**** BEGIN LOGGING AT Fri Jul 27 02:59:56 2007 Jul 27 03:07:49 nico * r8189 /trunk/ (12 files in 8 dirs): update ar7 to 2.6.22.1 (thanks to Matteo Croce for his great help) Jul 27 03:48:49 nico * r8190 /trunk/target/linux/ar7-2.6/files/include/asm-mips/ar7/ (ar7.h prom.h vlynq.h): more cosmetic fixes for ar7 Jul 27 08:09:01 rwhitby * r8191 /trunk/target/linux/generic-2.6/config-template: Add missing config symbols that stop a default 2.6.21.6 built - set them all to default values so nothing changes for anyone Jul 27 08:38:43 <[florian]> sn9: rebuilt using your changes, seems not to boot, I will have several tries Jul 27 08:52:15 florian * r8192 /packages/utils/fortune-mod/ (Makefile patches/ patches/01-strfile-64bit-fix): Fix fortune-mod compilation on 64bits hosts (#2136) Jul 27 12:59:36 rwhitby * r8193 /trunk/target/linux/generic-2.6/config-template: Removed the obsolete MTD_SPLIT_ROOTFS config symbol. Jul 27 15:49:55 pavlov * r8194 /packages/lang/python/Makefile: add readline support for python Jul 27 15:56:40 pavlov * r8195 /packages/libs/postgresql/Makefile: add readline support to postgres as well Jul 27 15:58:56 juhosg * r8196 /trunk/target/linux/adm5120-2.6/patches-2.6.22/ (14 files): [adm5120] rename patches to follow our naming conventions Jul 27 16:03:33 pavlov * r8197 /packages/libs/postgresql/Makefile: upgrade to postgres 8.2.4, adds some crash fixes Jul 27 16:37:46 nbd * r8198 /trunk/docs/wireless.tex: fix a typo Jul 27 16:57:16 juhosg * r8199 /trunk/target/linux/adm5120-2.6/files/drivers/leds/leds-adm5120.c: [adm5120] add LED definitions for the RB133/133C boards, thanks to Christophe Lucas Jul 27 17:22:00 <[florian]> sn9: ping ? Jul 27 17:22:06 ack Jul 27 17:23:25 <[florian]> sn9: so, I tried your changes {Nico} merged this morning, still no luck, you tried several builds, right ? Jul 27 17:23:49 nbd: i'm sorry, you were right. the problem does lie in the lzma after all Jul 27 17:23:49 every now an then, a build works, but other times, it doesn't, EVEN IF THERE WERE NO CHANGES WHATSOEVER Jul 27 17:24:03 <[florian]> ok Jul 27 17:24:23 <[florian]> would be good to find where the problem comes from Jul 27 17:24:39 that would require using gdb over jtag Jul 27 17:24:50 no other way Jul 27 17:25:02 look at the code if you believe otherwise Jul 27 17:30:32 i'm working on the wifi right now Jul 27 17:31:02 for some reason, kzalloc(0x100, GFP_KERNEL); fails Jul 27 17:31:23 i don't know why Jul 27 17:34:10 <[florian]> have you considered trying rt2x00-mac80211 ? Jul 27 17:34:26 <[florian]> could you provide me with a working squashfs image if you have such in near ? Jul 27 17:34:38 that is what fails Jul 27 17:35:02 <[florian]> ah ok, the mac80211 could be updated to a more recent snapshot Jul 27 17:35:28 no Jul 27 17:35:39 it's already too new for the kernel Jul 27 17:35:48 i had to patch that Jul 27 17:39:40 redownload rdc.patch and rdc.config Jul 27 17:40:37 tell me when you finish downloading http://homepage.mac.com/danielg4/openwrt-rdc-2.6-squashfs-ar525w.img so that i may take it down quickly Jul 27 17:41:30 [florian]: ^^ Jul 27 17:43:20 <[florian]> done Jul 27 17:43:30 <[florian]> thanks Jul 27 17:53:00 the r6040 is failing to obtain a dhcp lease Jul 27 17:53:38 could that have something to do with polling? Jul 27 17:55:51 <[florian]> yes Jul 27 17:55:59 <[florian]> seems to be working with static ip here Jul 27 17:56:08 here too Jul 27 17:56:33 but "udhcpc -nh ar525w -i eth1" fails Jul 27 17:59:45 <[florian]> yeah definitively related to polling and NAPI Jul 27 18:01:08 ok, i'll turn off NAPI Jul 27 18:01:28 <[florian]> until we fix it, but this is no priority for now Jul 27 18:01:38 <[florian]> could not we use a bzip2 compression instead of lzma Jul 27 18:01:55 lzma barely compresses enough Jul 27 18:02:15 and it would introduce additional variables Jul 27 18:03:25 for now, commit rdc.patch, but with NAPI turned off Jul 27 18:10:10 [florian]: still reviewing? Jul 27 18:10:43 <[florian]> no, turning off napi Jul 27 18:11:14 <[florian]> also finding your name ;) Jul 27 18:13:35 florian * r8200 /trunk/ (6 files in 6 dirs): Add the 93cx6 patch to generic-2.6, correctly fix dependencies on the rt2x00 drivers (thanks to Daniel Gimpelevich) Jul 27 18:15:52 <[florian]> here we go Jul 27 18:21:55 nbd: are you around? Jul 27 18:25:03 [florian]: do I need annex-a or -b for german dsl? Jul 27 18:25:12 B Jul 27 18:25:41 thanks Jul 27 18:43:06 sn9: yes Jul 27 18:44:10 nbd: have you used that lzma_misc code elsewhere? Jul 27 18:44:27 lzma_misc? Jul 27 18:45:59 nbd: we have a fair number of people wondering whent he full package set for 7.07 will be uploaded, any guesses? Jul 27 18:46:16 no idea Jul 27 18:46:22 i guess i can start building that stuff soon Jul 27 18:46:54 nbd: arch/i386/boot/compressed/lzma_misc.c Jul 27 18:47:35 nbd: you can sync them from the x-wrt repo :-) Jul 27 18:47:44 sn9: some of the code that we use on mips is similar Jul 27 18:47:53 sn9: but i've never really used that particular file anywhere else Jul 27 18:48:15 you said you saw similar problems on generic x86 Jul 27 18:49:14 yes. that patch was in generic-2.6 a while ago Jul 27 18:49:38 when i upgraded the x86-2.6 target, i found that the lzma loader did not work most of the time Jul 27 18:49:48 same on rdc Jul 27 18:50:05 so i moved that patch to the rdc target, because rdc was known to be broken and it was the only target that actually *needed* it Jul 27 18:50:22 but at least on x86-2.6 it can be easily debugged Jul 27 18:50:43 with qemu or the like Jul 27 18:51:04 the killer is "while (1);" Jul 27 18:51:45 there has to be some way to look at the function arg Jul 27 18:55:05 anyway, it seems that the very reason "rdc was known to be broken" was the broken lzma Jul 27 18:55:25 major bad form Jul 27 18:57:12 anyway, if you move the lzma patch back to generic-2.6, might as well move the git-newsetup*.patch along with it Jul 27 19:02:14 if you replace the "while (1);" with assembly code that loads the arg into a register and then halts the cpu, you can examine it in a virtual machine to see the error msg Jul 27 19:03:13 nbd: are you in a position to do that atm? Jul 27 19:07:35 sn9: i don't really have the time to debug this Jul 27 19:08:02 ok, i will see what i can do Jul 27 22:20:25 [florian]: r8200 doesn't boot on the W701V. I tried several things: http://pastebin.ca/636452 Jul 27 22:22:35 [florian]: anything else you want me to try? Jul 27 22:23:19 what does this output: Jul 27 22:23:22 nm build_mipsel/linux/vmlinux | grep 94073138 Jul 27 22:23:26 or this: Jul 27 22:23:26 nm build_mipsel/linux/vmlinux | grep 94073138 Jul 27 22:23:29 err Jul 27 22:23:30 nm build_mipsel/linux/vmlinux | grep 94073138 Jul 27 22:23:36 nm build_mipsel/linux/vmlinux | grep 9407313 Jul 27 22:24:13 both nothing Jul 27 22:24:28 (this is with the last change on pastebin though Jul 27 22:24:47 check the output of nm and find the next symbol to that one Jul 27 22:25:18 94073170 t s_start Jul 27 22:26:32 does that look good? Jul 27 22:27:14 i don't really know ;) there are many s_starts in the kernel Jul 27 22:28:57 Hi. 1st time here. I'm very new to OpenWrt/Linux and interested in hardware hacking. Jul 27 22:29:06 Ago: that's my LOADADDR? (or could be) Jul 27 22:29:39 that's the output of nm build_mipsel/linux/vmlinux | grep 9407313* Jul 27 22:33:17 i have to update my wc to tell you more, mine is still 2.6.21 Jul 27 22:33:35 I have a couple of kernel module dev questions which are *not* OpenWrt-specific. Can anyone recommend a good forum where I can ask? Jul 27 22:34:49 Darkcobra: lkml? or maybe kerneltrap.org Jul 27 22:35:07 Ago: okiā€¦ Jul 27 22:36:00 arghh... full rebuilding and merging a whole lot of stuff :\ grrrr Jul 27 22:37:04 Checking out both now, thanks Loswillios. Jul 27 22:37:16 loswillios, can you send me your vmlinux file ? Jul 27 22:43:15 sure Jul 28 01:12:58 nbd: you'll never believe what's causing lzma to fail Jul 28 01:13:52 after 2.6.19, the upstream kernel began to assume that the last four bytes of the compressed file are the uncompressed length Jul 28 01:14:29 that may be true for gzip, but gets random data for lzma Jul 28 01:15:02 so whether or not a kernel with lzma will boot is completely random Jul 28 01:32:10 sn9: lol Jul 28 01:32:21 sn9: good job tracking that one down ;) Jul 28 01:33:25 i am thinking of just appending the length to the compressed file, but can't quite think how to do that from within a makefile Jul 28 02:44:51 nbd * r8201 /trunk/ (Makefile include/subdir.mk tools/Makefile): initial attempt at cleaning up subdirectory handling. tools/Makefile is now being included instead of recursively called Jul 28 02:53:50 nbd * r8202 /trunk/ (include/subdir.mk tools/Makefile): don't force the install->compile, compile->prepare dependencies in subdir.mk Jul 28 02:54:46 nbd * r8203 /trunk/include/subdir.mk: add missing flag **** ENDING LOGGING AT Sat Jul 28 02:59:56 2007