> Andrew, > I'm also really busy, but managed to make a bootable hd on an alix board with > the older initrd based version and had success connecting to net and use most > of the stuff. > > A few observations: > > - aiccu segfaults with an error related to libpthread > > Ok' I'll look on it - possible it just require update. > - the atheros WLAN card can't be initiated - maybe related to pb's setting the > country zone with the ath5k driver; the net community says to just use the > madwifi driver as with the 2.4 kernel > > Hmm... Strange, but I use DWA-520 with ath5k on my home router/NAS/test platform (Gentoo box) in AP mode, and all is OK. Can you show me logs of error and your config files? > - access to the IDE-CF has changed from hda to sda > > Yes, I switched to SCSI-based drivers - for many reasons (old ATA aren't developed, also IMHO it'll be wrong to use old ATA-based drivers for ATA and SCSI for SATA/USB, and also new drivers make devices port-independent - it'll be no troubles if somebody connects storage that configured to work on 2nd IDE channel to 1st channel). I can see only one con of that decision - devices are initiated/detected only at driver loading, so for IDE hotplugging it'll be needed to reload kernel module. > - mounting sda results in UPPERCASE filenames on the mountpoint, though this > may have been introduced by the short way to copy all 2.6 files through my > dual setup via Bering-uClibc 3.1 to /dev/hda2. This has to be doublechecked > with a plain pxeboot setup/installation. > > By default it's mounted as VFAT - with LFN/uppercase support, but actually FS is case-insensitive (you can access to file by name in any case - like in Win/DOS). Also it's possible to specify FS type at mount - like in 3.1 version. > - I'd like to see a more generic kernel as default kernel. It's more tedious > to replace a kernel than in the 2.x times (with kernel-elan) than today. A > new kernel requires also a new initrd (at least as long as we built one), > moddb|modb_6.lrp and a new tarball with kernel modules. Not to mention that a > kernel with Geode optimization and without SMP support is about 100k smaller. > It'll also require at least to rebuild some packages for new kernel (that uses kernel modules - like accel-pptp), and use i586 optimization in all packages - that may cause some performance loss comparable to i686. Also, disabling SMP is really actual only for Geode or enough old hardware. For new hardware - even cheap Atom board (very useful for such applications) will use SMP. Possible it'll be easier/better to make distinct package tree for Geode, or even for generic Pentium MMX (which will also cover K6/K6-2/Geode)? > - I'm investigating if it's possible to run a more recent shorewall version > with uperl.lrp due to three reasons - shorewall-perl is the only shorewall > version supported in the future, shorewall-perl is the only that supports > multicast and shorewall-perl is the base of shorewall-ipv6 support. This will > allow us to replace 6wall, with shorewall, which is actively supported and > developed. > Ok' I'll look on it. > - some of the above mentioned changes requires to revise and rewrite the > documenation, which are a big plus for LEAF users. > > So I's vote to resolve a few issues mentioned above to release an alpha > version aiming at the developer community and then to start to creating the > new docs. Once those are in place (and probably with some new updated > packages) we can build a beta for interested users. > > kp > >
------------------------------------------------------------------------------ ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo _______________________________________________ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
