**** BEGIN LOGGING AT Thu Mar 30 03:00:02 2017 Mar 30 06:31:32 ant_home: white display? well looks like the issues around the SPI issue has not been fixed yet Mar 30 06:32:03 ant_home: this was present in 4.9 already. https://github.com/greguu/linux-4.9-rc3-c3x00/commit/56219f7f5f1817cd4de61c69ca99d8a5c2ad4310 Mar 30 06:33:22 ant_home: I had to revert to some 4.6 code for SPI as it was introduced somewhere then. Mar 30 06:34:34 ah, bloody hell, ant is not here Mar 30 06:35:58 I never investigated what exactly the issue was, just that reverting to 4.6.7 code fixed it as this was where the change happened after Mar 30 06:36:21 so it may be the same for 4.10 / 4.11 Mar 30 06:36:53 I guess I should send a bug report to the maintainer of that code , right? Mar 30 06:39:30 well something went wrong in steering the SPI LCD controller in drivers/spi/spi-pxa2xx* Mar 30 06:46:19 something SPI DMA stuff cocked up I am sure Mar 30 06:56:09 ant_work: ask the livelogs. I just responded to the white screen issue Mar 30 06:57:00 yes, the dmesg shows warning about spi using deprecated hooks Mar 30 06:57:19 and sharpsl power man is also rather broken Mar 30 06:58:02 I think I'll stay on 4.4 for the moment, let see if we can fix things for 4.12 Mar 30 06:58:45 yeah, better stay out of that mess. :) I might hack together a 4.11 or so once I got more time. Mar 30 06:58:55 who should we contact to sort this out? Mar 30 06:59:06 DMA is also to verify, after PXA_DMA was introduced Mar 30 06:59:34 was there any device tree issues ? did you disable it? Mar 30 07:00:35 official pxa guy today is Robert Jarzmik Mar 30 07:01:01 ok, I saw his name pop some times on kernel.org in the patches Mar 30 07:01:12 but the one converting stuff to devicetree was Linus Walleij Mar 30 07:01:16 do we need some .dtb for Zaurus at some stage Mar 30 07:01:19 ? Mar 30 07:01:25 ok Mar 30 07:01:42 pxxa 25x and 27x have already been converted Mar 30 07:01:50 and tested ? Mar 30 07:01:59 on Zaurus there are some special things Mar 30 07:02:03 of course Mar 30 07:02:09 yet to convert Mar 30 07:02:17 to bad Mar 30 07:02:37 look, I will be playing Thimbleweed park from tomorrow afternoon on anyway. so no worries Mar 30 07:03:15 this can wait Mar 30 07:03:46 talking with anarsoul, he said it is easy, as he did for another pxa iirc Mar 30 07:03:55 I can't find it though... Mar 30 07:03:56 https://github.com/anarsoul Mar 30 07:03:57 well, there is some hope Mar 30 07:04:17 maybe for the zipit Mar 30 07:05:13 I was asking _again_ on the alpine linux irc channel for an arm5tel port... but no luck Mar 30 07:05:38 musl, busybox and openrc, would be great on the Zaurus I believe, packed with grsec Mar 30 07:05:39 maybe this https://github.com/anarsoul/linux-2.6/commits/v4.4-z2 Mar 30 07:06:26 musl is already ok on Zaurus Mar 30 07:06:37 can be built with OE Mar 30 07:06:46 but then we have sysvinit or systemd Mar 30 07:07:01 systemd is a pain, too memory hungry Mar 30 07:07:14 it was ok a few years ago but bloated up. Mar 30 07:07:15 I didn't know openrc was used apart my beloved Gentoo Mar 30 07:07:35 I run alpine on my APU2C4 and I love it Mar 30 07:09:01 gentoo , yes, I wonder if I ever go back to it. I was great until I found arch Mar 30 07:09:16 I'm still on Gentoo for my server Mar 30 07:09:24 but it was better optimized by CFLAGS for my AMD back then Mar 30 07:09:25 since 2000 maybe Mar 30 07:09:47 yeah, thats when I started with it too, on my k2 AMD Mar 30 07:10:08 it still rocks imho Mar 30 07:10:20 great, I was thinking to go back because of systemd in arch Mar 30 07:10:39 heh, systemd is too pervasive Mar 30 07:10:41 it does not matter on my bigger machines but the Zaurus is killed by it Mar 30 07:11:02 I am still on sysvinit for my OE builds Mar 30 07:11:30 it surely is overkill, OOM just by creating eudev cache :( Mar 30 07:12:02 yes, I tried an experimental arch linux arm for the zaurus using busybox and mdev, much better Mar 30 07:12:12 but harder to set up Mar 30 07:12:53 saved heaps of RAM Mar 30 07:13:09 on arch now systemd takes almost half Mar 30 07:13:14 I am practically finishing the mtd parser for kernel Mar 30 07:13:41 last night I had first tests...it ends out bad with null pointer but starts to work! Mar 30 07:13:56 why still parsed the MTD ? Mar 30 07:14:04 just use the SD / CF Mar 30 07:14:12 does OE still fit on NAND ? Mar 30 07:14:22 yes, even two images :) Mar 30 07:14:43 musl ? Mar 30 07:15:07 with glibc a base image is 16-20 mb Mar 30 07:15:22 with musl I didn't yet see big savings, maybe 1 mb Mar 30 07:15:28 but no X11/openbox etc ? Mar 30 07:15:32 I'l lswitch to musl anyway Mar 30 07:15:50 no, that needs a little bit more but is ok on sharp clamshells Mar 30 07:15:58 (not on C3000 ofc) Mar 30 07:16:36 if the parser is done we clean much of the hardcoding done in the board files Mar 30 07:16:38 ok, I never tried to back to NAND, Arch is too demanding, Alpine would work but no ARM5vtel so, SD/CF only Mar 30 07:16:50 and we don't need to specify mtd in devicetree Mar 30 07:16:54 so can Varti boot now Mar 30 07:17:16 yep, let him test :) Mar 30 07:17:45 well, if he can boot kexec he can try 4.9 kernel or 4.4 at least Mar 30 07:17:52 reading the mtd/nand code I have seen that akita has different ecc/oob code than spitz/borzoi Mar 30 07:18:03 maybe there is still a little bug Mar 30 07:18:52 I removed the MTD/NAND parser from my kexecboot as it would slow down the boot for some reason, even I formated all as UBIFS, but that was years ago Mar 30 07:18:53 I gave him a C860 already flashed so he can start on it if akita fails Mar 30 07:19:43 C860 these were great. Mar 30 07:19:45 that was atemporary solution, definitive is in kernel Mar 30 07:19:46 never had one Mar 30 07:20:03 ah ok Mar 30 07:20:40 greguu, imagine I wrote that code basing on nandlogical code (pxa-xrom stuff) which was copied form the kernel 2.4 Mar 30 07:20:54 so now I'm grabbing directly kernel to kernel :) Mar 30 07:22:12 do you still aim to get X11 to work on these ? Mar 30 07:22:20 basically all these clamshells plus poodle/5600 have minimum a 20-30 mb partitions Mar 30 07:22:35 normally mtd2 is over 30 mb Mar 30 07:22:53 mtd3 can vary, 64MiB or 128MiB nand flashes Mar 30 07:23:20 X11 works but slow like hell after the removal of kdrive long ago Mar 30 07:23:40 yes, kdrive or what was the other name for it ? tiny X or something Mar 30 07:24:04 yes, that one Mar 30 07:24:35 I did some work on DirectFB a few years ago and got FB rotation to work and modified directfb code and SDL code to get games to run in FBDEV pretty well Mar 30 07:24:55 but had issues with mouse input and gave up after some time Mar 30 07:25:26 one guy in the OESF forum picked it up and was happy to play doom etc Mar 30 07:25:33 but it was never really optimized Mar 30 07:25:53 all that BBVD/iWMMXT stuff got lost Mar 30 07:26:28 like the accelerated fb for W100 on corgi Mar 30 07:26:30 now I think I can not be bothered to figure it out and modifiy SDL to work with it Mar 30 07:26:45 yes, and ZGROM had some of that in it too. Mar 30 07:26:52 but not shared open source Mar 30 07:27:33 also that guy who made the RailroodTycoon port years ago was very skilled and did his own iwmmxt bitmap bittling and rotation Mar 30 07:28:09 but sadly I do lack these skills and are to busy to optain them (great excuse yes) Mar 30 07:29:16 it was OpenTTD not Railroad Tycoon Mar 30 07:30:53 ZodTTD was the guy Mar 30 07:33:19 there was an optimized W100 SDL library yes, that could do this on stock rom I think Mar 30 07:33:31 but that was for the 760/860 only Mar 30 07:33:56 first step, we have to fix the touchscreen Mar 30 07:34:07 ah ? Mar 30 07:34:13 calibration ? Mar 30 07:34:24 on some models the calibration cannot be done Mar 30 07:34:34 i.e. with old images like OPIE Mar 30 07:34:50 kernel drivers have changed and nobody tested it Mar 30 07:35:10 i.e. ads7486 for c7x0 had fine tuning in the old driver Mar 30 07:35:25 I did extract some tslib calibration files from pdaxrom and they seemd to work if modified for X11 org Mar 30 07:35:41 there is a thread on osef.org Mar 30 07:35:52 not sure about 7xx/8xx Mar 30 07:36:10 I have to retest at least with 4.4 kernels Mar 30 07:36:27 yes, the tool I think does not exist anymore in tslib Mar 30 07:36:35 iir tslib was ok but deprecated while xinput-calibrator had issues Mar 30 07:36:43 but you can use legacy files to optain calibration data from a device , if you have it Mar 30 07:37:16 these you can then use in X11 Mar 30 07:37:31 but these also can vary from device to device, even if same model Mar 30 07:39:18 I think the matrix on how the coordination is represented is a bit different but the values can be used. But I have not played with it for a year or so Mar 30 07:39:19 there are wrong value readings, I have the logs somewhere Mar 30 07:39:29 maybe it is just unrotated matrix Mar 30 07:39:45 yes Mar 30 07:39:50 but kernel drivers did surely change a lot Mar 30 07:39:54 indeed Mar 30 07:42:02 found the old emails Mar 30 07:42:45 do you want the ones of 2011 or of 2016 ? :) Mar 30 07:43:11 both Mar 30 07:43:17 if worth it Mar 30 07:45:18 pls see if you maybe get some idea Mar 30 07:47:54 ok, I will check, I got some good feeback on X11 on Zaurus, Touchscreen worked but was just off some pixels.. see OESF.org Mar 30 07:47:59 I need to reproduce Mar 30 07:48:12 that was on ALARM Mar 30 07:48:32 i am sure it can be done Mar 30 07:48:49 did you get garbage on 4.4 and X11 ? Mar 30 07:49:31 or just from checking raw /dev ? Mar 30 07:50:25 I don't think I even tested X11 on 4.4 Mar 30 07:50:28 :/ Mar 30 07:50:40 just built images Mar 30 07:51:01 I'll build some tonight Mar 30 07:51:14 xinput set-prop "ADS7846 Touchscreen" --type=float "Coordinate Transformation Matrix" 1.2 0 -.0625 0 1.18 -.0834 0 0 1 Mar 30 07:51:21 this seemed to work well for ArchiMark Mar 30 07:51:49 but again, you may need to change the values from a legacy PDAXROM tslib calib file Mar 30 07:52:17 to match this Mar 30 07:53:04 as the calibration tool is missing Mar 30 07:53:36 I still boot to an old debian armel with tslib and calibration working to obtain some calibration file Mar 30 07:55:43 are they valid with the new kernels? Mar 30 07:56:09 this was from Oct 27 2016, so yes at least 4.3 and further Mar 30 07:56:16 I don't remember exactly right now, the issues is long in my TO-DO Mar 30 07:56:32 ok, sounds good Mar 30 07:56:50 but again, here there are bugs :) Mar 30 07:57:47 no surprise ? that is with 4.4.x ?] Mar 30 07:57:56 what version exactly ? Mar 30 07:59:01 ah in general. yes there are many bug indeed Mar 30 07:59:20 some say the online charging worked again, but I never tested Mar 30 07:59:48 * greguu is such a lousy zaurus port maintainer Mar 30 08:17:15 VartiWork: You may want to ask livelogs to catch up Mar 30 08:20:15 * VartiWork checks Mar 30 08:20:59 as of yesterday my build of 4.10.6 failed miserably on a loop of suspend/resume tests Mar 30 08:21:05 on C860 Mar 30 08:21:25 and btw hi greguu and ant Mar 30 08:23:09 I'd need a while to read all of it :) Mar 30 08:24:45 VartiWork: no rush mate, just take your time Mar 30 08:24:50 btw greguu, I did some more testing with kexec, found out how to make it bootable, but unfortunately it only boots once Mar 30 08:25:39 i.e. flash 2.6.2x first, boot, flash 4.4.8, it boots Mar 30 08:26:28 after a reboot or battery+power removal, it doesn't boot anymore, you need to repeat the above procedure to boot it again Mar 30 08:26:53 what kexec is that ? Mar 30 08:27:00 how did you flash it ? Mar 30 08:27:11 ant's latest one Mar 30 08:27:24 latest released one I mean, on LinuxPDA Mar 30 08:27:54 I have flashed it with the OK+Power button Sharp's menu, as always Mar 30 08:28:34 somewhere in the back of my head this issue seems familiar Mar 30 08:28:59 what updater.sh you use ? Mar 30 08:29:14 I'll make tests with the mtb2/3 partitions erased, or formatted with JFFSv2/UBI Mar 30 08:29:43 VartiWork, just for fun, pls repeat testings on c860 Mar 30 08:29:46 ant_work: is that kexecboot vanilla ? Mar 30 08:30:15 the ones supplied with the respective kexecs, I'll check now what versions are Mar 30 08:30:51 I build the one in OE Mar 30 08:30:57 there are still patches Mar 30 08:30:58 http://cgit.openembedded.org/meta-openembedded/tree/meta-initramfs/recipes-bsp/kexecboot Mar 30 08:31:13 iirc is my github fork Mar 30 08:31:45 ant_work: ok, I wanted to test there too indeed after I finish on the 1000 Mar 30 08:32:39 collie needs to be tested too heh Mar 30 08:32:53 ant_work: what is the reason the kexec kenrel needs to go to 4.x series ? As long as it can boot ext4/ubi etc it would not need an updat Mar 30 08:33:12 bluelightning, seeing the patches now, I remember how I protested against the oe-core change which made /boot/zImage absolute symlink Mar 30 08:33:29 bluelightning, imagine I've seen yesterday the symlink is back to relative... Mar 30 08:34:20 greguu, right, maybe some little ubi changes, need to check ABI Mar 30 08:36:30 * greguu wonders if ant_work and bluelightning use telepathy to communicate Mar 30 08:37:51 greguu, it's all more difficult now with time difference :/ He has a family too ;) Mar 30 08:38:08 now I'm in office and he's relaxing (I wish) Mar 30 08:38:10 greguu: the 3.x vanilla kexec binaries are btw lost, they'd need to be rebuilt Mar 30 08:38:17 I see Mar 30 08:38:40 this lost binary or source code thing I will not accept Mar 30 08:38:45 who to blame for Mar 30 08:38:46 thath's a big sh*t..these were rather ok Mar 30 08:39:08 were hosted on kexecboot.org Mar 30 08:39:10 just in case, can they be rebuilt? Mar 30 08:39:24 yes, but it is a pain today Mar 30 08:39:36 old distr, old gcc, binutils, etc Mar 30 08:39:43 I can imagine... Mar 30 08:39:53 but the sources are surely in OE Mar 30 08:40:02 ant_work: just because you use OE Mar 30 08:40:31 send me the source I compile on my debian from 2008 Mar 30 08:40:45 without OE ! Mar 30 08:40:58 I mean, the patches and the rest are not lost :) Mar 30 08:41:08 I'd of course prefer to see a working 4.x kexec, but if that would be too troublesome or require too much work, better to at least have a working 3.x one Mar 30 08:41:12 it was a 3.2 iirc Mar 30 08:41:26 a kexec based on 3.x is fine Mar 30 08:41:30 no need to go 4.x Mar 30 08:41:43 I remember I found the page on archive.org, but not the binaries, let me check Mar 30 08:41:50 I need to see the patches and get my 3.5 based kexec updated Mar 30 08:42:00 for akita Mar 30 08:42:09 spitz/borzoi are ok Mar 30 08:42:30 it might be trivial Mar 30 08:42:34 it was one point here: Mar 30 08:42:35 http://cgit.openembedded.org/meta-handheld/log/?ofs=350 Mar 30 08:42:52 but what if latest ext4 is not compatible with the 3.x one? (I hope this is not a silly question) Mar 30 08:43:04 ah, no worries Mar 30 08:43:13 fs do not change that quick Mar 30 08:43:48 it is just to boot the kernel anyway Mar 30 08:44:10 so basic read partition Mar 30 08:44:13 https://web.archive.org/web/20120615193033/http://kexecboot.org/download Mar 30 08:45:59 ant_work: I guess I have to digest this into seperate diff files ? Mar 30 08:46:35 (greguu: has Thimbleweed park been released? Are you playing it / is it playable on ScummVM on the Z?) Mar 30 08:46:45 no no no Mar 30 08:46:50 this is the answer Mar 30 08:47:14 It is not ScummVM compatible Mar 30 08:47:14 ok Mar 30 08:47:29 It has not been release, but will be in the next hours Mar 30 08:47:34 aha nice Mar 30 08:47:39 I have not played it yet Mar 30 08:47:48 but there will be a linux verison Mar 30 08:47:56 greguu, I remember I tested 3.8 and it was also ok Mar 30 08:47:57 I'll look forward to play it too someday Mar 30 08:48:18 ah, so there were 3.8 binaries too... Mar 30 08:48:43 greguu, looking at it, there aren't patches for Cxxx Mar 30 08:49:00 send me the defconf for akita Mar 30 08:49:05 3.x Mar 30 08:49:08 I never released the 3.8, was testing on collie Mar 30 08:49:12 I know you did already some time ago Mar 30 08:49:31 that 3.2 was ok, anyway Mar 30 08:50:02 I think I did add it in may/june 2012 Mar 30 08:50:28 someone must have the binary Mar 30 08:50:31 someone Mar 30 08:50:52 ask google or NSA , they must have cached it Mar 30 08:51:18 probably these defconfigs Mar 30 08:51:20 http://cgit.openembedded.org/meta-handheld/commit/?id=ac1c8702b33b71fe085188eb174e1565c8a52fa6 Mar 30 08:52:07 ok Mar 30 08:53:19 not to ask for too much but what is the diff of defconfig between spitz and akita Mar 30 08:53:38 for this kernel version Mar 30 08:54:12 wasn't already a symlink? Mar 30 08:54:13 greguu: I don't exclude they are stored in some of the still active feeds, but I couldn't find them Mar 30 08:54:25 iirc I added CONFIG_GPIO_PCA953X=y to spitz as well Mar 30 08:55:22 so was the MTD layout the only reason for having seperate kernels for spitz and akita then ? not the defconfig per se ? Mar 30 08:56:13 there is an unified kernel Mar 30 08:56:29 but for the buildsystem, it has a virtual/kernel for each machine Mar 30 08:56:47 I have to build two distinct kernels Mar 30 08:56:55 I tried to add Cxxxx to mine but still would not boot on akita Mar 30 08:57:26 for kexec Mar 30 08:57:55 how do you try? just try to kexec the kexecboot-kernel of akita Mar 30 08:57:57 anywaya, VartiWork will be the guinea pig Mar 30 08:58:16 I'm pretty sure akita kernel booted on my 3200 Mar 30 08:58:42 heh Mar 30 08:58:59 akita can boot on spitz/borzoi etc Mar 30 08:59:08 but did not work the other way around Mar 30 08:59:22 anyway, as of today, for the 3200 you need this patch Mar 30 08:59:25 http://cgit.openembedded.org/meta-handheld/commit/?id=a268810f7c2c1f13e01cc02bcddc23c46118551d Mar 30 08:59:36 I do not have a C3200 Mar 30 08:59:44 what is the deal? Mar 30 08:59:50 I really hate these nand details, I'm getting rid of these :) Mar 30 09:00:18 43 instead of 32 Mar 30 09:01:40 just scratch the NAND Mar 30 09:01:46 no need, no worries Mar 30 09:01:55 just use SD/CF Mar 30 09:01:58 done Mar 30 09:03:08 I mean I'm fixing the kernel. I do love having two small distros in flash Mar 30 09:06:23 sure Mar 30 09:07:10 but I would get the kernel to boot on SD/CF first and then troubleshoot NAND Mar 30 09:07:51 gives you more room to play then being stuck without a userland Mar 30 09:08:30 so a kernel that does not depend on NAND in the first instance would be good Mar 30 09:08:39 I did disbale this in mine because of that reason Mar 30 09:11:07 greguu: I'd like to set the topic back on #alarmz, could you please exit from there, for a moment? Mar 30 09:14:00 VartiWork: I responded in the #alarmz channel] Mar 30 09:18:29 * greguu will be in Brisbane next week and anyway playing Thimbleweekpark so catch up again later in April Mar 30 09:18:49 ah, have a nice trip then Mar 30 09:19:37 could you please set it to Mar 30 09:19:44 sorry, wrong channel Mar 30 09:22:22 all done Mar 30 09:28:29 thanks Mar 30 09:40:41 sorry I am not very good at responding am I Mar 30 09:41:26 I don't recall that symlink change I'm afraid Mar 30 09:43:55 greguu: forgot to answer you this bit: [10:28] what updater.sh you use ? Mar 30 09:44:26 I have done a check and 2.6's and 4.4.8's updater.sh are different Mar 30 09:45:26 could updater.sh be the culprit? Mar 30 09:50:33 the one I'm using to flash 4.4.8 is btw the same as the one you used for alarmz, both v0.1 and v0.2 releases **** ENDING LOGGING AT Fri Mar 31 03:00:01 2017