Most of these warnings were between int or ssize_t and size_t.
In particular, various POSIX system calls like read() and write()
return ssize_t and comparing that value with the sizeof a buffer
is a common source of these warnings. Another common source is
using an int as the iterator in a for loop with the limit being
a size_t.
With the type change, some printf() specifiers needed to change also.
This change was made to address GCC -Wshift-negative-value.
Shifting a negative value is undefined in the C programming language
because one's and two's complement representations give different
results.
Closes#5385.
rtems_message_queue_broadcast() may, under certain circumstances, copy
the message to tasks which were not waiting on the message queue when
the broadcast started, and may copy the message multiple times to the
same task.
This behaviour is, based on discussion in #4804, something that might
not change in the short term, so expose it in the user documentation.
The wording is copied with modifications from an existing internal
requirements note.
Also correct "none-atomic" -> "non-atomic" in said internal requirements
note.
The changes were auto-generated from rtems-central.
Code should not rely on the contents of local variables set before
setjmp() after the longjmp() returns. In this case, it was notpossible
to reset events after the return. But analysis determined that the
longjmp() should always returns to the same stack frame. Thus local
variables should be preserved. The warning was disabled.
This case was not addressed by adding the "returns_twice" attribute
to setjmp() in setjmp.h.
Updates #5364.
Code should not rely on the contents of local variables set before
setjmp() after the longjmp() returns. In this case, it was possible
to set cpu_self after the return from setjmp().
This case was not addressed by adding the "returns_twice" attribute
to setjmp() in setjmp.h.
Updates #5364.
Fixed many warnings for unused parameters. Some cases were
a simple matter of adding "(void) param" at the beginning
of the function, while others required ensuring that addition
was inside the proper conditional section. Found with GCC's
warning -Wunused-paramter.
On the genmcf548x BSP variants, the "request" local variable was
flagged as having a pointer taken and passed to subroutines. GCC
is unable to detect that the "request" variable is removed from
the chain in an ISR. Disabled -Wdangling-pointer for the single
test function CallWithinISR().
Test the proper serialization of explicit thread priority changes (for
example rtems_task_set_priority()) and thread priority changes carried
out by thread queues (locking protocols such as priority inheritance).
With RTEMS_DEBUG enabled, the new test case fails with an assertion. It
is fixed in a follow up commit.
Update #5159.
In _Terminate(), everything after invoking the fatal extensions is essentially
dead code. Simplify this code block and provide a reasonable safe fall-back
for badly configured applications. Just disable masked interrupts and execute
an idle loop.
Update #5067.
On some targets or configurations, the data cache cannot be disabled.
The data cache may be necessary to provide atomic operations. In SMP
configurations, the data cache may be required to ensure data coherency.
Close#5050.
If the RTEMS_INTERRUPT_LOCK_NEEDS_OBJECT define has a non-zero value,
then the interrupt lock operations require an object of type
rtems_interrupt_lock, otherwise no lock object is required.
This indication can be used to avoid the space overhead for lock objects
when they are not needed. In this case, the lock operations will not
use a lock objects parameter.
Update #4957 and #5038.
This test was broken by b031cf2b19 since
it expects the macros to evaluate to an empty string for non-SMP builds.
This updates the check for the new content.
Add directives to get and set the priority of an interrupt vector.
Implement the directives for the following BSP families:
* arm/lpc24xx
* arm/lpc32xx
* powerpc/mpc55xxevb
* powerpc/qoriq
Implement the directives for the following interrupt controllers:
* GICv2 and GICv3 (arm and aarch64)
* NVIC (arm)
* PLIC (riscv)
Update #5002.
Make the clock driver parameters configurable. Use the maximum counter
frequency to get the best time resolution. Decouple the CPU counter from the
timecounter. Make the tick catch up handling more robust. Add a validation
test for the tick catch up.
The TM27 support may define TM27_INTERRUPT_VECTOR_ALTERNATIVE to provide
an alternative software generated interrupt request which is raised by
_TM27_Raise_alternative() and cleared by _TM27_Clear_alternative().
Both functions shall return an RTEMS status code. This interrupt vector
may be used to test the interrupt controller support on targets which do
not provide generic software generated interrupts.
Update #3716.
For uniprocessor configurations, the selection of
RTEMS_MULTIPROCESSOR_RESOURCE_SHARING results in a priority ceiling
mutex.
Build the MrsP validation tests only if RTEMS_SMP is enabled.
Update #3716.
