**** BEGIN LOGGING AT Sat Aug 21 02:59:57 2010
Aug 21 04:26:10 <CIA-41> agb * r22745 /branches/backfire/package/broadcom-sdhc/: [backfire] merge r21702
Aug 21 09:34:10 <_trine> WELL
Aug 21 09:34:33 <_trine> who owning up to removing webif from trunk :P
Aug 21 09:37:01 <_trine> just did a distclean then updated and installed the feeds again and now webif seems to be missing
Aug 21 09:49:13 <_trine> I just tried it again and webif is still missing
Aug 21 10:04:09 <_trine> I just did a completely new install of trunk and now webif is back again ,, odd
Aug 21 10:08:23 <lautriv> blogic, found serial connector :-)
Aug 21 10:15:40 <KanjiMonster> _trine: perhaps a symlink was missing?
Aug 21 10:16:58 <_trine> well I dont know now I deleted it all and just downloaded new which seems to be ok now
Aug 21 12:38:34 <blogic> lautriv: i found my board, i will have a hack at it today
Aug 21 12:38:44 <blogic> lautriv: as it seems the thing has uboot on it already
Aug 21 12:39:17 <blogic> lautriv: if this is the case, all we need to do is setup a machtype and create an image
Aug 21 13:51:08 <mb__> Hm, I'm wondering if there's something wrong with target/linux-compile make target.
Aug 21 13:51:26 <mb__> It's supposed to recompile the kernel image, too (not just external modules), right?
Aug 21 13:51:33 <mb__> it doesn't do that for me, however
Aug 21 13:51:58 <mb__> Neither in kernel QUILT mode, nor in non-QUILT mode after a target/linux-{clean,prepare}
Aug 21 13:52:23 <mb__> The "all" target properly triggers recompile of the kernel image
Aug 21 14:02:09 <mb__> http://pastebin.org/668842
Aug 21 14:02:12 <mb__> strange
Aug 21 14:13:57 <lautriv> blogic, afaik there is u-boot, at least the GPL'd sources are pointing there, i found 4 pins with a small cmos-tiny-logic dual-inverter, with GND/IN/OUT/unk. pinout but it seems that'S TTL @ 3v3 or 5V.
Aug 21 14:14:46 <KanjiMonster> mb__: linux/compile never built the kernel for me, so I don't know whether this is intended or not
Aug 21 14:15:20 <mb__> KanjiMonster: hm, maybe I wasn't remembering correctly then
Aug 21 14:15:52 <mb__> though, as you can see in the pastbin I posted, it tries to compile the image and says there's nothing to do. Which is wrong.
Aug 21 14:16:55 <{Nico}> mb__: make target/linux/install is building everything that's not a module
Aug 21 14:17:07 <mb__> ah thanks
Aug 21 14:17:37 <{Nico}> but it also tries to build images
Aug 21 14:44:36 <CIA-41> acoul * r22746 /packages/net/ethtool/ (Makefile patches/01-ixp4xx_ethtool_autotools.patch): net/ethtool: update to ethtool-2.6.35
Aug 21 15:40:09 <[florian]> Wipster: ping?
Aug 21 15:40:27 <Wipster> [florian], afternoon
Aug 21 15:40:40 <[florian]> Wipster: I was just playing with the adm6996
Aug 21 15:40:55 <Wipster> [florian], ahyeh got ur email, how you get on?
Aug 21 15:42:30 <[florian]> Wipster: well, it's reading zeroes in the phy identification registers, so I am afraid we will have to stick with fixed phy for these
Aug 21 15:42:55 <Wipster> [florian], ahhh shame
Aug 21 15:43:27 <Wipster> worth disabling the ephy in that case and seeing if it helps?
Aug 21 15:43:44 <Wipster> internal I mean
Aug 21 15:43:45 <[florian]> Wipster: I might check that yes
Aug 21 15:44:47 <CIA-41> mb * r22747 /trunk/target/linux/omap24xx/ (2 files in 2 dirs): n810: Add retu/tahvo userspace interface debugging
Aug 21 15:49:33 <[florian]> Wipster: it worked with an ADM6996M :)
Aug 21 15:49:42 <[florian]> Wipster: now let me try with the ADM6996L
Aug 21 15:49:57 <[florian]> and it pings, awesome
Aug 21 15:50:00 <Wipster> [florian], oh brilliant
Aug 21 15:50:37 <[florian]> the switch, however is not controllable, but that's fine
Aug 21 15:51:48 <[florian]> I just stumbled over another "bug"
Aug 21 15:51:54 <[florian]> but that's not related to the switch
Aug 21 15:52:10 <[florian]> the device has a tnetd7200 and the cpu frequency is incorrect
Aug 21 15:52:13 <[florian]> but that's for later ;)
Aug 21 15:53:13 <Wipster> :)
Aug 21 15:53:30 <[florian]> Wipster: I need to check which revision of ar7 is exactly ohio
Aug 21 15:53:59 <[florian]> Wipster: but if it is as simple as tnetd7200, then we do not even have to first try to detect the PHY by "normal" means, we can just toggle the bit in the platform code directly
Aug 21 15:54:30 <[florian]> Wipster: however, I am not sure this is so simple
Aug 21 15:54:53 <Wipster> [florian], yeh I wasn't sure on that one didn't spot a chip name... might have been a revison only used by one company.
Aug 21 15:55:47 <[florian]> Wipster: well, that did not work on the tnetd7300 with adm6996M
Aug 21 15:56:01 <[florian]> Wipster: which is a linksys wag54g btw
Aug 21 15:56:15 <[florian]> ok, so the correct probing order should be
Aug 21 15:56:21 <[florian]> 1) try to detect the phy "normally"
Aug 21 15:56:26 <[florian]> 2) disable EPHY, then check again
Aug 21 15:56:33 <[florian]> 3) if that still does not work, fallback to fixed phy
Aug 21 15:56:55 <Wipster> [florian], when I looked at the remap address it was in the range specified by the watchdog in ar7.h, but I might have been readong wrong. why that put the external MII where the normal fixed eth1 connection is is beyond me
Aug 21 15:57:26 <Wipster> [florian], but with CONFIG_FIXED_PHY a fixed MDIO bus registers, can that be turned off?
Aug 21 15:57:41 <[florian]> Wipster: that's what I need to check
Aug 21 15:59:48 <[florian]> Wipster: you are right, it gets registered unconditionnaly, which is not so good
Aug 21 15:59:55 <CIA-41> lars * r22748 /trunk/target/linux/ifxmips/files-2.6.33/arch/mips/ifxmips/danube/irq.c: [ifxmips] Danube: Fix irq ack
Aug 21 15:59:56 <Wipster> :(
Aug 21 16:02:22 <[florian]> oh, but it may not be a problem after all, if we register it in cpmac_probe
Aug 21 16:02:49 <[florian]> I go to run now, see you later
Aug 21 16:05:31 <Wipster> cya
Aug 21 16:10:10 <builder> build #94 of avr32 is complete: Failure [failed compile]  Build details are at http://tksite.gotdns.org:8010/builders/avr32/builds/94
Aug 21 16:26:16 <CIA-41> mb * r22749 /trunk/package/mac80211/patches/800-nuke_led_code.patch: mac80211: Fix incorrect ifdef placement in 800-nuke_led_code.patch
Aug 21 16:44:56 <CIA-41> agb * r22750 /packages/net/znc/Makefile: [packages] net/znc: fixup compilation options. Closes #7786, thanks Obsy
Aug 21 17:03:17 <CIA-41> agb * r22751 /trunk/tools/qemu/Makefile: [tools] qemu: fix build failure on some systems. Closes #7767
Aug 21 17:04:08 <CIA-41> agb * r22752 /branches/backfire/tools/qemu/Makefile:
Aug 21 17:04:08 <CIA-41> [backfire] merge r22751
Aug 21 17:04:09 <CIA-41> [tools] qemu: fix build failure on some systems. Closes #7767
Aug 21 17:17:12 <[florian]> Wipster: re
Aug 21 17:17:52 <Wipster> [florian], welcome back
Aug 21 17:18:45 <[florian]> Wipster: I am cooking up the patch which does what I suggested earlier
Aug 21 17:19:06 <Wipster> [florian], was looking into how to detect Ohio or sangam having a gander through some gpl just incase
Aug 21 17:19:19 <[florian]> Wipster: yes, good idea
Aug 21 17:21:56 <[florian]> Wipster: if you can also stumble upon which register 0x08611A08 is associated too, that would be cool
Aug 21 17:21:58 <Wipster> [florian], also saw in ag300 bootlogs autodetecting external phy's and the like, going to see if there is anything juicy to read
Aug 21 17:38:03 <[florian]> Wipster: ah, this would be a good hint
Aug 21 17:38:12 <[florian]> Wipster: beware, the original ar7 code is pretty messy
Aug 21 17:38:33 <Wipster> [florian], I'm seeing this, my eyes heh
Aug 21 17:40:28 <[florian]> yeah, it hurts :)
Aug 21 17:45:11 <[florian]> Wipster: got it working with the fallback stuff
Aug 21 17:45:43 <KanjiMonster> realtek code is also quite messy (at least the sdk in a gpl tar I've seen was)
Aug 21 17:46:08 <[florian]> most vendors trying to do hardware abstraction stuff are usually pretty bad at it unfortunately
Aug 21 17:46:23 <[florian]> if you do not work with upstream directly it's pretty rare they do something nice
Aug 21 17:46:37 <[florian]> as far as I can tell, only ubicom has done a great job on that
Aug 21 17:47:14 <Wipster> [florian], nice bit of a hazard there if we scan normal and see nothing with the first interface and go to external and find something, if we try the second which could be internal on then the code leaves the chip with it on, no?
Aug 21 17:47:51 <Wipster> that made more sense in my head...
Aug 21 17:48:50 <[florian]> Wipster: yeah
Aug 21 17:50:15 <[florian]> humm, only the two first ports of the switch actually work with fixed phy, sounds weird
Aug 21 17:50:26 <[florian]> ah no, that may be the learning time
Aug 21 17:53:31 <Wipster> [florian], found retection of ohio or sangam
Aug 21 17:53:37 <Wipster> *detection
Aug 21 17:54:00 <[florian]> ah, can you check if it is the same as checking for the tnetd7300/7200?
Aug 21 17:54:26 <Wipster> sure, that in platform I guess?
Aug 21 17:54:43 <[florian]> yes, in arch/mips/include/asm/mach-ar7/ar7.h
Aug 21 18:01:52 <Wipster> [florian], yeh it recons an ohio is a 7100, might be reading wrong as I have seen boot logs from 7200's that say ohio
Aug 21 18:03:18 <Wipster> [florian], it checks the same address and compares to 0x18 everything else is sangam apparently
Aug 21 18:04:17 <Wipster> [florian], also #define AVALANCHE_DCL_BOOTCR (KSEG1ADDR(0x08611A00)) thats close to that magic remap
Aug 21 18:04:33 <[florian]> Wipster: okay, I got a device here which has a SoC printed as a 7300, but which is detected as a 7200 by this code, and on wihch the "external MII hack" is required
Aug 21 18:05:00 <[florian]> Wipster: I do not think we can reliably assume that tnetd7100 or 7200 = ohio
Aug 21 18:05:41 <Wipster> [florian], think you might be right there
Aug 21 18:08:12 <Wipster> ah that define's address is also defined as the AVALANCHE_DEVICE_CONFIG_LATCH_BASE
Aug 21 18:08:20 <[florian]> ah
Aug 21 18:16:25 <AndyI> <Wipster> sangam have 3 different version
Aug 21 18:19:12 <Wipster> AndyI, Ah right can detect which ones?
Aug 21 18:19:50 <[florian]> Wipster: okay the patch works on all 3 devices that I have, I will commit it later when I am back home
Aug 21 18:20:21 <Wipster> [florian], brilliant :)
Aug 21 18:24:56 <AndyI> <Wipster> http://pastebin.org/678469
Aug 21 18:29:43 <AndyI> <Wipster> show the patch please
Aug 21 18:31:23 <Wipster> AndyI, sorry?
Aug 21 18:32:52 <AndyI> patch for cpmac
Aug 21 18:34:48 <Wipster> AndyI, think [florian] has a good one for detecting the switch on his devices. My only device could get hold of the switch with a simple one which is trunked now
Aug 21 19:26:17 <CIA-41> mb * r22753 /trunk/ (7 files in 5 dirs): Update to 2.6.35.3
Aug 21 20:26:50 <Wipster> that code is giving me a bit of a headach, and close
Aug 21 20:27:31 <lautriv> blogic, something new ?
Aug 21 20:34:07 <blogic> ewwnot yet
Aug 21 20:34:11 <blogic> playing with other toys
Aug 21 20:35:15 <lautriv> blogic, don't hurry, was just curios...........my brute-force is also not finished, needs long if you get disconnected each 3 attempts :-)
**** ENDING LOGGING AT Sun Aug 22 02:59:58 2010