**** BEGIN LOGGING AT Wed Jul 25 02:59:57 2007 Jul 25 03:00:17 <[mbm]> go bug florian, he's the one playing with rdc boards Jul 25 03:00:22 he only has one Jul 25 03:00:29 and it's the same as mine Jul 25 03:00:59 you will not get an openwrt kernel to boot on the sitecom until i make a profile for it Jul 25 03:01:24 <[mbm]> figured I'd atleast get a few lines of dmesg Jul 25 03:05:13 also, don't forget that openwrt revisions after 7314 produce no dmesg on any rdc device Jul 25 03:10:36 <[mbm]> ? Jul 25 03:12:42 cya later Jul 25 03:21:43 [mbm]: what, you're unaware the port is broken? Jul 25 03:29:34 * [mbm] hasn't used the rdc port before Jul 25 05:31:42 [mbm], [florian]: re Jul 25 05:32:05 <[mbm]> . Jul 25 05:32:51 [mbm]: were you able to jtag the sitecom? Jul 25 05:33:23 <[mbm]> don't see anything that looks like a standard jtag header Jul 25 05:33:35 <[mbm]> which is why I was screwing around with serial Jul 25 05:34:29 it's labeled JP1 in crazy_imp's photos Jul 25 05:35:20 <[mbm]> a 1x6 connector? Jul 25 05:35:23 yes Jul 25 05:40:50 <[mbm]> hmm not used to that form factor Jul 25 05:41:23 <[mbm]> anyways, I'm more interested in getting redboot to tftp boot a kernel Jul 25 05:41:33 chicken and egg Jul 25 05:41:51 <[mbm]> not particularly Jul 25 05:42:26 but if you're that curious, at least for now, you need to roll back svn to r7314 Jul 25 05:42:27 <[mbm]> mostly the result of a braindead version of redboot Jul 25 05:42:36 <[mbm]> already did, didn't help Jul 25 05:42:52 no dmesg? Jul 25 05:43:22 <[mbm]> like I said, this is a braindead version of redboot Jul 25 05:43:34 <[mbm]> it only supports a few commands Jul 25 05:43:57 i had to look at the source to this redboot to figure it out Jul 25 05:44:15 you can get it from linksys's ftp Jul 25 05:44:33 <[mbm]> http://pastebin.ca/raw/632815 Jul 25 05:45:39 that looks slightly different Jul 25 05:45:58 i'd have a better idea with a complete flash image Jul 25 05:46:06 <[mbm]> tehre's only load and linux .. the go and reset commands are missing Jul 25 05:46:43 <[mbm]> you can tell it to load an elf file, and then run linux, but then it complains that the kernel doesn't have a bootsector Jul 25 05:47:33 i don't think any of the rdc redboots have "go" Jul 25 05:47:59 <[mbm]> it also has basic problems with the load command Jul 25 05:48:35 <[mbm]> seems that you're limitedf on your choice of base addresses, ram isn't properly mapped Jul 25 05:49:36 <[mbm]> ie, it blindly tries to load into the area of ram specified, but then reading it back fails when the ekrnel is executed Jul 25 05:50:18 <[mbm]> figured it should be: load -r -b 0x100000 bzImage; linux -c "console=ttyS0,38400" Jul 25 05:52:26 <[mbm]> but when I do that it ends up executing the 2.4 kernel from flash Jul 25 05:56:30 <[mbm]> anyway, sitecom site has the redboot sources Jul 25 05:56:50 good Jul 25 06:07:23 <[mbm]> any ideas? Jul 25 06:09:41 i thought you were looking at the sitecom source Jul 25 06:10:54 <[mbm]> yep Jul 25 06:11:22 <[mbm]> figured you were as well Jul 25 06:12:38 not atm Jul 25 06:15:11 <[mbm]> anyway, I do have a full flash dump of the nas Jul 25 06:15:32 <[mbm]> but unfortunately that one doesn't use redboot Jul 25 06:16:42 [florian]: replace the driver in the trunk with http://homepage.mac.com/danielg4/r6040.c Jul 25 06:27:13 glp_home: pong Jul 25 06:27:22 glp_home: Sorry - I've been away for a few days Jul 25 06:33:57 [mbm]: what is the load command you're using? Jul 25 06:34:35 <[mbm]> scroll up Jul 25 06:36:43 load -r -b 0x400000 bzImage Jul 25 06:36:44 ? Jul 25 06:37:08 <[mbm]> yep, but if you remember the pastebin output, the ram regions available are somewhat limited Jul 25 06:37:17 ah Jul 25 06:37:18 <[mbm]> so that doesn't actually work Jul 25 06:37:26 noz: hi'a - wanted to pick up on you'r suggestion about looking at svn tickets during WIP-Day? Jul 25 06:37:42 <[mbm]> RAM: 0x00000000-0x000f0000, 0x00070470-0x000a0000 available Jul 25 06:38:02 glp_home: OK. Jul 25 06:40:27 [mbm]: but isn't 0x400000 the address it loads the kernel from flash to? Jul 25 06:40:51 <[mbm]> yep Jul 25 06:43:31 noz: thought it maybe could be good to make a note in the forum, that if members would like to take a look at fixing tickets the're people around on #openwrt-devel to help? Jul 25 06:47:11 maybe adding a few suggestions for what to look at (ticket suggestions) would be good? Jul 25 06:47:50 [mbm]: i don't see the code for shadowing the kernel Jul 25 06:54:06 <[mbm]> figure I'll probably have to boot into a newer version of redboot Jul 25 07:05:36 RAM: 0x00000000-0x000f0000, 0x00072ed0-0x000a0000 available Jul 25 07:05:51 that's what the ar525w says Jul 25 07:06:05 doesn't affect the load command Jul 25 07:08:57 [mbm]: what's the output of "load -r -b 0x400000 bzImage" ? Jul 25 07:12:41 <[mbm]> http://pastebin.ca/raw/632844 Jul 25 07:12:48 <[mbm]> note that it boots the kernel from flash Jul 25 07:14:58 <[florian]> sn9: sir , yes sir Jul 25 07:15:49 along with the other changes in http://homepage.mac.com/danielg4/rdc.patch Jul 25 07:17:22 [mbm]: i found the explanation Jul 25 07:17:47 the kernel shadowing code is _inside_ the "linux" command Jul 25 07:18:08 so it can only boot from flash Jul 25 07:20:24 <[mbm]> ugh Jul 25 07:20:48 <[mbm]> what about booting another version of redboot? Jul 25 07:21:37 only from flash Jul 25 07:21:54 so, again, jtag first Jul 25 07:24:18 <[florian]> sn9: what about using real commands for the ioctl like SIOCGMII and friends ? Jul 25 07:28:26 [florian]: i don't think the ioctl is necessary for standard utils. it's apparently just for tuning the switch, so having its own userspace tool is appropriate. whether openwrt would have any use for it is another matter Jul 25 07:33:58 <[florian]> sn9: right, ideally it should go in kmod-switch, but that's when rdc really works Jul 25 07:35:38 for now, http://homepage.mac.com/danielg4/rdc.patch and http://homepage.mac.com/danielg4/r6040.c can go in the trunk Jul 25 07:38:39 <[florian]> sn9: is not the mac address for the ethernet devices stored on the flash ? Jul 25 07:39:20 eth1 has the same mac as the wifi, and eth0 is that +1 Jul 25 07:39:31 so, no Jul 25 07:44:25 <[florian]> the rt61 reads the mac adress from the intermal eeprom of the wifi card ? Jul 25 07:44:39 yes, evidently Jul 25 07:50:36 <[florian]> I see no cleaner way to read the mac addres than what you have done Jul 25 07:51:00 if i saw a cleaner way, i would have used it Jul 25 08:17:07 <[florian]> sn9: does not /sbin/int work as expected ? Jul 25 08:27:21 <[florian]> sn9: also, we do not enable NAPI with the r6040 driver Jul 25 09:16:18 florian * r8153 /trunk/target/linux/rdc-2.6/ (5 files in 5 dirs): Upgrade rdc to 2.6.22.1, include r6040 ethernet fixes by sn9, thanks ! Jul 25 09:22:22 florian * r8154 /trunk/package/kernel/modules/network.mk: Put r6040 ethernet driver as a module Jul 25 09:26:47 <[florian]> sn9: here we go Jul 25 11:46:03 * crazy_imp read the log Jul 25 11:48:59 [mbm]: i could tell you the pinout from the JP1 header, traced the lines to the SoC (and it's easy to prove the layout because you just need to go one (or two) pins ahead on the pins from the cpu if you already know the right order) Jul 25 12:57:27 [mbm]: JP1 |3.3V| (gnd) (tck) (tdo) (tdi) (tms). afk for about 3h Jul 25 13:51:42 florian * r8155 /packages/libs/nmeap/ (. Makefile patches/ patches/100-no_test.patch): Add libnmeap (#1999) Jul 25 14:00:41 florian * r8156 /trunk/package/aodv-uu/patches/005-linux_2.6.22_skbuff.patch: Fix aodv-uu compilation on 2.4 kernels (#2116) Jul 25 14:07:01 nbd * r8157 / (2 files in 2 dirs): merge madwifi r2591 for rxantenna selection Jul 25 14:19:56 [florian]: would you like to add the NAPI .config option to target/linux/rdc-2.6/patches/002-r6040_ethernet.patch yourself? Jul 25 14:20:27 <[florian]> sn9: yes, I will do so Jul 25 14:20:33 <[florian]> sn9: have you tested it already ? Jul 25 14:20:47 <[florian]> sn9: I really need to take more time hacking on the rdc stuff again Jul 25 14:20:50 NAPI? no Jul 25 14:21:27 but the .config option should be available to turn on and off Jul 25 14:21:52 kernel config, that is Jul 25 14:23:05 also, it would make more sense to rename Ralink.mk to ar525w.mk in svn now Jul 25 14:35:02 <[florian]> absolutely Jul 25 14:38:28 florian * r8158 /packages/net/freeradius/Makefile: Move freeradius libs to /usr/lib since the configure script does not honnor the libdir/libexecdir paths (#1837) Jul 25 14:47:13 juhosg * r8159 /trunk/target/linux/ (12 files in 6 dirs): [adm5120] uses AMBA PL010 driver for the serial ports from now Jul 25 14:52:12 [florian]: also, the file target/linux/rdc-2.6/image/preinit.arch can be deleted now Jul 25 14:52:21 <[florian]> sn9: yep Jul 25 14:53:51 heh i just svn updated to the current rev, and now i get "Error: Bad value (i486) for -march" Jul 25 14:54:09 in uclibc? Jul 25 14:54:15 yes Jul 25 14:54:25 known bug. try make toolchain/uClibc-clean world Jul 25 14:54:57 somehow, my global .config got hosed, too Jul 25 14:56:06 oh, i had broken platforms off, and now the platform is marked broken. duh. Jul 25 15:05:02 nbd: "make menuconfig" now spews tons of warnings about referring to undefined symbols, and all the ethernet kernel module options are gone Jul 25 15:05:12 florian * r8160 /trunk/target/linux/rdc-2.6/ (5 files in 4 dirs): Remove preinit.arch, not used, compile NAPI support for R6040, change the profile name to ar525w Jul 25 15:07:31 [florian]: you didn't make CONFIG_R6040_NAPI depend on CONFIG_R6040 Jul 25 15:08:14 <[florian]> sn9: oh right, grr doing too many thing at the same time Jul 25 15:09:31 florian * r8161 /trunk/target/linux/rdc-2.6/patches/002-r6040_ethernet.patch: NAPI for r6040 of course depends on r6040 Jul 25 15:24:40 sn9: what platform? Jul 25 15:24:53 rdc-2.6 Jul 25 15:25:11 no idea Jul 25 15:25:24 maybe someone forgot to enable something in the conig Jul 25 15:25:50 will check after the "make world" fails Jul 25 15:31:51 eh, screw it. i'll do a distclean Jul 25 15:36:33 aha!! ERROR: please fix package/kernel/Makefile Jul 25 15:37:03 sn9: run svn status Jul 25 15:37:27 whoops Jul 25 15:41:07 thanx Jul 25 16:39:03 http://pastebin.ca/633323 (building for rdc, r8154 with the default settings, only removed ppp) Jul 25 16:43:08 crazy_imp: don't build recent revisions of openwrt for rdc yet Jul 25 16:43:26 especially without a wl153 profile Jul 25 16:44:39 sn9: i know... :). but it's broken anyway Jul 25 17:37:13 nbd * r8162 / (2 files in 2 dirs): fix the size of the flash mem window on pronghorn metro (#2070) Jul 25 17:57:07 florian * r8163 /tags/kamikaze_7.07/target/linux/x86-2.6/config/default: Enable r8169 as a module (#2129) Jul 25 19:59:58 hey this afternoon I loged into my router to check things and noticed that there seems to be a mem leak in miniupnpd it was using 38% memory and on restarting the process its using 2.8%, anyone else noticed this? I'm going to post on their forums about it when I'm activated Jul 25 20:00:18 what kind of device Jul 25 20:00:25 ar7 Jul 25 20:00:33 hm Jul 25 20:00:35 no idea Jul 25 20:00:42 but bugs in miniupnpd... that's not unheard of ;) Jul 25 20:00:48 :) Jul 25 20:03:09 report it to the miniupnpd developers http://miniupnp.free.fr/ Jul 25 20:04:22 yeh I'm waiting for forum activation Jul 25 20:04:49 ok :-) Jul 25 21:06:21 pavlov * r8164 /packages/net/squid/ (6 files in 2 dirs): refresh patches and update to STABLE14 of squid Jul 25 21:42:40 Wipster: could take a while Jul 25 21:42:46 a few days :P Jul 25 22:05:10 no hurry, gives more time for my program to build up space again, so I can get an exact time frame vs size :) Jul 25 22:07:32 can u dump the space it occupies for debugging? or do I need to run it through something? Jul 25 22:08:26 wtf is changeset 8107 all about? as crazy_imp reported earlier, it breaks the build Jul 25 22:10:10 what is "@#" supposed to do to a comment in a makefile, anyway? Jul 25 22:15:11 anyone? Jul 25 22:18:58 bizzare problem Jul 25 22:20:13 what's more bizarre is that it was apparently originally a bizarre solution to an equally bizarre problem Jul 25 22:21:41 yeh Jul 25 22:23:07 unless something useful gets suggested within the next two minutes, i'll be back in an hour Jul 25 23:35:12 still nothing? **** ENDING LOGGING AT Thu Jul 26 02:59:56 2007