GitLab

This is not a runtime tool, it's for development. And as libxenomai-dev
has no dependency on xenomai-runtime (as there should be none), it is
missed in build-only installations.
Reported-by: Florian Bezdeka <florian.bezdeka@siemens.com>
Fixes: 6e77fe2b ("debian: drop xeno-config from -dev package")
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

The 5.4-arm64 build[1] needs a cache size greater than 400M.

[1]: https://gitlab.com/Xenomai/xenomai-hacker-space/-/jobs/1171197811Signed-off-by: Quirin Gylstorff <quirin.gylstorff@siemens.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

We better exclude this code path due to unknown impact on RT.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Provided by linux/crc32.h as well, now causing build errors. But it was
unused anyway.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Follows upstream commit c4cb99185b4c.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

This feature should be available directly from Dovetail.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

We don't need this with Dovetail, there is no timestamp arithmetics to
perform in userland whatsoever: we get timestamps from the regular
vDSO-based clock_gettime() call.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
[Jan: adjust comments]
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

dump_stack() also prints out the system information, which is always
good to have.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
[Jan: converted two more cases in analogy, indented consistently]
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Since kernel v5.8, __vmalloc() does not take protection bits as
PROT_KERNEL is now wired in. Therefore we cannot disable the cache for
the UMM segment via the allocation call directly anymore.

This said, we don't support any CPU architecture exhibiting cache
aliasing braindamage anymore either (was armv4/v5), so let's convert
to the new __vmalloc() call format without bothering for cache
settings.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

This macro may unroll to instantiating 'date' twice. If that is a call
to some get-time function, as via xnstat_exectime_switch, we needlessly
do that twice.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

In contrast to real RTDM devices, message queues have no fallback path
to Linux in userspace. Thus, EADV is neither needed nor detected by
libcobalt. Avoid leaking it from RTDM to the mq syscall interface.

Fixes: 2db562ad ("cobalt: switch hand over status to -EADV for non-RTDM fd")
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Make the package set arch-specific to save some bytes and seconds.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

This patch includes the driver that supports the CANFD interfaces of
the PCAN-PCIe FD family. This driver is largely inspired by the peak_pciefd
driver included in the kernel since version 4.12. Except for the
differences related to the RTDM model, this driver differs from the
socket-CAN driver by the following points:

. CAN 2.0 a/b support only (CAN-FD not available in rt-socket-can)
. all interrupts are required (or released) when loading (or unloading)
  the module, not when activating (or deactivating) the CAN interface as in
  the socket-CAN model. Only hardware interruptions are enabled (or
  disabled) when the rtcan interface is set up (or down).
. This way of operating allows the usage of the MSI mode: if CONFIG_PCI_MSI
  is defined while CONFIG_XENO_OPT_SHIRQ is not, then the driver activates
  the MSI mode by default. The legacy INTA mode is used otherwise. In case
  CONFIG_PCI_MSI is defined, the parameter module "usemsi" allows to change
  this default behaviour: usemsi=0 selects INTA mode while usemsi=1 selects
  MSI mode.
. the Tx flow is managed more easily (no echo management in rt-socket-can)
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Tested-by: Andy Tannenbaum <trb@bioniklabs.com>
[Jan: added to CI]
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

The former is not a target for latest Xenomai anymore, and the latter is
unmaintained by now.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

As internal interfaces are gradually being made y2038-safe, the
deprecated timeval type is now available as __kernel_old_timeval in
the mainline kernel for the sake of clarity. Switch to this type until
Cobalt is y2038-safe as well.

A wrapper is provided to build on older kernels not defining
__kernel_old_timeval yet.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

In the same move, fix places which depend on such inclusion but fail
to pull the related headers directly.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

The Kconfig parser became stricter, help stanzas must be introduced by
the naked "help" keyword, period. Let's drop all decorations.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

In order to intimately connect Cobalt to the kernel, Dovetail allows
us to extend the latter with data and procedures we need:

- we can embed our own context information into a set of critical
  kernel data structures. This information should be defined as a set
  of core-specific types, such as struct oob_thread_state which is
  going to be part of struct thread_info.

- we can define preparation and finalization handlers for out-of-band
  IRQ handling, which Dovetail should invoke right after entering the
  outer interrupt frame, then right before leaving it respectively.

Add the couple of interface headers we need to connect those elements
to the kernel.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Dovetail interposes on the in-band syscall entry, passing the request
on to Cobalt via different handlers depending on the calling context:

- via handle_oob_syscall() when the current thread is running
  out-of-band, and the request is not an in-band (i.e. regular Linux)
  system call, therefore shall be handled by Xenomai. This is the fast
  path.

- via handle_pipelined_syscall() in all other cases, e.g. when current
  is running in-band, and/or the system call shall be handled by the
  in-band kernel eventually (useful for relaxing the current thread in
  the process if out-of-band on entry).
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Since the low-level tick management code is pipeline specific, we may
flatten the call stack by using the underlying pipeline API directly.

At this chance, fix a uniprocessor build issue with IPIs.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
[Jan: fix non-SMP build issues on ppc32]
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

smp fixup

This is similar to upstream commit 9e6c3b63399dd ("e1000e: fix compiler
warnings" by David Ertman). Fix compile error:

  ERROR: "__bad_udelay" [drivers/xenomai/net/drivers/e1000e/rt_e1000e.ko] undefined!
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

A missing volatile caused some compiler versions (e.g. gcc-8) to skip
updating the variable inside the loop. At this chance, drop the
pointless and misleading usage of atomic_t for this variable.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Internal header, broken for mercury and not used anymore.
Reported-by: Florian Bezdeka <florian.bezdeka@siemens.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

As optimization can shrink __do_work to just two instruction, the
calibration loop needs to run longer.

At this chance, drop the pthread_kill to stop the thread and rely in
both cases on the throttle mechanism.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

More recent, and now also the one carrying a required build fix.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
[Jan: style tweak in detach_current]
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>