**** BEGIN LOGGING AT Tue Feb 02 02:59:58 2010 Feb 02 03:44:32 nbd: generated using the config I showed? I'd have to reconfigure the router and kick off my clients for that. Feb 02 03:44:41 wait Feb 02 03:44:49 what driver are you using? Feb 02 03:45:08 the broadcom one? Feb 02 03:45:09 I dunno Feb 02 03:45:12 whatever comes with 2.6 Feb 02 03:45:13 ah Feb 02 03:45:17 there lies the problem Feb 02 03:45:32 how do I tell what driver I'm using? Feb 02 03:45:33 b43 does not support multi-bssid Feb 02 03:45:51 so you can only configure one interface Feb 02 03:46:00 I do have b43 loaded Feb 02 03:46:02 i fixed the segfault in svn now Feb 02 03:46:09 ok, I see Feb 02 03:46:11 but you won't be able to use multiple interfaces anyway Feb 02 03:46:12 well, that sucks Feb 02 03:46:23 because the proprietary driver can support it Feb 02 03:46:56 it's probably not that much work to add it to b43 for somebody that knows the hardware Feb 02 03:47:04 is that in the roadmap? Feb 02 03:47:08 no idea Feb 02 03:47:23 i'm mainly working on ath9k these days Feb 02 03:47:29 oh Feb 02 03:47:47 and that ones does support multi-bssid Feb 02 03:47:53 yeah Feb 02 03:48:02 but it also requires a whole new device which costs money :) Feb 02 03:48:05 yeah Feb 02 03:48:13 but there are some faily cheap devices out there Feb 02 03:48:16 that can run openwrt with ath9k Feb 02 03:48:38 I'd rather just setup a full blow PC for routing after problems with openwrt Feb 02 03:48:44 lack of space is really annoying Feb 02 03:48:52 what kind of problmes? Feb 02 03:49:02 not enough space for ssl libraries... Feb 02 03:49:14 oh, well you could get a device with 8M flash Feb 02 03:49:15 plus constant breakage in versions Feb 02 03:49:22 constant breakage in versions? Feb 02 03:49:28 yeah Feb 02 03:49:40 what do you mean? Feb 02 03:49:53 in the past month, I've wasted hours upon hours fixing problems Feb 02 03:50:17 basically, what I mean is that the new releases fix some things while breaking others Feb 02 03:50:19 we're getting closer to releasing the next rable version Feb 02 03:50:24 s/rable/stable/ Feb 02 03:50:31 my typing sucks today Feb 02 03:50:35 np Feb 02 03:51:02 I had to figure out how to build my own firmware to fix most of the problems Feb 02 03:51:27 and if you think about it, that's quite too much for managing a personal network Feb 02 03:51:54 well, the next stable version isn't far away now Feb 02 03:52:00 yeah Feb 02 03:52:03 and it'll take core of most of the common issues Feb 02 03:52:05 there's always the next version :) Feb 02 03:52:19 well, there's a big difference between what we have now and the last stable release Feb 02 03:52:27 that's what they all say to keep you addicted :) Feb 02 03:52:48 the biggest difference is that mac80211 drivers are finally stabilizing Feb 02 03:52:55 but it's also very true in this case Feb 02 03:52:55 especially for 11n atheros stuff Feb 02 03:53:11 so we finally have a common infrastructure to support a lot of wifi hardware well Feb 02 03:53:32 that just wasn't there for the last stable release Feb 02 03:53:45 so there is a lot more to it than just having 'the next version' Feb 02 03:54:06 right, you may very well be correct in saying that Feb 02 03:54:22 but the fact is that I almost became an openwrt developer just so I can have a nicely running network Feb 02 03:54:49 I'm not blaming you or anyone in particular either Feb 02 03:55:19 well, obviously i can't really respond to that, since i don't know which issues you had with which version Feb 02 03:55:22 this is quite a common occurance for free/opensource probjects Feb 02 03:55:35 but i am pretty confident that we're close to having something that just works for most common use cases Feb 02 03:55:50 although maybe not as well on some kinds of hardware Feb 02 03:56:01 heh Feb 02 03:56:07 yeah, that's another thing Feb 02 03:56:24 by the time openwrt becomes stable, my hardware (wrt54gl) becomes obsolete and uncared for by the developers Feb 02 03:56:26 there is less motivation to work on the broadcom stuff than, say, atheros Feb 02 03:56:48 but that's mainly because of the lack of support from broadcom Feb 02 03:57:02 there's always a reason :) Feb 02 03:57:24 there's always a true, but meaningless statement ;) Feb 02 03:57:44 oh well Feb 02 03:57:48 I don't want to waste your time Feb 02 03:57:50 good night Feb 02 03:57:52 good night Feb 02 08:26:21 Any idea what might have broken in the usb stack in the past few days? Feb 02 11:11:27 hehe ;) Feb 02 11:26:56 kaloz * r19497 /trunk/package/kernel/modules/ (netsupport.mk other.mk): add kernel packages for the ixp4xx-beeper and appletalk -- closes #6618 Feb 02 11:38:43 mh, can i somehow make the kernel reset the system on panics? Feb 02 11:40:15 concerning the mini_fo filesystem Feb 02 11:40:41 if i delete a big file already present in the image, should some space free up? Feb 02 11:41:22 only if you duplicated, maybe with changes Feb 02 11:41:34 not if it's the original file Feb 02 11:46:22 i see Feb 02 11:46:45 and how can i see what's filling up my space ? Feb 02 11:47:24 root@OpenWrt:/jffs# du -h . | grep M Feb 02 11:47:27 ^___^ Feb 02 12:07:23 marek * r19498 /packages/net/batman-advanced/files/etc/init.d/batman-adv-kernelland: batman-adv: handle multiple configured interfaces Feb 02 14:41:33 <[Fate]> is there some standard way to define a second source archive for a package? Feb 02 14:49:46 a second source archive? Feb 02 14:51:20 <[Fate]> yes, a second .tar.gz I need to pick up files from Feb 02 15:34:06 nope Feb 02 15:53:32 jow * r19499 /packages/net/rtorrent/ (3 files in 2 dirs): [packages] rtorrent: update to r1123 (#6624) Feb 02 15:55:21 jow * r19500 /packages/libs/libtorrent/ (5 files in 2 dirs): [packages] libtorrent: update to r1123 (#6624) Feb 02 15:56:48 jow_laptop: I guess sav qualifies as maintainer for those packages, no? Feb 02 15:57:11 loswillios: exactly my thoughts, but haven't figured out how to contact him Feb 02 15:58:02 do you know whether he's in irc? Feb 02 15:58:30 no, I don't. Feb 02 15:58:56 replying to the ticket could be worth a try Feb 02 16:01:59 did that Feb 02 16:02:24 cool Feb 02 17:13:58 anyone know if there is a problem with arptables? I seem to get 'accept' instead of 'return' after processing a user defined chain Feb 02 17:15:10 I also tried an explicit return rule, didn't help Feb 02 18:49:56 Hello. Any idea what might have happened to the USB subsystem on ar71xx architecture in the past few days? I am testing a Huawei E1752 HSPA modem and after usb_modeswitch (that is supposed to switch from CD-ROM mode to modem mode) /dev/ttyUSB0 fails to show up. Feb 02 18:50:42 Creating it by hand does not work either. At the same time usb_modeswitch reports success (and indeed detaches the storage driver). Feb 02 18:51:47 Interesting log entry: OpenWrt user.warn kernel: usb 1-1: usbfs: process 1326 (usb_modeswitch) did not claim interface 0 before use Feb 02 18:53:02 This used to work, though I am not sure with which revision (I did not expect it to break, so I haven't paid attention to the old version when re-flashing -- my fault) Feb 02 18:58:48 Another, tangentially related question is whether or not it is planned to support udev in addition to hotplug2? Feb 02 18:59:33 Guest25031: did you load usbserial with vendor and product id params? Otherwise it will not claim the usb device after mode switch Feb 02 19:02:46 No. Feb 02 19:03:01 usbserial was actually already loaded, I believe. Feb 02 19:03:21 then edit /etc/modules.d/*-usbserial and add the params there Feb 02 19:05:58 What exactly are the implications of usbserial not claiming the device? Feb 02 19:06:37 no ttyUSB0 is created :) Feb 02 19:08:05 Isn't there some hot-pluggable way of doing this? Feb 02 19:08:19 none that I am aware of Feb 02 19:08:56 well it is hot-pluggable once usbserial is loaded with the current vendor/device ids Feb 02 19:09:10 s/current/correct/ Feb 02 19:10:34 Right, but the vendor/device ids are known to the kernel upon plugging in the device (they get logged). Isn't it possible to load usbserial with the correct parameters on the fly? Feb 02 19:10:48 How does it work in "big" linux? Feb 02 19:11:00 the ids usually change after modeswitch Feb 02 19:11:35 and big linux ships with big fat quirk tables to handle every different braindead way of circumenting the fake mass storage Feb 02 19:12:02 if at all Feb 02 19:12:22 just search for ubuntu + usb_modeswitch Feb 02 19:13:19 it usually boils down to some hand crafted solution Feb 02 19:22:31 Let's see... Feb 02 19:32:27 No bananas. :-( Feb 02 19:32:53 Actually, I believe that usb_modeswitch does not fully succeed. Feb 02 19:39:53 Yepp. That's the problem. Feb 02 19:40:26 On my desktop machine, after running usb_modeswitch, the usb product id changes to 0x1001 Feb 02 19:40:56 It used to do that last week on the router as well, but something changed. Feb 02 19:41:22 https://dev.openwrt.org/changeset/19354 Feb 02 19:41:55 On my desktop machine, I use 0.9.4 Feb 02 19:44:44 I'll roll back this change and try again. I suspect that switching to the new version of usb_modeswitch is the root of the problem. Feb 02 20:05:20 Ouch! The old version has been removed from the upstream archive. Feb 02 20:50:11 Nope, downgrade to 1.0.7 doesn't make it work either. Feb 02 21:16:55 I tried usb_modeswitch v 1.1.0 on my PC and it also works. So something else must have changed. Feb 02 21:23:04 Any ideas what might affect the operation of usb_modeswitch? Feb 02 21:23:47 The same version works on the PC and does not work on the router. Also, it worked a few days ago on the router, though that was an earlier build. Feb 02 21:24:19 Guest25031: there are verbose options for usb_modeswitch Feb 02 21:24:21 did you try those? Feb 02 21:35:32 Yes, of course. Feb 02 21:35:48 The output is essentially identical. Feb 02 21:36:06 Guest25031: could you show it to me? Feb 02 21:36:10 Guest25031: and dmesg? Feb 02 21:36:38 Of course. Feb 02 21:36:46 PC first: Feb 02 21:38:40 Reading config file: /etc/usb-modeswitch.conf Feb 02 21:38:40 * usb-modeswitch: handle USB devices with multiple modes Feb 02 21:38:40 * Version 1.1.0 (C) Josua Dietze 2010 Feb 02 21:38:40 * Based on libusb 0.1.12 Feb 02 21:38:40 ! PLEASE REPORT NEW CONFIGURATIONS ! Feb 02 21:39:20 :/ Feb 02 21:39:42 please use pastebin or something Feb 02 21:39:43 Where did it cut me off? Feb 02 21:41:14 How does pastebin work? Feb 02 21:42:10 openwrt.pastebin.com Feb 02 21:42:21 ping rtz Feb 02 21:42:41 hello cshore Feb 02 21:43:05 sorry, didn't have much time yesteray, had to work on a presentation for physics Feb 02 21:43:17 not a problem Feb 02 21:43:23 but I really hope to get it running in a few h Feb 02 21:45:17 rtz: Today I'm bouncing back between working on some stuff that's for our ODM's firmware and OpenWRT Feb 02 21:45:33 Done. Feb 02 21:45:51 Guest53135: I will need the links to your posts ;) Feb 02 21:45:57 both are output from a usb_modeswitch -W command using the same configuration (listed above). Feb 02 21:46:05 http://openwrt.pastebin.com/m70754a79 Feb 02 21:46:14 rtz: making sure the routers that go out today can be upgraded tomorrow Feb 02 21:46:32 to OpenWRT (when we make the switch) Feb 02 21:48:06 Guest53135: and the dmesg output? Feb 02 21:49:18 one sec Feb 02 21:56:02 It's there now. From the PC, I pasted only the relevant part, from the router the entire output. Feb 02 21:57:22 hello guys, Where I can read about $(INSTALL_DIR) $(INSTALL_DATA) $(INSTALL_BIN), is there any wiki or i need look any script to see how they work? Feb 02 21:57:59 INSTALL_DIR creates a directory, INSTALL_DATA and INSTALL_BIN use install to copy files Feb 02 21:58:01 Guest53135: I need the link a gain Feb 02 21:58:10 the difference being that INSTALL_BIN sets executable permissions Feb 02 22:01:06 hello and thanks ndb, if I have many directories and file I need do one INSTALL_DIR and INSTALL_DATA for each one? or only for top directory? Feb 02 22:02:08 http://openwrt.pastebin.com/m4009f366 Feb 02 22:08:06 Guest53135: sorry, I don't have much of an idea Feb 02 22:08:26 Guest53135: you could try the different options for usb-modeswitch Feb 02 22:08:41 Guest53135: or check if you pc has some additinal patches for it Feb 02 22:11:22 No, it's exactly the same stuff compiled from the same source. Feb 02 22:12:08 Vanilla upstream tarball, no patches. Feb 02 22:12:42 The configuration file for usb_modeswitch was also identical. Feb 02 22:12:44 maybe check the svn logs on OpenWRT to see if there have been changed to the usb stack? Feb 02 22:12:56 Guest53135: different kernel? Feb 02 22:15:06 Yes, the kernel is different. Feb 02 22:15:15 which versions? Feb 02 22:16:38 router: Linux version 2.6.32.7 Feb 02 22:17:06 pc: Linux version 2.6.27-16 Feb 02 22:19:01 :/ Feb 02 22:19:34 Guest53135: you could open a bug report for usb_modeswitch on thier bugtracker Feb 02 22:19:58 Guest53135: or you could try to debug this yourself, but this will be rather difficult Feb 02 22:20:08 It's not a usb_modeswitch bug, I think. Feb 02 22:20:59 Guest53135: maybe not, but those guys are the most likely one to figure it out Feb 02 22:21:02 Here's one thing that I have just discovered: even though /dev/sda appears when the device is plugged in, it cannot be mounted as a cdrom even though it should be possible. Feb 02 22:21:16 And that is before usb_modeswitch is even started. Feb 02 22:22:30 I think, I am missing the entire scsi subsystem and that might be THE problem. Feb 02 22:22:31 what about /dev/scd0 Feb 02 22:23:00 That might be the difference from last week's build. Feb 02 22:24:06 Guest53135: did the kernel change in the meantime? Feb 02 22:24:44 Guest53135: a new version I mean Feb 02 22:24:52 No, but the configuration has. Which is, of course, my fault. Feb 02 22:25:35 Guest53135: you could start over Feb 02 22:25:40 Among block devices, I do have kmod-scsi-core, though. Feb 02 22:25:53 Guest53135: if the problem is still there, try to bisect it Feb 02 22:26:05 Guest53135: it's timeconsuming but your best chance Feb 02 22:26:26 I know. Feb 02 22:26:48 Where the scsi cd-rom driver is supposed to be? Feb 02 22:29:52 OpenWrt user.notice kernel: sd 23:0:0:0: [sda] Attached SCSI removable disk Feb 02 22:29:52 but then Feb 02 22:29:52 root@OpenWrt:~# mount -t isofs /dev/sda /mnt Feb 02 22:29:52 mount: mounting /dev/sda on /mnt failed: No such device Feb 02 22:30:54 Guest53135: maybe it's named differently? Feb 02 22:31:24 There appeared a device named /dev/sda Feb 02 22:31:33 isn't it /dev/scd0? Feb 02 22:31:49 No. Feb 02 22:32:04 do you have the scsi-cdrom driver installed? Feb 02 22:32:34 I think, I don't, because I cannot find it. Where is it in the configuration? Feb 02 22:33:57 oh, I lied, it's not used anymore Feb 02 22:34:32 When I plug in the device, two nodes appear under /dev: 1-1 and sda Feb 02 22:34:58 BTW, what is 1-1? Feb 02 22:35:31 is there /dev/sr0? Feb 02 22:36:04 I think 1-1 is the usb device node Feb 02 22:36:21 bus 1-port 1 or some such Feb 02 22:36:27 No /dev/sr0 Feb 02 22:36:32 hmm.... Feb 02 22:37:16 I'm not sure what is supposed to create the device node, but there is a device major 11, minor 0, for my desktop cd drive Feb 02 22:40:13 maybe the device isn't being detected as the right type (e.g. mass storage instead of cd?) not sure how usb cd drives work that way Feb 02 22:44:13 hi Feb 02 22:44:28 where should i mail patches? Feb 02 22:47:32 see https://dev.openwrt.org/wiki/SubmittingPatches Feb 02 22:48:18 openwrt-devel@, thanks Feb 02 22:51:18 cshore: as you can see from the output, in both cases it is the usb-storage driver that gets detached. Feb 02 22:51:54 But again, one problem is that the virtual cdrom is not accessible before being detached. Feb 02 22:52:08 On the pc, I have no problems mounting it as a cdrom. Feb 02 22:53:30 Hmmm....what is SD Storage? Feb 02 22:53:39 what kernel versions? Feb 02 22:53:56 nv I see from before Feb 02 22:54:23 what patches on the desktop vs OpenWRT? Feb 02 22:54:55 Do you have two devices (an mass storage and CD-ROM)? Feb 02 22:55:04 Guest53135: bisect it, it will be faster then messing around blindly Feb 02 22:55:07 trust me on this Feb 02 22:56:03 Guest53135: if it worked before I got with rtz Feb 02 22:56:10 go with Feb 02 22:57:36 I can check out older versions of openwrt trunk, but how do I compel it to checking out older versions of packages? Feb 02 22:58:11 use svn://svn.openwrt.org/openwrt/packages Feb 02 22:58:23 and use a src-link in your feeds.conf Feb 02 23:00:22 Currently, I have src-svn svn://svn.openwrt.org/openwrt/packages in my feeds.conf.default. Would running ./scripts/feeds update check out an older version? Feb 02 23:00:33 no Feb 02 23:00:44 Guest53135: the main packages are in a subdir of trunk and the additional packages shouldn'T matter Feb 02 23:00:58 Guest53135: you could even try to uninstall the feeds Feb 02 23:01:15 Well, usb_modeswitch is an additional package to begin with. Feb 02 23:01:56 Guest53135: you need to check out svn://svn.openwrt.org/openwrt/packages Feb 02 23:02:14 and add Feb 02 23:02:14 src-link packages /path/to/packages/svn Feb 02 23:02:18 to feeds.conf Feb 02 23:02:27 Oh, I see. Thanks. Feb 02 23:02:27 copied from feeds.conf.default Feb 02 23:03:13 Guest53135: don't forget to comment out the packages you aren't using Feb 02 23:03:13 why the heck takes execv char *const argv[] as an argument and not const char *const argv[]? Feb 02 23:09:21 It's a secret signal from an alien terrorist from the centre of the galaxy Feb 02 23:12:28 if this fucks up my memory managment, I will do some ugly things to them Feb 02 23:13:11 you'd be happy if was only your memory management.... :-P Feb 02 23:14:18 Is hotplug2 a proper replacement for udev in these circumstances? Feb 02 23:15:40 Guest53135: no reason it shouldn't be...from dmesg the problem is the sr0 device isn't detected, not that hotplug is failing to do something with a detected sr0 Feb 02 23:26:25 compiling r19355 (I believe, that was about the one that worked) Feb 03 00:09:03 hi Feb 03 00:09:06 ping thepeople Feb 03 00:09:29 ping nbd Feb 03 00:10:10 pong Feb 03 00:10:49 nbd i read about individual packages that can be maintained by thepeople on the mailing list .. are still people accepted? Feb 03 00:10:55 yes Feb 03 00:11:00 i wanna have some Feb 03 00:11:08 just wrote a mail Feb 03 00:11:11 ok Feb 03 00:11:37 openssl, ctorrent, irssi, openssh would be nice Feb 03 00:11:47 should be no problem, except for openssl, i thin we're not currently accepting package maintainers for trunk Feb 03 00:12:04 i think a change to get busybox isnt available Feb 03 00:12:09 chance Feb 03 00:20:51 nbd do u know when thepeople is normally around? Feb 03 00:21:10 no Feb 03 00:21:26 nbd and did u have time to look at the kernel oops on brcm47xx? Feb 03 00:21:45 some advice would be nice .. then i could maybe fix it on my own Feb 03 00:22:04 no, didn't have time Feb 03 00:22:43 maybe its only a ocf error - then i should maybe ask on the mailing list .. but i dont know if its introduced by some patches Feb 03 00:27:18 puchu: did it work some time ago? Feb 03 00:27:49 rtz2 what worked? Feb 03 00:28:02 puchu: without oops Feb 03 00:28:25 rtz2: yeah it works right now when i insert the module befor hotplug2 is restart in /ettc/init.d/boot Feb 03 00:28:41 this error is weired Feb 03 00:28:55 puchu: do you have some additional informations I could look at? Feb 03 00:29:04 i send a mail to the mailing list Feb 03 00:29:06 some mails Feb 03 00:29:24 brcm 47xx 2.6.32 update patch Feb 03 00:29:35 im on 2.6.32.7 right now Feb 03 00:30:45 puchu: what subject? Feb 03 00:30:54 "brcm 47xx 2.6.32 update patch" Feb 03 00:31:32 the last one includes "all" Feb 03 00:32:43 rtz2 on the mailing list no one ever answered to help me on this problem Feb 03 00:37:13 rtz2 are u looking at it? Feb 03 00:37:18 puchu: yes Feb 03 00:37:31 weired aint it? Feb 03 00:37:59 and it only happens when i pivot root to the usb disk Feb 03 00:38:05 @rtz2 Feb 03 00:42:09 rtz2: i found the error Feb 03 00:42:14 its not a kernel bug Feb 03 00:42:22 its a bug on my own Feb 03 00:42:37 a "script" of mine was causing this Feb 03 00:42:51 nbd u can update brcm47xx to 2.6.32.7 Feb 03 00:42:53 its stable Feb 03 00:43:03 jupi ... Feb 03 00:43:12 nbd * r19501 /trunk/package/mac80211/ (14 files in 2 dirs): mac80211: update to latest wireless-testing from today Feb 03 00:43:16 later... working on wifi stuff at the moment Feb 03 00:43:21 okay Feb 03 00:43:32 rtz2: did u read this? Feb 03 00:43:38 yes Feb 03 00:43:41 ok, then Feb 03 00:43:51 a arch update on my own without a serial console :) Feb 03 00:52:27 nbd whatfor is devtmpfs needed? Feb 03 00:52:34 is it doing anything by now? Feb 03 00:52:43 when i umount it befor i insert the modules Feb 03 00:52:47 the error accures Feb 03 00:52:59 altough there is a devfs in /dev Feb 03 00:54:36 rtz2 do u know this? Feb 03 00:55:50 seems that hotplug2 is messing around when its not there any more Feb 03 00:57:23 puchu: it isn't used or needed at the moment Feb 03 00:57:37 puchu: it should get unmounted pretty early Feb 03 00:57:47 i tried this but this doesnt work Feb 03 00:57:56 the system is unbootable when i do this in preinit Feb 03 00:58:06 puchu: when exactly did you do it? Feb 03 00:58:07 i also wanted to use it instead of devfs Feb 03 00:58:14 in 20_device_fs_mount Feb 03 00:58:23 befor /dev is created Feb 03 00:58:35 and some time later also at different points Feb 03 00:58:44 always the same - system is unabootable Feb 03 00:58:54 maybe it should be used when its there Feb 03 00:59:02 nbd ment thats no good idea Feb 03 00:59:51 puchu: where exactly? Feb 03 01:00:00 rtz it should be used? Feb 03 01:00:10 20_device_fs_mount Feb 03 01:00:19 puchu: line number? Feb 03 01:00:39 "use devtmpfs if available and dont let it be unused" Feb 03 01:00:43 this mail Feb 03 01:00:48 it works Feb 03 01:01:32 I can understand, why nbd doesn't want to use it Feb 03 01:01:51 i also dont know .. he ment the device names maybe get fucked up Feb 03 01:01:52 puchu: could cause subtle breakage later on due to the fact that it create device nodes that may have the wrong name and/or permissions Feb 03 01:02:13 so it should be removed when its not used Feb 03 01:02:14 puchu: it's very simplistic Feb 03 01:02:33 hotplug2 seems to mess around with it Feb 03 01:02:38 or used Feb 03 01:02:51 I think it should be disabled in the kernel config Feb 03 01:03:11 i agree Feb 03 01:03:29 we've been doing fine without it for a long time now Feb 03 01:03:32 I has some pros, but somebody would have to spend some time and work out all the kinks and problems Feb 03 01:03:39 and in the end, it may not worth it Feb 03 01:03:44 here it works without problem Feb 03 01:03:55 but maybe in some pioint in time it could introduce porblems Feb 03 01:04:10 i don't want it to cause hidden surprises for us Feb 03 01:04:17 puchu: it's more a question of subtle breaks later Feb 03 01:04:28 also, i have yet to see a practical advantage to using it Feb 03 01:04:36 so lets remove it :) Feb 03 01:04:37 puchu: what do you gain by using it? Feb 03 01:04:46 puchu: where exactly did you try to unmount it? Feb 03 01:04:55 i dont know ... maybe a little speed up at boot time .. i used it because its there Feb 03 01:05:05 puchu: there's *no* speedup Feb 03 01:05:06 at all Feb 03 01:05:12 since hotplug2 has to do the coldplugging anyway Feb 03 01:05:15 rtz: befor and after devfs is mounted Feb 03 01:05:25 and befor exec init is called Feb 03 01:05:32 system is ubootable Feb 03 01:05:42 disable it and enable it when its ever needed Feb 03 01:05:50 puchu: devfs is 2.4 and brcm47xx uses 2.6 Feb 03 01:06:16 so, it will blow up Feb 03 01:06:17 mount devfs /dev -t devfs Feb 03 01:06:28 do_mount_devfs() { Feb 03 01:06:29 mount devfs /dev -t devfs Feb 03 01:06:29 } Feb 03 01:06:32 puchu: and? Feb 03 01:06:34 .t devfs Feb 03 01:06:38 thats 2.6 code Feb 03 01:06:45 it's not used Feb 03 01:06:48 it used Feb 03 01:06:54 if grep -q devfs /proc/filesystems; then Feb 03 01:06:54 do_mount_devfs Feb 03 01:07:00 2.6 does not have devfs Feb 03 01:07:03 so it never gets called Feb 03 01:07:07 devfs != devtmpfs Feb 03 01:07:09 ohh okay Feb 03 01:07:21 but i umounted it befor the if Feb 03 01:07:25 if grep -q devfs /proc/filesystems; then Feb 03 01:07:50 also umounting it after preinit causes the oops Feb 03 01:07:54 for cryptodev Feb 03 01:08:10 puchu: are you sure, it's mounted at all? Feb 03 01:08:18 devtmpfs? Feb 03 01:08:27 yes for shure Feb 03 01:08:43 rtz: it's automounted by the kernel if there is a rootfs= line to the kernel Feb 03 01:08:56 rtz: and devtmpfs is enabled in the kernel config Feb 03 01:09:03 the only sensible place to get rid of devtmpfs is just before hotplug2 does the coldplugging Feb 03 01:09:19 nbd: couldn't we just disable it in the kernel? Feb 03 01:09:20 there i did it Feb 03 01:09:28