**** BEGIN LOGGING AT Fri Nov 24 03:00:00 2006 Nov 24 06:50:34 03oleo * r4548 10optware/trunk/make/nmap.mk: nmap: bump ipk to force libuclibc++ relink Nov 24 07:45:07 03oleo * r4549 10optware/trunk/make/pcre.mk: pcre: bump ipk to force libuclibc++ relink Nov 25 00:53:34 does anyone here know how the NPE-B firmware is supposed to be loaded ? Nov 25 00:53:56 I am getting "Missing microcode for NPE-B" errors Nov 25 00:54:15 but the firmware is in /lib/firmware/NPE-B Nov 25 01:00:59 is this the correct size for the firmware: -rwxr-xr-x 1 root root 11964 Jan 1 1970 NPE-B* Nov 25 01:00:59 I read the magic was 0xfeedf00d yet od -a on the file shows: Nov 25 01:00:59 0000000 fe ed f0 cr soh nul stx soh nul nul vt ac nul nul nul nul Nov 25 01:01:42 never mind, the magic is correct Nov 25 01:02:43 well maybe it is Nov 25 01:03:02 busybox od like most bb utils is missing useful options Nov 25 01:03:19 root@brokenslug:/lib/firmware$ od -t x1 NPE-B |head Nov 25 01:03:19 od: invalid option -- t Nov 25 01:04:23 jacques: ping Nov 25 01:06:41 jacques: microcode loading will work on the lude (untested, but should) if you put the microcode in a partition called npe_ucode (or you can change nas100d-setup.c in the kernel to specify a different partition). Nov 25 01:07:04 on the nslu2, it's currently stored in FIS dir but will move to sysconfig Nov 25 01:07:14 holy crap I just found it Nov 25 01:07:30 root@brokenslug:/lib/firmware$ ls -l /dev/ixp4xx_ucode Nov 25 01:07:30 crw-rw---- 1 root root 10, 184 Jan 1 1970 /dev/ixp4xx_ucode Nov 25 01:07:30 root@brokenslug:/lib/firmware$ cat NPE-B > /dev/ixp4xx_ucode Nov 25 01:07:30 swapped image found Nov 25 01:07:30 npe: firmware loaded to NPE-B, func: 00, rev: 2.1, status: 80c00000, crc: 875e Nov 25 01:07:31 root@brokenslug:/lib/firmware$ ifconfig eth0 up Nov 25 01:07:33 eth0: link up, 100Mbps, full-duplex, lpa 0x45E1 Nov 25 01:07:46 Duplex mode eth0 =1 Nov 25 01:09:19 yeah, you can feed it in that way too Nov 25 01:09:34 (that's not done in the startup scripts yet) Nov 25 01:09:39 nod Nov 25 01:09:54 I'm just glad it appears to be working - no net is a major blocker :-) Nov 25 01:10:30 I gotta say I don't look forward to totally redoing all my flash partitions on the lude, it was rather traumatic last time Nov 25 01:11:09 now, did I hear that there are known issues with this driver in LE mode? problems with scp, nfs, etc. ? Nov 25 01:11:11 well, don't make the npe-ucode one - just wait for the move to sysconfig Nov 25 01:12:10 the idea is that the last 64k of the sysconfig partition will be microcode Nov 25 01:12:37 ok that sounds good Nov 25 01:13:03 it's currently 384k on my lude Nov 25 01:13:20 not sure why Nov 25 01:13:21 and the sysconfig handling scripts get updated to not overwite that area Nov 25 01:13:35 it seems to always be way too big Nov 25 01:13:43 now I'm wondering if I can get NPE working in angstrom same way Nov 25 01:14:09 should be able to - it's kernel support Nov 25 01:14:48 yeah, I'll give it a try after I get debian working again - gotta remember how to use turnup without touching the disk partition (which already has debian on it) Nov 25 01:15:42 just don't use -i Nov 25 01:16:05 cool - that sounds familiar Nov 25 01:16:20 going back to NPE - so the location of the firmware is hard-coded in the driver ? Nov 25 01:16:34 and the firmware is included in /lib/firmware just because ? Nov 25 01:16:48 in board setup code for slugos Nov 25 01:16:59 but we doubt that will be accepted upstream Nov 25 01:17:06 ah right Nov 25 01:17:09 hence the cat to dev method Nov 25 01:17:30 only way I found cat to dev was by googling and reading a thread on the arm list Nov 25 01:17:41 is there a better way I could have found it ? Nov 25 01:17:58 yep, none of this is documented or automated yet. Nov 25 01:18:24 ok, that matches my conclusion Nov 25 01:28:52 jacques: do you have a slug too? Nov 25 01:29:58 oh this is bad Nov 25 01:30:24 apparently there's no kernel support for ide disk on lude ? Nov 25 01:30:32 in current debianslug build? Nov 25 01:31:13 I have two slugs but they are not in usable states right now Nov 25 01:31:31 that could be true - I'm not sure anyone has tested that on 2.6.19 yet Nov 25 01:32:17 :-\ Nov 25 01:32:23 lots of regressions from feb Nov 25 01:32:28 from my point of view Nov 25 01:32:44 why would ppl take stuff out of the kernel config ? Nov 25 01:32:50 *big* changes - slugos 3 to slugos 4 Nov 25 01:33:20 cause it didn't work in 2.6.18 and no-one has refreshed the patch yet Nov 25 01:33:29 at this rate it will be just as fast for me to get angstrom working on the lude as debianslug :-| Nov 25 01:33:43 that's the idea ;-) Nov 25 01:33:56 didn't work in .18 ???? what does that mean? Nov 25 01:34:00 What IDE controller is in the Lude? Nov 25 01:34:07 does that mean it may not work in .19 either ? Nov 25 01:34:18 that would pretty much kill all my plans for the lude Nov 25 01:34:31 jacques - not didn't work, just patches didn't apply cleanly Nov 25 01:34:34 If the controller was in the huge IDE patchset from cox, it didn't make it into 19 Nov 25 01:35:28 the news just keeps getting better Nov 25 01:35:36 :-) Nov 25 01:35:56 you mean to get my lude working with current kernel I may need to much around with merging patches that nobody else could merge? Nov 25 01:36:00 in that case I am sunk Nov 25 01:36:01 you're trailblazing for the lude right now Nov 25 01:36:02 It is good news about the IDE patchset (I'm told) -- the support is supposed to be in the .19 kernel. Nov 25 01:36:16 mwester, OK I will check that out Nov 25 01:36:18 just needs enabling maybe Nov 25 01:36:25 I'll see what I can find out Nov 25 01:36:41 The ARTOP IDE controller for the dsmg600 is in the .19 release now, anywya. Nov 25 01:36:41 I'm just in shock how much is broken that worked in feb Nov 25 01:36:55 according to the wiki the lude has an ACARD Nov 25 01:37:00 Acard ATP865-A Nov 25 01:37:42 I kinda agree; all was stable and a relative kernel newbie such as myself was able to keep up with the stuff right around the 3.10 SlugOS release - but soon after that, developments began occuring so fast that I haven't been able to entirely keep up! :D Nov 25 01:37:59 I don;t know if I call breaking things "development: Nov 25 01:38:01 jacques: feb was slugos 3, ready for binary release Nov 25 01:38:18 Cliche warning: Nov 25 01:38:26 now is slugos 4 with major changes (apex, npe driver) Nov 25 01:38:30 "Gotta break some eggs to make an omelet!" Nov 25 01:38:48 * mwester ducks after using that horrible overused cliche Nov 25 01:39:03 looks like the lude is an orphan Nov 25 01:39:11 and you know what that means - not worth the time Nov 25 01:39:33 just second place - slug wasn't working until last week Nov 25 01:39:40 lude is next on my list Nov 25 01:40:10 oh yeah rwhitby-treo has one. that's good for me :-) Nov 25 01:40:19 I have two :-) Nov 25 01:40:33 ok that motivates me a bit Nov 25 01:40:40 What NIC on the lude? Nov 25 01:40:52 if I was the only one who cared, it would be better worth my time to look at the LSPro Nov 25 01:41:07 mwester, ixp4xx Nov 25 01:41:14 jacques: I have a lude with serial and remote power control and auto-on in my test setup at work now Nov 25 01:41:38 gonna put jtag on it too, and get openocd xscale debugging working eventually Nov 25 01:41:53 rwhitby-treo, were you asking about my slug before because you would like me to test latest build on it ? Nov 25 01:41:54 (single step the kernel, for instance) Nov 25 01:42:05 I suppose I could try to get one working Nov 25 01:42:18 jacques: just so you could get a feeling of success with the slug :-) Nov 25 01:42:20 one is my u-boot mule, other one lost its serial port Nov 25 01:43:04 lost as in I needed the level converter :-) Nov 25 01:43:59 ok I will investigate IDE support for lude in latest kernel Nov 25 01:45:23 excellent :-) cause it should work better now with latest patches from alan coz Nov 25 01:45:25 cox Nov 25 01:46:06 nice Nov 25 01:46:09 I'm gonne get apex support and npe auto firmware loading working on the lude when I get some time Nov 25 01:46:33 so apex as stage 2 bootloader ? Nov 25 01:46:44 instead of ludeboot ? Nov 25 01:47:30 yep Nov 25 01:47:52 marc singer is working on integrating open source npe driver into apex this weekend Nov 25 01:48:05 cooool Nov 25 01:48:16 I still want to get it working on slug in u-boot Nov 25 01:48:23 some day Nov 25 01:48:46 u-boot devs lost interest because of the microcode license Nov 25 01:50:01 that's why we put it in a separate partition Nov 25 01:50:18 maybe we can get apex to load it from there too Nov 25 01:50:31 what would be nice Nov 25 01:51:35 not all devices have 16MB flash like the lude :-) Nov 25 01:51:39 apex has fis dir support now, so it shouldn't be too jard Nov 25 01:51:49 hard Nov 25 01:52:02 of course, angstrom seems to assume all devices have at least 64MB flash :-| (most handhelds do) Nov 25 01:52:58 firmware is less than 30k Nov 25 01:53:06 microcode I mean Nov 25 01:53:23 the one I have is like 11k Nov 25 01:53:48 yep Nov 25 01:54:06 we need ucode with dual npe support for lude Nov 25 01:54:08 BTW, this big Alan Cox IDE patch - is it in the current 2.6.19 kernel ? Nov 25 01:54:19 the current one only has npe-b Nov 25 01:54:41 rwhitby-treo, what's the difference ? lude only has one network port Nov 25 01:54:47 dunno about the patch - mwester or swery no best Nov 25 01:55:22 hmm - I thought lude had a second internal for wifi, but now I remeber that's on pci Nov 25 01:55:39 Dunno the all the details, but it seemed when I enquired that I was told that "it was in the .19 kernel" -- but I don't seem to recall that the words were that his actual code was in the .19 kernel... Nov 25 01:55:59 I just scrolled back and re-read this: If the controller was in the huge IDE patchset from cox, it didn't make it into 19 Nov 25 01:56:25 I meant that the IDE patchset didn't make it in; device support may have made it in. Nov 25 01:57:30 Key components for the dsmg600 were the mechanism to make the IDE drives appear as /dev/sd instead of /dev/hd, and the ARTOP driver (PATA_ARTOP). That's all I needed for the dsmg600. Nov 25 01:58:46 I have to build the latest to try to reflash the dsmg600 anyway; I'll take a look to see if the Lude controller is listed. It'll take an hour or two before I can get there. Nov 25 01:59:02 I was about to look Nov 25 01:59:17 well, after I figure out the monotone issue on my t42 Nov 25 02:03:01 monotone issue appears to be that I uninstalled monotone :-) Nov 25 02:07:43 LOL! Yep, that'll do it. But make sure you re-install an older release, otherwise it will make you reformat the database, which (at least when I tried it) performed about at the same speed everything else about Monotone does -- meaning that a monk could transcribe the data with quill and ink at the same speed. Nov 25 02:07:58 hehehehe That's so true! Nov 25 02:08:36 heh Nov 25 02:11:25 wow, someone changed the colors in 'make menuconfig' to actually be readable Nov 25 02:11:33 that only took like 10 years! Nov 25 02:12:59 looks like this should do it for lude IDE : CONFIG_BLK_DEV_AEC62XX Nov 25 02:13:01 i liked the colors Nov 25 02:13:19 This driver adds explicit support for Acard AEC62xx (Artop ATP8xx) IDE controllers. Nov 25 02:13:49 yellow on gray doesn't work for me Nov 25 02:14:08 yellow on any color other than black or red doesn't work IMHO Nov 25 02:15:15 so I guess acard and artop are the same thing ? confused Nov 25 02:18:04 Must be. That's the driver I've tested on the dsmg600 - worked for me. Nov 25 02:18:14 cool Nov 25 02:18:57 Doesn't the default defconfig have that driver built as a module? (It was supposed to, but that might have been lost in the recent changes) Nov 25 02:19:25 let me check - either way, it wasn't installed in debianslug image Nov 25 02:20:59 # CONFIG_IDE is not set Nov 25 02:21:12 Oh dear. Nov 25 02:21:16 that's the first problem. Nov 25 02:22:09 hmm - that must date all the way back to when blaster8 did a new 2.6.18 config instead of upgradin the existing one :-( Nov 25 02:23:39 back in a while - hungry. :-) Nov 25 02:24:57 It's still on my list to go make the current HEAD run on the dsm-g600; I've just been distracted by a job change right now. Before we release 4.x I am committed to making sure we have at least rudimentary support for the beast. Nov 25 02:37:01 a Nov 25 02:37:05 same for lude Nov 25 02:37:19 slug, lude, loft, dsmg600 Nov 25 02:37:37 maybe fsg-3 too ;-) **** ENDING LOGGING AT Sat Nov 25 02:59:57 2006