JSON Lines text format (JSONL) [1] is a line-delimited JSON format where
objects are separated by the new line character (\n, LF) as opposed to
the JSON Sequence text format (json-seq) where JSON text is encapsulated
in an ASCII Record Separator (0x1E, RS) and ASCII Line Feed character
(0x0A, LF).
Whilst JSONL is also used for data streaming, this simpler format is
better suited for logging and works also well with traditional
line-oriented Unix tooling (e.g. grep or sed).
Add this format to util_fmt so that users have more choice and control
over formats that are required for their usecases.
Add helper functions that let the user determine whether a given format
type is JSON in general or a JSON streaming format (such as json-seq or
jsonl).
For better readability and more clarity use the helper function
util_fmt_is_json_stream() where the same decision is made for both
JSON streaming formats FMT_JSONSEQ and FMT_JSONL.
[1] https://jsonlines.org/
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Sending signal SIGINT/SIGTERM to a running pai process causes immediate
termination of that running process. This usually interrupts a
select() system call waiting for more input to read from the installed
events and its mapped memory buffers. As there is no signal handler
installed, a SIGINT or SIGTERM signal simply terminates the process,
sometimes leaving incomplete recorded output file paicryto.XXX
(where XXX is the CPU number).
Install a signal handler to intercept signal SIGINT or SIGTERM and run
one more data collection loop to read out pending data and close all
recording output files properly.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Sumanth Korikkar <sumanthk@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>
If pvsecret {add, list, retrieve} is executed with options on a non-s390
system the user gets misleading error messages as the options are not
defined.
> pvsecret add -i secret.bin
error: unexpected argument '-i' found
This may lead the user to think wrong arguments where chosen, which is
not entirely true as they are valid on s390. The more helpful error
message would be
error: Command only available on s390x
Which is already the case if no arguments are given.
Solve this by allowing non-s390 systems to parse the options:
> pvsecret add -i secret.bin
error: Command only available on s390x
Reported-by: Carlo Della Giusta <carlo.dellagiusta@suse.com>
Fixes: dd82c26f87 ("rust: Add tool to manage UV-secrets")
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>
If pvattest perform is executed with options on a non-s390 system the
user gets misleading error messages as the options are not defined.
> pvattest perform -i attestation_request.bin -o attresp.bin
error: unexpected argument '-i' found
This may lead the user to think wrong arguments where chosen, which is
not entirely true as they are valid on s390. The more helpful error
message would be
error: Command only available on s390x
Which is already the case if no arguments are given.
Solve this by allowing non-s390 systems to parse the options:
> pvattest perform -i attestation_request.bin -o attresp.bin
error: Command only available on s390x
While at it ignore some unused code warnings in the exchange format code
that appear on non-s390 systems as not all code is used.
Reported-by: Carlo Della Giusta <carlo.dellagiusta@suse.com>
Fixes: 16610a211f ("rust: pvattest-Rust")
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>
Force the autocompletion script generator to always write to a new
empty file, thus preventing the potential corruption of script contents.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Szabina Korbai <szkorbai@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With OpenSSL 4.0.0 function SSL_set1_host() is deprecated and should be
replaced by SSL_set1_ipaddr() and SSL_set1_dnsname().
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With OpenSSL 4.0.0 function X509_NAME_get_entry() returns a const pointer.
Make the local variable also const to avoid compile warnings like:
warning: assignment discards ‘const’ qualifier from pointer target type
[-Wdiscarded-qualifiers]
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With OpenSSL 4.0.0 function X509_NAME_get_entry() returns a const pointer.
Make the local variable also const to avoid compile warnings like:
warning: assignment discards ‘const’ qualifier from pointer target type
[-Wdiscarded-qualifiers]
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With OpenSSL 4.0.0 function X509_cmp_current_time() is deprecated and
should be replaced by X509_check_certificate_times().
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Generating and then including C source files from another C file can
lead to unexpected compilation errors in certain environments.
Switch the usage text generation from %_usage.c to %_usage.h. The
generated header provides the usage_text definition directly, and
chzdev.c and lszdev.c include the corresponding *_usage.h instead.
Update depfile prerequisites and the clean target to match the new
generated artifacts.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Polensky <japo@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Consolidate the duplicated sed command sequences used to generate
*_usage.c files into a shared CSTR_SED definition and a single pattern
rule:
%_usage.c: %_usage.txt
This removes the copy/pasted rules for chzdev_usage.c and lszdev_usage.c
and keeps the Makefile easier to maintain.
Also drop the explicit chzdev.o/lszdev.o prerequisites on the generated
sources as dependencies are already tracked via the .*.o.d depfiles.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Polensky <japo@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Remove the uname()/sscanf()-based kernel version gate (Linux < 2.6),
which is long obsolete and does not belong in user-space argument
validation.
While touching the code, replace the repeated warnx() + exit(1) pattern
with errx(EXIT_FAILURE, ...) in the option sanity checks to reduce
boilerplate and keep error paths consistent.
Behaviour is unchanged for supported environments; the version gate is
dropped because it is obsolete.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Polensky <japo@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
- Use __packed instead of __attribute__ ((packed))
- Rename EBCtoASC to ebc_to_asc following naming conventions
- Consolidate multi-line error message into single line
No functional changes.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Polensky <japo@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Normalize the man page source to UTF-8/US-ASCII and remove a mojibake
artifact that could not be represented cleanly.
Documentation only, no functional changes.
Reviewed-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Polensky <japo@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Future updates might change the outcome of the resulting scripts and the
comments describing the script output would need to be updated every
time as well. It's not worth the effort.
Remove the comments that list script examples from the functions
generate_bash_autocomp() and generate_zsh_autocomp().
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
For Bash, when compspecs are found the generated script is returned as
the full set of possible completions and default completions are
disabled [1]. This leads to the behaviour that command line arguments
are not completed, only options defined by the script.
Zsh has the same issue.
Fix the issue by always adding the bash defaults to the generated
script. For zsh the corresponding file completion is always added to the
end of the argument list and the -A "*" option is added to allow
completion after positional arguments.
[1] https://www.gnu.org/software/bash/manual/html_node/Programmable-Completion.html#Programmable-Completion-1
Fixes: 638cbbe332 ("libutil: Implement zsh and bash autocompletion")
Reported-by: Stefan Haberland <sth@linux.ibm.com>
Reviewed-by: Szabina Korbai <szkorbai@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Apply coding style changes by reducing line breaks for a more compactly
generated script and add double quotes to fix ShellCheck findings.
Furthermore, remove the unused variable previous_word and use mapfile to
read the output of compgen into the COMPREPLY array to avoid unwanted
splitting and glob expansion.
Reviewed-by: Szabina Korbai <szkorbai@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When installing IPL on mirrored targets, zipl builds multiple program
tables (one such table per each mirror). If the option "--add-files"
was specified, then zipl uses the in-bootmap files, that was written
at the prevoius iteration (for the mirror ID #0) to build program
tables for mirrors with ID #1 (and larger). The in-bootmap files
already don't contain trailers. Despite this, the building process
cuts off the tail of trailer size, which results in corrupted boot
data.
This bug may result in crashing the kernel when booting from mirrors
with ID #1 (and larger) and manifests only if the boot components are
signed and the option "--add-files" is specified for the installation
session.
Don't count the trailer, when building program tables using in-bootmap
files written at the previous iteration.
Fixes: 431e4542ca ("zipl/src: Reuse data of file components in bootmap")
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
New trace tags "sysuas*" added in kernel to trace zfcp sysfs unit add
store events.
Print the trace events under HBA with new record ID 6. The fields under
this tag are WWPN, FCP LUN and return value of the device having issues.
Example zfcpdbf output for unit add store events:
Timestamp : 2026-02-02-10:50:02:474983
Area : HBA
Subarea : 00
Level : 4
Exception : -
CPU ID : 0003
Caller : 0x000001d89a86bd0e
Record ID : 6
Tag : sysuas2
Description : sysuas2 HBA, Sysfs unit add store failure, unit add failed
Request ID : 0x00000000ffffffff
Request status : 0xffffffff
FSF cmnd : 0xffffffff
FSF sequence no: 0xffffffff
WWPN : 0x500507680b26c449
LUN : 0x01d0000000000000
Return Value : 0xfffffff4
Timestamp : 2026-02-02-10:50:02:475037
Area : HBA
Subarea : 00
Level : 4
Exception : -
CPU ID : 0003
Caller : 0x000001d89a86bd0e
Record ID : 6
Tag : sysuas2
Description : sysuas2 HBA, Sysfs unit add store failure, unit add failed
Request ID : 0x00000000ffffffff
Request status : 0xffffffff
FSF cmnd : 0xffffffff
FSF sequence no: 0xffffffff
WWPN : 0x500507680b26c449
LUN : 0x01d1000000000000
Return Value : 0xfffffff4
Signed-off-by: Chinmaya Kajagar <chinmayk@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The stage3 linker script places .sb.trailer so that it must end at
COMMAND_LINE_EXTRA. The current script derives the start address from
SIZEOF(.sb.trailer) before the section is emitted:
. = COMMAND_LINE_EXTRA - SIZEOF(.sb.trailer);
With binutils older than 2.39 before commit 648f6099d4dc ("-z relro
relaxation and ld script SIZEOF") this can result in .sb.trailer being
placed at COMMAND_LINE_EXTRA instead, moving the trailer into the
following area and breaking the expected layout.
The trailer has a fixed size, so use an explicit constant for the
placement calculation and keep the ASSERT to verify the final section
size. This makes the placement deterministic again.
Fixes: a1126352ec ("zipl/boot: Improve linker scripts")
Reviewed-by: Marc Hartmayer <marc@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Improvements for lszcrypt and chzcrypt:
* lszcrypt -b and lszcrypt -d now check for default domain
available and gives a warning if the current default domain
is not in the usage_domain_mask of the AP bus.
* lszcrypt without any further device also checks for the
default domain and emits a warning string if the default
domain is not available.
* chzcrypt --default-domain emits a warning if the newly
set default domain is not enabled in the usage_domain_mask
of the AP bus.
Suggested-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
Reviewed-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The blkiomon warnings "bad trace magic 0" and "blkiomon: bad trace" are
benign startup artifacts caused by early pipeline reads before blktrace
produces a complete record. These messages do not affect processing and
valid reports are still generated. Filter them out so only real errors
trigger failures.
Signed-off-by: Ajaykumar Rajappa <ajaykr@linux.ibm.com>
Reviewed-by: Nihar Panda <niharp@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Hyptop should be able to accept abbreviations of the long command line
options as getopt() is able to match them. Ex;
$ hyptop --batch-mode
$ hyptop --batch
$ hyptop --bat
From getopt(3) man page:
"""
Long option names may be abbreviated if the abbreviation is unique
or is an exact match for some defined option.
"""
After the introduction of commit c5695e43c4 ("hyptop/opts: Replace long
option formats for consistency") long command line options for hyptop
received additional definitions to support dash separated option formats.
Unfortunately these definitions were defined as new and unique options
and caused an ambiguity for getopt() when abbreviations matched both
definitions. Ex;
$ hyptop --batch
hyptop: option '--batch' is ambiguous;
possibilities: '--batch-mode' '--batch_mode'
Map both long option formats to the same short option to fix the
issue and restore the functionality.
Fixes: c5695e43c4 ("hyptop/opts: Replace long option formats for consistency")
Reported-by: Gorkem Kilinc <kilinc@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Mete Durlu <meted@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Hyptop's "--sort_field" command line option has always been documented
as "--sort", while the code _only_ explicitly has "--sort_field".
Specifying the shorter "--sort" happened to work due to an unnoticed
getopt() behavior.
From getopt(3) man page:
"""
Long option names may be abbreviated if the abbreviation is unique
or is an exact match for some defined option.
"""
With the addition of "--sort-field" as another unique identifier via
commit c5695e43c4 ("hyptop/opts: Replace long option formats for
consistency") "--sort" is no longer unique. getopt() won't be able to
use that as an abbreviation, since there is ambiguity between
"--sort_field" and new "--sort-field" as they are defined as separate
options.
Replace "--sort-field" and "--sort_field" with plain "--sort" to adhere
to the documented hyptop command line argument specification and resolve
the broken behavior.
Fixes: c5695e43c4 ("hyptop/opts: Replace long option formats for consistency")
Reported-by: Gorkem Kilinc <kilinc@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Mete Durlu <meted@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
An empty device list is not an error case, if `lszcrypt` is called
without a specific device list or device filter. Return with rc == 0 in
such cases.
Remove the message about the empty device list on stderr.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Reviewed-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Kernel support for LCS devices was removed with commit 6cccb3bb0561
("s390/net: Remove LCS driver") in kernel v6.15. Remove the associated
lsznet support for the lcs device type.
Reviewed-by: Aswin Karuvally <aswin@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
znetcontrolunits provided two arrays and a search function that were
only used by lsznet. Since lsznet has it's own implementation of this
function now and the CU array was a duplicate of CU_TCPIP anyway, remove
znetcontrolunits as it serves no purpose.
Reviewed-by: Aswin Karuvally <aswin@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
lsznet sources znetcontrolunits to use search_cu() and set the variable
cu_idx. lsznet's own function search_cu_tcpip() is doing the same thing
as search_cu() without setting cu_idx.
Declare and move cu_idx to the global variable CU_IDX and consolidate
the functions by letting search_cu_tcpip() set CU_IDX. Call
search_cu_tcpip() instead of search_cu() and replace cu_idx with CU_IDX
accordingly. search_cu() is removed and the CU_DEVDRV array is moved to
lsznet.
Reviewed-by: Aswin Karuvally <aswin@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With SECURE_BOOT_DISABLED always set in add_dump_program() no signature
entries for a dump kernel are written. This might lead to security
violation error from the IPL Loader on the systems with secure boot support
on the attempt to boot the dump kernel:
MLOLOA6269321F A security violation error was encountered when loading from device <device>
MLOLOA62693212 Audit: No signed components found for program 0 loaded from device <device>
Propagate '--secure' zipl option to add_dump_program() in order to sign
relevant components upon dump program installation and thus support
secure boot execution for List-directed dump kernels.
Signed-off-by: Mikhail Zaslonko <zaslonko@linux.ibm.com>
Acked-by: Eduard Shishkin <edward6@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Reviewed-by: Alexander Egorenkov <egorenar@linux.ibm.com>
Tested-by: Alexander Egorenkov <egorenar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Kernel support for LCS devices was removed with commit 6cccb3bb0561
("s390/net: Remove LCS driver") in kernel v6.15. Remove the associated
zdev support for the lcs device type.
Signed-off-by: Vineeth Vijayan <vneethv@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The Virtual Tape Server (VTS) only supports 3490 tape devices. Remove
all other older device types from the man pages.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Data compaction is only supported by 3490 tape devices. For Virtual Tape
Server (VTS) this is the only supported device. Reading device
characteristics and checking the type is unnecessary.
Remove the corresponding code and simply enable data compaction.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Load Display (LDD) X'9F' is still accepted by the Virtual Tape Server
(VTS) but does not perform any action. Remove code that still uses this
command in tape2dump.c.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
There is only one supported tape device type left. A filter command line
option doesn't make any sense anymore.
Remove the functionality and documentation of the --type option.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The device models 3480 and 3590 are no longer supported by the tape
device driver. Remove them from the device list
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The tape390_display tool was used to send text to the display of tape
storage server. The corresponding command is still accepted by the
Virtual Tape Server (VTS) but no action is performed. The tool is
useless, remove it.
The tape390_crypt tool was used to manage encryption for tape devices.
However, only 3590/3592 models did support encryption. Support for these
models is removed from the Kernel as there is no support for these tape
models in general anymore. The tool is not required anymore, remove it.
Note: VTS uses encryption transparantly for its virtualized 3490 models.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
