Introduce shift masks for multiple ioctl functions #32
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Daniel Wagner <[email protected]>
Abstract the file descriptor handle to a more abstract type so it's possible to use different 'backends' to send/receive commands, e.g. via MI instead direct. The name 'link' was choosen because this name is not in use and thus we avoid overloading existing names, such ctrl. The specification is using 'link' to describe the connection on a PCI bus which is kind of matching how the handle is used in the library. It is the means to talk to a controller. Signed-off-by: Daniel Wagner <[email protected]>
Since uring not supported for link type mi. Signed-off-by: Tokunori Ikegami <[email protected]>
There is no user left for the file descriptor in the args structs, thus remove it. Signed-off-by: Daniel Wagner <[email protected]>
When the user provides their own implemention of the passtrhu function, it is necessary to use the file descriptor for the ioctl call. Thus allow to retrieve the file descriptor from the link handle. Signed-off-by: Daniel Wagner <[email protected]>
nvme-cli wants to retrieve the name of the device, add a getter for this. Signed-off-by: Daniel Wagner <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
The MI code has the aligment and lenght check which the the direkt APIs are missing. Without these checks the download tests for MI will fail after using the generic APIs. Signed-off-by: Daniel Wagner <[email protected]>
With the introduction of the link handle, it's possible to use the generic nvme API. The transport selection happens on lower levels depending on the link object type. Note, the tests rely on using the ioctl interface and not the io_uring. Signed-off-by: Daniel Wagner <[email protected]>
The v1 design used the POSIX way to return errors, return -1 with errno set. Replace this design by returning directly the error code. That means non trivial function have to return any data via a pointer so the function is able to return proper error codes. Signed-off-by: Daniel Wagner <[email protected]>
MI doesn't access the complete identify data struct and thus only transfers a port of it. Add a generic version of this function. Signed-off-by: Daniel Wagner <[email protected]>
Use the generic nvme_identify{_partial} interface and drop
the MI specific version.
Signed-off-by: Daniel Wagner <[email protected]>
Since the warning message output by kernel-doc-check. Signed-off-by: Tokunori Ikegami <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
The next change will add a static inline implementation of nvme_get_log which uses nvme_get_log_page. Thus move the nvme_get_log_page function up so it can be used directly. Signed-off-by: Daniel Wagner <[email protected]>
Refactor nvme_get_log and nvme_get_log_page to use use struct nvme_passthru_cmd directly and drop the struct nvme_get_log_args. The struct args have are ABI thus it's hard to change it after a release. The nvme_passthru_cmd buffer is a generic buffer and it's possible to use the setters directly on the buffer in an inline function. Besides a reduced API to maintain it also generates better code and avoids one call less when issue a command. [ikegami.t: update nvme_uring_cmd_amdin_passthru_async] Signed-off-by: Tokunori Ikegami <[email protected]> Signed-off-by: Daniel Wagner <[email protected]>
The compiler warns with
../test/ioctl/logs.c: In function ‘test_get_log_reachability_groups’:
../test/ioctl/logs.c:725:28: warning: enum constant in boolean context [-Wint-in-bool-context]
725 | (!!TEST_LSP << 8) | (!!TEST_RAE << 15) |
| ^
../test/ioctl/logs.c:733:70: warning: enum constant in boolean context [-Wint-in-bool-context]
733 | err = nvme_get_log_reachability_groups(test_link, TEST_RAE, !!TEST_LSP,
| ^
Update the tests so the types are correctly handled. It also addresses
the problem with the wrong lsp value testing.
Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_format_nvm_args. [dennis: merge lbafl and lbafu into one argument] Signed-off-by: Dennis Maisenbacher <[email protected]> Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_ns_mgmt_args. Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_ns_attach_args. Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_fw_download_args. Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_fw_commit_args. Signed-off-by: Daniel Wagner <[email protected]>
This was referenced Sep 15, 2025
Drop struct nvme_security_send_args. [dennis: merge spsp0 and spsp1 into one argument] Signed-off-by: Dennis Maisenbacher <[email protected]> Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_set_features_args. Signed-off-by: Dennis Maisenbacher <[email protected]> [wagi: remove libnvme.map entries reorder fid and sv argument] Signed-off-by: Daniel Wagner <[email protected]> nvme_set_features: fix argument order Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_set_property_args. Signed-off-by: Tokunori Ikegami <[email protected]> [wagi: remove libnvme.map entry] Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_lockdown_args. Signed-off-by: Tokunori Ikegami <[email protected]> [wagi: remove libnvme.map entry] Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_capacity_mgmt_args. Signed-off-by: Tokunori Ikegami <[email protected]> [wagi: remove libnvme.map entry reordered/renamed arguments] Signed-off-by: Daniel Wagner <[email protected]>
Drop struct nvme_get_features_args. Signed-off-by: Dennis Maisenbacher <[email protected]> [wagi: reordered/renamed arguments] Signed-off-by: Daniel Wagner <[email protected]>
There is no user left of this type in ioctl.c. Remove it. Signed-off-by: Daniel Wagner <[email protected]>
This is to maintain nvme_get_log_partial easily. Signed-off-by: Tokunori Ikegami <[email protected]>
Since the following warnings output. $ ./scripts/kernel-doc-check test/zns.c test/zns.c:3: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst test/zns.c:10: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst Signed-off-by: Tokunori Ikegami <[email protected]>
examples/mi-mctp.c:3: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst Signed-off-by: Tokunori Ikegami <[email protected]>
Since checked by kernel-doc-check then the warning message output. Signed-off-by: Tokunori Ikegami <[email protected]>
To fix the warning message below. test/mi.c:3: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst Signed-off-by: Tokunori Ikegami <[email protected]>
igaw
reviewed
Sep 15, 2025
|
__u8 has not endianess problems, so no need to handle it special for the different architectures. |
Replace hardcoded MASK and SHIFT values with enum entries for nvme_capacity_mgmt. Signed-off-by: Dennis Maisenbacher <[email protected]>
Replace hardcoded MASK and SHIFT values with enum entries for nvme_lockdown. Signed-off-by: Dennis Maisenbacher <[email protected]>
Replace hardcoded MASK and SHIFT values with enum entries for nvme_io_mgmt_recv. Signed-off-by: Dennis Maisenbacher <[email protected]>
Replace hardcoded MASK and SHIFT values with enum entries for nvme_resv_*. Signed-off-by: Dennis Maisenbacher <[email protected]>
MaisenbacherD
force-pushed
the
libnvme2-dennis-shift-masks
branch
from
4e054b2 to
fea32c8
Compare
Author
|
I reverted the payload size from my initial proposal back to the correct size. |
igaw
force-pushed
the
libnvme2
branch
5 times, most recently
from
e8528de to
e10ee25
Compare
|
patches merged into libnvme2. thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes:
#27
#25
#21
#20
The nvme_resv_* functions were broken, and I need to backport the fixes to upstream after we reviewed these patches.
@igaw the nvme_resv_* payloads are 8bit chunks. This means we don't have to worry about the endianness - correct?
Users of this API changes must be updated.