**** BEGIN LOGGING AT Fri May 16 02:59:57 2008 May 16 04:14:14 Peteru: I was going to ask you about the weird busybox behaviour i was getting, but i worked it out May 16 04:18:56 turns out if the binary is not compat with the system and you type ./busybox to attempt to run said binary, the default path busybox kicks in... anyway, long story short, I have busybox 1.10.2 compiled and running on tgeh wiz May 16 04:21:53 wtih shell shortcuts rather than soft links for all the applets (due to being on the vfat HDD). Did this as Beyonwiz cut the busybox in 23x to nearrly next to no functionality. SOrry for typos, am typing blind in a remote desktop window and the chars are coming 10secs after typing. May 16 04:26:57 Hmmm, i thought they may start slimming down the system, now that they are running short on space. May 16 04:27:18 Next thing you'll know, there will be no telnetd :-( May 16 04:29:12 i hadn't noticed until eric mentioned it May 16 04:29:38 because i had only telnetted in once wiwth 235 May 16 04:33:32 They were wasting quite a lot of flash with having a bad install of busybox. The BB they shipped had a lot of functionality not compiled in, but they still created the filesystem links for all those commands. May 16 04:46:36 Does the Beyonwiz firmware have a working version of awk ? May 16 05:02:16 finally my irc window is back May 16 05:02:21 awk... hang on May 16 05:02:39 # awk May 16 05:02:39 awk: not found May 16 05:02:57 but if busybox supports awk i will compile it in May 16 05:02:59 hey eric May 16 05:03:06 i got the latest busybox working on the wiz May 16 05:03:37 Currently defined functions: May 16 05:03:37 [, [[, adduser, arp, arping, bunzip2, bzcat, bzip2, cat, chattr, chgrp, chmod, chown, chpasswd, chroot, May 16 05:03:37 cp, date, dd, deluser, df, diff, dirname, dmesg, dos2unix, du, dumpkmap, echo, egrep, env, ether-wake, May 16 05:03:37 expand, expr, false, fdisk, fgrep, find, findfs, free, freeramdisk, fsck, ftpget, ftpput, getopt, May 16 05:03:37 Bugger. That would be the most useful thing to have. Awk is a whole programming language that can be used to do quite a lot of powerfull stuff on embedded systems. May 16 05:03:38 grep, gunzip, gzip, head, hexdump, hostname, ifconfig, insmod, install, kill, killall, less, ln, loadkmap, May 16 05:03:41 logname, ls, lsattr, lsmod, makedevs, md5sum, mkdir, mkfifo, mknod, mkswap, mktemp, modprobe, more, May 16 05:03:43 mount, msh, mv, netstat, nice, nohup, nslookup, passwd, ping, pipe_progress, pivot_root, ps, pwd, May 16 05:03:45 rm, rmdir, route, sh, sleep, sort, split, stat, strings, su, swapoff, swapon, switch_root, tail, tar, May 16 05:03:48 tee, telnet, test, tftp, tftpd, time, top, touch, tr, traceroute, true, tty, umount, unix2dos, uptime, May 16 05:03:51 vconfig, vi, watch, wc, wget, which, who, whoami, xargs, yes, zcat May 16 05:03:53 May 16 05:07:41 done: # awk May 16 05:07:41 BusyBox v1.10.2 (2008-05-16 13:33:34 EST) multi-call binary May 16 05:07:41 May 16 05:07:41 Usage: awk [OPTION]... [program-text] [FILE...] May 16 05:07:41 May 16 05:07:42 Options: May 16 05:07:44 -v var=val Set variable May 16 05:07:46 -F sep Use sep as field separator May 16 05:07:48 -f file Read program from file May 16 05:08:37 it took me ages to work out how to compile it to work on the wiz... i can do simple things like a single .c file with a simple makefile, but the Makefile for Busybox was just crazy May 16 05:09:32 With a bit of hacking of one of the Makefile includes, this is how I achieved it (will only compile with gcc 3.x cross compiler): make CROSS_COMPILE=/opt/arm-uclinux-tools/bin/arm-uclinux-elf- LDFLAGS="-Wl,-elf2flt" May 16 05:12:05 I couldln't work out why the cross compile was working, but after the linking, the binary refused to work on the wiz, and it was because the makefile didn't have the -Wl,-elf2flt flags May 16 05:12:38 even though the executable created looked like the real deal with the elf header May 16 05:15:13 The joys of not having an official toolchain :-( May 16 05:15:16 luckily in rc.local I export a PATH to /tmp/mnt/idehdd/wiz/bin and then call the telnetd, and it appears that this path is OK for the shell/login, and will accept the new busybox (had to ditch the idea of sym links to call the busybox binary though with a vfat filesystem, and instead have little shell scripts: May 16 05:15:16 # cat ls May 16 05:15:16 #!/tmp/mnt/idehdd/wiz/bin/busybox May 16 05:15:59 so, the whole source & binaries are 5meg, can I upload anywhere, or how best do we want to deal with this> May 16 05:16:51 It should end up on OpenWiz one way or another - probably best get in touch with rwhitby to see if he's got any good ideas. May 16 05:17:19 He's not on irc, so perhaps shoot him an email. May 16 05:17:37 i am guessing if you diff the original release of 1.10.2 against my seriously hacked one, there will be a tonne of differences May 16 05:19:18 there is even a bug with the applet "link" creation, in that it defaults to building #!/bin/busybox rather than use the .config coded path for where you will have the busybox binary installed in the busybox source... so I hard coded the directory above, but it really needs to be dealt with by the busybox team May 16 05:40:42 Your source could go on Openwiz as it is (we would fork the upstream distribution) and over time patches could be sent to BB maintainers, until the two versions converge. May 16 06:42:05 tonymy01: As a fan of motorsport, you may enjoy this clip. http://www.snotr.com/video/1175 May 16 06:42:31 Gives a whole new meaning to Reverse Polish Driving ;-) May 16 07:18:49 hey, just got nfs working on the wiz too May 16 07:19:19 then i can pivot_root properly to a network drive and thus the HDD check routines will work May 16 07:20:11 (that is the theory... I am having hassles copying the WIz filesystem over without recursion of my shared paths and mounts) May 16 07:20:39 The kernel has nfs support built in? May 16 07:20:47 yep May 16 07:21:00 just had to add nfs support to the mount in busybox May 16 07:22:55 i was trying to work out a way of supporting softlinks given the kernel doesn't have ext2/3 support, in order to mirror the OS and enable easy tweaking without reflashing May 16 07:23:12 and saw it had nfs in /proc/filesystems May 16 07:25:54 from what I can determine, uclinux systems don't support loadable modules, so unless everything is compiled into the kernel, it makes things damn hard. I was going to insmod an ext module but that was dashed. May 16 07:28:52 I'd really like to figure out how to make the bootloader tftpboot the box. That way you could load the kernel from the network and put the root filesystem on nfs. May 16 07:29:39 There may be something along the lines of 1234+POWER perhaps a 9999+POWER or some such. May 16 07:30:11 would make trying to knock up a kernel a tiny bit easier for sure. May 16 07:30:44 presumably booting/loading this way simply loads to ram rather than flash? May 16 07:31:20 so you can keep on trying indefinitely until you get it right with no fears of breakage? May 16 08:02:42 Peter, didn't you say you tried hooking to the Wiz serial port? May 16 08:02:58 I just looked on the back of mine and there aint one :-/ May 16 08:07:09 Yes, tftpboot is essential for kernel tweaking and yes it does load the kernel to RAM via tftp May 16 08:07:40 There is a 4 pin header on the motherboard about half way between the CPU and the tuner cans. May 16 08:07:47 ahh, ok May 16 08:08:07 so i presume you used a voltage converter to go from TTL to RS232? May 16 08:09:02 hehe, i have seen that polish backwards car before, bloody crazy that driver going sideways up narrow roads! May 16 08:09:53 well, I stuck a multimeter on the pins and it looked like there was 3.3V ground and two pins that were most likely TX and RX. I hooked up my NSLU2 serial port dongle (with what I thought was the correct pinout) and turned on the Beyonwiz. About 3-5 seconds later there was a loud pop and a puff of blue magic smoke :-( May 16 08:10:17 Hey guys May 16 08:10:32 evening. May 16 08:10:38 My machine seems to be turning itself on without me May 16 08:10:45 not good... you can't go hooking RS232 +/-12V to a 3.3V system May 16 08:11:00 I have a TTL to serial converter May 16 08:11:06 but couldn't get a signal May 16 08:11:13 lucky you didn't let the blue smoke out the Wiz CPU May 16 08:11:30 I was worried that I blew the 8624, but when I disconnected everything and powered it on again, the wiz worked. May 16 08:11:49 hey eric... so you tried already hey May 16 08:11:56 yup May 16 08:11:59 I was using my 3.3V converter that I previously used for NSLU2 and wgt634u development May 16 08:12:06 so no diagnostics pop up? May 16 08:12:10 tried that before I found out how to load custom firmwares May 16 08:12:18 not that I could get running May 16 08:12:25 tried lots of combos May 16 08:12:33 with baud rate and pin selection May 16 08:12:44 ahh, peter, i thought you meant USB-serial dongle May 16 08:12:54 good, i didn't think you were that silly :-) May 16 08:12:54 I emailed jackie from BW and he told me they didn't have a serial connector May 16 08:13:39 * Yuv422 wonders what's waking his machine up May 16 08:13:42 hmm, no serial... usually doesn't stop the experienced, they just trace the pins out May 16 08:13:44 I could not get anything from that 4 pin header either, but I think it may just be disabled in software. May 16 08:14:05 yeah May 16 08:14:17 what f/w are you running eric? May 16 08:14:29 The latest May 16 08:14:30 are you using your wizremote to create timers still? May 16 08:14:37 yeah May 16 08:14:53 I find it's easier than using the rf remote. :) May 16 08:14:58 ir May 16 08:14:58 I had slightly more luck with the TF7000 serial port. I could see that it sent some data out, including ASCII colour, but I could never get it to react to anything. May 16 08:15:00 because i haven't had a look as yet, but PRL/peter suggests there is more info in the xml data to support the ICE stuff May 16 08:15:00 hehe May 16 08:15:16 hmm May 16 08:15:28 maybe that's only if you have an ice account May 16 08:15:33 let me check May 16 08:15:43 so maybe you have "bogus" timer info that is confusing it and causing it to come out of standby when it shouldn't May 16 08:16:01 sorry I was talking about my mac May 16 08:16:09 it woke up at 3 thisafternoon May 16 08:16:25 lololol... this is a wiz forum mate :-) May 16 08:16:37 I'd still blame IceTV ;-) May 16 08:16:54 Your mac is probably dialing home to Apple to download more DRM without your say so :-) May 16 08:17:04 hehe May 16 08:17:20 or a mouse ran over my keyboard May 16 08:17:21 or refresh your itune library May 16 08:17:25 They fucked up the whole thing after I and then Peter Vogel left. May 16 08:17:38 yeah, it is a bit of a mess May 16 08:17:48 lots of potential May 16 08:17:52 it is refusing to set a red dwarf timer for me at the moment May 16 08:18:07 mind you, the ep is marked as a repeat when it shouldn't be.... May 16 08:18:30 do you think I stand any chance of getting BW to give me a way of adding timers with wizremote without rebooting the machine? May 16 08:18:33 but i thought I would get smart and add it as a non-series timer, but both the web and widget just don't play ball May 16 08:19:07 i wonder if the ice protocol involves an intermediate file, or it talks direct May 16 08:19:30 iceUidHi="-1" iceUidLo="-35127295" iceCid="0" May 16 08:19:43 was ist das? May 16 08:20:27 from book.xml May 16 08:20:37 in my new timer entries May 16 08:20:55 there are still entries (yet to fire) that don't contain these new arguments May 16 08:21:12 so it looks like wizremote should still work for the time being May 16 08:21:32 until they make these new fields mandatory for something May 16 08:21:44 good May 16 08:22:20 BW is beeing pretty good to ICE in my opinion May 16 08:22:43 giving them exclusive access to their device May 16 08:22:57 yeah, essentially May 16 08:23:06 i wonder how much Jai is pushing this? May 16 08:23:37 and whether Jai has any clout with the Beyonwiz engineers to get some of our bugs fixed (add Dave to that also) May 16 08:26:17 I wonder what makes the stat file so prone to corruption? May 16 08:27:34 i am about to fix that May 16 08:27:48 the stat problem? May 16 08:28:07 either by hooking into the OS call for the dosfsck, or by simply putting the fix into rc.local May 16 08:28:19 yep, the stat prob May 16 08:29:00 hence my attempting to get nfs working properly, so i can muck around without reflashing all the time. May 16 08:29:26 192.168.0.105:/home/beyonwiz on /tmp/mnt/nfs type nfs (rw,v3,rsize=32768,wsize=32768,hard,udp,lock,addr=192.168.0.105) May 16 08:29:32 (my mount) May 16 08:30:06 I flashed a modded rc.sysinit which called a script on the config partition so I could mucck around with the boot sequence May 16 08:30:10 it seemed to work ok May 16 08:30:39 i found shell scripts calling shell scripts seemed to cause me problems May 16 08:31:26 rc.local is in config is it not? May 16 08:32:14 #!/bin/sh May 16 08:32:15 PATH=/tmp/mnt/idehdd/wiz/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/flash/bin; export PATH May 16 08:32:33 works fine for me for rc.local, and picks up the path properly May 16 08:32:45 yeah May 16 08:33:02 most of that was your idea actually May 16 08:33:08 I needed to change bits of rc.sysinit before the rc.local call May 16 08:33:16 :) May 16 08:33:18 ahh, ok, before mounting May 16 08:33:21 etc May 16 08:33:23 yeah May 16 08:33:34 so I could change the call to wizdvp May 16 08:33:39 I designed the IceLink protocol to use uint64 types, but it looks like BWiz convert them to a pair of int32 types. May 16 08:33:40 tot enable logging to a file May 16 08:33:58 does it log like that? May 16 08:34:02 anything interesting? May 16 08:34:13 it used to produce mountains of logging May 16 08:34:36 aha, hi & low. is 64bit a prob for the Wiz CPU? May 16 08:36:48 Its' not a native type, but they should not need to do that. gcc will handle 64bit values. May 16 08:41:41 frigging WPA bug on the WIz...grrrrrrrrrrrrr May 16 08:42:02 trying to copy the whole filesystem over WPA isn't quite working May 16 08:42:17 i need a low powered 3 port ethernet switch May 16 08:42:35 as my slug is next to the wiz and i have a lan cable there... May 16 08:43:39 I think the smallest you will find is 4 or 5 port May 16 08:44:00 i know, there is all sorts of wonderful usb gadgets, but not many ethernet ones May 16 08:44:35 i was thinking of using a usb-ethernet dongle on the slug, but the performance will probably be dismal (but could it be any worse than 802.11g??) May 16 08:44:56 I have a 5 port Ethernet hub with magnetic feet (so you can clip it to the side of a PC case) that I am not using, but it's only 10Mbit :-( May 16 08:45:21 i think i have a 10meg one somewhere too, with 10base-2 May 16 08:45:41 but if everything had that, i could daisychain with coax May 16 08:46:28 they should revive that standard and get it to gigabit speeds over coax... but i guess collisions are always the problem on a bus setup May 16 08:46:30 You must have a short memory or high threshold of pain if you even entertain Ethernet over coax. May 16 08:47:01 The standard where the bits really can fall off the end of the wire. ;-) May 16 08:47:16 that is the way we used to hookup in lan parties at about 1998 May 16 08:48:25 looks like i might yet again try WEP... May 16 08:49:22 I use WPA May 16 08:49:36 haven't had any issues with it (touch wood) May 16 08:49:45 billion wifi router May 16 08:50:00 but transferring more than about 80meg in one hit? May 16 08:50:13 i can be telnetted in for days on end May 16 08:50:31 but as soon as i transfer anything substantial, it goes lala May 16 08:51:54 The Billions had pretty poor WPA implementation until very recently. May 16 08:53:26 Mind you, the BWiz wireless sucks even more. I bet if they complied with the GPL, the WPA issues would have been fixed by one of us by now. May 16 08:54:35 i had a fiddle with iwconfig today, couldn't do much with it though May 16 08:57:25 I don't have a DP-S1, but WPA will most likely involve wpa_supplicant or some other alternative to it. iwconfig does not handle WPA, just WEP. May 16 08:57:42 hmm, ok May 16 08:58:27 it won't let me set the tx power more than 0dBm May 16 08:59:44 Of course not - 0dBm is the max. May 16 08:59:51 ahh May 16 09:00:03 The tx power is a misnomer, it's really attenuation May 16 09:00:10 just the man pages on the net show an example where they set it to 10dBm May 16 09:00:17 (not dB) May 16 09:00:51 Maybe it depends on each device, but I recall seeing figures like -18 May 16 09:01:21 yep, the man pages show you can use absolute power, or atten or percentage etc May 16 09:01:48 what the heck is cp -r copying at the moment to my nfs... would be nice to see which files are copying currently May 16 09:02:19 i unmounted the HDD first of course :-) May 16 09:03:43 try "cp -va src dst" May 16 09:04:11 ahh, verbose... silly me May 16 09:04:25 You should almost always prefer cp -a over cp -r, so that special files are handled correctly. May 16 09:04:42 and you retain time/date stamps and permissions too. May 16 09:07:38 no verbose option with busybox cp :-( May 16 09:09:07 grr, it recursed up the nfs mount May 16 09:09:21 hmm May 16 09:09:58 Does BW have find and xargs? May 16 09:10:08 yes, find May 16 09:10:16 What about xargs? May 16 09:10:34 i believe i compiled that into busybox too May 16 09:10:51 yep May 16 09:11:00 but not in the stock one May 16 09:11:28 well, you could use those to 1) replicate the dir structure and 2) copy individual files. May 16 09:12:05 or see if you could build BB with enhanced support for cp (if available) May 16 09:12:17 i thought i did... May 16 09:12:21 soo many options for it May 16 09:13:06 aha, lowercase r detects recursion, uppercase does not May 16 09:13:21 so help says -a is same as -dpR May 16 09:13:51 so hence i think it is now working properly with -dpr May 16 09:15:10 You may want something like "( cd src_dir ; find . -xdev -type d -print0 ) | ( cd dst_dir ; xargs -0 mkdir )" to recreate the dirs May 16 09:16:37 But I think you may be better off with a better cp and the right args :-P May 16 09:25:26 bbl May 16 10:06:49 How is it going tonymy01? May 16 10:07:03 kind of :-) May 16 10:07:38 i am doing my best to try and kill it May 16 10:07:39 Kind of +ve or kind of -ve? May 16 10:10:09 can you really copy /proc to another filesystem, it keeps kind of just stopping May 16 10:10:32 You should ignore /proc May 16 10:10:45 It's all virtual information generated by the kernel May 16 10:10:50 but when i pivotroot, of course it is looking for the other proc May 16 10:11:02 you need to remount proc May 16 10:11:07 hmm, interesting actually, /proc is a seperate mount May 16 10:11:18 ahh, thanks! May 16 10:11:37 i should dig out the slug pivot scripts and see how they do it May 16 10:11:42 mount -t proc none /proc May 16 10:12:00 but there are a few other mounts too May 16 10:12:04 usbfs May 16 10:12:06 ramfs May 16 10:12:07 or in your case: mount -t proc none /newroot/proc May 16 10:12:35 mount -t usbfs none /newroot/proc May 16 10:13:00 actually proc/bus/usb but I get the drift now May 16 10:13:33 just need to make sure all the devs are there, otherwise mounting is a bit difficult May 16 10:14:00 all this so i can fiddle a bit more May 16 10:14:21 best bet may be to have a look at the slug to get some ideas. May 16 10:14:43 i just get buried in script after recursive script though May 16 10:14:44 and if we can get rwhitby in here, he'll be able to give you tips as well May 16 10:15:02 but i did find hte crucial slug script about 7weeks ago May 16 10:15:13 and gave me the idea to try something similar with the wiz May 16 10:15:17 There's really not all that many things you need to do, but ordering them right is important (as is error recovery) May 16 10:15:27 but then without an ext partition and also ext support, it makes it hard May 16 10:16:59 the nfs mount is weird too, it hangs until i kill the shell May 16 10:17:10 then after that it is mounted ok! May 16 10:17:32 i am guessing a busybox quirk with the way i compiled it May 16 10:28:24 cool, mounting the /proc again brought up the other mounts May 16 10:30:33 That's because you need /proc/mounts to see the mounts ;-) May 16 10:30:40 yesp May 16 10:30:53 getting there: # mount May 16 10:30:53 rootfs on / type rootfs (rw) May 16 10:30:53 none on /wizfs/dev type devfs (rw) May 16 10:30:53 proc on /wizfs/proc type proc (rw) May 16 10:30:53 usbfs on /wizfs/proc/bus/usb type usbfs (rw) May 16 10:30:54 ramfs on /wizfs/tmp type ramfs (rw) May 16 10:30:56 192.168.0.105:/home/beyonwiz on / type nfs (rw,v3,rsize=32768,wsize=32768,hard,udp,lock,addr=192.168.0.105) May 16 10:30:58 none on /proc type proc (rw) May 16 10:31:13 wizfs is the pivoted original fs May 16 10:31:49 now to see if the hdd check works May 16 10:32:41 it is confused again May 16 10:33:20 Don't forget that you'll still have the standard BWiz processes trying to hang onto the HDD, but I guess they'll probably let go as part of starting the hdd check from the ui May 16 10:33:45 ahh crap, i just worked out why May 16 10:33:56 it is mounted to wizfs/blah May 16 10:34:12 i will have to unmount and mount directly to the /tmp May 16 10:35:04 you should be able to see if a process is holding onto a particular fs by looking in /proc//fd/ May 16 10:35:54 the wiz doesn't hang onto the hdd when in a menu, you can happily unmount May 16 10:36:06 # umount /wizfs/tmp/mnt/idehdd May 16 10:36:06 # mount -t vfat -o rw,sync,nosuid /dev/ide/host0/bus0/target0/lun0/part1 /tmp/mnt/idehdd May 16 10:36:14 it worked! May 16 10:36:18 and /proc/ also has a link called cwd that point to current dir May 16 10:36:22 and the hdd check is cruising along merrily May 16 10:36:55 hmm, seems to have got stuck 1/2 way May 16 10:37:23 hasn't frozen telnet wise May 16 10:37:55 that is good info to know peter, i have often dug a little into the /proc and wondered what the various extras were May 16 10:38:02 yay, fsck finished May 16 10:38:48 is your fsck a wrapper that creates stat files if they are missing? May 16 10:42:02 will be May 16 10:42:21 this is the main reason i wanted to do the pivot_root, so i can test it works without doing a dozen reflashes May 16 10:42:46 the wiz runs "dosfsck" when doing a HDD check May 16 10:43:16 but maybe it will just be easier to run the check for the dodgy stat files on bootup May 16 10:43:51 because i am guessing the Wiz uses an internal unmount prior to running dosfsck May 16 10:44:13 which means i will need to remount the drive, do my thing, then unmount and run the dosfsck May 16 10:44:45 the only issue with a vfat filesystem that i can see is that directory traversal is quite slow May 16 10:45:16 e.g. pulling up the filelist in the wiz itself, in WizFX and by simply doing an LS on the HDD May 16 10:47:15 I like the idea of having a hack that checks on boot if there is a usb drive mounted and if it contains a directory called "openwiz/etc/init.d". If it does, it should try to execute any scripts in that directory, sorted alphabetically. That way you could have all sorts of start up scripts, like "00_statfix" and "10_telnetd" May 16 10:47:41 brilliant as usual Peter May 16 10:48:18 i think that is one for eric :-) May 16 10:48:57 * Yuv422 catches up on the logs May 16 10:49:09 To remove the enhancements, you pull the usb dongle. Hell, one could even provide entire BWiz enhancement packages on a USB key. May 16 10:49:47 and the good thing about usb vs hdd is that the hdd checks etc won't baulk due to applications currently running on the drive May 16 10:49:49 Like an alternative EPG uploader and remote scheduling client on a USB stick ;-) May 16 10:50:16 which is why i am persuing the NFS option at the moment, as it means I don't have to leave my PC to do the work :-) May 16 10:50:23 and symlinks etc work May 16 10:50:45 but this is only for tweaking, obviously not everyone will be able to setup this environ May 16 10:51:18 the USB method is much more robust. i have been thinking about using it for a week or two, but dismissed it due to the filesystem not supporting symlinks May 16 10:51:19 Well, the usb key could contain a hack to nfsmount a dir from your PC - you wouldn't even need to chroot or pivot_root to the NFS drive, just mount-bind it where you need it. May 16 10:51:59 the only reason i am pivot_rooting is to get full write access to the whole wiz filesystem May 16 11:42:37 The networking thread on the forums has an "interesting" post. :-( May 16 11:44:26 hmm, hang on May 16 11:44:39 Not Work Safe! May 16 11:48:00 dunno what the wiz is doing but my LAN LED is flashing like buggery May 16 11:48:15 obviously it does a fair bit of writing in the background May 16 11:48:41 maybe i should keep the ramfs mounted accordingly May 16 11:49:26 are you at work? May 16 11:49:49 there is a lot of those cropping up now, same with on the topfield forum May 16 11:50:13 dave will have to pull his finger out and patch the forum s/w i think May 16 11:50:37 that happened to me today on another thread, quickly managed to hit the back button before anyone saw May 16 11:53:07 I'm trying to display the Kb/s and ETA on my WizMac app May 16 11:53:20 but the transfer speed is swinging wildly May 16 11:53:25 lol May 16 11:53:38 so the numbers jump around. :( May 16 11:53:43 just average it out over 5 seconds or so May 16 11:53:51 yeah May 16 11:53:59 and then let others be the judge of how lame the wiz networking it May 16 11:54:01 is May 16 11:54:08 I'm working at .5 second intervals atm May 16 11:54:16 too fast May 16 11:54:21 try 2 secs maybe May 16 11:54:32 I'm getting about 800KB/s May 16 11:54:39 less then the 5K May 16 11:54:55 over wifi May 16 11:55:00 aha May 16 11:55:25 people whinge about the 5K speed, but 21megabit/s with the turbo patch is damn good May 16 11:55:45 yeah that's not bad May 16 11:56:05 I'm getting 21megabit downlink from my ADSL2+ May 16 11:56:17 hopefully i get that soon May 16 11:56:23 i have the modem ready for it May 16 11:56:25 I must be nice and close to the exchange May 16 11:56:42 but optus have given me conflicting info about whether i will get it or not May 16 11:56:59 I'm with TPG at the moment May 16 11:57:21 are they ok? they used to be pretty dodgy a few years ago May 16 11:57:26 they are alright but they proxy their web traffic which bottlenecks during the peak periods May 16 11:57:32 ew May 16 11:57:47 you need to refresh alot! May 16 11:57:49 mind you, everyone is probably doing that to save a smidgin of b/w May 16 11:57:57 yeah :( May 16 11:58:03 it is really how much cpu can be thrown at it May 16 11:58:35 the wiz hdd checking option most definitely doesn't call the operating system umount May 16 11:58:39 It forces me to focus on my devel work May 16 11:58:42 but it does call the dosfsck May 16 11:58:47 I tend to distract easy May 16 11:58:51 hehe May 16 11:58:55 same here May 16 11:59:16 is there a umount syscall? May 16 11:59:37 i have about 5 hours of tv to watch on the hdd of the toppy and wiz... but decided to knuckle down and get a solution to fix the dodgy stat file issue that is bugging so many people May 16 11:59:42 syscall? May 16 12:00:15 the kernel userland api May 16 12:00:28 swi 0x9000xx May 16 12:00:53 the sound of whooshing passing over tony's head May 16 12:01:01 That's the first thing I look for when disassembling an app May 16 12:01:03 hey, how do you guys do the message May 16 12:01:13 irc message i mean May 16 12:01:20 *hello May 16 12:01:23 then I can find disk access etc May 16 12:01:24 not like that May 16 12:01:31 msg *hello May 16 12:01:31 * Yuv422 like this? May 16 12:01:33 ja May 16 12:01:39 slash me May 16 12:01:47 /me message May 16 12:02:09 hmm.. May 16 12:02:26 * tonymy01 hello May 16 12:02:29 yay May 16 12:02:48 :) May 16 12:03:00 * tonymy01 the sound of whooshing passing over tony's head May 16 12:03:17 hmm, that didn't work too well May 16 12:03:26 * tonymy01 "the sound of whooshing passing over tony's head" May 16 12:03:30 crap, worse May 16 12:03:54 * tonymy01 'test test' May 16 12:04:00 * tonymy01 test test May 16 12:04:03 huh May 16 12:04:09 ok, whatever May 16 12:05:21 syscalls are a great way to find file access in a binary May 16 12:05:40 then you can home in on the code you want to study May 16 12:05:53 thanks mr ARM/MIPS hacker May 16 12:05:54 want to find the umount logic May 16 12:06:06 :) May 16 12:06:22 I think I found some strings from the opensource one in there once... May 16 12:07:48 I think BW are worried that if the release any code under GPL license requirements then people will start to demand they open up wizdvp with all the gpl code contained therein May 16 12:07:59 :) May 16 12:08:20 hey, wow, huh? May 16 12:08:26 the wiz uses the OS mount May 16 12:08:41 the app? May 16 12:08:49 ep May 16 12:08:50 yep May 16 12:08:53 cool May 16 12:08:55 sorry, /bin/mount May 16 12:09:04 (soft linked to busybox) May 16 12:09:11 that means they won't be stripping that out any time soon. :) May 16 12:09:16 which is good. :) May 16 12:10:40 the strange thing is when I "wrapped" the mount (did a "touch /tmp/BULLDUST" in the wrapper for testing) the file appeared 1/2 way thru the check HDD sequence, but the mount showed no HDD mounted still at that point in time May 16 12:11:24 it appears to unmount/mount the drive twice in the process??? May 16 12:12:11 so, i can put the stat fix in like that... the only issue is what else uses the mount? May 16 12:12:57 maybe i can test to see if the hdd is mounted, if not, then that must be the hdd check, and thus trigger the stat fix May 16 12:13:08 (once it is mounted) May 16 12:15:28 the one thing that isn't working during this pivot_root to my nfs drive is video/audio on file playback May 16 12:15:51 it is just a black screen May 16 12:15:59 but the ff/etc navigation works May 16 12:16:08 hmm May 16 12:16:22 i am guessing one of the devs doesn't like being elsewhere May 16 12:17:30 it won't let me unmount or mount the devfs, so i am stuck with it being mounted in, now, /wizfs/dev May 16 12:17:45 so if anything directly refers to /dev, i guess it is caput May 16 12:18:46 yeah that might put a bit of a spanner in the works May 16 12:19:24 but the main thing i wanted to do was see what calls were made to tap into creating a simple means for people to fix their files May 16 12:28:25 I'm off to bed now May 16 12:28:27 cya May 16 12:28:40 cya May 16 16:33:03 Ok peter, finished your request of having usb trigger the tweaks, thanks for putting that idea forward May 16 16:33:15 will release something tomorrow, i need to sleep i think May 17 01:31:40 wb tony May 17 01:32:16 hi May 17 01:32:41 how's the stat fixer going? May 17 01:32:51 got the usb stick hacks working, but now i can't open the stick on my PC :-( May 17 01:33:12 do you think it is related? May 17 01:33:18 maybe I shouldn't write the bootlog to the usb stick and immediately unmount it? May 17 01:33:43 the stat fix works thankfully May 17 01:33:47 did you see they've removed the trccmd app May 17 01:33:48 but i liked peters idea May 17 01:34:00 they have removed the sync command too May 17 01:34:10 what is trccmd? May 17 01:34:22 it was used for setting the ctrace logging level May 17 01:34:29 ahh May 17 01:34:36 you could turn on or off thread traces May 17 01:34:47 trace debug or general msgs etc May 17 01:35:11 as i have said, i have the latest busybox with nearly everything ticked on running on the wiz May 17 01:35:12 I wonder if they've remove the ctrace logging from the wizdvp app? May 17 01:35:25 cool May 17 01:35:42 of course as a seperate binary off the hdd rather than replacing their one May 17 01:35:51 you could always get the trccmd from an eariler firmware if you want to change the logging style May 17 01:35:58 nice May 17 01:36:08 yep, that is how i got the bloated busybox across May 17 01:36:27 by copying the original 192 busybox first, so i could actually use tar etc May 17 01:37:10 hmm, my usb stick is ok now May 17 01:37:18 * Yuv422 is reworking the maths in wizmac to use long double May 17 01:37:34 could have been a faulty connection? May 17 01:37:37 i rebooted my pc just in case May 17 01:37:58 and now can access the stick perfectly. before it was timing out and giving io errors. bloody bill gates May 17 01:38:10 here is the log: May 17 01:38:19 Fri May 16 19:05:25 MDT 2008 May 17 01:38:19 drive mounted and etc/init.d found, lets go loading tweaks May 17 01:38:19 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S01telnet May 17 01:38:19 telnet hack May 17 01:38:19 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S02recfixer May 17 01:38:19 found stat /tmp/mnt/idehdd/recording/7 hd demo_May.10.2008_12.24+54596.44681.tvwiz/stat May 17 01:38:21 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S03wizremote May 17 01:38:23 wizremote starting May 17 01:38:25 Fri May 16 19:19:11 MDT 2008 May 17 01:38:27 drive mounted and etc/init.d found, lets go loading tweaks May 17 01:38:29 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S01telnet May 17 01:38:31 telnet hack May 17 01:38:33 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S02recfixer May 17 01:38:35 found stat /tmp/mnt/idehdd/recording/7 hd demo_May.10.2008_12.24+54596.44681.tvwiz/stat May 17 01:38:37 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S03wizremote May 17 01:38:39 wizremote starting May 17 01:38:54 cool May 17 01:39:12 how are you discovering the usb? May 17 01:39:16 the recfixer/statfixer works by scanning for the first stat file it can find, and then scans all directories in the recording directory looking for the stat May 17 01:39:29 i simply mount the usb May 17 01:39:37 righto May 17 01:39:47 from the front connector? May 17 01:39:57 #!/bin/sh May 17 01:39:57 PATH=/tmp/mnt/idehdd/wiz/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/flash/bin; export PATH May 17 01:39:57 sleep 2 May 17 01:39:57 echo `date` >>/tmp/tweakbootlog May 17 01:39:57 if [ $? -ne 0 ] May 17 01:39:58 then May 17 01:40:00 # try ntfs May 17 01:40:02 echo "usb not vfat, trying ntfs" >>/tmp/tweakbootlog May 17 01:40:04 /bin/mount -t ntfs /dev/scsi/host1/bus0/target0/lun0/part1 /tmp/mnt/usb/tmp May 17 01:40:06 fi May 17 01:40:08 May 17 01:40:10 if [ -d "/tmp/mnt/usb/tmp/etc/init.d" ] May 17 01:40:12 then May 17 01:40:14 echo "drive mounted and etc/init.d found, lets go loading tweaks" >>/tmp/tweakbootlog May 17 01:40:16 for z in /tmp/mnt/usb/tmp/etc/init.d/* May 17 01:40:58 it appears that the scsi id is independant of front or back connector!! May 17 01:41:15 ah k May 17 01:41:17 it worked just as well in the back as it did in the front i think when i tried it once May 17 01:42:30 that's handy May 17 01:42:50 i will try again now though, just in case i was being stupid at 2am May 17 01:44:26 yep, just plugged into the back and get: May 17 01:44:42 -/dev/scsi/host1/bus0/target0/lun0/part1 on /tmp/mnt/usb/Device1/Unit0_P1 May 17 01:44:59 same device... it must allocate sequentially May 17 01:45:04 cool May 17 01:45:22 i should plug another stick in and see which gets priority on boot May 17 01:45:47 now to find my ancient 256 May 17 01:46:15 killed my toppy last night btw May 17 01:46:23 oh no! May 17 01:46:27 how'd that happen? May 17 01:46:50 something didn't go right with my nfs mucking about, and the wiz wouldn't power to standby with the remote, so i went to pull the plug and pulled the toppy May 17 01:47:08 and then it booted up and hung May 17 01:47:12 so rebooted, it was ok May 17 01:47:24 then the 4am boot it has it booted and hung May 17 01:47:39 and today, started with no taps, it booted ok May 17 01:47:46 started again and loaded taps one by one May 17 01:47:58 and it was description extender doing the deed May 17 01:48:23 but now when i start it up, it does the files/folders check, and then after 4mins asks to format the drive..... ouch May 17 01:48:37 i have never had this behaviour before May 17 01:49:02 did you have any important data on the drive? May 17 01:52:59 kind of May 17 01:53:36 i am going to attempt to recover with tophd/rw (was doing that every second day when trying to get the timezone code working without silly crashes) May 17 01:54:11 i need slide out shelves on my home theatre cabinet May 17 01:59:43 grr, front stick priority over rear May 17 01:59:48 I never transfered data from my old 5k May 17 02:00:08 ok, will have to make the mounting algorithm attempt both front and rear May 17 02:00:12 I could never be bothered setting up the usb link May 17 02:00:25 on the 5k... it goes thru my slug May 17 02:01:47 I'm getting a qnap NAS May 17 02:01:58 1terra May 17 02:02:02 hopefully it will be as flexible as the slug May 17 02:02:12 ts-109 II May 17 02:02:19 single bay May 17 02:02:25 it's fan less though May 17 02:02:34 and only draws 14W May 17 02:02:46 hey, this wiz properly supports my 256meg usb stick which has a 1.4meg floppy emulator in it May 17 02:03:08 /dev/scsi/host1/bus0/target0/lun0/part1 on /tmp/mnt/usb/Device1/Unit0_P1 type vfat (rw,nosuid) May 17 02:03:20 nice May 17 02:03:30 it shows it as lun1/disc (the 1.4meg "partition" for want of a better word) May 17 02:03:50 the slug wouldn't even do that, it simply thought my 256 stick was only 1.4meg May 17 02:04:02 1.4Mb so much space! May 17 02:04:14 lol May 17 02:04:23 could copy 0.5seconds of house onto it May 17 02:04:29 why would anyone need more than 640k though! May 17 02:04:35 yeah yeah May 17 02:05:02 27686862848 May 17 02:05:15 THE GOOD, THE BAD AND THE UGLY May 17 02:05:16 the floppy emulation actually somehow depending on if a pc bios supports it will bootup off that drive May 17 02:05:24 ??? May 17 02:05:38 that's the biggest file on my wiz ATM May 17 02:05:53 27gig lol May 17 02:06:06 i hope you are rounding down to meg or gig in your application? May 17 02:06:18 yeah meg May 17 02:06:26 numbers that long i can't work out unless i highlight in groups of 3 May 17 02:06:46 same May 17 02:07:38 I'm testing my app for >4gig support at the moment May 17 02:07:47 hehe, good May 17 02:07:50 I was using ulong May 17 02:07:55 what filesystems do macs use? May 17 02:07:55 I need ulonglong May 17 02:08:03 HFS+ May 17 02:08:20 so this is ok for 4G+ also i presume? May 17 02:08:32 yeah no problems there May 17 02:08:55 it is case insensitive by default though. :( May 17 02:09:09 man, my dad is coming over soon and i have a toppy to fix, a new f/w to brew for the wiz, and release notes....grr May 17 02:09:14 I installed 10.5 with a case sensitive fs though May 17 02:09:25 hehe May 17 02:09:27 the wiz lowercases things too a bit May 17 02:10:06 my better half is out with friends so I'm taking the time to work on WizMac. ;) May 17 02:11:54 good one May 17 02:12:24 I might do a release soon May 17 02:17:16 I'm currently giving the ETA in seconds. ;) May 17 02:17:25 25000s May 17 02:17:26 bit optimistic May 17 02:17:33 hehe May 17 02:17:44 holy crap, this is a lot of time May 17 02:17:56 I'm going to write a function to automatically determine the best representation May 17 02:17:58 is that nearly 10hours May 17 02:18:20 8hours May 17 02:18:27 yeah May 17 02:18:29 using brain maths only May 17 02:18:32 6 May 17 02:18:36 (not wincalc) May 17 02:19:09 for a 4 hour recording May 17 02:19:24 it is pretty dismal May 17 02:19:40 1000KB/s May 17 02:19:40 but at least you can do other things at the same time, like the 5k in non-turbo mode May 17 02:20:06 i would hate for it to have a mode that shuts down everything during transfer like some other PVRs have May 17 02:20:23 I wonder if turning off the timeshift buffer would improve things? May 17 02:20:37 don't think so May 17 02:20:42 but going into the menu will May 17 02:20:59 no decoding May 17 02:21:50 lucky I can resume a partial download. :) May 17 02:22:21 that is a nice feature to have for sure May 17 02:22:33 I haven't added automatic retries yet May 17 02:22:46 but that shouldn't be hard May 17 02:23:03 you can also pause the transfer May 17 02:23:18 and queue transfers May 17 02:23:46 you can also start a transfer while downloading the index May 17 02:24:03 so the wiz http server accepts multiple connections. :) May 17 02:24:43 your reboot.sh script on the wiz: May 17 02:24:51 /bin/sync: not found May 17 02:25:01 ah May 17 02:25:15 do you think it is necessary? May 17 02:25:26 if you didn't hard code the path, then it would be fine, as i have my path to also look in /tmp/mnt/idehdd/wiz/bin May 17 02:25:26 I just added that to stop corruption May 17 02:25:40 did you get corruption? May 17 02:25:41 ah k May 17 02:25:52 or it was a safety measure May 17 02:26:01 I just wanted to make sure the .grace file was removed May 17 02:26:08 yeah just for safety May 17 02:26:25 I wasn't sure if the micom reboot unmounted the drive May 17 02:26:49 ideally we should unmount the drive then reboot May 17 02:27:00 then we wouldn't need any syncs May 17 02:27:11 woohoo: Fri May 16 20:24:44 MDT 2008 May 17 02:27:11 no etc/init.d found on host1, trying host2 USB May 17 02:27:11 drive mounted and etc/init.d found, lets go loading tweaks May 17 02:27:11 loading tweak /tmp/mnt/usb/tmp/etc/init.d/S01telnet May 17 02:27:11 telnet hack May 17 02:27:33 good stuff May 17 02:27:34 now works with 2 usb drives connected, doesn't matter which one May 17 02:27:47 how about the card reader slots? May 17 02:27:49 :) May 17 02:27:51 grr May 17 02:27:54 :-) May 17 02:27:54 hehe May 17 02:28:03 does the P1 have those? May 17 02:28:11 hmm, not sure May 17 02:28:34 will you support loading tweaks from the dvd drive? May 17 02:28:35 ok, work to do May 17 02:28:45 If not, I'll take my unit back for a refund!! May 17 02:28:50 ;-) May 17 02:28:51 haha May 17 02:29:11 shame the date is in korean timezone May 17 02:29:58 feel like doing some more fw hacking? ;-) May 17 02:30:16 hmm, if I export TZ=blah then it is in UTC.... better than a kick in the head May 17 02:39:11 morning May 17 02:39:14 hi May 17 02:39:17 Hi Rod May 17 02:39:19 I read all the backlog overnight already May 17 02:39:21 actually, date -u does the trick May 17 02:39:50 I think the usb key idea is a winner May 17 02:39:57 done May 17 02:40:11 just need to release as either binarys or a f/w patch May 17 02:40:17 ies May 17 02:41:04 is there any way we can inject stuff into /tmp/config/rc.local without modifying firmware? May 17 02:41:25 so supports fat and ntfs sticks, and on either usb port May 17 02:41:36 i don't think so May 17 02:41:43 prl found a way to create a file May 17 02:41:54 but we can't file it with data May 17 02:42:15 is there any settings backup facility which can be used? May 17 02:42:30 one it is in there, then it can be modified to your hearts content and then reboot.sh can keep the mods May 17 02:42:46 but i don't know any way of easity getting it in there May 17 02:43:07 according to beyonwiz, you don't need to factory reset when you f/w upgrade May 17 02:43:18 so no need to backup settings. May 17 02:43:24 there isn't a lot to save anyway May 17 02:43:42 (shame the wiz doesn't remember audio prefs per channel.... grr) May 17 02:43:47 the factory defaults are stored in a cramfs img in /etc May 17 02:47:06 So, shall we start a wiki page on the design of this WizExtensions USB stick concept, design and implementation? May 17 02:47:55 it is done and dusted May 17 02:48:01 i have completed the design May 17 02:48:09 of course if anyone wants to improve :-) May 17 02:50:07 ok, where is the doco? May 17 02:50:24 in this irc thread :-) May 17 02:50:45 OK, does it handle CF cards and SD cards too? May 17 02:50:59 possibly, if they plug in via an adapter to teh usb port May 17 02:51:14 * rwhitby is thinking of a page with software requirements, etc. May 17 02:51:22 I'll start one. May 17 02:51:33 I was referring to the CF and SD card slots in the front of the machine May 17 02:51:36 i don't want to make it a bloated script as it runs at startup May 17 02:51:54 i know, but if we had to accomodate everything, it will start filling the flash etc May 17 02:52:03 [ -e foo ] is not really bloating ... May 17 02:52:29 anyway, which repo is your source going in? May 17 02:52:49 or do you need a new one? May 17 02:52:51 but i have to mount each device, and also try mounting vfat and ntfs May 17 02:52:59 i don't have source, it is a shell script May 17 02:53:06 same thing May 17 02:53:09 hehe May 17 02:53:12 which repo is it going in? May 17 02:53:16 the wiki will do for a shell script May 17 02:53:23 i haven't distro'd anything in my life May 17 02:53:38 (well, not from an opensource publishing perspective) May 17 02:53:51 um, this is openwiz right? that means source goes into a svn system where multiple people can work on it, not on a wiki page. May 17 02:54:38 the wiz doesn't automount any devices on bootup except the hdd May 17 02:54:54 or is your intention for this to be a one-man-supported application? (no problem with that if that's what you intend - just want to know up front) May 17 02:55:12 so the script has to mount, then unmount, hence not wanting to make it support every device the wiz can tak3 May 17 02:55:34 can't you just check /proc/partitions? May 17 02:56:09 anyway, back to the source repo question - where's it going to live? May 17 02:56:18 that doesn't tell me what it currently has does it? May 17 02:56:40 (yes, I will keep harping on this question till I get an answer ;-) May 17 02:56:58 where peter puts his ones May 17 02:57:09 prl May 17 02:57:30 http://svn.openwiz.org/ May 17 02:57:38 ja May 17 02:57:45 does it fit into any of the 5 existing repos, or should we create a new one? May 17 02:58:12 (note that this is just the source repo - it can be delivered quite differently to end users) May 17 02:58:35 (i.e. packaged up with instructions and so forth and put on beyonwizsoftware.net) May 17 02:59:03 ok smarty, it appears that you are correct, /proc/partitions is an active view of currently installed drives May 17 02:59:14 maybe wee should see if BW will play nice first? May 17 02:59:25 maybe they will enable telnet **** ENDING LOGGING AT Sat May 17 02:59:57 2008