On Wed, Apr 26, 2017 at 2:06 AM, Ryan Tandy <[email protected]> wrote: > On Wed, Apr 26, 2017 at 01:56:34AM +0900, Roger Shimizu wrote: >> >> I installed -3 version of micro-evtd from snapshot.d.o >> add iomem=relaxed kernel option, and confirmed it can boot. >> So if the issue doesn't occur for a few days, we can guess it's caused by >> -4 version change. > > FWIW, with iomem=relaxed, -3 and -4 *should* have the same behaviour.
I've been running with "iomem=relaxed" kernel option, and -3 for nearly a week, everything is fine. So I guess we find a workaround, at least. >> If it still occurs, I think it should be the regression for the kernel. > > I suspect they'll tell us we need to get our driver into the kernel properly > and not use /dev/mem... Another option is to use GPIO interface by /sys But I don't have much experience for it. >> Is there any side effect if we add this "-a system_set_watchdog off" >> command >> to start_server() routine also? The watchdog is quite anonying AFAICS. > > IIRC it will be enabled again as soon as micro-evtd starts updating it. You > would have to patch the feature out of micro-evtd altogether. Thanks for your info. Seems better not to touch this part for now. Cheers, -- Roger Shimizu, GMT +9 Tokyo PGP/GPG: 4096R/6C6ACD6417B3ACB1
