**** BEGIN LOGGING AT Mon Oct 22 02:59:56 2007 Oct 22 05:57:05 hi, where is the binary for Kamikaze 7.09 for NSLU2 ? Oct 22 12:52:09 hi, all! Oct 22 12:52:15 ejka, ping Oct 22 13:02:49 ejka, answer please if you're here, I need some hint :( Oct 22 13:03:04 hello Oct 22 13:03:10 jonasg[1]? Oct 22 13:03:20 [florian], do you know something about lexra toolchains? Oct 22 13:03:32 <[florian]> slapin: hi Oct 22 13:03:42 [florian], hi! Oct 22 13:03:51 <[florian]> slapin: I know the realtek one works, as well as the gcc patched with non-mips Oct 22 13:04:32 [florian], I have problem with nonmips.sf.net toolchain emitting lwl/lwr instructions :( Oct 22 13:05:05 slapin: ?/ Oct 22 13:05:55 <[florian]> slapin: their toolchain work for me, what's wrong ? Oct 22 13:07:03 [florian]: Did you ever get hold of AR7 datasheets? Oct 22 13:07:32 <[florian]> farnz: I think so Oct 22 13:07:48 [florian], http://ossfans.org/DI524UP/death.txt Oct 22 13:08:03 slapin: 403. Oct 22 13:08:06 <[florian]> slapin: check your web perms Oct 22 13:09:09 ejka, any news regarding realtek? Oct 22 13:09:13 which part of the AR7 datasheet do you need? I have a TNETV data sheet, which is 80% similar... Oct 22 13:09:14 [florian], try again Oct 22 13:09:29 [florian]: I'm trying to work out whether the IP175A in my D-Link DSL-G604T is wired to the AR7's MDIO bus, and if it is, whether cpmac.c will let me get at it to reprogram the chip. Oct 22 13:09:55 farnz: it does, and it is exported as physomething Oct 22 13:10:06 ejka, is it possible to build realtek's firmware using gcc-4.1? Oct 22 13:10:09 farnz: but you don't really want to reprogram it Oct 22 13:10:19 slapin: never tried Oct 22 13:10:27 <[florian]> slapin: did you compile your kernel with the nonmips toolchain ? Oct 22 13:10:38 ejka: I can't persuade mii-tool to let me inspect the registers on the IP175A. Oct 22 13:10:48 ejka, what toolchain are you using to build original firmware Oct 22 13:10:57 ejka: I'd just like to split the switch into two VLANs, if I can. Oct 22 13:11:04 [florian], yes, I did Oct 22 13:11:06 farnz: you can't do it with mii-tool Oct 22 13:11:17 farnz: and it only has port-based vlans Oct 22 13:11:22 <[florian]> slapin: sounds weird Oct 22 13:11:25 slapin: I didn't Oct 22 13:11:41 BTW: on what software version from TI is the AR7 code in openwrt based? Oct 22 13:11:51 slapin: and btw, there was a working binary toolchain at ftp.dlink.com Oct 22 13:12:19 ejka: The datasheet I have for the IP175A says it can do tagged VLANs. Oct 22 13:12:22 I got some "original" TI linux software, PSP 7.4. something Oct 22 13:12:40 farnz: port-based, meaning all traffic will go through cpu Oct 22 13:13:03 ejka: That's fine for my usage. Oct 22 13:13:17 [florian], ejka: what is weird is that all the rest works, so it seems that it is due to bug in toolchain which happens only accidentally only in this place. Kernel boots and works, but USB disks doesn't work. Oct 22 13:13:32 farnz: then add support for it to package/switch Oct 22 13:14:10 blogic is working on a new switch api that integrates with the phy layer Oct 22 13:14:21 farnz: you basically need to phy_connect to exported phy Oct 22 13:14:27 ejka, something about closed toolchains looks terrific... Oct 22 13:15:08 ejka, I have enough binary crap trashing my hard drive.... Oct 22 13:15:35 jonasg[1]: the openwrt code is a rewrite. the TI code is utter crap Oct 22 13:15:46 ejka, any ideas on fix? Oct 22 13:15:55 slapin: no Oct 22 13:16:15 ejka: Thanks for the pointers. Oct 22 13:16:35 nbd: yes, that is true. however, the cpmac.c that I have from TI seems to be "newer" than that in OpenWrt Oct 22 13:16:51 jonasg[1]: The OpenWRT cpmac.c is a complete rewrite, mostly by ejka Oct 22 13:17:04 <[florian]> slapin: I guess you have read this page : http://rtl8181.sourceforge.net/rtl8181_toolchain.php Oct 22 13:18:15 [florian], yes, and I built from nonmips.sf.net sources :( Oct 22 13:20:08 jonasg[1]: I have something like 20 different cpmac sources here :) All of them are utter crap Oct 22 13:20:14 [florian], I used nonmips version because it contains exact versions which are used in realtek firmware (gcc-3.3.3, uclibc-0.9.26) Oct 22 13:20:36 eijka: ok Oct 22 13:21:36 <[florian]> slapin: I see, this toolchain is supposed not to emit any patented instructions Oct 22 13:22:16 [florian], I've tested, their patch is applied, so source is changed accordingly. I used their build system. Oct 22 13:22:28 ejka: on the other hand, on TNETV devices also the switch is supported and it seems to perform quite well. Just for your information: I have PSP 7.4.0.18. Oct 22 13:22:36 [florian], so there's something they forgot to patch. Oct 22 13:23:09 jonasg[1]: I am much more interested in hardware datasheets Oct 22 13:24:07 [florian], I'm mostly wondering that this is only, and non-fatal (while annoying) bug. Oct 22 13:24:37 <[florian]> slapin: well, yes, or it might be this file that did not get recompiled properly Oct 22 13:24:40 ejka: ok, I have: TNETV1060, TNETC4600, TNETD7300, TNETV1061-PETRA Oct 22 13:25:16 ejka: and some more, but those unfortunately only under NDA :( Oct 22 13:25:45 jonasg[1]: I have a TNETD7300; hardware datasheets would interest me for that, and I'm happy to host them if needed. Oct 22 13:25:55 [florian], I removed all generated files before build and checked timestamps. It seems definitely a toolchain bug. Oct 22 13:26:08 ejka: TNETD7300 = AR7 Oct 22 13:26:16 <[florian]> slapin: I guess so, this toolchain used to work for me, though I did not use usb stuff Oct 22 13:26:55 now I need some doc for backends so I could check which code emits the instruction. Oct 22 13:27:40 <[florian]> yes, you need to find a sample file wich would trigger lwl to be emitted for instance Oct 22 13:27:45 jonasg[1]: tnetd7300 and tnetv1060 are the most interesting for me Oct 22 13:28:13 farnz: have you tried to reach ago? He has some (leaked) data sheets IIRC Oct 22 13:28:22 loswillios: Tried and failed. Oct 22 13:29:38 I quickly looked into the datasheets: they are all preliminary (and as from TI, they are crap, nothing to be compared to freescale....) Oct 22 13:29:45 but they might help... Oct 22 13:32:26 [florian], it reproduces only in msdos.c, msdos_partition function. Oct 22 13:40:14 florian * r9399 /packages/net/ (crowdcontrol/Makefile mvprelay/Makefile updatedd/Makefile): Fix more download URLs (#2551) Oct 22 13:45:59 florian * r9400 /packages/net/radsecproxy/ (. Makefile files/ files/radsecproxy.init): Add radsecproxy (#2571) Oct 22 13:59:19 [florian], do you know of some doc and tool for gcc backend debugging? Oct 22 14:04:23 [florian], ejka: what do you think - if I setup public git repository with dlink firmware, collecting as much source stuff as possible, like some central location to gather data, will it be legal and/or useful? Oct 22 14:20:03 ejka * r9401 /trunk/ (3 files in 3 dirs): ar7: add eva image generation (thanks Axel Gembe) Oct 22 14:30:21 yay Oct 22 16:17:11 florian * r9402 /trunk/target/linux/au1000/ (config-2.6.23 image/Makefile): Resync kernel config, make kernel.{flash,ram}.srec available Oct 22 20:56:37 florian * r9403 /trunk/target/linux/adm5120/ (11 files in 6 dirs): Add preliminary support for the Routerboard 153 CF slot (#2550) Oct 22 22:40:29 kaloz * r9404 /trunk/package/madwifi/patches/320-karma.patch: remove madwifi karma patch until it's fixed Oct 23 00:55:34 * jk- patchbombs Oct 23 00:59:03 no way, openwrt on the ps3! Oct 23 01:03:51 :D Oct 23 01:05:31 jk-: what are you planning to do with it? Oct 23 01:05:54 or is it just "because I can"? ;-) Oct 23 01:06:07 it makes it a lot easier to have a ps3 with no monitor Oct 23 01:06:25 since we can ssh to the bootloader Oct 23 01:07:19 I see. so it's better for guys hacking the ps3? Oct 23 01:07:39 no, not hacking Oct 23 01:08:53 the current bootloader (still linux-based, but not openwrt) only has a local console Oct 23 01:09:19 so you need to be physically in front of the machine to do anything other than 'boot the default kernel' Oct 23 01:09:30 this allows us just to leave the PS3s in the machine room :D Oct 23 01:09:40 heh cool Oct 23 01:10:10 jk-: building ps3 clusters? Oct 23 01:10:31 ejka: I'm not, but i'm sure some people are interested Oct 23 01:10:39 it's amazing on what variety of devices openwrt already runs Oct 23 01:11:08 it has so much potential Oct 23 02:40:59 slow day today **** ENDING LOGGING AT Tue Oct 23 02:59:57 2007