**** BEGIN LOGGING AT Sun Apr 08 02:59:57 2007 Apr 08 07:12:58 <[florian]> sn9_: ok, sounds like a good explanation Apr 08 10:15:21 florian * r6893 /trunk/target/linux/adm5120-2.6/ (6 files in 3 dirs): Rename mipsIRQ.S to int-handler to match the kernel naming convention, implement basic board detection, rewrite usb HCD driver (to be tested) Apr 08 15:50:42 nico * r6894 /packages/libs/axtls/ (Makefile files/config): update axtls to 1.1.4, rename awhttpd to axhttpd, (un)install dev files Apr 08 16:13:31 <[florian]> {Nico}: ping ? Apr 08 16:18:36 in mussing about inside the device, i broke the solder joints i made at the unpopulated serial header Apr 08 16:19:14 i resoldered them, but i think i might have made a solder bridge somewhere Apr 08 16:20:10 because it works for a few seconds, then fails. if i wait some time, the same thing happens, and if i don't wait, it stays not working Apr 08 16:20:54 it seems like i am charging up that mystery capacitor more quickly than before Apr 08 16:22:57 there are three electrolytic capacitors on the board, one of which goes up to 15V when the power is on, but none of the three seem to be the culprit. there are lots of surface-mount things. any way i can get some idea what needs to be discharged? Apr 08 16:23:15 <[florian]> sn9_: no idea Apr 08 16:23:49 in that case, is there an easy way to tell whether i've made a solder bridge? Apr 08 16:24:41 if i've made one, it's most likely Vcc to ground Apr 08 16:24:42 <[florian]> I guess near the serial pinout Apr 08 16:24:43 sn9_: use your multimeter Apr 08 16:26:19 crazy_imp: i tried, but it shows what looks like negligible resistance among everything, so it doesn't yet really tell me anything i didn't already know Apr 08 16:27:24 sn9_: :/ Apr 08 16:32:13 crazy_imp: what would be the obvious effect of a direct Vcc->GND short? Apr 08 16:32:52 <[florian]> board reset I guess Apr 08 16:33:11 as in a reboot? Apr 08 16:33:18 <[florian]> yes Apr 08 16:33:35 i'm not referring to a momentary short, here Apr 08 16:34:14 <[mbm]> vcc-gnd just generates heat Apr 08 16:34:40 btw, i found the datasheet for the SoC, and it has lots of info that will be useful: http://www.rdc.com.tw/files/productfiles/R8610-G_D01_20051207.pdf Apr 08 16:34:46 sn9_: guess the voltageregulator would get hot until the thermal shutdown kicks in and cuts the power off Apr 08 16:34:53 <[mbm]> what you need to be careful of is running vcc at a higher voltage than the component expects Apr 08 16:35:37 crazy_imp: to the whole device? Apr 08 16:35:50 <[mbm]> obviously the router isn't going to work while vcc is connected to ground but it shouldn't cause serious issues after the short has been corrected Apr 08 16:36:19 hmm, that must mean i haven't really made such a short after all Apr 08 16:36:55 sn9_: i think so. even connecting a gpio direct without any loat to ground reboots any device i have :/ Apr 08 16:37:26 <[mbm]> crazy_imp: doesn't help that the power supplies are so cheap Apr 08 16:38:05 <[mbm]> some devices just turning on wifi is enough to cause a nice brownout on the router Apr 08 16:38:17 my gateway 7001 survived me doing something pretty stupid Apr 08 16:38:26 <[mbm]> just because the power supply is too close to the minimum limit Apr 08 16:38:39 h3sp4wn: plugging a 120V transformer into a 240V outlet? Apr 08 16:38:59 Bartman007: with a convertor but not a step down one Apr 08 16:39:08 and it still works perfectly Apr 08 16:40:14 there are two unpopulated connector headers in the device: JP1 and CON1 Apr 08 16:40:38 the openwrt wiki gives a partial pinout for CON1 Apr 08 16:41:27 it says that 1 is Tx, 2 is Vcc, 6 is GND, and 7 is Rx Apr 08 16:45:08 however, it is obvious from looking at the board that 8 is also GND Apr 08 16:47:02 on JP1, it is similarly obvious that 2 is GND Apr 08 16:48:34 each of 3 through 6 on JP1 is wired to one end of a different resistor Apr 08 16:50:10 3, 4, 5, and 9 on CON1, and 1 on JP1 seem to go only to lower layers of the board Apr 08 16:52:03 if the need later arises to jtag, which would of those sets of pins would it be? the ones going to resistors, or the ones going to lower layers? Apr 08 16:52:04 does someona know an alternative to the subversion package? (i need a server, not a client) Apr 08 16:52:15 s/someona/someone/ Apr 08 16:52:31 <[florian]> mercurial ? Apr 08 16:52:40 bazaar-ng Apr 08 16:52:48 rcs Apr 08 16:52:51 * Bartman007 ducks Apr 08 16:53:34 crazy_imp: are you asking for an alternate scm system, or an alternate implementation of svn itself? Apr 08 16:54:14 sn9_: alternative implemention of svn without apache :) Apr 08 16:55:10 for openwrt, or a glibc-based distro? Apr 08 16:55:23 sn9_: openwrt Apr 08 16:55:38 crazy_imp: why not use the svn server itself. Apr 08 16:55:48 not required to use it with apache. Apr 08 16:56:05 Bartman007: i can't compile subversion for openwrt :/ Apr 08 16:56:50 Bartman007: or show mbe a way to get around the apache part in the build process :) /usr/lib/libaprutil-1.so: could not read symbols: File in wrong format Apr 08 16:58:08 sounds like a path problem in ./configure Apr 08 16:59:24 <[florian]> crazy_imp: svn will require many deps to be satisfied Apr 08 16:59:33 sn9_: not really, it uses the file from my host system, but if i would build apr-util for brcm it can't be uses in the build process i think (?) Apr 08 17:00:01 [florian]: atm it's the only problem i have :D Apr 08 17:00:12 crazy_imp: that's what i mean. ./configure is using host paths, which it shouldn't Apr 08 17:01:48 sn9_: you think i should build apr-util and point it towards the right location? Apr 08 17:02:27 or make ./configure not see the host one, and assume it's not there Apr 08 17:03:25 pavlov * r6895 /packages/net/squid/ (5 files in 2 dirs): Apr 08 17:03:25 bring in squid 2.6, this disables winbind as it is no longer part of squid... Apr 08 17:03:25 epoll is forced via configure so for anyone using 2.4, you might have an issue Apr 08 17:03:25 and need to disable that. let us know if this poses a problem. max fds was Apr 08 17:03:25 bumped to 4096 as well so as to not use the system defaults, that might need Apr 08 17:03:27 tweaked. please test. Apr 08 17:04:17 sn9_: the file is there, svn itself needs it, if i remove it i can't use svn any longer and configuring subversion in the trunk fails because it's checking for the file at the beginning Apr 08 17:04:41 then your course of action is clear Apr 08 17:05:06 [10:00:52] crazy_imp sn9_: you think i should build apr-util and point it towards the right location? <-- YES Apr 08 17:05:28 sn9_: okokok, one moment please :D Apr 08 17:06:09 now, jtag uses four data pins, right? Apr 08 17:07:34 sn9_: i don't think it would work, it trys to execute the file and that wont work i think if it's build for mipsel Apr 08 17:08:25 execute a .so??? Apr 08 17:09:07 sn9_: ok, stupid me :D Apr 08 17:26:46 florian * r6897 /trunk/target/linux/brcm63xx-2.6/files/arch/mips/bcm963xx/setup.c: Re-enable MPI/PCI code Apr 08 17:58:28 can someone please answer the last question i asked about jtag? Apr 08 18:00:21 sn9_: http://en.wikipedia.org/wiki/JTAG 4 pins if the diagram is right Apr 08 18:02:06 oops, my mistake -- i meant the second to the last question; i forgot that i asked another one Apr 08 18:03:35 sn9_: the question about where to connect? dunno Apr 08 18:04:09 the question about resistors vs. lower layers Apr 08 18:10:12 the other end of each of the resistors goes to the SoC. i will attempt to determine what those pins on the SoC mean Apr 08 18:20:24 it looks like pin 4 of JP1 is wired directly, without a resistor, to A20GATE_n/GPIO54 on the SoC Apr 08 18:31:18 similarly, pin 6 is CTS_n/TMS/GPIO11 Apr 08 18:34:45 pin 5 is DSR_n/TCK/GPIO12 Apr 08 18:38:31 pin 3 is RTS_n/TDO/GPIO9 Apr 08 18:44:21 the DCD_n/TDI/GPIO13 pin on the SoC is connected only to resistor R21 Apr 08 19:26:49 oops Apr 08 19:27:14 make that 6 is DSR_n/TCK/GPIO12 and 5 is DCD_n/TDI/GPIO13 Apr 08 19:32:15 pin 4 is CTS_n/TMS/GPIO11 Apr 08 19:32:36 A20GATE_n/GPIO54 is connected to resistor R6 Apr 08 19:32:59 i was looking at a bad angle the first time Apr 08 19:59:38 nbd * r6898 /packages/net/aircrack-ptw/ (. Makefile patches/ patches/001-smaller.patch): add aircrack-ptw (patch by Jens Muecke) Apr 08 20:00:06 :-/ Apr 08 20:02:29 florian * r6899 /trunk/target/linux/brcm63xx-2.6/files/arch/mips/pci/pci-bcm96348.c: Fix warnings Apr 08 20:10:19 florian * r6900 /packages/net/ssmtp/Makefile: Enable SSL support in ssmtp (#1567) Apr 08 20:17:05 florian * r6901 /trunk/target/linux/adm5120-2.6/config/default: Enable bluetooth modules (#1529) Apr 08 20:21:51 florian * r6902 /trunk/package/yamonenv/patches/ (. 001-yamonenv_mtd_partition.patch): Fix yamonenv errors on accessing environment (#1542) Apr 08 20:35:38 florian * r6903 /packages/net/ntpd/Makefile: Create /etc/services for ntp if not existing(#1557) Apr 08 20:36:00 any ideas why such a thing happens: build_mipsel/apr-1.2.8/ipkg-install/home/mj/trunk/build_mipsel/apr-1.2.8/ipkg-install/usr ? Apr 08 20:38:48 <[florian]> symlink problem Apr 08 20:38:55 hm, looks like i overread some parts of the prefix stuff.... we'll see :D Apr 08 20:39:01 <[florian]> or absolute path being passed somewhere Apr 08 20:40:43 [florian]: i passed --includedir="$(PKG_INSTALL_DIR)", simply missed the fact that it already is [PREFIX/include] Apr 08 20:41:27 yupp, works now :) Apr 08 20:42:30 time to get apr-util working Apr 08 20:58:04 hm, $(CP) includes -r, or? Apr 08 20:59:30 florian * r6904 /trunk/package/iproute2/ (Makefile patches/004-darwin_fixes.patch): Fix MacOSX build errors with iproute2 (#1533) Apr 08 21:00:35 <[florian]> it includes -r,yes Apr 08 21:00:40 <[florian]> no, sorry Apr 08 21:00:44 <[florian]> -fpR Apr 08 21:00:58 but it's also necrusive, or? Apr 08 21:01:11 s/necrusive/recrusive/ Apr 08 21:03:07 <[florian]> it is Apr 08 21:12:32 florian * r6905 /trunk/ (2 files in 2 dirs): Add ISDN support (#1498) Apr 08 21:13:28 cool, i can reuse my fritzcard now? Apr 08 21:14:47 <[florian]> I did not enable any device support yet Apr 08 21:15:07 [florian]: ack Apr 08 21:18:20 florian * r6906 /trunk/ (2 files in 2 dirs): Enable codepage 1250 support (#1554) Apr 08 21:22:23 florian * r6907 /branches/whiterussian/openwrt/package/base-files/default/etc/init.d/ (S10boot rcS): Set hostname before logging (#1552) Apr 08 21:50:07 florian * r6908 /packages/net/iodine/Makefile: Upgrade iodine Apr 08 21:55:58 following the traces on the board, it appears that all the reset switch does is ground a gpio line Apr 08 21:56:46 sn9_: :D Apr 08 21:58:42 sn9_: hm, so it doesn't behave like any resetbuttons i know from routers where you can read the status (guess it reboots your device without any interaction with the os, right?) Apr 08 22:02:00 crazy_imp: huh? Apr 08 22:03:15 with the stock firmware, it reboots and/or resets nvram; does nothing in redboot or openwrt Apr 08 22:04:10 i would have thought it was some kind of hard reset, but it only grounds a gpio line, so that the OS does whatever in response Apr 08 22:05:17 the dmz led is the adjacent gpio line Apr 08 22:05:28 hmmm, i remember it caused a hard reset as i connected the gpioline from the powerled from a wl500gp direct to ground Apr 08 22:06:11 the power led, OTOH, is apparently not connected to gpio Apr 08 22:06:38 so it's always on? Apr 08 22:06:47 in series with a resistor, it just shunts Vcc to ground Apr 08 22:09:20 it kinda looks like the serial header has two Vcc pins Apr 08 22:09:46 i wonder why Apr 08 22:12:53 for some strange reason, the cpu's PWRGOOD signal is connected to the ethernet switch chip Apr 08 22:16:11 i just made a ticket with my apr makefile - i have problems with apr-util, configure works fine, generates the Makefile with CPP = mipsel-linux-uclibc-gcc -E but later it's using i686-pc-linux-gnu-gcc *confused*. should i upload it too (somewhere)? Apr 08 22:17:13 sn9_: so if you turn PWRGOOD of (is it possible?) the switch dies? Apr 08 22:17:31 s/of/off/ Apr 08 22:17:41 um, that's a signal to the cpu Apr 08 22:18:15 sn9_: thougt you said it's connected to the switch? Apr 08 22:18:56 yeah, somehow, the switch someway can initiate a hard reset, it looks like Apr 08 22:26:45 sn9_: would be nice if they ship with routers something like an eaglelayout ;D Apr 08 22:27:41 grounding that line from the switch indeed does a hard reset Apr 08 22:29:17 it is the tenth pin clockwise from the top left of the switch chip Apr 08 22:31:06 sn9_: btw, which devices? Apr 08 22:31:15 s/devices/device/ Apr 08 22:31:24 airlink101 ar525w Apr 08 22:34:29 ack Apr 08 22:36:21 in the switch datasheet, the pin is labeled RESETB Apr 08 22:38:51 hmm, that's an input on the switch, too Apr 08 22:41:03 oh, wait, i didn't notice before, but the line also goes to a little black component labeled U6 Apr 08 22:46:12 it also goes to the flash chip Apr 08 22:48:43 pin 12 of it Apr 08 22:51:15 the hardware reset pin Apr 08 22:52:04 U6 is apparently a transistor or such Apr 08 22:54:17 or a shunt of somekind Apr 09 00:50:55 nbd * r6909 /trunk/ (3 files in 3 dirs): add magicbox 2.0 cf slot support - kmod-ide must not be used on magicbox 1.1 (no working hw detection yet) Apr 09 01:02:47 nbd * r6910 /trunk/include/scan.mk: fix dependencies for metadata scanning **** ENDING LOGGING AT Mon Apr 09 02:59:56 2007