**** BEGIN LOGGING AT Fri Sep 14 02:59:56 2007
Sep 14 04:47:38 <CIA-12> ejka * r8772 /trunk/target/linux/ar7/files/drivers/char/watchdog/ar7_wdt.c: fix watchdog seek test (#2383)
Sep 14 04:50:50 <CIA-12> ejka * r8773 /trunk/target/linux/ar7/files/arch/mips/ar7/platform.c: add *dma_mask's to cpmacs
Sep 14 05:31:16 <CIA-12> ejka * r8774 /trunk/target/linux/ar7/files/arch/mips/ar7/platform.c: fix previous commit
Sep 14 08:19:52 <CIA-12> florian * r8775 /trunk/package/kernel/modules/network.mk: Package support for IP over ATM
Sep 14 13:01:47 <sn9> {Nico}: http://homepage.mac.com/danielg4/rdc.patch updated
Sep 14 13:27:20 <CIA-12> rwhitby * r8776 /trunk/target/linux/ixp4xx/ (config/ config/default config/profile-FSG3 config-2.6.21): Reinstated the kernel configuration for the FSG3 profile.
Sep 14 14:00:55 <matteo> ejka: good work, wifi finally works
Sep 14 14:01:41 <ejka> matteo: actually, I was fixing bug for cpmac :)
Sep 14 14:02:00 <matteo> ejka: th IPv6 panic?
Sep 14 14:02:05 <matteo> can you reproduce it?
Sep 14 14:02:25 <ejka> hmm. no I can't. though I have quite a lot of changes.
Sep 14 14:04:01 <ejka> matteo: can you try http://uejikov.g-service.ru/cpmac.diff
Sep 14 14:04:04 <ejka> ?
Sep 14 14:04:05 <matteo> sure
Sep 14 14:04:17 <matteo> a moment, the router is in the apartment in front of mine
Sep 14 14:04:22 <matteo> i was testing wifi range
Sep 14 14:04:28 <matteo> i have a patch to force 20 dbm
Sep 14 14:04:35 <ejka> at least ping6 and telnet other ipv6 work here
Sep 14 14:04:49 <matteo> ejka: it panics when forwarding packets
Sep 14 14:04:52 <matteo> not when sending
Sep 14 14:05:12 <matteo> ejka: install radvd, wait for your pc to get an IPv6
Sep 14 14:05:28 <matteo> and when ping6ing from your PC you will get the panic
Sep 14 14:07:03 <matteo> ejka: are you following suggestions from linux-mips mailing lists?
Sep 14 14:07:08 <matteo> i hope that there were useful
Sep 14 14:08:40 <ejka> yes, that too :)
Sep 14 14:08:58 <matteo> good
Sep 14 14:09:11 <matteo> lol, i see the comment about non-standard hash
Sep 14 14:09:20 <matteo> to stop other devs complain ;)
Sep 14 14:10:20 <ejka> well, actually, such thing should be documented
Sep 14 14:10:48 <CIA-12> olli * r8777 /packages/net/updatedd/files/ (updatedd.config updatedd.init): updatedd: be more human-friendly :)
Sep 14 14:14:24 * matteo go to pick up his router
Sep 14 14:14:28 <matteo> *goes
Sep 14 14:18:54 <matteo> ejka: booting
Sep 14 14:19:15 <matteo> ejka: why don't you add a version to your driver?
Sep 14 14:19:19 <matteo> cpmac XXX loaded
Sep 14 14:19:38 <matteo> could be useful when reading old logs
Sep 14 14:19:46 <ejka> I will always forget to increment it :)
Sep 14 14:20:32 <matteo> np, it wasn't fundamental
Sep 14 15:03:53 <matteo> ejka: it works
Sep 14 15:06:27 <matteo> ejka: line like this:
Sep 14 15:06:29 <matteo> while (cpmac_read(bus->priv, CPMAC_MDIO_ACCESS(0)) & MDIO_BUSY);
Sep 14 15:06:36 <matteo> you wait for the device to be ready?
Sep 14 15:07:33 <matteo> what about sleeping a usec?
Sep 14 15:07:45 <matteo> while (cpmac_read(bus->priv, CPMAC_MDIO_ACCESS(0)) & MDIO_BUSY) usleep(1);
Sep 14 15:08:28 <blogic> usleep does not schedule
Sep 14 15:08:48 <blogic> i think it is only a calibrated while(true) loop
Sep 14 15:08:57 <farnz> matteo: cpu_relax() is what you're looking for, I think.
Sep 14 15:09:14 <matteo> farnz: i wish soemthing like yeld()
Sep 14 15:09:14 <ejka> matteo: as I said already, you can't sleep there
Sep 14 15:09:19 <blogic> cpu_relax(is the idle loop
Sep 14 15:09:26 <matteo> ok
Sep 14 15:09:36 <blogic> matteo: you could do a wait interruptable
Sep 14 15:10:21 <blogic> matteo: you got the /msg filter turned on
Sep 14 15:10:30 <blogic> http://freenode.net/faq.shtml#privmsg
Sep 14 15:10:37 <matteo> blogic: i know
Sep 14 15:10:46 <matteo> i should add to the init cmd
Sep 14 15:10:50 <blogic> well turn it off :-)
Sep 14 15:33:41 <ejka> matteo: still can't reproduce ipv6 problem
Sep 14 15:39:22 <ejka> matteo: using tspc tunnel + radvd
Sep 14 15:40:00 <matteo> ejka: connect to pastebin.ca from your pc
Sep 14 15:40:04 <matteo> with ipv6
Sep 14 15:41:52 <ejka> works fine
Sep 14 15:51:41 <matteo> damn
Sep 14 15:51:59 <matteo> ejka: i'll try it when you will post the final release
Sep 14 15:52:06 <matteo> and I'll put my router online again
Sep 14 15:52:21 <matteo> i connect with a 2nd router now
Sep 14 15:59:18 <ejka> it can be a problem with ppp/atm on dsl
Sep 14 16:20:27 <matteo> ejka: no, the error is in the 7cp/ip stack
Sep 14 16:20:40 <matteo> tcp/ip
Sep 14 16:20:58 <matteo> ejka: https://dev.openwrt.org/ticket/1671
Sep 14 16:21:12 <matteo> cpmac_poll+0x3d8/0x5bc
Sep 14 16:21:23 <matteo> it doesn't panics with disable_napi = 1
Sep 14 16:24:04 <matteo> "Kernel unaligned instruction access[#1]:"
Sep 14 16:24:05 <matteo> mmm
Sep 14 16:24:21 <matteo> you removed all those structs
Sep 14 16:24:28 <matteo> maybe the issue doesn't happens
Sep 14 16:24:39 <matteo> i'll try soon
Sep 14 16:24:43 <matteo> gonan study now
Sep 14 16:24:45 <matteo> bye :)
Sep 14 16:25:11 <ejka> the fault is in dev_hard_start_xmit, and I doubt it is cpmac's start_xmit
Sep 14 16:26:00 <matteo> a function can fault if I pass it struct with wrong data
Sep 14 16:26:06 <matteo> it could be something like this
Sep 14 16:43:55 <ejka> the skb fields are filled in cpmac_rx_one, whether you use napi or not.
Sep 14 20:28:05 <CIA-12> matteo * r8778 /trunk/target/linux/ar7/patches-2.6.22/500-serial_kludge.patch: don't broke other archs, we're close to mainstream (i hope)
Sep 14 21:38:24 <CIA-12> rwhitby * r8779 /trunk/target/linux/ixp4xx/ (config/default config-default): Moved config back to config-default, but left the FSG3 profile config non-working in the config directory as a reminder that it has to be changed over to the new scheme when that's ready.
Sep 14 22:27:34 <crazy_imp> hmmm, small typo-like thing here: openwrt-brcm-2.4-2.4-squashfs.trx
**** ENDING LOGGING AT Sat Sep 15 02:59:57 2007