On Wed, Nov 19, 2008 at 11:31:42AM +0900, James Bottomley wrote: > On Tue, 2008-11-18 at 21:04 +0100, Helge Deller wrote: > > I've tried various -rc candidates in the last few weeks, and sadly the > > parisc kernel is in pretty bad shape. > > > > At the beginning I started to get problems only when I had USB > > keyboard/mouse attached to the PA machine. This lead to bugzilla > > http://bugzilla.kernel.org/show_bug.cgi?id=11913 > > > > Now, with -rc4 and -rc5 things got even worse. > > > > Crashes happen often just directly at the beginning. > > I'm sure it's not USB related any more, and I assume we have some > > strange mm-problem somewhere. > > I'd be really dubious about that. Our pa8900 ... the most mm unstable > system we have ... is running -rc5 just fine. The only warning before > boot is the usual SMP one. I'm not. During a normal kernel build: /bin/sh: line 1: 2985 Trace/breakpoint trap ( cat /dev/null ) > sound/mips/modules.order make[2]: *** [sound/mips/modules.order] Error 133 Restarting the build allowed it to finish. This is 2.6.28-rc5 trying to build itself again. I didn't see this with -rc3 though it possible it's just chance that I didn't. Config file is on: http://iou.parisc-linux.org/~grundler/2.6.28-rc5-config-j6000 (or gsyprf11:~grundler/public_html/2.6.28-rc5-config-j6000) > Linux ion 2.6.28-rc5 #20 SMP Tue Nov 18 18:16:35 PST 2008 parisc64 > > Why don't we compare configs? This is mine for ion: > > http://parisc-linux.org/~jejb/config-2.6.28-rc5-ion > > I suspect we have some nasty interaction in the device driver subsystem > (ion, the pa8900, isn't running either USB or PDC stable). That's possible. But the IOMMU code hasn't changed. I'm inclined to believe another VM bug has crept in. I agree with Dave that kernel hasn't gotten more stable since 2.6.22 kernel that we keep running on gsyprf11. hth, grant -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
