**** BEGIN LOGGING AT Mon Apr 02 02:59:56 2007 Apr 02 06:58:14 <[florian]> sn9__: ping ? Apr 02 06:58:20 pong Apr 02 06:58:59 <[florian]> any progress on the flash map driver ? Apr 02 06:59:04 yes Apr 02 06:59:23 <[florian]> does it work as you want? Apr 02 07:00:26 i will get it to compile with a few fewer warnings, then "make clean distclean" and recompile overnight before trying. hopefully, what i will try next in the build process will fix the symbol error Apr 02 07:01:40 but the code _looks_ like it might be a marvel of ingenuity Apr 02 07:01:56 <[florian]> if you don't manage fixing the symbol error,that I still did not fix, switch to rdc-2.6 :) Apr 02 07:02:25 many thanx to nbd for some of the ideas that i failed to imagine for the driver Apr 02 07:02:57 oh, you were getting the symbol error, too? Apr 02 07:03:06 <[florian]> yes Apr 02 07:03:34 i was under the impression your 2.4 worked fine; otherwise i would have switched sooner Apr 02 07:04:35 <[florian]> I think there is something badly cross-compiled when your build machine is x86 and the target too Apr 02 07:05:01 oh, did it work for you from another arch? Apr 02 07:05:28 <[florian]> this is what I will test Apr 02 07:05:46 i can try on ubuntu-ppc Apr 02 07:05:58 <[florian]> I wanted to try on gentoo-ppc :p Apr 02 07:06:08 go ahead Apr 02 07:07:31 <[florian]> I am not sure airlink.c is endian safe Apr 02 07:07:37 <[florian]> anyway, I will have a try Apr 02 07:11:23 i will pack up my updated target for you Apr 02 07:11:23 and get started on 2.6 while you mess with 2.4, for a change Apr 02 07:11:30 i have endian-patched it a bit more since the last commit Apr 02 07:11:34 after which, i don't see further endian issues in it Apr 02 07:58:32 i am now making a first attempt at 2.6 Apr 02 07:59:20 there is xargs breakage in the buildroot on my system, btw. next patch will also include a fix Apr 02 08:11:36 if you'd like to see things before my build completes (or after, but before i come back), http://homepage.mac.com/danielg4/big.patch and http://homepage.mac.com/danielg4/rdc-2.4.tar.bz2 Apr 02 08:12:33 [florian]: ok? Apr 02 08:15:16 <[florian]> sn9__: ok, I am having a look right now Apr 02 08:15:26 at which? Apr 02 08:19:45 <[florian]> in 30 minutes Apr 02 08:19:57 ? Apr 02 08:20:17 <[florian]> oh sorry, at both Apr 02 08:20:56 what do you think so far? Apr 02 08:22:08 <[florian]> I think you could have get rid of the gemtek code, and I do not like the -DRDC3210_NVRAM_IS_JFFS2 Apr 02 08:22:49 <[florian]> your xargs substitution is highly welcome, another user met problems with xargs too as far as I remember Apr 02 08:23:18 nvram as jffs2 is my own contribution Apr 02 08:23:37 <[florian]> sn9__: I think, like in the brcm47xx-2.6 flash map driver, we can autodetect the fs type by reading the magic number Apr 02 08:23:51 i already do that for root Apr 02 08:24:05 <[florian]> ok Apr 02 08:24:42 the reason i included that option was to allow a jffs2 union with a squashfs root Apr 02 08:25:18 <[florian]> it is already possible if you name the root partition as rootfs and the corresponding kernel option is enabled Apr 02 08:25:31 <[florian]> the split is automatically done, and the overlay is managed by mini_fo Apr 02 08:25:57 <[florian]> this would actually just mean renaming the romfs to rootfs Apr 02 08:26:00 squash and jffs2 in the same partition? Apr 02 08:26:33 <[florian]> yes Apr 02 08:26:59 that means more padding for the kernel Apr 02 08:28:09 after all, the jffs2 must begin on an erase block boundary Apr 02 08:29:12 <[florian]> yes Apr 02 08:29:50 i think there is room for a compromise: have the map driver do the split like this, then have mini_fo still overlay Apr 02 08:30:13 that way, squashfs can begin earlier Apr 02 08:30:30 <[florian]> on brcm the squashfs is first if I remind correctly Apr 02 08:31:13 well, then it's not all that useful a comparison Apr 02 08:32:20 i see this: block0 == kernel, block1 == squashfs, block2 == jffs2, block3 == redboot Apr 02 08:32:52 i took gemtek's lead in making an additional partition optional Apr 02 08:33:05 <[florian]> that is a good start yes Apr 02 08:34:27 still don't like -DRDC3210_NVRAM_IS_JFFS2 ? Apr 02 08:34:57 i had a separate such option for romfs until nbd's idea Apr 02 08:35:17 but that does not extrapolate to nvram Apr 02 08:38:58 <[florian]> if the bootloader does not even need nvram to be there, let's extend the romfs to it if possible ? Apr 02 08:39:29 <[florian]> that is what you want to do, right ? Apr 02 08:39:37 i already extended block0 to it in the presence of the option Apr 02 08:39:57 i think that's a nice compromise Apr 02 08:41:42 <[florian]> yeah, right now I cannot imagine a better solution Apr 02 08:42:31 i thought about that, and you see what i came up with Apr 02 08:44:20 <[florian]> next time do not hesitate to put your name in the the header Apr 02 08:46:08 functionality first Apr 02 08:50:17 i also wanted to maintain backward compatibility and interoperability in the driver Apr 02 09:06:51 <[florian]> I will have a test Apr 02 09:34:42 florian * r6817 /trunk/target/linux/adm5120-2.6/ (4 files in 4 dirs): Fixes from Compex to make OpenWrt run on WP54AG, thanks ! Apr 02 09:40:02 nbd * r6818 /trunk/package/busybox/patches/350-httpd_redir.patch: patch for busybox httpd that adds an option for redirecting all requests to a specific page (e.g. for making a splash screen) Apr 02 09:41:57 nbd * r6819 /trunk/package/base-files/files/lib/network/config.sh: network config: ignore nonexistant interfaces Apr 02 10:20:42 florian * r6820 /trunk/target/linux/adm5120-2.6/config/default: Config is now more complete : firewalling, filesystems ... Apr 02 10:21:39 florian * r6821 /trunk/target/linux/adm5120-2.6/Makefile: Enable squashfs for adm5120 as well Apr 02 10:27:12 florian * r6822 /packages/multimedia/motion/ (3 files in 2 dirs): Update motion to a recent snapshot (#1538) Apr 02 11:31:55 florian * r6823 /trunk/target/linux/adm5120-2.6/config/default: Another more complete config change Apr 02 11:32:16 florian * r6824 /trunk/target/linux/adm5120-2.6/profiles/105-Texas.mk: Add a TI profile Apr 02 11:38:57 [florian]: the AR7 image you ran on the conceptronic machine, just default 2.4 target with squashfs, right? Apr 02 11:40:17 <[florian]> armijn: 2.6 target with squashfs Apr 02 11:40:28 but then there is no DSL, correct? Apr 02 11:41:08 nor serial port Apr 02 11:41:10 or... Apr 02 11:41:24 <[florian]> serial port is working, but dsl, not yet Apr 02 11:41:28 at least, when I look at https://dev.openwrt.org/wiki/ar7-2.6 Apr 02 11:41:29 ok Apr 02 11:42:03 florian * r6825 /trunk/include/prereq-build.mk: Add gawk as a preriquisite (needed for ramdisk images) Apr 02 11:42:32 florian * r6826 /trunk/include/prereq-build.mk: Typo, nice Apr 02 11:44:50 [florian]: but there is no ETA on the DSL drivers? :) Apr 02 11:46:23 <[florian]> armijn: I do not know, ask ejka Apr 02 11:47:31 [florian]: ok Apr 02 11:47:48 don't get me wrong, I'm very pleased I can report progress to Conceptronic, so they can see there is progress Apr 02 11:48:11 <[florian]> armijn: it "just works without DSL" Apr 02 11:48:34 great :) Apr 02 11:48:48 <[florian]> been playing with it Apr 02 11:48:57 <[florian]> I will a comment on the flashing procedure in the wiki Apr 02 11:49:11 ok, thanks Apr 02 12:05:59 nbd * r6827 /trunk/target/linux/ixp4xx-2.6/image/ (Makefile apex/Makefile npe-ucode/Makefile): fix xscale compile (#1495) Apr 02 12:51:54 armijn: dsl will take some time Apr 02 12:52:06 and I will need a guinea pig for testing Apr 02 12:53:22 i will test occasionally Apr 02 12:53:33 as soon as it works reasonably well, i will start using it for my main line :) Apr 02 12:53:53 ejka: how often do you need a pig? Apr 02 12:53:57 guinea pig even Apr 02 12:54:21 armijn: sometimes, I don't have dsl line here, and probably won't get one until summer Apr 02 12:55:47 ok Apr 02 13:18:46 ejka: I am happy to be your guinea pig; I'm finding making sense of the TI driver hard going. Apr 02 13:22:07 Is there a reason why the ldconfig in utils in the uclibc source is not packaged - http://wiki.openwrt.org/UsbStorageHowto the ipkg-link symlink hack is horrible. (works I have just tested it - is there a sensible Makefile to add it to ? (it needs a full built gcc afaik) Apr 02 13:31:25 kaloz * r6828 /trunk/target/linux/ixp4xx-2.6/image/apex/patches/100-slugos-nslu2-armeb_config.patch: fix apex commandline for squashfs on the nslu2 Apr 02 13:32:08 <[florian]> h3sp4wn: space I guess, and few ppl are likely to use it Apr 02 14:20:48 [florian]: The binary is 27k Apr 02 14:40:18 <[florian]> h3sp4wn: ok Apr 02 14:41:12 kaloz * r6829 /trunk/target/linux/ixp4xx-2.6/patches/ (200-gateway_7001.patch 210-gateway_7001_mac_setup.patch): clean up Gateway 7001 support Apr 02 14:43:34 kaloz * r6830 /trunk/target/linux/ixp4xx-2.6/ (3 files in 2 dirs): Netgear WG302 v2 / WAG302 v2 support Apr 02 14:44:21 kaloz * r6831 /trunk/target/linux/ixp4xx-2.6/patches/400-pronghorn_metro.patch: add basic support for the ADI Engineering Pronghorn Metro Apr 02 14:45:02 kaloz * r6832 /trunk/scripts/arm-magic.sh: enable zImage generation for the new boards Apr 02 14:45:39 kaloz * r6833 /trunk/target/linux/ixp4xx-2.6/patches/ (2 files): sync patch naming schema with the other boards Apr 02 15:32:45 anybody: ping Apr 02 15:32:52 net/sched/sch_esfq.c:163: error: 'struct sk_buff' has no member named 'nfmark' Apr 02 15:33:07 two more such errors in the same file Apr 02 15:33:24 when compiling 2.6 kernel Apr 02 15:34:11 is this a target issue, or in generic-2.6? Apr 02 15:35:16 [mbm] ? [florian] ? Apr 02 15:35:52 it looks more like a kernel config issue Apr 02 15:36:09 in what way? Apr 02 15:36:38 CONFIG_NETFILTER is not defined Apr 02 15:36:54 that module is pulled in by generic-2.6/config-template Apr 02 15:37:28 i guess there is a dependency; thanx Apr 02 16:05:41 florian * r6834 /trunk/tools/firmware-utils/src/airlink.c: Make airlink.c more endianness robus (sn9) Apr 02 16:07:31 what does G mean in "svn up" now? Apr 02 16:07:41 florian * r6835 /trunk/target/linux/rdc-2.6/files/drivers/mtd/maps/rdc3210.c: Use sn9 mtd driver Apr 02 16:07:57 florian * r6836 /trunk/target/linux/rb532-2.6/patches/500-Nand.patch: Add missing \n Apr 02 16:09:17 florian * r6837 /trunk/ (6 files in 5 dirs): Use find -exec instead of xargs Apr 02 16:12:02 <[florian]> sn9__: I don't exactly remember Apr 02 16:12:27 must be that the new svn version is identical to the changed local Apr 02 16:12:35 <[florian]> something like that, yes Apr 02 16:12:50 you screwed up the driver Apr 02 16:12:57 you changed = to == Apr 02 16:13:02 florian * r6838 /trunk/target/linux/rdc-2.6/Makefile: Add squashfs to rdc, add nvram partition as jffs2 partition (sn9) Apr 02 16:13:35 <[florian]> sn9__: ah, you should have told me, because with gcc4, it should be done in two lines : ref = func() Apr 02 16:13:38 <[florian]> if (ret) ... Apr 02 16:13:55 no other way? Apr 02 16:14:12 <[florian]> the other way generates a warning, and I do not like warnings Apr 02 16:14:15 if ((ret = func()))... Apr 02 16:14:28 ejka: that works in 4? Apr 02 16:14:42 it should Apr 02 16:14:48 i will generate a new patch Apr 02 16:15:01 i do not like the #if 0, either Apr 02 16:15:23 <[florian]> sn9__: well, the symbol was not defined Apr 02 16:15:56 undefined symbols in #if always evaluate to 0; you should know that Apr 02 16:16:18 <[florian]> ok Apr 02 16:29:05 you also pulled out an #else Apr 02 16:32:29 <[florian]> yeah Apr 02 16:33:15 i will resubmit the patch if there are no build errors Apr 02 17:00:14 florian * r6839 /trunk/target/linux/rb1xx-2.6/ (68 files in 23 dirs): Add preliminary RouterBoard RB1xx support Apr 02 17:01:48 <[florian]> sn9__: your flash map driver is crashing here Apr 02 17:02:04 hold on Apr 02 17:02:54 <[florian]> ok :-) Apr 02 17:10:51 florian * r6840 /trunk/tools/firmware-utils/ (Makefile src/csysimg.h src/mkcsysimg.c): Add ADM5120 based edimax image generation utilities Apr 02 17:14:25 florian * r6841 /trunk/target/linux/adm5120-2.6/image/lzma-loader/ (6 files in 2 dirs): Entry point is now configurable via a Makefile variable. Entry point is detected automatically (Gabor Juhos) Apr 02 17:15:13 florian * r6842 /trunk/target/linux/adm5120-2.6/image/Makefile: Generate images for edimax devices (Gabor Juhos) Apr 02 17:40:50 <[florian]> sn9__: if you have an update for me to test, just ping, off to eat Apr 02 17:41:13 i am testing myself right now Apr 02 17:42:34 any particular reason you chose CONFIG_X86_PC instead of CONFIG_X86_RDC in the kernel config? Apr 02 17:49:11 <[florian]> no, kernel_menuconfig merging made that for me, I put back CONFIG_X86_RDC back Apr 02 18:11:40 [florian]: i notice that you cite others by name in the commit log, but not me (/who sn9) Apr 02 18:21:58 wow, gcc4 is much, much slower than gcc3... Apr 02 18:35:48 <[mbm]> thought compilers were rated on the code they produced, not the speed that the compiler ran Apr 02 18:36:26 are you saying gcc4 produces much better code than gcc3? Apr 02 18:37:30 gcc4 binary was not particularly better than gcc3 binary output when I was running my tests on it Apr 02 18:37:37 infact it was significantly worse in a number of areas Apr 02 18:39:50 that is exactly why i don't buy the justtification [mbm] just proposed Apr 02 18:39:58 s/tt/t/ Apr 02 18:43:32 <[florian]> sn9__: I think I did mention you quite few times Apr 02 18:44:43 [florian]: yes, but i seem to be the only one in the logs cited by anything other than full name Apr 02 18:44:48 <[mbm]> I'm not saying gcc4 is better or worse Apr 02 18:45:17 <[mbm]> I'm just saying that judging it based on the time it takes to run is somewhat insignifigant vs the actual code produced Apr 02 18:45:40 <[florian]> sn9__: if only your whois would return a real name, I could put you in the logs Apr 02 18:45:54 it does if you leave off the __ Apr 02 18:50:40 <[florian]> ok Apr 02 18:51:42 there is also the nickserv info command Apr 02 19:11:35 i know it makes no sense to include zydas drivers for rdc, but they should not make the build fail Apr 02 19:12:06 zd1205 compilation spews a screenful of errors Apr 02 19:13:02 well, warnings, but one error: Apr 02 19:13:42 build_i386/linux-2.6-rdc/zd1211-driver-r83/src/zd1205.c:4777: error: 'struct driver_stats' has no member named 'iw_stats' Apr 02 19:21:12 nbd? {Nico}? Apr 02 19:31:27 kaloz * r6843 /trunk/target/linux/ixp4xx-2.6/ (config/default patches/400-pronghorn_metro.patch): cleanup Apr 02 19:35:20 <[florian]> sn9__: have you enabled wirelss support ? Apr 02 19:36:08 i start off with the kitchen sink and disable stuff. it's a more thorough test Apr 02 19:36:21 it appears zd1211 is only for 2.4 Apr 02 19:36:35 it should be marked as such Apr 02 19:36:47 <[florian]> why care for now ? Apr 02 19:37:56 same reason as the xargs: best to fix problems as they are discovered whenever possible, instead of grouping them Apr 02 19:49:26 [florian]: build error in the flash map driver, introduced by you... Apr 02 19:50:24 <[florian]> sn9__: of course Apr 02 19:50:44 <[florian]> sn9__: let me remind you that what you sent me was not 2.6 compatible, do you remember ? Apr 02 19:51:48 then neither was the driver as you had it in 2.6 at the beginning. doesn't matter now -- will fix Apr 02 20:00:27 <[florian]> sn9__: compiles fine here Apr 02 20:38:17 [florian]: you were right about endianness; fix forthcoming Apr 02 21:06:06 [florian]: ping Apr 02 21:07:58 besides what was already encountered, there is yet another 2.4/2.6 difference that i'm a bit unclear on how to overcome Apr 02 21:08:48 the driver uses flashdrv_get_size() to determine how much flash there is. it is apparently there under 2.4, and not under 2.6 Apr 02 21:09:11 ld does not like that Apr 02 21:10:11 if you haven't run into this in your builds, i guess you're still using the static mapping Apr 02 21:17:51 in the 2.4 target, that and two other functions used were in drivers/mtd/chips/flashdrv.c, part of target/linux/rdc-2.4/patches/001-rdc3210_flash_map.patch Apr 02 21:19:06 <[mbm]> sn9__: might work better if you just write all your comments in a ticket and attach the patch Apr 02 21:19:35 i don't run into the problems all at once Apr 02 21:20:04 <[mbm]> what does that have to do with it? Apr 02 21:20:35 [florian]: for now, please redownload big.patch, apply to current svn, and investigate Apr 02 21:20:56 [mbm]: you said "all" comments Apr 02 21:21:12 <[mbm]> you're allowed to comment more than once Apr 02 21:21:41 i thought of the tickets as a more long-term thing Apr 02 21:21:53 <[mbm]> just means that instead of scrolling through an irc log I can just look at the status of the ticket Apr 02 21:22:42 some of these issues are very quickly addressable, without requiring wide attention Apr 02 21:24:01 <[mbm]> you mean "woul be quickly addressable" .. florian doesn't appear to be around Apr 02 21:25:51 he was an hour and a half ago, and it is not yet the time at which he was previously announcing that he was going to sleep, nor has he announced so today Apr 02 21:26:09 <[mbm]> lmao Apr 02 21:26:24 <[mbm]> didn't know you were stalking him Apr 02 21:27:03 also: [10:40:41] [florian] sn9__: if you have an update for me to test, just ping, off to eat Apr 02 21:27:27 just noticing patterns... Apr 02 21:44:14 florian * r6844 /trunk/target/linux/adm5120-2.6/files/ (2 files in 2 dirs): Add the infrastructure to detect RouterBOOT and use USB **** ENDING LOGGING AT Tue Apr 03 02:59:56 2007