**** BEGIN LOGGING AT Sat Feb 14 02:59:57 2009 Feb 14 03:03:44 RoundSparrow: ping Feb 14 03:06:20 i want to test ath9k n mode, but my 802.11n adapter is using g mode, is there a way for force n-mode only? i did add 80211n to hostapd.sh Feb 14 03:07:01 maybe i should force the usb 80211n adapter to n mode only, but...don't know how Feb 14 03:14:56 johnrw, still there? Feb 14 03:15:02 yes Feb 14 03:15:08 done? Feb 14 03:15:17 i get the same error i used to have on other os: mipsel-openwrt-linux-uclibc-ld:arch/mips/ld.script:6: syntax error Feb 14 03:15:19 :( Feb 14 03:17:26 any idea what it could be? Feb 14 03:18:27 http://rafb.net/p/B0F9fK93.html Feb 14 03:19:38 nbd, are you there too? Feb 14 03:21:32 what does ld.script line 6 say? Feb 14 03:24:48 /* Read-only sections, merged into text segment: */ Feb 14 03:24:48 . = ; Feb 14 03:25:21 johnrw, exactly the same error : http://forum.openwrt.org/viewtopic.php?id=18076 Feb 14 03:26:37 in that case his problem was that he set QUILT_PATCHES in his env but in my case i did not set any env variables Feb 14 03:29:37 yassine: maybe you will have to look at your env vars... and see if there are any extraneous stuff ??? Feb 14 03:30:23 http://rafb.net/p/iGAiv884.html Feb 14 03:30:54 I can't tell what Ubuntu does set... and how it handles things... but I know there are a couple of quirks that depart from debian. Feb 14 03:30:55 nothing wrong i would say Feb 14 03:31:09 that's a debian system Feb 14 03:32:12 ok try this... Feb 14 03:32:24 dpkg -l | grep quilt Feb 14 03:32:32 as su maybe... Feb 14 03:33:08 nothing is installed Feb 14 03:33:21 no quilt? Feb 14 03:33:26 install quilt Feb 14 03:33:37 hey xMff! Feb 14 03:33:42 okay Feb 14 03:34:11 hmm looks like quilt is missing as prereq Feb 14 03:34:15 hey johnrw Feb 14 03:34:37 ii quilt 0.45-6 Tool to work with series of patches Feb 14 03:34:57 xMff, the problem is that prereq does not required it Feb 14 03:35:20 [2009-02-13 20:30] I remember I had to install - quilt 0.46-4 Tool to work with series of patches Feb 14 03:35:23 yassine: grep CONFIG_BCM947XX build_dir/linux-brcm-2.4/linux-2.4.35.4/.config Feb 14 03:35:55 nothing is in there xMff Feb 14 03:36:10 yassine: how did you wind up with 0.45 ? Feb 14 03:36:43 johnrw, i just installed it Feb 14 03:37:09 will it now resume if i make V=99 ? Feb 14 03:37:11 okay well then it's just a heads up that there is a later version... Feb 14 03:37:14 yes Feb 14 03:37:27 yassine: try Feb 14 03:37:41 yassine: maybe make target/linux/clean before Feb 14 03:37:56 okay Feb 14 03:38:04 does he need to remove tmp// Feb 14 03:38:13 tmp/ ? Feb 14 03:38:27 no Feb 14 03:38:31 ok Feb 14 03:39:29 okay re make V=99 Feb 14 03:39:58 i guess quilt was the piece i always missed (assuming it will now compile) Feb 14 03:40:06 yeah, it should be quick Feb 14 03:40:33 yeah... and don't say i didn't tell ya so ! :P Feb 14 03:41:12 Support for Broadcom MIPS-based boards (CONFIG_MIPS_BRCM) [N/y/?] (NEW) Feb 14 03:41:17 y or n ? Feb 14 03:41:28 lol Feb 14 03:41:35 allright Feb 14 03:41:37 lol Feb 14 03:41:41 make dirclean Feb 14 03:41:44 i assume y Feb 14 03:41:48 and start from scratch Feb 14 03:41:58 ohh no :( why? Feb 14 03:42:49 just hope and pray THAT works :) Feb 14 03:42:57 to get a sane default config Feb 14 03:43:37 can he just do a make menuconfig and kernel_menuconfig ??? Feb 14 03:44:16 yassine: well, just answer the config questions Feb 14 03:44:42 i did now its compiling again Feb 14 03:44:44 yassine: but not sure whether other stuff is missing too Feb 14 03:45:32 xMff: you going to add quilt to make prereq ? Feb 14 03:46:07 while it's on your mind? Feb 14 03:46:10 LOL Feb 14 03:46:19 I'll ask whether it should be there in the first place Feb 14 03:46:25 it burned me... and burned yassine Feb 14 03:46:50 johnrw, more than 10 times :) Feb 14 03:47:24 it's a real 'gotcha' on systems that don't have it... Feb 14 03:47:53 xMff, i will provide this VM for download once its compiling successfully (well assmung someone want it ) Feb 14 03:48:12 s/assmung/assuming/ Feb 14 03:48:20 what vm VirtualBox or vmWare ? Feb 14 03:48:29 kvm Feb 14 03:48:33 johnrw: I'm not sure whether it's an accident that brcm-2.4 does not build without quilt Feb 14 03:49:03 xMff: ic Feb 14 03:49:33 still i needed it for other platforms.... Feb 14 03:50:32 on a machine that didn't have it... and i always do make prereq before starting a buildroot's final make Feb 14 03:51:03 maybe I was building for my wrtsl54gs... a broadcom... Feb 14 03:52:06 Making dependencies: please wait... Feb 14 04:12:01 xMff, there? Feb 14 04:12:18 its not quilt :( Feb 14 04:12:27 mipsel-openwrt-linux-uclibc-ld:arch/mips/ld.script:6: syntax error Feb 14 04:13:18 I#m giving up for tonight its 5:15 morning maybe tomorow/today evening Feb 14 04:13:26 xMff, johnrw, thanks guys Feb 14 04:13:34 good night Feb 14 04:14:30 http://rafb.net/p/zWWsWQ86.html Feb 14 04:14:54 yassine: dirclean + from scratch Feb 14 04:15:02 with quilt this time Feb 14 04:15:37 okay Feb 14 04:15:52 xMff, you mean make clean and then make V=99 ? Feb 14 04:16:00 yes Feb 14 04:16:10 and then i can go sleep :) Feb 14 05:30:13 hmm. what's the right way to use an external kernel module for <2.6.28, but the vanilla kernel module for >=2.6.28? Feb 14 07:03:45 I got my first data from JTAG ;) The trendnet 652 seems to have a missing solder point required to get JTAG responding. Details here: http://forum.openwrt.org/viewtopic.php?pid=81858#p81858 Feb 14 07:04:26 nice Feb 14 07:07:34 one step closer Feb 14 07:42:27 #define IXP4XX_FEATURE_RESET_NPEA (1 << 11) Feb 14 07:42:41 is that a binary 11 or eleven? Feb 14 07:45:38 * sn9 says it's two to the eleventh power Feb 14 07:48:22 the reason i ask... is in kcalc as soon as I click (x) binary the calculator shuts off everything but a 1 or a 0... so there would not be a way to say raise it to the 11th power without using 1011 as the second argument... which indeed left shifted it 11 times. Feb 14 07:49:02 100000000000 Feb 14 07:49:14 it's strange... Feb 14 07:50:16 it's only strange in binary mode... so i just checked it in decimal and it works out okay there... Feb 14 07:50:39 how is that strange? Feb 14 07:51:04 stupid of me to click the binary option i guess... but i naturally thought that to do a left shift... i would HAVE to be in binary Feb 14 07:51:31 use hex to preserve sanity Feb 14 07:51:46 or that's where i check my results! Feb 14 07:51:52 lol Feb 14 07:52:18 you can't convert between hex and binary in your head at a glance? Feb 14 07:52:25 it just required the args to be entered in binary format... which wasnt so hard Feb 14 07:52:58 if only binary had commas!!!!!!! Feb 14 07:53:09 like the bean counters use... Feb 14 07:54:31 although i'd say it should be in every 4th place... Feb 14 07:55:13 8 nibbles is a strain on the damn eyes... Feb 14 07:55:15 lol Feb 14 10:43:17 nbd * r14506 /trunk/package/opkg/patches/005-array_warnings.patch: opkg: work around a gcc warning bug related to array bounds checking Feb 14 13:21:48 xMff, johnrw even after a make clean its again mipsel-openwrt-linux-uclibc-ld:arch/mips/ld.script:6: syntax error Feb 14 13:22:04 im clue less :s Feb 14 13:22:54 make distclean Feb 14 13:38:14 Alright, so how should I handle a one-port broadcom device in netconfig? Feb 14 13:38:29 It's the asus 330ge **** BEGIN LOGGING AT Sat Feb 14 14:03:53 2009 Feb 14 14:10:51 nbd * r14507 /trunk/include/kernel-defaults.mk: kernel: add missing definition for CONFIG_KALLSYMS_ALL Feb 14 14:11:16 nbd * r14508 /trunk/package/opkg/ (Makefile patches/005-array_warnings.patch): opkg: apparently array bounds checking is gcc 4.3 specific, move -Wno-array-bounds into TARGET_CFLAGS instead of a patch and make it conditional Feb 14 16:08:15 guys Feb 14 16:08:24 to have regex I need PCRE correct ? Feb 14 16:08:39 kind of, yes Feb 14 16:09:14 xMff: I'm having weird issues in my builds Feb 14 16:09:23 xMff: can you merge this one please: http://jan.willies.info/0001-update-p54-firmware-and-fix-mac80211-dependancy.patch Feb 14 16:29:50 RoundSparrow: got 19.6Mbits on wireless here Feb 14 16:46:15 RoundSparrow: you there ? Feb 14 16:48:19 yha, one sec Feb 14 16:49:08 RoundSparrow: could I ask you a favor please Feb 14 16:50:17 if you can, upload a copy of your binary somewhere so I can look it over/compare. What was your client chipset,driver, OS? Feb 14 16:50:42 RoundSparrow: I'm running the 652BRP dude Feb 14 16:50:51 gave me 19.6Mbits almost 20Mbits Feb 14 16:50:53 it's fine for me Feb 14 16:50:58 the issue is not there Feb 14 16:51:05 it's building dansguardian Feb 14 16:51:11 I can build it fine Feb 14 16:51:22 but I'm having an error with pages like newegg.com Feb 14 16:51:32 enconding or compression error Feb 14 16:56:53 Anyone interested in hacking this Cavium multicore 64-bit mips 80211n router: http://www.dlink.com/products/?pid=683 Feb 14 17:04:40 :( Feb 14 17:04:57 i'm here Feb 14 17:05:14 I'd still like a copy of your firmware... just to make sure I'm not building mine some mistaken way Feb 14 17:06:34 xxiao: not unless there is a serious discount ;) That's the price of a netbook these days. Feb 14 17:06:42 hey Feb 14 17:06:59 RoundSparrow: do you want my .config or the .bin ? Feb 14 17:07:08 both ideally Feb 14 17:07:18 RoundSparrow: ahh you want it to test your wireless speed ? Feb 14 17:07:25 yes Feb 14 17:07:30 okay let me copy it for you Feb 14 17:07:42 RoundSparrow: please would you give me a hand with dansguardian ? Feb 14 17:07:45 What was your client chipset,driver, OS? Feb 14 17:08:01 so does your network (newegg,etc) work fine without that? Feb 14 17:08:19 I've never even heard of it before, so not sure how much help I'll be ;) Feb 14 17:09:07 RoundSparrow: that's true, but get cavium to openwrt sounds interesting, and I believe the price will come way down soon Feb 14 17:09:48 RoundSparrow: also, it's the only active POE router I can find, which means you can remotely reset it Feb 14 17:10:47 RoundSparrow: client driver ? it's the 652BRP dude Feb 14 17:10:55 RoundSparrow: it works perfect Feb 14 17:11:03 RoundSparrow: the issue is my build of dansguardian Feb 14 17:11:11 RoundSparrow: When I build it fails Feb 14 17:11:19 RoundSparrow: when I use the build from packetprotector, it works Feb 14 17:11:33 the issue is , they only have builds for Broadcom Feb 14 17:11:55 RoundSparrow: I was going to ask to see if you can build it Feb 14 17:12:04 xxiao: netbooks keep coming down in price too ;) What do you mean only POE router? There are tons of solutions... or do you mean something else? http://www.ubnt.com/products/picostation.php Feb 14 17:12:14 gramulhao: which device are you talking about? Feb 14 17:12:15 gramhulhao, sure, I can try Feb 14 17:12:25 RoundSparrow: widesol.com/charles/openwrt-ar71xx-tew-652brp-squashfs.bin Feb 14 17:12:29 RoundSparrow: widesol.com/charles/openwrt-ar71xx-tew-652brp-squashfs.bin_config Feb 14 17:13:07 gramulhao: client ;) client ;) Wifi does require two systems. What was the other system in your 20Mbps results? Feb 14 17:13:19 RoundSparrow: my laptop Feb 14 17:13:44 RoundSparrow: I put my 652BRP as ap mode and the laptop as client no encryption Feb 14 17:14:25 RoundSparrow: i mean it's the only standard 802.11 af POE product, the "passive POE" is non-standard. certain env requires a standard PoE Feb 14 17:14:52 RoundSparrow: broadcom 4321AG Feb 14 17:15:10 driver, OS? Feb 14 17:15:19 RoundSparrow: vista Feb 14 17:16:03 RoundSparrow: do you know if there is a way to force n-only-mode on ath9k? Feb 14 17:16:09 stock MS drivers? Feb 14 17:16:18 xxiao: no, if you find out, let me know please Feb 14 17:16:52 my SR71 is g/n mode, my 80211n adapter will choose g instead of n, and i found no way to verify the packets(n or g) Feb 14 17:17:00 tried a sniffer to no avail Feb 14 17:17:28 i was trying to use sniffer to tell if it's g or n, tcpdump certainly can not do that Feb 14 17:17:30 xxiao: I have no idea where OpenWRT UCI / LUCI work stands on picking 802.11n things like that {mixed mode, G only, N only}, wide band (20, 40, auto}. I don't want to duplicate effort, but I suspect nobody has started on it Feb 14 17:17:36 RoundSparrow: broadcom drivers v5 Feb 14 17:18:40 RoundSparrow: that's right, /etc/config/wireless will only take 'hwmode a/b/g', nothing 'n', hostapd can add 'n' to b/g, but nowhere to force n-only Feb 14 17:18:48 ok gramulhao: I don't think I actually have a bradcom client chipset... other than I could put a N router in client mode. All my laptops are Intel AGN, Atheros or RALINK. I've tested Vista on Intel, haven't gotten past 11Mbps Feb 14 17:18:55 RoundSparrow: we need some uci support for that first, /lib/wifi/mac80211.sh needs some updates to accept 11n params and /lib/wifi/hostapd.sh needs some work to write the 11n stuff Feb 14 17:19:29 RoundSparrow: I can test atheros to atheros later Feb 14 17:19:40 RoundSparrow: I have two brp652 here Feb 14 17:19:53 client mode is a horror no matter what I talk to Feb 14 17:20:03 (wiht brp652 ath9k in client) Feb 14 17:20:22 I did have it in client mode yesterday and it wasn't that bad dude Feb 14 17:20:25 2MB/s Feb 14 17:20:36 about 16Mbits on wpa2 Feb 14 17:20:49 a client from my broadcom buffalo Feb 14 17:21:19 there is a thread on client mode: http://forum.openwrt.org/viewtopic.php?id=18377 Feb 14 17:21:27 what rate does iwconfig show? Feb 14 17:21:33 xMff: check hostapd.sh, it will set hwmode=g if it's empty, there is also a ieee80211n=y in hostapd for 'n', iwconfig shows 'abgn' though 'iw' can not display speed more than 54mbps, it's a known bug Feb 14 17:21:40 RoundSparrow: all the way to 54mbits Feb 14 17:22:44 RoundSparrow: did you tried my build ? Feb 14 17:22:44 xMff, nabend Feb 14 17:23:11 yassine: $ grep CONFIG_BCM947XX build_dir/linux-brcm-2.4/linux-2.4.35.4/.config Feb 14 17:23:18 gramulhao: as geminijatt reported on that thred, it drops to 1Mbps for me, and I get very erratic ping times Feb 14 17:24:02 xMff, # CONFIG_BCM947XX is not set Feb 14 17:24:04 unboxing my trendnet 652 :) Feb 14 17:24:20 yassine: that's also the reason for your ld.script error Feb 14 17:24:29 RoundSparrow: encrypted or clear ? Feb 14 17:24:30 gramulhao: give me a minute here, like 3 conversations going on at once ;) Feb 14 17:24:35 both Feb 14 17:24:38 should be active? Feb 14 17:25:29 yassine: yep, should be =y, please to a completely fresh build from a completely fresh checkout to ensure a sane default config. Because quilt was missing, some important patches were not applied, like kernel support for broadcom boards etc. Feb 14 17:25:50 I took it to a users group meeting last monday and tried for 30 minutes to talk to a WRT54GL running dd-wrt... would immediately drop to 1Mbps after connect and ping times all over the place. A dozen laptops sitting there doing great ping times and 54Mbps Feb 14 17:25:56 last Wednesday Feb 14 17:26:05 xMff, ok i got your point ( now that the deös Feb 14 17:26:13 let alone the reboots and freezes in client mode ;) Feb 14 17:26:27 RoundSparrow: I will try client in a few Feb 14 17:26:28 oups, now that the deps are fine it should be okay to restart from the scrach Feb 14 17:26:32 re-flashing here Feb 14 17:26:38 I've seen this behavior with all new builds in past 3 weeks since I first started trying client mode Feb 14 17:27:03 juhosg reported a bug with a locking issue in the driver Feb 14 17:27:11 xMff, just one question since the compiler is downloaded and build each time is there an option to keep the compiler for every build and not rebuilding it each time? Feb 14 17:27:34 yassine: none that I'm aware of right now Feb 14 17:27:45 ok thanks Feb 14 17:28:23 As far as I can tell, this detailed bug report from juhosg hasn't gotten a peep of attention: https://lists.ath9k.org/pipermail/ath9k-devel/2009-February/001154.html Feb 14 17:29:55 gramulhao: Ok, so you want my helping recreating a compile problem? What do I pick, etc? Feb 14 17:30:24 I need to update and rebuild my trunk tree, I hadn't bothered as I was waiting for next compat-wireless drop Feb 14 17:30:31 RoundSparrow: dansguardian Feb 14 17:30:40 where is it in the menus? Feb 14 17:30:43 RoundSparrow: network Feb 14 17:30:46 which kernel you picking? Feb 14 17:30:51 RoundSparrow: check dansguardian and tinyproxy Feb 14 17:30:53 RoundSparrow: i hope ath9k has a bug tracker system instead of just the mailing list Feb 14 17:31:36 RoundSparrow: didn't sellected a kernel, used default 2.6.28-2 Feb 14 17:31:41 xxiao: if it does, let me know. I wish our damn wiki worked well, it would be nice to have a group focus on the 802.11n / AP81 progressions Feb 14 17:31:45 I mean 2.6.28-2 Feb 14 17:31:56 ok, well the default changes a lot ;) Feb 14 17:32:17 RoundSparrow: I'm using the 2.6.28-2 Feb 14 17:32:36 give me 10 minutes to get some tea made and sit down... i'll try a build Feb 14 17:32:45 is it compile errors you are running into? pastebin them? Feb 14 17:32:49 RoundSparrow: i don't know how to change the wiki scheme, is it a good idea to put the 80211n efforts in a different location for now, say, my site? Feb 14 17:32:53 what's your host OS ? Feb 14 17:33:18 RoundSparrow: centos5 / 64bits Feb 14 17:33:26 RoundSparrow: i have a few drupal sites Feb 14 17:33:36 RoundSparrow: it compiles fine, just does not open some websites Feb 14 17:34:55 xxiao: I'm not sure people are going to read what's on another site consistently. We already have the IRC, forum, mailing list, Wiki, etc. I think the forum is doing OK... as long as someone is editing the messages and keeping up to date. We somewhat abandoned this thread: http://forum.openwrt.org/viewtopic.php?id=18113&p=3 Feb 14 17:35:17 I guess a compromise is to have a Wiki page, but have it point to 5 or 10 forum threads on the various topics / issues Feb 14 17:35:28 RoundSparrow: the first error that you are going to notice is when you run dansguardian Feb 14 17:35:38 my biggest frustration with the wiki: 1) very slow to work on, 2) the AtherosPort page has some damn lock that won't let me revise it! Feb 14 17:35:54 it complays about /usr/share/dansguardian/transparent1x1.gif Feb 14 17:36:02 but the file is there Feb 14 17:36:19 gramulhao: execute this: sysctl -w net.ipv4.tcp_ecn=0 Feb 14 17:36:23 on the router Feb 14 17:36:25 and try again Feb 14 17:38:06 we should probably take some of this over to #OpenWRT Feb 14 17:40:42 ecn is congestion control ? Feb 14 17:41:44 also called: 'drop the package if this is set, because I'm an idiotic router'-bit Feb 14 17:41:44 gramulhao: yes Feb 14 17:41:52 xMff: the issue is not with the network Feb 14 17:42:03 xMff: I can get to any site fine without dansguardian Feb 14 17:42:09 xMff: and tinyproxy also works good Feb 14 17:42:18 xMff: the issue is my build of dansguardian are failing Feb 14 17:42:45 I have a Broadcom WL-500G V1 and I have packetprotector on it running smooth Feb 14 17:42:57 gramulhao: and your builds of dansguardian have ecn off? Last time I played with tinyproxy it was the problem Feb 14 17:43:03 So I build openwrt from scratch and my dansguardian fails Feb 14 17:43:13 If I copy the dansguardian from PacketProtector it works fine Feb 14 17:43:24 ok Feb 14 17:43:46 the problem packetprotector only builds for Broadcom and I'm trying the Atheros Feb 14 17:44:07 maybe some endianess issue? Feb 14 17:44:09 everything works well, 95% of the websites are working Feb 14 17:44:20 few websites faill Feb 14 17:44:23 and firefox says Feb 14 17:44:29 Content Encoding Error Feb 14 17:44:37 aren't broadcom and Atheros both MIPS? Feb 14 17:44:46 RoundSparrow: i agree. we should talk about this at #openwrt really, see u there Feb 14 17:44:47 The page you are trying to view cannot be shown because it uses an invalid unsupported for of compression Feb 14 17:45:08 gramulhao: i think we should talk about this at #openwrt ? Feb 14 17:45:13 RoundSparrow: broadcom mipsel, atheros mips Feb 14 17:46:04 ok Feb 14 17:46:27 much to learn, much to forget ;) Feb 14 17:46:38 xMff: the interesting thing that I get to notice Feb 14 17:46:51 If I run (PACKETPROTECTOR) dansguardian it loads without any error Feb 14 17:47:02 If I run my build Feb 14 17:47:28 it starts by complaying : Error reading custom image file: /usr/share/dansguardian/transparent1x1.gif Feb 14 17:47:44 but the file is perfectly there Feb 14 17:47:53 and I'm using the same config Feb 14 17:48:23 gramulhao: file permission foo? Feb 14 17:48:28 xMff: no dude Feb 14 17:48:32 I'm not that lame :P Feb 14 17:48:41 it's something in the binary Feb 14 17:48:48 well, strace then Feb 14 17:48:54 I did Feb 14 17:49:03 but there is too much junk Feb 14 17:49:19 I see an fopen Feb 14 17:49:29 and FILE NOT FOUND Feb 14 17:50:04 strace ... 2>&1 | grep -C4 transparent1x1.gif Feb 14 17:52:34 rebuilding strace one sec Feb 14 17:59:22 write(2, "Error reading custom image file: "..., 33Error reading custom image file: ) = 33 Feb 14 17:59:22 write(2, "/usr/share/dansguardian/transpare"..., 42/usr/share/dansguardian/transparent1x1.gif) = 42 Feb 14 18:03:16 gramulhao: this are the only occurences? Feb 14 18:03:26 nope Feb 14 18:03:27 gramulhao: search for open() Feb 14 18:03:41 fcntl(6, F_SETFL, O_RDONLY) = 0 Feb 14 18:03:51 = OK Feb 14 18:04:04 let me take the grep out Feb 14 18:06:43 xMff: still looking Feb 14 18:10:05 xMff: doing a full log again Feb 14 18:11:39 xMff: first is parses the config file Feb 14 18:11:57 fcntl(7, F_GETFL) = 0 (flags O_RDONLY) Feb 14 18:11:57 fcntl(7, F_SETFL, O_RDONLY|O_NONBLOCK) = 0 Feb 14 18:11:57 read(7, "n/transparent1x1.gif'\nfiltergrou"..., 32) = 32 Feb 14 18:11:57 fcntl(7, F_SETFL, O_RDONLY) = 0 Feb 14 18:11:58 nk Feb 14 18:12:00 then Feb 14 18:12:05 the next time it appears Feb 14 18:12:33 fcntl(7, F_SETFL, O_RDONLY) = 0 Feb 14 18:12:33 close(7) = 0 Feb 14 18:12:34 write(2, "Error reading custom image file: "..., 33Error reading custom image file: ) = 33 Feb 14 18:12:34 write(2, "/usr/share/dansguardian/transpare"..., 42/usr/share/dansguardian/transparent1x1.gif) = 42 Feb 14 18:12:34 write(2, "\n"..., 1 Feb 14 18:12:34 ) Feb 14 18:13:24 weird Feb 14 18:13:58 looks like it doesn't even try to test the file Feb 14 18:14:59 xMff: I see erros like ioctl(7, TIOCNXCL, 0x7fb41a20) = -1 ENOTTY (Inappropriate ioctl for device) in other parts Feb 14 18:15:18 first occurence is reading the config file, second is spitting out an error message Feb 14 18:15:30 xMff: would strace show if the software instead of using a system call used a library ? Feb 14 18:15:41 gramulhao: maybe use strace -f Feb 14 18:15:46 that one follows forks Feb 14 18:17:08 yep Feb 14 18:17:16 let's see if it finds it Feb 14 18:18:47 same Feb 14 18:18:56 parses the config file Feb 14 18:19:04 does a close Feb 14 18:19:21 and write's the error reading custom image file Feb 14 18:19:56 the program is lying :D Feb 14 18:20:02 xMff: is ioctl(8, TIOCNXCL, 0x7f859b50) = -1 ENOTTY (Inappropriate ioctl for device) Feb 14 18:20:08 is that common ? Feb 14 18:20:17 gramulhao: hm yes, see that pretty often Feb 14 18:20:30 I see the other open for the other files Feb 14 18:20:34 gramulhao: I'd look at the source now to see what makes the binary print this error Feb 14 18:20:41 maybe it does some other tests Feb 14 18:20:42 open("/usr/share/dansguardian/languages/ukenglish/fancydmtemplate.html", O_RDONLY) = 8 Feb 14 18:21:56 cpp :( Feb 14 18:22:00 reading it now Feb 14 18:23:06 maybe it does some totally unrelated check Feb 14 18:25:05 image = new char[imagelength + 1]; Feb 14 18:25:05 imagefile.read(image, imagelength); Feb 14 18:28:50 xMff: I see that in the source Feb 14 18:28:58 before the error checking Feb 14 18:29:03 there is a imagefile.open(filename, std::ifstream::binary); Feb 14 18:29:41 and is it checked somehow? Feb 14 18:29:43 xMff: do you know anything about the type std::ifstream ? Feb 14 18:29:50 gramulhao: nope :( Feb 14 18:30:08 guglis Feb 14 18:30:10 gramulhao: but on what does the actual error message depend? Feb 14 18:30:21 it tries to open Feb 14 18:30:23 I mean Feb 14 18:30:30 std::ifstream imagefile; Feb 14 18:30:30 imagefile.open(filename, std::ifstream::binary); Feb 14 18:30:30 imagefile.seekg(0, std::ios::end); Feb 14 18:30:30 how does the condition looks like? Feb 14 18:30:30 imagelength = imagefile.tellg(); Feb 14 18:30:31 imagefile.seekg(0, std::ios::beg); Feb 14 18:30:47 it checks Imagelenght Feb 14 18:30:52 and checks image ! null Feb 14 18:31:10 and checks image.good Feb 14 18:31:16 hmpf Feb 14 18:31:23 fancy high-level foo :P Feb 14 18:31:28 yep Feb 14 18:44:54 xMff: when making changes to the source, what is the correct place to do Feb 14 18:45:08 I see that some stuff is overwritten after a make Feb 14 18:45:27 gramulhao: best is to make a patch and throw it in patches/ Feb 14 19:02:22 xMff: hey Feb 14 19:02:34 xMff: strace doesn't show everything Feb 14 19:02:48 somehow they software is not using system calls Feb 14 19:03:28 imagefile.open(filename, std::ifstream::binary); Feb 14 19:03:40 that funcion is working fine but strace does not report it Feb 14 19:10:45 gramulhao: hm Feb 14 19:11:40 xMff: I'm still debugging it Feb 14 19:11:52 xMff: it could be a problem with c++ compiler maybe Feb 14 19:13:56 gramulhao: my next best advice would be printf() debugging... Feb 14 19:15:17 xMff: I'm doing that Feb 14 19:15:35 xMff: I added lots of comment to stdout Feb 14 19:15:45 rebuilding it and going to debug Feb 14 19:24:41 if (!imagefile.good()) { Feb 14 19:24:59 fails on ifstream.good Feb 14 19:25:45 well... god know what qualifies a stream as "good()" Feb 14 19:26:07 C++ manual says: Check if the state of the stream is good for i/o operations. (public member function) Feb 14 19:28:08 I can't understand Feb 14 19:28:33 It checks the length Feb 14 19:28:47 but the file is not good to be read Feb 14 19:28:48 strange Feb 14 19:29:28 dansguardian runs as root? Feb 14 19:29:40 xMff: yep i have it as root Feb 14 19:31:21 it's weird Feb 14 19:31:24