GitLab

Align naming of the kthread termination-related calls on the in-band
counterparts. At this chance, further clarify the interface by having
evl_kthread_should_stop() explicitly return a boolean status.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

We cannot safely run the stage switching machinery while running in
NMI context.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

The trap_entry handler may expect hard irqs off on entry. It may
demote the current context to the in-band stage, may return with hard
irqs on.

The trap_exit handler only runs if trap_entry was called for the same
trap occurrence. It may expect hard irqs off on entry, may switch the
current context back to the oob stage. Must return with hard irqs
off. This departs from the previous implementation which allowed the
trap_exit handler to return with hard irqs on.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

This change first asserts that the FIFO class is the topmost
scheduling class by design.  From this point, we may check this class
upfront when looking for the next runnable thread to pick, without
going though the indirection of its .sched_pick handler.

This allows the compiler to fold most of the FIFO picking code into
the generic __pick_next_thread() routine, saving an indirect call.
This is nicer to the I-cache in all cases, and spares the cycles which
would otherwise be consumed by some vulnerability mitigation code like
retpolines.

On a highly cache-stressed i.mx6q, the worst case latency figures with
this change in dropped from about 5%.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

There is no user for this one.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

This reverts commit 7c81fc4f

.

Reinstate the dovetail/poll.h wrapper files we had to drop from the
mere Dovetail code.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Some drivers may need to poll multiple heads concurrently on a single
oob_poll() invocation. Replace the single wait.next backlink to the
poll head by an array of connectors multiple heads. The oob_poll()
handler of a driver can now attach the wait descriptor it receives
(struct oob_poll_wait) to up to EVL_POLL_NR_CONNECTORS different poll
heads (currently set to 4).
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Core EVL headers should be readable from the asm-generic/ section
(e.g. evl/wait.h), which among other things requires the tracepoint
definitions not to be pulled in. The latter should be read from the
files implementating them instead.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

The syscall_exit tracer appeared in v5.9, so prefix our tracer
definitions with evl_ to avoid naming conflicts.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

# Please enter the commit message for your changes. Lines starting
# with '#' will be ignored, and an empty message aborts the commit.
#
# On branch wip/evl-5.9-rc3
# Changes to be committed:
#	modified:   include/trace/events/evl.h
#

Suggested-by: lio <liu.hailong6@zte.com.cn>
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

We need to stop the sampling timer as soon as the test sitter unblocks
from waiting on the flag due to SIGINT/SIGALRM which denotes the end
of the measurement, so that we won't push extraneous samples to the
histogram.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

A thread might resume oob from a CPU which is excluded from the set
dedicated to oob scheduling (evl_cpu_affinity), due to a migration
which occurred while in-band (e.g. sched_setaffinity()).

Make sure the thread is allowed to run shortly on the target CPU
nevertheless, so that it can exit cleanly before returning from
evl_switch_oob(). The previous implementation would leave it hanging
in limbo.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

We need a way to peek at a flag state without clearing it if set.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

All pointers passed as 64 bit long integer in the ABI in order to deal
with compat mode refer to user memory. Annotate their description
accordingly.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Enable drivers to create EVL factories in order to extend the core
services with new elements/features, benefiting from the object
lifetime management brought by the factory support.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

We need to wait for the synthetic irq fired by irq_work_queue() to be
delivered on the current CPU, so we'd better not stall the in-band
stage prior to waiting for the handler to complete (flush_work()).
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

This mode measures the delay between the moment a synthetic interrupt
is posted from the oob stage and when it is eventually received by its
in-band handler. When measured under significant pressure, this gives
the typical interrupt latency experienced by the in-band kernel due to
local interrupt disabling.

As a result of this change, the current ABI revision is bumped to 25.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

We should be able to read a complete message of the requested length
if O_NONBLOCK is clear. If set and some bytes are available, return
them.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Just like in the in-band case, we cannot fully rely on testing the
ring buffer for (non-)emptiness in order to detect data
availability. If so, kick the worker thread in order to perform some
readahead, as a last resort.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

!FMODE_STREAM does not tell us anything about atomic fpos handling,
although FMODE_STREAM would denote a non-seekable
file. FMODE_ATOMIC_POS is what we need to check instead.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Testing the input ring buffer for (non-)emptiness tells us nothing
about the readability of the target file, which may have data
available we did not pull into the buffer yet. We have to relay such
request to the driver actually handling this file in addition to
checking the ring buffer.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

A proxy should be usable for reading from an in-band file without
encurring a stage switch, just like it is for the output direction.

These changes introduce the EVL_CLONE_INPUT, EVL_CLONE_OUTPUT creation
flags for a proxy, which can be used in a mutually exclusive manner,
or combined in order to have a bidirectional channel to the target
file.

As a by-product, the output direction gains in-band poll() support.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Observables enable the observer design pattern, in which any number of
observer threads can be notified of updates to any number of
observable subjects, in a loosely coupled fashion. In the same move,
an EVL thread becomes in and of itself an observable which can be
monitored for events.

As a by-product, provide support for channeling SIGDEBUG notifications
to threads through their own observable on request instead of, or in
addition to issuing SIGDEBUG. The built-in runtime error detection
capabilities the core provides can feed health monitors, based on the
observability of threads.

The core switches to ABI 23 as a result of these changes.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>

Multiple concurrent writers could cause a missed wake up of the relay
worker, which would keep them blocked on the drain flag upon
contention indefinitely.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>