**** BEGIN LOGGING AT Mon Mar 05 03:00:03 2018
Mar 05 04:55:48 <Oksanaa> Is there a leste image for Allwinner A33 tablet?
Mar 05 05:21:06 <DocScrutinizer05> Oksanaa: forget about systemd on any maemo that used ohm(?) managed cgroups, systemd wants to rape those and doesn't work without exclusive control over them afaik
Mar 05 05:22:06 <DocScrutinizer05> ooh sixwheeledbeast beat me on it
Mar 05 05:26:46 <DocScrutinizer05> iirc system D even just segfaults when you try to start it (PID1, remember?) on a kernel without cgroups support
Mar 05 05:30:26 <DocScrutinizer05> http://without-systemd.org/wiki/index.php/Arguments_against_systemd
Mar 05 05:31:01 <DocScrutinizer05> >>Unchecked null pointer dereferencing in PID 1 not considered a serious issue - "To make this work we'd need a patch, as nobody of us tests this.", "I will not work on this" - Systemd requires cgroups and segfaults if there is no cgroups support.<<
Mar 05 05:32:23 <DocScrutinizer05> https://bugs.freedesktop.org/show_bug.cgi?id=74589
Mar 05 05:34:17 <DocScrutinizer05> https://bugs.freedesktop.org/show_bug.cgi?id=74589#c4
Mar 05 05:39:53 <Maxdamantus> to be fair, I'm pretty sure you can still create your own cgroups alongside systemd.
Mar 05 05:41:11 <Maxdamantus> I'm not too familiar with them, though I'm fairly sure I tried that a few years ago and it didn't seem like my cgroup would have interferred with systemd's ones or vice versa.
Mar 05 05:44:13 <Maxdamantus> https://wiki.archlinux.org/index.php/Cgroups#Ad-hoc_groups
Mar 05 05:45:15 <Oksanaa> Still, init system should not require this, that and everything? https://www.heise.de/forum/iX/News-Kommentare/Red-Hat-Enterprise-Linux-for-SAP-Solutions/Neulich-auf-einem-RedHat-Server/posting-31123843/show/ 
Mar 05 05:45:47 <Maxdamantus> I agree with that, though the other software should also generally not care.
Mar 05 05:46:06 <Maxdamantus> I'm also not familiar with ohmd, but it sounds like that probably relies on cgroups as well.
Mar 05 05:46:30 <Maxdamantus> and I imagine if it doesn't work alongside systemd, it's probably because ohmd is trying to control systemd's cgroups.
Mar 05 05:47:49 <Oksanaa> At least, ohmd doesn't try to be an init system at the same time :-)
Mar 05 05:48:30 * Oksanaa doesn't understand cgroups well enough to know if it makes sense to "grant some daemon exclusive access to this functionality"
Mar 05 05:52:09 <KotCzarny> it would be like creating firewalld that takes exclusive control of iptables, networking, app network rights, that will run in init 1, had a lot of bugs, 'simplified' config for users while at the same time made it more complicated, required bunch of custom binaries to control etc
Mar 05 05:52:49 <KotCzarny> i'm still amazed those guys didnt went that way yet
Mar 05 05:53:55 <sicelo> Juesto: mer was not Nokia's project and so doesn't count as a 'Maemo.' I don5 recall that it even used Hildon
Mar 05 05:54:28 <DocScrutinizer05> KotCzarny: oh they didn't? are you sure?
Mar 05 05:54:40 <KotCzarny> dont know, havent checked
Mar 05 05:54:51 <KotCzarny> but i suspect it would make some news and fuss
Mar 05 05:54:58 <DocScrutinizer05> I think system D ate iptables
Mar 05 05:55:52 <DocScrutinizer05> networkd anybody?
Mar 05 05:56:17 <DocScrutinizer05> s/firealld/networkd/
Mar 05 05:56:43 <DocScrutinizer05> https://www.phoronix.com/scan.php?page=news_item&px=systemd-networkd-IP-Forward
Mar 05 05:57:19 <Juesto> Oksanaa: nothing about allwinner a33 tablet
Mar 05 05:57:22 <Juesto> yet
Mar 05 05:57:43 <Juesto> sicelo: oh... derp
Mar 05 05:57:53 <KotCzarny> eh
Mar 05 05:58:49 <sicelo> :-)
Mar 05 05:58:56 <DocScrutinizer05> >>The firewall manipulation helpers are used for establishing NAT rules. This support in systemd is provided by libiptc, the library used for communicating with the Linux kernel's Netfilter and changing iptables firewall rule-sets. <<
Mar 05 05:59:44 <sicelo> KotCzarny: they're one step ahead :-)
Mar 05 05:59:47 <DocScrutinizer05> so they didn't grab netfilters/iptables/whatever the name completely yet since it's kernel land
Mar 05 06:00:16 <Maxdamantus> netfilter = kernel, iptables = userspace stuff that configures it
Mar 05 06:00:41 <KotCzarny> as i've said, they try to be both in kernel and userland, breaking separation and security of that model at the same time
Mar 05 06:01:32 <KotCzarny> so, did they implement remote management yet?
Mar 05 06:01:55 <KotCzarny> or just stick to exploiting own bugs for that? ;)
Mar 05 06:01:58 <DocScrutinizer05> yep, of course. It pissed lennart to no end that linus doesn't eat lennart's BS without bitching about it and rejects it instead
Mar 05 06:02:57 <DocScrutinizer05> remote management? only via HTML frontend ;-P
Mar 05 06:03:32 <KotCzarny> which reminds me i should add html frontend to oscp-core
Mar 05 06:03:41 <KotCzarny> har har
Mar 05 06:04:01 <DocScrutinizer05> you're free to do that, OSCP is an app. ÜID1 shouldn't
Mar 05 06:04:04 <KotCzarny> that would make it accessible from any html capable device
Mar 05 06:04:05 <DocScrutinizer05> PID*
Mar 05 06:04:18 <KotCzarny> yup
Mar 05 06:11:18 <Juesto> still curious about mer and all that tbh
Mar 05 06:15:11 <sicelo> -- sailfish 
Mar 05 06:18:27 <Juesto> oh?
Mar 05 06:18:44 <Juesto> so it's sailfish clone?
Mar 05 06:19:00 * Juesto shakes head
Mar 05 06:19:16 <KotCzarny> other way i think
Mar 05 06:19:22 <Juesto> hmm
Mar 05 06:19:29 <Juesto> *perplexed*
Mar 05 06:19:49 <KotCzarny> Sailfish OS (also styled as SailfishOS[4] or abbreviated to SFOS) is a general purpose Linux distribution used commonly as a mobile operating system combining the Linux kernel for a particular hardware platform, the open-source Mer core stack of middleware, a proprietary UI contributed by Jolla or an open source UI, and other third-party components.[1][2]
Mar 05 06:20:10 <KotCzarny> https://en.wikipedia.org/wiki/Sailfish_OS
Mar 05 06:20:25 <KotCzarny> just read the first few paragraphs that explain what comes where
Mar 05 06:29:24 <sicelo> thanks KotCzarny 
Mar 05 07:11:21 <DocScrutinizer05> or just look at http://neo900.org/stuff/cccamp15/ccc2015talk/talk.pdf#6
Mar 05 07:40:33 <Juesto> moblin is defunct or...? idk just remembered about that
Mar 05 07:42:00 <sicelo> tizen :)
Mar 05 07:44:14 <Juesto> Samsung smarttv....
Mar 05 09:21:40 <KotCzarny> anyway, oscp-0.9.81 uploaded. special n8x0 release
Mar 05 09:22:17 <KotCzarny> isnt it nice to have app that can play current formats on a device 10+ years old?
Mar 05 15:45:55 <drathir> KotCzarny: bump of ssl/tls i guess probably woulld be nice too...
Mar 05 15:46:09 <KotCzarny> ssl?
Mar 05 15:47:30 <drathir> openssl^
Mar 05 15:48:25 <KotCzarny> that topic was discussed on the channel few times, it's not trivial because higher openssl versions aren't compatible with old apps
Mar 05 15:49:45 <drathir> KotCzarny: yea im aware of that he best choice probably two version provided old and new, and old app against old build and new against new one...
Mar 05 15:50:20 <drathir> but kinda lot of work/rebuild...
Mar 05 15:55:12 <sicelo> answer - maemo leste
Mar 05 15:55:13 * sicelo hides
Mar 05 15:55:29 <KotCzarny> it's not showtime ready yet
Mar 05 15:55:45 <drathir> or even not sure if still possible build new openssl with flags enabled for old functions but not sure how lon them will be still available, and still first idea better two  separate...
Mar 05 16:05:09 <Juesto> depends
Mar 05 16:09:03 <Juesto> ssl 2.0 doesn't go well with tls 1.2 or sth
Mar 05 17:11:52 <DocScrutinizer05> sicelo: rather maemo leste? answer: what drathir suggested and leste doesn't really do: go ahead without dropping backward compatibility. AIUI in leste you would need to recompile *all* (SSL-depending?) apps, so leste isn't the answer but rather the duplication and potentiation of the situation on fremantle in that regard
Mar 05 17:15:00 <freemangordon> I don;t see anything stopping us from providing openssl 0.9.8 in leste
Mar 05 17:17:03 <DocScrutinizer05> you should :-)
Mar 05 17:17:30 <DocScrutinizer05> though with armhf you will need to recompile every app anyway
Mar 05 17:17:40 <DocScrutinizer05> aiui
Mar 05 17:18:08 <freemangordon> yes. and not only for armhf but also for x64 and arm64 ;)
Mar 05 17:18:22 <DocScrutinizer05> then it doesn't matter when you provide/link against newest openssl
Mar 05 17:18:35 <freemangordon> it does, as APIs are not compatible
Mar 05 17:18:42 <freemangordon> so it would mean less porting
Mar 05 17:20:34 <freemangordon> which remands me we're near to the point we'll have to port certman :). That would be fun
Mar 05 17:20:41 <DocScrutinizer05> I think an OS of this (sorry) limited scale regarding users and devs will fail like a 2 dozen others I've seen failing, when you don't have an existing convenient pool of readily usable apps
Mar 05 17:21:40 <DocScrutinizer05> it's the chicken/egg problem, always
Mar 05 17:22:20 <freemangordon> DocScrutinizer05: sure we may fail. Though, we HAVE those apps already in the repos
Mar 05 17:22:31 <freemangordon> they just need to be recompiled
Mar 05 17:23:30 <DocScrutinizer05> yes, I hope all the best for leste that >90% will just build without any issues
Mar 05 17:23:41 <freemangordon> see, if "users" don;t want/need free mobile OS, so be it
Mar 05 17:23:46 <freemangordon> exactly
Mar 05 17:24:05 <freemangordon> also, I expect similar percentage to be binary compatible on armel
Mar 05 17:24:29 <freemangordon> but we still don;t have enough of the "bas" to start such tests
Mar 05 17:24:30 <KotCzarny> hard part is getting image/installer generic enough for the majority
Mar 05 17:24:56 <KotCzarny> most users just ask 'where is the image for my XYZ device'
Mar 05 17:24:56 <DocScrutinizer05> I admit I'm not really savvy to judge on the ABI-compatibility between armel and armhf
Mar 05 17:25:09 <freemangordon> KotCzarny: which is a problem of HW/kernel
Mar 05 17:25:26 <freemangordon> DocScrutinizer05: In short - there is no such compatibility :)
Mar 05 17:25:27 <KotCzarny> some of them posses ability to piece/compile os given the recipe
Mar 05 17:25:43 <KotCzarny> but most just want readily working thingy
Mar 05 17:26:28 <freemangordon> KotCzarny: I think we'll just provide online installer for devuan, so, as long as your device can boot to devuan, you're fine
Mar 05 17:26:31 <KotCzarny> and it's especially sore in arm land
Mar 05 17:27:36 <DocScrutinizer05> freemangordon: do you work on arm64 yet?
Mar 05 17:27:37 <freemangordon> I know, but without us doing something in the userland it will never change. Not that we're guaranteed to have that change with leste, but still
Mar 05 17:28:06 <KotCzarny> i'm doing h3droid, which is targetting one specific soc, and there are people without the skill to just write the .img.xz to sdcard, boot it and select their board model
Mar 05 17:28:21 <freemangordon> DocScrutinizer05: no, I don;t have HW, but I am pretty sure it will be fine as most of the develpmnet done by me was on x64
Mar 05 17:28:55 <KotCzarny> and it's sad, but there is a chance with few specific devices supported it will fly
Mar 05 17:29:27 <DocScrutinizer05> freemangordon: what do you think about this (SoC)? https://www.olimex.com/Products/OLinuXino/A64/A64-OLinuXino/open-source-hardware
Mar 05 17:29:31 <freemangordon> KotCzarny: I think we can provide image, given you can boot upstream kernel on the particular device
Mar 05 17:30:34 <freemangordon> DocScrutinizer05: afaik Wizzup has one (or plans to get)
Mar 05 17:30:52 <freemangordon> that'd be one of the supported targets afaik
Mar 05 17:30:54 <freemangordon> see https://maemo-leste.github.io/pages/allwinner-a20-lime2-status.html
Mar 05 17:32:22 <DocScrutinizer05> Allwinner A20[1]	(sun7i)	 2 x Cortex-A7 CPU-cores ||  Allwinner A64[4]	(sun50i) 	 4 x Cortex-A53 CPU-core
Mar 05 17:32:45 <freemangordon> Maili GPU, right?
Mar 05 17:33:06 <KotCzarny> A33 is also sweet target, plenty of sub 50usd ips display tablets
Mar 05 17:33:13 <freemangordon> yes
Mar 05 17:33:40 <freemangordon> but, we still have problems with Mali there, I really hope this will be solved soon
Mar 05 17:33:50 <KotCzarny> and the nice thing is that support for mali and video accel engines are being done
Mar 05 17:33:58 <freemangordon> where?
Mar 05 17:34:04 <freemangordon> ah
Mar 05 17:34:08 <freemangordon> "being" done
Mar 05 17:34:19 <KotCzarny> yeah, did you see the VE kickstarter?
Mar 05 17:34:25 <freemangordon> no
Mar 05 17:34:35 <KotCzarny> https://www.kickstarter.com/projects/bootlin/allwinner-vpu-support-in-the-official-linux-kernel
Mar 05 17:35:26 <freemangordon> ah, VPU (DSP)
Mar 05 17:35:26 <KotCzarny> first stretch goal is also ticked off
Mar 05 17:35:33 <DocScrutinizer05> >>GPU: Mali400, Mali400-MP2, SGX544 or PowerVR G6230 Graphics Procesor Unit, supporting OpenGL ES2.<<
Mar 05 17:35:52 <KotCzarny> so basically any Axx series will get hw video decoding open sourced and mainlined
Mar 05 17:35:54 <DocScrutinizer05> friggin useful¡
Mar 05 17:36:15 <freemangordon> KotCzarny: and what about GPU? any news there?
Mar 05 17:36:46 <DocScrutinizer05> http://linux-sunxi.org/Allwinner_SoC_Family
Mar 05 17:36:49 <KotCzarny> apart from that mripard blog entry
Mar 05 17:37:12 <freemangordon> mhm, anything else?
Mar 05 17:37:23 <DocScrutinizer05> yep, Mali400: http://linux-sunxi.org/Mali400
Mar 05 17:37:28 <freemangordon> as drivers are broken ATM
Mar 05 17:37:38 <DocScrutinizer05> Mali-400 MP2	 1	 2	 256 KiB	A20 (sun7i), A23 (sun8i), A33, H3 (sun8i), A64 A64
Mar 05 17:37:50 <KotCzarny> that amd guy is still the hope
Mar 05 17:38:12 <freemangordon> which one, link?
Mar 05 17:38:25 <KotCzarny> https://github.com/yuq/linux-lima/tree/lima-4.13/drivers/gpu/drm/lima
Mar 05 17:38:30 <freemangordon> ah, lima
Mar 05 17:38:37 <KotCzarny> but not the old lima
Mar 05 17:38:50 <KotCzarny> yuq took over 
Mar 05 17:39:34 <freemangordon> does it work at all?
Mar 05 17:39:57 <KotCzarny> haven't really checked, probably wizzup or folks on #linux-sunxi might say more
Mar 05 17:40:34 <freemangordon> Wizzup was too busy with leste stuff lately ;)
Mar 05 17:40:39 <freemangordon> but I'll ask him
Mar 05 17:40:52 <freemangordon> as I have A33 in front of me
Mar 05 17:41:09 <freemangordon> gathering dust because of unusable GPU accell
Mar 05 17:41:42 <KotCzarny> well, you might get that tree and try out
Mar 05 17:41:53 <freemangordon> 4.13?
Mar 05 17:41:56 <KotCzarny> yup
Mar 05 17:42:28 <KotCzarny> master seems to be unused
Mar 05 17:42:36 <freemangordon> what about userspace? mesa?
Mar 05 17:43:02 <KotCzarny> since it's being done as generic drm driver, probably
Mar 05 17:43:16 <freemangordon> and modesetting I gues
Mar 05 17:43:19 <freemangordon> *guess
Mar 05 17:43:29 <KotCzarny> there is mesa-lima repo too
Mar 05 17:43:37 <KotCzarny> https://github.com/yuq/mesa-lima
Mar 05 17:43:49 <KotCzarny> so probably need both
Mar 05 17:51:04 <Wizzup> I am not so sure that lima does anything useful in mesa yet
Mar 05 17:51:37 <Wizzup> freemangordon: I also have Olimex OLinuXino A64
Mar 05 17:52:39 <Wizzup> specifically regarding providing old and unsupported openssl libs -- what could go wrong :)
Mar 05 17:53:19 <freemangordon> actually they ARE supported by centos AFAIK ;)
Mar 05 17:53:23 <DocScrutinizer05> freemangordon: so what's your personal "sweetspot" SoC?
Mar 05 17:53:44 <DocScrutinizer05> max grunt and still open enough
Mar 05 17:53:50 <freemangordon> omap3 :)
Mar 05 17:53:55 <DocScrutinizer05> LOL
Mar 05 17:53:56 <KotCzarny> :>
Mar 05 17:54:06 <DocScrutinizer05> why not OMAP5?
Mar 05 17:54:09 <Wizzup> qualcomm socks with adreno gpus are not that badf
Mar 05 17:54:13 <Wizzup> s/badf/bad/
Mar 05 17:54:13 <infobot> Wizzup meant: qualcomm socks with adreno gpus are not that bad
Mar 05 17:54:17 <Wizzup> s/socks/socs/ wat
Mar 05 17:54:40 <KotCzarny> aren't qualcomm inception socs? computer in a computer?
Mar 05 17:55:12 * freemangordon have to go afk
Mar 05 17:55:13 <freemangordon> bbl
Mar 05 17:55:32 <DocScrutinizer05> qualcomm sounds nasty. I seem to recall they got a record for not being open
Mar 05 17:55:46 <Wizzup> ifc6410 works quite nicely on mainline
Mar 05 17:55:52 <Wizzup> including opengl and opengles by a free driver
Mar 05 17:55:53 <KotCzarny> i suspect their recent move to openness doesnt involve sharing all details
Mar 05 17:56:01 <Wizzup> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm/boot/dts/qcom-apq8064-ifc6410.dts
Mar 05 17:56:08 <KotCzarny> ie. they could just show you what they want you to
Mar 05 17:56:23 <KotCzarny> and keep real cpus hidden and working behind
Mar 05 17:56:34 <Wizzup> right - there's many aspects to this of course
Mar 05 17:56:45 <Wizzup> but if you want a powerful yet mostly supported (with foss gpu driver) soc, that one works
Mar 05 17:57:22 <Wizzup> freemangordon: I will try mali again a bit later on a20, I got module and dts to work fine, just had trouble with userspace and debugging it wasn't obvious IIRC
Mar 05 18:01:13 <DocScrutinizer05> Wizzup: who created that DT and based on what? 
Mar 05 18:01:22 <Wizzup> linaro did a lot of work
Mar 05 18:01:35 <Wizzup> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/arch/arm/boot/dts/qcom-apq8064-ifc6410.dts
Mar 05 18:02:43 <Wizzup> main issue I had with the board that is gets too hot and downclocks
Mar 05 18:02:56 <Wizzup> but the opengl driver is best on all of arm platforms :) 
Mar 05 18:03:01 <Wizzup> of all of*
Mar 05 18:03:06 <DocScrutinizer05> I always got the same problem with that stuff - there's no reference to any datasheet or TRM
Mar 05 18:03:41 <Wizzup> yes, that is a problem with I think all major SoC vendors now
Mar 05 18:04:35 <DocScrutinizer05> basically somebody throws some code over the wall and you can check if it works or not, but not really fix or improve it, or understand what it really does
Mar 05 18:06:20 <DocScrutinizer05> "the code is the documentation" comes to a dead end when looking at device drivers
Mar 05 18:08:32 <Wizzup> it's impressive how well it works though 
Mar 05 18:09:20 <KotCzarny> as long the code is correct
Mar 05 18:09:40 <KotCzarny> and matches the binaries
Mar 05 18:09:55 <KotCzarny> 'oh, hey, we forget to include few essential bugfixes'
Mar 05 18:10:24 <DocScrutinizer05> from a security audit perspective as well as from a FOSS perspective it's a failed approach
Mar 05 18:11:48 <Wizzup> yeah, so it depends on the definition of open -- open drivers or open specs
Mar 05 18:11:53 <Wizzup> (or both)
Mar 05 18:12:21 <KotCzarny> both
Mar 05 18:12:31 <Wizzup> well, sure, but he asked about a sweetspot
Mar 05 18:12:39 <Wizzup> he didn't even ask me, but still, I like talking about arm socs
Mar 05 18:12:49 <DocScrutinizer05> what REALLY pisses me off is FSF supporting that approach
Mar 05 18:13:03 <Wizzup> yeah, fsf is not fhf :P 
Mar 05 18:13:07 <Wizzup> kinda sad
Mar 05 18:24:22 * drathir remember was mentioned that openssl get build flag to enable old one but it need extra switch to enable and in future will be deleted from build flags too...
**** ENDING LOGGING AT Tue Mar 06 03:00:02 2018