**** BEGIN LOGGING AT Thu Mar 05 02:59:56 2009 Mar 05 05:11:58 i want to pass "--target=mips64-linux" to gcc/binutils, however I don't wan to change the existing mips-openwrt-linux-uclibc- prefix, how can I do that? Mar 05 05:12:54 there are some hard-code to check for mips-openwrt-linux-uclibc-readelf while my binary is mips64-linux-readelf Mar 05 05:13:12 the current toolchain stuff is hard-coded for 32-bit Mar 05 11:48:08 FYI, near the top of the Forum page it says Latest: 7.09 Mar 05 12:48:19 <_trine> RoundSparrow: have you ever thought about changing the flash cip on the trendnet routers to a larger one? from an hardware point of view it would be relatively easy but but require the boot loader to be modified Mar 05 12:48:26 <_trine> chip* Mar 05 12:49:44 juhosg * r14749 /trunk/package/mac80211/ (33 files in 2 dirs): [package] mac80211: update compat-wireless to 2009-03-04, and refresh patches Mar 05 13:57:26 _trine: no, not really. I think a mmc or USB would be more productive and produce much more usable Mar 05 14:22:59 hmmm, I started a build last night and it was waiting for me to answer some questions... with this warning .config:2030:warning: symbol value 'm' invalid for SND_PPC Mar 05 14:23:17 * ARM sound devices Mar 05 14:23:17 * Mar 05 14:23:17 ARM sound devices (SND_ARM) [Y/n/?] (NEW) Mar 05 14:24:32 sound ? Mar 05 15:05:02 <_trine> RoundSparrow: but would it not be much harder to implement USB on them than it would to just change the flash chip? Mar 05 15:06:21 _trine: I just have no desire to add a few MB when i can add GB Mar 05 15:06:34 <_trine> RoundSparrow: ok Mar 05 15:06:41 _trine: I have had no limitations with 4MB on OpenWrt that I see. Luci, IPV6, etc all fits fine Mar 05 15:06:53 the key is to install on original flash and not install as packages Mar 05 15:07:14 <_trine> yes I understand that Mar 05 15:08:49 <_trine> but that in itself does not give you very much more Mar 05 15:09:47 <_trine> it occurred to me that 8M would be all that most people wanted and a flash chip change would be the ideal solution Mar 05 15:10:12 <_trine> for me that would be true Mar 05 15:10:38 _trine: I personally think surface mount re-soldering is something only a very few dedicated users would do ;) I don't want to discourage your personal efforts, but I don't see it as time economical for most people Mar 05 15:11:20 _trine: my overall goal is to show the router vendors picking new chipsets that here is a well supported one and maybe some of them will get a clue and add the extra $0.75 in parts to come with 16MB base Mar 05 15:11:38 <_trine> well the flash chip is one of the easier components to change much easier than the rams chips which I have already done Mar 05 15:11:51 _trine: already d-link switched away from this platform to RALink which seems an industry-wide trend on affordable models Mar 05 15:12:27 <_trine> I don't follow that last comment Mar 05 15:12:57 <_trine> RALink being the wifi component ? Mar 05 15:13:04 RoundSparrow: that's a waste of effort. Mar 05 15:13:09 RALInk is an atheros competitor Mar 05 15:14:39 <_trine> I just thought that a small change in the loader could allow those that wanted to to change the flash to a bigger one. In the belief that the work to do this would be minimal Mar 05 15:15:02 <_trine> just a thought really Mar 05 15:15:04 _trine: it is, it has been done on other u-boot systems Mar 05 15:15:36 _trine: you asked about my personal thoughts. I don't discourage your efforts Mar 05 15:16:03 <_trine> although I can easily do the change of chip I don't have the expertise or know how to change the code Mar 05 15:16:59 <_trine> I guess that's why we all collaborate our individual skills Mar 05 15:17:00 it should be nothing more than a hex edit of the binary u-boot and reflash. Bartman007 I believe is using this approach on the WNR2000 to remove a checksum... but not sure. I think once that has been demonstrated, shouldn't be too hard to do this Mar 05 15:17:10 we already have one model with 8MB of flash to dump the u-boot and compare Mar 05 15:17:33 and we have the source code for two models with u-boot Mar 05 15:17:54 <_trine> I also think this would be a good idea for the fonera 2100's too Mar 05 15:18:06 <_trine> which uses redboot Mar 05 15:18:27 wonder when Fonera will ship their AP81 ;) Mar 05 15:19:24 <_trine> the foneras have been one of my main toys since they came out Mar 05 15:19:42 _trine: i think i've seen people add more flash before. it would be most useful for models with 2mb flash imo. there are plenty of them Mar 05 15:20:10 <_trine> danage: thas a great idea for the wrt's Mar 05 15:20:30 <_trine> the version 5 and above Mar 05 15:20:39 there are many many more that could be supported if it weren't for the small flash Mar 05 15:20:51 <_trine> danage: exactly Mar 05 15:21:03 we continue to give economic incentive to vendors to continue to screw us on hardware. Mar 05 15:21:16 <_trine> it's one area that has been neglected in my view Mar 05 15:21:28 it's not trivial smd solderin isnt Mar 05 15:21:33 they aren't selling hardware for charity, economic selection of the models that are more LInux friendly should be an objective of the community Mar 05 15:22:11 you keep forgetting they arent making these devices for the linux community. if their firmware is only 2mb, they will put 2mb chip Mar 05 15:22:41 <_trine> I think some of the efforts of some manufactures have driven along designs to stop the implementation of better firmware Mar 05 15:23:02 The WRT54GL has been an economic bad joke on the LInux community. Only putting 16MB. They had a factory still pumping out 32MB/8MB models for T-Mobile. so the hardware support and tooling was all done. Maybe $1 total in parts difference for them Mar 05 15:23:16 while i doubt they had that in mind, there is certainly that effect, yes Mar 05 15:24:02 <_trine> so they can sell more expensive models Mar 05 15:24:25 _trine: only one they sell I know of with Tux on the box Mar 05 15:24:48 <_trine> danage: I would have thought that if they really wanted to sell lots of routers they should have in mind the attractiveness of what they are producing to theor customers Mar 05 15:25:10 I mean come on, what would it have cost them to integrate a USB serial on the 610N or 310N... $2 extra, then sell a special "GL" edition and allow root via the serial Mar 05 15:25:41 <_trine> it beggers belief that they continue to produce hardware that year on year is less attractive to the customer base Mar 05 15:25:52 the customer base of reflashing linux freaks is marginally small. plus, they will keep updating their software and will only rarely need to buy a new router. if customers use stock firmware, the vendor can just stop updating the firmware and force the customers to buy a newer model Mar 05 15:26:08 that's their real incentive Mar 05 15:26:16 <_trine> look whats just happened to the price and worth of the Trendnet routers Mar 05 15:26:31 the economic disposable / planned obsolescence. Mar 05 15:26:37 of course it is Mar 05 15:26:49 after all, the world is a capitalistic one Mar 05 15:27:02 Well economic rules have changed, waste may be on the way out, I sure know there won't be 45 brands of routers with 180 models Mar 05 15:27:10 <_trine> a few months ago they could not give them away now when they realize that others have figured a way to improve them the price has soured Mar 05 15:27:11 consumers are meant to consume products! not use them! Mar 05 15:27:25 we shoudln't be on -devel for this :) Mar 05 15:27:30 they are supposed to pay and then shut up Mar 05 15:28:04 <_trine> RoundSparrow: this is a devel. issue it's a hardware dev issue Mar 05 15:28:07 i'll end by saying that N is a key transition period, as was 64bit for Linux and Microsoft Mar 05 15:28:54 larkw: hello! Mar 05 15:29:08 RoundSparrow: hello :) Mar 05 15:30:02 larkw: so Juhosg checked in new ath9k stuff, I'll probably do another binary today Mar 05 15:31:58 RoundSparrow: great. I haven't test last build you did. Yesterday night, I was trying integrating openwrt tree into linus tree's tip. Still have problem to compile Mar 05 15:32:42 I will concentrate on kernel test. Mar 05 15:35:32 RoundSparrow: I will hang on here tonight. When you finish your new build, drop me a note. But now I must do some kernel tweaking on ppc for one of my friends. Mar 05 15:43:31 anyone knows if uclibc works with 64bit mips? Mar 05 15:44:04 i'm trying to compile uclibc with 64-bit gcc and it complains on ABI compatiblity Mar 05 15:44:30 mips-openwrt-linux-uclibc-ld: ldso/ldso/ld-uClibc_so.a(ldso.oS): ABI is incompatible with that of the selected emulation Mar 05 15:46:20 looks like mips64 was integrated in late 2006 http://www.nabble.com/status-of-0.9.29-td6653167.html Mar 05 15:48:01 thanks. struggling to get a mips64 box working under openwrt(dlink 2590) Mar 05 15:48:19 the kernel has to be 64-bit,. rootfs can be either 32-bit or 64-bit Mar 05 15:48:46 as openwrt uses the same gcc to compile both kernel and rootfs, i'm trying to go 64-bit all the way Mar 05 15:51:09 3 weeks old: https://dev.openwrt.org/ticket/4603 Mar 05 15:54:14 RoundSparrow: thanks. that's very helpful, checking it Mar 05 15:54:53 dev.openwrt.org is really slow here Mar 05 15:55:02 everywhere Mar 05 15:55:21 must be behind a dial-up link Mar 05 15:56:12 thl * r14750 /trunk/target/linux/ifxmips/files/arch/mips/ifxmips/prom.c: [ifxmips] make selection of ASC for prom working Mar 05 16:10:55 someone could have a look at my failing build log ? openssl fails in a wierd way : http://oomz.net/openssl.log Mar 05 16:35:10 BusError: This looks like you are linking as shared make -f ../Makefile.shared -e \ but in my Makefile i see CFLAGS= -DMONOLITH $(INCLUDES) $(CFLAG) Mar 05 16:40:33 trunk/build_dir/target-armeb_uClibc-0.9.30/openssl-0.9.8j/appsMakefile :also has CFLAG= -static Mar 05 16:43:31 johnrw, I;m also having trouble with openssl Mar 05 16:44:31 BusError: I didn't have any trouble with openssl... Mar 05 16:45:05 it built... even though I didn't specify it in my image. Mar 05 16:45:22 dependency ? Mar 05 16:45:42 lemme see what menuconfig says Mar 05 16:48:30 I dunno... this buildroot got messed up a little... and so... it was selected as a module [M] there... but it built. menuconfig says : Selected by: PACKAGE_openvpn-easy-rsa && PACKAGE_openvpn Mar 05 16:52:57 hrm what does xserver-kdrive needs with openssl :/ Mar 05 16:56:24 -> Xorg ? Mar 05 16:56:24 ? -> server ? Mar 05 16:56:24 ? Selects: PACKAGE_libopenssl && Mar 05 16:57:09 That was info for... Symbol: PACKAGE_xserver-kdrive-xvesa [=n] ; among others... Mar 05 17:39:29 juhosg: ping Mar 05 17:45:31 how to generate target/linux/ar71xx/config-2.6.28, is it a standard kernel config? Mar 05 17:46:13 I am wondering... when you added the dsa support to ar71xx you seem to have done so with C style code... but when I look at my usr8200-setup.c.. it is mostly done as static 'structures' that must be glued to some code somewhere else... Mar 05 17:51:37 in mine... trunk/target/linux/ar71xx/config-2.6.28 exists... Mar 05 17:58:10 ? alot of unknown magic happens with those structures... trunk/build_dir/linux-ixp4xx_generic/linux-2.6.28.7/arch/arm/mach-ixp4xx/usr8200-setup.c Mar 05 17:59:23 so where is the code that winds up using those structures? Mar 05 18:01:25 johnrw: for ARM specific DSA examples, you should look into the arch/arm/mach-orion5x/common.c and/or arch/arm/mach-kirkwood/common.c Mar 05 18:03:18 juhosg: you know I am waiting for that to hit the mainline kernel... and now MateIn4 is on vacation for a week... so thanks... ! Mar 05 18:20:05 johnrw: i'm sorry, but currently i have enough fun without any ixp stuff Mar 05 18:20:54 juhosg: Oh God Forbid! I was not asking you to do it. Mar 05 18:21:54 *cough* ralink SoC *cough* Mar 05 18:22:15 hey, you keep juhosg focused on ath9k, he is the universe's best hope for an entirely free wireless driver that works well ;) Mar 05 18:23:15 for the moment, it is... Mar 05 18:23:17 sn9_: which one? Mar 05 18:23:24 mips Mar 05 18:23:45 but the arm one needs support, too Mar 05 18:24:00 I just like to stare at the code... wishing I could... but inevitably wind up doing some background work for those that can like noz or matein4 Mar 05 18:25:39 sn9_: we don't have anything useful yet Mar 05 18:26:39 i know Mar 05 18:27:03 sn9_: I wanted to say thank you for giving me the impetus to try out openocd from the trunk (again), although I never tried to construct anything more, than the most basic openocd.cfg, it did work! So... Thank You... Mar 05 18:27:26 yw Mar 05 18:28:28 johnrw: did you try to poke Kaloz already? Mar 05 18:30:45 build/..../openssl-0.98/apps/Makefile id borken if you use ccache the CC-$${CC} should have quotes Mar 05 18:31:34 sn9_: I wound up using a gdb_force_hardware_breakpoint option, in openocd.cfg... and lo and behold! All of a sudden, I could use gdb. So I made a patch to build the gui interface - Insight for it... but it hasn't been put in yet. juhosg: nah... I guess I'll just tell MateIn4 or noz... what you told me... one of them will be able to do it... when they have time. Mar 05 18:31:58 great Mar 05 18:33:34 sn9_: It is in ticket 4701 if you wanna have a look... All I know... is with Insight... I could tell it to show me mixed assembly and C source very easilly... Mar 05 18:34:35 i could tell gdb to do the same Mar 05 18:36:38 I know Insight can only tell gdb to do it... and that means gdb can do it also... but man... I probably should have just printed the pdf cheatsheet I was using... but since I didn't... It was kind of awkward. Mar 05 18:38:12 I do like using Insight... it shows the register changes... etc... but that is all besides the point... as I was just trying to say Thanks... for getting me to try out openocd again. Mar 05 19:19:35 i guess developers in Europe are having dinner now Mar 05 19:20:00 it's pretty late in europe Mar 05 19:20:27 7:20PM there? i thought it's 6 hours early than CST in US Mar 05 19:20:39 that's just england Mar 05 19:20:47 right... Mar 05 19:20:52 21:20 here Mar 05 19:21:23 which country Mar 05 19:21:47 finland Mar 05 19:21:59 (you should even spot that from my host, it's not masked :P) Mar 05 19:22:21 lazy...dizzy after lunch Mar 05 19:24:51 maligor: do you know how to generate the config file under target/linux/board-name Mar 05 19:25:14 it's not a fully kernel config, neither an openwrt config, there must be some scripts to do that Mar 05 19:25:23 s/fully/full/ Mar 05 19:26:01 i'm adding a mips64 to openwrt Mar 05 19:27:24 there's embedded devices with mips64? Mar 05 19:27:34 cavium that is Mar 05 19:27:39 but no, I don't really understand the board configuration either Mar 05 19:28:56 mips64 can't run mips? Mar 05 19:29:17 since I imagine openwrt won't really gain much from 64bit instructions :P Mar 05 19:29:56 the kernel is 64-bit so it requires a 64-bit gcc to compile, but the 32 bit applications should be able to run once the 64-bit kernel is up Mar 05 19:30:26 meanwhile, i hope this downturn will not kill cavium for good Mar 05 20:48:20 hrms. common someone must have built curl/kdrive and thus openssl recently ? here it fails. it just can't work, and I even after fixing the double quotes error I get link errors and am getting nowhere :/ Mar 05 20:57:51 BusError: talk to larsc or mirko Mar 05 21:46:28 hauke * r14751 /trunk/ (3 files in 2 dirs): [kernel] Some symbols were missing or wrong. Tested with x86. Mar 06 01:10:48 florian * r14752 /packages/lang/luasec/Makefile: [package] suppress whitespace in luasec makefile (#4739) Mar 06 01:11:47 florian * r14753 /trunk/package/spidev_test/ (. Makefile): [package] add spidev_test utility from the kernel documentation Mar 06 01:13:12 florian * r14754 /trunk/target/linux/brcm63xx/files/include/asm-mips/mach-bcm63xx/bcm63xx_regs.h: [brcm63xx] fix SPI register definitions Mar 06 01:15:03 florian * r14755 /trunk/target/linux/brcm63xx/files/ (5 files in 2 dirs): [brcm63xx] prepare for SPI controller driver Mar 06 01:52:15 hey guys Mar 06 01:52:19 anyone ever worked with libpcre ? Mar 06 02:28:25 anyone ever worked with libpcre ? **** ENDING LOGGING AT Fri Mar 06 02:59:57 2009