Libnvme

Things are calming down, which indicates that we are slowly reaching the 1.0 release.

Arunpandian was busy again adding a lot of documentation and fixing a few small issues in utils.
Then there are bunch of build fixes but nothing really dangerous.

Wendy fixed PEL and I updated 'gen-hostnqn' to hand out the correct UUID. Also ported
a missing patch from the monolithic branch which made sure traddr are case compared
ignoring the case.


Arunpandian J (8):
types: All MI command related structure doc field update
types: Add doc field for admin opcodes, FIDs, LIDs, Identify commands
types: Add doc section for enum fields
types: Add missing doc for ZNS, NVMe OF, NVMe command set
types: status code type value update as per NVMe specification
util: Add status code values in generic status to errno conversion
util: Congregate status code values to return EINVAL
util: Add status code values in order to return errno

Boris Glimcher (4):
pyproject.toml instead of setup.py for new packaging
Add workflow to publish to PyPI and TestPyPI
Workflow: temporary disable publish to PyPI, starting with TestPyPI
CI: rename branch main to master

Daniel Wagner (6):
fabrics: Do not swap bytes for system uuid
tree: Ignore traddr case in nvme_lookup_ctrl()
build: Depend on libssl when using the fallback dependency
linux: Include private.h
linux: Add log.h back
doc: Regenerate all documentation

Klaus Jensen (1):
meson: do not leak internal include directories

Martin Belanger (1):
Move ccan definitions out of public headers.

Wen Xiong (1):
ioctl: Set lsp to action in nvme_get_log_persistent_event

In this release candidate we have a lot of documentation updates. Arunpandian is busy adding missing documentation. Thanks a lot!

Then there were a few small bug fixes and one missing backport from the monolithic branch. A few API changes in this candidate again, but I hope we slowly have these sorted out.

The biggest change is the inclusion of the TP8010 support which is mostly just new code so it should not regress (famous last words, I know).


Akhilesh RN (1):
linux: Update size when telemetry controller initiated data is unavailable

Arunpandian J (8):
types: Added doc description for fields used
ioctl.h: Add description for fields used
linux: Streamline documentation
log: Fix whitespacing in documentation
tree: Add missing documentation
types.h: Add missing documentaion for all the log pages
types.h: doc update for ZNS structures
types: doc field update for NVM commands, Feature Identifiers

Daniel Wagner (5):
log: Allow nvme_msg usage with invalid root pointer
fabrics: add default port number for NVMe/TCP I/O controllers
doc: Do not precomple HTML doc anymore
doc: Regenerate all documentation
build: Use boolean type for docs-build default

Hanumanthu H (1):
add cdw13 for set_feature_args structure

Jeff Lien (1):
linux: Change nvme_get_log_page to use nvme_get_log_args parm

Martin Belanger (8):
python: Add dump_config
tree: Add host symbolic name
types: Add DCTYPE to nvme_id_psd
tree: Add cntrltype and dctype to the controller object
types: Add Discovery Information Managment types
util: Add key-value helpers and OS identification tools
ioctl: Add nvme_dim_send()
fabrics: Add support to register with a DC

Sami Waheed (1):
Update ioctl.h

This release candidate brings mostly cleanups and a bunch of bug fixes. The main change is that
we are now properly creating the precompiled documentation. Again some small API changes but
nothing big this time. It looks like things are calming down.

Please keep testing and check if the API is okay. After the v1.0 release things will be way harder
to change.


Arunpandian J (2):
nvme: Add Supported Capacity Configuration List log page(LID: 0x11)
nvme: Add Enhanced Controller Meta Data(FID: 0x7D)

Daniel Wagner (12):
doc: Remove unused make.bat file
tree: Remove doublicated declaration
types: Fix doc description
doc: Mesonize documentation generation
doc: Fix sphinx setup
doc: Regenerate all documentation
build: Make CI build fail on warnings
build: Add kernel-doc CI build
tree: Remove dead !LIBUUID code
tree: Remove CONFIG_JSONC ifdefs
tree: Fix memleaks in __nvme_free_ns() and nvme_scan_subsystem()
build: make Python memory allocations valgrind friendly

Hannes Reinecke (25):
Remove nvme_reset_topology()
Rename nvme_path_get_subsystem()
tree: document function prototypes
Return error from nvme_read_config()
Ignore errors when opening JSON configuration file
ioctl.h: fixup kernel-doc comments
linux.h: fixup kernel-doc comments
tree.h: fixup kernel-doc comments
types.h: fixup kernel-doc comments
fabrics.c: fixup kernel-doc comments
util.h: fixup kernel-doc comments
Fixup sphinx warnings
Regenerate libnvme.rst
Add documentation section descriptions
Fixup wrong kernel-doc annotations
tree: fixup memory leak in nvme_scan_ctrl()
tree: restart controller lookup
Add scripts to regenerate documentation
tree: loop controllers can have transport addresses
tree: loop address may be NULL
tree: do not set dhchap_key to 'none'
tree: do not print function twice
json: remove wrapper definitions
Make JSON-C mandatory during built
tree: add nvme_dump_tree()

Jeff Lien (1):
linux: Add size and data area parm to nvme_get_telemetry_log

Martin Belanger (1):
fix license info

Steven Seungcheol Lee (6):
ioctl: Modify identify command argument based on spec
types: Add NVM Command Set id-ns structure
ioctl: Add identify ioctl for CNS 09h, 0Ah
nvme: Add nulbaf(Number of Unique Capability LBA Formats) field on nvmd_id_ns
nvme: get log domain id included in Log Specific Identifier
ioctl: nvme_io_args.dspec(Directive Specific) value is 16 bits

Another release candidate for libnvme. We are still moving bits around in the API and hence we need a bit more time to stabilize all things. Don't expect a release next week as Hannes already told me that he has some more cleanup patches in the queue. Anyway, overall I think we are moving into the right direction.

Please give it a test and report the bugs you find. Thanks!


Arunpandian J (1):
Clean up Patch for types.h, ioctl.h

Daniel Wagner (6):
ioctl: Set errno when command args check fails
util: Set errno in error case nvme_get_directive_receive_length
ioctl: Fix struct nvme_get_log_args memberm alignment
ioctl: Sort members their natural size
linux: Rename __nvme_get_log_page()
libvnme: Remove __nvme_msg from export map

Hannes Reinecke (10):
fabrics: Remove nvmf_add_ctrl_opts()
fabrics: Update prototype documentation
fabrics: remove documentation for nvmf_add_ctrl_opts()
Update README.md
Merge hostname2traddr()
Add 'nvme_root_t' argument to nvme_msg()
Move global logging variables into nvme_root_t
Remove stale definition for nvme_log_message
libnvme/tests: fixup create-ctrl-obj.py
Rework nvme_scan_filter()

Busy week, we got a bunch of cleanups and bug fixes. The API has changed slightly due
to the added Media Unit Status log page feature and dropping the disable_sqflow
argument for nvmf_add_ctrl.

Here the shortlog:


Arunpandian J (1):
nvme: Add Media Unit Status log page(LID: 0x10)

Daniel Wagner (3):
linux: Include missing stddef.h
src: Use PRIu64 format specifier for uint64 types
ioctl: Initialize all members in nvme_get_log_args

Hannes Reinecke (5):
Implement nvme_errno_to_string()
fabrics: do not treat ctrl_loss_tmo values of '-1' as invalid
fabrics: add fabrics config option 'tls'
tree: only open controller device node if required
fabrics: kill 'disable_sqflow' argument to nvmf_add_ctrl()

Martin Belanger (1):
Define ccan as a static library

Steven Seungcheol Lee (2):
meson: ssl version checking allowed version only
util: modify flbas rsvd bits to be used

Tomas Bzatek (2):
docs: Fix a typo in README.md
types: TP8013 and TP8014 doc strings updates

This is a pre release for the upcoming v1.0 release.

The API should be fairly stable from now on but some things might still change until the v1.0 release. Also there are still some hickups but nothing serious which prevents broader testing. So please give it a go and report bugs you find.

Thanks!