12:49 <wpwrak> freemangordon, Pali: hmm, does anything in fremantle actually use ECI ? i can see ECI[5] (eci_sw_gpio) getting initialized in the kernel, but that's all i can find.

12:49 <Pali> ECI? do you mean Nokia ECI headsets?

12:50 <Pali> IIRC they are not supported by N900 HW

12:50 <wpwrak> yes, that famout "Enhancement whatever Interface". well, there's a fancy bit of circuit for it in n900.

12:51 <Pali> but I read somewhere it N900 HW could support it somehow

12:51 <Pali> but answer from Nokia kernel developer was "unsupported by HW"

12:52 <wpwrak> kewl :)

12:54 <wpwrak> DocScrutinizer05: are you sure we want to keep that mess ? the protocol specification is some secret nokia took to its grave, it seems. the msc thesis about eci says nothing about the protocol. the kernel driver nokia once submitted assumes the use of an unknown controller chip. i couldn't find any other sources / leaks / reverse-engineering either. so for all practical purposes, this seems unusable.

12:55 <Pali> eci headsets are supported on Nokia N9 and Nokia N950

12:55 <Pali> and support is in Nokia kernel (2.6.32)

12:55 <wpwrak> aah, where would i find that kernel ?

12:56 <Pali> it was on harmattan DVD

12:56 <wpwrak> grr :)

12:56 <Pali> and was in harmattan SDK

12:56 <Pali> but I have copy in linux-n900 git tree

12:56 <Pali> https://github.com/pali/linux-n900

12:56 <Pali> branch v2.6.32-nokia

12:59 <wpwrak> thanks ! let's clone the whole bugger then ...

12:59 <Pali> there is also v2.6.28-nokia branch in which is official nokia n900 kernel source code

13:00 <Pali> (versioned with all publically available versions)

13:00 <wpwrak> i suspect what i'll find is that driver that talks to the mystery chip. but let's see ...

13:00 <Pali> IIRC that code from 2.6.32 just export some memory via sysfs

13:01 <Pali> and userspace ohmd plugin parse it

13:01 <Pali> ohmd with all plugins were opensourced in harmattan

13:01 <Pali> on gitorious

13:01 <Pali> so now when gitorious is working in R/O mode again, you can look into source code

13:02 <wpwrak> when / if ever :)

13:04 <wpwrak> pity that they didn't just a do a "forced" migration to gitlab. granted, some people would have complained, but there's also a lot who just can't be bothered to migrate anywhere

13:05 <wpwrak> anyway, that's a great lead. thanks a lot !

14:16 <DocScrutinizer05> wpwrak: ECI is also generic AV jack detection, so please don't try tro optimize that out

14:24 <DocScrutinizer05> ((answer from Nokia kernel developer was "unsupported by HW")) they seen there is a driver that uses an external MCU and they answered "we don't have any such MCU so we don't support ECI"

14:26 <DocScrutinizer05> the SoC has a way to enable and disable the micbias (I suggest video analog switch) and it has an ADC to read in the inbound signal. That's honestly all that's needed for ECI

14:29 <wpwrak> (optimize out) what i'm trying to figure out are the timing requirements. so far, all i have is that the msc thesis 1) says that ECI is basically I2C, and that 2) it operates I2C at 400 kbps. thus, there is nothing to indicate that ECI does not operate at a fairly high speed.

14:30 <wpwrak> and twl5031 (undocumented, right ?) appears to have a dedicated hardware block (ACI) for it.

14:30 <wpwrak> so that would also suggest non-trivial speed

14:31 <DocScrutinizer05> there were scope pics (iirc from jacekowski) that didn't look like highspeed

14:32 <DocScrutinizer05> iirc they looked more like HDQ

14:32 <wpwrak> would you have some bit rate ? :)

14:33 <DocScrutinizer05> I first would need the scope snapshots

14:34 <DocScrutinizer05> anyway for the IO extenmder stuff it's fair to assume all three GPIO being low-latency IRQ

14:35 <DocScrutinizer05> maybe it's more economic when I evaluate the IRQ timing?

14:36 <wpwrak> hmm, isn't ECI[5] an output ? (goes to 2SEL of the switch)

14:36 <DocScrutinizer05> so?

14:37 <wpwrak> that seems to contradict "all three GPIO being low-latency IRQ"

14:38 <wpwrak> (irq timing) well, i was preparing a detailed analysis since you rejected my first quick assignment. but sure, if you want to do it, please go ahead

14:39 <wpwrak> all i really want is to place the io expander chip in the block diagram ;-)

14:41 <DocScrutinizer05> no, it doesn't, see [5]

14:42 <DocScrutinizer05> we have two comparator outputs plus one direct ADC line

14:42 <DocScrutinizer05> all THREE are low-latency, the rest (4, 5) are most likely outputs

14:43 <DocScrutinizer05> you're looking at the schematics (I hope), while I talk from top of my head

14:44 <wpwrak> where is [4] ?

14:44 <wpwrak> (sch) yes, of course

14:50 <DocScrutinizer05> and where is [0]?

14:50 <DocScrutinizer05> thje Nokia schematics are a mess sometimes

14:51 <DocScrutinizer05> err, I found 0, 1, 5

14:52 <DocScrutinizer05> so rtaher "where's 2,3,4?"

14:53 <DocScrutinizer05> funny enough ECI_AD doesn't count

14:53 <DocScrutinizer05> so we allegedly have 7 lines

14:57 <DocScrutinizer05> sorry, busy with fixing a vital box

15:05 <DocScrutinizer05> ok, 2GB deleted, stuff fixed

15:28 <wpwrak> yes, 2, 3, and 4 are another mystery. "the eci bermuda triangle" :)