On 10/05/2017 11:57 AM, Philippe Gerum wrote: > On 10/04/2017 10:57 PM, Jackson Jones wrote: >> I am running on an i.MX6 quad core. It is running Debian Jessie with an >> NXP/Varscite kernel 4.1.15. So it is a 32bit ARM platform like the Pi. >> >> What I meant by freeze is it hangs the entire OS/SoC, a hard reset is the >> only way out. I will check the status of all the return codes (I did not >> write this example but I should have cleaned it up when the problem was >> tossed to me). >> > Confirmed, I can badly break a couple of SoCs here using this test > program, provided I run it twice. The first run passes, second time is > always lethal. The Cobalt core seems to be doing silly things in this case. > > An additional bug also affects the interrupt pipeline for any release > starting with kernel 4.4 onward actually hides the issue in Cobalt, > making the affinity setting a nop, sort of. > > Ok, I'll follow up on this. >
That one was pretty ugly, and affects all Xenomai releases, 2.x, 3.x, all of them. Fixed in 3.x by this commit: http://git.xenomai.org/xenomai-3.git/commit/?id=59344943b0c58d5f0c8b58d2ad97f2d894d33354 Thanks for the test code, that helped a lot. -- Philippe. _______________________________________________ Xenomai mailing list [email protected] https://xenomai.org/mailman/listinfo/xenomai
