**** BEGIN LOGGING AT Wed May 04 23:59:56 2005
May 05 01:59:59 <rwhitby-away> pushing unslung and openslug linuxrc changes now
May 05 02:00:07 <rwhitby-away> wonder where CIA is ... ?
May 05 02:01:44 <ka6sox-zzzz> MIA?
May 05 02:05:20 <ka6sox-zzzz> okay I just moved the Packages file to Unslung/Cross
May 05 02:05:27 <ka6sox-zzzz> can somebody test?
May 05 03:13:56 <rwhitby-away> macsat: hi!
May 05 03:14:26 <rwhitby> bbiab
May 05 03:15:35 <macsat> Hi rwhitby-away :)
May 05 03:15:59 <lennert> hello
May 05 03:20:53 <rwhitby-away> lennert: did your slug arrive safely?
May 05 03:21:15 <lennert> i _think_ so
May 05 03:21:17 <rwhitby> macsat: nice howto's - we need to get you writing NSLU2 howto's for us :-)
May 05 03:21:26 <rwhitby> lennert: think?
May 05 03:21:26 <lennert> i've spent some time trying to get it to work, but failed so far
May 05 03:21:35 <lennert> it doesn't look damaged or anything though
May 05 03:21:40 <lennert> well
May 05 03:21:44 <lennert> the redboot trick doesn't work
May 05 03:21:49 <lennert> i.e. telnet to 192.168.0.1
May 05 03:22:01 <rwhitby> it's got serial, hasn't it?
May 05 03:22:09 <lennert> the ethernet led on the nslu2 does flash according to network activity
May 05 03:22:27 <lennert> i put the serial on 115k2 but it outputs just a bunch of garbage at bootup
May 05 03:22:43 <lennert> i.e. things like "................../),..o..((-.[2"
May 05 03:23:24 <lennert> i don't 100.1% trust the machine that the nslu2 is connected to atm, so i'll try connecting it to a different pc later on
May 05 03:23:24 <rwhitby> 8n1?
May 05 03:23:29 <lennert> 8n1, no flow control
May 05 03:23:37 <macsat> heeh thanks rwhitby, unfortunately I dont have a NSLU2 :-)
May 05 03:23:59 <dyoung-zzzz> link?
May 05 03:24:09 <lennert> 'link'?
May 05 03:24:15 <rwhitby> lennert: did Christian test it after soldering and before sending it to you?
May 05 03:24:17 <lennert> dyoung: sleepwalking again? :)
May 05 03:24:22 <lennert> rwhitby: i didn't ask.. :(
May 05 03:24:34 <rwhitby> maybe it's on his LAN's IP address
May 05 03:24:44 <dyoung> macsat's Wiki Entry Link.
May 05 03:24:46 <rwhitby> but the redboot and serial should work
May 05 03:24:59 <rwhitby> you've got a level converter hooked up?
May 05 03:25:17 <lennert> rwhitby: i wanted to mess with it for a while myself before asking stupid questions
May 05 03:25:18 <macsat> my howtows are at http://www.macsat.com/wlguide dyoung
May 05 03:25:42 <rwhitby> lennert: there are no stupid questions, just stupid people ;-)
May 05 03:25:57 <lennert> rwhitby: in the serial connector, i think -- i try to open it to make sure
May 05 03:26:17 <dyoung> rwhitby, I was just gonna sasy the same thing; but i was gonna qualify it that lennert cannot be stupid, because he is the Man.
May 05 03:26:36 <dyoung> thus any question he asks is valid. ;-)
May 05 03:26:38 <lennert> rwhitby: well, asking 'stupid' questions and getting really trivial and obvious answers is somewhat damaging to one's self-esteem :)
May 05 03:26:44 <rwhitby> yeah, lennert is *the* man.
May 05 03:26:50 <rwhitby> ~praise lennert
May 05 03:26:51 <lennert> dyoung: oh come on, i'm as useless as they come when it comes to hardware :)
May 05 03:26:54 <jbot> All hail lennert!
May 05 03:27:07 <lennert> ok ok, this is too much honor, please :)
May 05 03:27:27 <dyoung> sounds like me and soldering.
May 05 03:27:35 <rwhitby> so did Christian send the serial connector ?
May 05 03:27:54 <dyoung> macsat: nice work!
May 05 03:28:09 <macsat> tnx
May 05 03:28:10 <rwhitby> (I don't remember if he was going to put a CompSys kit in there too, or just solder on the connector)
May 05 03:28:26 <lennert> rwhitby: there's an extra wire coming out of the nslu2 which is connected to a 9 pin female d connector
May 05 03:28:49 <lennert> rwhitby: i've opened the nslu2 and there are three wires connected to tx/rx/gnd
May 05 03:29:31 <lennert> rwhitby: the connector was already there
May 05 03:29:36 <dyoung> I'll be sure to follow these howto's when I become Wiley-Enabled.
May 05 03:29:54 <rwhitby> any circuitry on the DB9 connector?
May 05 03:30:02 <lennert> rwhitby: maybe inside, didn't check yet
May 05 03:30:26 <lennert> rwhitby: it has a sticker "c/s/m/sl series" on it
May 05 03:30:40 <lennert> rwhitby: so it's probably some kind of cable for mobile phones
May 05 03:30:46 <lennert> rwhitby: s/is/was/
May 05 03:31:42 <rwhitby> Hmm, ok.  Can you take a picture for us?
May 05 03:32:22 <lennert> rwhitby: yeah, hang on
May 05 03:32:29 <lennert> rwhitby: what time is it over there now?
May 05 03:33:00 <rwhitby> 8pm
May 05 03:33:06 <dyoung> Time is Relative.
May 05 03:33:53 <lennert> what i should have asked was, "how long until you go to sleep?"
May 05 03:33:57 <lennert> dyoung: indeed
May 05 03:34:12 <rwhitby> lennert: at least two or three hours
May 05 03:35:04 <dyoung> I should get back to hacking some hardware soon.
May 05 03:42:26 <lennert> connector is rather hard to open
May 05 03:46:47 <lennert> ok
May 05 03:47:07 <lennert> it has a 16-pin surface mount IC in it
May 05 03:47:11 <lennert> on a tiny bit of pcb
May 05 03:47:21 <lennert> SIPEX SP232ACN  \delta 0149
May 05 03:47:34 <lennert> and three surface mount components
May 05 03:47:57 <rwhitby> http://www.sipex.com/products/pdf/sp231_312a.pdf
May 05 03:48:38 <lennert> yeah, that one
May 05 03:50:13 <dyoung> this extra wire... where does it go to in the nslu2?
May 05 03:50:39 <lennert> there's a white, black and red wire
May 05 03:50:46 <lennert> let me open it again
May 05 03:51:35 <lennert> red to pin 2, white to pin 3, black to pin 4
May 05 03:51:45 <lennert> (on the J2 connector in the nslu2)
May 05 03:52:01 <rwhitby> dyoung: you thinking VCC?
May 05 03:52:22 <dyoung> thats my initial thought....
May 05 03:52:27 <rwhitby> lennert: only three wires going into the slug?
May 05 03:52:56 <rwhitby> 2 is RXD, 3 is TXD, 4 is GND.  What is connected to pin 16 on the SP232 chip in the connector?
May 05 03:53:11 <rwhitby> we're looking for how that chip get's it's 5V power supply.
May 05 03:54:40 <dyoung> and actually if its runnign 5V, thats asking for trouble.
May 05 03:54:52 <rwhitby> yeah, if it's connected to 3.3V
May 05 03:55:08 <rwhitby> or if it's connected to 5V and putting 5V on the ixp pins
May 05 03:55:56 <dyoung> Either of those cases will cause an issue.
May 05 03:56:35 <lennert> ping 15 is GND, and that's connected to a rather wide PCB trace
May 05 03:56:54 <lennert> which eventually connects to serial port pin.. 5?
May 05 03:57:07 <lennert> pin 16 goes through the pcb
May 05 03:57:44 <lennert> i'll just take a picture of this
May 05 03:57:55 <dyoung> that part looks like itst a 5V only part.
May 05 04:01:49 <rwhitby> yeah
May 05 04:02:13 <rwhitby> lennert: are there really only three wires going to the slug?
May 05 04:02:20 <lennert> rwhitby: yeah
May 05 04:03:04 <lennert> ok, took some pictures
May 05 04:03:07 <rwhitby> ok, so that chip must be getting power from the serial port
May 05 04:03:59 <dyoung> which means its outputting 5V.
May 05 04:04:18 <rwhitby> or worse ...
May 05 04:05:01 <rwhitby> lennert: do you have any way of checking continuity?  can you test which pin on the DB9 is connected to pin 16 on the chip?
May 05 04:08:18 <lennert> rwhitby: i don't think it's connected directly
May 05 04:08:39 <lennert> rwhitby: unfortunately i don't have a multimeter or anything alike
May 05 04:09:01 <lennert> things like this are a good excuse for getting one, though
May 05 04:09:07 <lennert> let me copy those pictures
May 05 04:09:21 <dyoung> lennert, is it easy to visually inspect to see where pin16 on that SIPEX part goes to?  ie: a trace going to the serial side, or a trace going to the "slug side" ?
May 05 04:11:16 <dyoung> lennert, hehe I had to break down and break aparty my 'scope and fix it, just to diagnose another hardware problem I've been havng.
May 05 04:11:34 <dyoung> the scope's been acting funny for at least a year.
May 05 04:13:47 <lennert> darn, picture's unsharp
May 05 04:14:06 <lennert> ok, my sp232acn is slightly different from the ones in the datasheet
May 05 04:14:31 <lennert> in the datasheet, there's a U-shaped thingy between pin 1 and 16/20
May 05 04:14:40 <lennert> my chip has a dot, presumably near pin 1
May 05 04:14:59 <dyoung> yes, the dimple indicates pin1.
May 05 04:15:00 <lennert> i can't see pin 16 (opposite one of where the dot is) going anywhere
May 05 04:15:29 <lennert> i think it might be going underneath the ic itself
May 05 04:15:47 <dyoung> it must go somewhere... udner the chip is possible/.
May 05 04:16:04 <dyoung> does it pop out the other side, or underneath somewhere?
May 05 04:17:34 <lennert> http://www.liacs.nl/~buytenh/nslu2/ will have the pics in a few minutes
May 05 04:18:14 <lennert> it's very hard to see :(
May 05 04:19:24 <dyoung> we'll try to squint real hard to make sense of it all. ;-)
May 05 04:23:03 <lennert> can you see the pics?
May 05 04:23:38 <lennert> i'll redo the second one
May 05 04:25:30 <dyoung> no need.
May 05 04:25:35 <dyoung> I can see whats happening already.
May 05 04:25:39 <lennert> ok
May 05 04:26:28 <NAiL> lennert: The cable is a Siemens cable, if you didn't know already ;)
May 05 04:26:45 <lennert> NAiL: i didn't :)
May 05 04:26:51 <dyoung> that 78L05 is a 5V regulator, and is outputting 5V to pin16 through a via.
May 05 04:26:57 <dyoung> thats bad.
May 05 04:27:32 <lennert> dyoung: isn't the chip a different one than in the datasheet?
May 05 04:27:37 <dyoung> what we need for the nslu2 is a 3.3V (LVTTL) level converter, as the i/o on the IXP are not 5V tolerant.
May 05 04:28:33 <dyoung> nah, SP232 is enough to know what it is.  What you really need is a 3232 which is the 3.3V part.
May 05 04:28:48 <dyoung> however, if you feel like soldering, I have a solution for you. ;-)
May 05 04:29:05 <lennert> soldering involves a soldering iron, no? :)
May 05 04:29:12 <dyoung> But its likely gonna be a PITA.
May 05 04:29:22 <lennert> so did this blow up the serial port on the ixp?
May 05 04:29:33 <rwhitby-away> probably not
May 05 04:30:05 <dyoung> lennert, does it boot without the serial adapter attched to a serial port?
May 05 04:30:38 <lennert> dyoung: the leds do blink, but the redboot method still doesn't work without the serial, so i'm not sure
May 05 04:31:05 <dyoung> does it ever beep?
May 05 04:31:14 <lennert> i don't think it ever beeped
May 05 04:31:25 <dyoung> interesting...
May 05 04:31:35 <rwhitby> it takes up to 3 minutes to beep
May 05 04:32:02 <lennert> if i boot it without a usb disk attached, the ready/status led eventually goes off, and the ethernet one keeps blinking along with network traffic
May 05 04:32:25 <lennert> with an usb disk attached, the ready/status led is on pretty much as soon as i press the on button, and doesn't blink anymore
May 05 04:33:00 <rwhitby> lennert, can you boot with no disk attached, and describe the led patterns and any beeps ....
May 05 04:33:10 <lennert> ok, hang on
May 05 04:34:08 <lennert> i'll reassemble it
May 05 04:36:15 <NAiL> hrrm
May 05 04:36:42 <lennert> ok, no serial, no disk
May 05 04:36:51 * NAiL found an alternative serial cable, but the name of the chip has been removed
May 05 04:37:51 <lennert> ready/status goes orange or yellow
May 05 04:37:58 <lennert> ethernet is blinking (greenish)
May 05 04:38:03 <lennert> r/s stays on
May 05 04:38:08 <lennert> now it starts blinking green
May 05 04:38:10 <lennert> now it's off
May 05 04:38:23 <lennert> still off
May 05 04:38:27 <dyoung> hmm
May 05 04:38:31 <lennert> ethernet still blinking occasionally
May 05 04:39:07 <lennert> r/s blinking again
May 05 04:39:11 <lennert> now r/s is constantly green
May 05 04:39:15 <dyoung> really?
May 05 04:39:26 <dyoung> you should be able to ping 192.168.1.77 now then.
May 05 04:39:41 <rwhitby> lennert: do you have the NSLU2 Setup Wizard program?
May 05 04:39:52 <lennert> 192.168.1.77 doesn't ping
May 05 04:39:56 <rwhitby> (on the CD which came with it) and a PC connected to the network?
May 05 04:39:59 <lennert> now r/s is off again
May 05 04:40:25 <lennert> rwhitby: i guess i could get that running in a win98 virtual machine?
May 05 04:40:27 <lennert> r/s blinking again
May 05 04:40:31 <lennert> now 192.168.1.77 pings
May 05 04:40:49 <lennert> so what's happening? :)
May 05 04:41:46 <lennert> tcp port 6789 is open
May 05 04:43:36 <rwhitby> lennert: can you run upslug?
May 05 04:44:13 <rwhitby> (on linux :-)
May 05 04:46:13 <lennert> [root@vi buytenh]# ./upslug 
May 05 04:46:13 <lennert> Using eth0 to find slugs...
May 05 04:46:13 <lennert> Finding Slugs...
May 05 04:46:13 <lennert> Which slug?:
May 05 04:47:16 <rwhitby> can you still ping 192.168.1.77 ?
May 05 04:47:28 <lennert> yes, still pinging
May 05 04:49:34 <lennert> is there a faq that i've managed to miss?
May 05 04:50:44 <rwhitby> nope
May 05 04:51:09 <NAiL> I had that problem too
May 05 04:51:17 <NAiL> upslug couldn't find any slugs
May 05 04:51:23 <NAiL> but still ping
May 05 04:51:33 <rwhitby> lennert: direct connection or through router?
May 05 04:52:05 <lennert> just a switch
May 05 04:52:20 <rwhitby> not 10MB, right?
May 05 04:52:40 <lennert> 100Mb
May 05 04:52:59 <rwhitby> dyoung: wanna try the reset button method?
May 05 04:52:59 <lennert> it broadcasts a packet with ethertype 0x8888
May 05 04:53:05 <lennert> and indeed receives no response
May 05 04:56:43 <lennert> it stopped pinging now, btw
May 05 04:56:58 <rwhitby> lennert, can you try this: http://www.nslu2-linux.org/wiki/HowTo/UseTheResetButtonToEnterUpgradeMode
May 05 04:57:20 <lennert> it was pingable for exactly 854 sec, now the r/s light is just off
May 05 04:57:27 <lennert> rwhitby: will do
May 05 04:57:46 <rwhitby> if you get red/green alternating, then upslug should work
May 05 04:58:28 <lennert> i don't see the difference between red and green so i'll need the significant other for that
May 05 04:59:10 <lennert> so.. what's up with my unit?
May 05 04:59:45 <rwhitby> dunno, I think it's booting into linux, but not completing the boot sequence, and maybe watchdog timing out and rebooting ...
May 05 05:00:11 <lennert> and what's up with 192.168.1.77 versus 192.168.0.1 ?
May 05 05:00:28 <lennert> sorry, i'm just very curious
May 05 05:00:33 <lennert> 192.168.0.1 is redboot ip
May 05 05:00:38 <lennert> 192.168.1.77 is ... default linux ip?
May 05 05:01:08 <rwhitby> normal boot sequence is orange (redboot), then blinking green (kernel booting), then solid green (boot complete) and a beep.
May 05 05:01:40 <rwhitby> the orange should be about 2 to 5 seconds, then blinking green for a couple of minutes
May 05 05:02:02 <rwhitby> 192.168.1.77 is the default linksys firmware ip
May 05 05:03:16 <lennert> and when r/s is permanently off and ethernet led is still blinking according to traffic?
**** BEGIN LOGGING AT Thu May 05 05:08:31 2005
May 05 05:09:11 <lennert> r/s on
May 05 05:10:10 <NAiL> rwhitby: Do you know if a max232epe would work? 
May 05 05:10:31 <rwhitby> if it's 3.3V it will work - I'm not familiar with that one
May 05 05:10:56 <dyoung> it will work, but will be flaky.
May 05 05:11:30 <NAiL> dyoung: Flaky? You've tried using one?
May 05 05:11:32 <dyoung> I powered my MAX232CPE with 3.3V and it "sorta" worked... but was flaky.
May 05 05:12:00 <lennert> ok, so, hit power, wait 17 sec, press reset 2 sec
May 05 05:12:04 <lennert> doesnt seem to do anything
May 05 05:12:40 <rwhitby> lennert: since the led are changing, that means that redboot is still intact, which means you should be able to telnet to it
May 05 05:12:48 <rwhitby> (if you get the timing right)
May 05 05:13:15 <lennert> that's 192.168.0.1, right?
May 05 05:14:57 <dyoung> the arping method works best for me.
May 05 05:15:15 <NAiL> Has worked for me every time :)
May 05 05:15:16 <lennert> ready/status now just goes on after power-up and stays on
May 05 05:15:26 <lennert> i've never managed to make this slug respond to arping 192.168.0.1
May 05 05:15:32 <lennert> not even if i arping it 5 times a second
May 05 05:23:02 <rwhitby> hey Jacmet
May 05 05:23:11 <rwhitby> hey jacques too :-)
May 05 05:23:25 <rwhitby> jacques: lennert has his slug, but is having problems with it
May 05 05:23:34 <lennert> ready/status is staying orange after power-up now
May 05 05:23:48 <lennert> it does spew out 139 characters of garbage to the serial port every time i boot it
May 05 05:23:58 <lennert> a la
May 05 05:24:02 <lennert> 0000000 df b3 73 65 7b df 8b fb 7b fb bf ff bf bf bf bf
May 05 05:24:02 <lennert> 0000020 ff ef ef ab 81 eb df fb db fb db eb fb fb db bf
May 05 05:24:02 <lennert> 0000040 fd 7f fb db ff ff fb bf ff ff fb 7f ac eb fe ef
May 05 05:25:22 <rwhitby> that's the redboot boot message,
May 05 05:25:36 <rwhitby> push the reset button for three seconds as soon as you see that
May 05 05:25:53 <jacques> my comcast just came back
May 05 05:29:49 <lennert> rwhitby: doesnt seem to do anything
May 05 05:30:24 <jacques> so this slug has been modified ?
May 05 05:30:36 <rwhitby> yeah, serial
May 05 05:30:41 <jacques> serial port added? anything else? I assume done before it was sent to lennert 
May 05 05:30:52 <rwhitby> but with a 5V level converter, powered from the serial port, not the slug
May 05 05:31:05 <jacques> hi lennert - thanks for the armeb support in qemu
May 05 05:31:16 <lennert> jacques: you're welcome *bows*
May 05 05:31:47 <rwhitby> lennert, does the boot behaviour change depending on whether the serial is connected or not?
May 05 05:31:55 <jacques> yeah good test
May 05 05:32:04 <rwhitby> and is it repeatable on each boot ?  (based on leds)
May 05 05:32:19 <jacques> surely this was tested after modification?
May 05 05:32:33 <jacques> I'm hoping it's something simple
May 05 05:32:35 <rwhitby> dunno, and stop calling me Shirley
May 05 05:32:40 <jacques> lol
May 05 05:33:49 <dyoung> putting 5V into the IXP cant be good.
May 05 05:34:20 <rwhitby> lennert, give us the repeatable led sequence on a non-serial boot, and a serial boot, and describe the spurts of chars on the serial boots
May 05 05:39:11 <lennert> no serial: r/s goes on (orange says the wife)
May 05 05:39:21 <lennert> ethernet goes on, blinks a few times
May 05 05:39:29 <lennert> ethernet goes off, and on again, blinks a few more times
May 05 05:39:35 <lennert> ready/status still the same color
May 05 05:39:49 <lennert> ethernet is continuously on but blinks off whenever there is network traffic
May 05 05:39:58 <lennert> it stays like this for at least a few minutes, afaics
May 05 05:40:14 <rwhitby> r/s doesn't start blinking?
May 05 05:40:23 <lennert> r/s does not start blinking
May 05 05:40:30 <rwhitby> (we can ignore ethernet, cause it's not controlled by the processor)
May 05 05:41:11 <rwhitby> so r/s stays orange
May 05 05:41:36 <lennert> can it even be orange?
May 05 05:41:54 <lennert> what i've read seems to suggest that it can be either green, red or yellow
May 05 05:42:25 <rwhitby> try upslug when it is orange
May 05 05:42:36 <dyoung> green + red = yellowish-orange dull glow.
May 05 05:42:51 <lennert> upslug doesn't find anything still
May 05 05:43:34 <rwhitby> orange is reboot command line
May 05 05:44:02 <lennert> so it's at the command line
May 05 05:44:03 <rwhitby> upgrade mode is red/green flashing
May 05 05:44:11 <lennert> it seems to behave the same if i have the serial plugged in
May 05 05:44:23 <lennert> but everything coming out of the serial port is just garbage
May 05 05:44:37 <dyoung> thats notnn suprising.
May 05 05:44:55 <rwhitby> dyoung: wonder if typing "upgrade" at the serial would do anything
May 05 05:45:03 <dyoung> couldnt hurt.
May 05 05:45:13 <dyoung> dont forget to control-C a few times.
May 05 05:46:00 <lennert> will try
May 05 05:46:02 <rwhitby> lennert: when you see the first spurt of characters, and then a couple of seconds pause, are there more spurts after that, or is there just one spurt at the start and then nothing
May 05 05:46:37 <dyoung> is there an idiots guide to string handling in C somewhere?
May 05 05:46:48 <lennert> dyoung: what do you want to know?
May 05 05:47:04 <rwhitby> man strstr :-)
May 05 05:47:25 <dyoung> if I have a string "+123+456+789" and I want the 456 only....
May 05 05:47:45 <lennert> memcpy(dest, original+4, 3); dest[4] = 0;
May 05 05:47:50 <lennert> or can it be variable length?
May 05 05:47:58 <dyoung> it can be variable length.  
May 05 05:48:05 <rwhitby> strchr twice
May 05 05:48:23 <jacques> use perl  (j/k)
May 05 05:48:36 <dyoung> hehe is that considered "ugly" to do strchr twice?
May 05 05:48:47 <dyoung> cuz I'm doing strtok twice in another place.
May 05 05:48:48 <lennert> dyoung: just check if you don't get NULL
May 05 05:49:10 <lennert> dyoung: and don't do things like strchr(string+1, ...) without checking strlen(string)>1 first
May 05 05:49:27 <dyoung> Yes Master!
May 05 05:49:40 <rwhitby> strong with the force, he is.
May 05 05:49:58 <lennert> but can't solder himself out of a wet paper bag
May 05 05:50:19 <dyoung> bah.
May 05 05:50:30 <lennert> dyoung: hm?
May 05 05:50:43 <rwhitby> lennert: did you see my question about the spurts of chars?
May 05 05:50:48 * dyoung looks at a pile of dyoung broken beta boards broken after dyoung soldering.
May 05 05:51:07 <lennert> rwhitby: one spurt of characters if i hammer ctrl-c
May 05 05:51:28 <dyoung> if you keep hitting control-C does it spew more stuff?
May 05 05:51:48 <dyoung> something that may look like redboot> ^c
May 05 05:51:53 <rwhitby> a little bit each time
May 05 05:52:02 <lennert> it seems like just one spurt
May 05 05:52:19 <lennert> dyoung: you're brave enough to try soldering :)
May 05 05:52:25 <rwhitby> and nothing else for further Ctl-C's ?
May 05 05:52:52 <dyoung> the initial spurt comes when you first apply power, or like a few seconds later?
May 05 05:53:06 <lennert> ok, if i stay off the keyboard:
May 05 05:53:24 <lennert> power on
May 05 05:53:27 <lennert> wait
May 05 05:53:28 <lennert> wait
May 05 05:53:36 <lennert> spurt
May 05 05:53:43 <lennert> quite long spurt was this
May 05 05:53:50 <rwhitby> that's the redboot banner
May 05 05:53:51 <dyoung> /hug rwhitby
May 05 05:53:51 <lennert> nothing anymore now
May 05 05:54:04 <dyoung> strchr is my friend.
May 05 05:54:07 <lennert> just one spurt
May 05 05:54:18 <lennert> hitting space/enter/ctrl-c doesn't show anything
May 05 05:54:25 <rwhitby> lennert: if you type stuff at it, does it respond?
May 05 05:54:38 <lennert> it doesn't respond at 115200
May 05 05:54:42 <rwhitby> try typing "upgrade" return
May 05 05:54:54 <dyoung> we need to get lennert a proper 3.3V level converter.
May 05 05:54:59 <rwhitby> the upgrade command should turn the leds alternating red and green
May 05 05:55:39 <lennert> if i turn the serial port to 9600 and keep space bar pressed, it shows me about two . characters per second
May 05 05:55:59 <lennert> but in 115200 i can't get anything out of it at all
May 05 05:56:02 <dyoung> tried 115200 7e1 ?
May 05 05:56:12 <lennert> will try now
May 05 05:56:24 <lennert> holding space bar pressed gives me
May 05 05:56:29 <rwhitby> so the spurt means that redboot is intact, but the lack of another spurt says it's not booting the kernel
May 05 05:56:48 <lennert> 115200 7e1 gives me "pp       p p       " when i hold space bar
May 05 05:56:59 <dyoung> this sounds so famliar....
May 05 05:56:59 <rwhitby> and the lack of red/green says it's not going into upgrade mode by itself
May 05 05:57:26 <dyoung> rwhitby, the kernel must be booting, he could ping .77 earlier.
May 05 05:57:29 <rwhitby> so it's hung at the "== Executing boot script in 2.000 seconds - enter ^C to abort" prompt
May 05 05:57:41 <lennert> dyoung: it's not booting now, it seems
May 05 05:57:44 <dyoung> Oh.
May 05 05:58:13 <dyoung> lennert, do you have a free USB port?
May 05 05:58:18 <rwhitby> and my slug used to do that when I didn't have serial plugged in
May 05 05:58:29 <lennert> dyoung: yes?
May 05 05:58:30 <rwhitby> (hang at that prompt)
May 05 05:59:03 <dyoung> in the slug, are those connections soldered to the header?  or can they come off?
May 05 06:00:07 <lennert> they are soldered on
May 05 06:00:19 <dyoung> Hmm.
May 05 06:00:27 <dyoung> that poses an interesting challenge.
May 05 06:00:56 <rwhitby> really, they are not just pushed onto the header near the battery?
May 05 06:01:45 <lennert> the three wires are each soldered to one pin of the header
May 05 06:01:54 <rwhitby> ah, yeah, they look soldered and heatshrunk
May 05 06:02:06 <dyoung> Hmm.
May 05 06:02:09 <lennert> with a bit of that black shrinking stuff over them
May 05 06:02:09 <lennert> yes
May 05 06:02:15 <rwhitby> we should have got someone else to do the serial :-(
May 05 06:02:20 <dyoung> Do you have a friend with a soldering iron?
May 05 06:02:34 <lennert> so the problem is that it's putting 5V into the ixp?
May 05 06:02:36 <dyoung> (that has 5 min spare?)
May 05 06:03:41 <rwhitby> lennert, can you try telnettting to redboot port 9000 at the exact moment that the first spurt comes out?
May 05 06:04:38 <lennert> dyoung: i don't think so, no
May 05 06:05:28 <lennert> rwhitby: ok
May 05 06:05:36 <lennert> i'll try another PC as well, just to be sure
May 05 06:06:13 <rwhitby> linux or windows?
May 05 06:06:38 <rwhitby> perhaps try a direct connect cross-over cable if you have one
May 05 06:07:44 <rwhitby> Now is the time to send an email to Christian too :-)
May 05 06:14:11 <rwhitby> Hmm.  This is our correspondence with Christian about the serial interface:
May 05 06:14:17 <rwhitby> >What sort of serial interface do you put on?
May 05 06:14:17 <rwhitby> >
May 05 06:14:17 <rwhitby> >
May 05 06:14:17 <rwhitby> It is a selfbuild one, one max3232, 5 caps. This would cost about about
May 05 06:14:17 <rwhitby> 4 Euro (if i use my so max package, but this does not look very
May 05 06:14:18 <rwhitby> professional) or 10 euro if i use a new max3232 in dip package(including
May 05 06:14:20 <rwhitby> 4.40Euro for the IC).
May 05 06:14:45 <rwhitby> He has done something completely different
May 05 06:15:03 <lennert> let's check my email
May 05 06:15:44 <lennert> christian wrote me:
May 05 06:15:47 <lennert> do you want a hole in your nslu2 or should i use a ribbon cable for the
May 05 06:15:47 <lennert> serial interface (through one usb hole, no drilling required then)?
May 05 06:15:50 <lennert> -- 
May 05 06:15:56 <lennert> but he didn't end up using a ribbon cable at all
May 05 06:16:12 <lennert> on 21 mar he wrote that he ordered it
May 05 06:16:20 <lennert> on 21 apr i asked him if it arrived yet
May 05 06:16:21 <macsat> Anyone here cross-compile gurus ? ;-)
May 05 06:16:28 <macsat> after doing the : cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/nslu co unslung    
May 05 06:16:34 <lennert> 5 minutes later he wrote "sorry... of course, i solder it now."
May 05 06:16:50 <macsat> where do I put the ASUS GPL tarball and OLEGS toolchain ?
May 05 06:16:59 <macsat> (wl-500g here, not NSLU2)
May 05 06:17:12 <rwhitby> macsat: you install the toolchains as per Oleg's instructions
May 05 06:17:33 <rwhitby> in /opt/bcrm/.....
May 05 06:17:38 <rwhitby> brcm
May 05 06:17:53 <lennert> hmmmmm
May 05 06:18:02 <lennert> rwhitby: remember quentin wright was going to do my slug at first?
May 05 06:18:22 <rwhitby> yeah
May 05 06:19:27 <lennert> quentin wrote:
May 05 06:19:30 <lennert> This morning I've just successfully added a serial connection to my
May 05 06:19:30 <lennert> NSLU2 with a "no brand" Siemens S35 data cable costing 10 UKP and found
May 05 06:19:30 <lennert> it was fairly easy. The task that required the most dexterity was
May 05 06:19:33 <lennert> desoldering the through-holes in the PCB at J2. To make a neat job I
May 05 06:19:33 <lennert> drilled a push-fit hole for the cable in the rear of the case below the
May 05 06:19:33 <lennert> power socket, but not so low as to foul the support foot. Then a small
May 05 06:19:33 <lennert> tie-wrap was used on the cable on the inside as a strain relief.
May 05 06:19:35 <lennert> ---
May 05 06:19:41 <lennert> but that's what christian seems to have done
May 05 06:19:48 <dyoung> Okay, I need to think about something else now... but if neccessary, lennert can send it to me and I'll make it work right.  I have boatloads of USB-Serial adapters that are slug compatible, or I have spare compsys 3232 based unit.
May 05 06:20:42 <rwhitby> lennert: no luck with telnet to redboot?
May 05 06:21:21 <rwhitby> I have one last thing to try before we send it off to someone for repair
May 05 06:22:07 <rwhitby> this one last thing involves cutting one of the wires ....
May 05 06:23:19 <macsat> and what about the original ASUS firmware tar, where do I place that ?
May 05 06:23:32 <lennert> hmmm, ready/status is blinking now
May 05 06:23:33 <lennert> now it's off
May 05 06:23:45 <lennert> hum
May 05 06:23:58 <rwhitby> macsat: you only need the tools out of it
May 05 06:24:02 <lennert> i was keeping it off for ~20 secs (without power) before turning it on again, but apparently that's not enough?
May 05 06:24:21 <macsat> okidoki
May 05 06:24:23 <rwhitby> that's usually enough
May 05 06:24:28 <macsat> so I guess that OLEGs is enough :)
May 05 06:24:40 <lennert> now ready/status is blinking
May 05 06:24:42 <lennert> now it's on again
May 05 06:24:50 <macsat> his binary package includes the toolchain and the patches as far as I can see
May 05 06:24:50 <rwhitby> macsat: no, you need the original tools, and also Oleg's additional tools
May 05 06:24:57 <macsat> hmm, k
May 05 06:25:08 <lennert> Unicast reply from 192.168.1.77 [00:0F:66:7F:68:1A]  3.196ms
May 05 06:25:22 <lennert> and now it beeped
May 05 06:25:33 <lennert> the very first time i ever heard it do that
May 05 06:25:37 <macsat> any info on how to extract only the tools from the firmware, and there to put it ?
May 05 06:25:41 <NAiL> Ugly beep, innit? ;)
May 05 06:25:45 <dyoung> hmm, you should be able to upslug discover it now.
May 05 06:25:51 <lennert> very unfriendly type of beep
May 05 06:26:03 <lennert> [root@vi buytenh]# ./upslug 
May 05 06:26:03 <lennert> Using eth0 to find slugs...
May 05 06:26:03 <lennert> Finding Slugs...
May 05 06:26:03 <lennert> 1: 00:0f:66:7f:68:1a
May 05 06:26:03 <lennert> Which slug?:
May 05 06:26:09 <rwhitby> lennert: try to telnet in normally first
May 05 06:26:20 <lennert> which port?
May 05 06:26:24 <rwhitby> normal
May 05 06:26:35 <rwhitby> 21
May 05 06:26:41 <lennert> is closed
May 05 06:26:45 <lennert> ports 80 and 139 are open, though
May 05 06:26:47 <rwhitby> try ssh
May 05 06:26:53 <lennert> last time i had it, it didn't beep and port 6789 was open
May 05 06:27:06 <rwhitby> ok, open it in your browser
May 05 06:27:09 <lennert> ssh closed too -- why would it be running that?
May 05 06:27:13 <rwhitby> must still be stock firmware
May 05 06:27:27 <rwhitby> http://192.168.1.77:80/
May 05 06:27:29 <lennert> nice web page
May 05 06:27:29 <lennert> yes
May 05 06:27:35 <lennert> V2.3R24
May 05 06:27:44 <jf-away> macsat: Asus tarball contains either firwmare source plus original libc and uclibc toolchain. you'll need from that tar the original firmware ource + lib toolchain. Olegs uclibc toolchain then replace asus uclibc toolchain
May 05 06:27:51 <dyoung> R24 even.
May 05 06:27:57 * dyoung icks
May 05 06:28:24 <lennert> dyoung: hm
May 05 06:28:29 <rwhitby> lennert: what's the status of your serial connection at the moment? plugged in? unplugged?
May 05 06:28:29 <lennert> dyoung: ?
May 05 06:28:41 <lennert> rwhitby: it was unplugged for a few minutes
May 05 06:29:01 <lennert> rwhitby: on previous attempts it was either plugged in or had been unplugged for ~5 seconds
May 05 06:29:08 <rwhitby> so you got it to boot with no serial attached for a long time
May 05 06:29:27 <lennert> rwhitby: yes
May 05 06:29:37 <lennert> rwhitby: and it actually beeped as well
May 05 06:29:43 <macsat> WTF - is the asus sources some 120+ MB ?
May 05 06:29:43 <rwhitby> that is consistent with my hung redboot prompt theory
May 05 06:29:52 <jf-away> macsat: yep
May 05 06:29:59 <macsat> ffs
May 05 06:30:00 <rwhitby> lennert: leave the serial off, and try power cycling it now
May 05 06:30:02 <lennert> rwhitby: first time i tried booting it (just after i got it) it would do a similar thing but never beep
May 05 06:30:10 <lennert> rwhitby: preferred method of power cycling?
May 05 06:30:11 <jf-away> source + toolchains (the big part)
May 05 06:30:39 <macsat> 273 MB actuately..
May 05 06:30:40 <rwhitby> lennert: try the power switch first :-)
May 05 06:30:51 <macsat> now is the time where I could use more than 512 down speed
May 05 06:30:55 <rwhitby> it should beep and power down
May 05 06:31:20 <lennert> beep it did
May 05 06:31:55 <rwhitby> and powered itself off?
May 05 06:32:02 <lennert> not yet
May 05 06:32:08 <lennert> ready/status still on
May 05 06:32:11 <lennert> ethernet blinking
May 05 06:32:15 <lennert> still multicasting to port 1900
May 05 06:32:19 <lennert> 15:32:10.651305 IP 192.168.1.77.1025 > 239.255.255.250.1900: UDP, length 323
May 05 06:32:19 <lennert> 15:32:10.761091 IP 192.168.1.77.1025 > 239.255.255.250.1900: UDP, length 359
May 05 06:32:55 <rwhitby> anyway, the flash is intact, so the weird behaviour is being caused by the botched serial
May 05 06:33:03 <jf-away> macsat: asus tarball willl uncompress under a broadcom/ directory. In that directory, you'll find a subdir named opt which contains the toolchain. you have to move this directory to root /
May 05 06:33:10 <lennert> it's still not off
May 05 06:33:22 <rwhitby> lennert: pull the power plug
May 05 06:33:36 <lennert> is it normal that it doesn't go off?
May 05 06:33:39 <lennert> dyoung: g'nite
May 05 06:33:39 <rwhitby> no
May 05 06:33:50 <lennert> pulled the plug
May 05 06:33:52 <lennert> turn it on again?
May 05 06:33:55 <jf-away> macsat: then untar Oleg's toolchain usings info @ wl500g.dyndns.org
May 05 06:33:58 <rwhitby> then power on again
May 05 06:34:03 <jf-away> 'Nite dyoung-zzzz
May 05 06:34:06 <dyoung-zzzz> rwhitby, lennert, mail me if you decide to utilize my services.
May 05 06:34:16 <lennert> r/s on, ethernet on
May 05 06:34:19 <lennert> dyoung: ok!
May 05 06:34:37 <macsat> I will jf-away, once I get it downloaded :-)
May 05 06:34:51 <jf-away> macsat: okay
May 05 06:35:06 <macsat> will take 1-2 hours on this crap connection
May 05 06:35:10 <lennert> r/s still on, didn't start blinking yet
May 05 06:35:41 <lennert> i have a feeling it's not going to boot
May 05 06:35:49 <macsat> hopefully I can manage to build a few packages....would be nice to have more people helping out in making packages, wouldnt it jf-away  ? ;-)
May 05 06:36:12 <rwhitby> lennert, got some wire-cutters handy?
May 05 06:36:18 <macsat> maby (once I get that far) you can even help me in how to build my own testing-repository ?
May 05 06:36:27 <lennert> rwhitby: tiny scissors will do?
May 05 06:36:27 <dyoung-zzzz> lennert, if you can get your hands on a soldering iron, I can send you a compatible USB/Serial adapter tommrow.
May 05 06:36:51 <jf-away> macsat: most of people here can help you, once you ask for package creation related question
May 05 06:37:04 <macsat> cool
May 05 06:37:12 <rwhitby> lennert - they have to be sturdy enough to cut through the connections just above the heat-shrink 
May 05 06:37:24 <lennert> dyoung: i'll ask around if someone i know knows someone that maybe knows someone that can solder
May 05 06:37:37 <rwhitby> macsat: we're always helpful to people who are developing new packages
May 05 06:37:45 <lennert> rwhitby: all wires?
May 05 06:37:57 <dyoung-zzzz> Package Developers are our friends.
May 05 06:38:14 <lennert> dyoung: i'll just make sure before you send me anything, ok?
May 05 06:38:22 <lennert> rwhitby: or just one of them
May 05 06:38:23 <rwhitby> lennert: one sec
May 05 06:38:31 <jf-away> macsat: ppl there are not wl500g / olegs firmware gurus, so asking for howto installing asus toolchain may not get a lot of answers
May 05 06:38:42 <lennert> rs232c is.. 5V?
May 05 06:39:23 <dyoung-zzzz> RS232 is +-12V actually....
May 05 06:39:46 <dyoung-zzzz> the TTL side of your SIPEX part is 5V.
May 05 06:39:51 <rwhitby> lennert: power off, cut the red wire only, make sure the bare wire is not touching anything, then try powering on again
May 05 06:39:51 * jf-away should start introducing wiley at chupa ;)
May 05 06:39:54 <dyoung-zzzz> The IXP is a 3.3V part.
May 05 06:40:11 <macsat> but once the asus toolchain is in place, building for NSLU2 and ASUS WL500g is the same? (except maby a few directories on the end-packages)
May 05 06:40:39 <dyoung-zzzz> WileyWare too!
May 05 06:40:40 <rwhitby> lennert: with just the white and black wires, you are no longer pumping 5V into the IXP, but you should still be able to see the serial output
May 05 06:40:54 <lennert> red wire is connected to pin 2
May 05 06:41:06 <rwhitby> I asked Oleg about "WileyWare" vs "Oleg's firmware", and he said he prefers the latter at the moment
May 05 06:41:11 <rwhitby> lennert: yes, pin 2 red
May 05 06:41:16 <jf-away> macsat: building for each platform require two different toolchain. if you build for wiley (asus wl family), you just need asus toolchain
May 05 06:41:41 <rwhitby> But we can still call it "Wiley" and "WileyWare" here :-)
May 05 06:41:42 <macsat> what is, and can, "WileyWare" compared to Olegs firmware ?
May 05 06:41:46 <dyoung-zzzz> Okay, so that remains our petname.
May 05 06:41:58 <jf-away> macsat: they are the same
May 05 06:41:59 <rwhitby> macsat, our name for Oleg's firmware
May 05 06:42:01 <dyoung-zzzz> ~wiley
May 05 06:42:03 <jbot> extra, extra, read all about it, wiley is the petname for the Asus WL-500 family of products, including the WL-HDD, WL-500gx, and WL-500g
May 05 06:42:12 <macsat> ohh ok
May 05 06:42:23 <rwhitby> like "slug" and "Unslung"
May 05 06:42:34 <jf-away> macsat: people here loves to gives petname or adjectives for everything...
May 05 06:42:36 <rwhitby> Unslung is to slug as WileyWare is to Wiley
May 05 06:42:47 * dyoung-zzzz hides
May 05 06:42:51 <perlguru> lsmod and insmod are missing. In which package is it included?
May 05 06:43:01 <jf-away> ~nevarious
May 05 06:43:06 <dyoung-zzzz> ~nefarious
May 05 06:43:06 <NAiL> modutils?
May 05 06:43:07 <jbot> nefarious is probably to be infamous by way of nastyness
May 05 06:43:08 <macsat> okidoki I see.....just dont get the connection from wl500g -> Wiley, but anyway.....;-)
May 05 06:43:21 <jf-away> *hide*
May 05 06:43:30 <dyoung-zzzz> WL without the vowels?
May 05 06:43:32 * lennert cuts red wire
May 05 06:43:38 <rwhitby> KABO
May 05 06:43:38 <jf-away> wl expands to wi ley
May 05 06:43:41 <rwhitby> oOOOOMMMMM
May 05 06:43:42 <dyoung-zzzz> *BOOM*
May 05 06:43:53 <macsat> ok
May 05 06:43:57 <lennert> i'm still here :)
May 05 06:44:20 <rwhitby> yeah, we figured 5V wouldn't stop your heart :-)
May 05 06:44:23 <NAiL> lennert: That's because there's a 30sec delay from the wire-cutting till the boom ;)
May 05 06:44:26 <lennert> lol
May 05 06:44:35 <lennert> you're a real crazy bunch :)
May 05 06:44:59 * NAiL looks around
May 05 06:45:29 <rwhitby> lennert: any boot activity sans red wire?
May 05 06:45:40 <NAiL> Hmm... How on earth am I going to cut the plastic on the slug?
May 05 06:45:51 <rwhitby> NAiL: Stanley knife
May 05 06:45:56 <dyoung-zzzz> NAiL, I used a file
May 05 06:46:48 <jaro> lennert: any boom activity sand red wire?
May 05 06:47:08 <jaro> s/sand/sans/
May 05 06:47:23 <rwhitby> this is almost as suspenseful as jacques's jtag exploits
May 05 06:47:26 <perlguru> lsmod and insmod are missing. In which package is it included? (modutils does not exist)
May 05 06:47:41 <lennert> i'm putting the thing back together now
May 05 06:47:50 * NAiL googles for Stanley knife
May 05 06:47:54 <perlguru> trying to install cups
May 05 06:47:55 <dyoung-zzzz> "put back together"......
May 05 06:48:07 <rwhitby> oh, leave it apart, cause you might have to cut the other two :-)
May 05 06:48:09 <dyoung-zzzz> perlguru, are we talking about openslug or unslung?
May 05 06:48:20 <lennert> oh, okay
May 05 06:48:24 <perlguru> unslung 4.13
May 05 06:48:38 <dyoung-zzzz> its part of the base package.
May 05 06:48:55 <lennert> so you want the serial plugged into the pc now?
May 05 06:48:58 <rwhitby> # ls -l /sbin/insmod
May 05 06:48:59 <rwhitby> lrwxrwxrwx    1 root     root           14 Apr 25 23:03 /sbin/insmod -> ../bin/busybox
May 05 06:49:06 <rwhitby> lennert: yes
May 05 06:49:17 <rwhitby> actually, try it  without first
May 05 06:49:24 <lennert> ok
May 05 06:49:25 <rwhitby> shouldn't matter now
May 05 06:49:45 <perlguru> dyoung-zzzz, which base package?
May 05 06:50:05 <dyoung-zzzz> part of the stock firmware which is instantiated into unslung
May 05 06:50:06 <rwhitby> as it should only be recieving the output from the slug now - worst that can happen is garbage on the output - can't hurt the slug anymore
May 05 06:50:11 <lennert> but... how can the red wire having been connected have made any difference when the serial wasn't plugged in?
May 05 06:50:18 <dyoung-zzzz> Pullup.
May 05 06:50:24 <lennert> ready/status starts blinking now
May 05 06:50:28 <lennert> and now the buzzer
May 05 06:50:42 <lennert> so it should shut down and power up properly now?
May 05 06:50:44 <lennert> dyoung: ?
May 05 06:50:47 <rwhitby> web?
May 05 06:50:57 <perlguru> ok, i found them.   S88cups does not pick them up. maybe path is incorrent
May 05 06:51:04 <lennert> it does shut down properly, within a few seconds
May 05 06:51:06 <lennert> power up again
May 05 06:51:13 <lennert> i'll try web
May 05 06:51:26 <dyoung-zzzz> the Serial line (sometimes) requires a pullup to VCC (3.3V) in order to function properly.
May 05 06:51:48 <lennert> dyoung: so the convertor chip was getting power that way?
May 05 06:51:55 <dyoung-zzzz> otherwise the line may float resulting in unpredictable results.
May 05 06:52:04 <lennert> beep, it successfully booted
May 05 06:52:07 <rwhitby> and a floating input can suck power
May 05 06:52:27 <rwhitby> that red wire was your problem :-)
May 05 06:52:29 <dyoung-zzzz> Okay so we now know what the problem is.....
May 05 06:52:38 <dyoung-zzzz> solution is up in the air....
May 05 06:52:41 <lennert> web works fine
May 05 06:52:54 <rwhitby> lennert: try it with serial plugged into PC
May 05 06:52:56 <dyoung-zzzz> lennert, you should be able to telnet into the thing now.
May 05 06:53:01 <rwhitby> and watch the output
May 05 06:53:03 <lennert> sorry, could one of you explain that one more time to me please?
May 05 06:53:16 <dyoung-zzzz> Okay.
May 05 06:53:29 <lennert> uhm
May 05 06:53:40 <lennert> i'm seeing characters on the serial port (in minicom) even though the nslu2 is off
May 05 06:53:55 <lennert> the power adapter is plugged in
May 05 06:54:04 <rwhitby> that's weird
May 05 06:54:05 <dyoung-zzzz> Integrated Circuits may have 3 possible states.  High (VCC), Low (gnd) and "undefined".
May 05 06:54:13 <lennert> ...o?=.~......P.I.."..o?=.o?=o?=......o?=......
May 05 06:54:18 <lennert> dyoung: yes
May 05 06:54:50 <lennert> rwhitby: i turned it on and it outputs more garbage now
May 05 06:54:51 <dyoung-zzzz> now usually you want to specifically define what your inputs and outputs are to prevent unpredicatble results, and to save power.
May 05 06:54:52 <lennert> and things like
May 05 06:54:54 <lennert> .#{###.NetWinder Flgatif# Point Em5$!.#...^97 (double
May 05 06:54:57 <lennert> nt [X1226] registered to adaptero?=.o?=o?=o?={Uo?=.
May 05 06:55:10 <rwhitby> ok, that's garbled boot messages
May 05 06:55:11 <dyoung-zzzz> Ahh, thats the kernel booting.
May 05 06:55:12 <NAiL> Well.. it's boting ;)
May 05 06:55:17 <lennert> omiscuous mode                     
May 05 06:55:32 <lennert> dyoung: yes
May 05 06:55:53 <lennert> .|..........#W.##u###.....j..+.W. .a#.......#@##....@XYY..rom flash to RAM
May 05 06:55:54 <lennert> .......#..#..#u}#{u}u###]o..55R.......        P.Hh.
May 05 06:55:56 <dyoung-zzzz> and the serial input into the IXP should be pulled up to VCC.
May 05 06:55:57 <lennert> #th 0x00.d39#.5##...#d..#.# .B! P.Hh.
May 05 06:56:02 <lennert> HT.a.H...$.S.V..*..5....U.o?=..o?=Z.o?=.o?=o?=o?=c01b9f8c to?=}co?=5
May 05 06:56:04 <lennert> kuo?=o?=o?=-.o?=o..             .&$HH.         .A.2..$..        "D...@00000000 c01d26e8 
May 05 06:56:04 <lennert> 00000549 
May 05 06:56:04 <lennert> 9fa0:o?=o?=.........
May 05 06:56:04 <lennert> 9fe0: ......E".....~..4.WWwo?=_Wo?=[8;33HSo.o?=o?=&M.c01ebc34 c01c6fec o?=}o?=..o?=xtdd to ko?=[8;1Hmmo?=umuo?={].]o?=o?=m.-HK.oot 3yncing..n
May 05 06:56:12 <rwhitby> lennert: try and get repeatable telnet into redboot now
May 05 06:56:35 <rwhitby> if you can get repeatable telnet into redboot, then we can cut the serial cable off completely for the moment
May 05 06:56:37 <dyoung-zzzz> the specific reasons for that are beyond what I can explain right now because its way past my bedtime.
May 05 06:57:06 <lennert> dyoung: so there's a resistor on the pcb between +3.3V and the serial input on the J2 ?
May 05 06:57:17 <dyoung-zzzz> aha, excellent point.
May 05 06:57:24 <dyoung-zzzz> there isnt one, but there should be.
May 05 06:57:50 <dyoung-zzzz> and more than one person has experienced "bizzare behaviour" when one is not installed.
May 05 06:57:57 <lennert> dyoung: ok
May 05 06:58:23 <rwhitby> lennert: a floating input can oscillate between 1 and 0, and that can cause the IXP to see garbage on the serial input, causing framing errors.  I expect that RedBoot (whilst looking for Ctrl-C at the boot wait banner message) must get confused by the serial errors and just hangs.
May 05 06:58:40 <lennert> rwhitby: ok, makes sense
May 05 06:58:50 <dyoung-zzzz> There you go.  Oscillate.
May 05 06:58:57 <lennert> hmmm
May 05 06:58:59 <rwhitby> Once we remove the long antenna (the red wire) which is picking up all sorts of EMI and induced voltages, then it settles to a stable value
May 05 06:59:10 <lennert> it's solid orange now
May 05 06:59:14 <lennert> so again it's stuck in redboot?
May 05 06:59:16 <lennert> hm :(
May 05 06:59:28 <rwhitby> yep
May 05 06:59:40 <lennert> rwhitby: why is that still happening?
May 05 06:59:43 <lennert> Unicast reply from 192.168.0.1 [00:0F:66:7F:68:1A]  9.825ms
May 05 06:59:43 <lennert> Unicast reply from 192.168.0.1 [00:0F:66:7F:68:1A]  3.214ms
May 05 06:59:48 <lennert> i should be able to get into redboot now
May 05 07:02:04 <lennert> i'm in redboot
May 05 07:02:38 * dyoung-zzzz zzzz's
May 05 07:02:56 <rwhitby> was it easy enough that you think you can repeat it without looking at the serial output for timing?
May 05 07:03:09 <lennert> i didn't look at the serial output
May 05 07:03:19 <lennert> just arping
May 05 07:03:26 <lennert> i'll try again, but i think so
May 05 07:04:06 <lennert> yeah, it's fairly easy to do
May 05 07:04:20 <lennert> want me to cut the other wires too?
May 05 07:04:22 <rwhitby> ok, cut the other two wires (from pin 3 and 4)
May 05 07:04:40 <rwhitby> free your slug from the tyranny of a botched serial port!
May 05 07:04:48 <dyoung-zzzz> lennert, if you have a friend of a friend with a soldering iron that can clean up the header, mail me your address later and I'll send you one of these USB/Serial things.
May 05 07:05:16 <dyoung-zzzz> or a CompSys serial Adapter.
May 05 07:05:22 <dyoung-zzzz> your choice.
May 05 07:05:37 <lennert> it's free!
May 05 07:05:37 <rwhitby> lennert: if one was to desolder the stuff inside the shrink-wrap off the header pins, do you think there would be enough header pin sticking up to plug something onto it?
May 05 07:06:06 <rwhitby> i.e. are they all the same length as the header pin on pin 1 ?
May 05 07:06:30 <rwhitby> (which looks like it is on an angle, but has enough length to plug something onto it)
May 05 07:06:37 <rwhitby> (in your photo
May 05 07:07:14 <lennert> pin 2 has about 8mm of pin sticking out
May 05 07:07:23 <lennert> the rest slightly longer
May 05 07:07:28 <lennert> with pieces of solder still on there
May 05 07:07:42 <rwhitby> ok, that should be enough.
May 05 07:07:57 <rwhitby> you need to find someone who can get the solder off the pins, without removing the pins from the PCB
May 05 07:07:59 <lennert> i should be able to get the solder off in some way
May 05 07:08:09 <rwhitby> then dyoung will send you something to plug into the 4 pin header
May 05 07:08:16 <lennert> right
May 05 07:08:23 <lennert> that would be great
May 05 07:08:37 <rwhitby> and we ask Christian to pay for the shipping :-)
May 05 07:08:55 <lennert> rwhitby: hmmm :)
May 05 07:09:05 <dyoung-zzzz> I should be able to cram that into the envelope.
May 05 07:09:17 <lennert> dyoung: you don't have my address?
May 05 07:09:22 <dyoung-zzzz> "the envelope" is $5 of shipping.
May 05 07:09:24 <rwhitby> cause if we knew he was going to botch it, then we would have got Quentin or Tiersten to do it
May 05 07:09:37 <dyoung-zzzz> no, but you were gonna email it to me. ;-)
May 05 07:09:57 <lennert> then you gonna tell me your email address once again
May 05 07:10:04 <lennert> rwhitby: well.. yeah
May 05 07:10:15 * dyoung-zzzz bahs
May 05 07:10:20 <dyoung-zzzz> okay I'll mail it to you in a sec
May 05 07:10:28 <lennert> rwhitby: then again, we didn't pay him
May 05 07:10:38 <rwhitby> yeah we did
May 05 07:10:43 <lennert> rwhitby: do we couldn't really expect anything
May 05 07:10:48 <lennert> rwhitby: yeah, ok, for the parts
May 05 07:11:18 <dyoung-zzzz> sent.
May 05 07:11:50 <rwhitby> oh, right.  it's ok for him to be generous and all that, but he specifically said he was going to do something (which we took at good faith and chose him to do the job) and then did something else.  And we had other volunteers to do the right job.
May 05 07:12:11 <lennert> dyoung: sent
May 05 07:12:18 <lennert> rwhitby: you're right, i guess
May 05 07:12:27 <rwhitby> lennert, so booting is repeatable when the slug is free?
May 05 07:12:50 <lennert> rwhitby: trying
May 05 07:12:58 <dyoung-zzzz> assuming I dont oversleep, it goes out tommorow.
May 05 07:13:12 <lennert> dyoung: thanks
May 05 07:13:23 <rwhitby> lennert: you feel gratitude to him cause he sent you a slug.  But he deceived me, so I don't need to feel gratitude.
May 05 07:14:07 <dyoung-zzzz> lennert, which do you prefer?  USB/serial adapter or "standard" serial adapter?
May 05 07:14:27 <lennert> dyoung: no preference, really..
May 05 07:14:31 <NAiL> rwhitby: I've cut a hole for the display now, mostly. Upload a pic now ;)
May 05 07:14:34 <rwhitby> lennert: get the USB/serial
May 05 07:14:38 <lennert> rwhitby: you're right
May 05 07:14:45 <rwhitby> cause then you can have another slug or an asus router control it
May 05 07:14:50 <dyoung-zzzz> Youre gonna get a USB one then, because I have boatloads of them.
May 05 07:15:00 <lennert> rwhitby: bit sad that it went this way
May 05 07:15:42 <lennert> rwhitby: will you contact him?
May 05 07:16:02 <lennert> dyoung: okay
May 05 07:16:38 <rwhitby> lennert: I think it's best to just call it bad luck and move on.
May 05 07:16:53 <lennert> rwhitby: ok, i thought you were really going to contact him
May 05 07:17:04 <NAiL> rwhitby: If you wanna see what I want, check http://slug.repvik.info/lcd.png :)
May 05 07:17:06 <lennert> rwhitby: my thought was just to let it pass
May 05 07:17:33 <lennert> it booted and powered down fine four times in a row now
May 05 07:17:40 <lennert> definitely better than what it was
May 05 07:17:56 <lennert> thanks for all your help
May 05 07:17:57 <rwhitby> cool.  Now, are you gonna load Unslung 4.x or OpenSlug 1.x ?
May 05 07:18:23 <lennert> not decided yet :)
May 05 07:18:29 <dyoung-zzzz> openslug.
May 05 07:18:36 <lennert> i've been out of the loop for a while, i'll read up on everything
May 05 07:18:43 <lennert> a 2.6 kernel would be my preference ;)
May 05 07:18:46 <rwhitby> oh, yeah, openslug, to fix the EHCI problem :-)
May 05 07:18:55 <dyoung-zzzz> gotta hunt down that "nobody cared" issue...
May 05 07:19:00 <rwhitby> yeah
May 05 07:19:08 <dyoung-zzzz> and Lennert is the MAN after all....
May 05 07:19:31 <rwhitby> lennert thinks we're all altruistic and all that, but now he learns that that slug has some strings attached :-)
May 05 07:19:35 <NAiL> It's the only problem I'm having with openslug, so ;)
May 05 07:20:23 <dyoung-zzzz> ok ttfn.
May 05 07:20:56 <NAiL> ta ta
May 05 07:20:56 <lennert> rwhitby: i love good challenges :)
May 05 07:20:57 <rwhitby> what lennert doesn't know is that I asked Christian to botch it, so that lennert would have to come here for help, and then feel so much in gratitude that he would just have to fix the EHCI and "nobody cared" issues ... ;-)
May 05 07:21:06 <lennert> rwhitby: ah :))
May 05 07:21:41 <dyoung-zzzz> ~nefarious
May 05 07:21:42 <jbot> well, nefarious is to be infamous by way of nastyness
May 05 07:21:49 <lennert> so let's see.. pci interrupts are implemented with GPIOs?
May 05 07:21:54 <rwhitby> dyoung-zzzz: he hasn't realised yet that the whole nslu2-linux project is just a way for us to get other people to fix our stuff.
May 05 07:22:09 <dyoung-zzzz> rwhitby, Uh huh. ;-)
May 05 07:22:29 <lennert> so you work for linksys, hm? :)
May 05 07:22:55 <rwhitby> nope
May 05 07:23:01 <dyoung-zzzz> I think we're gonna have to export the openslug source tree for lennert (and others) to look at.
May 05 07:23:05 <lennert> can't it just be that it's going
May 05 07:23:10 <lennert> pci interrupt
May 05 07:23:15 <lennert> enter interrupt handler
May 05 07:23:18 <lennert> another pci interrupt
May 05 07:23:24 <lennert> interrupt handler finishes up all pending work
May 05 07:23:25 <dyoung-zzzz> rwhitby, whats the bk command to do that?
May 05 07:23:27 <lennert> interrupt handler exits
May 05 07:23:34 <lennert> interrupts handler immediately gets reentered again
May 05 07:23:40 <lennert> no work to do -> error messsage
May 05 07:23:41 <lennert> ?
May 05 07:23:46 <rwhitby> dyoung-zzzz: we have an rsync repo don't we?
May 05 07:23:50 <dyoung-zzzz> remind me later and I'll put a tarfile for lennert to download.
May 05 07:23:53 <NAiL> lennert: Isn't that your job to figure out? ;)
May 05 07:23:57 * NAiL has got no idea
May 05 07:24:16 <dyoung-zzzz> I think the rsync repo broke.  but I dunno
May 05 07:24:26 <lennert> NAiL: i will :)
May 05 07:24:40 <rwhitby> ~worship lennert
May 05 07:24:44 * jbot genuflects before lennert.  "I'm not worthy!"
May 05 07:24:50 <dyoung-zzzz> ~worship lennert
May 05 07:24:57 <dyoung-zzzz> ~worship lennert
May 05 07:24:57 * NAiL bows before lennert
May 05 07:24:59 * jbot kneels before lennert.  "I'm not worthy!"
May 05 07:25:25 <lennert> oh well :)
May 05 07:26:14 <NAiL> rwhitby: Did you see the picture? What do you think? ;)
May 05 07:26:18 <lennert> if i would be the genius you think i am, i would have thought of cutting those wires myself, wouldn't i? ;)
May 05 07:26:28 <NAiL> Can't connect it until I get that hub from dell, so it'll take a while
May 05 07:26:46 <lennert> i think it's hung in redboot again after the 7th boot attempt
May 05 07:26:48 <rwhitby> NAiL: nice work!
May 05 07:28:16 <lennert> maybe it's still picking up some noise?
May 05 07:28:39 <rwhitby> shouldn't be without any wires attached
May 05 07:29:24 <rwhitby> NAiL: do you reckon you've got enough room inside the case?
May 05 07:30:03 <NAiL> rwhitby: Don't think that'll be a problem. At most it's 2-3mm inside the case
May 05 07:30:14 <rwhitby> lennert: can you send me an ssh2 rsa or dsa key? and a preferred login name
May 05 07:30:45 <NAiL> (actually, I think I can fit *another* display just above this one)
May 05 07:30:46 <rwhitby> we're gonna give you an account on the official build machine so you can see the source tree
May 05 07:32:33 <lennert> sent
May 05 07:32:35 <lennert> great, thanks
May 05 07:32:49 <lennert> i've got to run now for my daily exercise
May 05 07:33:24 <rwhitby> ok, I'm setting up the account now
May 05 07:33:50 <lennert> now it boots properly
May 05 07:33:58 <lennert> it just needs to be off power for ~1 min
May 05 07:34:00 <lennert> ok, gotta run!
May 05 07:42:53 <NAiL> hmm
May 05 07:43:05 <NAiL> Has anyone compiled gphoto2 on openslug?
May 05 07:44:34 <NAiL> Sounds like a nice way to rid myself of my dependency on windows for my Ixus :)
May 05 07:45:12 <rwhitby> NAiL: put a hires LCD in instead of the one you have, and then you can preview the pictures on the slug
May 05 07:45:50 <NAiL> hah! Don't need hires! I'll use AALIB! ;)
May 05 07:46:21 <NAiL> well... preview would be nice, but that's a bit overkill ;)
May 05 07:50:26 <rwhitby> night all
May 05 07:51:20 <rwhitby-asleep> jp30, I just promoted libusb, libtopfield and ftpd-topfield for nslu2
May 05 07:51:23 <rwhitby-asleep> (cross)
May 05 07:51:51 <jp30> hi rwhitby, good, want me to push them
May 05 07:52:56 <rwhitby-asleep> anytime in the next 8 hours will be fine :-)
May 05 07:53:08 <rwhitby-asleep> libusb seems to work now
May 05 07:53:14 <jp30> i'll take a look
May 05 07:53:15 <rwhitby-asleep> but I didn't change anything
May 05 07:53:23 <jp30> mysterious
May 05 07:53:29 <jp30> what was broken about it before?
May 05 07:53:42 <rwhitby-asleep> endianness
May 05 07:53:55 <jp30> ah
May 05 07:54:26 <rwhitby-asleep> see you on the other side of the zzzzz's.
May 05 07:54:33 <jp30> 'night
May 05 07:55:07 <ByronT-Away> rwhitby-asleep, before I head into work... notification from berlios,de - project approved
May 05 07:56:22 <rwhitby-asleep> cool.  one or boh?
May 05 07:56:25 <rwhitby-asleep> both?
May 05 08:16:02 <caplink811-away> jp30: if i changed the software version of a package (openldap goes from 2.23 to 2.26), is it necessary to put it back in the section ready_for_testing?
May 05 08:17:05 <jp30> hi, caplink811; no, its not necessary, but it might be a good way of calling my attention to it
May 05 08:19:55 <caplink811-away> jp30: thx
May 05 08:21:08 <jp30> ...likewise if you make any large sweeping change to a .mk file
May 05 09:17:13 <jp30> morning, pTweety
May 05 09:17:26 <pTweety> morning
May 05 09:17:48 <macsat> jf-away, what excately do you mean by the test of /usr/sbin/test ?
May 05 09:18:03 <pTweety> my develslug arrived, but I'm having issues with unslung 4.18 alpha
May 05 09:18:46 <jp30> pTweety, what was the reasoning behind having unslung-devel have the same version number as the unslung firmware?
May 05 09:19:35 <pTweety> well, I thought it can be a pointer to the firmware version of the official build system
May 05 09:20:02 <pTweety> but you can choose any versioning sheme you want to have
May 05 09:20:10 <pTweety> I don't mind at all
May 05 09:20:36 <jp30> ok, the idea of a pointer to the firmware version sounds good
May 05 09:21:23 <jp30> i'll promote it now. sorry to hear about your issues with unslung 4
May 05 09:21:52 <pTweety> I used it to setup my develslug with firmware 3.18 beta and it was a nice job with the new unslung-devel package
May 05 09:22:18 <pTweety> but I'll try with 4.x later again
May 05 09:26:06 <jp30> jf-away, rwhitby-asleep: there were some makefile-variable issues with adduser. i've fixed them and documented some guidelines for cross-package variable references in the BestPractices wiki page
May 05 09:37:11 <jf-away> macsat : 'test' seems broken on wileyware
May 05 09:49:24 <macsat> oleg answered....he dont agree 100% :)
May 05 09:52:37 <jf-away> macsat: it's ok now.; /usr/sbin comes befoire /usr/bin in the path so 'test' refer to /usr/sbin/test which is not the standard one
May 05 09:54:00 <jf-away> '/usr/bin/test is correct but as package doesn't use the full path we may have to wait for Olegs next fw to have clean package installation/startup script
May 05 09:59:50 <jf-away> jp30-work: I see your change in adduser.mk. Could you explain me when you have time why this is now correct and what's wrong with mine (personal curiosity & skill extension) ? ty :)
May 05 10:08:18 <macsat> where can I find the builds of his FW that he refers to? The one "before the -5 release"
May 05 10:08:19 <macsat> ?
May 05 10:08:32 <macsat> ..and is there a changelog anywhere for thoose?
May 05 10:47:20 <jf-away> macsat: dunno.
May 05 10:53:52 <jf-away> macsat: http://wl500g.dyndns.org/WL500g-1.9.2.7-4a-pre14.trx
May 05 10:58:18 <jf-away> cia's back ? nice !
May 05 11:13:02 <jp30-work> jf-away, the trouble with your $(ADDUSER_BUILD_DIR)/.configured rule was that $(BUSYBOX_SOURCE) is not yet defined when the rule is parsed (make has not read busybox.mk yet), so it cannot appear in the prerequisites
May 05 11:14:57 <jp30-work> in general, variables in makefile heads are dangerous because they are evaluated during parsing
May 05 11:15:44 <jp30-work> by contrast, variables in the body of a makefile rule are evaluated when the rule is invoked, after all the makefile has been parsed
May 05 11:29:01 <CIA-1> 03caplink811 * 10unslung/Makefile: clamav added to section native_ready_for_testing, and stated while it not compile cross
May 05 12:22:34 <jf-away> jp30-work: is it because adduser.mk is included before busybox.mk ? So your solution is a solution not to depend on alphabetic order inclusion ?
May 05 12:26:01 <jf-away> and refering own package variables in a Makefile header is not critical as far as they have been defined in the makefile, before the header where it appear ?
May 05 12:27:19 <jf-away> jp30-work: did my package use to break at building in your  build system ?
May 05 12:27:30 <jp30-work> jf-away, yes and yes. of course by "makefile head" i meant "makefile rule head" i.e. the target and prerequisites
May 05 12:27:56 <jp30-work> adduser broke on my personal build system because it didn't download busybox
May 05 12:28:24 <jf-away> ok. It didn't on my caus I didn't clean sources/ - as usual :(
May 05 12:29:25 <jf-away> so you can't have prerequisite in a target referring an other package, that's right ?
May 05 12:30:06 <jp30-work> not unless you're prepared to make assumptions about the order in which makefiles are included
May 05 12:30:45 <jp30-work> it's ok to refer to a target defined anywhere, just not a variable
May 05 12:31:33 <jp30-work> e.g. "a: b, c" can occur before "b: d, e", but "a: $(b)" can't occur before "b=xyz"
May 05 12:40:56 <copperbeech> is there anyone here with a cross build environment at their fingertips and 10 minutes to spare?
May 05 12:43:17 <jp30-work> copperbeech, what do you want tested?
May 05 12:44:05 <copperbeech> jp30: appweb has just released version 2 and I want to get someone to test a make file on the cross build...
May 05 12:44:41 <jp30-work> what changes are needed from the cvs version appweb.mk? just bumping the version?
May 05 12:45:18 <copperbeech> can I stick it somewhere where you can pick it up easily?
May 05 12:46:15 <jp30-work> email it as an attachment to me: jbparsons@ucdavis.edu
May 05 12:50:11 <copperbeech> jp30: on it's way - much appreciated
May 05 12:52:01 <copperbeech> jp30: have you been involved in the php work for apache and thttpd ?
May 05 12:52:56 <jp30-work> copperbeech, i can't read a application/ms-tnef attachment. better try sending again as text; yes i have been involved in php, i'm the main contributor of php and apache on unslung
May 05 12:55:21 <copperbeech> jp30: great I really would like to pick your brains about this...
May 05 12:58:38 <jp30-work> hmm... now the .mk has messed up line breaks... i have a lunch meeting now.
May 05 12:59:30 <jp30-work> ...can you try again with the attachment if you can get it to send in some way that readable on a linux system, or put the file on a web server and send me the uri
May 05 12:59:49 <jp30-work> i'll be back in an hour or so and will look at it then
May 05 13:02:40 <copperbeech> jp30: Thanks here's the url: http://home.copperbeech.me.uk/download/appweb.mk
May 05 14:36:15 <jp30-work> copperbeech, your version of appweb dies during "make appweb" with "ld: cannot find -lappWebStatic"
May 05 14:37:08 <copperbeech> jp30: hmmm it seems to be having a problem locating a library to link...
May 05 14:38:08 <copperbeech> if it's building some of it's own libraries as part of the process will it need to specify the path to these libraries anywhere in the appweb.mk file? 
May 05 14:38:40 <copperbeech> The funny thing is that it worked with the previous version of appweb, and I wouldn't have thought things would have changed that much...
May 05 14:39:08 <copperbeech> jp30: thanks for testing the cross build for me...
May 05 14:39:21 <jp30-work> that shouldn't be necessary. in fact there is no libappWebStatic that i can see in the build tree
May 05 14:40:10 <jp30-work> there is a libAppWeb though
May 05 14:40:15 <copperbeech> hmmm true! I just checked my build tree and you're right.
May 05 14:40:23 <copperbeech> yeah
May 05 14:40:53 <jp30-work> you did configure with --disable-static, perhaps that is the problem
May 05 14:41:41 <copperbeech> on my build it failed with a: undefined reference to `MprThread::MprThread[in-charge](void (*)(void*, MprThread*), int, void*, char*, int)'
May 05 14:42:49 <copperbeech> that configure was carried across from the previous version 1.2.3. The question is why is it trying to link a static library when it was configured static diabled?
May 05 14:43:29 <copperbeech> hmmm I need to keep looking into this...
May 05 14:44:11 <copperbeech> jp30: do you have a few moments to answer a couple of questions about php?
May 05 14:44:33 <jp30-work> i don't know... but this is not a libtool based package. --disable static might not have the same semantics it does with libtool and automake
May 05 14:44:41 <jp30-work> about php: fire away
May 05 14:45:11 <copperbeech> ok it's related to appweb... I would like to get php working with it... 
May 05 14:45:25 <jp30-work> ok...
May 05 14:45:39 <copperbeech> it seems fairly straight forward (perhaps) : http://www.appwebserver.org/products/appWeb/doc/common/source/packages.html#php
May 05 14:46:28 <copperbeech> jp30: I would like your first thoughts on it with your php build experience... would it be difficult do you think?
May 05 14:47:23 <jp30-work> depends on how php and appweb interact - let me take a look at that web page
May 05 14:53:29 <jp30-work> hmm... looks like it uses the php embed sapi. it might make sense to have php.mk build and stage the embed sapi, then have appweb pick that up from staging
May 05 14:53:46 <jp30-work> that way you wouldn't have to rebuild php
May 05 14:55:34 <jeanfabrice> hi jp30-work: ty for answering my questions and correcting adduser.mk
May 05 14:55:52 <jp30-work> no probs, jf
May 05 14:56:14 <macsat> was wondering (now that I saw some adduser stuff ;-) ), is it possible to add more shellusers to wl500g, or am I stuck with one ? :)
May 05 14:56:55 <copperbeech> jp30: ok - would you say that's a fairly simple tweak to the existing php make file? 
May 05 14:57:01 <macsat> (I could run debian or something on it, I know.....but with Wiley...
May 05 14:57:08 <jeanfabrice> macsat: you can add as many as you want till they survive the reboot
May 05 14:57:26 <copperbeech> jp30: Should I create a new file for appweb specifically?
May 05 14:57:36 <macsat> so no way to store multiple user accounts in the flash area ?
May 05 14:58:22 <macsat> would be cool with /usr/local/etc/passwd  or something :)
May 05 14:59:16 <jp30-work> copperbeech, no, i don't know if it will be simple. i'd need to experiment
May 05 14:59:19 <jeanfabrice> link /etc/passwd to /usr/local/etc/passwd then add /etc/passwd to /usr/local/.files
May 05 15:00:20 <copperbeech> jp30: ok thanks for looking at it - I'll continue trying to get 2.0 to build in the meantime :-)
May 05 15:04:33 <macsat> how do I generate the crypted passwords then jeanfabrice  ?
May 05 15:05:31 <jeanfabrice> macsat: passwd <username>
May 05 15:05:41 <macsat> ok
May 05 15:05:42 <macsat> tnx
May 05 15:05:51 <macsat> kindda obvious ;)
May 05 15:06:14 <macsat> and /usr/local/.files is just a file containing a list of files that is stored in flash ?
May 05 15:06:59 <jeanfabrice> jp30-work: did you follow the discussion about creating user in rc script for daemons optware ?
May 05 15:07:37 <jp30-work> jf, no didn't, though i gathered that was the purpose of adduser. what was the upshot of the discussion?
May 05 15:08:33 <jeanfabrice> well, the question was first on user creation in postinst script and how to make it survive the reboot
May 05 15:08:55 <jaro> May I ask where does the Oleg based firmware holds the initialization scripts? Is there any way how to customize processess which are to be started on boot? I've noticed the support ov rcS, but it doesn't seeem to exist anywhere in the target filesystem produced by the compilation process.
May 05 15:09:31 <macsat> jaro : scripts /usr/local/sbin/post-boot and /usr/local/sbin/post-firewall  are ran just after boot
May 05 15:09:32 <jeanfabrice> rwhitby-asleep did preconize having rc script recreating the user at each boot, which seems a good/simple/easy solution
May 05 15:09:53 <macsat> I believe that /usr/local/sbin/pre-boot would be run just after kernal load.....correct jeanfabrice  ?
May 05 15:10:14 <jp30-work> it's complicated to make persistent passwd changes on unslung at least
May 05 15:11:41 <jeanfabrice> jp30-work: the same for wileyware. Daemons optware or any optware which needs a user should then : 1)depend on adduser and 2) provide a rc script where user creation is done
May 05 15:11:41 <macsat> true, a script could do job jeanfabrice...what did you mean with : /usr/local/.files   though ?
May 05 15:12:34 <jeanfabrice> jp30-work: and maybe 3) having postinst create the user for some chown chgrp needs
May 05 15:13:07 <jp30-work> ok; i'm not convinced that we have any daemons yet that need their own users
May 05 15:13:57 <macsat> hmm....adduser dosnt seem to be a valid command here....is that in a ipkg package or...?
May 05 15:14:06 <jeanfabrice> jp30-work: I know, I know.. eggdrop needs one, really. and caplink/ka6sox found some
May 05 15:14:57 <jp30-work> all this complication is a heavy price for a little extra security on a system that's not very secure to begin with. but there's no harm in having the functionality there if people want it
May 05 15:15:06 <jeanfabrice> jp30-work: I tried suid eggdrop. didn't work
May 05 15:15:34 <macsat> I'm just playing around jp30, I dont NEED an extra user, but it would be nice to have the possibility....:-)
May 05 15:15:54 <jaro> macsat: /usr/local/ - this portion of filesystem seems to be contained in the ramfs. But where does the image for ramfs come from?
May 05 15:16:27 <jp30-work> macsat, the adduser ipk is in the feeds if you want it.
May 05 15:16:47 <macsat> what do you mean "come from" ? It is generated when you flash your router with olegs latest firmware I supposed jaro
May 05 15:16:54 <macsat> tnx jp30, found it :)
May 05 15:17:10 <jp30-work> jf, it sounds like we agree - the functionality should be there; daemons should avoid it if they don't need it
May 05 15:17:39 <macsat> Im on my way to bed....but can anyone before I leave what jf meant with the /usr/local/.files thingy ?  I dont know that feature :)
May 05 15:18:10 <jaro> Yes. That is true. But I want to know, where exactly does the ramfs image lies in the file tree which is to be compressed into trx ;)
May 05 15:18:20 <jeanfabrice> jp30-work: btw : did you remember my problem with pear and the broken 'test' command ? well, that is because test is not the real 'test' but, as I'm root, I've /usr/sbin before /usr/bin in the PATH and the broken test sits in /usr/sbin/test
May 05 15:18:29 <macsat> I have absolutely no idea jaro :)
May 05 15:19:05 <jp30-work> jf, is /usr/sbin/test some other program by the same name, or is it a broken version of the standard unix test?
May 05 15:19:24 <jaro> macsat: however I've produced something sligtly different from what you describe, because /usr/local contains only 1 empty directory : root
May 05 15:19:38 <jeanfabrice> jp30-work: oleg's told me it is the Asus_test command ;)
May 05 15:19:41 <jaro> macsat: nothing more, no sbin/pre-boot
May 05 15:21:01 <jp30-work> jf, i guess the question is, should this be the job of php to fix, or of wileyware. a lot of standard bourne shell scripts will fail miserably if "test" does not mean test
May 05 15:21:27 <jp30-work> we could easily patch php to use "[" instead, but the problem would just show up elsewhere
May 05 15:21:51 <jeanfabrice> non root user normally don't have sbin in their path
May 05 15:22:00 <macsat> I know jaro....just create the /sbin dir, and the files :)
May 05 15:22:07 <jp30-work> right, but even root needs to run bourne scripts
May 05 15:22:29 <jeanfabrice> this is just a poor excuse, just to justify having unprivilege account runniing daemons
May 05 15:22:46 <macsat> for a "demo" of what the post-boot and post-firewall can do jaro, take a look in the wiki pages
May 05 15:22:59 <macsat> on the chupa.nl forums I mean
May 05 15:23:17 <macsat> also I made a couple of small examples on www.macsat.com/wlguide/
May 05 15:24:17 <jp30-work> jf, :)
May 05 15:25:07 <jp30-work> jf, /opt/bin/pear needs to be able to be run by root
May 05 15:25:12 <jeanfabrice> jp30-work: anyway, Oleg said it had drop asus 'test' out of its next release fw
May 05 15:26:09 <jp30-work> ok, i think that if users complain we explain that it is a known fw bug and tell them to make sure /usr/sbin is not in their path
May 05 15:26:12 <jeanfabrice> jp30-work: ok, no pb. shouldn't it redefine the path at its beginning ?
May 05 15:27:31 <jeanfabrice> Oleg just publish its next release fw. everything sould be ok now ;)
May 05 15:27:55 <jeanfabrice> "ASUS test program is no longer installed, so busybox version is now used"
May 05 15:28:43 <macsat> I'm off....cu tomorrow
May 05 15:28:45 <jp30-work> i think that pear doesn't bother to redefine the path because it doesn't think it needs to run anything "special" 
May 05 15:29:28 <jp30-work> on many shells, "test" is built-in
May 05 15:29:40 <jaro> macsat: The other thing I'd like to figure out is how all those stupid-ftpd / httpd and other stuff spawns itself upon boot. I suppose this to be caused by some "binary busybox magic" or am I wrong and some modifiable script for these services exists as well?
May 05 15:30:11 <jp30-work> jaro, look in /opt/etc/init.d
May 05 15:30:53 <ka6sox-office> jeanfabrice, is there a version of oleg's software that I *should* be putting into my WL-HDD?
May 05 15:31:26 <jeanfabrice> ka6sox-office  last one just released
May 05 15:31:47 <jaro> jp30-work - nothing like this exists on my wlhdd.
May 05 15:32:12 <ka6sox-office> jeanfabrice, thanks.
May 05 15:32:31 <jeanfabrice> ka6sox-office: running 1.9.2.7-4 without any problem.  1.9.2.7-5 just appear and fix a great bug : Asus test program was buggy !
May 05 15:33:04 <jp30-work> copperbeech, php embed sapi seems to build very easily. i'm going to check that building it doesn't break php at runtime, and then stage and package it.
May 05 15:33:53 <copperbeech> jp30:that would be great! I'll test it with the old version of appweb.
May 05 15:34:22 <ka6sox-office> jeanfabrice, ah...has there been any progress on the WL500G DLX bootstrap and other issues or are we still avoiding it?
May 05 15:34:35 <jp30-work> copperbeech, i will let you know when i have the embed sapi support tested and checked into cvs
May 05 15:35:55 <copperbeech> ok - it's almost midnight here and I'm in 2 minds as to whether I am really keen enough to see what happens with the election to stay up to watch it
May 05 15:36:10 <jeanfabrice> ka6sox-office: never take a look at it as I only own a cheap wl500g
May 05 15:36:17 <jp30-work> copperbeech, the UK election?
May 05 15:36:22 <copperbeech> yeah
May 05 15:36:34 <copperbeech> 2 results in so far
May 05 15:37:20 <jeanfabrice> ka6sox-office : plus a BIG NICE 16MB cache / 80GB wl-hdd ;)
May 05 15:38:27 <jaro> oh.. i've just lost my wlhdd :( set boot_dev to external disk, rebooted and it won't respond now ;(((
May 05 15:40:03 <jeanfabrice> jaro: send it to me for electronic part spare instead of throwing in at the trash
May 05 15:40:58 <ka6sox-office> jaro: pull the HD out and reboot.
May 05 15:41:37 <jeanfabrice> 'Nite all
May 05 15:41:45 <jp30-work> 'night jf
May 05 15:43:08 <rwhitby-asleep> jaro: a lot of your questions over the last few hours are answered at wl500g.dyndns.org ....
May 05 15:45:18 <[g2]> rwhitby-asleep, good morning 
May 05 15:47:36 <rwhitby> g'day
May 05 15:47:44 <jp30-work> morning rwhitby
May 05 15:47:59 <rwhitby> morning.
May 05 15:48:24 <rwhitby> Looks like I have to find some down-time to upgrade the firmware in my InternetRouter now :-)
May 05 15:48:34 <rwhitby> (Oleg has released a new version)
May 05 15:49:40 <jf-asleep> building appropriate uclibc toolchain on the fly !
May 05 15:52:29 <jp30-work> copperbeech, php embed sapi support is now in cvs.
May 05 15:52:46 <copperbeech> wow! that's great!
May 05 15:53:16 <copperbeech> is there a way of just pulling that file out of the cvs rather than pulling the whole tree down every time?
May 05 15:53:30 * copperbeech is new to CVS
May 05 15:53:48 * Tomilius is new to just about everything
May 05 15:54:33 <jp30-work> copperbeech, once you have a working copy, then "cvs update" and "cvs checkout" just retrieve changes since your last update
May 05 15:55:03 <copperbeech> ok - it's just that I have been editing things locally - will an update overwrite those changes?
May 05 15:55:39 <jp30-work> no, it will try to merge them. welcome to the wonderful world of revision control
May 05 15:56:40 * copperbeech grins
May 05 15:56:45 <jp30-work> copperbeech, http://cvsbook.red-bean.com/ is a good place to start learning to use cvs
May 05 15:57:04 <copperbeech> jp30: thatnk I'll start my education there :-)
May 05 16:00:46 <copperbeech> jp30: so basically I just need to put a dependency in for php-stage ?
May 05 16:01:42 <rwhitby> Tomilius: welcome!
May 05 16:02:16 <jp30-work> you'll need to 1) stage php; 2) configure appweb to use it; 3) add an ipk dependency on php-embed. step 2 will be the hardest
May 05 16:03:00 <rwhitby> I like how Oleg released -5 on 05/05/05 :-)
May 05 16:05:44 <copperbeech> jp30: for (1) it is a matter of adding " $(MAKE) openssl-stage php-stage " into the package
May 05 16:06:02 <copperbeech> jp30: for (2) it will be a matter of learning the configure switches etc
May 05 16:06:50 <copperbeech> jp30: for (3) are you suggesting that I will also have to add "APPWEB_DEPENDS=php-embed" to the .mk file ?
May 05 16:07:00 <jp30-work> copperbeech, that's right.
May 05 16:07:34 <copperbeech> jp30: what is the difference between (1) and (3) - excusing my ignorance?
May 05 16:08:42 <jp30-work> make php-stage causes php's libraries and include files to be copied into the staging directory...\
May 05 16:09:36 <jp30-work> adding php-embed to APPWEB_DEPENDS will cause ipkg to install php-embed when appweb is installed by the user
May 05 16:10:35 <jp30-work> copperbeech, i think it's great that you're learning how to do this, but you've chosen a complicated package to work on
May 05 16:12:04 <copperbeech> jp30: yes I must be mad, but since no-one else seems to be interested and I want appweb with php -  I feel like I had to roll up my sleeves and dive in the deep end. 
May 05 16:12:27 <jp30-work> one gotcha is that you will need to learn how staging works so that you can configure appweb correctly
May 05 16:13:15 <copperbeech> yes - there seems to be an example already since it needs to stage openssl so I'm going to model it on that and see how I go...
May 05 16:13:47 <copperbeech> I have an old appweb.mk file which builds 1.2.3 which I'm going to try and get working first.
May 05 16:14:13 <jp30-work> i mean that you need to understand what staging is and why we do it. the instructions on the appweb web pages suggest that you configure appweb with a path into the php build dir, but i would not recommend trying that with the unslung build system
May 05 16:18:20 <copperbeech> jp30: oh you mean in case it 'hard codes' the path into the binary?
May 05 16:18:33 <copperbeech> how would you approach it?
May 05 16:21:22 <jp30-work> well, that's what staging is for - it mimics the directory tree of an unslung slug with all the packages installed. it should be possible to configure appweb to get the libraries it needs from there
May 05 16:22:20 <jp30-work> the trouble isn't so much hardwritten paths as the fact that it's unreliable to make assumptions about the states of other packages' build dirs
May 05 16:22:34 <rwhitby> agreed
May 05 16:23:27 <copperbeech> jp30: I've probably hit a bit of a dead end anyway... I currently only have a native build environment (much to rwhitby's perplexity) and I probably need to get the php-embed installed in there without the need to compile them. I'm pretty sure this must be possible - I think libstdc++ staging was a simple copy of binary files rather than a recompile. 
May 05 16:24:23 <jp30-work> copperbeech, again, this is why you need to understand how the build system works. you shouldn't have to install php-embed to develop appweb, just stage it
May 05 16:25:06 <jp30-work> make php-stage will build php and install it (including php-embed) in the staging dir
May 05 16:25:26 <copperbeech> jp30: I realise that, the real madness is that I only have anative build environment, so to stage php I have to compile it natively which doesn't work...
May 05 16:25:56 <jp30-work> what goes wrong when you try?
May 05 16:26:18 <jp30-work> (aside from it taking several hours :)
May 05 16:26:26 <copperbeech> make[1]: *** No rule to make target `/share/hdd/data/public/cvs/unslung/sources/php/threads.m4.patch', needed by `/share/hdd/data/public/cvs/unsl
May 05 16:26:52 <copperbeech> lung/builds/php/.configured'.  Stop.
May 05 16:26:58 <jp30-work> you need to cvs update -dP
May 05 16:27:20 <jp30-work> ...to get all the directories up to date
May 05 16:27:41 <copperbeech> aahhh
May 05 16:29:24 <jp30-work> i can't guarantee that it will work. the problems are not likely to be with php itself, but it has many and varied dependencies. expect a 12 hour build.
May 05 16:30:26 <copperbeech> btw - eno and I fixed a couple of bugs in the existing appweb make file which would be quite nice to get into the CVS - could you upload it? It can be found at:   http://homepage.mac.com/brian_zhou/nslu2appweb/appweb.mk
May 05 16:31:34 <copperbeech> jp30: is there no quicker way of staging? Is there a way I can simply get the existing compiled ipkg into the staging area without actually having to build it?
May 05 16:32:32 <jp30-work> copperbeech, two things...
May 05 16:32:43 <copperbeech> ok - fire away :-)
May 05 16:34:21 <jp30-work> if appweb's to be modified in cvs, i'd like the changes to be checked in by a developer with write access who can take responsibility for ensuring that they work and fixing matters if they don't. so if eno wants to check them in that's fine
May 05 16:35:59 <jp30-work> 2) the purpose of your efforts is to develop a build procedure for appweb that can be replicated on the official build system. that's why you need to use php-stage to set up staging. if you set it up manually, you can't be sure that what you did can be replicated by php-stage
May 05 16:36:00 <NAiL> hmm... Anyone here got a remote idea what a 4x20 char LCD pulls of power? ;)
May 05 16:39:38 <copperbeech> jp30: The appweb.mk file is (dis)owned by nslu2-linux and no one is picking up any bugs for it. eno was helping me by testing the mk file build on the cross system, so I guess I should try and get developer access for this make file and try to keep up with any bugs.
May 05 16:40:13 <copperbeech> jp30: I understand (2) above. Good point. I'll try building it over the weekend. :-)
May 05 16:43:26 <copperbeech> jp30: Thanks for your help this evening - have learned much
May 05 16:43:33 * copperbeech is off to bed
May 05 16:44:35 <jp30-work> 'night copperbeech. yes, you should get write access if you want to maintain appweb. send me a working .mk file :)
May 05 16:45:20 <copperbeech> see this link for a working make file for version 1.2.3  http://homepage.mac.com/brian_zhou/nslu2appweb/appweb.mk
May 05 16:45:49 <copperbeech> the previous make file we had a look at this evening which failed was for version 2.0.0
May 05 17:47:54 <[g2]> Tiersten, around ?
May 05 19:05:01 <NAiL> jbowler :)
May 05 19:05:59 <jbowler> hi!
May 05 19:06:37 <NAiL> <-- repvik
May 05 19:13:50 <[g2]> hey jbowler 
May 05 19:14:03 <[g2]> nice couple posts on the ml
May 05 19:14:49 <jbowler> [g2]: wot's an ml?
May 05 19:14:58 <[g2]> mailing list :)
May 05 19:16:44 <jbowler> [g2]: oh, duh...
May 05 19:16:46 <jp30-work> jbot, seen eno
May 05 19:16:47 <jbot> eno <~eno@adsl-68-120-128-150.dsl.snfc21.pacbell.net> was last seen on IRC in channel #nslu2-linux, 4d 21h 59m 1s ago, saying: 'that's great'.
May 05 19:17:59 <[g2]> jbowler, so what are the current issue wrt OpenSlug and /linuxrc ?
May 05 19:18:31 <jp30-work> anyone here know anything about the unslung samba package? looks like eno and goodnewscd have been making conflicting changes to it. it will get demoted soon unless someone takes responsibility for making sure it works
May 05 19:22:50 <jbowler> [g2]: openslug-kernel_2.6.11.2.bb still has the initrd stuff in there and /linuxrc attempts to exec /bin/init (it should be /sbin/init)
May 05 19:23:21 <jbowler> [g2]: then the hardware ether is not set in the NFS boot stuff (disaster - it uses the build in single RedBoot one)
May 05 19:24:36 <jbowler> [g2]: the pivot_root makes three copies of >/dev/console (it should <>/dev/console >&0 2>&0, I don't know how much this actually costs).
May 05 19:26:58 <CIA-1> 03jp30 * 10unslung/ (6 files in 3 dirs): apache, php-thttpd: use update-alternatives for htpasswd
May 05 19:27:31 <jp30-work> morning, CIA-1, haven't seen you for a while
May 05 19:28:52 <jbowler> [g2], rwhitby-away: ok, I have a working serial-less boot of openslug (up to date pull) - want me to check in the changes?  (back in a bit, cooking...)
May 05 19:29:07 <[g2]> jbowler, OK... I see what you mean about the initrd parameter in the .bb file for the OpenSlug kernel
May 05 19:29:29 <[g2]> I'm way out-of-date, but I *don't* see a /bin/init
May 05 19:30:08 <NAiL> <-- looking for that as well
May 05 19:30:10 <[g2]> you're saying the it tries to do /bin/init and really should do /sbin/init correct ?
May 05 19:30:26 <[g2]> the linuxrc is incorrect
May 05 19:31:18 <NAiL> ah
May 05 19:32:23 <NAiL> jbowler: So, if I do those two changes, it should work just nicely? I don't care about NFS at all anyways.
May 05 19:33:04 <jbowler> [g2]: correct, it never gets out of the linuxrc
May 05 19:33:42 <jbowler> NaiL: it is sufficient to use the CMDLINE_ROOT which is comment out in the .bb file, if you don't have init=/linuxrc on the command line it won't exec it...
May 05 19:33:52 <[g2]> jbowler, Ok so the linuxrc should test for /bin/init and then try /sbin/init
May 05 19:34:24 <[g2]> then we'll be compatible with Unslung and Openslug or OpenSlug could add a /bin/init link to point to /sbin/init
May 05 19:34:26 <jbowler> [g2]: not a lot of point to that, the build always puts it in /sbin/init.
May 05 19:34:51 <jbowler> [g2]: the /linuxrc is in openslug-init, so it is openslug specific.  The unslung one is somewhere else.
May 05 19:35:11 <[g2]> ah...
May 05 19:35:37 <NAiL> Ok, I'll start a build and go to bed. It's 04:35am here now, so I should go to bed :P
May 05 19:35:37 <jbowler> [g2]: oh, the linuxrc must be different openslug/unslung, 'cause the ifconfig won't work on openslug (it's eth0 not ixp0)
May 05 19:36:06 <NAiL> jbowler: Again, thanks for the help
May 05 19:36:36 <[g2]> jbowler is such and ass-kicker :)
May 05 19:37:01 <[g2]> and how did you do the bolding ? (total side note here) but it is cool
May 05 19:37:29 <jbowler> [g2]: I was asked that last time ;-)  It's a '*' either side of the word.
May 05 19:37:39 <NAiL> ctrl-b?
May 05 19:38:01 <[g2]> hmmmm.... jbowler is *such* and ass-kicker
May 05 19:38:09 <jbowler> [g2]: what I'd really like to do is to move the reading of SysConf from openslug-init (sysconfsetup) and just do it once in the linuxrc
May 05 19:38:15 <[g2]> that doesn't come out bold for me :(
May 05 19:38:25 <NAiL> [g2]: Comes out bold with * here
May 05 19:38:26 <jbowler> Hum, must be my IRC klient (ksirc)
May 05 19:38:50 <jbowler> Could always try <ESC>[33m etc...
May 05 19:38:53 <[g2]> NAiL, X-chat :)
May 05 19:39:00 <NAiL> [g2]: Try pressing ctrl-b before and after what you want bolded ;)
May 05 19:39:31 <[g2]> the NSLU2 community Rocks!
May 05 19:39:37 <NAiL> There's a way to _underline_ as well
May 05 19:39:47 <[g2]> ctrl-u eh ? :)
May 05 19:39:52 <NAiL> dunno :P
May 05 19:39:55 <[g2]> ok back to business
May 05 19:41:28 <dyoung-web> moring [g2]
May 05 19:41:36 <[g2]> hey dyoung-web 
May 05 19:42:28 <[g2]> dyoung-web, I'd like to talk to you later about the x1205
May 05 19:42:41 <[g2]> and getting that all fixed up properly
May 05 19:43:13 <dyoung-web> Okay, I need to spend around 3 hours with it to make it all better.
May 05 19:45:28 <[g2]> dyoung-web, that'd be great
May 05 19:45:54 <[g2]> dyoung-web, do you play with NTP ?
May 05 19:46:20 <dyoung-web> Yes, I was the one that talked Christopher into building ntp in the first place.
May 05 19:46:33 <dyoung-web> But in general I like NTP.
May 05 19:46:37 <dyoung-web> NTP is my friend.
May 05 19:46:56 <dyoung-web> Unless its a TrueTime Stratum1 server.  Then NTP isnt my friend.
May 05 19:47:39 <[g2]> ok NTP the standard stuff is in OE
May 05 19:48:24 <[g2]> we should add it to the packages as soon as we figure out why it locks up in the 12-24 hour period
May 05 19:48:52 <[g2]> but before we straighten that out, we need to fix the RTC stuff to get us on the right month and day
May 05 19:49:03 <[g2]> and look at the adjtick
May 05 19:49:11 <[g2]> are you using the 10100 value ?
May 05 19:51:23 <dyoung-web> I havnt tried fooling with it under openslug.
May 05 19:51:34 <dyoung-web> It would be more of a priority to fix the RTC routine.
May 05 19:51:54 <[g2]> good lets get that right.... It can't hurt NTP too much :)
May 05 19:58:00 <CIA-1> 03jp30 * 10unslung/sources/ (4 files in 2 dirs): fix alternatives work
May 05 19:58:22 <dyoung-web> basically I fixed only half the problem.  I need to revisit the code; but it was a issue of 0-23 vs 1-24 for hours or maybe it was 0-11 vs 1-12 for months, I forget.
May 05 19:59:40 <[g2]> yeah I remember that last time we were here :0
May 05 19:59:41 <[g2]> :)
May 05 20:00:26 <dyoung-web> is 10100 really the good value for the tickadj?
May 05 20:00:58 <[g2]> that's was I was trying today on ccsmart's recommendation
May 05 20:28:27 <jbowler> [g2]: the correct value for ticks is 10000
May 05 20:28:29 <CIA-1> 03jp30 * 10unslung/ (4 files in 2 dirs): apache: to version 2.0.54, fix apachectl
May 05 20:28:54 <[g2]> jbowler, that would make sense to me
May 05 20:28:57 <jbowler> I've run an NSLU2 free for several days at 10000 and not had significant drift.
May 05 20:29:20 <[g2]> which ntp source are you using ?
May 05 20:29:59 <[g2]> I mean as in source code not clock source :)
May 05 20:30:12 <jbowler> [g2]: I'm not - that's what I meant by 'free'
May 05 20:30:41 <jbowler> [g2]: with the only versions on the NSLU2 I had the 12/24 hr problem because of incorrect handling of a negative adj
May 05 20:30:45 <CIA-1> 03jp30 * 10unslung/ (make/samba.mk sources/samba/control): samba depends on openldap-libs now
May 05 20:31:24 <jbowler> I have ntp_4.2.0.bb apparently working, but I haven't checked it in to OE because it's core.
May 05 20:31:44 <[g2]> jbowler, that's the version I'm trying to get working properly
May 05 20:31:53 <[g2]> do you have a patch for that ?
May 05 20:32:14 <[g2]> dyoung-web, is working on getting the RTC code to not be a month off on startup
May 05 20:32:34 <[g2]> is there something else also ?
May 05 20:37:58 <jbowler> [g2]: It's necessary to persuade it not to check a machine dependent check of the layout (i.e. it tries to run a target binary on the build host), but that's an easy patch.  I've built but not, I think, tested
May 05 20:38:34 <[g2]> can you post the patch on the wiki or send me a copy ?
May 05 20:39:07 <[g2]> it build fine for me today
May 05 20:39:13 <[g2]> s/build/built
May 05 20:40:25 <jbowler> ntpdc/Makefile.am has BUILT_SOURCES= maybe-layout, I just commented out the maybe-layout - without that it will only build native.
May 05 20:43:07 <jbowler> I'll push it with PREFERENCE=-1 and PREFERENCE_openslug=0
May 05 20:43:49 <[g2]> Ah.... It built fine with glibc
May 05 20:44:17 <[g2]> this is exactly why I'd like to have an automated matrix setup
May 05 20:45:12 <[g2]> so we could compare the gcc xxxx with glibc/uclibc and ntp 4.2.0 ---- 1) Compile  2) Run
May 05 20:46:26 <[g2]> jbowler, and finally the fix for the negative adj was where ?
May 05 20:46:34 <jbowler> My build is on glibc too.  Look at 'ntpdc/Makefile' - search for 'maybe-layout'.  If you have perl installed it will execute ${MAKE} check-layout which executes ntpdc-layout, which is an ARM binary!
May 05 20:47:15 <[g2]> I was running the ntpdc client I installed today
May 05 20:47:18 <jbowler> [g2]: I'm not sure it was fixed, but it seemed to be an FP problem on unslung and there was a kernel fix in this area.
May 05 20:47:23 <[g2]> and built today
May 05 20:47:37 <[g2]> however, my snapshot was *old*
May 05 20:47:56 <[g2]> so it's more likely a problem that got introduced lately
May 05 20:48:15 <[g2]> I just pulled an hour or so ago and I'm doing a fresh build with tmp wiped
May 05 20:48:42 <[g2]> glibc is staging now
May 05 20:48:45 <jbowler> [g2]: take a look at the Makefile in the ntpdc tree.  Something must have stopped the rule for 'layout.here' being executed.  The problem has always been there.
May 05 20:49:53 <[g2]> ok will do.
May 05 20:50:17 <[g2]> and btw rwhitby said that NFS should work fine
May 05 20:53:00 <jbowler> [g2]: how can it?  Line 88 of /linuxrc: /sbin/ifconfig ixp0 up $ipaddr netmask $netmask, but there is no interface 'ixp0' on openslug!
May 05 20:53:36 <[g2]> well that'll have to be change obviously
May 05 20:53:47 <[g2]> you are correct that it won't work now
May 05 20:54:04 <[g2]> just like the /bin/init versus /sbin/init
May 05 20:55:47 <jbowler> [g2]: it also won't work because of lockd and portmap.  They will be running as a result of the kernel insmod when the NFS mount occurs, they need to be killed before the pivot_root.
May 05 20:56:59 <[g2]> hmmmm....... 
May 05 21:06:00 <dyoung-web> we get around that by specifying nolock
May 05 21:06:16 <CIA-1> 03jp30 * 10unslung/Makefile: demote clamav
May 05 21:06:34 <dyoung-web> that was tested and working before we started screwing around with it in the past week.
May 05 21:08:36 <jbowler> dyoung-web: doesn't the kernel insmod pull in lockd/statd/portmap, or does nolock stop that too?
May 05 21:09:51 <jbowler> [g2]: with a straight (unpatced) ntp_4.2.0.bb I get a compile fail with: /bin/sh: line 1: ./ntpdc-layout: cannot execute binary file
May 05 21:11:55 <[g2]> jbowler, I think it was 4.2 that built earlier today
May 05 21:12:55 <dyoung-web> jbowler: it didnt used to.
May 05 21:13:02 <[g2]> I trashed tmp and I'm rebuilding everything so I can look there, but... I've still got the ipkg around
May 05 21:13:23 <jbowler> [g2]: that's with a simple 'inherit autotools' in the .bb file?  I.e. as in the 4.1.2.bb?  If so maybe autoconf isn't finding perl on your build system.
May 05 21:14:21 <jbowler> [g2]: anyway, it's moot.  The layout.here stuff is bogus unless done on the target system (it's checking that the C compiler lays out structs in the way it expects).  I've verified that gcc/arm is ok.
May 05 21:15:12 <jbowler> dyoung-web: ok, when I tried this on a flash rootfs (i.e. when I tried NFS mounts and checked the modules via lsmod) I wasn't using nolock.
May 05 21:17:27 <jbowler> [g2]: blah, the ntp package is pulling in curses.  That's got to be fixed...
May 05 21:17:42 <jp30-work> hi, eno
May 05 21:17:54 <eno> hi jp30, i'm looking at samba package right now
May 05 21:18:44 <jp30-work> excellent. i patched it a little to depend on openldap, but i wasn't sure whether the correct thing to do might have been to configure --without-ldap
May 05 21:19:45 <jp30-work> (sorry, to depend on openldap-libs)
May 05 21:19:51 <eno> i c
May 05 21:20:30 <jp30-work> i was also worried about the runtime error that was reported on the list, but since i don't run samba myself i couldn't check it out
May 05 21:20:55 <eno> i can replicate the problem
May 05 21:21:16 <eno> and i am comparing config.log of native vs. cross
May 05 21:21:37 <eno> the problem is between the native to cross change
May 05 21:21:49 <jp30-work> ah
May 05 21:22:00 <jp30-work> does samba make a config.cache?
May 05 21:22:48 <jp30-work> if it's a cross-configuration issue, that might be a good place to look for solutions
May 05 21:23:32 <eno> find builds/samba -name config.cache
May 05 21:23:33 <eno> builds/samba/packaging/Debian/debian-stable/config.cache
May 05 21:23:35 <eno> builds/samba/packaging/Debian/debian-unstable/config.cache
May 05 21:23:53 <eno> this is on the cross build machine
May 05 21:24:37 <jp30-work> hmm. well, config.log should do
May 05 21:24:50 <eno> the same on the native build box
May 05 21:25:31 <eno> when autoconf says a setting is "cross", what does that mean?
May 05 21:26:10 <jp30-work> that means that it couldn't tell what the answer was because it was cross compiling.
May 05 21:26:29 <eno> and what value will it actually use?
May 05 21:26:46 <jp30-work> that depends on how the particular test was written
May 05 21:27:18 <eno> k
May 05 21:29:32 <jp30-work> often, configure will store the results of many tests in config.cache. this file is also in a suitable format to be fed back into another run of configure. that's why i was asking about it
May 05 21:32:10 <jp30-work> back in 30 mins...
May 05 21:43:03 <jbowler> eno: you can tell configure the answer, sometimes by an explicit option (e.g. --with-yielding-select=yes to openldap), sometimes by direct change of the config.cache variable (e.g. I use lt_cv_dlopen_self=yes to openldap to get libtool to work correctly)
May 05 21:45:06 <eno> ok, i often just set do something like samba_cv_have_setresgid=yes before configure
May 05 21:46:04 <jbowler> configure parses options like that - i.e. you can just put it on the configure command line
May 05 21:47:03 <eno> yeah, "cv_var=yes configure", this works most of the time, if the configure.in was written the correct way
May 05 21:47:37 <eno> in some cases, i had to patch the configure.in and rerun autoconf first
May 05 21:47:52 <[g2]> jbowler, THX for checking in the OpenSlug fix.
May 05 21:47:58 <[g2]> did you bump the PR # ?
May 05 21:48:12 <jbowler> [g2]: I hope so ;-)
May 05 21:48:31 <[g2]> my fresh build just completed
May 05 21:49:37 <jbowler> [g2]: aargh, I did on openslug-init but not openslug-kernel.
May 05 21:50:45 <[g2]> NOTE: package ntp-4.1.2-r2: task do_build: completed
May 05 21:50:45 <[g2]> NOTE: package ntp-4.1.2: completed
May 05 21:51:00 <[g2]> that builds out-of-the-box
May 05 21:51:19 <jbowler> [g2]: yes, it does - they added the maybe-layout stuff in 4.2.0
May 05 21:51:56 <[g2]> jbowler, cool.  can you let me know when you push that PR bump, then I'll pull and build agin
May 05 21:52:01 <[g2]> again
May 05 21:52:23 <jbowler> [g2]: just did it, 1.3226
May 05 21:52:39 * [g2] goes to pull :)
May 05 21:55:31 <[g2]> who added the nslu2-2.6.11.8 kernel ?
May 05 21:56:53 <jbowler> [g2]: me - I just tested it, you have to set PREFERRED_PROVIDER_virtual/kernel to nslu2-kernel to pull it in.
May 05 21:57:23 <[g2]> cool.
May 05 21:57:30 <[g2]> I was waiting for .12
May 05 21:57:39 <jbowler> [g2]: it's partial, I've got the reiser4 changes to check in but that needs an mm kernel, nslu2-kernel.inc should just do it (totally untested so far...)
May 05 21:57:55 <[g2]> but I'm glad you did it
May 05 21:58:17 <[g2]> my kernel's rebuilding now
May 05 21:58:20 <jbowler> [g2] try copying nslu2-kernel_2.6.11.8.bb -> nslu2-kernel_2.6.12-rc3.bb ;-)
May 05 21:58:37 <jbowler> [g2] (the patches will almost certainly fail!)
May 05 21:58:44 <[g2]> that's the problem
May 05 21:59:06 <[g2]> I think I applied that patch to the wrong base 
May 05 22:02:34 <jbowler> [g2]: ok, I have an ntp_4.2.0 with no curses or readline dependencies (that just means no command line history in ntpdc...)
May 05 22:03:40 <dyoung-web> okay I'm all stringed out.  I hate strings.
May 05 22:04:48 <jbowler> [g2], dyoung-web: tick = 10000 in openslug 2.6.11.8 (sic) with ntp 4.2.0
May 05 22:05:09 <jbowler> oops, no, that was openslug 2.6.11.2 - I went back to the released version
May 05 22:06:02 <jbowler> date output seems to be correct too, 5am on May 6 UTC?
May 05 22:07:59 <dyoung-web> jbowler: that sounds like a better value to me.
May 05 22:08:34 <dyoung-web> the date should be sorta correct; but its really a hack job.  I want to fix it right beucase I dunno what happens when you get to december.
May 05 22:09:22 <jbowler> dyoung-web: well, that sounds fine, that's six months away
May 05 22:09:37 <dyoung-web> hehe that was my thinking too. ;-)
May 05 22:11:15 <jbowler> With ntp 4.2.0 I get some interesting debug messages from xl205, e.g.:
May 05 22:11:24 <jbowler> raw x1205 read data  - sec-11 min-04 hr-85 mday-06 mon-05 year-05 wday-05 y2k-20
May 05 22:12:12 <dyoung-web> oh, thats in the x1205 driver.
May 05 22:12:33 <dyoung-web> you can turn it on and off too.
May 05 22:15:57 <[g2]> dyoung-web, how does one do that ?
May 05 22:16:04 <[g2]> turn it on and off
May 05 22:16:16 <dyoung-web> I forget, maybe something in /sys.
May 05 22:16:27 <jbowler> [g2]: 4.2.0 works a whole lot better on openslug than it did on unslung, ntpdc and ntpq aren't crashing the server now ;-)
May 05 22:16:55 <[g2]> jbowler, good to hear that :)
May 05 22:17:35 * [g2] note to self find /sys entry
May 05 22:18:53 <ka6sox-away> I use ntpdate to do gross changes and then nptd to keep it ontime.
May 05 22:19:49 <[g2]> it should just be adjtick 10000; ntpdate; ntpd 
May 05 22:20:27 <ka6sox-away> I've used that before and it works good.
May 05 22:20:39 <dyoung-web> the stock openslug tickadjustment should be at 10000 already no?
May 05 22:20:50 <[g2]> probably
May 05 22:21:08 <dyoung-web> it was detuned in unslung to 9990 or sometihng silly like that.
May 05 22:21:34 <[g2]> Ok I've got a riddle here
May 05 22:21:42 <dyoung-web> 42
May 05 22:21:47 <[g2]> :)
May 05 22:22:08 <[g2]> I've nfsmounted a dir that has a swapfile init it
May 05 22:22:26 <[g2]> I've insmod'ed loop.ko
May 05 22:22:47 <[g2]> and I've losetup /mnt/swap/swap128 /dev/loop0
May 05 22:22:53 <[g2]> and it worked the first time
May 05 22:23:13 <[g2]> then the "swapon /dev/loop0" failed
May 05 22:24:23 <dyoung-web> this is a silly question, but you do have write permisison to that file right?
May 05 22:24:56 <[g2]> :)
May 05 22:25:05 <[g2]> not so silly 
May 05 22:25:27 <[g2]> hmmm...
May 05 22:25:31 <[g2]> I think I do
May 05 22:25:52 <dyoung-web> cd /mnt/swap; touch xx
May 05 22:26:28 <[g2]> ls -l xx
May 05 22:26:28 <[g2]> -rw-r--r--    1 root     root            0 May  6  2005 xx
May 05 22:27:31 * dyoung-web ponder
May 05 22:28:00 * [g2] reboots to try again
May 05 22:28:22 <[g2]> It's *so* close
May 05 22:29:18 <[g2]> Then I'll have a swap for native compiles and I'll be able to setup the RV with the 533 or the Avila with the 64MB and Gig NFS to do native compiles
May 05 22:30:09 <dyoung-web> does it tell you any good error messages in either /var/log/message or at the console?
May 05 22:33:37 <[g2]> AHHH....
May 05 22:33:46 <[g2]> dyoung-web, finds a clue
May 05 22:34:02 <[g2]> Unable to handle swap header version 16777216
May 05 22:34:14 <[g2]> Ah....
May 05 22:34:20 <[g2]> Endianess
May 05 22:34:37 <[g2]> how do you un losetup something ?
May 05 22:35:02 <dyoung-web> reboot ? ;-)
May 05 22:35:09 * [g2] reboots :)
May 05 22:35:13 <dyoung-web> but sinceyou have a zillion RPM drive...
May 05 22:35:35 <dyoung-web> or even why not just re mkswap it?
May 05 22:35:57 <[g2]> I'm gonna zero it out and make if from the TARGET
May 05 22:36:10 <dyoung-web> from the Revvie right?
May 05 22:36:39 <dyoung-web> I think all thats important is you mkswap it from the revvie.
May 05 22:37:14 <[g2]> I'm on the slug right now, but I'll be doing this from the 533 on the revie
May 05 22:37:32 <dyoung-web> Okay.
May 05 22:37:47 <[g2]> I'm not sure which will be faster
May 05 22:38:10 <dyoung-web> I would think the revvie would be faster.
May 05 22:38:18 <[g2]> 533 on the RV082 with 32MB mem and 100Mb nfs mount or
May 05 22:38:35 <[g2]> 266 on Avila with 64MB and GigE nfs mount
May 05 22:38:54 <dyoung-web> Good point.
May 05 22:39:23 <dyoung-web> how fast does the GigE really go on the avila?
May 05 22:39:58 <[g2]> dyoung, THX
May 05 22:40:07 * [g2] does the happy dance !
May 05 22:40:24 <dyoung-web> worked?
May 05 22:41:15 <[g2]> swap's mounted
May 05 22:41:38 <[g2]> SwapTotal:      131064 kB
May 05 22:41:40 <[g2]> :)
May 05 22:41:42 <dyoung-web> Cool!
May 05 22:42:09 <[g2]> robbat2 in #gentoo-embedded explained how he did it
May 05 22:42:24 <[g2]> with the loop back mount 
May 05 22:42:27 <dyoung-web> you should be able to twiddle the rsize and wsize for better performance too.
May 05 22:42:35 <[g2]> nod.
May 05 22:42:41 <[g2]> I was reading about that
May 05 22:42:48 <[g2]> I'll probably make them 8K
May 05 22:43:03 <dyoung-web> do you just dd (options) of=swapfile; and mkswap it?
May 05 22:43:19 <dyoung-web> mount it and mkswpa it.
May 05 22:43:21 <[g2]> because on the gigE I can have 9K jumbo frames
May 05 22:43:37 <dyoung-web> ooh, nevermind, I see your point.
May 05 22:43:40 <dyoung-web> thats a neat trick.
May 05 22:43:48 <[g2]> one packet
May 05 22:44:30 <dyoung-web> okay dinner time.  See you later.
May 05 22:44:36 <[g2]> OK
May 05 22:44:41 <[g2]> THX again
May 05 22:44:47 <dyoung-web> np.
May 05 22:44:49 <dyoung-web> ;-)
May 05 22:44:53 <[g2]> I'll write it all up on the wiki tomorrow
May 05 22:45:15 <dyoung-web> Coolio.
May 05 22:57:06 <jbowler> [g2]: I believe the NTP code is still adjusting the clock the wrong way - i.e. instead of slewing +1 second it slews -1 second
May 05 22:57:23 * [g2] hugs jbowler 
May 05 22:58:09 <jbowler> [g2]: I'm going to leave it running with no NTP overnight to see how much slew I get (if any), then run with NTP and watch it go exponentially wrong...
May 05 22:58:39 <jbowler> [g2]: if I'm right there is probably a simple bug in one of the ARM kernel calls (adjtime, or whatever it uses).
May 05 22:59:38 <[g2]> jbowler, awesome work
May 05 22:59:51 <[g2]> I'd love to have this nailed
May 05 23:00:08 <jbowler> [g2]: oh, the hardware clock is, how can I put it?  Not very good?  It's gained 4s in about 10 minutes...
May 05 23:01:03 <[g2]> maybe we do want the tick to be 10100 :)
May 05 23:03:20 <jbowler> Ah, I meant the RTC, i.e. the thing which runs all the time (even when the power is off).  I.e. the output of 'date' is good, that of 'hwclock' is bad.
May 05 23:04:19 <jbowler> [g2]: eh... when I run 'ntpdate -q' it gives me the offset from the RTC, not from the internal clock???  That must be wrong.
May 05 23:07:47 <[g2]> jbowler, I was hanging in #ntp with the guys that developed that sw
May 05 23:07:51 <[g2]> today.
May 05 23:07:54 <jbowler> [g2]: I got that wrong, I was misreading the output of ntpdate.  I need to time the system clock over a few hours.
May 05 23:08:20 <[g2]> I don't know enough to ask questions I'm still ramping up to speed
May 05 23:08:59 <[g2]> I've we've got some good questions, I"m sure those are the guys with answers
May 05 23:13:27 <jbowler> [g2]: I suspect it's easy to track down.  I'll do the timing first to make sure tickadj really is correct (and find out how much the RTC slews).
May 05 23:13:49 <jbowler> [g2]: anyway, that's a good excuse to go to sleep now...
May 05 23:13:56 <[g2]> nod.
May 05 23:19:44 <[g2]> nite all
May 05 23:28:44 <CIA-1> 03bzhou * 10unslung/make/samba.mk: force cross compile to USE_SETRESUID
May 05 23:29:35 <eno> jp30, i believe i've fixed the samba setgid problem
May 05 23:29:59 <eno> nite all
May 05 23:30:03 <jp30> hi, eno
May 05 23:30:16 <jp30> have you checked in the fixes to samba?
May 05 23:30:34 <eno> just checked in make/samba.mk
May 05 23:30:45 <eno> and bumped ipk-version
May 05 23:31:02 <eno> on top of 3.0.14a
May 05 23:31:20 <eno> and tested the package on my slug
May 05 23:31:57 <jp30> great, and it's still cross-compiling ok?
May 05 23:32:06 <eno> right
May 05 23:32:31 <jp30> ok, i'll build and push it on the official build system. thanks for your efforts!
May 05 23:33:10 <eno> sorry for the problem. I can sleep now
May 05 23:33:16 <jp30> 'nite
**** ENDING LOGGING AT Thu May 05 23:59:56 2005