**** BEGIN LOGGING AT Mon Dec 24 02:59:58 2012
Dec 24 03:00:22 <Russ> don't ask?
Dec 24 03:00:53 <Russ> kryogenifex
Dec 24 03:01:57 <Russ> http://kryogenifex.com/
Dec 24 03:05:06 <emeb_mac> aha - that would make sense.
Dec 24 03:33:44 <mranostay> Liquid Nitrogen? why would you need that?
Dec 24 03:33:57 <mranostay> industry grade cooling?
Dec 24 03:35:06 <emeb_mac> special effects, really cold drinks, etc.
Dec 24 03:54:20 <mranostay> emeb_mac: fog machines i guess?
Dec 24 03:54:31 <emeb_mac> yep
Dec 24 05:29:32 <ds2> \
Dec 24 05:30:24 <emeb_mac> /
Dec 24 07:10:03 <_av500_> mranostay: !
Dec 24 07:13:00 * mranostay hides
Dec 24 07:14:53 <emeb_mac> run awaaaayyyyyy
Dec 24 07:18:28 * mranostay uses emeb_mac as a human shield
Dec 24 07:20:19 * emeb_mac chants "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn"
Dec 24 07:23:23 * _av500_ shrugs
Dec 24 07:25:57 <mranostay> how is it in _av500_ world?
Dec 24 07:33:56 <chainz> Hi everyone, any one handled the AM3354 with TI PMIC TPS65910A, I am meeting some problem with that..
Dec 24 07:41:32 <mranostay> which are?
Dec 24 07:53:25 <mranostay> heh i'm enjoying this FSF story
Dec 24 08:54:04 <_av500_> the Furry Software Foundation?
Dec 24 09:07:55 <mranostay> heh
Dec 24 09:08:24 <mranostay> panto: aren't you suppose to be doing xmas things?
Dec 24 09:09:28 <panto> hi mranostay
Dec 24 09:09:42 <panto> mranostay, I did lots of them over the weekend
Dec 24 09:09:58 <panto> will do more during the evening
Dec 24 09:14:14 * panto just found that no-one ever bothered making dt platform devices unregister them selves :/ 
Dec 24 09:15:53 <mranostay> panto: rmmod kernel panics?
Dec 24 09:16:18 <panto> no even rmmod
Dec 24 09:16:24 <panto> just unregistering the device
Dec 24 09:16:57 <panto> resource allocation is stuck in the middle ages (no parent/sibling handling)
Dec 24 09:17:38 <dm8tbr> you can check out any time you like, but you can never leave...
Dec 24 09:17:55 <panto> pretty much
Dec 24 09:18:17 * dm8tbr continues to humm hotel california
Dec 24 09:18:43 <mranostay> panto: is this why the kernel doesn't reboot on AM335x?
Dec 24 09:18:56 <panto> no, that's a different b0rkeness
Dec 24 09:19:08 <mranostay> little annoying to have to use the BDI to reset the damn board
Dec 24 09:19:20 <panto> we're slowly getting there
Dec 24 09:21:26 <mranostay> dm8tbr: i have this new thing everytime i see someone mention a song i pull up Youtube is see if someone has done a dubstep version :)
Dec 24 09:23:07 <mranostay> panto: *ahem* ecap :P
Dec 24 09:23:24 * panto kicks mr
Dec 24 09:23:28 * panto kicks mranostay 
Dec 24 09:27:13 <_av500_> mranostay: !!
Dec 24 09:28:44 <panto> hi _av500_
Dec 24 09:32:04 <mranostay> panto: it is just software :)
Dec 24 09:36:33 <_av500_> hi panto
Dec 24 09:36:36 <_av500_> nice tree
Dec 24 09:36:44 <_av500_> is it balanced?
Dec 24 09:36:58 <mranostay> he had to chop the top off
Dec 24 09:37:35 <panto> _av500_, it's a P-tree* (A panto tree with a star on top)
Dec 24 09:37:40 <_av500_> :)
Dec 24 09:37:45 <_av500_> mranostay: top chop I did too
Dec 24 09:38:01 <panto> no point in the holidays if you don't go completely overboard
Dec 24 09:39:16 <_av500_> panto: kids getting submarines for presents?
Dec 24 09:39:26 <panto> no kids :)
Dec 24 09:39:31 <_av500_> ah
Dec 24 09:39:32 <_av500_> what?
Dec 24 09:39:55 <panto> I don't have any kids (at least yet)
Dec 24 09:40:02 <panto> I do have a god-son
Dec 24 09:40:03 <mranostay> or you know of? :P
Dec 24 09:40:12 <panto> he got a toy car and a sweater
Dec 24 09:40:19 <_av500_> no submarine?
Dec 24 09:40:27 <_av500_> think of hte children
Dec 24 09:40:30 <_av500_> the german children
Dec 24 09:40:30 <panto> do you want a submarine? :)
Dec 24 09:40:45 <panto> a u-boot submarine perhaps?
Dec 24 09:41:08 <panto> we can arrange a sale?
Dec 24 09:41:15 <_av500_> no, we sel to you
Dec 24 09:41:17 <_av500_> sell
Dec 24 09:41:29 <_av500_> nice submarines
Dec 24 09:41:34 <_av500_> every greek should have one
Dec 24 09:41:36 <_av500_> or two
Dec 24 09:41:44 <panto> if a bit leany during turns
Dec 24 09:42:13 <mranostay> i'm sure the greeks can totally afford that
Dec 24 09:42:20 <_av500_> mranostay: they have to
Dec 24 09:42:26 <panto> or else...
Dec 24 09:42:29 <_av500_> their eastern neighbours can attack at any time
Dec 24 09:42:42 <_av500_> (we sell them subs too)
Dec 24 09:42:59 <mranostay> what do they pay for it in. gyros? :)
Dec 24 09:43:05 <_av500_> no
Dec 24 09:43:12 <mranostay> islands? :)
Dec 24 09:43:18 <_av500_> deal
Dec 24 09:44:21 <mranostay> panto: so i've heard taking pictures of military bases in .gr is a good way to get locked up?
Dec 24 09:44:51 <panto> mranostay, excellent way
Dec 24 09:45:08 <panto> especially if your game company also makes military training simulators
Dec 24 09:45:22 <mranostay> heh
Dec 24 09:45:45 <panto> ah, and on top of that you were caught once doing it and let go
Dec 24 09:45:59 <panto> those military guys don't have a sense of humour
Dec 24 09:46:28 <panto> I'm sure you can totally do that same with .us military bases
Dec 24 09:46:28 <mranostay> possibly worse than police :)
Dec 24 09:47:10 <panto> it's military police probably, so policemen with better guns
Dec 24 09:47:16 <mranostay> the police aren't suppose to shoot people generally :)
Dec 24 09:47:23 <_av500_> no?
Dec 24 09:47:38 <_av500_> isnt there an amendment for that?
Dec 24 09:47:39 <panto> what about all the .us movies I've seen
Dec 24 09:47:52 <panto> I'm sure those guys shooting at the baddies were policemen, no?
Dec 24 09:48:21 <mranostay> yeah well most cops never use their guns in their careers hate to break it to you :)
Dec 24 09:48:36 <mranostay> i think something like 95% never do
Dec 24 09:49:03 <mranostay> but writing tickets and filing paperwork doesn't sell movies :)
Dec 24 09:50:09 <panto> it does fill the state's coffers
Dec 24 09:50:27 <_av500_> not coffins
Dec 24 09:59:04 * panto 's mind boggles with the OF resource handling code 
Dec 24 11:15:29 * panto boggles again 
Dec 24 11:15:44 <panto> how much more fscked can this get?
Dec 24 12:47:58 <mdp> panto, stayed tuned...there's more treasures that await you.
Dec 24 13:05:31 <panto> hi mdp
Dec 24 13:07:46 <mdp> hi panto
Dec 24 13:10:00 <mdp> working on !capebus things?
Dec 24 13:10:52 <panto> hi mdp
Dec 24 13:11:02 <panto> that is complete
Dec 24 13:11:11 <panto> now I'm trying to unfsck the platform device layer
Dec 24 13:11:27 <panto> guess what? no-one ever tried unregistering a platform device
Dec 24 13:12:04 <mdp> hrm
Dec 24 13:12:11 <mdp> that sounds vaguely familiar for some reason
Dec 24 13:12:30 <panto> yeah, all that nice resource handling in the devices?
Dec 24 13:12:42 <panto> it's sorta working on OF but not really
Dec 24 13:12:56 <panto> does resource DMA & friends ring a bell?
Dec 24 13:13:02 <mdp> the vision is more important than reality
Dec 24 13:13:10 <mdp> yes
Dec 24 13:13:15 <panto> more like blindness
Dec 24 13:13:33 <panto> for how long does nested resources exist?
Dec 24 13:13:38 <panto> seems like for ages
Dec 24 13:13:59 <panto> platform code just doesn't bother with it, and when you try to release them, boom
Dec 24 13:14:10 <mdp> nested resources?
Dec 24 13:14:26 <panto> resources can now be in tree
Dec 24 13:14:31 <mdp> right
Dec 24 13:14:40 <mdp> you mean in the DT?
Dec 24 13:14:45 <panto> so there's a parent, sibling relationship
Dec 24 13:14:57 <panto> no, in core device model resources
Dec 24 13:15:19 <panto> take a look at the definition at ioport.h
Dec 24 13:15:26 <mdp> oh, well, we were doing that correct prior to the generic device model when it was just in PCI ;)
Dec 24 13:15:53 <panto> well, the generic device model does it right now
Dec 24 13:16:01 <panto> but the platform code is fscked
Dec 24 13:16:12 <mdp> so let's see...'98ish on a 603e based VME board ;)
Dec 24 13:16:15 <panto> hehe
Dec 24 13:16:23 <panto> but you lacked vision!
Dec 24 13:16:26 <panto> and synergy!
Dec 24 13:16:45 <mdp> well yeah, the PCI subsystem had this done right first, then Pat took the bigger vision and made that resource model across everything
Dec 24 13:17:03 <panto> I see a whole bunch of memcpy(resource*) and playing with it
Dec 24 13:17:06 <mdp> of course, platform devices were only visioned for PeeCee stuff at that time
Dec 24 13:17:14 <panto> which is totally bogus when there's a tree structure
Dec 24 13:17:36 <panto> since all this is mostly static, no-one noticed
Dec 24 13:18:07 <mdp> you are probably the first fresh set of eyes on the core OF stuff
Dec 24 13:18:41 <panto> it's archaic
Dec 24 13:19:07 <mdp> I immediately ran into problems just on the periphery, myself, though. if you'll recall my concern over the omap "external h/w reset" platform device function pointer callbacks
Dec 24 13:19:14 <panto> yes
Dec 24 13:19:29 <panto> it's even worse at places
Dec 24 13:19:43 <mdp> I feel bad now in a way, because the guy I was arguing with got laid off :(
Dec 24 13:19:51 <panto> :(
Dec 24 13:19:56 <panto> which was it? Tony?
Dec 24 13:20:03 <mdp> Tony is not a TIer
Dec 24 13:20:12 <mdp> that was Omar Luna Ramirez
Dec 24 13:20:21 <mdp> another shameful loss
Dec 24 13:20:26 <panto> :/
Dec 24 13:20:29 <mdp> but let's not go there ;)
Dec 24 13:20:50 <panto> I feel this !capebus thing got legs
Dec 24 13:21:11 <panto> I can imagine a whole bunch of places where it's useful
Dec 24 13:21:12 <mdp> disagreements are just that...he just didn't see that creating a framework that involves callbacks is archaic and not acceptable in OF-land
Dec 24 13:21:44 <panto> it's better to argue during development than cry during deployment
Dec 24 13:21:59 <panto> engineers should have a thick skin IMO
Dec 24 13:22:15 <mdp> it's another thing I suppose I need to pick up else I'll never have pruss on am335x in mainline ;)
Dec 24 13:22:26 <panto> heh
Dec 24 13:22:41 <mdp> am180x is lovely for regression testing but it's not *useful* to me for projects
Dec 24 13:23:38 <mdp> did you see my idea for a spi slave zero device?
Dec 24 13:23:58 <mdp> I think I have an elegant solution for real spi testing
Dec 24 13:24:12 <panto> must've missed it
Dec 24 13:24:13 <panto> link?
Dec 24 13:24:27 <panto> I have some ideas for testing as well
Dec 24 13:25:32 <mdp> panto, just shared friday on another channel
Dec 24 13:25:54 <panto> on freenode?
Dec 24 13:26:25 <mdp> basically, I started some f/w for the PRU with the idea of making something similar to usb gadget zero...but for spi controller testing
Dec 24 13:26:40 <panto> oh, that's nice
Dec 24 13:27:04 <mdp> I was debugging some problem reported at work with spidev_test.c and looked how the common use is just looping mosi/miso
Dec 24 13:27:26 <mdp> well, that doesn't test enough, since it ignores clocking modes..
Dec 24 13:28:16 <mdp> so basically you need some definition of a sink/source device so you can test all the modes
Dec 24 13:28:21 <panto> I think many SPI controllers have loopback modes
Dec 24 13:28:32 <panto> perhaps you can use that too in some way
Dec 24 13:28:36 <mdp> some do, but not mine
Dec 24 13:29:06 <mdp> there's only one in the kernel that implements the SPI_LOOP support
Dec 24 13:29:15 <mdp> so anyway, I need it, so that matters ;)
Dec 24 13:29:17 <panto> what about looping two different spi controllers?
Dec 24 13:29:27 <panto> that way you could test potentially everything
Dec 24 13:30:23 <mdp> hrm...that could be a more universal approach
Dec 24 13:30:46 <panto> so more like spi 0 master, spi 0 slave,
Dec 24 13:30:52 <mdp> except, one problem
Dec 24 13:31:43 <mdp> slave support in the kernel is non-existent (attempts have been made) however, the biggest problem is responding to transfers appropriately when the master controls the clock
Dec 24 13:32:02 <mdp> that's the number 1 reason I'm doing PRU for this simple slave stuff
Dec 24 13:32:16 <panto> I don't get it
Dec 24 13:32:32 <panto> so what if the master controls the clock? isn't that a common use case?
Dec 24 13:33:02 <mdp> ok, let's try it this way...what are you looping to?
Dec 24 13:33:11 <mdp> there's no slave support drivers in the kernel
Dec 24 13:33:24 <panto> aha, there's for you first item :P
Dec 24 13:33:28 <mdp> no
Dec 24 13:33:45 <panto> slave mode is super useful IMO
Dec 24 13:34:18 <panto> cheapest fastest on board point to point link there can be
Dec 24 13:34:35 <mdp> sure, but I'm not sure I want to reopen that battle..I've already followed the existing failed attempts at that
Dec 24 13:35:05 <mdp> plus, I'd have to doublecheck that our IP is even good enough to deal with the kernel limitations
Dec 24 13:35:38 <mdp> the biggest problem, panto, is that you can't do much of anything in s/w at a 48MHz spi bus rate
Dec 24 13:35:46 <panto> right
Dec 24 13:35:52 <panto> and it can go much higher
Dec 24 13:35:53 <mdp> you can't handle interrupts due to lack of determinism..etc. etc.
Dec 24 13:36:07 <panto> so SPI controller using PRUSS?
Dec 24 13:36:11 <mdp> best you can do is fill a response fifo and let it event trigger
Dec 24 13:36:30 <mdp> panto, so that's why I said it needs to be implemented in PRUSS to be a useful test tool
Dec 24 13:37:24 <mdp> you *could* even implement a defined "slave zero" device in any uC...just with a lower max clock speed it could handle
Dec 24 13:37:52 <mdp> so what I'm trying to do is make something that's easy to implement elsewhere too
Dec 24 13:38:02 <mdp> on beaglebone..you just jump a few wires and use it
Dec 24 13:38:19 <mdp> test all clock speeds up to 48MHz ideally
Dec 24 13:38:55 <mdp> I did get our test prople (after some prodding) to add a couple more useful options to spidev_test.c, btw, which I'll proxy upstream now
Dec 24 13:39:37 <panto> what's the highest SPI clock?
Dec 24 13:39:42 <panto> on am?
Dec 24 13:39:57 <mdp> mcspi goes to 48MHz (11!)
Dec 24 13:40:58 <mdp> basically, I want to have some OOB commands that change word width
Dec 24 13:41:44 <mdp> which is easy in the PRU context..other implementations of slave zero could whack their f/w over usb connection to the uC
Dec 24 13:42:18 <mdp> then you do some standardize xmit/recv pattern thing
Dec 24 13:42:25 <mdp> supporting various lengths
Dec 24 13:42:49 <mdp> OOB commands to change word width && clock mode that's sampled
Dec 24 13:43:15 <mdp> so then on the host side you still just use the spidev_test.c and call it in your N different variations
Dec 24 13:44:15 <mdp> but you have a chameleon slave device that can behave as a device that's N different word widths and M different clock modes
Dec 24 13:45:17 <mdp> panto, in the PRU implementation if you want to change word/clock mode you just write a newmode/reset command to pru memory from userspace which it would detect and restart in the new mode
Dec 24 13:45:31 <mdp> that's what I mean by an OOB command to the slave zero device
Dec 24 13:45:46 <panto> right
Dec 24 13:50:45 <mdp> a fast m3/m4 with it's spi slave device might make a nice lab unit to implement this same thing
Dec 24 13:52:12 <mdp> panto, you can also make it do i2c too... ;)
Dec 24 13:52:32 <panto> heh
Dec 24 13:52:52 <mdp> uart...
Dec 24 13:53:27 <mdp> the biggest problem with all these drivers is that the code paths aren't test properly, IMHO
Dec 24 13:53:47 <mdp> it's the usual problem of people not writing test code for every line of code they write
Dec 24 15:57:40 <panto> http://pastebin.com/D7gSrsnA
Dec 24 19:20:50 <ds2> 22z
Dec 24 19:30:59 <emeb_mac> 11y
Dec 24 19:33:23 <mranostay> can i use floating points here?
Dec 24 19:33:39 <mranostay> 5.5x
Dec 24 19:34:16 <dm8tbr> 56aa
Dec 24 19:36:54 <mranostay> heh preproduction r-pis...
Dec 24 19:57:26 <emeb_mac> what _did_ the world do before rpi showed up?
Dec 24 19:58:16 <mranostay> suck less?
Dec 24 19:58:47 <emeb_mac> haha
Dec 24 20:26:50 <mranostay> emeb_mac: more old broadcom chips has have a home? :)
Dec 24 20:27:22 <emeb_mac> instead of cluttering up landfills?
Dec 24 20:28:47 <mranostay> except the fact you had to pop a PCB on the old chips
Dec 24 23:27:45 * mranostay smacks head
Dec 24 23:30:09 <emeb> why?
Dec 24 23:31:25 <mranostay> emeb: well i was writing PRU code to the same PRU
Dec 24 23:31:34 <mranostay> of course the first one never worked :)
Dec 24 23:34:28 <emeb> oops
Dec 24 23:44:14 <mranostay> debugging this assembly is a pain
Dec 25 00:44:06 <mranostay> av500: is that bear photo yin yang? :)
Dec 25 01:31:03 <mranostay> woot standalone clock app is working
**** ENDING LOGGING AT Tue Dec 25 02:59:58 2012