**** BEGIN LOGGING AT Tue Jan 13 02:59:57 2009 Jan 13 03:00:11 p54 doesn't look too bad and it's actively maintained Jan 13 03:00:21 ok... cause rwhitby said he's still looking... Jan 13 03:00:24 and it's the only driver besides ath5k that supports multi-rate retry for minstrel ;) Jan 13 03:00:25 yeah, p54 claims to work in AP mode Jan 13 03:00:31 It DOES Jan 13 03:00:51 oh, cool. Jan 13 03:00:55 I pinged it all day and night... with real losses. Jan 13 03:01:11 with NO real losses* Jan 13 03:01:15 ;) Jan 13 03:01:19 johnrw: haha, minor difference. Jan 13 03:01:20 ooops Jan 13 03:01:38 i was like now why'd my fingers say THAT! Jan 13 03:03:55 because they can Jan 13 03:03:56 ;) Jan 13 03:04:31 Wireless USB has kinda been my Holy Grail... and I am happy with the solution... I thought it was a great New Year's Day present from linux-wireless; although I am thinking that the crappy plastic cover needs some ventilation if I am going to give it 24/7 duties... Jan 13 03:10:51 rwhitby: I'd say... Go ahead and get one... but I have yet to test how fast it is... I was just happy to connect! Finally! If you build from openwrt, and get the compat-wireless... it will work. You may have to use hostapd... as I think I recall some cmdline tools failing when I did a mode master. hostapd just got busy and set it up. Jan 13 03:14:05 I guess I'll sign off with this... http://wiki.openwrt.org/OpenWrtDocs/Hardware/USRobotics/USR8200#head-44859a8edf25cb34e4282d3e2ee6735ee715b77e Jan 13 03:14:13 g'night guys Jan 13 03:21:07 nbd * r14017 /trunk/Config.in: clean up build system settings and make some of them more accessible Jan 13 03:21:28 nbd * r14018 /trunk/ (5 files in 3 dirs): add support for an optional staging directory for debug binaries Jan 13 03:21:41 nbd * r14019 /trunk/toolchain/uClibc/Makefile: do not strip uclibc Jan 13 03:40:19 rwhitby: This Bud's for you! Jan 13 03:40:19 root@OpenWrt:~# usb 1-3: new high speed USB device using ehci_hcd and address 3 Jan 13 03:40:19 usb 1-3: configuration #1 chosen from 1 choice Jan 13 03:40:19 usb 1-3: reset high speed USB device using ehci_hcd and address 3 Jan 13 03:40:19 usb 1-3: firmware: requesting isl3887usb Jan 13 03:40:40 phy1: p54 detected a LM87 firmware Jan 13 03:40:40 p54: rx_mtu reduced from 3240 to 2376 Jan 13 03:40:40 phy1: FW rev 2.13.24.0 - Softmac protocol 5.9 Jan 13 03:41:27 phy1: cryptographic accelerator WEP:YES, TKIP:YES, CCMP:YES Jan 13 03:41:27 phy1: hwaddr 00:0f:66:6b:9f:c9, MAC:isl3887 RF:Frisbee Jan 13 03:41:55 neat, it has hwaccel for all three? Jan 13 03:43:56 I dunno... but that's what it says... (I just unplugged it and re-inserted... so that's why it's on phy1) I do know I told hostapd to use wpa-psk (version 1) and it did so. Jan 13 03:45:13 Ok... now I am gonna setup wpa-psk and test... throughput... Is there a preferred tester to use linux to linux? Jan 13 04:21:05 who can give me a thumbnail sketch of the make file hackery, the "blatant disregard and abuse of the traditional make format" part. anything to get me started understanding the thing? Jan 13 04:21:38 russell_: http://downloads.openwrt.org/kamikaze/docs/openwrt.html#x1-310002 Jan 13 04:22:05 though you actually want: http://downloads.openwrt.org/kamikaze/docs/openwrt.html#x1-380002.1.2 Jan 13 04:22:34 no, that's how to create a makefile for a package Jan 13 04:22:44 i want to understand how the templating works Jan 13 04:23:59 that is, a primer on the contents of $(TOPDIR)/include/*.mk Jan 13 04:24:43 russell_: $(TOPDIR)/include/package.mk did it for me after a few head implosions. Jan 13 04:26:33 who is "responsible" for all that stuff? no one seems to have taken credit. Jan 13 04:26:50 * russell_ will look btw, and thanks for the pointer Jan 13 04:26:52 * Bartman007 points at nbd Jan 13 04:26:56 he did it! he did it! Jan 13 04:28:13 i'm interested in things like getting parallel builds working, for example, for which some comprehension of all that stuff would seem essential. Jan 13 04:28:49 when i try -j i get a bunch of failures that go away in serial mode Jan 13 04:28:59 russell_: with trunk? Jan 13 04:29:02 yeah Jan 13 04:29:14 what is your value of ? Jan 13 04:29:18 like 17 Jan 13 04:29:22 * Bartman007 tries Jan 13 04:29:36 russell_: I blame you if I have to reboot the domain. Jan 13 04:29:38 ;) Jan 13 04:29:39 last time i tried was a couple weeks ago, olsrd was one Jan 13 04:29:59 in fact, just restarting seemed to fix each failure Jan 13 04:30:09 typically that is due to missing deps, do you happen to have the errors laying around? Jan 13 04:30:33 I only have access to four core machines so I may not be able to parallelize enough to reproduce a lot of them Jan 13 04:31:03 yeah, mine is a 4 core too, with hyperthreading thingy so it looks like 8 cpus Jan 13 04:31:19 russell_: I'll trade you for that i7. Jan 13 04:31:19 and then i double that for good measure Jan 13 04:31:36 a q6600 is better, I swear! Jan 13 04:31:48 i'd consider buying you one if you'd explain the it all to me. Jan 13 04:32:10 i've got the 920, bottom of the line i7 Jan 13 04:33:36 still costs 150% of what I paid for the q6600 Jan 13 04:34:37 yeah, not cheap yet. Jan 13 04:34:50 and motherboards kick you in the shorts too Jan 13 04:35:24 yeah. I built most of this machine for the cost of your processor ;) Jan 13 04:35:48 the 4x 500gb drives cost as much as the rest of the machine combined. Jan 13 04:35:56 that's what got me started. then it creeped towards "pennies more to go first class" Jan 13 04:37:39 then there was the high wattage power supply Jan 13 04:38:07 I think I picked up a quality 600W seasonic for $50 after MIR Jan 13 04:38:48 shrinkwrap fever had me popping for something retail for $100+ Jan 13 04:41:27 doing a fresh cleandir build on latest svn, I'm getting failure in make[3] -C toolchain/kernel-headers prepare . I'm picking .28 kernel, getting error: scripts/kconfig/mconf.c:91: error: static declaration of 'current_menu' follows non-static declaration Jan 13 04:42:15 svn 14019 Jan 13 04:43:13 oops, changed makefile to kernel .8 instead of .28 by accident Jan 13 04:52:17 ok... Testing this p54 wireless... i am just copying a file stored on a drive connected to the router... through samba. 1GB Windows estimated it at 17 minutes... How far below normal is that? Jan 13 04:55:46 russell_: yay, and the carnage of -j17 on 4 cores begins, it's untarring the kernel ;) Jan 13 04:56:01 nice Jan 13 04:56:59 i haven't bothered to actually test throughput on different N yet. my use of 17 is just a wildassed guess Jan 13 04:57:01 load instantly shot up from 1.04 to 8.21, and climbing Jan 13 04:57:30 run top and press "1" Jan 13 04:57:53 I'm running htop Jan 13 04:58:23 neat, sadly I never knew about that top functionality Jan 13 04:58:43 yeah, someone just pointed it out to me recently Jan 13 04:58:56 apparently it's provided on a need-to-know basis Jan 13 04:59:17 try htop, it's kinda piggish on the cpu, but I like it. Jan 13 04:59:46 * russell_ trying Jan 13 04:59:55 cool Jan 13 05:00:11 highly customizable output Jan 13 05:00:20 omg i'm drowning in iceweasels Jan 13 05:00:40 may want to disable the display of user threads Jan 13 05:04:20 russell_: heh, I couldn't reproduce your olsrd failure. Jan 13 05:05:13 well, it was somewhat random Jan 13 05:05:16 did anything die? Jan 13 05:05:28 oh. it was a "select all" build too Jan 13 05:05:49 I wanted to single olsrd out first Jan 13 05:06:18 modulo stuff that build-fails Jan 13 05:06:26 now going back for the full build Jan 13 05:07:28 i kept restarting and it would make progress Jan 13 05:14:58 agb * r14020 /trunk/package/kernel/modules/fs.mk: Jan 13 05:14:58 fix vfat.ko in 2.6.28 Jan 13 05:14:58 Signed-off-by: Jose Vasconcellos Jan 13 09:00:25 Bartman007: i have replicated the -j17 failure Jan 13 09:01:05 seems to bump between olsrd and openswan Jan 13 09:01:40 without -jN, it builds fine Jan 13 09:04:58 oh. well, i am not sure those are actually what is failing. could be their messages are just adjacent to the failure that occurs elsewhere. i haven't figured that out yet. Jan 13 09:16:53 make[1]: *** [/aux/src/openwrt/staging_dir/target-i386_uClibc-0.9.29/stamp/.package_compile] Error 2 Jan 13 09:21:46 olsrd and openswan rebuild every make, so that's probably why i was seeing those Jan 13 10:15:52 juhosg * r14021 /trunk/target/linux/adm5120/files/arch/mips/adm5120/cellvision/nfs-101.c: [adm5120] NFS-101U/WU: add PCI irq map and change vlan map Jan 13 10:18:46 juhosg * r14022 /trunk/tools/firmware-utils/src/mkcasfw.c: [tools] firmware-utils/mkcasfw: fix magic value of the TN-U200 board Jan 13 13:01:40 juhosg * r14023 /trunk/package/mac80211/patches/ (11 files): [package] mac80211: update ath9k patches Jan 13 13:26:48 hi Jan 13 13:26:54 hello Jan 13 13:37:38 rwhitby: this thing is holding steady at 300+ KB/sec in a ftp transfer from the web, through the p54. Sometimes it jumps to around 400 KB/sec. Jan 13 13:38:28 but it sometimes disappears... Jan 13 13:39:07 Windows says the ap is gone. Jan 13 13:41:29 I probably need to do some usb chip setup... as on mine it sits at 0ms latency. Factory firmware had it reporting 128ms latency. Jan 13 13:41:44 ymmv Jan 13 14:04:13 hi guys, I need something like nohup to execute a script, How i can do that? Jan 13 14:04:19 welcome to #openwrt-devel! Jan 13 15:35:22 Gee... I get this when I try to mount a drive that ordinary users can mount... EXT3-fs: Unrecognized mount option "user" or missing value Jan 13 16:26:13 anybody knows about TEW-652BRP support added soon to openwrt ? Jan 13 16:28:00 nbd * r14024 /trunk/package/ (5 files in 3 dirs): replace the gdbserver package with a full gdb package that also includes gdbserver Jan 13 16:30:48 wuhu! Jan 13 16:35:17 nobody ? Jan 13 16:35:41 yeah... it's dead in here Jan 13 16:35:49 not good Jan 13 16:35:53 i have this stupid router Jan 13 16:36:03 it is very bad on the wifi part Jan 13 16:36:32 keep disconnect me ... and if 2 stations try to connect on it, sometimes works, sometimes it crashes both connections Jan 13 16:36:38 very strange behaviour Jan 13 16:37:09 what driver does it use for wifi? Jan 13 16:38:40 what does this mean? "TEW-652BRP support added soon to" Is there support? Is it just basic, not finished support? Jan 13 16:38:55 not sure :( , it seems it has linux on it but telnet/ssh is disabled Jan 13 16:39:03 where did u found that sentence ? Jan 13 16:39:08 on openwrt site ? Jan 13 16:39:12 you said it Jan 13 16:39:25 aaaa Jan 13 16:40:01 i was asking if somebody from the developers plans to add support for 652BRP router ... that's all Jan 13 16:40:18 so you are running factory firmware... Jan 13 16:40:34 yes Jan 13 16:41:04 is there a wiki page for that model? Jan 13 16:41:09 nope Jan 13 16:41:32 then there, you have your answer. Jan 13 16:41:32 the most advanced trendnet router that is in the wiki appears to be TEW-632BRP Jan 13 16:41:46 yah , sort of :) Jan 13 16:42:20 cage_31: http://forum.openwrt.org/viewtopic.php?id=18314 ;-) Jan 13 16:44:17 i mean: cool Jan 13 16:45:30 yah thanks juhosg Jan 13 16:45:51 np Jan 13 16:46:11 strangely i searched the forum for 652brp but it came up no results , or perhaps i forgot to check some boxes Jan 13 16:46:44 google site search probably works better for needles in a haystack Jan 13 16:47:59 so juhosg, what about my question? Jan 13 16:49:49 is this a case of filesystem drivers being gutted, or is this a case of a really deceptive error message... ie something is wrong somewhere else... but has no message to indicate the source of the trouble? Jan 13 16:57:17 thanks for your help. best regards. Jan 13 17:22:47 bad bad design... guess I'll just have to settle for guest account = root Jan 13 17:23:22 johnrw: 'user' and 'users' mount options are disabled in busybox Jan 13 17:24:01 juhosg: It seems that even in the ext3 driver... Jan 13 17:25:00 I found that out when I snipped the /bin/mount link to busy box, and installed the real mount command. Jan 13 17:30:07 besides... there seems to be some policy file that keeps creating the mount option onerror=continue... and that has to be a great idea... like I want a disk with errors on it to silently continue Jan 13 17:38:33 juhosg: in my menuconfig under the busybox tree in Linux System Utilities [*] Support lots of -o flags in mount was selected... Jan 13 17:41:25 the help for it says... Without this, mount only supports ro/rw/remount. With this, it supports nosuid, suid, dev, nodev, exec, noexec, sync, async, atime, noatime, diratime, nodiratime, loud, bind, move, shared, slave, private, unbindable, rshared, rslave, rprivate, and runbindable Jan 13 17:42:26 user and users is mysteriously just "not there" Jan 13 17:42:55 so where is that option set? Jan 13 17:56:03 nbd * r14025 /branches/8.09/package/opkg/files/opkg.conf: opkg: add the missing overlay_root option for 8.09 Jan 13 20:02:35 agb * r14026 /branches/8.09/target/linux/ (16 files in 14 dirs): Jan 13 20:02:35 merge r13819 to 8.09 Jan 13 20:02:35 Add hostapd-mini to default images where applicable Jan 13 20:04:16 juhosg: when you said the mount options are disabled in busybox you weren't kidding around... Jan 13 20:15:52 johnrw: it depends on the DESKTOP option: http://openwrt.pastebin.com/d72fe496 Jan 13 20:25:35 juhosg: Thank You. In menuconfig... I saw that option and said to myself... naaah... it can''t because of that one not being checked... Jan 13 20:25:50 [ ] Enable options for full-blown desktop systems Jan 13 20:26:40 [ ] Ubuntu mode ;) Jan 13 20:26:41 johnrw: I heard that when you enable that buildroot compiles all of ubuntu instead ;) Jan 13 20:26:46 xMff: damn you! Jan 13 20:26:53 hehe Jan 13 20:26:59 hahaha Jan 13 20:27:10 lol Jan 13 20:27:49 no... ubuntu... please....... I can't search for anything linux related on google anymore... it's all ubuntu users! Jan 13 20:27:59 johnrw: is this still samba related ? I mean your mount problems Jan 13 20:28:35 xMff: Well, it is now like... samba started it... but it's gotten deeper. Jan 13 20:28:58 ic... the dependency chain Jan 13 20:30:08 MateIn4 said to me... I don't think the router should be considered a multiuser system... (I guess that was his way of hinting where the answer was) Jan 13 20:30:59 I asked myself... if that were true... then why maintain file permissions at all? Jan 13 20:31:01 have you tried to make samba shares belong to the nobody user? Jan 13 20:31:24 it's at least an account that is already present on a fresh system Jan 13 20:32:33 I think the thing that actually started this quest... was the Nice looking Network Shares... in LuCI... and the read only toggle Jan 13 20:32:51 hmm but that would force clients to connect as nobody... hmm Jan 13 20:33:11 default config was security = user ? Jan 13 20:34:26 As juhosg showed... i am not the most agile linux user... of that it is obvious... but damn... it should not be this hard to write to a drive... YES, the default option is user in /etc/config/smb.conf.template Jan 13 20:35:54 A newer version of samba build with the apache c++ standard library might be nice Jan 13 20:36:03 then maybe the default config should be redone to implement a security = share approach, that's also more sane in a quasi single-user environment imho Jan 13 20:36:36 esp. when thinking about a nas-like use case Jan 13 20:36:40 I tried changing /dev/sda1 to brw-rw-rw but no cigar... mount still tells me that I must be a Super User to mount. Jan 13 20:38:07 I think most routers rely on their built in firewall... and that most people would simply want to hook a drive to it... if possible, to read and write files... Jan 13 20:38:13 so we agree here... Jan 13 20:38:30 yes, therfore I think that security = user is too much hassle to setup Jan 13 20:38:41 a Fort Knox filesystem I don't need... Jan 13 20:39:26 yeah... I'd bring it up at the next commitee meeting... If I was there... (BY Example) Jan 13 20:39:52 But I don't even know what security = share does... Jan 13 20:40:06 I think it should use kerberos for security Jan 13 20:40:44 kerberos ? Jan 13 20:41:12 Active Directory uses kerberos 5 Jan 13 20:41:25 Its pretty good then you don't need to have local user accounts Jan 13 20:42:43 Its needed for nfs4 as well Jan 13 20:42:49 johnrw: security share is a much less secure permission model, you don't need per-client unix accounts then Jan 13 20:43:29 Does busybox have proper extended attributes ? Jan 13 20:44:27 xMff: But as juhosg has showed... the shell is by default built to prevent users from being owner of their own drives... (floppy cdrom /dev/sda1) Jan 13 20:44:56 removable media should not be guarded like Fort Know. Jan 13 20:45:03 Fort Knox* Jan 13 20:45:10 hmm I'll find a way once I have time to look into it Jan 13 20:45:11 Did they add mountd yet ? Jan 13 20:45:18 not yet Jan 13 20:47:18 xMff: I'd appreciate it... if I could just plug in a drive... and be able to write to it by default... it is a cold day in hell when I plug in a drive... and want it read only... Jan 13 20:48:26 well, I am gonna build a Desktop busybox and see how large it is... Jan 13 20:48:55 a Full Blown busybox... :) geeze Jan 13 20:49:24 I'd call it fatbox then Jan 13 20:50:10 Bartman007: Can you close - https://dev.openwrt.org/ticket/3988 - I don't think its relevent anymore (maybe add https://dev.openwrt.org/ticket/4367) eglibc 2.9 and binutils 2.19 is stable for me (gateway is working better than its ever been) Jan 13 20:50:52 hey my box has lots of unused flash and ram... Jan 13 20:54:45 hi. I had a problem with upnp. I think there might be a problem with /lib/miniupnpd/firewall.sh Jan 13 20:55:13 tzim: which version? from 8.09rc1 or later? Jan 13 20:55:22 it opens the firewall on the lan zone instead of the wan Jan 13 20:55:28 bleeding edge Jan 13 20:55:34 r13960 Jan 13 20:55:53 wasn't working with MSN Jan 13 20:56:58 I changed the script (in a surely ugly way) to call the MINIUPNPD iptables anchor on the wan instead of lan. And it seems to work, now Jan 13 20:57:28 moment Jan 13 20:58:10 waht did you changed exactly? Jan 13 20:59:25 the chain calling the MINIUPNPD chain (and not the anchor - sorruy) Jan 13 21:00:13 can you post a sample rule? Jan 13 21:02:18 I can post whole iptables and modified script Jan 13 21:02:31 or that, throw it in pastebin Jan 13 21:03:23 besides that, you could add/swap "option internal_iface" and "option external_iface" in /etc/config/miniupnpd Jan 13 21:03:42 then you can change the interface without having to edit the script Jan 13 21:05:23 the values are not physical interface names but that from /etc/config/network ("lan" and "wan") Jan 13 21:06:25 http://pastebin.com/m6b6dd921 Jan 13 21:06:46 I posted iptables, script and configs files Jan 13 21:07:14 (just hidden ip ) Jan 13 21:09:22 xMff: no, couldn't have swap the interfaces in upnp config file. It would have swap the miniupnpd deamon arguments Jan 13 21:09:51 ah ic, so it added a jump for lan->MINIUPNPD instead of wan->MINIUPNPD ? Jan 13 21:10:29 xMff: yep Jan 13 21:10:48 okay Jan 13 21:11:03 it's the incomming which has to be allowed when using upnp Jan 13 21:11:22 yes of course, must have been on drugs when I committed that ;) Jan 13 21:11:31 I'll fix it Jan 13 21:11:34 did the same for -t nat rules, as you can see Jan 13 21:11:56 as the portforwarding is also on the incomming Jan 13 21:13:48 okay, thanks for the report, I'll include your changes Jan 13 21:14:56 check what I did. I'm not sure I did it properly Jan 13 21:15:56 I was wondering if I could use upnp on both lan and wlan2 (a isolated wlan for my neiboor). I had problem with same miniupnp on many interfaces on freebsd and pf. Might work better with iptables Jan 13 21:19:04 hmm I tried to support multiple internal interfaces with the scripts, but not sure if it actually works Jan 13 21:19:25 therfore the whole hotplug fuzz instead of a few plain rules Jan 13 21:24:40 hi Jan 13 21:24:45 ping CyrusFF Jan 13 21:24:47 hi Jan 13 21:25:02 phaidros: pong Jan 13 21:25:26 CyrusFF: qry Jan 13 21:27:11 ~. Jan 13 21:36:16 xMff: How would you specify multiple lan on /etc/config/upnp ? (tried 2 lines, 2ifaces separated by space or comma) Jan 13 21:36:24 space Jan 13 21:36:47 I get /etc/rc.common: eval: line 1: syntax error: Bad substitution Jan 13 21:39:44 mmh Jan 13 21:40:07 it's my script that's bogus Jan 13 21:44:37 yeah, I had a loop there Jan 13 21:44:54 but maybe I overlooked sth. Jan 13 21:45:51 my things still loads the internal_iface although it's useless (and i did it badly) Jan 13 21:46:46 I will test on a actual device tomorrow before committing any changes Jan 13 21:47:59 http://pastebin.com/m44c12579 Jan 13 21:48:19 should be cleaner (and seems to work as it should) Jan 13 21:48:26 almost Jan 13 21:50:07 but I think that forwarding rules for lan are still needed as traffic forwarding for lan is forbidden by default Jan 13 21:50:32 xMff: yes indeed. Jan 13 21:51:20 I think the for loop should go over both intif and extif Jan 13 21:51:28 even if I do not see the point enabling upnp in ths case Jan 13 21:51:52 hehe yeah Jan 13 21:52:12 I don't think miniupnpd allow more than one extif Jan 13 21:52:20 yep Jan 13 21:52:24 I chaked Jan 13 21:52:27 *checked Jan 13 21:52:37 it only accepts one -i Jan 13 22:01:08 xMff: http://pastebin.com/m2d6dab73 < this last one seems to work as it shoulf Jan 13 22:01:34 but I think I got the same problem that I got on freebsd Jan 13 22:02:25 okay, I'll look into this tomorrow Jan 13 22:02:34 I have no test device right now Jan 13 22:02:35 Got to check, but miniupnpd annouces itself with the first -a ip address Jan 13 22:03:07 (so windows ignores it as it's not on it's subnet) Jan 13 22:03:39 xMff: thanks for the work anyway Jan 13 22:27:48 Bartman007: your last build from last night still had symbol errors on ath9k Jan 13 22:36:18 xMff: well I built the [*] ubuntu mode Jan 13 22:37:00 is it brown-orange now? ;) Jan 13 22:37:33 lol, still this thing does not honor the 'users' flag. Jan 13 22:37:54 hm I only know "user" Jan 13 22:38:02 man mount Jan 13 22:38:02 to allow users to mount a device Jan 13 22:38:34 ok ic Jan 13 22:38:36 user may allow a user to mount... but it also requires that user to umount... Jan 13 22:39:04 but how is mounting related to samba? Jan 13 22:39:36 it is a problem that has a couple of phases... Jan 13 22:39:56 the first problem... is that only root can mount. Jan 13 22:40:18 do you hope to give a non-root user write access when the drive is mounted by that user? Jan 13 22:40:51 but in that scenario... root is supposed to be able to grant permissions to the kernel to allow ordinary users those priveledges with the users flag in /etc/fstab Jan 13 22:41:40 but that is clearly broken. Jan 13 22:41:57 yes to yoour last question Jan 13 22:42:30 johnrw, before getting into such low level details, it would be good for you to define the problem you're trying to solve. Need a clear use case. Jan 13 22:42:32 a mounted filesystem does not inherit the permissions from the mount point folder for it's root dir Jan 13 22:43:00 MateIn4: It should operate like every man page says Jan 13 22:43:13 if you want to change the filesystem permissions you need to do chmod/chown after the filesystem is mounted Jan 13 22:44:24 and I don't think that mounting as non-root will solve the write problem as long as the permissions don't match Jan 13 22:45:08 my drive has files from ordinary users on it... from other machines I've hooked it up to... Jan 13 22:45:28 and who owns it's root folder? Jan 13 22:45:30 usually uid 1000(the knoppix user's uid) Jan 13 22:46:09 lan@OpenWrt-USR8200:/$ ls -al /dev/sda1 Jan 13 22:46:10 brw-rw-rw- 1 root root 8, 1 Jan 13 17:24 /dev/sda1 Jan 13 22:46:27 no, do the ls on the mount point Jan 13 22:46:30 that is how it is after i chmod'ed it Jan 13 22:47:25 ls -l /mnt/something Jan 13 22:47:25 drwxr-xr-x 2 root root 0 Jan 12 14:54 sda1 Jan 13 22:47:39 that's /mnt Jan 13 22:47:39 so you see, it's r/o for non-root Jan 13 22:48:11 now do chmod 0777 /mnt/sda1 Jan 13 22:48:41 btw... I tried to stick some /dev/sda* special files... in trunk/files/dev and you can imagine how that blew up Jan 13 22:48:49 hehe Jan 13 22:48:51 yep Jan 13 22:49:13 i was not root when doing the build... so it barfed at the end... Jan 13 22:50:10 Now if the buildroot could manage to do a sudo cmd beforehand... it would have worked... Jan 13 22:50:26 but anyhow... Jan 13 22:50:42 lemme do that Jan 13 22:50:49 brb Jan 13 22:51:36 what? compiling openwrt as root? it will refuse to do that Jan 13 22:51:47 lan@OpenWrt-USR8200:/$ mount /dev/sda1 Jan 13 22:51:47 mount: must be superuser to use mount Jan 13 22:52:09 you should try this instead: chmod 0777 /mnt/sda1 Jan 13 22:52:11 i chmod'ed the mnt/sda1 to 666 Jan 13 22:52:21 666 is wrong for directories Jan 13 22:52:38 dirs need executable bits to allow users to browse them Jan 13 22:54:12 ok... did that... still the same super user message... when I am in the "lan" account Jan 13 22:54:47 yeah, my chmod was not meant to solve your mount problem, but "lan" should now be able to touch /mnt/sda1/foo Jan 13 22:55:15 ok... lemme check that at least Jan 13 22:56:33 well that allowed "lan" to write... Jan 13 22:57:08 next step would be to check whether "lan" can also write via samba Jan 13 22:57:51 so /mnt/sda1 needs to reflect what the root's intention is... and fstab is meaningless... Jan 13 22:58:17 yes Jan 13 22:58:21 and here i thought it was the device special file... Jan 13 22:58:43 no, it's the mount point bot only *after* sth. is mounted on it Jan 13 22:58:51 *but Jan 13 22:59:16 if I create a trunk/files/mnt/sda1 with the right permissions... will that get included in the image? Jan 13 22:59:23 nope Jan 13 22:59:38 moment Jan 13 23:00:58 xMff: mmh, sorry if I'm annoying you, but there also something minor regarding the bandwith setting on /etc/config/upnp Jan 13 23:02:20 miniupnpd -h states that bandwidth are in bits/s, down, then upload. Jan 13 23:02:31 johnrw: http://pastebin.com/m4536809d Jan 13 23:03:18 as you see, the mount point's permissions are meaningless if sth. is mounted on it Jan 13 23:03:30 /etc/init.d/miniupnpd use upload first, then download, and the * 1024 / 8 thing doesn't mean anything for a bandwidth setting Jan 13 23:04:09 tzim: the options are meant to be in kBytes/s Jan 13 23:04:16 ( AFAIK, bandwidth use 1000 powers and not 1024 ) Jan 13 23:04:18 tzim: there for they are converted Jan 13 23:04:38 ugh, dunno if it's that newstyle 1000 thing now Jan 13 23:05:02 xMff: if options are meant to be in KBytes, you should multiply by 8 *1024 to get bits Jan 13 23:05:04 I - in fact - never touched that part, I only changed the firewall stuff from broken to broken again :) Jan 13 23:05:31 (and as I said, 10Mbit/s are 10.000.000 bit/s ) Jan 13 23:06:53 yeah maybe Jan 13 23:07:06 (no problem, I shouldn't be so finicky) Jan 13 23:07:16 I never understood why that stupid si stuff is now enforced in compuder science Jan 13 23:07:39 xMff: My goal is to make a pleasant experience for LuCI users... where when a Mountpoint read only checkbox is not checked... the damn thing is writable! :) Jan 13 23:07:53 Telecom world is not SI :) Jan 13 23:08:30 I mean, Telecom world is not computer science Jan 13 23:08:34 I don't know why those values are converted in the first place Jan 13 23:08:54 why they are not jsut passed through to the miniupnpd flags Jan 13 23:09:15 (and in the wrong order :) ) Jan 13 23:09:21 yeah Jan 13 23:10:15 so how is the power-of-2 thing called now correctly? Mibit? Jan 13 23:11:17 yeah Mibit... sounds strange Jan 13 23:11:21 I do not know. but in telco world, all bandwidth are in powers of 10. Jan 13 23:12:20 okay so in fact both the firewall and the bandwidth units are swapped Jan 13 23:12:47 it seems Jan 13 23:13:25 and powers of ten Jan 13 23:13:31 okay so I will fix that too Jan 13 23:16:13 in fact, it depends... if you convert kB/s to bit/s , you may want to multiply by 1024. because in kB/s, it's a transfer speed Jan 13 23:17:03 * 1024 * 8 Jan 13 23:17:07 (you transfer a kB (1024 bytes) in 1 s) Jan 13 23:17:12 yep Jan 13 23:18:13 okay, I have to take some sleep now Jan 13 23:18:18 me too Jan 13 23:18:20 thanks again for your report Jan 13 23:18:37 np. new to openwrt, but great stuff Jan 13 23:19:33 johnrw: I think we'll solve that ;) But not today anymore Jan 13 23:19:36 good night Jan 13 23:20:06 ok... g'night Jan 13 23:52:32 hmm.. problem with the build. scripts/feeds uses the staging_dir/host/bin/find .. even when it doesn't exist. Jan 13 23:53:03 I did: ln -s $(which find) staging_dir/host/bin Jan 13 23:53:10 which seemed to get around it :| Jan 14 00:39:37 Kaloz: Did you specifically test gcc-4.1.2 + eglibc 2.9 (EABI) because it seems to be working great on my gateway Jan 14 00:40:18 (with nptl everything is working openvpn (hw crypto) ath1 and ath0 is bridged all fine) Jan 14 00:53:06 nbd * r14032 /trunk/toolchain/uClibc/patches/200-libpthread_fix.patch: work around problems with the libpthread onexit() handler Jan 14 00:56:08 h3sp4wn: what's the size difference between a uclibc and eglibc build? Jan 14 01:01:01 Bartman007: uclibc is about 500k - eglibc is about 2MB Jan 14 01:01:49 eglibc fits pretty comfortably into 8MB though (I read about the tls and nptl in uclibc still being pretty flaky) Jan 14 01:02:09 (the uclibc-nptl svn branch) Jan 14 01:03:39 nbd * r14033 /trunk/target/linux/generic-2.6/ (2 files in 2 dirs): add some extra kernel exports to be used by ashmem and binder later Jan 14 01:04:21 h3sp4wn: tls should be working with the new linuxthreads implementation in uclibc Jan 14 01:04:27 i just added some fixes for that Jan 14 01:04:41 testing on it has been very light, though Jan 14 01:04:47 so it'll probably still have some issues Jan 14 01:05:12 since android is making heavy use of threads, i'll probably hit some of the remaining issues soon ;) Jan 14 01:05:32 Bartman007: the squashfs.img is 5.2MB with this package set - http://paste.debian.net/25944 Jan 14 01:07:09 nbd: Interesting is the new linuxthreads implimentation comparable in quality to nptl ? Jan 14 01:07:20 dunno Jan 14 01:07:25 it's supposedly based on glibc code Jan 14 01:07:54 h3sp4wn: not bad... Jan 14 01:08:09 maybe it would be worth merging in some code from bionic (the android libc) at some point Jan 14 01:08:26 The last linuxthreads for glibc was pretty bad (at least the way it was being abused by cadence) Jan 14 01:09:05 (That was from 2.5) had to use 2.3.6 and abusing the linker Jan 14 01:10:23 btw. the dalvik vm runs on openwrt now ;) Jan 14 01:11:06 nbd: how long until luci is written in java/dalvik? ;P Jan 14 01:11:30 not. going. to. happen. :) Jan 14 01:12:02 i really don't like java as a language Jan 14 01:12:13 but google has done a nice job on the framework Jan 14 02:34:52 hmm.. having strange problems with 386 target - getting signal.h - no such file or directory Jan 14 02:38:04 http://pastebin.ca/1307744 Jan 14 02:38:54 .. while compiling toolchain/gcc !!! Jan 14 02:42:10 huh? it's using gcc 3.4.6? Jan 14 02:42:14 did you change something? Jan 14 02:45:17 nope Jan 14 02:46:08 Bartman007: your last build of ar71xx last night still had symbol problems on ath9k. I have no clue as to the cause. I'm doing fresh checkouts on empty directory and have no such issue. Jan 14 02:46:22 nbd: as in I didn't change anything :) Jan 14 02:47:47 did you update? Jan 14 02:48:03 you should probably run make dirclean Jan 14 02:48:09 to nuke the toolchain and package builds Jan 14 02:48:15 nbd: I checked, and there's no usr/include/signal.h just linux/include/signal.h Jan 14 02:48:37 I'll update and remove build_dir and staging_dir Jan 14 02:55:55 Bartman007: did you see my parallel build observation from last night? Jan 14 02:56:42 that 'find not found' is a bit of a pain. Jan 14 02:57:39 hmm.. find wasn't built :| Jan 14 02:59:06 that's a bit weird really - what's meant to build find ? Jan 14 02:59:30 * frogonwheels links it in. **** ENDING LOGGING AT Wed Jan 14 02:59:57 2009