17:07 <Joerg-Neo900> [2017-11-10 Fri 05:51:33] <pabs3> "An update on the Android problem" https://lwn.net/SubscriberLink/738225/28d2098ce220b2ee/ [2017-11-10 Fri 16:53:41] <DocScrutinizer05> wow, and HAHA yeah! >>Greg Kroah-Hartman started by saying that he has been working for some time with the system-on-chip (SoC) vendors to try to resolve this problem, which he blames primarily on Qualcomm for having decided not to work upstream. Qualcomm has since concluded

17:07 <Joerg-Neo900> that this decision was a mistake and is determined to fix it, but the process of doing so will take years. The other SoC vendors are also committed to closing the gap between the kernels they provide and the mainline but, again, getting there will take a while.<<

17:07 <Joerg-Neo900> btw the approach of SoC manufs providing kernels instead of proper datasheets for community to develop true FOSS kernels based on that info, that's so radically wrong I fail to appreciate it in my rants

18:47 <Joerg-Neo900> the problem starts earlier: with tivoization (locked bootloaders, forbidding reflashing with unsigned software images) and hardware manufs not providing proper datasheets/schematics. When you got a complex circuit for e.g. battery management (let's assume it been made from a Silego GP5 mixed signal matrix 'FPGA' and a few analog components like regulators, current limiters, thermosensors and so on) then it doesn't matter if the manufacturer

18:47 <Joerg-Neo900> provides the software needed to make that stuff work in form of a userland closed blob or a customized kernel driver or even upstreams their drivers to mainline kernel. Nobody will really be able to do anything about that "gibberish" even when plan C sourcecode, since the datasheets and schematics for the hardware are missing so a FOSS sw developer could understand *what* to do about the code and *how* it's actually supposed to work. Pretty

18:47 <Joerg-Neo900> much exactly same for GPUs where the drivers are worth nothing even in FOSS sourcecode when you don't have decent docs about how the internal gear in the GPU actually works. Same for cameras and you-name-it

18:52 <Joerg-Neo900> in the article linked above - and the comments there - some folks insinuate that such lockdown like tivoization was a regulatory requirement for phones. That's BS, it's not. Only the modem's radiostack is really subject to FCC et al approval

18:53 <Joerg-Neo900> it's a pity that guys like Kroah-Hartman and Torvalds seemingly fail to understand this

18:56 <Joerg-Neo900> sn0wmonster: hint: this ^^^ is pretty much related to the "security" topic

19:06 <Joerg-Neo900> btw the Librem5 is following exactly same broken approach of providing their "own firmware" which despite they explicitly claim willingness to upstream stuff to mainline, is worth nothing since they obviously cannot (or don't want to) provide the schematics and datasheets for their hardware

19:09 <Joerg-Neo900> ok, a FOSS driver makes it easier to RE the hardware, but it's only gradually different in its brokenness, while the correct approach would be to provide schematics and datasheets first and only on next lower priority provide some software reference implementation demonstrating how to use the hardware *in a limited selection of operation modes* (no sw implementation ever can cover *all* operation modes of hardware, only the datasheets +

19:09 <Joerg-Neo900> schematics can do this)

19:27 <atk> Joerg-Neo900: do you think it's possible to get the binary graphics blobs for the n900 working on the mainline?

19:29 <atk> Joerg-Neo900: also, I remember you were once giving out a link to some instructions for "what to do when you get a used n900"

19:32 <Joerg-Neo900> atk: yes, I think TI's "gfx blob" stuff as they provide it on their website should compile on mainline kernel, maybe after a few patches

19:33 <atk> cool

19:34 <Joerg-Neo900> a link to "what to do with a used N900" was possibly a link to #maemo chanlog - I don't think there's a concise comprehensive web page covering that

19:34 <Joerg-Neo900> you may try

19:34 <Joerg-Neo900> ~jrtools

19:34 <infobot> i guess jrtools is http://wiki.maemo.org/User:Joerg_rw/tools

19:34 <Joerg-Neo900> not exactly geared to the topic, but might help

19:36 <atk> re graphics: any link you could give me to that? *tries googling "TI gfx blob for n900"*

19:36 <Joerg-Neo900> mompls

19:36 <atk> I'll have a look

19:38 <Joerg-Neo900> http://omappedia.org/wiki/Graphics#SGX_software_components (dead)http://listas.gnu.org.ve/pipermail/powervr-devel/2012-September/000036.html https://pandorawiki.org/SGX_drivers

19:39 <Joerg-Neo900> "TI SGX toolkit" or somesuch

19:41 <atk> http://processors.wiki.ti.com/index.php/AM35x-OMAP35x_Graphics_SDK_Getting_Started_Guide - I found this

19:41 <Joerg-Neo900> http://www.ti.com/product/OMAP3530/toolssoftware

19:42 <Joerg-Neo900> and finally http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/gfxsdk/index.html

19:44 <atk> yes, I found that too

19:44 <atk> the wiki page claimed that no registration was required

19:44 <atk> and what a load of bullshit that was

19:44 <atk> wait no, it's on a js alert, I clicked the wrong button

19:45 <atk> it really is no registration required!

19:45 <atk> although it looks like this is an installer.... I guess I'll have to binwalk it so that I can avoid "installing" things over my pristine filesystem

19:56 <Joerg-Neo900> actually I think somebody "leaked" the complete PowerVR software sourcecode, just the FOSS community rejects to use leaked copyrighted code

19:59 <Joerg-Neo900> atk: about "what to do when you get a used n900" please ask in #maemo, there are a few nice folks willing to help

19:59 <atk> cool

19:59 <Joerg-Neo900> for a start:

20:00 <Joerg-Neo900> ~lazyflashing

20:00 <infobot> lazyflashing is, like, http://wiki.maemo.org/Updating_the_tablet_firmware#The_Lazy_Approach

20:06 <atk> Joerg-Neo900: what do you think of postmarketOS?

20:06 <atk> (I'm considering helping them get the hwaccel working for the n900 so I can eventually maybe help port the rest of the n900 stuff onto a more recent kernel and set of libraries

20:07 <atk> they have hildon UI working but it's slow because of the lack of hwaccel

20:15 <Joerg-Neo900> I didn't look into postmarketOS, so I can't really comment on it

20:16 <atk> I see

20:17 <Joerg-Neo900> a quite unfortunate side effect of Neo900 UG - I hardly have any spare time and energy for following other aspects of mobile phones or anything

20:27 <atk> how's the prototype going anyway?

20:37 <Joerg-Neo900> metacollin had a family obligation last 5 weeks, so not much progress.

20:38 <Joerg-Neo900> he'll prolly be back next few days

20:40 <Joerg-Neo900> that's the downside of tiny projects like this one: so many "single points of failure", basically every single person is such

20:41 <Joerg-Neo900> in a larger project you wouldn't even notice when one engineer catches the flu or otherwise isn't available for a few weeks

20:43 <atk> I've caught the plague myself

20:44 <atk> or something