On Sun, Jan 10, 2010 at 11:16:34AM +0100, Petr Vanek wrote: > MJ> > MJ> Quite big FSO upgrade, upstream is now using separate configs > MJ> > MJ> for daemons in /etc/freesmartphone/* instead of huge > MJ> > MJ> frameworkd.conf with is nice. (In shr-u it's fsodeviced, > MJ> > MJ> fsousaged and fsonetworkd now). > > seems like fsogsmd is already in? (i don't think it was in before, > right?)
Yes its in, but not used by default (its not started as service). See /etc/frameworkd.conf there is still ogsmd enabled, but you can play with fsogsmd if you want now. The same for fsotdld, fsomusicd, fsodatad. > MJ> > MJ> Also ubifs is worth of some one-liner there.. > MJ> > > MJ> > is it now proven to work? i haven't really followed the > MJ> > conversations throughly enough to get a full picture. > MJ> > MJ> I know about 3 people with working rootfs including me. > > i know very little about ubifs but understand it should bring faster > performance and more sophisticated uSD card wearing? True, I haven't used NAND for long (even with jffs2) so I cannot compare, but running it 4 times for few minutest give me feeling that its maybe a bit faster (not objective compare at all!) and I cannot reproduce bonnie++ results max_posedon posted (I had such a low numbers that it cannot be true based on using system from the same system where bonnie++ said something about read 4Kb/s :). > MJ> > MJ> > On the same token, are there any plans to incorporate the > MJ> > MJ> > kernel speed ups s discussed on kernel list? If yes, would > MJ> > MJ> > make sense to > MJ> > > MJ> > MJ> First change (MTD_NAND_VERIFY_WRITE) pushed > MJ> > MJ> > http://git.openembedded.org/cgit.cgi/openembedded/commit/?id=f72efbb544c45594650cd5380d8dd417aa257f5f > > i opkg update upgrade-ed now, how can i see i have this already in? > > this is my kernel: > r...@om-gta02 ~ $ opkg list_installed | grep kernel-image > kernel-image-2.6.29-rc3 - > 2.6.29-oe11+gitr119862+a15608f241a40b41fed5bffe511355c2067c4e88-r7.4 >From that commit you can see that I bumped kernel revision to a15608f241a40b41fed5bffe511355c2067c4e88 which is what you have (see +gitrNNNNN+SOURCE_GIT_REVISION). > are we planning to have the debug info removed too? can this be done > also just on shr-kernel-config level? I would like to let this decission on kernel guys, PaulFertser already sent his opinion. So it can be done even on shr-kernel-config, but the decission is theirs (well they are fixing bugs for shr-kernel). And the amount of debug options for newer kernels is not that bad at all, on 2.6.32 from shr zcat /proc/config.gz | grep -i debug | grep -v ^# | wc -l => 10 2 of them are about ubifs, where is BGT disabled by default (max_posedon reported that ubi is much more reliable with that), but we can disable BGT from running system through /sys.. but nobody tried/cared yet. > MJ> As I sent later.. it works only for images built > 2009-12-09, but > MJ> yesterday I added updated update-alternatives-cworth to feeds and > MJ> upgrading that and then reinstalling busybox should work in all > MJ> images. > > i am safely past reflash && upgrade now :) Good :) > Petr -- uin:136542059 jid:[email protected] Jansa Martin sip:[email protected] JaMa _______________________________________________ Shr-devel mailing list [email protected] http://lists.shr-project.org/mailman/listinfo/shr-devel
