Yes, rtems 5 also has it. I can create a ticket. Regarding the proposed solution, I wanted to start a thread for discussing it (maybe there is better was to do it).
>Is this a bug in rtems 5 also? > >If so, does it warrant a back-port fix and a ticket? > >On Mon, Apr 12, 2021 at 3:16 AM Moyano, Gabriel <gabriel.moy...@dlr.de> wrote: >> >> Hello everyone, >> >> I've found what can be an issue in the function genirq_set_active(): under >> some conditions it can return a value greater than 1. >> >> This function is used by genirq_enable() and genirq_disable() and both of >> them returns the value returned by genirq_set_active(). According to the >> documentation in genirq.h, they should return -1, 0 or 1. >> >> When this issue can happen? If there are 3 entries in the list of IRQ and 2 >> of them are already enabled, the variable `enabled` would be 2, because of >> `enabled += isrentry->enabled`. >> >> As a possible solution, the value of `enabled` can changed to 1 if it's >> greater than 1 (see the patch) or maybe improve the search. >> >> Thanks in advance, >> >> Gabriel Moyano >> >> >> >> Moyano, Gabriel (1): >> grlib/genirq: Taking into account that it could be more than one ISR >> enabled/disabled >> >> bsps/shared/grlib/irq/genirq.c | 5 +++++ >> 1 file changed, 5 insertions(+) _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
