Allow the creation of SE images using headers with hybrid (=quantum safe) keys.
This results in using the headers in version 2 (0x200).
Co-developed-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Timo Keller <tkeller@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Refactor test code and increase coverage. The new test code tests for
the correct error kind instead of simply for an error.
Assisted-by: IBM Bob:1.0.6
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Replace version-specific flag enums (PcfV1, ScfV1) with a unified
SeHdrFlag enum and generic ControlFlagsModel. This refactoring provides:
- Single SeHdrFlag enum for all control flags across SE header versions
- Generic SeHdrControlFlagsModel with version-specific configurations
- EffectiveControlFlags type replacing PlaintextControlFlagsV1/SecretControlFlagsV1
- FlagsOverride mechanism for customizing flag configurations
- Enhanced FromStr implementation for flag parsing
- Comprehensive test coverage for flag operations and conversions
The new design improves maintainability by eliminating code duplication
and provides a more flexible API for working with SE header control
flags.
Assisted-by: IBM Bob:1.0.4
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Add a generic flags framework for managing (Secure Execution) header
control flags. This implementation provides:
- Clap compatibility for auto-completion and proper help messages
- ControlFlagTrait for defining flag types with bit positions and version support
- ControlFlagsModel for managing default flags and validation rules
- FlagsOverride for customizing individual flag states
- EffectiveControlFlags representing the final flag configuration
- UnknownFlags handling for forward compatibility
- Comprehensive error handling via FlagValidationError
- Display implementations for both human-readable and binary formats
- Type-safe conversions between flag representations
The framework supports version-aware flag validation, allowing flags to
be marked as supported only in specific Secure Execution header
versions. It enables flexible flag configuration through overrides while
maintaining type safety and validation.
Assisted-by: IBM Bob:1.0.4
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Add a HostKey enum (currently V1(PKey<Public>)) and introduce a
versioned Keyslot enum (V1(KeyslotV1)). Rename the existing Keyslot type
to KeyslotV1 to prepare for future format extensions.
Update pv, pvattest, pvimg, and pvsecret to use the new enums.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Timo Keller <tkeller@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add more CCK tests to make sure that new changes do not introduce a
regression.
Assisted-by: IBM Bob:1.0.4
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
- Remove useless type conversion in uvdevice.rs
- Replace useless comparison in hostname.rs
- Replace unnecessary unwrap patterns in pvapconfig
- Use sort_by_key instead of sort_by in pvimg example
Command line used to get the findings:
$ clippy --all-features -- --cap-lints=warn
warning: useless conversion to the same type: `u64`
--> pv_core/src/uvdevice.rs:56:28
|
56 | rc = ioctl(raw_fd, cmd.try_into().unwrap(), cb.as_ptr_mut());
| ^^^^^^^^^^^^^^
|
= help: consider removing `.try_into()`
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#useless_conversion
= note: `#[warn(clippy::useless_conversion)]` on by default
warning: this comparison involving the minimum or maximum element for this type contains a case that is always true or always false
--> utils/src/hostname.rs:60:13
|
60 | assert!(isize::try_from(buf_len).unwrap() <= isize::MAX);
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= help: because `isize::MAX` is the maximum value for this type, this comparison is always true
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#absurd_extreme_comparisons
= note: `#[warn(clippy::absurd_extreme_comparisons)]` on by default
warning: `utils` (lib) generated 1 warning
Checking pvebc v0.12.0 (/home/mhartmay/git/s390-tools/rust/pvebc)
warning: consider using `sort_unstable_by_key`
--> pvapconfig/src/ap.rs:177:9
|
177 | self.0.sort_unstable_by(|a, b| b.gen.cmp(&a.gen));
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#unnecessary_sort_by
= note: `#[warn(clippy::unnecessary_sort_by)]` on by default
help: try
|
177 - self.0.sort_unstable_by(|a, b| b.gen.cmp(&a.gen));
177 + self.0.sort_unstable_by_key(|b| std::cmp::Reverse(b.gen));
warning: called `unwrap_err` on `r` after checking its variant with `is_err`
--> pvapconfig/src/main.rs:55:29
|
54 | if $r.is_err() {
| -------------- help: try: `if let Err(<item>) = r`
55 | eprintln!("{}", $r.unwrap_err());
| ^^^^^^^^^^^^^^^
...
87 | on_error_print_and_exit!(r);
| --------------------------- in this macro invocation
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#unnecessary_unwrap
= note: this warning originates in the macro `on_error_print_and_exit` (in Nightly builds, run with -Z macro-backtrace for more info)
warning: this `repeat().take()` can be written more concisely
--> pvimg/src/se_img_comps/bootloader/ipl.rs:95:21
|
95 | let comps = iter::repeat(ipl_pb0_pv_comp::default())
| _____________________^
96 | | .take(num_comp)
| |___________________________^ help: consider using `repeat_n()` instead: `std::iter::repeat_n(ipl_pb0_pv_comp::default(), num_comp)`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#manual_repeat_n
= note: `#[warn(clippy::manual_repeat_n)]` on by default
warning: this `repeat().take()` can be written more concisely
--> pvimg/src/se_img_comps/bootloader/ipl.rs:113:21
|
113 | let comps = iter::repeat(comp).take(num_comp).collect();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: consider using `repeat_n()` instead: `std::iter::repeat_n(comp, num_comp)`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#manual_repeat_n
Reviewed-by: Timo Keller <tkeller@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Bump MSRV to 1.85.0 and upgrade all dependencies using 'cargo upgrade'
from cargo-edit. The bump to 1.85.0 is requiered to get at least openssl
version 0.10.79 including fixes for a couple security issues.
Link: https://crates.io/crates/cargo-edit
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
+ Sort and group the imports
+ Normalize and format comments (100 characters width)
Command used:
$ cargo +nightly fmt --
Acked-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The associated type bound for Iterator::Item was unsatisfied for Self.
Fix this by requiring Sized for IntoEnumIterator.
Fixes: 1d2a89b387 ("pvimg: Improve the readability of Display output for control flags")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add documentation for the JSON structure outputted by 'pvimg info
--format=json ...'.
Created with assistance from IBM Bob AI.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add a new '--print-schema <FORMAT>' option to the 'pvimg info' command
that prints the schema, if available, describing the given output
format. For example, 'pvimg info --print-schema json' prints the JSON
schema for the 'pvimg info' command output.
Acked-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add '--show-secrets' flag to 'pvimg info' to make secret output explicit
and avoid accidental disclosure.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Auto-detect default format for 'pvimg info' command. If stdout is a
terminal, use 'text', otherwise 'json'.
Adapt the tests accordingly.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add human-readable output format to 'pvimg info' command. The format
'text:normal' shows only basic information about the Secure Execution
header, but skips the keys and other binary data; the format 'text:full'
shows everything.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The Display implementation should produce human-readable output. Convert
the flags into a descriptive flag list to improve readability.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Use the s390-tools JSON meta data for the JSON output of 'pvimg info' to
make the JSON output more stable and to provide an stable API.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Improve error messages for errors when reading the SE header protection
key or when the decryption/verification of the SE header has failed.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add two JSON output variants: pretty and minify. The desired variant can
be selected via '--format json:pretty' and '--format json:minify'. Using
'--format json' without a variant defaults to pretty.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add a warning to notify the user when the Secure Execution (SE) header
is present but its integrity and authenticity has not been verified.
This makes the lack of validation explicit and helps avoid unintended
use of untrusted data.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Improve the documentation of flags and secured components.
Created with assistance from IBM Bob AI.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The definition NO_PIE_CFLAGS has only two users and specifies only a
single compiler flag. There is no point having this definition anymore.
Remove the definition and specify -fno-pie directly for the two users.
Reviewed-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Commit c5a91199e3 ("zipl: Always build and link without PIE.")
introduced -fno-pie (for compilation) and -no-pie (for linking) for
zipl. At the time the linker was still used directly before commit
5e46632767 ("zipl: Use the compiler for linking instead of ld")
eventually switched to calling the compiler for the linking step.
During that adaption -static was introduced to the linker flags. -no-pie
was carried over as well. However -static implies -no-pie and it is
therefore not required. For GCC see also man 1 gcc (-static) [1]. Whilst
not explicitly documented, Clang shows the same behaviour. Clang also
complains when -static and -no-pie are specified in the linker step at
the same time with the following warning:
clang: warning: argument unused during compilation: '-no-pie' [-Wunused-command-line-argument]
Since -no-pie is not required, remove it and get rid of the warning.
[1] https://gcc.gnu.org/onlinedocs/gcc/Link-Options.html#index-static
Reviewed-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The 'clang' compiler generates a section with the name '".loader_parms"'
and gcc '.loader_parms' if the double quotes are used. Fix this by
removing the superfluous double quotes, this results in a section with
the name '.loader_parms' for both clang and gcc.
Reported-by: Richie Buturla <richie@linux.ibm.com>
Reviewed-by: Euan Bourke <euan@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add support for '--image-key'. This new option can be used to select the
components encryption key (e.g. kernel, initrd, and kernel command
line). Previously, this was only available as an experimental
option ('--x-comp-key').
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Use 'PVIMG_PKGDATADIR' as it is used by the 'pvimg_pkg_data' macro rule.
The problem only shows up if someone changes 'PVIMG_PKGDATADIR' to
something different than the default.
Fixes: f4cf4ae6eb ("rust: Add a new tool called 'pvimg'")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Move the check_hostkeydoc script to scripts. This eliminates the last file
in the genprotimg directory. Additionally, add a deprecation warning to
that script. Every pv tool can verify the chain itself using the pv
library.
Acked-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Adding a new tool (tamper_pvimg) for helping to test if the expected
RC and RRC values are received for several different reasons when
DIAG 308 subcode does not complete successfully. The reasons can include
issues related to host key hash, SE header, ALD/PDL/TLD mistmatch, etc.
This tool can read the SE image and manipulate one of the SE header
fields so that the SE image will fail to enter secure mode during the
stage3a booting process. The stage3a bootloader should receive UVC
command code id, RC and RRC values corresponding to the SE header
field that was manipulated. These values will be checked for
consistency in the tests for correctness of DIAG 308 subcode 10 RCs
and RRCs.
[hoeppner@linux.ibm.com: shortened commit subject]
Signed-off-by: Gautam Gala <ggala@linux.ibm.com>
Signed-off-by: Steffen Eiden <sediden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Converting several variables to public and exporting some structures
for use in SE header tampering tool.
Signed-off-by: Gautam Gala <ggala@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When DIAG 308 subcode 10 is specified, and the configuration fails to enter
the secure mode, then the stage3a bootloader reads the DIAG response code,
UV RC, RRC and command id from bit positions 48-63, 32-47, 16-31, and
0-15 of the general register R1 + 1.
The bootloader now prints the error codes to help pinpoint the error cause.
For some cases that are reported to occur often the bootloader also prints
the cause when possible or prints hints to help understand the cause.
QEMU patch "DIAG 308: extend subcode 10 to return UVC cmd id, RC and RRC
values upon failure to enter secure mode" (commit-id: )
is needed with this patch.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Gautam Gala <ggala@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Create a symbolic link to the 'pvimg-create' manpage of for 'genprotimg'
during the manpage installation. This helps the users to migrate to
'pvimg create'.
Suggested-by: Jan Höppner <hoeppner@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Describe the breaking changes to genprotimg in the notes section. This
helps the users to adapt to the new behavior.
Fixes: f4cf4ae6eb (rust: Add a new tool called 'pvimg')
Acked-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The CCK will be able to be updated by an ultravisor call. For security
reasons, this feature must be enabled at Secure Execution header
creation time.
Because this makes it possible to dump without having set a CCK
earlier, the requirement to specify a CCK when creating an image with
guest dump support is extended to also allow for CCK update instead.
Change a CLI test for this that was a duplicate anyway.
[seiden@linux.ibm.com: Allow --cck & --enable-cck-update simultaneously]
Reviewed-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jakob Naucke <naucke@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
