From: Steve S. <sa...@gm...> - 2010-10-20 18:17:39
|
On Tue, Oct 19, 2010 at 10:15 AM, Steve Sakoman <sa...@gm...> wrote: > On Tue, Oct 19, 2010 at 9:49 AM, Ash Charles <as...@gu...> wrote: >> On Tue, Oct 19, 2010 at 7:32 AM, Philip Balister <ph...@ba...> wrote: >>> I also see this on a kernel that comes from the gumstix 2.6.35 branch. >>> At one point, I was blaming my driver, but it is good to hear other >>> people see this. >> >> Hi Alex and Philip, >> >> Steve and I are struggling to consistently replicate this problem; >> Steve has never seen it and I've only had it twice on a card that may >> have seen some memory corruption. >> I've passed the bug onto one of our testers to see if he can replicate >> it. Any suggestions you have to make the problem occur would be most >> welcome (and necessary in order that we can actually have a chance of >> figuring it out). This is the setup so far: >> -> Tobi + Overo Earth with microSD using Steve's pre-built images from Oct.13th >> -> Boot the board and use to dhclient to make sure the ethernet has an >> IP address >> -> Let the board run for 15 minutes >>> issue a 'shutdown -r now' and look for a crash (not the known PVR crash) > > I've been able to reproduce it with the console image. > > It seems to be occurring in generic DSS code, so I don't think it is > anything Overo specific. I'll investigate further. Yes, this is a generic issue in upstream Linux 2.6.34 (and even in 2.6.35). This error will occur if you do a shutdown when the display is disabled by the power management code after 10 minutes of inactivity. If the display is active you won't see this error. That is why Ash didn't see it unless he waited 10 minutes. I've push a fix and triggered my build machine to generate a new set of images. Steve |