IRC log of #cubox of Tue 01 Oct 2013. All times are in CEST < Back to index

08:26 jnettlet wumpus, ping
08:31 wumpus jnettlet: pong
08:32 jnettlet wumpus, wondering if you wanted that XO-4. I am going to send dv_'s info over to get his shipped.
08:33 wumpus ah yes, I'll send my info
08:33 wumpus but just so you know I can't promise anything, I'm extremely busy already
08:34 jnettlet wumpus, I am not asking for anything. Just want to make sure you have the hardware on hand down the road.
08:34 wumpus so it may be very useful for testing, but I can't spend much time getting devices up and running and such
08:34 wumpus okay :D
08:35 jnettlet Most of the devices are up and running. Make sure you include your phone number.
08:35 wumpus well there was a certain point months ago many people were trying to send me devices 'hey get etnaviv to work on this', I just feel guilty then :)
08:36 wumpus I send my mail and phone
08:37 wumpus sent*
08:37 jnettlet I am more interested in if I have a problem with etnaviv you will be able to replicate it to help debug if you have time.
08:37 wumpus I understand
08:37 rabeeh which gc does mmp3 use?
08:37 jnettlet GC1000
08:38 wumpus GC1000 is great, should be supported already
08:39 jnettlet it is a big.LITTLE tri-core MMP3 (although I still haven't finished the multicore stuff), 2GB's RAM, and GC1000. Which I admit I still need to test etnaviv on
08:39 wumpus tchebb did some work with that on the chromecast, he had no problems with etnaviv, however getting a working frame buffer was a bigger challenge though
08:39 cbxbiker61 damn, i just spent an hour and a half looking for one of my spare usb to ttl converters, finally found them
08:39 jnettlet wumpus, plus what geek wouldn't want a sunlight readable arm convertable notebook with touchscreen and hdmi out?
08:39 wumpus cbxbiker61: I finally managed to mcgyver something to connect to the board, but can't test it yet, still have to wait for a sdcard to arrive
08:40 cbxbiker61 now i'm finally equipped to work with the dev box
08:40 wumpus jnettlet: true, true, :)
08:40 jnettlet cbxbiker61, hurrah
08:41 cbxbiker61 wumpus, i bought three of them from amazon, with jumper connects on one end, they're beautiful
08:42 wumpus cbxbiker61: jumper connects on the usb<->ttl converter are a great idea, I had to temporary repurpose the connector of my logic analyser for that
08:43 wumpus I hope to only need the serial connection temporary anyway until I can do everything over network
08:44 cbxbiker61 i got mine mainly for debugging routers
08:44 wumpus that's why I got mine too
08:48 wumpus jnettlet: marvell, right? I think the chromecast has the same/similar SoC, except that one of the cores is locked down as "secure presentation engine"
08:49 jnettlet wumpus, the chromecast is slightly different. It is an Armada 1500 I believe.
08:49 wumpus yes, slightly different
08:50 jnettlet do you know the specs on the "secure presentation engine"? All the Marvell chips have the security core.
08:51 wumpus specs? are you kidding me :) this is marvell we're talking about
08:52 wumpus there is some vague presentation about it somewhere, maybe I can find it
08:53 jnettlet I found this which is not the mini variant. http://www.marvell.com/digital-entertainment/armada-1500/assets/Marvell-ARMADA-1500-Product-Brief.pdf
08:53 jnettlet It has the qdeo video post processing, which I wish our hardware had.
08:54 jnettlet the MMP3 has the dual-core pj4b smp, and then a third low power core which is also a Cortex-A9 just clocked down. They call it the "low power core"
08:55 jnettlet unfortunately it has been quite buggy. Just starting to get it up and fully functional. At some point I also shelved it as I wanted to go straight to Linaro's big.LITTLE work.
08:58 wumpus this is the one: http://www.marvell.com/digital-entertainment/armada-1500/assets/Marvell_%20Android_for_CE_devices-02_whitepaper_r3.pdf
11:22 shesselba wumpus: armada 1500 is quite similar to mmp3, dual Cortex-A9 with "System Manager" CPU which I suppose is Mohawk (PJ1, XScale)
11:23 shesselba but the SoC is locked down, you probably need something signed to get it started.. I know the bootrom code for BG2 (Armada 1500), BG2CD (Armada 1500-mini) might have changed
11:24 shesselba currently I am running a Sony NSZ-GS7 with gtvh's u-boot
11:24 shesselba SoC's IP is either Synopsys DesignWare or PXA/MMP stuff
11:53 jnettlet shesselba, the mmp3's third core is also a Cortex-A9
11:54 shesselba jnettlet: yeah, possible that it is also for BG2
11:55 jnettlet so it isn't big.LITTLE in the truest of terms as the LITTLE is just a low power core clocked down.
11:56 shesselba I thought the bright idea of big.LITTLE is that is a compatible core clocked down
11:57 jnettlet I think LITTLE is supposed to be a lesser core though. Like an A15 and an A7, rather than just clocking down.
11:59 jnettlet Since all the A9 cores can scale down to that same speed, I don't know why they just didn't give us 3 full speed cores and then let software cripple one if we so choose.
12:03 shesselba wumpus: make sure to get gpl sources from asus cube, the u-boot has a bunch of defines that might be interesting
12:03 shesselba mostly soc's ip registers
12:06 wumpus shesselba: thanks, I'll forward it to the guy working on chromecast, I'm sure he'll appreciate it
12:08 wumpus there he is :)
12:08 shesselba thebb? I am sure he knows about asus' u-boot includes
12:08 wumpus oh you meant for vivante in general?
12:08 shesselba with firmware defines for bg2 socs
12:09 tchebb shesselba: I haven't looked at ASUS's sources in particular. Have they got something the other bootloader sources don't?
12:10 shesselba tchebb: u-boot with register layout includes, a whole bunch of for all socs
12:15 tchebb Oh, these do look a bit more complete than the ones in the Chromecast source. Thanks!
12:16 shesselba that is what I used for mainline support patches
12:19 jnettlet rabeeh, is there anywhere on the board where I can wire up a reset switch?
12:21 tchebb shesselba: How are those coming along? Is there enough support for me to use them for my Chromecast development? I'd love to be able to use a kernel newer than 3.8.
12:23 shesselba tchebb: I kind of wasted my time to understand the bootrom, I now know how boot-up procedure works.. but haven't been able to successfully boot from spi or re-enable jtag
12:23 shesselba I guess I can get SMP, SDIO, GBE working
12:23 shesselba it's all pxa ip
12:24 shesselba s/GBE/ETH/
12:25 shesselba + USB
12:27 tchebb Nice. I'm definitely interested in trying those out when they're ready.
13:25 dv_ damnit, the yocto imx kernel isnt working
13:25 dv_ gah
13:25 dv_ kernel panics left and right
13:33 dv_ rabeeh: sometimes, after unplugging the serial output, if I re-plug immediately, the board doesnt start
13:33 dv_ I have to wait about a minute
14:34 rabeeh dv_: ?
14:35 rabeeh which usb to uart are you using now?
14:56 rabeeh jnettlet: i'v updated the wiki with the 8 pin header (has reset signal)
14:56 rabeeh http://imx.solid-run.com/wiki/index.php?title=Carrier-One_Hardware#8_pin_header_layout
14:57 jnettlet rabeeh, thanks.\
14:58 jnettlet took a break but back to fighting with the u-boot HEAD imx header generation.
14:58 jnettlet I think I have everything else sorted out, just need to get the stupid image to load.
15:00 rabeeh ?
15:01 rabeeh remember the seek and skip in the dd command
15:01 jnettlet apparently with the 2013 u-boot you don't need the skip, just the seek.
15:01 jnettlet they have changed the way the imx flash headers are generated.
15:11 rabeeh jnettlet: ddr timing is tricky
15:12 dv_ rabeeh: cp2102
15:12 rabeeh dv_: i remember something about having the uart signals connected when i.MX6 is powered off
15:12 dv_ rabeeh: btw I recommend using this figure for UART pins: http://elinux.org/images/2/2a/GPIOs.png
15:12 jnettlet rabeeh, I will double check my ddr-setup.cfg to make sure I didn't miss anything.
15:12 dv_ its much more clear
15:13 rabeeh jnettlet: send me a url please
15:13 dv_ jnettlet: how far along is your patched uboot?
15:13 rabeeh trust me; it's very tricky
15:13 jnettlet dv_, working on getting it to boot.
15:14 dv_ rabeeh: how extensively did you have to patch buildroot? I need an environment where I can run & test my gstreamer VPU plugins in
15:14 dv_ I am willing to wait until otavio & co. add support for the c1 to their layer, but I need something else in the meantime
15:16 rabeeh dv_: not that much
15:16 rabeeh i can send you mine
15:17 dv_ yes please
15:17 dv_ I dont have that much time, and need to focus my efforts on the plugins
15:17 dv_ (for now)
15:18 rabeeh dv_: sent
15:18 rabeeh it's based on nitrogenx buildroot
15:18 rabeeh ok
15:19 rabeeh mine specifically is buildroot 2013.05
15:21 dv_ alright
16:18 dv_ rabeeh: thanks, got the config
16:18 dv_ but what about kernel patches and your uboot fork?
16:18 dv_ how are these integrated into buildroot?
16:19 dv_ or did it work just fine with the default nitrogenx stuff?
17:01 rabeeh dv_; those patches are without u-boot stuff
17:50 nikitis Hello.
17:51 nikitis Anyone know where i can contact someone regarding my order?
17:55 nikitis Anyone there?
18:19 dv_ nikitis: doesnt the page have a contact section?
18:19 nikitis Yes, but i've used it and no response for days
18:20 dv_ you could also try the forums
18:20 nikitis I'm just wondering why I placed an order but wasn't charged
18:22 nikitis Do they not charge until it's shipped?
18:27 rabeeh nikitis: please contact [email protected]
18:29 dexter93 Hello rabeeh. Are there any news rearding the cubox overlay for Gentoo? (1st gen)
18:30 nikitis rabeeh, i did already
18:35 rabeeh dexter93: i'm not familiar with that issue
18:41 dexter93 http://www.solid-run.com/mw/index.php/Gentoo . The DoveFB Xorg driver overlay
18:41 dexter93 TODO: ebuild will be provided in an overlay later
18:47 dexter93 all that TODO are like they were some months ago. No maintainers for the device I guess?
19:08 rabeeh dexter93: the Gentoo stuff is not maintained.
19:08 rabeeh the tarball of the drivers is there; it needs to be integrated into gentoo ebuild
19:08 rabeeh the driver support pre and pos xorg 13 (there were some changes in the API that jnettlet kindly provided)
19:09 dv_ rabeeh: did you build uboot and the kernel with buildroot?
19:09 rabeeh dv_: nop.
19:09 rabeeh separately
19:09 rabeeh i mean the u-boot is completely out of buildroot; but the kernel is
19:09 rabeeh the kernel is part of the buildroot and hence the initramfs
19:10 dv_ so, just cloned linux-imx , then set CROSS_COMPILE= and called make ?
19:10 dv_ (and applied the patches)
19:11 rabeeh dv_: https://github.com/SolidRun/linux-imx6
19:11 rabeeh cloned from boundary devices and applied patches on top
19:12 rabeeh it's kernel 3.0.35 + bsp 4.1.0 + c1 patches
19:12 dv_ yeah
19:12 rabeeh the defconfigu should imx6_defconfig
19:12 rabeeh so in 'buildroot' you tell it to use standard imx6_defconfig
19:12 dv_ ah, from boundarydevices
19:12 dv_ I cloned from freescale directly
19:12 dv_ perhaps thats why I saw problems
19:12 rabeeh where from freescale?
19:12 rabeeh their git? or some github?
19:13 rabeeh because i couldn't find any github from freescale that has lk 3.0.35 + bsp 4.1.0 !!!
19:13 dv_ git://git.freescale.com/imx/linux-2.6-imx.git
19:44 jnettlet I am back. What did I miss?
19:45 dv_ some pain regarding faulty kernels, not much more
19:45 rabeeh tomlohave is working on getting xbmc up and running on c1
19:45 nikitis And a question about orders
19:46 jnettlet excellent.
19:47 dv_ btw. I noticed some 3.10 linux-imx branches
19:48 dv_ it seems that freescale is working on updating their patches to 3.10 as well
19:48 jnettlet dv_, _rmk_ has the c1 booting 3.12 with device-tree
19:48 dv_ oh wow
19:48 dv_ with VPU and galcore too?
19:48 jnettlet He posted his patch before he went on vacation.
19:49 jnettlet woah there...not that far into it
19:49 dv_ :)
19:49 dv_ is this part of the joint effort?
19:50 dv_ I think you mentioned it
19:52 cbxbiker61 amd just published documentation for their gpus that covers pass-through, it's really cool how amd has opened up
19:56 nikitis They only did it because they don't have a choice
19:57 cbxbiker61 hehe, well we need to make the rest of the hardware vendors "feel" they don't have a choice
19:58 nikitis With announcment of SteamOS, All serious computing is going over to Linux. And Linux demands open source
19:59 nikitis if AMD wants to sell anything past gaming consoles they have to do it
20:15 rabeeh dv_: they should be working on 3.10lts
20:16 rabeeh dv_: probably with ipu only; maybe vpu
20:16 Juggie rabeeh, will the shipping unit have a mini pci-e?
20:16 rabeeh gpu probably not (rmk stuff)
20:16 rabeeh Juggie: ?
20:16 rabeeh CuBox-i doesn't have mini pcie
20:16 rabeeh why?
20:17 rabeeh oh; about the second gig?
20:25 jnettlet rabeeh, I am porting rmk's patches into the IMX tree. I spent a good chunk of the weekend cleaning up the 4.x vivante codebase. I started bringing in my device-tree and common clock work for scaling.
20:26 rabeeh jnettlet: which tree is this?
20:27 jnettlet rabeeh, 3.11 and 3.10 lts
20:28 jnettlet still reviewing the imx6 KMS driver that was submitted but never merged.
20:29 jnettlet although that will mostly just be for Linux as Android doesn't have support for it yet.
20:35 jnettle 20:35 * jnettlet wonders if Android can run on top of the fbdev helper functions
20:44 wumpus imx6 kms driver? that sounds interesting
22:24 dv_ damn
22:25 dv_ even when I use rabeeh's forked kernel, I get "Unable to handle kernel paging request at virtual address 35797474"
22:25 dv_ (briefly tried my modified yocto here again)
22:25 dv_ well, its buildroot or openbricks for now I guess
22:32 rabeeh dv_: do you suspect there is hardware issue?
22:32 dv_ hah. no. found a big clue.
22:32 rabeeh i mean if you run the kernel and u-boot i'v sent you; do you see any issues?
22:32 dv_ yocto dora is using gcc 4.8
22:32 dv_ and apparently there are problems with kernels built with that version
22:32 dv_ rabeeh: no, your stuff works fine
22:32 rabeeh :)
22:33 rabeeh thank you; jnettlet already said that; high quality engineering sample :)
22:33 dv_ :)
22:33 dv_ I showed it to several people in the office
22:33 rabeeh and ...?
22:33 dv_ first they were wondering, "oh you got a rpi?"
22:33 dv_ then I explained things
22:34 dv_ the hw guy wanted to rip out the micro SOM to get a closer look
22:34 dv_ I said a big no :)
22:34 rabeeh hehe
22:34 rabeeh please let him see it
22:34 dv_ he was wondering about EMC
22:34 rabeeh oh
22:34 dv_ oh, the microSOM can be removed ?
22:34 rabeeh yes
22:34 rabeeh gently ofcourse
22:34 dv_ ah. well then he will be happy
22:35 rabeeh http://cubox-i.com/the-mainline-kernel-is-on/
22:35 rabeeh rmk gets the credit to get mainline on
22:36 dv_ yay
22:38 Juggie rabeeh, are you still shipping out/selling more dev boards?
22:38 Juggie or are they all gone?
22:39 rabeeh we are shipping every week few tens of boards
22:39 Juggie how does one qualify as a developer?
22:41 dv_ I dont think there are clear qualifications for this
22:41 rabeeh Juggie: sends out what porting intention he is willing to do on C1
22:41 rabeeh and internally we have 12 people setting around and discussing weather it's worthy or not :)
22:41 rabeeh hehe.. just joking
22:41 dv_ "is this one worthy of our blessing? the council shall discuss"
22:42 rabeeh there is no qualification; we simply want to avoid putting the hardware on people that don't know how to handle embedded systems in early stages
22:43 Goophy Nice to finally se some working code for the i!
22:43 Goophy I've got big hopes for this to be the first ARM HTPC that actually works for everything.,
22:43 dv_ yeah, I think so too
22:43 dv_ the ingredients are there
22:44 dv_ esp. esata and gb lan are important
22:44 Goophy And decent gpu drivers.
22:44 Goophy That's been a huge issue for others.
22:44 dv_ yeah. and eventually open source ones
22:45 Goophy I used a imx6 devboard for a while a year or two ago. The drivers were really shitty then, but it seems like they have improved heaps.
22:46 dv_ also, the overall situation with regards to (future) support is quite nice (buildroot, openbricks, archlinuxarm, OE through meta-fsl-arm)
22:46 dv_ oh yes, the GPU drivers improved a lot
22:46 dv_ there is still enough left to do though, just look at the race conditions _rmk_ found (and fixed)
22:47 Goophy Hehe, always something. But if there is a usable base, there is hope.
22:47 nikitis When is this looking to go out?
22:47 Goophy Can't wait for November. :)
22:48 nikitis November still looking good?
22:48 dv_ seems so to me
22:48 nikitis What order number are you guys?
22:49 dv_ I didnt order yet. the dev board is good enough for now
22:49 dv_ (I will order one though)
22:49 nikitis I ordered, but it didn't take my cash
22:49 nikitis So i guess it will draw from it later
22:49 dv_ I dimly remember this happening to me once
22:49 dv_ did you pay with credit card?
22:49 nikitis paypall express
22:50 dv_ hmm never used that one
22:50 dv_ I remember a delay with mastercard in a transaction
22:50 nikitis i can't get any answers from [email protected]
22:51 dv_ when did you sent the message?
22:51 nikitis i ordered mine on September 10th
22:51 nikitis i sent awhile back and today again
22:51 dv_ they are probably overloaded
22:51 dv_ patience, I guess
22:52 nikitis If I were running it I'd check 10 times daily.
22:52 nikitis that's money involved
22:52 dv_ heh. true
22:52 nikitis i want that cubox but i don't think they want my money so much :(
22:53 nikitis i was hoping whoever answers that email address was on here
22:54 nikitis I'd hate to not get one because of this if there is a shortage
22:54 dv_ did you post this in the forum?
22:54 nikitis no
22:54 dv_ try that
22:54 nikitis doesn't seem to be a proper forum
22:54 nikitis for this issue
22:55 dv_ http://imx.solid-run.com/forums/index.php ?
22:55 dv_ just post in General discussion
22:55 dv_ it makes the issue more visible, I think
22:55 dv_ also, perhaps other customers have had the same problems?
22:56 dv_ in addition, it could be a problem with paypal
22:57 nikitis maybe
22:57 nikitis but would be nice to be able to talk to someone
22:57 dv_ yes! with gcc 4.7, the kernel boots. I didnt add a devtmpfs though. but its progress.
22:57 dv_ oh I know what you mean
22:57 nikitis Get latest kernel working on it?
22:58 dv_ but as said, just be patient
22:58 dv_ cant really do anything else anyway
22:58 dv_ no, the imx kernel
22:58 dv_ but built with yocto
23:00 _rmk 23:00 * _rmk_ must teach Rabeeh to spell my name with two l's too :)
23:00 _rmk_ hi from Norfolk btw.
23:00 rabeeh _rmk_ go for the spa
23:01 dv_ rabeeh go to bed
23:01 dv_ :)
23:01 _rmk_ btw, I brought the c1 with me
23:01 rabeeh _rmk_: for a massage?
23:01 rabeeh enjoy your holiday; i'll ask Waseem to fix the 'll'.
23:02 rabeeh i'm heading to bed now; ttyl
23:02 _rmk_ their internet is not the best (it keeps getting really laggy)
23:03 nikitis dv_, i just saw on the screenshot I took that paypal express was not available at this time.
23:03 rabeeh nikitis: please send my pm with your email and order ID
23:03 nikitis ok
23:03 dv_ _rmk_: if you are in the UK, and got problems with the internet connection, you must read this: http://www.rense.com/general43/cabel.htm
23:03 rabeeh or to my email - [email protected]
23:03 dv_ its a timeless classic
23:04 _rmk_ dv: there's two problems here - one is that the closest AP doesn't always talk to the main AP, and secondly the hotel's connection isn't that good
23:05 _rmk_ dv: yes, you know NTL's alternative name? Not Til Later
23:05 dv_ :)
23:05 _rmk_ they apply that to everything they did, whether it be support or the delivery of IP packets :)
23:06 dv_ at least they are consistent :)
23:10 _rmk_ that they are
23:10 _rmk_ or were
23:12 nikitis Did the guy think to march down to their offices?
23:12 nikitis That's what I had to do to get service here in USA
23:12 nikitis Now it's great
23:13 nikitis i know i'm kind of missing the point but that's what I would have done
23:13 _rmk_ nikitis: its entirely possible that their offices were a distance away from the guy
23:14 nikitis Can't be that far. There's like limit to how long those lines can go.
23:14 _rmk_ NTL is cable, so fibre in the street
23:14 nikitis I would love fiber....
23:15 _rmk_ and coax from the a very local box
23:15 _rmk_ from back in the 1990s era
23:15 _rmk_ well...
23:15 nikitis coax has to be within a mile to the nearest ctms
23:15 nikitis or it degrades
23:16 nikitis especially in the 90's
23:16 nikitis I used to cable modem hack back in the day before it was deemed illegal
23:16 dv_ coax for the last mile, fiber for the rest?
23:16 nikitis maybe
23:16 nikitis most isp's aren't setup for fiber like that
23:17 nikitis i'm probably one of the few people who even know how to cable modem hack nowadays
23:18 _rmk_ well, today at home, I get 4-6mbps down via copper from the local office (which is about 2 miles in terms of cable distance away)
23:18 _rmk_ if I go for their fibre, I can get that up to around 13mbps as the cabinet is then about 1 mile away
23:18 nikitis can always just grab a config file from someone on a CTMS that you are not on. As the hmac's are only verrified at that level. If you are on a different ctms but same ISP, you can use their config.
23:19 nikitis that's half the battle
23:19 nikitis The other half is getting it onto your cable modem
23:19 nikitis some are still easily accessed via snmp protocol
23:19 nikitis others have to use a jtag and flash the nand
23:20 nikitis But i won't go further into detail
23:20 nikitis I don't do it anylonger anyway
23:20 nikitis since they passed laws
23:21 nikitis I'll stay with my 10Mbps down 1.5Mbps up
23:24 jnettle 23:24 * jnettlet feels spoiled with 45Mbps down and 4.5Mbps up
23:25 _rmk_ I think we'll stay with ours, because the fibre is more expensive... and just to get ADSL2+ speeds, its not worth it
23:25 _rmk_ jnettlet: if I lived directly next to one of these cabinets, that would be possible
23:26 _rmk_ or even 300mbps if you're prepared to pay
23:26 jnettlet They are rolling out fiber in Aarhus but not down near me. I am on DSL and just got speed bumped.
23:27 jnettlet I am actually on a 50/5 plan but you never get those speeds. Works out well because I can traffic shape the house wifi to 20Mbps and use the rest for my work network.
23:28 jnettlet however I rarely need all the bandwidth.
23:29 jnettle 23:29 * jnettlet is going to be now. have an early morning tomorrow.
23:29 jnettlet ttyl.
23:29 _rmk_ cu