**** BEGIN LOGGING AT Fri Dec 15 03:00:02 2017
Dec 15 11:35:00 <greguu> ant_work: hi, what is the kernel status like ? 4.16 still the goal?
Dec 15 11:36:25 <ant_work> yes, I don't see yet the lcd backlight fix in -next / 4.15 trees
Dec 15 11:38:26 <ant_work> so unless new regriessions are introduced we should have a vanilla, patch-free 4.16 for pxz zaurus
Dec 15 11:38:38 <ant_work> *pxa
Dec 15 11:45:17 <greguu> great, so 4.15 or -next is already good to go ? Anything outstanding ?
Dec 15 11:46:21 <greguu> or hold the horses till 4.16 ?
Dec 15 11:50:47 <ant_work> it's just that silly white lcd...
Dec 15 11:51:10 <ant_work> I didn't test 4.15-rc yet
Dec 15 11:52:41 <ant_work> btw I fear there are issues with 'new' touchscreen drivers: we need to check in userspace
Dec 15 11:53:05 <ant_work> the old driver used tslib but now we only have the input/event interface
Dec 15 11:53:26 <greguu> so the white blank issue is still present ?
Dec 15 11:53:29 <ant_work> afair the TS did spit strange values and I did not finish the calibration
Dec 15 11:53:44 <greguu> yes , you need to use Xorg now to calibrate the touchscreen
Dec 15 11:55:07 <ant_work> about lcd: https://marc.info/?l=linux-fbdev&r=1&b=201712&w=2
Dec 15 11:55:50 <ant_work> still in limbo... "Didn't I already ack this one?"
Dec 15 11:56:05 <ant_work> today ;)
Dec 15 11:56:48 <greguu> hm, yes I can see, this SPI madness is still ongoing, I wonder what regression introduced it in the first place
Dec 15 11:56:53 <ant_work> greguu, calibration fails in OPIE images (old tslib) and in newer X11 images with matchbox/sato
Dec 15 11:57:33 <ant_work> ..but works on poodle
Dec 15 11:57:53 <ant_work> so I think it is the kernel driver
Dec 15 12:01:21 <greguu> wow, tslib has a new release from 12th of Nov 2017. Does it include the calibrate tool ?
Dec 15 12:02:46 * greguu got a bit out of touch in the last months
Dec 15 12:06:29 <ant_work> afair there was some interpolation done by tslib which is supposedly done now by the kernel driver
Dec 15 12:06:40 <greguu> afaik touch screen can be configured bu xinput
Dec 15 12:07:29 <ant_work> it's now th einput subsytem so userspace uses uevent, evdev, evtest
Dec 15 12:08:06 <ant_work> basically just need some time and logging ;)
Dec 15 12:09:16 <ant_work> note that it is broken since 3,2 kernels iirc, with the new ts drivers, PATA, etc
Dec 15 12:09:57 <ant_work> (in OpenEmbedded, with our userspace)
Dec 15 12:11:07 <greguu> hm, so there are two ways, use Xorg and xinput, only working in a X session
Dec 15 12:11:41 <greguu> or use tslib, also allowing out of X touchscreen
Dec 15 12:11:43 <greguu> ?
Dec 15 12:12:35 <ant_work> right, OPIE did not use X, just tslib
Dec 15 12:13:30 <greguu> so is tslib really needed ? most console apps would not benefit from touch interface .?
Dec 15 12:15:03 <ant_work> let me remember..I think it was deprecated
Dec 15 12:15:13 <greguu> indeed
Dec 15 12:15:42 <ant_work> http://zaurus-devel.linuxtogo.narkive.com/TVu3p2Z9/touchscreen-calibration-issues-again
Dec 15 12:15:51 <ant_work> 7 years ago ;)
Dec 15 12:16:17 <greguu> yeah, nostalgia
Dec 15 12:16:41 <greguu> wayland for pxa2xx framebuffers ?
Dec 15 12:17:05 <greguu> what is tha futua ?
Dec 15 12:17:11 <ant_work> heh
Dec 15 12:18:01 <ant_work> we have anyway to check "1. Fix ads7846 kernel driver noise" has been done. iirc yes
Dec 15 12:18:03 <greguu> Xorg is hardly usable on the Zaurus, dont you agree
Dec 15 12:18:22 <ant_work> is heavy, yes, we used xserver-kdrive
Dec 15 12:18:40 <greguu> what ? this is still maintained ? kdrive?
Dec 15 12:18:55 <greguu> X11?
Dec 15 12:18:58 <greguu> mean
Dec 15 12:18:59 <ant_work> nope, that is the issue
Dec 15 12:19:24 <ant_work> once kdrive was removed we found ourselves w/o ts support
Dec 15 12:19:39 <ant_work> (it was a special kdrive)
Dec 15 12:19:40 <greguu> stranded
Dec 15 12:19:46 <greguu> I know
Dec 15 12:20:07 <ant_work> in the meanwhile OPIe, QT based, still works
Dec 15 12:20:19 <ant_work> so I agree, we don't need X
Dec 15 12:20:31 <ant_work> but seems the only viable choice atm
Dec 15 12:21:01 <greguu> hm, sad that DirectFB left the game
Dec 15 12:21:14 <ant_work> :/
Dec 15 12:21:49 <ant_work> and ùfbdev is regularly deprecated
Dec 15 12:21:50 <greguu> was good, had pxa3 hardware support even
Dec 15 12:21:59 <ant_work> *fbdev
Dec 15 12:23:11 <greguu> fbdev was never really good, DirectFB tried to fill the gap, for PXA2 not much gain but at least rotation, for PXA3 hardare
Dec 15 12:23:49 <greguu> no it is KMS or nothing
Dec 15 12:24:08 <ant_work> I understand modern hw needs DRM/KMS
Dec 15 12:25:45 <greguu> I patched DirectFB and a SDL 1.2 version many years ago to get games going without X
Dec 15 12:26:09 <greguu> worked quite well, but I struggled getting mouse coordinates going
Dec 15 12:26:14 <ant_work> epic times
Dec 15 12:26:20 <greguu> due to rotation
Dec 15 12:26:58 <greguu> well, now I am happy to boot Arch in console only
Dec 15 12:27:25 <ant_work> in the last years I have been fixing the console-images only
Dec 15 12:28:07 <ant_work> once kernel has stabilized again we can finally cook up something
Dec 15 12:28:39 <greguu> I tried to convince Alpine Linux developer Nathan to release a armv5tel port, but no luck.
Dec 15 12:28:52 <greguu> can we make a musl based port happen with OE ?
Dec 15 12:29:14 <ant_work> btw one kernel for all akita/spitz/borzoi/terrier and one for the 7/8xx clamshells
Dec 15 12:29:35 <ant_work> I already build musl
Dec 15 12:29:55 <ant_work> is ok on armv4/armv5 but must be built with ARM ISA
Dec 15 12:29:59 <ant_work> not thumb
Dec 15 12:30:16 <greguu> thumb ? never really used it
Dec 15 12:30:40 <greguu> native musl arm 5 would be great
Dec 15 12:31:03 <ant_work> we have the privilege, the OE maintainer is one of the lead developers of musl ;)
Dec 15 12:31:13 <greguu> ha!
Dec 15 12:31:37 <ant_work> I recently sent him the last two bulid fixes for armv4/5
Dec 15 12:32:16 <greguu> I belive Alpine Linux is agreat distro, just a shame they do not provide armv5tel support. Maybe put some pressure on
Dec 15 12:32:20 <ant_work> about musl, I still don't see big size advantages
Dec 15 12:32:28 <greguu> not about size
Dec 15 12:32:35 <ant_work> I remain on klibc for kexecboot stuff
Dec 15 12:32:37 <greguu> memory footprint
Dec 15 12:32:49 <ant_work> ah, that's different story
Dec 15 12:32:52 <greguu> ah, so klibc vs musl
Dec 15 12:33:09 <ant_work> klibc-static is unbeatable
Dec 15 12:33:29 <ant_work> kexecboot+kexec+ubiattach less than 80KiB
Dec 15 12:35:09 <greguu> true, this is a win in static file size
Dec 15 12:35:16 <greguu> good for kexecboot
Dec 15 12:35:37 <greguu> but for OS, musl seems to be the most lean libc
Dec 15 12:35:51 <greguu> on low RAM system
Dec 15 12:36:25 <greguu> OE build on musl yet?
Dec 15 12:38:06 <ant_work> it one setting
Dec 15 12:38:21 <greguu> tested ?
Dec 15 12:38:27 <greguu> I am interested ?
Dec 15 12:39:08 <ant_work> yes, widely
Dec 15 12:39:17 <ant_work> https://github.com/kraj/meta-musl
Dec 15 12:39:58 <ant_work> nowadays it is merged in oe-core so you don't need an extra layer
Dec 15 12:40:23 <greguu> just some flags ?
Dec 15 12:40:24 <ant_work> you just set  TCLIBC=musl   in local.conf
Dec 15 12:40:29 <greguu> hm I see
Dec 15 12:40:41 <ant_work> default is glibc still
Dec 15 12:41:03 <greguu> any feedback on musl on armv5tel ?
Dec 15 12:41:10 <ant_work> I use the same setup/instructions but add meta-handheld and meta-initramfs layers
Dec 15 12:41:28 <ant_work> maybe I still have some musl images uploaded...let me check
Dec 15 12:41:38 <greguu> so no compile issues at all ?
Dec 15 12:41:57 <ant_work> not at all
Dec 15 12:42:57 <greguu> like to test a basic busybox system based on musl (armv5tel)
Dec 15 12:43:11 <ant_work> with latest musl, just my pending patch here
Dec 15 12:43:12 <ant_work> https://patchwork.openembedded.org/patch/146600/
Dec 15 12:43:33 <greguu> ah, now you try to drag my again to use bitbake and OE dev tools.
Dec 15 12:43:39 <greguu> :)
Dec 15 12:44:01 <ant_work> well, I bet the flags are named differently for buildroot
Dec 15 12:44:26 <ant_work> but here we have very reproducible builds now
Dec 15 12:45:07 <greguu> buildroot a good alternative for testing but not for release
Dec 15 12:45:23 <ant_work> when I had issues with gcc7 I tried it out all
Dec 15 12:45:28 <ant_work> linaro, buildroot, ..
Dec 15 12:45:38 <ant_work> I had to go back to OE ;)
Dec 15 12:45:52 <greguu> haha
Dec 15 12:46:00 <ant_work> buildroot did not support armv5
Dec 15 12:46:05 <ant_work> in the latest rel
Dec 15 12:46:20 <greguu> really ? I thought it is possible, even as we speak with musl
Dec 15 12:46:34 <ant_work> I could not find any toolchain
Dec 15 12:46:45 <ant_work> with gcc7/armv5
Dec 15 12:46:59 <greguu> hm, maybe this is a build yourself part
Dec 15 12:47:48 <greguu> I wonder how long Arch will support armv5. At the moment it is still pretty good up to date packages.
Dec 15 12:47:54 <ant_work> in fact OE is unique, supporting EABI for armv4. Nobody does that
Dec 15 12:48:09 <greguu> armv4 and EABI, what the hell.
Dec 15 12:48:23 <ant_work> most packages are shareable btw 4/5
Dec 15 12:48:32 <ant_work> you only build once
Dec 15 12:48:53 <ant_work> if you change to OABI you are alone
Dec 15 12:49:27 <greguu> OABI Debian still alive?
Dec 15 12:49:33 <greguu> hehe
Dec 15 12:50:05 <greguu> OpenBSE hat trouble to change to EABI and dropped Zaurus , nio ?
Dec 15 12:50:23 <greguu> no pun intended
Dec 15 12:50:26 <greguu> :)
Dec 15 12:50:43 <ant_work> people 'only' patched gcc and binutils then we are here
Dec 15 12:51:39 <ant_work> armv4 and 5 are almost deprecated by gcc devs
Dec 15 12:51:52 <ant_work> happily there is still a strong kernel community around
Dec 15 12:52:03 <ant_work> so gcc6 and gcc7 are still ok
Dec 15 12:52:06 <greguu> true, and we are stil here debating how we can support this obsolete platform
Dec 15 12:52:07 <greguu> :)
Dec 15 12:52:11 <ant_work> he he
Dec 15 12:52:37 <ant_work> once Russel King ditches armv4 we are lost
Dec 15 12:52:39 <greguu> strong kernel community ?
Dec 15 12:52:51 <greguu> well one or two dedicated devs
Dec 15 12:52:55 <ant_work> the nr 1 ARM kernel developer
Dec 15 12:53:07 <greguu> send Russel some roses
Dec 15 12:53:11 <greguu> make him happy
Dec 15 12:53:17 <ant_work> RMK is a strange guy ;)
Dec 15 12:53:30 <ant_work> lot of enemyes, lot of pride ;)
Dec 15 12:53:41 <greguu> well send tulips
Dec 15 12:54:24 <ant_work> http://armlinux.simtec.co.uk/whoswho.html
Dec 15 12:54:26 <greguu> good developers always use emotions, look at linus
Dec 15 12:54:36 <ant_work> looks like Woody Allen young
Dec 15 12:54:38 <greguu> not a bad thing per se
Dec 15 12:54:46 <greguu> lol
Dec 15 12:55:35 <ant_work> he probably knows the ARM Linux kernel better than anyone else
Dec 15 12:55:37 <ant_work> scary
Dec 15 12:55:51 <greguu> Russel ? sure he does..
Dec 15 12:56:32 <greguu> we had some pretty good contributors for Z kernels, all are gone now
Dec 15 12:57:07 <ant_work> right, but most of the work is done
Dec 15 12:57:32 <greguu> yes, true, it is done, and then they drop it because ther are no users..
Dec 15 12:57:45 <greguu> :(
Dec 15 12:57:46 <ant_work> shame that the 16patches c/o lumag for collie/poodle are bitrotting
Dec 15 12:58:07 <greguu> ah, collie / poodle are dead
Dec 15 12:58:23 <greguu> spitz, akita maybe aklive
Dec 15 12:58:24 <ant_work> I might try to rebump those but I really don't know much about the matter
Dec 15 12:58:25 <greguu> alive
Dec 15 12:59:09 <ant_work> you see, most is about collie SD card reader
Dec 15 12:59:27 <ant_work> driver never went upstream
Dec 15 12:59:33 <greguu> what is the drama ?
Dec 15 12:59:39 <greguu> not compatible ?
Dec 15 13:00:00 <ant_work> C cards are disappearing, you can find a n user with collie and SD and you're f***ed
Dec 15 13:00:24 <ant_work> collie only boots from CF or MTD
Dec 15 13:01:00 <ant_work> kexecboot 4.x for collie is fixed
Dec 15 13:01:19 <ant_work> but the last working kernel (with bitrotting patchset) is 4.4
Dec 15 13:01:32 <greguu> not too bad
Dec 15 13:01:41 <greguu> 4.4 is still LTS
Dec 15 13:01:41 <ant_work> that's why I am still on 4.4
Dec 15 13:01:48 <greguu> :)
Dec 15 13:02:08 <ant_work> then huge SPI and GPIO changes :/
Dec 15 13:02:15 <ant_work> ...as you know
Dec 15 13:02:37 <greguu> yes
Dec 15 13:02:40 <greguu> ..
Dec 15 13:03:15 <greguu> darn, it is getting late here
Dec 15 13:03:24 <ant_work> we'll have then 4.4 for collie and maybe poodle (battery driver is also in that patchset) and 4.16 for corgi/spitz/tosa
Dec 15 13:03:52 <greguu> I hold back until 4.16
Dec 15 13:03:52 <ant_work> as sonn 4.16 is out (Jan-Feb I'd say)
Dec 15 13:03:57 <greguu> hah
Dec 15 13:04:02 <ant_work> yes, it's the best
Dec 15 13:04:10 <ant_work> maybe late
Dec 15 13:04:14 <ant_work> r
Dec 15 13:04:44 <ant_work> ...neverthless we can start playing with the touchscreens :=)
Dec 15 13:04:49 <greguu> the community is pretty weak, active users still can cope with 4.12-rc1
Dec 15 13:05:00 <ant_work> no doubts
Dec 15 13:05:52 <greguu> with Arch the issue is that a base rootfs needs to be replaced frequently otherwise the signatures are too old
Dec 15 13:06:25 <greguu> I may release anotherone early next year
Dec 15 13:08:06 <greguu> systemd seems to be a memory hog on arm
Dec 15 13:08:30 <greguu> Alpine uses OpenRC, what is OE using ?
Dec 15 13:08:40 <ant_work> I still use sysvinit
Dec 15 13:08:51 <ant_work> minimal scripts
Dec 15 13:08:56 <greguu> good
Dec 15 13:09:12 <greguu> busybox too ?
Dec 15 13:09:23 <ant_work> I can't convince me to like systemd and its pervasive attitude
Dec 15 13:09:41 <ant_work> OpenRC here on gentoo server in office
Dec 15 13:10:03 <ant_work> we have a configurable busybox
Dec 15 13:10:14 <ant_work> stripped down basically
Dec 15 13:10:25 <greguu> I use systemd on my arch machines, but I move my laptop to Aline recently. Servers all run Alpine.
Dec 15 13:11:12 <greguu> one day I go back to Gentoo I am sure..
Dec 15 13:11:42 <ant_work> never tried Alpine, I am lazy and the last 3 servers have been gentoo
Dec 15 13:11:58 <ant_work> btw updated two days ago from .13 to .17
Dec 15 13:12:09 <ant_work> gentoo server profile
Dec 15 13:12:18 <greguu> I love Alpine, I started with Suse, the Gentoo, Arch , Alpine
Dec 15 13:13:06 <greguu> Alpine is as minmal as it gets, of course gentoo is always what you want
Dec 15 13:13:15 <ant_work> I lost inrterest testing differnent distros, back the all had buggy config tools
Dec 15 13:13:24 <ant_work> so I did prefer to edit all by hand
Dec 15 13:13:54 <ant_work> you see why I embraced OE then ;)
Dec 15 13:14:04 <greguu> no, I do not see that
Dec 15 13:14:13 <ant_work> self built, self configured
Dec 15 13:15:19 <greguu> I like the "fetch source / build" wau
Dec 15 13:15:30 <greguu> but
Dec 15 13:16:08 <greguu> it takes a friggint long time to compile and I pay for the power bill
Dec 15 13:16:33 <ant_work> not anymore
Dec 15 13:16:39 <ant_work> I build OE on a laptop
Dec 15 13:16:56 <ant_work> ok, its an i7 quad but you get the idea
Dec 15 13:17:36 <ant_work> I talked with Varti about hosting a package feed
Dec 15 13:17:50 <greguu> yes good idea
Dec 15 13:17:52 <ant_work> to host these prebuilt packages
Dec 15 13:18:03 <greguu> also host a build bot :)
Dec 15 13:18:13 <ant_work> I can put it initially on my NAS and let Varti fetch
Dec 15 13:18:38 <ant_work> I have asked for a slot on the big buildserver but no chanche atm...
Dec 15 13:26:13 <greguu> I got myself a old Z600 , dual Xeon 2.5 Ghz, 24 GB RAM - fast compile, big power bill /  warm room.
Dec 15 13:30:53 <ant_work> good for austral winter
Dec 15 13:33:24 <greguu> haha, yes, I want to have a threadripper, but shy of spending money...
Dec 15 13:35:14 <greguu> Irony is, recently I'am using my aarch64 tv box to compile aarch64 packages..
Dec 15 13:37:37 <ant_work> aarch64...will they fix kexec one day?
Dec 15 13:38:27 <ant_work> I have build issues with kexec-tools and older mtd-utils when built with klibc
Dec 15 13:38:57 <ant_work> actually I am masking out aarch64 form the kexecboot recipes :)
Dec 15 13:40:25 <greguu> never tried it with kexecboot, just using it for LibreElec
Dec 15 13:41:21 <ant_work> ah, for that I use the old Xeon  (Lakka for my kid)
Dec 15 13:43:15 <ant_work> I risk to get even involved with that...on my TV box runs enigma2 with Kodi 16.x and I need 17.x
Dec 15 13:43:32 <ant_work> it is a mipsel
Dec 15 13:44:04 <greguu> hm, kodi 17.x is way better than 16.x
Dec 15 13:44:30 <ant_work> seems so but afais there are issues with the proprietary drivers
Dec 15 13:44:38 <greguu> yes
Dec 15 13:44:51 <ant_work> mine is a vuplus duo2
Dec 15 13:44:59 <ant_work> broadcom
Dec 15 13:45:28 <ant_work> they got the driver for 16.x but never upddated. kernel is also 3.x
Dec 15 13:45:36 <greguu> I run it on a s905 AmLogic
Dec 15 13:46:04 <ant_work> does it run mainstream kernel?
Dec 15 13:47:17 <greguu> not if you run Android
Dec 15 13:47:37 <ant_work> ah, I see... mali
Dec 15 13:47:43 <ant_work> same issue
Dec 15 13:47:55 <greguu> yes, depend on Andrio
Dec 15 13:49:41 <greguu> LibreELEC for this platform is pretty good but they must use libhybrid to get access to the hardware grafix
Dec 15 13:49:56 <greguu> a shame..
Dec 15 13:51:36 <greguu> /s/libhybris
Dec 15 13:52:02 <greguu> however, the performance is pretty good.
Dec 15 13:56:23 * greguu is playing kraftklub
Dec 15 13:57:34 <greguu> its 2.55 am here, so good night. ant_work
Dec 15 13:57:46 <ant_work> ouch...
Dec 15 13:57:53 <greguu> hehe
Dec 15 13:58:29 <ant_work> have a nice week-end
Dec 15 13:58:35 <greguu> wake me up  when 4,16 is out :)
Dec 15 13:58:38 <greguu> you too
Dec 15 13:58:44 <ant_work> heh, bye
**** ENDING LOGGING AT Sat Dec 16 03:00:01 2017