Add a new monitor which uses libnvme to collect SMART data from NVMes.
This monitor only implements the .collect_adapter_data operation since
it carries no state across data collections. Nevertheless for future
symmetry and possible future expansion also add an empty struct
nvmemon_ctx.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Some of the attributes printed as JSON by opticsmon are common to all
PCI adapters factor these out into reusable helper functions
zpci_adapter_json_print_start() and zpci_adapter_json_print_end().
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The get_device_node() function for getting the name of an NVMe's
controller device e.g. "nvme0" as implemented in zpcictl is generally
useful and will be needed for NVMe SMART data collection in zpcimon as
well. Move it to libzpci and rename it to zpci_get_nvme_device_node() to
account for the change in namespace.
It would be tempting to pass a struct zpci_dev* instead of a const char*
but as zpcictl uses an incompatible struct zpci_device* this will
require a larger rework. Prioritize minimizing code duplication for now.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Introduce the concept of a generic zPCI device monitor by creating the
zpcimon_ops operations struct turning all optics specific calls into
abstract monitor calls. Handle monitors as a static array of zpcimon_ops
based monitors of which the now split out optics monitor is currently
the only one. All operations are in principle optional though a monitor
which provides an init operation must also provide the corresponding
destroy operation.
Keep the base64 based JSON pair as non optics specific for later re-use
and make it just skip the output in the very unlikely case that encoding
fails. In follow on commits a monitor for NVMe devices collecting SMART
data will be added.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Introduce separate '--format' option to set output format. Use the new
JSON Lines format explicitly in the systemd unit. Since JSON Lines
format only uses printable characters this removes the need for passing
'-a' to systemctl status. At the same time JSON-SEQ is more easily
parsable using 'jq --seq' so keep it the default for easier scripting.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add a function to get the textual name of an enum util_fmt_t value. To
make this robust to changes in the order of elements in the format array
initialize this using named indices.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The opticsmon tool started out as a tool for monitoring the health of
optical modules in directly attached PCI NICs. In the future however it
will also monitor the health of other PCI devices. In particular in
a first step it will monitor the health of directly attached NVMe
devices.
To reflect this broadening of its scope rename opticsmon to zpcimon. Add
zpcimon.service and install it both under the new name and symlinked as
opticsmon.service for backwards compatibility. Since users are expected
to mostly just enable the service this keeps old instructions just
working.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
So far it was possible for the user to specify an interval of 0 which
would result in not re-arming the timer at all thus disabling interval
based data collection. While this would still do optics data collection
on startup and link changes the intention is to collect data at least
every 24 hours. If interval is given as 0 treat this as minimum update
interval of 1 second.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The code accesses OPTICS_QSFP28_LOS_IMPLEMENTED_OFFSET so the size needs
to be one larger for the LOS to be accessible.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Don't just ignore the return of ethtool_nl_connect(). This would also
otherwise lead to invalid socket accesses later. Also make
ethtool_nl_connect() return negative error values instead of
EXIT_FAILURE.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When ethtool_nl_get_optics() fails after ethtool_nl_cb() has allocated
oi->raw this memory would leak because only the struct optics* itself
was freed. Fix this by using optics_free() and relying on free(oi->raw)
being a no-op for NULL pointers in combination with using util_zalloc()
to handle errors both before and after oi->raw was allocated.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Don't ignore the return value of monitor_wait_loop() but instead forward
it to the caller.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The epoll fd (epfd) in monitor_wait_loop() is never closed. Close it in
all return paths. Also there is no explicit check for an error return of
epoll_create1() add that too.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The "tx_fault" field in the JSON output accidentally contained the
rx_los value instead of the tx_fault one. Change it to the correct
value. Note that this does not impact the data relayed to firmware as
that uses the raw module data.
Fixes: c34adb9cab ("opticsmon: Introduce opticsmon tool")
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The audit directory ts-shell configuration and also the output of
getpwuid() are considered tainted when running in Perl taint (-T) mode.
Provide untaint checks with validating the content.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The updateConfiguration() function updates and checks the ts-shell
configuration for each line in the ts-shell configuration file.
Run the checks just once after the configuration is completely read.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Introduce a configuration setting to allow administrators to specify
path to the iucvconn binary as part of the ts-shell configuration.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Perl provides a Taint (-T) mode which tracks data that is obtained
by external means (e.g. arguments, reading file contents, environment
variables, ...)
Introduce a first set of corrections to clear and validate data
obtained from external sources.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Do not obtain the pager program through the PAGER environment
variable. Instead, introduce a configuration setting to specify
the pager program as part of the ts-shell configuration.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When ts-shell creates the authorization for current user, the
user name and group names are interpreted as regex. This might
create additional authorizations which are not permitted for
current user.
Correct this behavior by quoting the user name in the regex.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Check the return code of fcntl() calls to ensure the close-on-exec
setting succeeded. Otherwise, the PTY and server file descriptors
would become available to the forked client process.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
iucvtty expects to receive the TERM environment information right
after a connection has been established. Reading this information
can lead to a connection stall (DoS). Introduce a temporary timeout
and fall back to the default TERM environment.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The iucvtty_read_msg() now receives the entire message header.
Perform message header checks for specific message types where the
payload length is clearly defined. Also this needs to be done prior
starting the message chunk processing because the message datalen
field will be adjusted based on the read chunks.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The inclusion of termio.h leads to the following compilation error:
CC iucvterm/tests/test_functions.o
test_functions.c:16:10: fatal error: termio.h: No such file or directory
16 | #include <termio.h>
| ^~~~~~~~~~
The termio ioctl interface has been obsolete for a very long time and
was replaced by termios. termio.h was still present for compatibility
but was removed with glibc 2.42.
iucvterm uses termios.h correctly in the tool. The test however not only
includes termio.h erroneously, it doesn't even require it. Remove the
inclusion of termio.h.
Reviewed-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
If iucvtty_read_data() receives message data and reads less than
MSG_DATA_OFFSET, out-of-bound reads will happen.
Correct and improve the receiving path by always reading the entire
message header. Also improve the "residual" message data handling by
replacing and processing every message now as chunk.
Acked-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The PTY slave file descriptor is the controlling terminal
allocated for the child process. When all file descriptors
are being closed, a SIGHUP is triggered for the child process.
Keeping the file descriptor in the parent open would prevent
the SIGHUP to be delivered to the child process.
For proper SIGHUP processing, close the file descriptor in
the parent process.
Acked-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Improve handling of TERM environment processing by validating
the received terminal name. If the terminal name is not valid,
a message will be displayed and the default terminal will be used.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Explicitly unmount the root partition on unit stopping to prevent still
mounted boot partition on switch root which results in boot being unable
to be mounted after switch root.
Reviewed-by: Holger Dengler <dengler@linux.ibm.com>
Signed-off-by: Finn Callies <fcallies@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Since this unit cannot have a dependency on the actual needed
dev-disk-by2xdlabel-boot.device unit because it might not exist if the
label is never set but this unit has to run after it when it does exist
wait for udev-settle which creates the symlink if it does exist.
Reviewed-by: Holger Dengler <dengler@linux.ibm.com>
Signed-off-by: Finn Callies <fcallies@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Update osasnmpd to support Object ID Sub-IDs with length up-to and
including 10 digits, in line with SNMP specification. While at it, also
replace sprintf() calls with snprintf(). Together, these changes
contribute to better memory safety.
Suggested-by: Alexandra Winter <wintera@linux.ibm.com>
Reviewed-by: Hidayath Khan <hidayath@linux.ibm.com>
Signed-off-by: Aswin Karuvally <aswin@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
These messages indicate verification failures and are therefore reported
as errors rather than debug output. Unfortunately,
X509StoreContext::init(...) expects the callback to return an OpenSSL
ErrorStack, so it's not possible to propagate these failures through our
own error hierarchy.
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Enforce root CA organization name requirements during certificate chain
validation to prevent accepting certificates from unexpected root CAs.
Add RootCaVerification enum with RootCaOrganizationPinning and
SkipPinning variants. Extend verify_chain() to validate root CA
organization. Default to pinning "DigiCert" for standard trust chains,
skip pinning when custom root CA is provided.
Assisted-by: IBM Bob:1.0.6
Signed-off-by: Marc Hartmayer <marc@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The command line length value is not properly checked and limited to the
documented 1024 bytes, even though a MAX_CMD_LEN is already defined but
not used.
With this, an overflow of the static char mon_record[] buffer is possible
in read_cmdline(), corrupting adjacent .bss data. The data is sanitized to
printable ASCII bytes, but in theory a crash of the mon_procd daemon could
be possible.
Fix it by adding a check and truncation similar to the other restricted
fields like e.g. ruser_len or cmd_len.
Reviewed-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The yes_no() function was leaking memory when returning
early from the loop, as the 'answer' buffer allocated by
getline() was not freed before the return statements.
Restructure the function to use a single exit point, ensuring
free(answer) is always called before returning.
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Volkan Unal <vunal@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
sort_bls_entries() already checks for the keyword "title"
duplication. Add also checks for keywords "linux", "initrd", and
"options" duplication (Rule #5).
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Check decimal prefixes, representing site values in on-disk
environment block. Reject any values different from {0, ..., 9}
as invalid ones.
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Before extracting a file trailer, check that the file is big enough;
Check that signature length stored in the extracted trailer doesn't
exceed the size of the file without the trailer. Treat the case of
the failed checks as unsigned file.
Without the checks, memcmp() may read the area before the allocated
@buffer, dumping heap into bootmap or crashing.
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When parsing environment block, check each its line for the
compliance with the 'foo=bar\n' pattern. In case of missing '=',
or '\n' abort the parsing procedure with a warning message.
Without this check, stage3 may write a zero byte to the absolute
address 0 (robustness defect)
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
