunha@solidigm.com>

doc: Regenerate all docs for v2.9.1

Signed-off-by: Daniel Wagner <dwagner@suse.de>

build: Update version to v2.9.1

Signed-off-by: Daniel Wagner <dwagner@suse.de>

plugins/spdk: drop the plugin

The plugin depends on the private header of libvnme and breaks the build
if the library is not a builtin dependency.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

doc: Regenerate all docs for v2.9

Signed-off-by: Daniel Wagner <dwagner@suse.de>

build: Update version to v2.9

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-print: expand the maximum field length to prevent misalignment

Fields like Generic and Device are longer for SPDK-controlled
devices, which causes misalignment.

Signed-off-by: Sebastian Brzezinka <sebastian.brzezinka@intel.com>
Signed-off-by: Daniel Wagner <dwagner@suse.de>

plugins/spdk: plugin to list SPDK subsystems

Since `nvme-cli` start using `sysfs` to gather information about
subsystems, SPDK devices that are represented using CUSE can
no longer be listed.

Support for the JSON format is disabled cause `nvme_ctrl_get_state` use
sysfs to get value, causing segmentation fault when it's missing.
It could be solved by providing an empty string instead of a NULL pointer.

This patch adds SPDK devices to the nvme_root topology tree,

     $ nvme spdk list-subsys

Signed-off-by: Sebastian Brzezinka <sebastian.brzezinka@intel.com>
Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-print: check if controller belong to SPDK

Check if controller belongs to SPDK cuse devices and compare paths
to make it possible to distinguish SPDK devices and not rely on
order and existence of a file.

Signed-off-by: Sebastian Brzezinka <sebastian.brzezinka@intel.com>
Signed-off-by: Daniel Wagner <dwagner@suse.de>

plugins/spdk: plugin to list SPDK controllers

Since libnvme and nvme-cli(2.0) are continuing as separate projects,
and libnvme use sysfs exclusively to gather information about NVMe
devices, legacy paths that allow specifying a custom directory ('-d`)
is no longer available.

SPDK, as a userspace driver, does not create a sysfs entry but makes it
possible to expose such devices via NVMe-cuse and since it behaves as
character device, most commands should be intact, except
listing devices and subsystems.

This patch adds SPDK devices to the nvme_root topology tree, which
allows for keeping functionality where both SPDK and kernel devices
are printed.

    $ nvme spdk list

Signed-off-by: Sebastian Brzezinka <sebastian.brzezinka@intel.com>
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-print-stdout: fix ASQB, ACQB and BMBBA prints

The fields needed be output only as the 52 most significant bits.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

build: bump libnvme wrap

Update to v1.9 release.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvmf-autoconnect: fix string encoding

systemd reports that the arguments are not correctly encoded. Replace
the = char with \x3 encoding.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme: identify primary ctrl do not accept nsid

The identify primary ctrl command do not accept the namespace argument.
Thus remove it.

While at it, also update identify secondary ctrl function to use same
argument name.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-wrap: remove unused nvme_cli_get_log_ana_groups()

nvme_cli_get_log_ana_groups() is not called anywhere.
In preparation for a libnvme change that changes the signature
of nvme_get_log_ana_groups(), just remove the function.

Signed-off-by: Caleb Sander Mateos <csander@purestorage.com>

nvme-print-json: use admin queue attributes definitions

Replace hard coded register shift and mask values instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: use CSTS definitions

Replace hard coded register value, shift and mask values.
Also add CSTS.ST output.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: use POWER_OF_TWO macro

Already used for json print.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: use controller configuration definitions

Replace hard corded register value, shift and mask values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json, nvme-print-stdout: add tertiary version number

Supported from version 1.2.1.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: use NVME_MAJOR and NVME_MINOR definitions

Replace hard coded shift and mask values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: use NVME_BOOT_PARTITION_INFO field definitions

Change it instead of hardcoded field mask and shift values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: add support telemetry log create event in PEL

Already supported but only for stdout print.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: add support set feature event in PEL

Already supported but only for stdout print.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: add obj_d() function

Also change json_d() to use the function.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: add obj_add_uint_0nx() function

Supports field width to add "0x" prepended and zero padded hexadecimal.
obj_add_uint_02x() also changed to use the function as field width 2.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

meson.build: Fix spelling of sed-opal.h filename

Add missing 'l' in sed-opal.h in description in meson.build.

Signed-off-by: Warner Losh <imp@bsdimp.com>

sed_opal: Only build when we really have sed_opal headers

There's already code to not build the sed opal plugin when the necessary
headers aren't present. However, it doesn't work, since HAVE_SED_OPAL is
always defined to be either 0 or 1, so the meson has() return true, even
when the value is 0, causing the sed-opal plugin to build, even when the
header files it needs are missing.

Signed-off-by: Warner Losh <imp@bsdimp.com>

nvme: fix error path if write() fails in get_telemetry_log()

Print an error message and return the error code to the caller.

Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>

plugins/solidigm: Additional LID for temperature statistics.

Temperature statistics log moved to different LID on some products.

Signed-off-by: Leonardo da Cunha <leonardo.da.cunha@solidigm.com>

plugins/solidigm: Improved Solidigm UUID index detection.

Code clean-up on solidigm log-page-directory.
UUID function calls following POSIX style.

Signed-off-by: Leonardo da Cunha <leonardo.da.cunha@solidigm.com>

nvme-print-stdout: Add CSTS.ST register stdout print

Added by NVMe revision 2.0 multiple controller shutdown enhancement.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Use NVME_CMIC definitions

Use libnvme definitions to print CMIC data.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print: Use NVME_UNIT definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Use NVME_CMBMSC/STS/EBS/SWTP definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

build: bump libnvme wrap

Include libnvme CMBSWTP property definitions changes.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

plugins/ssstc: add a new plugin for SSSTC driver

Signed-off-by: kobyou <kobyou@126.com>

nvmf-connect: systemd hardening effort

Apply the recommended hardening settings as recommended by openSUSE
and Fedor project. A few of the hardening option have to turned of
because nvme-cli needs write access to sysfs and /dev/nvme devices.

Links: https://en.opensuse.org/openSUSE:Security_Features#Systemd_hardening_effort
Links: https://fedoraproject.org/wiki/Changes/SystemdSecurityHardening
Signed-off-by: Daniel Wagner <dwagner@suse.de>

plugins/ocp: Adjusted variable type size carrying UUID index to 8 bit.

Changed uuidx type to __u8 in the related function calls,
because that's the type that will feed struct nvme_get_log_args.

Signed-off-by: Leonardo da Cunha <leonardo.da.cunha@solidigm.com>

nvme: fix security buffer allocation size

Signed-off-by: Wilfred Mallawa <wilfred.mallawa@wdc.com>

nvme: Change fw-download xfer initialization to set 0

Since by the default the FWUG value not checked.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme: Add fw-download warning message for FWUG alignment error

Check the firmware file size if not aligned with FWUG value.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

plugins/ocp: Split ocp_get_uuid_index() into find and get functions.

Enable reuse of OCP UUID discovery logic from other plugins.
Enable single call to nvme_identify_uuid() when dealing with multiple UUIDs

Signed-off-by: Leonardo da Cunha <leonardo.da.cunha@solidigm.com>

Revert "build(deps): bump AppImageCrafters/build-appimage from 1.3 to 2"

This reverts commit 158eb8d82c497c104cf9c2d4665983084cc8ae0e.

This breaks the build with

'appimagecrafters/appimage-builder:0.8.2' should be either '[path]/Dockerfile' or 'docker://image[:tag]'.

v2 was released a couple of years ago, so I don't expect this gets fixed
at all. Instead go back to last good version.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme: fix mem leak in multiple key related functions

Valgrind revealed mem leaks in multiple key related functions
including gen_tls_key() and check_tls_key() during insert, and
gen_dhchap_key(). This was traced to the respective cfg pointer
not getting freed after fetching the hostnqn string from the
hostnqn file. Fix the same.

Signed-off-by: Martin George <marting@netapp.com>

fabrics: fix mem leak in nvmf_check_hostid_and_hostnqn()

Valgrind revealed a mem leak in nvmf_check_hostid_and_hostnqn()
during nvme discover & connect. Leak traced to both hostid
pointers not getting freed for cases where the hostid value
matches the strings from both these respective hostid pointers.
Fix the same.

Signed-off-by: Martin George <marting@netapp.com>

build(deps): bump AppImageCrafters/build-appimage from 1.3 to 2

Bumps [AppImageCrafters/build-appimage](https://github.com/appimagecrafters/build-appimage) from 1.3 to 2.
- [Release notes](https://github.com/appimagecrafters/build-appimage/releases)
- [Commits](https://github.com/appimagecrafters/build-appimage/compare/v1.3...v2)

---
updated-dependencies:
- dependency-name: AppImageCrafters/build-appimage
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

nvme-print-stdout: Use NVME_BPRSEL definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Use NVME_BPINFO definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Use NVME_CMBSZ definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Use NVME_CMBLOC definitions

Use libnvme register definitions instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

Revert "logging: fix error handling for ioctl passthru wrappers"

This reverts commit a2b1aecb9075c7a8f9bad627fc45171bb533c8bd.

The result field is undefined for many commands and thus can have random
values. Thus we can't blindly evaluate these for all commands.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

fabrics: align identify command payload

The payload should be properly aligned.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

ocp: Unify line start spacing as tabs

Most of the ocp-nvme.c file already uses tabs in this way, but some
functions are using spaces instead. Unify this to follow the checkpatch
preference for code indents using tabs where possible.

Signed-off-by: Brandon Paupore <brandon.paupore@wdc.com>

ocp: support OCP DSSD Async Event Config feature

Adds two new OCP plugin functions for getting and setting this feature,
based on the 2.5 OCP spec.

Signed-off-by: Brandon Paupore <brandon.paupore@wdc.com>

nvme: Add support for get-reg and set-reg commands

The get-reg command is to output register(s), it's possible to show
several registers when register name are used as command line option.

The set-reg command is to write nvme register(s). Again, the command is
able to set several registers when the register name are used.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-json: Add register print functions

To prepare the state for the get register feature.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: Add register print functions

To prepare the state for the get register feature.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print: Add register print functions

To prepare the state for the get register feature.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme: Refactor mmap_registers to add writable parameter

To prepare the stage for the set register feature.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print: Change PMRSZU string function as unit string

To use the function by other registers print functions.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print: Refactor CMBLOC and PMRSTS registers function

Move the mmio_read up the call chain.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

logging: fix error handling for ioctl passthru wrappers

libnvme fixed the error handling for the passthru commands. Update the
nvme-cli version accordingly.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

build: bump libnvme wrap

Include error handling changes for the passthru commands.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nbft: Fix scandir() error check

read_nbft_files() returns negative errno values in case
of a failure.

Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>

nbft: Refactor connection attempt out

With planned NBFT discovery work, a common connection
function would be handy. Besides, the DHCP fallback mostly
duplicated the connection code anyway, let's refactor that.

Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>

nvme-print: Use EOM optional data present macros changed

Also use the EOM optional data present reserved field macro added.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print: Print cntlid number for controller

Print the controller identifier number (cntlid) for the nvme controller.
The cntlid is printed if nvme list verbose is specified.

Signed-off-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Link: https://lore.kernel.org/r/20240405121057.297071-1-nilay@linux.ibm.com
Signed-off-by: Daniel Wagner <dwagner@suse.de>

build: bump libnvme wrap

Get nvme_ctrl_get_cntlid changes.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme: check-tls-key fails despite having hostnqn file

Nvme check-tls-key requires both hostnqn and subsysnqn to be
specified along with the keydata. Ideally, it should be able to
fetch the hostnqn string from the hostnqn file if no hostnqn
option is passed, irrespective of whether the insert option is
specified or not. Also passing the subsysnqn option is required
not just for identity v1, but for v0 as well, irrespective of
the insert option. Fix accordingly.

Signed-off-by: Martin George <marting@netapp.com>

nvme: print retained key for gen-tls-key's -i option too

Printing the retained or encoded key is useful not only when
generating the tls key but also while inserting it into the
keyring as well.

Signed-off-by: Martin George <marting@netapp.com>

fabrics : allow host to create duplicate connections to target

Apparently the nvmf connect code doesn't allow creating
duplicate connections from a host to the same target even
though user specifies option "--duplicate-connect" or "-D".
This patch help fix this issue.

Fixes: 07d6b911e081 ("fabrics: Do not attempt to reconnect to already connected ctrls")
Reported-by: Amit Engel <Amit.Engel@Dell.com>
Signed-off-by: Nilay Shroff <nilay@linux.ibm.com>
Link: https://lore.kernel.org/r/20240403133438.161916-1-nilay@linux.ibm.com
Signed-off-by: Daniel Wagner <dwagner@suse.de>

fabrics: Fix fast_io_fail_tmo option

The option was partially removed by a refactoring commit.
Return it back, because it is still supported by the kernel
driver.

Fixes: 18de3a6d61a7 ("Convert to libnvme")
Signed-off-by: Israel Rukshin <israelr@nvidia.com>
Reviewed-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Link: https://lore.kernel.org/r/1712060292-11054-1-git-send-email-israelr@nvidia.com
Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-print-stdout: use admin queue attributes definitions

Replace hard coded register shift and mask values instead.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: use CSTS definitions

Replace hard coded register value, shift and mask values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: use controller configuration definitions

Replace hard corded register value, shift and mask values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: use NVME_MAJOR and NVME_MINOR definitions

Replace hard coded shift and mask values.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

nvme-print-stdout: use FDP configuration attributes definitions

Replace the hard coded shift and mask values in the print function.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

build(deps): bump sand4rt/ftp-deployer from 1.7 to 1.8

Bumps [sand4rt/ftp-deployer](https://github.com/sand4rt/ftp-deployer) from 1.7 to 1.8.
- [Release notes](https://github.com/sand4rt/ftp-deployer/releases)
- [Commits](https://github.com/sand4rt/ftp-deployer/compare/v1.7...v1.8)

---
updated-dependencies:
- dependency-name: sand4rt/ftp-deployer
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

nvme: add new function 'tls_key'

Add a new function 'tls_key' to import and export keys from the
keyring.

Signed-off-by: Hannes Reinecke <hare@suse.de>

build: bump libnvme wrap

To include missing symbol nvme_scan_tls_keys.

Signed-off-by: Hannes Reinecke <hare@suse.de>

nvme: use library functions for importing and exporting TLS keys

As we now have library functions for importing and exporting TLS
keys we should be using them.

Signed-off-by: Hannes Reinecke <hare@suse.de>

nvme: extend error message for ns scan failures

When the scan operation fails, let's also report the error code in order
to help understand why the operation failed.

Signed-off-by: Daniel Wagner <dwagner@suse.de>

nvme-print-json: Bugfix status json key of sanitize-log

There is a history of combining duplicated json keys before.
Fixes: 1d7c092f9dac ("nvme-print-json: Combine duplicated json key and val string variables")
Fixes: 59a0f125cb08 ("nvme-print-json: Delete static const char string global variables")
But there is a local variable with the same name, so the key and value
are the same.

Reviewed-by: Sangkuk Lee <sangkuk0.lee@samsung.com>
Reported-by: Kiryong Kim <kiryong1.kim@samsung.com>
Signed-off-by: Minsik Jeon <hmi.jeon@samsung.com>
Co-authored-by: Steven Seungcheol Lee <sc108.lee@samsung.com>

sed: update SED password when initalizing

The SED key and kernel keyring were not being updated by the
initialization operation. Add password update after the other
initialization tasks are completed.

Signed-off-by: Greg Joyce <gjoyce@linux.ibm.com>

plugins/memblaze: add performance stats

Add performance stats feature

Signed-off-by: jinhua.huang <jinhua.huang@memblaze.com>

plugins/memblaze: fix code mis-type error

Change "flash_error_media_count" to "inflight_write_io_cmd".

Signed-off-by: jinhua.huang <jinhua.huang@memblaze.com>

plugins/memblaze: code refactor on applying __packed on more structs

applying __packed on more structs

Signed-off-by: jinhua.huang <jinhua.huang@memblaze.com>

plugins/memblaze: code refactor on using cleanup_nvme_dev

using cleanup_nvme_dev to open/close device

Signed-off-by: jinhua.huang <jinhua.huang@memblaze.com>

nvme: Fix to cleanup free raw_secret memory allocated

The memory is allocated by the gen-tls-key and check-tls-key commands.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

plugins/memblaze: code-format on leading indent and align

1. Set leading indent to tab
2. Set proper alignment as code len limitation

Signed-off-by: jinhua.huang <jinhua.huang@memblaze.com>

plugins/wdc: Fix narrow index variable type in for loop to __u32

Fix the comparison of narrow type with wide type in loop condition.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>

plugins/wdc: Whitespace changes

Delete the loop variable since not changed the value.
Also delete the some variables initial values not necessary.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>