**** BEGIN LOGGING AT Sun Apr 02 03:00:01 2017 Apr 02 09:21:06 Pali: the more I look at maemo code, the less I wonder why it was not opened in its whole :( https://github.com/community-ssu/libhildonfm/issues/1 Apr 02 09:22:29 ;) Apr 02 09:23:18 please, somebody fix that and make a pull request, it should be easy Apr 02 11:11:10 hi all Apr 02 11:11:39 i have a old n900, i wanted to know what repositories are people using these days Apr 02 11:12:13 i was thinking since its been years that I have updated it, there must be some repos outdated or closed and maybe there r new ones that the community setup Apr 02 11:12:20 so let me know thanks Apr 02 11:12:32 cssu most notably Apr 02 11:12:39 if you missed that Apr 02 11:12:48 yes i missed Apr 02 11:12:52 ćssu Apr 02 11:12:54 ~cssu Apr 02 11:12:55 from memory, cssu is http://wiki.maemo.org/Community_SSU, or (Community Seamless Software Update) Apr 02 11:13:30 also i am not using the cssu rom, i have the last nokia rom that they released that had some security bug fixed, dont know maybe it was related to ssl or what, many years back Apr 02 11:15:08 you definitely want cssu then Apr 02 11:15:24 as cssu is not a rom, but kind of security.debian.org repos Apr 02 11:20:13 btw Apr 02 11:20:25 who is paying for tmo to keep it hosted and all? Apr 02 11:20:35 maemo foundation Apr 02 11:20:35 is it all community members? Apr 02 11:20:40 yup Apr 02 11:20:40 ok Apr 02 11:20:44 cool Apr 02 11:29:04 you still need mirrors for the old nokia repositories though Apr 02 11:29:15 (since a few things still depend on those) Apr 02 11:29:57 bencoh, its explained in article Apr 02 11:30:13 oh? neat Apr 02 12:39:26 miasma: point is, they could've done better for the same price, ie. better usb-sata chip and use box as a big radiator Apr 02 12:39:31 eh. echan. Apr 02 13:44:32 Is there a way to get a verbose output from NOLO? Apr 02 13:45:27 yes, use debug uart Apr 02 13:46:36 So I need some hardware... Apr 02 13:46:57 quick tip: it's not that easy to get proper hw Apr 02 13:47:33 In fact, I was wondering whether there was some log file with NOLO's output. Apr 02 13:48:35 have some fun: https://wiki.maemo.org/Category:N900_Hardware Apr 02 13:49:13 https://wiki.maemo.org/N900_Hardware_Hacking/serial_dump Apr 02 13:50:38 Ok, so basically I think I can't do that (at least for the moment). Apr 02 13:51:31 yup Apr 02 13:53:49 So let me tell you what I am doing (as usual, I'm experimenting). I'm trying to boot a standard Linux distribution. It's just a test. I have found that mainline kernel has some capabilities for the N900. I tried with the OMAP2plus conif file, and built the kernel. Now I'd try to upload it by Nokia's flasher (upload, not flash). I have been able to do that with Rescue OS's kernel. But no sign of life (with standard Apr 02 13:53:49 Linux). Apr 02 13:54:14 *config file Apr 02 13:54:51 So I was thinking about NOLO's log to understand whether something was happening or not. Apr 02 13:54:54 see instructions on http://elinux.org/N900 Apr 02 13:55:07 if you missed something Apr 02 13:55:46 I have read that page. Let me check again. Apr 02 13:56:04 see u-boot paragraph Apr 02 13:58:34 Ok, it'd like to avoid using u-boot for the moment. I already installed it and in the end bricked the device (not caused by u-boot), and it's been a hard task to unbrick it. Apr 02 13:58:56 I'd like to use the flasher-3.5 as it is outlined in the instructions for Rescue OS. Apr 02 13:59:04 (To upload and boot.) Apr 02 14:04:43 Probably I'm missing the modules. But I'd like to understand whether the kernel is booting at all. Apr 02 14:31:33 ~tell ovaltine_ about lazyflashing Apr 02 14:32:04 ~tell ovaltine_ about BM Apr 02 14:32:31 ~tell ovaltine_ about jrrepos Apr 02 14:32:38 ~tell ovaltine_ about jrtools Apr 02 14:32:40 he doesnt want flashing, he wants to experiment Apr 02 14:36:13 Enrico_Menotti: ((it's been a hard task to unbrick it)) WHAT? Apr 02 14:38:47 Well, I was unable to reflash it after having installed u-boot. I don't give the cause to u-boot - I had messed up some file in Maemo's file system. But in any case, the flasher wasn't recognising the N900 at the usb port. Apr 02 14:39:17 in a chain x-y-z you played with y and nuked it and you found it a hard task to regain it using x (which it is evidently not), so you prefer to play with x now (NOLO) which you don't even know the mystery (coldflashing) you need to use for fixing it? Apr 02 14:40:03 Cold flashing was not working anymore. Apr 02 14:40:16 bullshit Apr 02 14:40:21 Now it's ok - this happened a few weeks ago. Apr 02 14:40:49 coldflashing was not needed in your case Apr 02 14:41:01 It was my first trials with this device - I did not know what I was risking. Apr 02 14:41:18 and coldflashing is only brickable with a sledge hammer, it's in ROMBOOT Apr 02 14:42:18 Ok, maybe cold flashing was working, but I could not flash the device with the usual chain eMMC image - FIASCO image. Apr 02 14:42:38 you did it wrong Apr 02 14:42:56 ~flashing Apr 02 14:42:57 hmm... maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware, or - on linux PC - download&extract http://maemo.cloud-7.de/maemo5/patches_n_tools/maemo-my-private-workdir.tgz, cd into it, do sudo ./flash-it-all.sh; or see ~flashing-cmdline, or see ~lazyflashing Apr 02 14:43:05 has it *ALL*, since ages Apr 02 14:43:20 Maybe, but in the end, if I remember the details, I had to flash just the kernel. Apr 02 14:43:57 well, good luck. Apr 02 14:44:47 In any case, now that's not so relevant anymore, since it's an issue solved. A couple of days ago I also reflashed the whole thing with stock eMMC and FIASCO image, and it's working properly. I followed the instructions on the page you linked right now. Apr 02 14:45:15 (Including removing the battery in-between the two flashes.) Apr 02 14:45:24 it seems you're one of those curious users that are determined to do learning by try&error, not bothering about verifying the assumptions your experiments are based on Apr 02 14:47:38 for booting a kernel you need to provide a few environment items, often kernel commandline. make sure you understand how or even if at all you could adjust this in NOLO to match what your kernel needs Apr 02 14:48:13 Well, the problem is not that I don't want to verify my assumptions. The problem is that I found a bit difficult to find a comprehensive documentation about what I was trying to do. Let me explain: at the beginning, I did not know about Maemo.org, neither about this channel on IRC. I just tried. Now I'm step by step checking what I do. Apr 02 14:48:19 hint: on OMAP3 kernel it seems it's ATAGS not kernel cmdline Apr 02 14:49:04 let me help you out a little bit maybe Apr 02 14:49:16 Ok. Apr 02 14:49:39 the absolute basics: http://www.omappedia.org/wiki/Bootloader_Project Apr 02 14:50:10 on N900 Nokia called their version of uBoot "NOLO" Apr 02 14:51:18 Ah, did not know it's a version of uBoot. I know NOLO is a bootloader, but I thought it was just developed by Nokia. Apr 02 14:51:40 the maemo implementation of uBoot by pali is a combined uBoot+kernel in one "package" that gets loaded by NOLO as if it was a kernel but executes a 2nd layer uBoot that can do all the awesome things a uBoot is meant to be able to do Apr 02 14:53:02 ~uboot Apr 02 14:53:02 N900 uBoot is a siamese twin binary [uBoot+stockMaemoKernel] that resides in kernel NAND partition /dev/mtd3 aka "kernel". You can't uninstall it, rather you'll nuke it when you flash/install another kernel like stock maemo kernel or powerkernel. To start other than stock maemo kernel via uBoot, you have to provide the according kernel image files, or http://talk.maemo.org/showthread.php?t=81613 Apr 02 14:53:33 Ok, I already read something about this. Apr 02 14:54:56 Now for the moment I'd avoid flashing again u-Boot. I'd just want to understand how to put standard Linux kernel in a format suitable to be uploaded (not flashed) and booted by the flasher-3.5. Apr 02 14:56:32 maemo by default has kernel in "an own partition" (on NAND). Pali's maemo-uBoot either loads the stock kernel other siamese half it comes with, or you load kernel.bzimage whatever the name, which can live in maemo rootfs or on eMMC or uSD in a file Apr 02 14:57:29 to boot a kernel from NOLO you need special kernel patches aiui, to fix the atags/cmdline thing Apr 02 14:58:24 uboot cannot load kernel image from maemo rootfs (nand, ubifs) due to huge size of code needed for red-black tree implementation Apr 02 14:58:24 https://www.ixquick.com/do/search?q=uboot+atags Apr 02 14:58:50 so nand+ubifs is disabled for n900 uboot... as there is no more place in nand for it Apr 02 14:59:07 and no special kernel patches are needed to boot directy from nolo Apr 02 14:59:47 hah, now we have an expert here! \o/ thanks Pali :-) Apr 02 15:00:35 Pali As it is done in Rescue OS documentation, right? flasher-3.5 -b"" Apr 02 15:06:49 >>put standard Linux kernel in a format suitable to be uploaded (not flashed)<< alternative kernels are not flashed, they are files in a filesystem and uBoot loads them like on your x86 arch pc Apr 02 15:08:57 so if your concern is about avoiding the flashing of kernels, you don't need to use the rather intricate RAM load from flasher Apr 02 15:11:27 see http://www.omappedia.org/wiki/Bootloader_Project, Pali's uBoot(+stock kernel) replaces the 'Kernel' in that scheme, allowing you to either boot the attached stock kernel OR boot alternative kernels and thus OS from kernel image files you provide on e.g. uSD Apr 02 15:11:43 uBoot is tested and works, no issues with that Apr 02 15:13:09 using RAMloaded initrd based alternative kernel is needlessly complicated and not worth anything regarding safety of the procedure Apr 02 15:14:33 (that would actually depends on the system you're trying to boot ... in some cases you need some initrd for rootfs to become available) Apr 02 15:15:22 well, we are talking about N900. Or what's meant by "system"? Apr 02 15:16:19 n900 with rootfs sitting on a dmcrypt device would require cryptsetup at boot, for instance Apr 02 15:16:49 also note that initrd is a parameter to kernel telling kernel what to load and execute, it's one step later than loading kernel Apr 02 15:17:44 we're talking about loading kernel right now, so initrd or not is not even a issue at that stage Apr 02 15:19:57 DocScrutinizer05 Right. I could flash u-boot and use that as a loader. But for the moment I'd like to try directly from NOLO. Maybe that's not possible. However, it is done with Rescue OS. I'd like to understand how. Apr 02 15:19:57 >>I'd just want to understand how to put standard Linux kernel in a format suitable to be uploaded (not flashed) and booted by the flasher-3.5<< Apr 02 15:21:09 that's documented in rescueOS, why don't you check that and learn how it works? Apr 02 15:22:57 I don't understand what's the use to learn about RAMloaded kernel before you got your kernel booting 'in a normal way', but... that's pretty much up to you Apr 02 15:23:36 Well, there is something about the initrd, as it has been pointed out to me by NIN101, but I did not find the procedure for the kernel. Apart from the flasher-3.5 -b"". Maybe I did not check well enough. Apr 02 15:28:00 https://github.com/quitesimpleorg/N900_RescueOS/blob/master/HACKING Apr 02 15:28:37 https://github.com/quitesimpleorg/N900_RescueOS/blob/master/howtocreate Apr 02 15:29:06 anyway >>I tried with the OMAP2plus conif file, and built the kernel. Now I'd try to upload it by Nokia's flasher (upload, not flash)<< sounds like a pretty poor approach to me, mixing two unrelated problems A and B (aka ramload and new kernel) that both have many possible issues causing the whole thing to fail. A good approach is to first try make one part work (your new kernel loaded via uBoot), and only after that's accomplished try to Apr 02 15:29:08 accomplish the other part, maybe even with a known good unrelated part (try ramloading a kernel that is *known* to work) Apr 02 15:32:26 Rescue OS kernel works, e.g. (RAMloaded). Yes, I should check the new kernel by another procedure, but that would require to flash u-boot, and I'd like to avoid that for now. Apr 02 15:34:30 WHY??? Apr 02 15:36:20 why not, I'd say? booting a new kernel directly from flasher does seem simpler, and should just work, I'd say Apr 02 15:36:32 Just because I'd like not to flash anything on the device for now. As I said, I had to struggle quite a lot for reflashing it. Apr 02 15:36:45 and it requires one less step (updating kernel in emmc/microsd) Apr 02 15:36:55 bencoh Yes, this is my opinion. Apr 02 15:37:54 Enrico_Menotti: you should still flash uboot nonetheless, btw ... even if you don't need it for what you're trying to achieve right now Apr 02 15:37:56 Enrico_Menotti: this proves you lack the knowledge to handle simple scenarios, so you try to avoid the problems in simple scenarios by opting for a complex scenario. doesn't make sense. There is no problem at all in installing uBoot Apr 02 15:38:15 DocScrutinizer05: I don't see how running flasher should be a complex scenario Apr 02 15:38:44 Yes, that's my point as well. Apr 02 15:39:11 I mean, we have two working implementations of it Apr 02 15:39:17 they're both well tested Apr 02 15:39:25 and require no device modification Apr 02 15:39:27 oh yeah, ramloading a system seems sooo simple. It just doesn't work with my leete new stuff so PLZ HELP! Apr 02 15:39:38 ... Apr 02 15:41:46 Ok, the fact I don't know how to do it doesn't mean it is not simple, in my opinion. It's a lack of knowledge from my side, not a higher complexity of the problem. This is my humble opinion. Apr 02 15:45:20 and what makes you think your lack of knowledge would be a minor obstacle when it comes to a very süpecial procedure that's rarely ever used and thus not widely tested and only few experts and documents available to help you out? Apr 02 15:47:45 Yes, this is true as well. If I can't find out how to do it this way, I'll do the other way. But first I'd try to understand how this way works. (And it works at least for one system, Rescue OS.) Apr 02 15:48:27 you want to avoid problems form lack of knowledge how to fix a fsckd up uBoot config (which is a very well known issue easily fixed and you can find a dozen answers to any question you have with it), and you decide the best way is to use a method that basically only Pali, nin101 and fmg have used on a regular basis Apr 02 15:49:20 DocScrutinizer05: and pretty much everyone that has been using rescueOS (?) Apr 02 15:49:46 no, those USED it but don't necessarily understand shit of how it works Apr 02 15:51:05 let me just point out that it's based on NOLO which in turn isn't FOSS like uBoot Apr 02 15:51:33 NOLO and flasher Apr 02 15:51:53 don't know if Pali inverstigated it and implemented it into 0xffff Apr 02 15:52:39 Well, if they may help me, I hope Pali and NIN101 will do. Don't want to disturb too much, however. Apr 02 15:53:54 he did (implement kernel loading in 0xffff) Apr 02 15:53:58 nin101 already helped you,pointing you to his stuff how rescueOS works and how to prepare the image to load Apr 02 15:55:42 [2017-03-30 Thu 14:13:06] https://github.com/NIN101/N900_RescueOS/blob/master/howtocreate https://github.com/NIN101/N900_RescueOS/blob/master/HACKING Apr 02 15:57:52 Yes, I read it many times. I will google for "buildroot" and see what comes out. Apr 02 15:58:18 >> So let me tell you what I am doing (as usual, I'm experimenting). I'm trying to boot a standard Linux distribution<< WONT WORK with intird!!! NOT ENOUGH RAM! Apr 02 16:00:02 you're aware why it's called initRamDisk? because it takes away RAM from system to store files in it that are usually stored on e.g. BABD or eMMC or uSD Apr 02 16:00:21 NAND, not babd Apr 02 16:02:02 and btw no normal user would refer to installation of uBoot as "flashing", it's a normal package install from user perspective Apr 02 16:02:47 this is not android where you need to flash custom ROM Apr 02 16:04:33 This is how I came to yesterday's problem of how to transfer files from my computer to the N900. I put the file system on the sd card. But what for the kernel? Without u-boot I cannot upload it from eMMC or sd card. I need to upload it to RAM by the flasher and boot it, right? Apr 02 16:05:33 the eror lies in "without uboot" Apr 02 16:09:04 not even trying to explain how you will run into further special customization need when trying to make your initrd pivotroot onto the system on uSD then. Or musing about if or if not flasher+NOLO can load kernel from uSD file instead uploading i Apr 02 16:11:18 honestly what's so unbearably terrible about `apt-get install uboot-pr13` ? Apr 02 16:12:19 As I said, if I can't avoid it, I will do it. Apr 02 16:12:45 and yes, to get rid of it in case you messed up completely (hard to imagine how you'd do that) you simply flash stock kernel and everything back to normal, which is a nobrainer just like the installation Apr 02 16:14:48 Pali: when uboot missing its config files, would it break or simply boot straight into default aka stock kernel? Apr 02 16:15:31 uboot has hardcoded in its binary some defaults Apr 02 16:15:41 ta! :-) Apr 02 16:15:45 which are used when boomenu.src file is not available Apr 02 16:15:58 so yes, it boots default attached kernel Apr 02 16:16:05 Enrico_Menotti: see, it's basically impossible to break uboot Apr 02 16:16:22 ... Apr 02 16:16:57 correct package is u-boot-flasher: http://maemo.org/packages/view/u-boot-flasher Apr 02 16:19:13 The first time, I flashed everything ok, installed u-boot, messed up, and then I was unable to boot stock kernel. Nor normally flash the device. I also ended up with empty battery. Pali and NIN101 helped me in order to boot rescue OS from the sd card. This allowed charging the battery. Then I had to flash just the kernel, if I remember it right, and then I was able to flash normally. Apr 02 16:19:54 Enrico_Menotti: anyway, you can;t boot omap2plus_defconfig build kernel iirc Apr 02 16:20:09 check the one on Pali's tree on github Apr 02 16:20:14 ~elinux Apr 02 16:20:14 [elinux] http://eLinux.org/ or the Embedded Linux project. cross compilers. build systems, hardware hacking etc. Apr 02 16:20:47 Enrico_Menotti: https://github.com/pali/linux-n900 Apr 02 16:21:33 >>The first time, I flashed everything ok, installed u-boot,<< ok, so far >>messed up<< whatever you did, it been most likely because of lack of experience and info >>and then I was unable to boot stock kernel<< well, quite possible if you installed another kernel the wrong way >>Nor normally flash the device.<< PEBCAK Apr 02 16:22:58 freemangordon This is something in fact I would like to understand. Maybe I did the wrong thing in choosing mainline kernel. But there is a branch dedicated to the N900. And it's incorporated into the mainline. I tried to make that config file and then make menuconfig. Seems there are options for the N900. Apr 02 16:23:46 >>I also ended up with empty battery<< which is clearly PEBCAK since you been warned a dozen times at least during your study of docs and howtos that you need a fully charged battery before you mess around with the system in any way. And you shouldn't let system run in a bootloop for hours Apr 02 16:24:43 DocScrutinizer05: I honestly don't think this is helping Apr 02 16:25:39 It helps correct user attitude that dispises uBoot based on unrelated bad experience Apr 02 16:26:05 DocScrutinizer05 What's that acronym PEBCAK? Sorry, I'm a poor Italian. :) Apr 02 16:26:16 ~wtf pebcak Apr 02 16:26:18 Gee... I don't know what pebcak means... Apr 02 16:26:23 ~wtf peblac Apr 02 16:26:24 Gee... I don't know what peblac means... Apr 02 16:26:28 ~wtf pebkac Apr 02 16:26:30 PEBKAC: problem exists between keyboard and chair Apr 02 16:27:11 ... Apr 02 16:28:03 Enrico_Menotti: get rx51_defconfig from pali's tree on github (the one for linux 4.9) amd us that config with mainline kernel Apr 02 16:28:11 *and Apr 02 16:28:43 *and use Apr 02 16:28:50 mainline has no branch dedicated to n900 last time I've checked Apr 02 16:28:55 DocScrutinizer05: yep, thanks Apr 02 16:29:00 :-) Apr 02 16:30:07 freemangordon: thanks for chiming in. experts urgently needed here to make happy a user eager to learn and no hood docs around Apr 02 16:30:22 s/hood/good/ Apr 02 16:30:24 DocScrutinizer05 meant: freemangordon: thanks for chiming in. experts urgently needed here to make happy a user eager to learn and no good docs around Apr 02 16:30:47 Pali: did you see the issue I opened against libhildonfm? Apr 02 16:31:01 yes, but did not have time look at it deeply.. Apr 02 16:31:43 (now reading PDF spec and playing with generating PDF files... :-)) Apr 02 16:31:48 :) Apr 02 16:32:13 malware pdfs? ;-P Apr 02 16:32:16 and sha stuff? :* Apr 02 16:32:17 well, the fix should be trivial, I may take on it these days if I find some spare time Apr 02 16:32:43 DocScrutinizer05 Referring to the PEBKAC, I won't take this as an insult. I'm just trying to learn. But as bencoh said, this is not helping. Apr 02 16:32:43 freemangordon: you mean, removing const attributes? Apr 02 16:33:00 freemangordon About the defconfig file, I will try, thanks. Apr 02 16:33:38 bencoh: not only that, see https://github.com/community-ssu/libhildonfm/blob/master/hildon-fm/hildon-file-system-storage-dialog.c#L1268 Apr 02 16:33:56 Enrico_Menotti: wasn't an insult. I appreciate you want to learn. FOSS people and particularly me olf fart are sounding grumpy often when trying to convey facts as compressed as possible Apr 02 16:34:45 freemangordon: I suppose you're not referring to the fact that this is an awfully long name for a function :D Apr 02 16:34:46 bencoh: total = _("sfil_va_total_size_removable_storage"); and next 2 lines should be total =g_strdup( _("sfil_va_total_size_removable_storage")); IIUC Apr 02 16:35:21 hm, could be yeah Apr 02 16:35:24 Enrico_Menotti: you have a dislike of uBoot which lets you make not very wide decisions about how to proceed which will lead to more similar negative experience like you laready had in unrelated contect of uboot Apr 02 16:35:37 s/wide /wise/ Apr 02 16:35:37 DocScrutinizer05 meant: Enrico_Menotti: you have a dislike of uBoot which lets you make not very wisedecisions about how to proceed which will lead to more similar negative experience like you laready had in unrelated contect of uboot Apr 02 16:35:52 bencoh: because we have g_free(total); couple of lines down ;) Apr 02 16:35:53 freemangordon: haven't played much with glib/gtk recently Apr 02 16:35:55 yeah Apr 02 16:36:00 :) Apr 02 16:36:04 rogjt Apr 02 16:36:08 right*. Apr 02 16:36:11 _() macro resolves to dgettext() Apr 02 16:36:39 bencoh: from man dgettext " The resulting string is statically allocated and must not be modified or freed." Apr 02 16:36:46 huhu Apr 02 16:37:02 so the issue is way biger than just const modifier :) Apr 02 16:37:08 *bigger Apr 02 16:37:10 indeed Apr 02 16:37:17 hmm.. Apr 02 16:37:27 I wonder if we really want to g_free Apr 02 16:37:31 * freemangordon checks with gtranslate :) Apr 02 16:37:49 yeah, double g Apr 02 16:38:10 bencoh: yes, we want, as hildon_format_file_size_for_display allocates memory Apr 02 16:38:35 bencoh: http://maemo.org/api_refs/5.0/beta/hildon/hildon-Helper-Functions.html#hildon-format-file-size-for-display Apr 02 16:38:36 searching for "gtranslate" on google .... meh :) Apr 02 16:38:36 PEBKAC is just a short term for "user did probably something wrong, maybe because of lack of good instructions or documentation, maybe because of a typo or whatever other oopsie that everybody does from time to time" Apr 02 16:38:43 (silly me) Apr 02 16:39:12 ah, in that case ... Apr 02 16:39:37 at least this is how I read "This function is similar to g_format_file_size_for_display() " Apr 02 16:39:57 how did you stumble on that? Apr 02 16:40:17 while porting libhildonfm from gnomevfs to gio Apr 02 16:40:25 ouch Apr 02 16:40:33 gtk3 port? Apr 02 16:40:38 gtk2 Apr 02 16:40:42 ah Apr 02 16:41:01 but it should be the same/similar for gtk3 IMO Apr 02 16:42:29 DocScrutinizer05 Ok, no pb. In any case at this point I will read about buildroot, which is used for the building of rescue OS. Also I may try with the defconfig by Pali . If none of these ways will be successful, I will install u-boot and try with that. Apr 02 16:43:22 defconfig and uBoot/buildroot are orthogonal Apr 02 16:44:02 Enrico_Menotti: there is a bootable n900 devuan image http://46.249.74.23/devuan/image/ with kernel 4.something (4.8 iirc) Apr 02 16:44:32 Enrico_Menotti: uimage http://46.249.74.23/devuan/kernel-n900/ Apr 02 16:44:56 i have devuan on a n900 :) Apr 02 16:44:58 Pali and freemangordon are very experienced kernel devs who know their shit, when they tell you what works and what doesn't, i'd suggest to faithfully trust in their word :-) Apr 02 16:45:02 kinda nice but needs moar tweaks Apr 02 16:45:16 still havent implemented right-clicking for instance :P Apr 02 16:45:23 i just debootstrapped plain though Apr 02 16:45:24 buZz: did you use parazyd's builder? Apr 02 16:45:49 i used https://github.com/dderby/debian900 Apr 02 16:46:02 Enrico_Menotti: odds are a lot of upstream N900 kernel stuff you find has been upstreamed by Pali and freemangordon Apr 02 16:46:36 buZz: yeah, that one Apr 02 16:46:52 hmm, wait Apr 02 16:47:01 worked well, kinda amazed at how nice it is to run devuan on this device Apr 02 16:47:11 no, not that one Apr 02 16:47:21 well , thats the one i used :D Apr 02 16:47:47 yeah, but there is a newer one :) Apr 02 16:47:52 buZz: see https://github.com/dyne/arm-sdk Apr 02 16:48:32 freemangordon: when you got a kernel that is supposed to work according to freemangordon and you can't ramload it, nothing wrong with asking about it and get help :-) Apr 02 16:49:03 DocScrutinizer05: was that pointed to me? Apr 02 16:49:15 oops sorry nope, ETAB Apr 02 16:49:17 i might try it someday :) Apr 02 16:49:19 Enrico_Menotti: ^^^ Apr 02 16:49:22 starred Apr 02 16:49:38 buZz: the image ^^^ is build with it, albeit a couple of months ago Apr 02 16:51:18 * DocScrutinizer05 eventually will patch konversation to have two separate LRY stacks for tan nick expansion at start of line and inline Apr 02 16:51:47 LRU* and tab* Apr 02 16:52:14 DocScrutinizer05 freemangordon Ok, I am looking at that kernel. I need a moment. Apr 02 16:52:26 :-)) Apr 02 16:53:25 * freemangordon is starving, going to find something to bite Apr 02 17:02:06 damn I seriously need a second n900 to play with those things Apr 02 17:02:13 (mainline kernel, devuan, ...) Apr 02 17:02:20 (well I'd need time, too, but ...) Apr 02 17:05:06 bencoh: and a third Apr 02 17:06:55 freemangordon: ((devuan image)) armhf? Apr 02 17:07:05 yes Apr 02 17:07:24 freemangordon Ok, I got the kernel and system image. A first question: this is a uImage, so it is supposed to be loaded by u-boot, right? What if I want to try to RAMload it with the flasher? Apr 02 17:07:51 Pali: ^^^ Apr 02 17:08:24 Enrico_Menotti: though, you can ximply try Apr 02 17:08:28 *simply Apr 02 17:08:49 kernel not booting is the worst that can happen Apr 02 17:09:00 but yeah, it is supposed to be loaded by uboot Apr 02 17:09:18 No way to convert it to zImage? Apr 02 17:09:21 yeah, but what's supposed to happen if it does load and run Apr 02 17:10:09 Enrico_Menotti: none I am aware of Apr 02 17:10:26 lemme try to find zImage Apr 02 17:10:34 Thank you. Apr 02 17:10:58 http://unix.stackexchange.com/questions/122526/how-to-convert-a-zimage-into-uimage-for-booting-with-u-boot Apr 02 17:12:57 Enrico_Menotti: I found some newer, you may try with it. it is on http://46.249.74.23/devuan/kernel-n900/ Apr 02 17:13:33 https://linux.die.net/man/1/mkimage Apr 02 17:13:49 DocScrutinizer05: he wants the opposite uImage->zImage Apr 02 17:14:03 I know it's wrong way around, but explains the diffs Apr 02 17:16:12 hmm, seems pretty nontrivial Apr 02 17:17:01 mkimage has no inverse option to uimg->zimg Apr 02 17:18:00 freemangordon Ok, thank you, got it. Just in order to avoid doing strange things, I'd have a second question. The system image - I can't upload it to RAM (2GB). So should I mount it on a Linux system and then extract the contents to the SD card? Apr 02 17:20:28 cant you use dd ? Apr 02 17:20:41 also the image might be writeable? Apr 02 17:20:51 uimage -> zimage is a simple dd skip=64 bs=1 Apr 02 17:21:02 :-D \o/ Apr 02 17:21:57 Zungo Ok, let me try. Apr 02 17:22:00 and building zimage (vmlinuz) from a kernel tree is quite simple considering you've succeeded in building uimage :) Apr 02 17:22:25 easy to verify with mkimage applied to the result and check if it creates the original uimage Apr 02 17:22:50 DocScrutinizer05: you'd need to know mkimage parameters Apr 02 17:22:53 but yeah Apr 02 17:23:40 uboot (simple) header contains loadaddress and image name among other things Apr 02 17:32:41 Enrico_Menotti: on ubuntu you can write compressed image directly to SD card with "disks" UI. No idea about debian, but there should be similar functionality Apr 02 17:34:04 Enrico_Menotti: but simple "extract" won't do the job, there is more than just a files Apr 02 17:36:06 freemangordon Thanks. I used rsync for transferring the Debian file system to the sd card. I was thinking about the same way here. Clearly, it is easier to write directly the zImage. Now, I have (as usual :) ) a quite convoluted situation. Let me explain. Apr 02 17:37:07 My main computer is a Mac. I have a Debian live distribution running on another computer, which is unable to read sd cards (doesn't have a slot for that). I'd use directly the N900 for writing the disk image to the sd card. Apr 02 17:37:53 Enrico_Menotti: I am not sure rsync will do the job, but I might be wrong. just to be sure, use dd Apr 02 17:38:23 So now I'd resort to dd. But I think the zImage expects the sd card to be the right size, or a partition to be prepared in advance, am I right or wrong? Apr 02 17:39:41 doing dd of uncompressed image to sdcard is all you need, this is why it is disk image, but not a file archive Apr 02 17:39:54 *files archive Apr 02 17:40:40 you can always use CLI Apr 02 17:40:45 and theres GParted Apr 02 17:40:53 Zungo On N900? Apr 02 17:41:05 idk Apr 02 17:41:10 i do not have a n900 Apr 02 17:41:11 yet Apr 02 17:41:51 Enrico_Menotti: no, on debian Apr 02 17:42:06 Zungo: you'll be very happy when you finally get it :) Apr 02 17:42:15 install gparted and use it to restore the image onto the sdcard Apr 02 17:42:22 Don't have the GUI on Debian. Just command line (small computer). Apr 02 17:42:23 debian is like ubuntu :p Apr 02 17:42:28 ah Apr 02 17:42:29 you can always install X Apr 02 17:42:35 well, dd is your friend Apr 02 17:42:37 ubuntu is debian without training wheels Apr 02 17:42:41 with* Apr 02 17:42:43 dangit Apr 02 17:42:53 mhm Apr 02 17:43:15 but I like it, in most of its part Apr 02 17:43:50 yeah, i like devuan aswell Apr 02 17:43:57 actually gparted doesnt restore images but can attempt to restore lost partitions Apr 02 17:44:09 photorec or testdisk for that Apr 02 17:44:13 Enrico_Menotti: BUT - don;t forget to uncompress the image first (well, you can uncompress it on-the-fly) Apr 02 17:44:16 they're a bit old through Apr 02 17:45:53 freemangordon Ok, I'm still so handicapped with Linux & friends that I have to google for the right commands. But I'll manage to find the way. Just a question about dd: I found it is usually invoked with an option "bs=4096". Is this choice standard? Apr 02 17:46:15 I think yes Apr 02 17:47:05 but I am not exactly dd guru, so don't quote me on that one :) Apr 02 17:47:19 Ok, I'll just try. Apr 02 17:47:25 check the man pages for that Apr 02 17:47:47 but it should be something like dd if=$image of=/dev/$sdcard bs=4096 Apr 02 17:48:19 you'd have to figure out with what default commandline is dd compiled with Apr 02 17:48:24 what image is he restoring by the way? sorry i missed the first part of discussion Apr 02 17:48:24 Zungo: what is the shell tool to uncompress .xz images? Apr 02 17:48:33 Sicelo: devuan n900 Apr 02 17:48:35 because most of the time the defaults are fine with dd Apr 02 17:48:38 ok Apr 02 17:49:02 xz decompresses .xz :) Apr 02 17:49:09 lel Apr 02 17:49:22 thancs Apr 02 17:49:29 *thanks Apr 02 17:49:55 Sicelo: it doesn;t need parameters to decompress it and to put it on stdout? Apr 02 17:50:02 i'm happy Enrico_Menotti is installing a proper linux now (not Nitdroid) Apr 02 17:50:09 where' the link for devuan n900? Apr 02 17:50:32 well, a quick check says for stdout, you use -c Apr 02 17:51:07 Enrico_Menotti: some notes - devuan image is without X, login is usr=root pwd=toor, IIRC Apr 02 17:51:27 link please :) Apr 02 17:52:31 Enrico_Menotti: so, you may try with "xz -c $image.xz | dd -of=/dev/$sdcard bs=4096", without quotes Apr 02 17:52:38 Sicelo: link to what? Apr 02 17:52:49 devuan image Apr 02 17:52:56 see backscroll Apr 02 17:55:28 thanks got it now Apr 02 18:01:17 so there is a devuan+maemo image for testing? Apr 02 18:02:24 with maemo embedded into devuan? Apr 02 18:02:33 that sounds nice Apr 02 18:02:48 or just devuan images for n900..? Apr 02 18:02:51 18:44:02 < freemangordon> Enrico_Menotti: there is a bootable n900 devuan image http://46.249.74.23/devuan/image/ with kernel 4.something (4.8 iirc) Apr 02 18:02:54 18:44:32 < freemangordon> Enrico_Menotti: uimage http://46.249.74.23/devuan/kernel-n900/ Apr 02 18:03:02 ah Apr 02 18:03:04 i think these are an image anyway Apr 02 18:03:06 freemangordon: there was a newer i made with Wizzup Apr 02 18:03:09 sec Apr 02 18:03:30 https://pub.parazyd.cf/tmp/devuan_jessie_1.0.0-RC_armhf_n900.img.xz Apr 02 18:03:37 https://pub.parazyd.cf/tmp/devuan_jessie_1.0.0-RC_armhf_n900.img.xz.sha (sha256) Apr 02 18:03:45 4.10 mainlie Apr 02 18:03:55 parazyd: thanks Apr 02 18:03:59 *mainline. uncompresses to ~2GB Apr 02 18:04:01 Enrico_Menotti: ^^^ Apr 02 18:04:28 Zungo: no, embedding maemo with devuan is WIP Apr 02 18:05:09 Zungo: see https://github.com/fremantle-gtk2 Apr 02 18:06:23 right Apr 02 18:06:31 has those updated yet? Apr 02 18:07:34 buZz: Zungo: see videos at http://46.249.74.23/allwinner/ for some very very early results. This is gtk3 hildon-desktop running on top of devuan n900 image prepared with parazyd's armsdk. Apr 02 18:07:41 Zungo: updated? Apr 02 18:07:54 allwinner n900? YES PLEASE Apr 02 18:08:08 :) Apr 02 18:08:33 ^_^ Apr 02 18:08:34 no, it is not allwinner n900, but some cheapo alwinner tablet I bought for the purpose of POC Apr 02 18:08:45 hoping the ia32 variants are ready soon Apr 02 18:08:46 lel Apr 02 18:09:07 Zungo: well, actually I am using x86-64 VM for all the development ;) Apr 02 18:09:17 and testing Apr 02 18:09:43 Zungo: so, in theory, all you need to do is to git clone;dpng-buildpackge Apr 02 18:09:58 freemangordon: A13 ? :D Apr 02 18:10:06 *dpkg-buildpackge Apr 02 18:10:10 i have some 7" A13 tablet somewhere Apr 02 18:10:11 buZz: A33 Apr 02 18:10:18 freemangordon: its juest Apr 02 18:10:38 Zungo: sorry, can;t parse Apr 02 18:10:48 that i am juest Apr 02 18:10:50 -_- Apr 02 18:11:06 ah Apr 02 18:11:50 buZz: you need upstream kernel support AND working 3D accell Apr 02 18:12:29 A10/13/20 afaik all have upstream & 3d accell sorta working Apr 02 18:12:30 well, maybe upstream is not a must, but I don;t know how well devuan will play with some OEM kernel Apr 02 18:12:43 with blobbed drivers though Apr 02 18:12:51 buZz: yeah, blobbed Apr 02 18:13:07 but a10/a13/a20 are nowhere to be found Apr 02 18:13:37 which nowhere are you refering to? Apr 02 18:13:45 ebay Apr 02 18:13:51 oh no clue Apr 02 18:14:08 A13 = whats inside R8 Apr 02 18:14:13 so 'CHIP' is A13 Apr 02 18:14:19 no idea Apr 02 18:14:27 i'm telling you :) Apr 02 18:14:32 however, having it working on A33 is better Apr 02 18:14:42 yeah? Apr 02 18:15:04 yeah, those are some 50 euros - 4 cores/512MB Apr 02 18:16:11 i'm aware what the SOC is :P Apr 02 18:16:42 ok Apr 02 18:16:57 but I was not aware if you are :p Apr 02 18:17:24 thats ok <3 Apr 02 18:17:27 appreciated Apr 02 18:18:49 anyway, with help of Wizzup and parazyd, we may have maemo rebased on devuan repo on devuan servers. ofc not full maemo, but at least hildon-desktop, hildon-status-menu and hildon-home Apr 02 18:19:14 in 2-3 weeks from now, if they manage to find the needed time to do it Apr 02 18:19:29 quite nice Apr 02 18:19:55 mhm. I hope to attract more devs once there is something to be shown besides a POC Apr 02 18:20:06 \o/ Apr 02 18:20:34 yes i hope to do amprolla this month Apr 02 18:20:42 yeah, there are piles of porting work waiting to be done, not a one-man-show task Apr 02 18:20:43 also devuan RC will be released Apr 02 18:21:23 and then hopefully the other devuan people enable the CI Apr 02 18:21:45 cool Apr 02 18:22:23 so for other ppl interested: this way we don't "taint" devuan repos, but have our own downstream Apr 02 18:22:50 which takes precedence over devuan then so one can even override a devuan package Apr 02 18:23:04 (automagically) Apr 02 18:23:19 I wonder if we can make some more noise. After all the final goal is to have a real mobile/tablet linux, so I somehow expect a bit more enthusiasm from FOSS people. Apr 02 18:23:49 maybe on some conferences Apr 02 18:24:27 mhm. But still, there should be something at least close to functioning to be shown there Apr 02 18:25:15 however, I am one function away from finishing libhildonfm porting to gio, and IIRC that was the last missing hildon-home dependency Apr 02 18:25:30 <3 Apr 02 18:25:34 devuan RC? Apr 02 18:25:56 jessie release candidate Apr 02 18:26:09 yeah, it's slow... but steady i guess Apr 02 18:26:16 once having those 3 (h-d, h-h and h-s-m) I guess it will get easier Apr 02 18:26:31 though there remains stuff like systemui etc Apr 02 18:26:41 i don't like that it's moving _so_ slow though. debian already froze jessie and is getting ready for stretch Apr 02 18:26:48 but this is not the place to bitch about it Apr 02 18:26:48 not THAT hard, but time consuming Apr 02 18:27:59 IT would have been great if there was at least one more dev to work on sporting stuff that we have source code of, so me to be able to focus on REing work, but... Apr 02 18:28:07 *porting Apr 02 18:28:56 fmg: a10/a20 is present in those MIDxxx cheapo tablets from few years ago, that are laying around in people's houses Apr 02 18:28:59 we'll see Apr 02 18:29:23 and all of them would be very easy to make supported Apr 02 18:30:58 i think so too Apr 02 18:31:28 i can get to trying on this A13 tablet i own once something gets testable? Apr 02 18:31:44 yup Apr 02 18:31:53 you can do it now Apr 02 18:32:10 KotCzarny: I have not been able to find any online source to buy anything older than a33 Apr 02 18:32:35 fmg: second hand goods? unless you are interested in only something that can be bought new? Apr 02 18:32:37 buZz: it is testable ATM Apr 02 18:33:09 KotCzarny: yep, I wanted for it to be new, I don;t really need HW failures to mess with OS development Apr 02 18:34:57 buZz: you only need to install debian/devuan, X, 3d drivers gcc and git Apr 02 18:35:05 i have a quite successful theory, that if you buy something used that is working, it wont fail for a long time. new things, straight from the factory, otoh, can have a high chance of failure Apr 02 18:35:19 could be Apr 02 18:35:49 but still, it was way easier to buy it from BG merchant, with warranty and 3 days for delivery :) Apr 02 18:36:07 and if it fails, 2-3 weeks of being without device? ;) Apr 02 18:36:20 and with return policy covered by EU rules Apr 02 18:36:35 for that price, I can buy 2 per week :) Apr 02 18:36:41 ;) Apr 02 18:37:05 yeah, that's winning argument Apr 02 18:37:28 BUT, if you manage to find a10/a13 for 5usd .. ;) Apr 02 18:37:41 the key being "if you manage" Apr 02 18:37:46 I failed Apr 02 18:38:08 there were only 2 or 3 on some Russian sites, I'd rather not mess with Apr 02 18:38:26 no EU to cover my ass there Apr 02 18:38:32 most people who own those old devices are usually not knowing what they have Apr 02 18:39:21 Wizzup searched for used a10/a13 for a week or so, he found almost nothing Apr 02 18:39:58 and, I think it is better to try the newest as it is probable to be supported/available for some time Apr 02 18:40:11 not that a33 is the newest, but still Apr 02 18:40:16 erm, that's not true in arm land Apr 02 18:40:34 newest arent supported for 3-12 months Apr 02 18:40:39 KotCzarny: what is the point to port maemo to HW you can;t buy? Apr 02 18:40:50 and vendor bsp is usually crap Apr 02 18:41:30 fmg, afair, requirement is 'mainline kernel with not locked bootloader and 3g accel' Apr 02 18:41:49 *3d Apr 02 18:42:30 well, yeah Apr 02 18:43:45 Sorry, I have a small issue. I have been able to use dd to write the disk image to sd card, but when I try to mount the sd card itself I get "invalid argument". Apr 02 18:44:24 dd image? Apr 02 18:44:30 youcan;t mount it, as there are several partitions Apr 02 18:44:39 what dd image Apr 02 18:44:44 devuan n900 Apr 02 18:44:49 ahm Apr 02 18:45:00 be sure to uncompress it first Apr 02 18:45:16 yeah, warned him several times already :) Apr 02 18:45:34 just making sure... why can't it be mounted then? Apr 02 18:45:48 because you need -o param Apr 02 18:45:50 i get mmcblk0p1 and p2 on my laptop after dd-ing Apr 02 18:45:53 if you dont specify partition Apr 02 18:45:59 parazyd, you need hdparm -z too Apr 02 18:46:07 KotCzarny: hmm? Apr 02 18:46:08 or other way to reread partition table Apr 02 18:46:18 ah, yeah Apr 02 18:46:30 partprobe, yes Apr 02 18:46:43 Enrico_Menotti: just reconnect sdcard Apr 02 18:47:01 I decompressed the .xz file on the Mac. Got a .img file. Transferred that to the N900. dd the image to the sd card. Under /dev I see mmcblk1 and mmcblk1p1. Apr 02 18:47:03 that's the easiest way to force debian to reread the partition table Apr 02 18:47:22 reinsert card Apr 02 18:47:28 and check for new partitions Apr 02 18:47:33 Enrico_Menotti: emove the back cover and then put it back Apr 02 18:47:34 No other way than opening the back cover and extracting/reinserting the card? Apr 02 18:47:42 just remoive tha back cover Apr 02 18:47:59 or use hdparm -z /dev/mmcblk0 (or 1) Apr 02 18:48:01 no need to remove/instert the card Apr 02 18:48:22 damn typos :( Apr 02 18:48:22 KotCzarny Ok, will try this way. Apr 02 18:49:09 No, hdparm not found on the N900. I will manage with the cover. Apr 02 18:54:19 Ok, I have two partitions. The first may be mounted, but there's just a uImage inside, and the second cannot be mounted: "invalid argument". Apr 02 18:54:57 maybe its ext4 Apr 02 18:55:06 ie. unsupported by n900 Apr 02 18:55:17 s/n900/maemo/ Apr 02 18:55:17 KotCzarny meant: ie. unsupported by maemo Apr 02 18:55:34 Ah ok. Apr 02 18:55:46 you can always check on pc Apr 02 18:56:41 No, I have problems with this: the pc where I run Debian does not have a sd card reader, and the other is a Mac, which, as far as I know, is unable to read ext fs. Apr 02 18:56:49 KotCzarny: it is Apr 02 18:57:09 p1 is fat32, p2 is ext4 (purposely) Apr 02 18:57:41 parazyd So what parameter do I have to pass to the kernel? rootfs=/dev/mmcblk1p2 ? Apr 02 18:57:55 you can't mount it under maemo Apr 02 18:58:10 he asks how to boot it Apr 02 18:58:11 on n900 Apr 02 18:58:16 Yes, I will boot the Devian kernel. Apr 02 18:58:19 my reasoning for choosing ext4 is because of this exactly. so maemo doesn't touch it Apr 02 18:58:33 in uboot console type 'run sdboot' Apr 02 18:58:44 No uboot. Upload by flasher. Apr 02 18:58:56 I need to know the parameter to be passed to the kernel. Apr 02 18:59:09 i wouldn't know Apr 02 18:59:17 (you might want to start giving u-boot some love - or reinforce your USB) Apr 02 18:59:17 try root=/dev/mmcblk1p2 i guess Apr 02 18:59:47 but indeed, get uboot ;) Apr 02 18:59:49 parazyd Ok if the rootfs is in the second partition, this should work. Apr 02 18:59:57 yes Apr 02 19:00:04 After some testing I will get u boot, of course. Apr 02 19:00:06 the fstab has an entry for /boot Apr 02 19:00:10 but it's an uimage Apr 02 19:00:35 i dunno what else besides uboot can use it Apr 02 19:01:13 Zungo: gearcm flashed :) Apr 02 19:01:23 congrats Apr 02 19:01:46 sicelo: you wont need odin again, afaik Apr 02 19:01:57 yes, i only used it once Apr 02 19:02:55 parazyd: re maemo won;t touch ext4 - iirc kernel-power has ext4 support Apr 02 19:03:24 oh ok Apr 02 19:03:32 though ext4 is still a sane choice, no? Apr 02 19:03:42 Enrico_Menotti: keep in mind I am not sure zImage i provided is the correct one Apr 02 19:03:50 btrfs maybe Apr 02 19:03:51 uImage is though Apr 02 19:03:58 freemangordon: a friend of mine said he had an old a10 tablet, but they are very hard to source Apr 02 19:04:06 KotCzarny: ^^^ Apr 02 19:04:14 :) Apr 02 19:04:15 20:19 < freemangordon> mhm. I hope to attract more devs once there is something to be shown besides a POC Apr 02 19:04:18 this :) Apr 02 19:04:36 who wants my a10 tablet? (touch is dead) Apr 02 19:04:43 freemangordon: let's make noise once we have something that people can dd & apt-get upgrade, so to speak Apr 02 19:04:52 something that has at least h-d + xterm or so Apr 02 19:04:56 yeah Apr 02 19:05:10 BTW, I have xterm running in my VM Apr 02 19:05:21 allwinner has particular advantage of having sdcard at higher boot priority Apr 02 19:05:32 i saw firefox in your tablet vid :) Apr 02 19:05:33 I use fel Apr 02 19:05:37 parazyd: yeah Apr 02 19:06:48 parazyd: and debian file manager (with some strange name) and wicd Apr 02 19:07:49 there is networkmanager btw, in jessie-proposed iirc Apr 02 19:08:04 deb http://auto.mirror.devuan.org/devuan jessie-proposed main Apr 02 19:08:21 i haven't tried it yet Apr 02 19:10:19 parazyd: isn;t it too heavy for embedded? Apr 02 19:10:47 i hate networkmanager (when it doesnt work) Apr 02 19:10:50 Ok, signs of life. The kernel boots, it seems, but gets panic. Unable to mount the rootfs. May it be just because it is ext4? Apr 02 19:11:16 Enrico_Menotti: no, this means you've passed wrong root= parameter Apr 02 19:11:24 i don't know. i know it's nifty for users though. not sure how maemo does networking, but NM has vpn stuff, wifi and eth Apr 02 19:11:50 and it usually broke less often than wicd in my experiences Apr 02 19:12:21 could be, but at least on ubuntu, the applet is FUBAR Apr 02 19:12:30 Enrico_Menotti: you can also pass 'rootfstype=ext4' Apr 02 19:12:37 though it is canonical to be blamed here, no NM Apr 02 19:12:43 parazyd: no need Apr 02 19:12:46 iirc Apr 02 19:13:03 * freemangordon checks his n900 u-boot config Apr 02 19:13:17 freemangordon: i'm in love with netifrc (from openrc) Apr 02 19:13:20 serves me pretty well Apr 02 19:13:38 freemangordon root=/dev/mmcblk1p2 Apr 02 19:13:49 Should be right. It's the ext4 partition. Apr 02 19:14:12 Enrico_Menotti: try with root=/dev/mmcblk0p2 rootwait Apr 02 19:14:31 I have also a rootdelay. Apr 02 19:15:04 Enrico_Menotti: sd card numbering differs in maemo compared to upstream kernel Apr 02 19:15:20 Ah that's the point so. Apr 02 19:15:30 mmcblk1 in maemo is mmcblk0 in upstream, iirc Apr 02 19:15:37 ah so it IS the n900 doing that Apr 02 19:15:45 i was wondering what board did it :D Apr 02 19:16:08 parazyd: maemo does it Apr 02 19:16:20 it is not the board or the kernel Apr 02 19:16:31 freemangordon There is also the mmcblk0 under maemo, which is the eMMC. Are they switched in Devian? Apr 02 19:16:33 yes, that's what i meant Apr 02 19:16:41 wrong choice of words Apr 02 19:16:44 Enrico_Menotti: yes Apr 02 19:16:46 iirc Apr 02 19:16:56 Enrico_Menotti: *devuan, yes Apr 02 19:16:58 yes they are Apr 02 19:17:32 its 'devUan' Apr 02 19:17:39 not devian(t) Apr 02 19:19:54 Enrico_Menotti: if that root= does not work, try with no command line at all Apr 02 19:20:08 iirc the kernel should have the correct cmdline Apr 02 19:20:32 Sorry phone call. Apr 02 19:20:50 Enrico_Menotti: this is from the config used to build the kernel: Apr 02 19:20:50 Devuan, right, sorry. Apr 02 19:20:50 CONFIG_CMDLINE="root=/dev/mmcblk0p2 rw mtdoops.mtddev=log console=tty0 console=ttyO2 omapfb_vram=7M omapfb.mode=lcd:848x480-16 nokia-modem.pm=0" Apr 02 19:21:13 so, do not pass kernel command line at all Apr 02 19:22:06 Ok, let's try. Apr 02 19:23:09 freemangordon: btw, if you have a sane kernel config for the n900, let me know. the current one lives here: https://github.com/dyne/arm-sdk/raw/next/boards/kernel-configs/n900.config Apr 02 19:23:15 but it can be improved imo Apr 02 19:25:48 21:10 < freemangordon> parazyd: isn;t it too heavy for embedded? Apr 02 19:25:52 I think it might not be Apr 02 19:26:09 I dpn't care about the politics, but I think networkmanager may be easy to adept for the n900, but maybe not. Apr 02 19:26:20 I figured we could take the nm-applet and modify it to fit our needs Apr 02 19:26:23 but wicd may be fine too... Apr 02 19:26:35 s/adept/change/ Apr 02 19:27:00 isbt nm relying on some weird libs? Apr 02 19:27:32 well, let's get h-d on there first and then figure it out Apr 02 19:30:08 parazyd: well, https://github.com/dyne/arm-sdk/commit/7d9cc10223e07bf21b0370ae3fe3a043802b9b55 Apr 02 19:30:21 The sane n900 config I have is there Apr 02 19:30:54 it it almost the same as the one from Pali's repo Apr 02 19:31:10 ah ok Apr 02 19:31:12 freemangordon I tried a few times, reading the output from the kernel. Right, the external sd card is mmcblk0, two partitions. If I don't specify a root=, it says "unable to mount unspecified rootfs" or something of this kind. If I specify the root=/dev/mmcblk0p2, it's still complaining he can't mount it. I will try with also rootfstype=ext4. Apr 02 19:31:23 yeah in that one the watchdogs were modules Apr 02 19:31:30 now they're in the kernel Apr 02 19:31:40 other than that i don't remember changing anything Apr 02 19:32:57 but ok, eventually we'll figure out if there's things that need to be added Apr 02 19:41:34 Ok, something more. I tried with "rootwait root=/dev/mmcblk0p2 rootfstype=ext4". This now ends with "Mount failed for sys/fs/selinux: no such file or directory". Apr 02 19:41:56 wth are you booting? Apr 02 19:42:24 O.o Apr 02 19:43:16 KotCzarny After some discussion about how to try booting a standard Linux distribution, I am trying to boot Devuan. Apr 02 19:46:06 Well, at least I know there is something wrong with the mainline kernel I built, since it does not give signs of life. This other one does... :) Apr 02 19:46:28 why dont you use devuan kernel on your sdcard? Apr 02 19:47:07 Enrico_Menotti: If you want linux on your n900: 1) flash maemo, install cssu, install u-boot from maemo 2) dd the devuan image from parazyd to the sd card Apr 02 19:47:18 3) profit, but you'll need to install xorg and others, and configure the touch screen Apr 02 19:47:29 I have some instructions on how to do that with xinput Apr 02 19:47:36 freemangordon: where should we document things, maemo wiki? Apr 02 19:48:13 I am trying to upload and boot by using the flasher. To this aim, I need to upload the kernel to RAM from the computer and boot it. The file system image on the other hand is on the sd card. Apr 02 19:48:54 Wizzup Yes, there has been a long discussion about how to test a new kernel. I'd try to upload it via flasher, instead of using u-boot. Apr 02 19:49:17 Enrico_Menotti: why would you do that? Apr 02 19:49:55 Wizzup: i guess yes, maemo wiki Apr 02 19:50:13 Wizzup It's just a preliminary test. I would like to test without flashing anything. Apr 02 19:50:26 Enrico_Menotti: does it complain about missing modules? Apr 02 19:50:55 as it is possible that zImage I provided does not match the contents of /lib/modules Apr 02 19:50:58 Don't know. The messages stay on the screen just for a while, without backlight. I am unable to read all. Apr 02 19:51:20 Enrico_Menotti: try to make zImage from Umiage Apr 02 19:51:25 *uImage Apr 02 19:53:08 Enrico_Menotti: or, just follow the instruction on https://github.com/dyne/arm-sdk to create your own zImage and kernel Apr 02 19:53:16 and devuan image Apr 02 19:53:38 Ok, a minute. I think I am missing something. I downloaded zImage and the Devian .img.xz file. Unpacked the latter and used dd to write it to sd card. I upload the zImage by flasher and booted it. Apr 02 19:53:45 sidenote; it defaults to a uImage Apr 02 19:54:01 freemangordon: should i include both zImage and uImage on the /boot partition perhaps? Apr 02 19:54:11 parazyd: but, there is zImage in the kernel build directory Apr 02 19:54:23 yes there is Apr 02 19:54:50 Enrico_Menotti: what you did is correct Apr 02 19:55:01 There is also the lib.tar.gz. What about it? Apr 02 19:55:17 that one corresponds to the zImage Apr 02 19:55:22 you need it Apr 02 19:55:27 lemme check Apr 02 19:55:34 Yes, but how to use it? Apr 02 19:55:43 extract it in the rootfs Apr 02 19:55:56 those modules match the uImage Apr 02 19:56:18 parazyd I cannot access the rootfs since it is ext4. Apr 02 19:56:18 oh i might have something... Apr 02 19:56:23 and I think those modules should already be in the devuan image Apr 02 19:56:57 freemangordon: if he downloaded my latest image, that's kernel 4.10 Apr 02 19:57:07 Enrico_Menotti: you can install u-boot and then test whatever you want from u-boot ;) Apr 02 19:57:17 freemangordon: btw, did you ever try to record anything with arecord on mainline? Apr 02 19:57:34 Enrico_Menotti: which image do you use? mine or parazyd's? Apr 02 19:57:43 Parazyd's. Apr 02 19:58:00 then you need an according zImage Apr 02 19:58:10 well then, my zImage is linux 4.8, parazyd's devuan image is for linux 4.10 Apr 02 19:58:13 Ah ok. Apr 02 19:58:16 you can get it from the uImage that's on the fat32 partition (mmcblkXp1) Apr 02 19:58:35 Ok, and convert it to zImage, right? Apr 02 19:58:37 parazyd: care to give the command to convert uImage to zImage? Apr 02 19:59:00 if only... Apr 02 19:59:14 Enrico_Menotti: right Apr 02 19:59:24 google shoudl know, i don't Apr 02 20:00:12 "bencoh: uimage -> zimage is a simple dd skip=64 bs=1" Apr 02 20:00:41 so it should be dd if=uImage of=zImage skip=64 bs=1 Apr 02 20:00:43 freemangordon Ok, thank you, I was unable to find it again. Apr 02 20:00:46 no idea if it works Apr 02 20:01:10 i can pop a new build if needed, and give you a zImage Apr 02 20:01:38 let him first try with dd, it seems easy Apr 02 20:01:55 parazyd No need for now, thanks. Apr 02 20:02:02 ok Apr 02 20:02:27 Enrico_Menotti: you may also try with my devuan image if you fail with parazyd's one Apr 02 20:02:37 * freemangordon is afk Apr 02 20:02:42 Ok. I will try. Apr 02 20:10:37 Enrico_Menotti: any luck? Apr 02 20:11:02 A minute... Apr 02 20:11:15 i started a 4.10 build Apr 02 20:11:23 so i can give you the correct zImage in a few minutes Apr 02 20:12:13 No, the same mount failed for that file. Apr 02 20:14:03 "Mount failed for sys/fs/selinux: no such file or directory" Apr 02 20:14:40 No, wait, I've got a login!!! Apr 02 20:14:49 :) Apr 02 20:15:29 parazyd Could you please tell me again user and pass? Apr 02 20:15:33 root:toor Apr 02 20:16:13 Yes, it's working. But no backlight. Apr 02 20:16:39 that's strange, should be on full Apr 02 20:16:54 maybe it's some magic that u-boot does Apr 02 20:18:01 Anyway, now I'm sure my standard Linux kernel, the one I built, has something wrong. Apr 02 20:18:16 https://pub.parazyd.cf/tmp/zImage Apr 02 20:18:33 here's a zImage just built Apr 02 20:18:45 but i think it's the same thing as what you have now Apr 02 20:19:21 Probably. Thank you anyway. Apr 02 20:19:28 I will try, at any rate. Apr 02 20:19:32 https://github.com/dyne/arm-sdk/blob/next/boards/nokia-n900.sh#L78 Apr 02 20:19:40 here's how the kernel is built Apr 02 20:31:49 Guys, I have a problem. Apr 02 20:32:20 Seems my N900 doesn't go to update mode anymore. Maemo boots, however. Apr 02 20:32:39 broken update mode? Apr 02 20:32:48 what's updated mode? Apr 02 20:33:14 I mean flash mode... the mode you go in if you press "u" and connect the usb. Apr 02 20:34:14 Also, if I just connect the usb without doing anything, it does not charge. It does if I press the power button and wait a while. Apr 02 20:35:08 tried taking the battery out? Apr 02 20:35:19 No. I will try. Apr 02 20:35:29 that usually helps Apr 02 20:36:20 In this case not, seems. Apr 02 20:37:52 Ok, I will say what I did. After booting Devuan, I entered the shell, and issued "poweroff". Maybe wrong thing? Should I have pressed the power button to turn it off? Apr 02 20:42:37 Ok, I have been able to go again in flash mode. Pressing and holding "u" AND pressing and holding the power button. God knows what happened. I fear I will have to flash it again. :( Apr 02 21:26:49 (( but it should be something like dd if=$image of=/dev/$sdcard bs=4096)) in which case you could ommit the blocksize Apr 02 21:28:04 blocksize has no effect on the reslt in a dd command without skip or similar, it just tells buffer size of process Apr 02 21:28:32 if you use count= or skip= then bs= is relevant Apr 02 21:29:07 xz -c $image.xz | dd -of=/dev/$sdcard bs=1M Apr 02 21:29:09 faster Apr 02 21:37:06 (( try root=/dev/mmcblk1p2 i guess)) no there's a gotcha, kernel thinks uSD is mmcblk0 Apr 02 21:37:22 yes, we've resolved that Apr 02 21:40:32 one thing to knmow: you can't power off N900 when charger or PC is connected to USB Apr 02 21:40:57 as soon as there are 5V VBUS the device boots up Apr 02 21:41:39 it however enters a runlevel dubbed ACT_DEAD that mimics a shut down device with only LED flashing yellow Apr 02 21:44:21 actually the amber flashing LED is unrelated to charging on a hw level, maemo middleware does that. And maemo BME process tickles the watchdog timer in charger chip which keeps the chip from stopping charging. That's why you can't charge battery when your system is fubar, and you can't fix your fubar system (flashing) when battery empty Apr 02 21:44:41 ~bootloop Apr 02 21:44:41 hmm... bootloop is when your device has broken rootfilesystem, so during reboot it fails on some service startup or kernel module load and thus reboots. This *drains* battery! And you can't reflash to stop bootloop when battery is drained. Recharge your battery by other means before reflashing. E.g. using ~rescueOS. Or external charger or BL-5J compatible other device. Apr 02 21:44:55 ~flatbatrecover Apr 02 21:44:55 Remove battery for 1 minute. Insert battery. Plug powered ***NOKIA WALLCHARGER*** to device. Watch steady amber. Let sit and charge. Do NOT try to boot. After 30 min, you got either a) a booted up N900, b) flashing amber which means you can boot, c) steady amber shut off -> start over again with ~flatbatrecover while already searching for a new battery. CAVEAT! Only works when ~rootfs OK (no ~bootloop) Apr 02 21:45:54 WATCH OUT when using alternative OS, it usually does _not_ know how to tickle chargerchip watchdog Apr 02 21:46:55 http://maemo.cloud-7.de/maemo5/patches_n_tools/charge21_500mA-USBlimit__UNTESTED.sh Apr 02 21:47:23 ~charge21 Apr 02 21:47:28 hmm Apr 02 21:47:44 ~listvalues charge21 Apr 02 21:47:46 Factoid search of 'charge21' by value returned no results. Apr 02 21:48:10 ~charge21 is http://maemo.cloud-7.de/maemo5/patches_n_tools/charge21_500mA-USBlimit__UNTESTED.sh Apr 02 21:48:10 okay, DocScrutinizer05 Apr 02 21:49:09 * DocScrutinizer05 idly wonders what's the 21 in ShadowJK's charge21 Apr 02 21:49:39 sth like charge69 and charge18 ? ;-) Apr 02 21:51:05 oh charge24 another very common ... I don't even have a word for this marketing name strategy. *24 for car/friend/whatever URLs, 18 and 69 for XXX Apr 02 21:52:37 maybe 18 is specific german since you're adult at 18 here, might differ with other countries and their sexual intercourse legislative. 69 is obvious ;-) Apr 02 21:56:19 https://www.youtube.com/watch?v=70rBunO4uuk https://www.youtube.com/watch?v=3RHa4xdEAiE Apr 02 22:05:27 https://www.youtube.com/watch?v=z4pPboh9hDQ rotfl. I guess you need to have been annoyed by the original TV commercial a few times to enjoy this Apr 02 22:08:49 Ok, I went to charge mode by connecting the usb AND holding the power button for a while. Now battery's up. Before I flash the device again, does anybody have any idea why now it doesn't go to charge mode as soon as I connect the usb, but I need to press the power button? (And the same for flash mode.) Apr 02 22:11:12 charge mode with N900-maemo is always automatically enabled when you connect a charger or USB host Apr 02 22:12:24 I just explained you can't power down the device when charger is connected since device automatically boots as soon as 5V applied. No matter what you do with power button, it won't stay off Apr 02 22:12:35 Ok, but the LED doesn't flash. And the device doesn't give signs of life. And if I press and hold the "u" and connect the usb, nothing happens. Apr 02 22:13:23 then you managed to put the device into a limbo state where it's on but not respinsive Apr 02 22:13:59 What would you do? Flashing? Apr 02 22:14:02 I'd urgently suggest to remove battery to recover and not drain battery by keeping it in that state Apr 02 22:14:47 a 10s power button press *might* work too Apr 02 22:15:04 Well, I already charged the battery. If I press the power button (while connected to usb) it starts charging. Apr 02 22:15:15 no Apr 02 22:15:19 And if I hold "u" and press power it goes in flashing mode. Apr 02 22:15:45 No? Apr 02 22:15:58 as I already explained 2 times, it always starts charging as soon as you connect charger Apr 02 22:16:16 Even if the LED is not flashing? Apr 02 22:16:49 if the LED is not flashing then something is wrong with your system Apr 02 22:17:07 LED is not directly related to charging Apr 02 22:18:14 In any case, now I got the LED flashing by pressing the power button, and in the end it got green. I think I may assume that the battery is full. Apr 02 22:18:14 there's even an app to adjust (or disable completely) the LED flashing pattern for device states incl 'charging' state. It's an OS feature Apr 02 22:18:36 yes, at least 60% Apr 02 22:18:40 80% Apr 02 22:19:01 Ok, so now? Should I flash again, to restore normal behaviour? Apr 02 22:19:06 green is not physical end of charge but some logical thing defined by OS Apr 02 22:19:30 what's the problem with non-normal behavior? Apr 02 22:19:55 please describe what you think is non.normal Apr 02 22:20:25 The fact that the LED is not flashing as soon as I connect the USB and the fact that in order to go to flash mode I have to hold "u" AND press the power button. Apr 02 22:20:40 This happened after I've been able to boot Devuan. Apr 02 22:21:15 sorry, the first is normal and the second a handling problem or a hw defect Apr 02 22:21:53 please refer to Apr 02 22:21:56 ~flashing Apr 02 22:21:56 [maemo-flashing] http://wiki.maemo.org/Updating_the_tablet_firmware, or - on linux PC - download&extract http://maemo.cloud-7.de/maemo5/patches_n_tools/maemo-my-private-workdir.tgz, cd into it, do sudo ./flash-it-all.sh; or see ~flashing-cmdline, or see ~lazyflashing Apr 02 22:22:21 For the first, it does not flash until I press the power button. In any case, before booting Devuan these things did not happen. Apr 02 22:22:22 http://wiki.maemo.org/Updating_the_tablet_firmware#the_complete_foolproof_procedure Apr 02 22:23:08 there's no way to stop device from booting on a hw level when you attach 5V USB Apr 02 22:23:30 Ok, I know how to flash, I did it two days ago. But I wonder whether this is the case, or I could recover the thing without flashing. Apr 02 22:23:43 so no such thing like "I need to press power button to make it charge" Apr 02 22:24:04 Ok, let's say that I need to press power button to see any sign of life. Apr 02 22:24:43 you pressing power button while device already linked to charger will SHUT DOWN the device and it reboots instantly since it _cannot_ be powered down with charger attached Apr 02 22:25:40 so my suspicion is you boted into a kernel that leaves the device in a state where it doesn't show signs of life and you interpreted that as "powered down" Apr 02 22:26:23 attaching charger doesn't change that, so you have to press power button to reboot into sane working OS Apr 02 22:27:40 Ok, I understand. The strange thing is that I just uploaded the kernel for Devuan, not flashed it. It should not be in RAM anymore, right? Apr 02 22:27:49 also allow for 30s for power up *and* shut down. The device is still cleaning up for 20s after LED and display went dark, on power off Apr 02 22:29:06 RAM isn't magically cleared on random triggers. When experimenting with kernels you should get used to removing abttery and unplugging from USB to reach a defined state Apr 02 22:29:22 battery, even Apr 02 22:29:58 How much time removing battery? I already tried... Apr 02 22:30:57 actually I'm not even sure if the system reboots at all when pressing power button as long as charger attached, or simply does a "init 5" or "init 1" etc to change form "active" state to "ACT_DEAD" state and vice versa Apr 02 22:31:09 10s should suffice Apr 02 22:31:31 actually I guess 1s suffices Apr 02 22:32:04 please study http://wiki.maemo.org/Updating_the_tablet_firmware#the_complete_foolproof_procedure Apr 02 22:32:42 Ok, I disconnected the battery for more than half a minute. This finally worked. Thank you. No need to reflash. Apr 02 22:32:47 removing battery is the only failsafe way to get device into a known state after flashing Apr 02 22:32:55 (incl EAMload) Apr 02 22:33:02 RAMload* Apr 02 22:33:35 Yes, I just did not expect to have to do that for such a long time. Apr 02 22:34:17 I guess it's not needed for a long time, the system should crash hard from blackout within milliseconds Apr 02 22:34:40 unless you keep USB attached ;-P Apr 02 22:35:11 charger will power the system until watchdog kicks in, which is ~15s Apr 02 22:35:53 No, I disconnected and reconnected the battery, without usb, many times, without success, until this last time (when I wated ~30 s). Apr 02 22:36:08 and actually seen many devices come up with amber steady LED after 20min on charger without battery inserted Apr 02 22:36:50 wow, then N900 is a better hardware than I ever expected :-) Apr 02 22:36:59 :) Apr 02 22:38:13 In any case, now everything's ok. Next step: trying to understand why the new kernel I built is not booting at all. Apr 02 22:38:18 But this is for tomorrow. Now, bed time. Apr 02 22:38:20 when CPU enters zeroclock mode and RAM does suspend powersafe autorefresh, the power consumption may be very low indeed Apr 02 22:39:38 the lowest standby current we seen with fully booted system is ~2mA iirc Apr 02 22:39:56 when doing suspend2ram Apr 02 22:40:44 where the latter more helps with keeping CPU un unclocked mode than with cutting down on CPUs power consumption Apr 02 22:41:01 s/ un / in / Apr 02 22:41:01 DocScrutinizer05 meant: where the latter more helps with keeping CPU in unclocked mode than with cutting down on CPUs power consumption Apr 02 22:41:51 another kernel might be more relaxed and not start clocks at all until e.g. you connect charger Apr 02 22:42:53 a trick that usually helps: press power button while _battery_removed_ --> instant blackout from CPU ramping up and draining buffer capacitors Apr 02 22:46:53 no idea if your kernel you RAMloaded might have enetered suspend2ram state, or whatever. So yes it's possible a loaded kernel survives even 10 or more seconds of battery removal Apr 02 22:56:22 >>why now it doesn't go to charge mode as soon as I connect the usb, but I need to press the power button? (And the same for flash mode.)<< for flash mode all explained above applies accordingly: flash mode only entered during bootloader pjase. When device is still running a kernel it won't even enter bootloader. You pressing power button for 8?s however powered down the device and in the next millisecond it gets booted up again, which Apr 02 22:56:23 means it enters bootloader and the bootloader detects "u" held down etc Apr 03 02:10:10 DocScrutinizer05, 21 = version 2.1 **** ENDING LOGGING AT Mon Apr 03 03:00:01 2017